Academic Edi o : Alessand o
Gaspa e o
Recei ed: 19 No embe 2024
Re ised: 11 Janua y 2025
Accep ed: 13 Janua y 2025
Published: 16 Janua y 2025
Ci a ion: Pa ício, L.; Va ela, L.;
Sil ei a, Z. P oposal o a Sus ainable
Model o In eg a ing Robo ic P ocess
Au oma ion and Machine Lea ning in
Failu e P edic ion and Ope a ional
E iciency in P edic i e Main enance.
Appl. Sci. 2025,15, 854. h ps://
doi.o g/10.3390/app15020854
Copy igh : © 2025 by he au ho s.
Licensee MDPI, Basel, Swi ze land.
This a icle is an open access a icle
dis ibu ed unde he e ms and
condi ions o he C ea i e Commons
A ibu ion (CC BY) license
(h ps://c ea i ecommons.o g/
licenses/by/4.0/).
A icle
P oposal o a Sus ainable Model o In eg a ing Robo ic P ocess
Au oma ion and Machine Lea ning in Failu e P edic ion and
Ope a ional E iciency in P edic i e Main enance
Leonel Pa ício 1,* , Leonilde Va ela 1and Zilda Sil ei a 2
1
Depa men o P oduc ion and Sys ems, Algo i mi/LASI, Uni e si y o Minho, 4804-533 Guima ães, Po ugal;
[email p o ec ed]
2Depa men o Mechanical Enginee ing, Sao Ca los School o Enginee ing, Uni e si y o Sao Paulo,
Sao Paulo 13566-590, B azil; [email p o ec ed]
*Co espondence: [email p o ec ed]
Abs ac : This pape p oposes a sus ainable model o in eg a ing obo ic p ocess au oma-
ion (RPA) and machine lea ning (ML) in p edic i e main enance o enhance ope a ional
e iciency and ailu e p edic ion accu acy. The esea ch iden i ied a key gap in he li e a u e,
namely he limi ed in eg a ion o RPA, ML, and sus ainabili y in p edic i e manu ac u ing,
which led o he de elopmen o his model. Using he PICO me hodology (Popula ion,
In e en ion, Compa ison, Ou come), he s udy e alua ed he implemen a ion o hese
echnologies in Alpha Company, compa ing esul s be o e and a e he model’s adop-
ion. The in e en ion in eg a ed RPA and ML o imp o e ailu e p edic ion accu acy
and op imize main enance ope a ions. Resul s showed a 100% inc ease in mean ime
be ween ailu es (MTBF), a 67% educ ion in mean ime o epai (MTTR), a 37.5% dec ease
in main enance cos s, and a 71.4% educ ion in unplanned down ime cos s. Challenges
such as ini ial implemen a ion cos s and he need o con inuous aining we e also no ed.
Fu u e esea ch could explo e in eg a ing big da a and AI o u he imp o e p edic ion
accu acy. This model demons a es ha in eg a ing RPA and ML leads o ope a ional im-
p o emen s, cos educ ions, and en i onmen al bene i s, con ibu ing o he sus ainabili y
o indus ial ope a ions.
Keywo ds: RPA; machine lea ning; in eg a ion sys ems; p edic i e main enance; SIRPM
1. In oduc ion
P edic i e main enance has eme ged as an inc easingly impo an s a egy o op i-
mizing he ope a ion o complex indus ial sys ems, enabling he an icipa ion o ailu es
and he p e en ion o un o eseen down ime. Howe e , cu en solu ions ace signi ican
challenges. The in eg a ion o da a om di e se sou ces, he complexi y o p edic i e
models, and he lack o scalabili y in con en ional app oaches con inue o pose subs an ial
obs acles. Fo ins ance, in he au omo i e indus y, ailu es in high-p ecision machine y
o en lead o ex ended down imes, inc easing ope a ional cos s and a ec ing p oduc ion. In
he ca dboa d packaging indus y, ailu es in c i ical equipmen can hal en i e p oduc ion
lines, esul ing in un o eseen cos s, deli e y delays, and a signi ican loss o compe i i e-
ness. These issues unde sco e he u gen need o mo e e ec i e and adap able solu ions
in p edic i e main enance [1].
Al hough cu en echnologies enable mo e accu a e ailu e p edic ions, he implemen-
a ion o solu ions emains a challenge, pa icula ly due o da a in eg a ion, he complexi y
Appl. Sci. 2025,15, 854 h ps://doi.o g/10.3390/app15020854
Appl. Sci. 2025,15, 854 2 o 47
o p edic i e models, and he lack o scalable app oaches. The digi iza ion o indus ial
ope a ions and he g owing adop ion o au oma ion ha e d i en inno a ion in his ield.
Robo ic p ocess au oma ion (RPA) has eme ged as an e icien solu ion o au oma e epe i-
i e, ule-based asks, eeing up esou ces o mo e s a egic ac i i ies and enabling g ea e
p ecision in ope a ions. When combined wi h machine lea ning (ML) echniques, RPA
no only imp o es ailu e p edic ion accu acy bu also enhances ope a ional agili y and
e iciency [2].
In his con ex , his a icle p oposes an inno a i e and sus ainable model ha in e-
g a es RPA and ML o ailu e p edic ion and he imp o emen o ope a ional e iciency in
p edic i e main enance. The model aims no only o inc ease he accu acy o p edic ions
bu also o p omo e sus ainabili y in i s h ee dimensions, which a e en i onmen al, social,
and economic. En i onmen ally, he in eg a ion o hese echnologies can signi ican ly
educe esou ce consump ion and CO
2
emissions by p e en ing unexpec ed ailu es and
op imizing ene gy and ma e ial use. Socially, he model p oposes a be e alloca ion o
human esou ces, enabling wo ke s o ocus on mo e s a egic asks, such as da a analysis
and inno a ion, a he han dealing wi h epe i i e asks o eme gency issues. Economi-
cally, he educ ion in ope a ional and main enance cos s can enhance compe i i eness and
p o i abili y, con ibu ing o long- e m inancial sus ainabili y [2].
The speci ic p oblem add essed by his s udy is he ine iciency o adi ional p e-
dic i e main enance sys ems. These models o en ail o an icipa e ailu es e ec i ely,
leading o un o eseen down imes and high main enance cos s. The p oposed RPA and ML
model seeks o o e come hese limi a ions, allowing o accu a e ailu e p edic ion and
op imized esou ce managemen . Fo example, in a company wi hin he ca dboa d packag-
ing sec o , he implemen a ion o his model esul ed in a 66.67% educ ion in unplanned
down ime e en s and a 52% dec ease in a e age epai cos s, leading o g ea e e iciency
and p o i abili y.
The need o adop he RPA and ML model is e iden , gi en he inc easing eliance on
echnological solu ions in indus ies. T adi ional p edic i e main enance models a e o en
nei he scalable no su icien ly accu a e o he demands o mode n indus ial ope a ions.
The in eg a ion o au oma ion and eal- ime ad anced da a analy ics enables swi and
e ec i e in e en ion while also o e ing lexibili y o adap o di e en con ex s and
indus ial sec o s [3].
The unique con ibu ion o his wo k lies in he p ac ical and inno a i e applica ion
o he in eg a ed RPA and ML model in p edic i e main enance. This s udy no only seeks
o enhance he e iciency o indus ial p ocesses bu also demons a es how eme ging
echnologies can con ibu e o sus ainabili y, expanding he possibili ies o adap a ion and
implemen a ion ac oss a ious indus ies. The p oposed model o e s signi ican alue o
small and medium-sized en e p ises, such as Alpha, which ace simila challenges ela ed
o unplanned down ime and high main enance cos s.
The s uc u e o his a icle is as ollows: Sec ion 2p o ides a comp ehensi e e iew
o he s a e o he a , add essing he concep s o p edic i e main enance, RPA, and ML
and explo ing he in eg a ion o hese echnologies in ailu e p edic ion and ope a ional
e iciency imp o emen . Sec ion 3desc ibes he me hodology adop ed, jus i ying he
choice o he PICO app oach and explaining he hypo hesis o mula ion p ocess. Sec ion 4
p esen s he p oposed model and i s ad an ages o e exis ing models. Sec ion 5illus a es
he p ac ical applica ion o he model h ough a case s udy. In Sec ion 6, he esul s a e
analyzed, discussing he implica ions o RPA and ML in eg a ion in ailu e p edic ion and
ope a ional e iciency, wi h a ocus on sus ainabili y aspec s. Finally, Sec ion 7concludes
he a icle, summa izing he main indings and sugges ing di ec ions o u u e esea ch.
Appl. Sci. 2025,15, 854 3 o 47
In summa y, his a icle aims o deepen he unde s anding o he in eg a ion be ween
RPA and ML o ailu e p edic ion and he op imiza ion o ope a ional e iciency in p e-
dic i e main enance, highligh ing he social, economic, and en i onmen al bene i s o
applying hese eme ging echnologies.
2. S a e o he A
P edic i e main enance has eme ged as one o he mos e ec i e app oaches o
imp o ing ope a ional e iciency and inc easing equipmen du abili y ac oss a ious in-
dus ial sec o s. This p ac ice elies on a con inuous moni o ing o ope a ional pa ame e s
and he use o ad anced analy ical models o p edic equipmen ailu es o deg ada ion,
allowing o main enance o be ca ied ou a he igh momen be o e a ailu e occu s [
4
].
The main goal o p edic i e main enance is o maximize equipmen a ailabili y, educe
ope a ional cos s, p e en unexpec ed down ime, and imp o e sys em sa e y and eliabil-
i y [5].
P edic i e main enance is cha ac e ized by a da a-d i en and p edic i e analysis
app oach, which di e s om adi ional me hods such as co ec i e and p e en i e main-
enance. Ins ead o pe o ming main enance a ixed in e als o wai ing o equipmen
ailu e, he aim is o an icipa e ailu es based on eal- ime da a and analy ical models,
enabling p ecise, scheduled in e en ions. This ype o main enance no only educes
cos s bu also inc eases e iciency by op imizing asse li ecycles and imp o ing o e all
ope a ional pe o mance [6].
The au oma ion o p ocesses and ad anced da a analysis ha e played an inc eas-
ing ole in p edic i e main enance, wi h obo ic p ocess au oma ion (RPA) and machine
lea ning (ML) being he wo mos impo an echnologies in his con ex .
Robo ic p ocess au oma ion (RPA) e e s o he use o so wa e o au oma e epe i i e,
ule-based asks wi hin compu a ional sys ems. RPA is essen ial o collec ing and p ocess-
ing la ge olumes o da a om moni o ing senso s, main enance managemen sys ems,
and In e ne o Things (IoT) de ices. Au oma ing hese p ocesses elimina es he need o
human in e en ion in adminis a i e asks, allowing ope a o s o ocus on deepe analyses
and s a egic decisions [7].
On he o he hand, machine lea ning (ML), a sub ield o a i icial in elligence (AI),
enables compu e sys ems o lea n om da a and dynamically adjus as new in o ma ion is
acqui ed. In indus ial se ings, ML is used o analyze senso da a, moni o ope a ional
condi ions, and p edic ailu es by iden i ying pa e ns and anomalies ha migh go unno-
iced du ing adi ional inspec ions. This p edic i e capabili y con ibu es o a signi ican
shi in asse main enance, making i mo e e ec i e and cos -e icien [8].
The in eg a ion o RPA and ML in p edic i e main enance has b ough signi ican
bene i s o a ious indus ies. A clea example o RPA applica ion is he au oma ion o da a
collec ion p ocesses, whe e equipmen condi ion senso s ansmi eal- ime in o ma ion o
main enance managemen sys ems. RPA can p ocess hese da a and au oma ically o wa d
hem o main enance eams o moni o ing sys ems o u he analysis. In e ms o machine
lea ning, neu al ne wo ks a e o en used o p edic ailu es by analyzing la ge olumes o
his o ical da a and iden i ying ends and pa e ns ha indica e he onse o an impending
ailu e [9].
Howe e , despi e hese echnological ad ancemen s, some limi a ions emain. The
main issue conce ns he dependence on da a quali y. The e ec i eness o ML algo i hms
depends on he quali y o he da a used o ain he models. Inaccu a e o incomple e da a
can comp omise he accu acy o p edic ions, leading o inco ec diagnoses. Addi ionally,
implemen ing RPA and ML equi es signi ican in es men s in in as uc u e and he need
o specialized aining o sys em ope a ion and main enance [
10
]. Ano he limi a ion is
Appl. Sci. 2025,15, 854 4 o 47
he need o cus omize ML models o each ype o asse and indus ial en i onmen , which
can be cos ly and complex.
P edic i e main enance, suppo ed by RPA and ML echnologies, has p o en pa icu-
la ly bene icial in he manu ac u ing sec o , whe e ope a ional e iciency and minimizing
unplanned down ime a e c i ical o p oduc i i y. RPA plays a key ole in au oma ing
adminis a i e and ope a ional p ocesses, such as scheduling main enance, sending ale s,
and managing spa e pa s in en o y. The in eg a ion o RPA wi h p edic i e main enance
sys ems allows ope a ions o be mo e e icien and less p one o human e o s, which is
essen ial in high-demand p oduc ion en i onmen s.
The use o ML in manu ac u ing, in u n, is one o he g ea es ad ances in p edic i e
main enance. ML sys ems can moni o a ange o ope a ional pa ame e s o equipmen in
eal ime, such as empe a u e, ib a ion, p essu e, and wea , iden i ying ailu e pa e ns
ha may indica e imminen ailu es. These sys ems enable mo e e ec i e in e en ions,
educing co ec i e main enance cos s and ex ending equipmen li espan while also mini-
mizing p oduc ion cos s by a oiding unplanned down ime [11].
Fu he mo e, he combina ion o RPA and ML c ea es a closed-loop cycle o moni-
o ing and ac ion, whe e ailu es can be p edic ed and in e en ions can be ca ied ou
au onomously wi hou he need o cons an manual in e ac ion. This no only imp o es
ope a ional e iciency bu also con ibu es o sus ainabili y by educing esou ce was e
and ex ending asse li espans. This cycle op imizes equipmen usage and minimizes he
en i onmen al impac o indus ial ope a ions, aligning wi h he p inciples o a ci cula
economy [11].
The applica ion o p edic i e main enance esul s in signi ican imp o emen s in
ope a ional e iciency, as i enables he iden i ica ion o ailu es be o e hey occu , p e en ing
unexpec ed down ime and he associa ed cos s o eme gency epai s. Main enance is
ca ied ou only when necessa y, educing ope a ional cos s and imp o ing esou ce
u iliza ion [
11
]. Sus ainabili y is also suppo ed, as p edic i e main enance helps educe
ma e ial and ene gy was e by ensu ing ha equipmen ope a es op imally and p e en ing
ca as ophic ailu es ha could lead o la ge amoun s o was e.
Addi ionally, he use o RPA and ML, by au oma ing p ocesses and op imizing da a
analysis, con ibu es o he e iciency o main enance and ope a ional p ocesses, educing
human e o s and imp o ing decision-making accu acy. These echnologies a e unda-
men al o he digi al ans o ma ion o indus ial main enance, c ea ing in elligen and
au onomous sys ems ha ensu e g ea e eliabili y and e iciency in ope a ions [11].
Sus ainabili y, a c i ical aspec o mode n indus ial p ac ices, is buil on h ee ounda-
ional pilla s, namely economic, social, and en i onmen al pilla s. These pilla s emphasize
e icien esou ce u iliza ion, equi able socie al impac , and en i onmen al p ese a ion,
espec i ely. P edic i e main enance aligns wi h hese p inciples by enabling ope a ions
ha a e bo h cos -e ec i e and en i onmen ally esponsible. Th ough i s ocus on e iciency
and p oac i e managemen , p edic i e main enance con ibu es o sus ainable p ac ices,
ensu ing ha indus ies mee p esen needs wi hou comp omising u u e esou ces [11].
The in eg a ion o p edic i e main enance echnologies, such as obo ic p ocess au-
oma ion (RPA) and machine lea ning (ML), u he enhances sus ainabili y ou comes.
Economically, hese echnologies educe cos s by op imizing equipmen li ecycles and
minimizing unplanned down ime. Socially, hey imp o e wo kplace sa e y by iden i ying
and mi iga ing po en ial haza ds be o e hey lead o ailu es. En i onmen ally, p edic i e
main enance lowe s ene gy consump ion and ma e ial was e, ein o cing a commi men
o educing indus ial oo p in s. This app oach is pa icula ly ele an in he con ex
o a ci cula economy, whe e was e is minimized and esou ce alue is p ese ed ac oss
ope a ional p ocesses [12–14].
Appl. Sci. 2025,15, 854 5 o 47
By seamlessly embedding sus ainabili y in o p edic i e main enance s a egies, in-
dus ies can achie e a balanced app oach ha maximizes ope a ional e iciency while
add essing ecological and socie al esponsibili ies [
15
,
16
]. The p oac i e na u e o p edic-
i e main enance, suppo ed by ad anced analy ical ools, allows o ganiza ions o ex end
asse li espans, educe was e, and ope a e mo e sus ainably. This in eg a ion no only sup-
po s long- e m business esilience bu also unde sco es he ole o p edic i e main enance
in ad ancing b oade sus ainabili y objec i es wi hin indus ial ecosys ems [12,17].
In summa y, he in eg a ion o RPA and ML echnologies in p edic i e main enance
ep esen s a signi ican ad ancemen in ope a ional e iciency, ailu e p edic ion, and
sus ainabili y. These echnologies no only imp o e asse managemen bu also p o ide a
mo e sus ainable app oach o companies seeking o op imize hei esou ces and educe
en i onmen al impac while espec ing he pilla s o sus ainabili y, namely he economic,
social, and en i onmen al pilla s.
3. Me hodology
3.1. Me hod
The PICO me hodology (Popula ion, In e en ion, Compa ison, Ou come) has been
widely used in scien i ic esea ch due o i s abili y o p o ide a clea and well-de ined
amewo k o sys ema ically analyzing esea ch ques ions. When applied igo ously,
his app oach enables an objec i e and de ailed e alua ion o he ela ionships be ween
key elemen s o he s udy, ensu ing mo e accu a e in es iga ions ee om he subjec i e
biases o en p esen in na a i e e iews [
18
]. PICO has p o en o be a aluable ool
no only in medical and heal hca e ields bu also ac oss a ious o he a eas o esea ch,
such as enginee ing, social sciences, and echnology, as highligh ed by ecen s udies
(include upda ed e e ences) [
19
–
21
]. This me hodology acili a es he o mula ion o
speci ic esea ch ques ions and he selec ion o ele an s udies, ensu ing ha da a a e
handled consis en ly and compa ably [19].
In his s udy, he use o he PICO me hodology is based on he need o s uc u e he
analysis in a way ha in eg a es a ious dimensions o he esea ch p oblem, o e ing a
comp ehensi e and objec i e iew o he adop ion o obo ic p ocess au oma ion (RPA) and
machine lea ning (ML) in p edic i e main enance sys ems. Adap ing his me hodology
o he speci ic con ex o au oma ion and a i icial in elligence echnologies allows o a
clea ocus on he a iables o in e es and p o ides a p ecise assessmen o he impac o
implemen ing hese echnologies on o ganiza ions’ ope a ional pe o mance [20].
In his s udy, he “popula ion” e e s o o ganiza ions ha ha e adop ed o a e con-
side ing implemen ing RPA and ML solu ions wi hin he ealm o p edic i e main enance.
De ining he popula ion is c ucial o ensu ing ha he esul s a e ele an and applicable o
sec o s acing simila challenges ela ed o main enance and he in eg a ion o in elligen
sys ems [
22
]. Focusing on companies ha a e applying ad anced echnologies o op imize
p ocesses and inc ease ope a ional e iciency allows o an in-dep h analysis o adop ion
p ac ices and hei impac on speci ic indus ial en i onmen s.
The “in e en ion” in his s udy e e s o he adop ion o RPA and ML echnologies
aimed a imp o ing ailu e p edic ion models, he eby inc easing he e iciency o main-
enance ope a ions. The choice o his in e en ion is suppo ed by g owing e idence
ha in eg a ing RPA and ML can p o ide signi ican bene i s, such as imp o ed accu acy
in ailu e p edic ions and educed ope a ional cos s. Addi ionally, i is c ucial ha he
implemen a ion o hese echnologies is ca ied ou in a sus ainable manne , adhe ing o
en i onmen al, social, and economic p inciples, con ibu ing o a mo e esponsible busi-
ness model aligned wi h sus ainabili y goals (include upda ed e e ences on RPA and ML
applied o p edic i e main enance) [23–25].
Appl. Sci. 2025,15, 854 6 o 47
The “compa ison” in his s udy is made be ween he s a e o o ganiza ions be o e and
a e he implemen a ion o RPA and ML echnologies. To de elop he analy ical model, a
e iew o p e ious s udies was conduc ed, co e ing ele an opics and app oaches. This
e iew helped iden i y gaps in he li e a u e and bes p ac ices, which we e inco po a ed
in o he p oposed model. The compa ison aims o assess he impac s o adop ing hese
echnologies on o ganiza ional ope a ional e iciency, wi h a ocus on imp o ing ailu e p e-
dic ions and op imizing main enance p ocesses. The s udy also seeks o de e mine how he
implemen a ion o hese echnologies can add ess common issues aced by companies, such
as lack o main enance con ol, and iden i y angible bene i s, such as educed unexpec ed
ailu es, lowe ope a ional cos s, and inc eased eliabili y in main enance p ocesses.
The “ou come” expec ed om his s udy is he iden i ica ion o subs an ial imp o e-
men s in he ope a ional e iciency o o ganiza ions ha adop he p oposed model. The
applica ion o he PICO me hodology will allow o he o mula ion o obus hypo heses
and conduc ing an analysis based on conc e e da a, acili a ing he objec i e measu emen
o esul s. This will p o ide a clea e unde s anding o how he implemen a ion o RPA
and ML can op imize p edic i e main enance while p omo ing sus ainabili y in indus ial
ope a ions ac oss en i onmen al, social, and economic dimensions.
The me hodology used in his s udy in ol ed a de ailed analysis o a igo ously
selec ed se o ele an da a sou ces. The PICO app oach was applied o selec he a icles,
wi h well-de ined inclusion and exclusion c i e ia. Ini ially, he a icles we e iden i ied
h ough a bibliog aphic sea ch, and he sc eening p ocess in ol ed e iewing i les and
abs ac s, p io i izing s udies ha di ec ly add essed he in eg a ion o RPA, ML, p edic i e
main enance, and sus ainabili y. A icles ha did no mee he ele ance c i e ia o we e
no di ec ly ela ed o he opic we e disca ded, esul ing in a inal selec ion o s udies ha
signi ican ly con ibu e o he esea ch [26].
The in o ma ion ex ac ed was based on con ibu ions om expe s in he ields o
RPA, ML, p edic i e main enance, and sus ainabili y. The collec ion o a icles was ca ied
ou om enowned scien i ic da abases, such as he “B-on” pla o m, h ps://www.b-on.
p /) (accessed on 10 No embe 2024), which p o ides access o pee - e iewed scien i ic
publica ions. “B-on” is a widely accessible pla o m, including a icles indexed in da abases
such as ISI WOS and Scopus, ensu ing ha he selec ed sou ces a e comp ehensi e and
ele an o he subjec ma e .
Thus, he use o he PICO me hodology o e s a clea amewo k o he analysis and
in e p e a ion o he collec ed da a, acili a ing he connec ion be ween he o mula ed
hypo heses and he e idence ound in he li e a u e.
The inclusion c i e ia we e as ollows:
•
A icles di ec ly add essing he implemen a ion o e alua ion o RPA and ML in
p edic i e main enance.
•
Publica ions discussing he ela ionship be ween hese echnologies and sus ainabili y
in i s a ious aspec s (en i onmen al, social, and economic).
•S udies conduc ed be ween 2010 and 2024.
•A icles published in pee - e iewed jou nals and a ailable in ull ex .
The ollowing ypes o publica ions we e excluded:
•
A icles ha do no di ec ly add ess he implemen a ion o RPA and ML o ha co e
a eas un ela ed o p edic i e main enance.
•Wo ks ha do no signi ican ly discuss sus ainabili y aspec s.
•Publica ions ha a e no a ailable in ull ex o a e di icul o access.
•
S udies ha a e no based on empi ical me hodologies o ha do no p o ide clea
quan i a i e and quali a i e da a on he impac s o he echnologies.
Appl. Sci. 2025,15, 854 7 o 47
This igo ous selec ion p ocess ensu ed ha he analysis was based on high-quali y
sou ces di ec ly ele an o he s udy o he e ec s o adop ing RPA and ML in p edic i e
main enance and sus ainabili y.
The cen al esea ch ques ion and he hypo heses ha guided his s udy we e app o-
p ia ely o mula ed.
Cen al Resea ch Ques ion (CRQ)
CRQ: how can he in eg a ion o obo ic p ocess au oma ion (RPA) and machine lea n-
ing (ML) imp o e ailu e p edic ion and ope a ional e iciency in p edic i e main enance
sys ems, conside ing sus ainabili y in i s en i onmen al, social, and economic dimensions?
Hypo heses (H)
H1. The in eg a ion o RPA and ML imp o es accu acy in p edic ing ailu es in indus ial sys ems,
educing down ime and ope a ional cos s.
H2. The implemen a ion o a sus ainable RPA and ML model in p edic i e main enance con ibu es
o ope a ional e iciency, p omo ing en i onmen al (such as educed was e and ene gy), social (be e
alloca ion o human esou ces), and economic (cos educ ion and inc eased p o i s) bene i s.
This s udy aims o p o ide a de ailed iew on how he combina ion o obo ic p ocess
au oma ion (RPA) and machine lea ning (ML) can enhance ailu e p edic ion and inc ease
ope a ional e iciency in p edic i e main enance sys ems while p omo ing, a he same
ime, sus ainabili y in i s en i onmen al, social, and economic aspec s.
The wo hypo heses p esen ed suppo he idea ha he combina ion o RPA and ML
b ings ope a ional bene i s and can also be a undamen al app oach o a mo e sus ainable
main enance model. The i s hypo hesis (H1) sugges s ha he in eg a ion o RPA and
ML imp o es ailu e p edic ion accu acy and educes ope a ional cos s by imp o ing
main enance p ocesses, esul ing in less down ime and g ea e equipmen longe i y. This
p og ess is essen ial o educing was e and cos s, aligning wi h he economic objec i es
o sus ainabili y.
The second hypo hesis (H2) is based on he p emise ha he adop ion o a sus ainable
model, which combines RPA and ML in p edic i e main enance, op imizes ope a ional
e iciency and also b ings conside able bene i s in sus ainabili y. The expec a ion is ha by
educing ailu es and op imizing main enance in e en ions, i will be possible o minimize
he was e o na u al esou ces and he use o ene gy, a o ing mo e sus ainable business
p ac ices. Fu he mo e, by enabling mo e e ec i e managemen o human esou ces, his
model can imp o e social well-being in o ganiza ions by eeing p o essionals o asks o
g ea e s a egic and analy ical alue while con ibu ing o a mo e p o i able and esilien
economic model.
To ca y ou he sea ch p ocess cen al o his s udy, he esea che s u ilized he
online scien i ic lib a y o e ed by he Po uguese Founda ion o Science and Technology,
concen a ing on h ee speci ic g oups (G oup 1, G oup 2, G oup 3, and G oup 4), as
de ailed in Table 1.
The esea ch es s we e ca ied ou using he “B-on” pla o m, u ilizing he OR ope a o
o link ei he he i le, keywo ds (KWs), o abs ac (AB) wi hin he h ee de ined g oups.
Following his, il e s we e applied o he se s o publica ions ob ained du ing he
esea ch p ocess, and he esul s, in e ms o he numbe o publica ions, a e summa ized
in Table 2.
Appl. Sci. 2025,15, 854 8 o 47
Table 1. G oups sea ched h ough “B-on”.
G oup 1 G oup 2 G oup 3 G oup 4
“RPA” OR “Robo ic
P ocess
Au oma ion” OR
“In elligen P ocess
Au oma ion” OR
“Digi al P ocess
Au oma ion” OR
“Business Wo k low
Au oma ion”
“Machine
Lea ning” OR “ML
Algo i hms” OR
“Supe ised
Lea ning” OR
“Unsupe ised
Lea ning” OR
“Rein o cemen
Lea ning” OR
“Deep Lea ning”
“P edic i e
Main enance” OR
“Condi ion-Based
Main enance” OR
“Main enance
Op imiza ion”
“Sus ainabili y” OR
“Sus ainable” OR
“Social
Sus ainabili y” OR
“En i onmen ” OR
“En i onmen al
Sus ainabili y” OR
“Economic
Sus ainabili y” O
“Sus ainable
De elopmen ”
Table 2. Publica ions ob ained h ough B-on a e he applica ion o some il e s.
Se 1 Se 2 Se 3
Ini ial Resul : 436 825 2970
1—Res ic o Pee -Re iewed 352 792 2318
2—F om 2010 o 2024 351 780 2300
3—Language: English 238 200 1667
4—Res ic o Full Tex 227 200 1530
Following he applica ion o he il e s, he i les, keywo ds, and abs ac s o each
pape we e examined o selec hose mos closely ela ed o he esea ch opic. Ini ially, a
o al o 4231 a icles was e ie ed. A e he il e s we e applied, 1957 emained, o which
only 37 we e di ec ly ele an o he s udy’s ocus.
Figu e 1shows a lowcha ou lining he p ocess o he li e a u e sea ch and he
sc eening p ocedu e ollowed in his s udy.
Figu e 1. Flow diag am o he li e a u e sea ch and espec i e sc eening.
3.2. A icles Syn hesis and Analysis
This sec ion p o ides a de ailed summa y and analysis o he a icles mos pe inen o
he opic being in es iga ed. Table 3, shown below, lis s he 37 selec ed a icles along wi h
Appl. Sci. 2025,15, 854 9 o 47
he models discussed in each. This able was designed o classi y he con ibu ions o each
s udy and was ca e ully cons uc ed based on a ho ough sea ch o academic da abases.
Table 3. Iden i ied a icles and he espec i e hemes o he a icles ound.
Themes o he A icles
A icles
(Au ho /Yea /Re .)
P edic i e
Main enance
Robo ic
P ocess
Au oma ion
Machine
Lea ning
Type o
Con ibu ion
Pilla s o Sus ainabili y
Iden i ied in he
A icles
[26] Zhang, X.; Liu, T.;
Wang, H. (2020) x x Re iew En i onmen al
[
27
] Bai, X.; Li, J.; Zhang, Y.
(2020) x x Model En i onmen al
[28] Vasiliadis, L.;
Mani sa is, S.; Kapsalis, A.
(2021)
x Case S udy Economic, Social
[29] Lubis, L.; Sembi ing,
D. (2023) x Case S udy Economic
[30] Ma ínez-Gomez, J.;
Ma ínez, P.; Ramos, F.
(2021)
x x Re iew En i onmen al
[31] Singh, R.; Singh, A.;
Pandey, P. (2020) x Re iew En i onmen al
[32] Xu, B.; Yu, D.; Hu, X.
(2021) x Re iew En i onmen al
[
33
] San os, G.; Oli ei a, J.;
Sil a, P. (2021) x Case S udy Economic
[34] Chen, W.; Liu, Z.;
Yang, F. (2020) x x Re iew En i onmen al
[35] Anwa , A.; Mohd Ali,
N.; Ali, I. (2020) x x Case S udy En i onmen al
[36] Chen, J.; Zhang, H.;
Ma, Q. (2020) x Case S udy Social, Economic
[37] Li, Y.; Ma, Y.; Liu, L.
(2021) x x Re iew En i onmen al
[38] Pisacane, O.; Po ena,
D.; An oma ioni, S.;
Be ilacqua, M.; Cia apica,
F.; Diaman ini, C. (2020)
x Case S udy En i onmen al
[39] Bala aman, K.;
Palaniappan, S.; Zhuang, J.
(2021)
x Re iew En i onmen al
[40] Ga cía, F.; López, J.;
Díaz, R. (2021) x Re iew En i onmen al
[41] Lee, J.; Da a i, H.;
Singh, J. (2021) x Case S udy En i onmen al
[42] Ribei o, P.; Sil a, F.;
Rocha, A. (2021) x Case S udy En i onmen al
[43] Sha ma, N.; Ga g, H.;
A o a, A. (2021) x Case S udy En i onmen al
[44] Li, X.; Zuo, H.; Wang,
J. (2020) x Re iew En i onmen al
[
45
] Basu, R.; Kumbha , D.;
Acha ya, R. (2021) x Re iew En i onmen al
Appl. Sci. 2025,15, 854 16 o 47
•
Cloud-based s o age and con ol: The model p io i izes cloud solu ions o ensu e
esilience, scalabili y, and access o ad anced machine lea ning (ML) ools. How-
e e , i is adap able, allowing o ini ial con ol in on-p emise in as uc u es and a
p og essi e e olu ion o cloud solu ions as o ganiza ions ad ance in he p ocess o
echnological mode niza ion.
•
Implemen a ion o machine lea ning logic: The ML logic in SIRPM is designed o be
implemen ed bo h in he cloud and on-p emise, in a scalable and lexible way. In he
con ex o I4, cloud echnologies o e access o ad anced ML lib a ies and obus
compu a ional capabili ies, o e coming in e nal limi a ions ha many companies
s ill ace.
•
Mobili y and accessibili y in in as uc u e: The hyb id a chi ec u e o he model
ensu es ha e en wi hou a comple e mig a ion o he cloud, o ganiza ions can enjoy
he bene i s o connec i i y and mobili y, essen ial o I4. Fo example, he au oma ion
o da a collec ion and p ocessing by RPA can be achie ed locally bu wi h in e aces
eady o u u e in eg a ion in o cloud-based solu ions, p omo ing a con inuous low
o in o ma ion be ween machines, sys ems, and people.
•
S uc u ed ansi ion o cloud-based Indus y 4.0: The model p epa es he echnologi-
cal en i onmen o o ganiza ions o a con inuous and sus ainable e olu ion owa ds
he p inciples o I4. Each phase o SIRPM, om da a collec ion o he analysis and
in eg a ion o esul s, is designed o be ully compa ible wi h cloud in as uc u es,
ensu ing a ha monious and e icien ansi ion.
SIRPM combines lexibili y and inno a ion o mee he immedia e needs o p edic i e
main enance while o e ing a s uc u ed pa h o digi al ans o ma ion based on Indus y 4.0.
The unc ioning o he SIRPM model will be de ailed s ep by s ep in he ollow-
ing phases:
The SIRPM model p o ides a lexible, s uc u ed app oach o in eg a ing RPA and
ML in o p edic i e main enance sys ems. I emphasizes he sus ainable applica ion o
hese echnologies while ensu ing in e ope abili y wi h exis ing sys ems. The lexibili y
embedded in he model allows i o be adap ed o di e en o ganiza ional con ex s, ensu ing
a esponsible and e icien app oach o p edic i e main enance.
By ollowing he de ined c i e ia o each phase, he in eg a ion o RPA and ML
will align wi h bes p ac ices o sus ainabili y and ope a ional e iciency while allowing
o ganiza ions o cus omize echnology choices based on hei speci ic needs.
4.2. Cha ac e is ics and Bene i s o he Model (SIRPM)
The SIRPM model p esen s cha ac e is ics (Table 10) and bene i s (Table 11) ha make
i a obus and well-sui ed solu ion o p edic i e main enance sys ems in indus ial en i-
onmen s.
This sec ion highligh s he main echnical ea u es and he ad an ages associa ed wi h
i s implemen a ion.
The SIRPM model o e s an ideal balance be ween au oma ion, in elligence, and
lexibili y, enabling companies o maximize p oduc i i y while minimizing cos s and
en i onmen al impac s. The modula s uc u e and con inuous lea ning make he model a
long- e m solu ion o companies seeking inno a ion and sus ainabili y in managing hei
indus ial asse s.
Below, he compa a i e Table 12 is p esen ed be ween he exis ing s udies and he
p oposed model (SIRPM).
Appl. Sci. 2025,15, 854 17 o 47
Table 10. Model (SIRPM) cha ac e is ics.
Model (SIRPM) Cha ac e is ics
Cha ac e is ics Desc ip ion
1. In eg a ion o Ad anced
Technologies
•Combines obo ic p ocess au oma ion (RPA),
machine lea ning (ML), and he IoT.
•Enables con inuous da a low be ween senso s,
analysis pla o ms, and ope a ional eams.
2. In elligen Au oma ion
•Au oma es da a collec ion, p ocessing, analysis, and
execu ion o main enance- ela ed ac ions.
•Reduces manual in e en ion, minimizing human
e o s.
3. Con inuous Lea ning
•The machine lea ning model is con inuously
upda ed wi h main enance eedback, becoming mo e
e icien and accu a e o e ime.
4. Modula Con igu a ion
•
Clea ly di ided in o phases ha can be implemen ed
g adually, depending on he company’s needs.
•Suppo s di e en algo i hms and senso s, ensu ing
lexibili y o in eg a ion wi h exis ing sys ems.
5. Real-Time Moni o ing •
Allows eal- ime da a analysis o anomaly de ec ion
and immedia e ale gene a ion.
6. Sus ainabili y and Ene gy
E iciency
•
P omo es was e educ ion and esou ce op imiza ion
h ough de ailed p edic i e analyses.
Table 11. Model (SIRPM) bene i s.
Model (SIRPM) Bene i s
Bene i s Desc ip ion
1. Inc eased Ope a ional
E iciency
•Minimizes unplanned down ime h ough p oac i e
p edic i e main enance.
•Enhances p oduc i i y wi h as e and mo e p ecise
in e en ions.
2. Cos Reduc ion
•P e en s ca as ophic ailu es and ex ends he equipmen ’s
li espan.
•Op imizes human and ma e ial esou ces based on eliable
o ecas s.
3. Da a-D i en
Decision-Making
•Gene a es de ailed insigh s o imp o e asse managemen .
•P o ides s uc u ed da a ha can be used o c ea e
pe o mance me ics and s a egic epo s.
4. Co po a e
Sus ainabili y
•Reduces ene gy consump ion and minimizes was e,
con ibu ing o en i onmen al goals.
•Helps mee egula ions ela ed o indus ial sus ainabili y.
5. Con inuous
Imp o emen and
Adap a ion
•Feedback om ale s allows o an e alua ion o hei
use ulness, adjus ing he sys em as needed.
•Machine lea ning capabili y ensu es he sys em s ays
upda ed wi h changes in he ope a ional en i onmen .
6. Team Empowe men
•P omo es adap a ion o ope a ional eams o mode n
echnologies, inc easing hei e iciency and echnical
knowledge.
7. Scalabili y and
Flexibili y
•Can be expanded o o he a eas o sec o s o he company
wi h simila needs.
•
Suppo s a wide ange o indus ies, om manu ac u ing o
logis ics and ene gy.
Appl. Sci. 2025,15, 854 18 o 47
Table 12. Compa a i e analysis be ween he exis ing s udies and he p oposed SIRPM model.
Technology
In eg a ion
(RPA, Machine
Lea ning, IoT)
Me hod/
Algo i hm
In elligen
Au oma ion
Con inuous
Lea ning
Flexibili y
and
Scalabili y
Real-Time
Moni o ing
Ene gy
E iciency and
Sus ainabili y
[26] Zhang, X.; Liu,
T.; Wang, H. (2020) Deep Lea ning A i icial Neu al
Ne wo ks (ANNs) Mode a e
[27] Zhang, W.;
Yang, D.; Wang, H.
(2019) Deep Lea ning A i icial Neu al
Ne wo ks (ANNs) Mode a e
[28] Vasiliadis, L.;
Mani sa is,
S.;Kapsalis, A.
(2021)
RPA Rule-Based
Au oma ion, UI
Au oma ion x High
[29] Lubis, L.;
Sembi ing, D.
(2023) RPA Rule-Based
Au oma ion, UI
Au oma ion x High
[30]
Ma ínez-Gomez,
J.; Ma ínez, P.;
Ramos, F. (2021)
Machine
Lea ning
Random Fo es ,
Decision T ees,
SVM x Mode a e
[31] Singh, R.;
Singh, A.; Pandey,
P. (2020)
Machine
Lea ning
Random Fo es ,
Decision T ees,
SVM x Mode a e
[32] Pech, M.;
V cho a, J.; Bednᡠ,
J. (2021)
Machine
Lea ning, IoT Random Fo es ,
KNN, SVM x Mode a e Limi ed
[33] San os, G.;
Oli ei a, J.; Sil a, P.
(2021) RPA Rule-Based
Au oma ion, UI
Au oma ion High
[34] Chen, W.; Liu,
Z.; Yang, F. (2020) Machine
Lea ning Random Fo es ,
Decision T ees x Mode a e
[35] Anwa , A.;
Mohd Ali, N.; Ali, I.
(2020)
Machine
Lea ning Random Fo es ,
Decision T ees x Mode a e
[36] Ribei o, J.,
Lima, R., Eckha d ,
T., & Pai a, S.
(2020)
RPA, AI Rule-Based
Au oma ion x x Mode a e x Mode a e
[37] Kuma , P.,
Khalid, S., & Kim,
H. (2023)
Machine
Lea ning ANN x Mode a e Mode a e
[38] Kuma , R.;
Nai , R. (2021) Machine
Lea ning
Random Fo es ,
Decision T ees,
SVM x Mode a e Limi ed
[39] Bala aman, K.;
Palaniappan, S.;
Zhuang, J. (2021)
Machine
Lea ning Random Fo es ,
Decision T ees Mode a e
[40] Zon a, T.;
Cos a, C.; Righi, R.;
Lima, M.; Da
T indade, E.; Li, G.
(2020)
AI, IoT
Gene ic Algo i hms
x Mode a e x
[41] Lee, J.; Da a i,
H.; Singh, J. (2021) AI
Rein o cemen
Lea ning
Algo i hms, Sea ch
Algo i hms
x High x x
[42] Ribei o, P.;
Sil a, F.; Rocha, A.
(2021) AI, IoT
Gene ic Algo i hms
x Mode a e x Mode a e
Appl. Sci. 2025,15, 854 19 o 47
Table 12. Con .
Technology
In eg a ion
(RPA, Machine
Lea ning, IoT)
Me hod/
Algo i hm
In elligen
Au oma ion
Con inuous
Lea ning
Flexibili y
and
Scalabili y
Real-Time
Moni o ing
Ene gy
E iciency and
Sus ainabili y
[43] Sha ma, N.;
Ga g, H.; A o a, A.
(2021)
Machine
Lea ning Random Fo es ,
SVM x Mode a e
[44] Çına , Z.;
Nuhu, A.; Zeeshan,
Q.; Ko han, O.;
Asmael, M.; Sa aei,
B. (2020)
AI
Bayesian Ne wo ks
x Mode a e x
[45] Abidi, M.;
Mohammed, M.;
Alkhale ah, H.
(2022)
Machine
Lea ning Random Fo es ,
Decision T ees x Mode a e Limi ed
[46] Sobczak, A.;
Zio a, L. (2021) RPA Rule-Based
Au oma ion x High Mode a e
[47] Cheng, Y.;
Zhou, J.; Zhao, Y.
(2020) AI Sea ch Algo i hms,
Dynamic
P og amming Mode a e x
[
48
] Al es, S.; Sil a,
J.; Lima, J. (2021) RPA Rule-Based
Au oma ion High
[49] Pa il, A.; Nai ,
M.; Thomas, S.
(2021)
IoT, Machine
Lea ning
Random Fo es ,
Decision T ees,
KNN x Mode a e Mode a e
[50] Teoh, Y.; Gill,
S.; Pa likad, A.
(2021)
Machine
Lea ning Random Fo es ,
Decision T ees x Mode a e
[51] Meye , D.;
Redi, J.; Smi h, T.
(2021) AI
Rein o cemen
Lea ning
Algo i hms Mode a e
[52] Klies ik, T.;
Nica, E.; Du ana, P.;
Popescu, G. (2023) AI, IoT Rein o cemen
Lea ning
Algo i hms x Mode a e x Mode a e
[53] E-Fa ima, K.;
Khandan, R.;
Hosseinian-Fa , A.;
Sa wa , D. (2023)
RPA Rule-Based
Au oma ion, UI
Au oma ion High
[54]
Ruiz-Sa mien o, J.;
Mon oy, J.; Mo eno,
F.; Galindo, C.;
Bonelo, J.; Jiménez,
J. (2020)
Machine
Lea ning Random Fo es ,
Decision T ees x Mode a e
[55] Wang, H.,
Zhang, W., Yang,
D.; Xiang, Y. (2023).
Deep Lea ning,
IoT Deep Neu al
Ne wo ks (DNNs) x x High x x
[56] Pa ício, L.;
Va ela, L.; Sil ei a,
Z. (2024) RPA, AI Rule-Based
Au oma ion, UI
Au oma ion x High x Mode a e
[57] Cos a, C.R.S.;
Pa ício, L.;
Fe ei a, P.; Va ela,
L.R. (2023)
RPA Rule-Based
Au oma ion x High x Mode a e
[58] Pa ício, L.;
Cos a,
C.R.S.;Fe nandes,
L.P.; Va ela, M.L.R.
(2023)
RPA Rule-Based
Au oma ion x High x Mode a e
Appl. Sci. 2025,15, 854 20 o 47
Table 12. Con .
Technology
In eg a ion
(RPA, Machine
Lea ning, IoT)
Me hod/
Algo i hm
In elligen
Au oma ion
Con inuous
Lea ning
Flexibili y
and
Scalabili y
Real-Time
Moni o ing
Ene gy
E iciency and
Sus ainabili y
[59] Pa ício, L.;
A ila, P.; Va ela, L.;
C uz-Cunha, M.M.;
Fe ei a,
L.P.;Bas os, J.;
Cas o, H.; Sil a, J.
(2023)
RPA Rule-Based
Au oma ion x High x Mode a e
[60] Pa ício, L.;
Cos a, C.R.S.;
Va ela, L.;
C uz-Cunha, M.M.
(2024)
RPA Rule-Based
Au oma ion x High x Mode a e
[61] Daase, C.,
Pandey, A.,
S aegemann, D.;
Tu owski, K. (2023)
RPA Rule-Based
Au oma ion x High x Mode a e
[62] Pa ício, L.;
Cos a, L.; Va ela, L.;
Á ila, P. (2023) RPA Rule-Based
Au oma ion x High x Mode a e
[This wo k] RPA, Machine
Lea ning
Rule-Based
Au oma ion,
Decision T ees x x High x High
The compa a i e analysis be ween exis ing s udies and he p oposed SIRPM model
clea ly demons a es he dis inc ad an ages ha ou model o e s. Table 12 shows ha
while many p e ious models in eg a e echnologies such as machine lea ning, he IoT, o
RPA in isola ion o wi h a ying le els o au oma ion, he SIRPM model s ands ou o i s
ad anced combina ion o RPA and machine lea ning, p o iding in elligen au oma ion
and con inuous lea ning. In addi ion, SIRPM s ands ou o i s lexibili y and scalabili y,
allowing o e icien adap a ions and expansions acco ding o he company’s needs. This
in eg a ion esul s in mo e accu a e and e ec i e eal- ime moni o ing in addi ion o an
imp o emen in ene gy e iciency and sus ainabili y, aspec s ha a e o en add essed in a
limi ed way in exis ing models. A i icial in elligence (AI) is a b oad ield encompassing
echniques o simula e human in elligence in machines. Machine lea ning (ML) is a subse
o AI ocused on sys ems capable o lea ning and imp o ing om da a wi hou he need
o explici p og amming. Deep lea ning (DL), in u n, ep esen s an ad anced app oach
wi hin ML, u ilizing deep neu al ne wo ks o sol e complex p oblems such as pa e n
ecogni ion in la ge olumes o da a. In his model, ML is p ima ily used o iden i y
pa e ns and p edic ailu es, while DL can be applied o he analysis o complex da a, such
as images o ime se ies. Al hough DL is a p omising ield, he decision was made o adop
ML due o he na u e o he p oblem and he a ailable da a. DL is pa icula ly e ec i e in
con ex s whe e la ge olumes o uns uc u ed da a, such as images o ideos, a e common.
Howe e , in he p esen s udy, he amoun o abula da a a ailable o analysis does no
jus i y he use o DL models, which equi e conside able compu a ional esou ces. ML, on
he o he hand, o e s an e icien , accessible, and sui able app oach o add ess p edic i e
main enance, p o iding sa is ac o y pe o mance wi h a mo e agile implemen a ion.
The dis inc ion be ween AI, ML, and DL, as p esen ed in Table 12, can be obse ed
based on he app oaches adop ed in s udies ela ed o a i icial in elligence as ollows:
•
AI (a i icial in elligence) encompasses a ious echniques o simula e human in elli-
gence. Se e al a icles men ioned in Table 12 u ilize AI h ough di e se app oaches,
Appl. Sci. 2025,15, 854 21 o 47
including machine lea ning algo i hms, gene ic algo i hms, and dynamic p og am-
ming.
•
ML (machine lea ning), a subse o AI, ocuses on sys ems ha lea n om da a. A icles
ha explici ly use ML echniques, such as ein o cemen lea ning algo i hms, sea ch
algo i hms, and Bayesian ne wo ks, aim o iden i y pa e ns and make p edic ions
based on da a.
•
DL (deep lea ning) ep esen s an ad anced app oach wi hin ML, u ilizing deep neu al
ne wo ks. The use o deep neu al ne wo ks is mo e ele an in con ex s in ol ing
la ge olumes o uns uc u ed da a, such as images o ideos, which a e no di ec ly
add essed in he s udies in he able.
Thus, he di e en ia ion is clea : AI encompasses a ious app oaches, ML is a subse o
AI ha ocuses on lea ning om da a and is he mos widely used in he s udies p esen ed,
while DL is a mo e ad anced and speci ic o m o ML, ypically applied o uns uc u ed
and mo e complex da a. Shown below, i is possible o analyze each o he a icles in e ms
o hei desc ip ion ega ding hei unc ion and echnological applica ion.
•
[
26
] Deep lea ning in p edic i e main enance, which is used o p edic ailu es in
indus ial equipmen , helping o op imize p ocesses and educe main enance cos s.
•
[
27
] Deep lea ning in p edic i e main enance employs deep lea ning models o
analyze da a and p edic ailu es in indus ial sys ems, inc easing e iciency and
sus ainabili y.
•
[
28
] RPA in business p ocess op imiza ion in ol es he au oma ion o business p ocesses
o inc ease e iciency, educe e o s, and con ibu e o o ganiza ional sus ainabili y.
•
[
29
] RPA in digi al ans o ma ion, which is used o imp o e business p ocess e i-
ciency, educing manual e o s and accele a ing digi al ans o ma ion.
•
[
30
] Machine lea ning in p edic i e main enance uses machine lea ning algo i hms
o p edic ailu es in indus ial sys ems, helping o educe cos s and inc ease ope a-
ional e iciency.
•
[
31
] Machine lea ning in p edic i e main enance employs machine lea ning models o
p edic i e main enance s a egies, inc easing p oduc i i y and minimizing down ime.
•
[
32
] Technologies o sma manu ac u ing and p edic i e main enance in eg a es AI
and machine lea ning echnologies o p edic ailu es and imp o e sus ainabili y in
manu ac u ing.
•
[
33
] RPA in Indus y 4.0: RPA implemen a ion o he au oma ion o indus ial
p ocesses, p omo ing e iciency and sus ainabili y in he sec o .
•
[
34
] Machine lea ning in p edic i e main enance uses machine lea ning echniques o
p edic ailu es in indus ial sys ems, educing cos s and inc easing ope a ional e i-
ciency.
•
[
35
] Machine lea ning in p edic i e main enance applies machine lea ning algo i hms
o op imize manu ac u ing ope a ions and educe unexpec ed ailu es.
•
[
36
] RPA and AI in Indus y 4.0 combines RPA and AI o he in elligen au oma ion
o indus ial p ocesses, op imizing ope a ions and p omo ing sus ainabili y.
•
[
37
] Deep lea ning o he p edic i e main enance o indus ial obo s uses deep
neu al ne wo ks o p edic ailu es and op imize he main enance o indus ial obo s.
•
[
38
] Da a-d i en algo i hms o p edic i e main enance applies machine lea ning
algo i hms o op imize p edic i e main enance s a egies and educe cos s.
•
[
39
] Sus ainable echnologies in Indus y 4.0 in eg a es echnologies o imp o e
main enance and educe en i onmen al impac s in indus ial p ocesses.
•
[
40
] Machine lea ning in p edic i e main enance uses machine lea ning o imp o e
e iciency and p edic ailu es in indus ial sys ems, p omo ing sus ainabili y.
Appl. Sci. 2025,15, 854 22 o 47
•
[
41
] P edic i e main enance and sus ainabili y: he applica ion o p edic i e main enance
o educe en i onmen al impac s and imp o e sus ainabili y in indus ial sys ems.
•[42] AI and he IoT o p edic i e main enance in eg a es AI and he IoT o op imize
p edic i e main enance, con ibu ing o sus ainabili y and ope a ional e iciency.
•
[
43
] Machine lea ning in p edic i e main enance uses machine lea ning o p edic ailu es
and op imize main enance, ensu ing g ea e e iciency in he indus ial en i onmen .
•
[
44
] Indus ial AI and p edic i e main enance applies AI o p edic ailu es and
op imize p ocesses, p omo ing sus ainabili y in manu ac u ing.
•
[
45
] P edic i e main enance and AI uses AI o op imize p edic i e main enance,
imp o ing p oduc ion and p omo ing mo e sus ainable indus ial p ac ices.
•
[
46
] RPA and sus ainabili y in Indus y 4.0: RPA as a ool o imp o e sus ainabili y
and e iciency in indus ial p ocesses.
•
[
47
] AI and p edic i e main enance: he applica ion o AI o p edic i e main enance
in indus ial en i onmen s, imp o ing he e iciency and sus ainabili y o p ocesses.
•
[
48
] RPA in sus ainabili y in supply chains: RPA o p ocess au oma ion in supply
chains, aiming a sus ainabili y and ope a ional e iciency.
•
[
49
] The IoT and machine lea ning o p edic i e main enance in eg a es he IoT and
machine lea ning o op imize he main enance o indus ial sys ems and educe ailu es.
•
[
50
] Sma p edic i e main enance wi h AI and machine lea ning applies AI and
machine lea ning o imp o e p edic i e main enance in indus ial en i onmen s.
•
[
51
] P edic i e main enance and en i onmen al impac uses p edic i e main enance
o educe he en i onmen al impac o indus ial p ocesses and p omo e sus ainable
p ac ices.
•
[
52
] In eg a ion o AI and he IoT o p edic i e main enance combines AI and he
IoT o op imize main enance and ensu e sus ainabili y in indus ial sys ems.
•
[
53
] Sus ainable RPA: he implemen a ion o sus ainable RPA in indus ial p ocesses,
p omo ing e iciency and educing en i onmen al impac .
•
[
54
] Machine lea ning o p edic i e main enance in Indus y 4.0 uses machine lea n-
ing o op imize main enance and imp o e he e iciency o indus ial sys ems.
•
[
55
] Deep lea ning o p edic i e main enance applies deep lea ning o op imize
main enance and p edic ailu es in indus ial sys ems.
•
[
56
] The in eg a ion o AI and RPA p oposes a sus ainable model o in eg a ing AI
and RPA in indus ial p ocesses, op imizing ope a ions and p omo ing sus ainabili y.
•
[
57
] RPA and ene gy e iciency: RPA used o op imize indus ial p ocesses, imp o ing
ene gy e iciency and educing consump ion.
•
[
58
] F amewo k o RPA implemen a ion: de elopmen o a amewo k o RPA p ojec
implemen a ion and con ol, ocusing on sus ainabili y and ope a ional e iciency.
•
[
59
] Decision models o sus ainable RPA implemen a ion: e iew o decision models
o he sus ainable implemen a ion o RPA, aiming o op imiza ion and educ ion in
en i onmen al impac s.
•
[
60
] Sus ainable RPA implemen a ion in heal hca e adminis a i e se ices: analysis
o RPA implemen a ion in heal hca e, ocusing on sus ainabili y and e iciency.
•
[
61
] Uni e sal model o sus ainable RPA implemen a ion: p oposal o a uni e sal
model o sus ainable RPA implemen a ion, p omo ing e iciency and sus ainabili y.
•
[
62
] Sus ainable RPA implemen a ion: de elopmen o a ma hema ical model o he
sus ainable implemen a ion o RPA ac oss di e en indus ial con ex s.
The IoT plays a c ucial ole in eal- ime da a collec ion and ansmission. Many o
he solu ions highligh ed in Table 12 adop wi eless echnologies due o hei e iciency
and mobili y. Howe e , in speci ic cases, wi ed ne wo ks may be employed depending on
secu i y and bandwid h equi emen s. Table 12 illus a es a combina ion o echnologies
Appl. Sci. 2025,15, 854 23 o 47
such as RPA, machine lea ning, he IoT, and AI, wi h a ocus on in elligen au oma ion and
ene gy e iciency, among o he aspec s. The in e ac ion be ween he IoT and echnologies
such as RPA and machine lea ning is no ewo hy, especially in he de elopmen o sma
solu ions o au oma ion and moni o ing. The IoT, by in eg a ing connec ed de ices,
acili a es da a collec ion and communica ion be ween sys ems, allowing o he applica ion
o machine lea ning algo i hms o p edic i e analysis and p ocess op imiza ion. Rega ding
IoT solu ions, he able e eals ha i is o en associa ed wi h machine lea ning echnologies,
especially in con ex s in ol ing echniques such as andom o es , decision ees, and
suppo ec o machines (SVMs), applied o p ocess da a collec ed om IoT de ices. These
sys ems may include bo h wi ed and wi eless solu ions, depending on he con ex ’s needs.
Wi eless solu ions (g ea e mobili y and e iciency): These solu ions a e p edominan ly
used in IoT echnologies, being mo e common in en i onmen s ha equi e g ea e lexibil-
i y and mobili y. Examples include wi eless senso ne wo ks and IoT de ices connec ed
ia Wi-Fi o 5G, o e ing eedom o mo emen and ease o implemen a ion. Wi ed so-
lu ions: These a e applied in scena ios equi ing mo e s able communica ion and highe
bandwid h, such as indus ial ins alla ions o local a ea ne wo ks (LANs). Wi ed IoT
de ices a e ypically used when mo e eliable and secu e connec ions a e needed. In he
case o IoT solu ions associa ed wi h echnologies such as RPA and machine lea ning, he
end is owa ds wi eless sys ems. This choice is cha ac e is ic o scena ios ha equi e
g ea e mobili y and e iciency, as wi eless echnologies o e g ea e lexibili y and ease o
implemen a ion, especially in dynamic and scalable en i onmen s. Technologies such as
Wi-Fi, Blue oo h, Zigbee, LoRa, and 5G a e equen ly used in IoT de ices o ensu e e icien
communica ion wi hou he need o complex physical in as uc u e. These echnologies
enable eal- ime da a collec ion and emo e moni o ing, which is c ucial o applica ions in
in elligen au oma ion and sus ainabili y, as e idenced in he able. While wi ed solu ions
(such as E he ne o indus ial cables) p o ide g ea e s abili y and bandwid h, hey a e
less common in IoT solu ions gea ed owa ds mobili y and e iciency. Wi ed solu ions a e
mo e equen ly adop ed in speci ic en i onmen s equi ing g ea e eliabili y and secu i y,
such as complex indus ial ins alla ions. The IoT, in mos cases men ioned in he able, is
associa ed wi h wi eless solu ions, o e ing g ea e mobili y, lexibili y, and e iciency.
The SIRPM model posi ions i sel as a obus and inno a i e solu ion in he ield
o p edic i e main enance. I o e s a se o ea u es and bene i s ha p o ide g ea e
ope a ional e iciency, cos educ ion, and be e da a-d i en decision-making. The in e-
g a ion o ad anced echnologies such as RPA, machine lea ning, and he IoT esul s in a
sys em capable o con inuously adap ing o changes, ensu ing sus ainabili y and educing
en i onmen al impac s, in addi ion o p omo ing he aining o ope a ional eams.
In he con ex o he p oposed model, scalabili y is an essen ial cha ac e is ic o assess
he sys em’s abili y o handle inc eases in demand wi hou comp omising i s e iciency.
Limi ed scalabili y e e s o a sys em ha can only suppo small expansions, equi ing sig-
ni ican in as uc u e changes o handle a signi ican inc ease in load. Mode a e scalabili y,
on he o he hand, allows o some deg ee o adjus men bu s ill equi es modi ica ions
o accommoda e subs an ial inc eases in demand. High scalabili y indica es ha he sys-
em is capable o suppo ing la ge expansions wi hou he need o s uc u al changes,
main aining e iciency and pe o mance wi hou in e up ions, ega dless o he wo kload.
Wi h ega d o ene gy e iciency, his is de ined as he sys em’s abili y o pe o m
i s unc ions wi h minimum ene gy consump ion, op imizing he use o esou ces and
hus educing ope a ing cos s. Ene gy e iciency no only seeks o minimize he need o
ene gy bu also o ensu e ha he sys em ope a es sus ainably. When in eg a ed wi h
sus ainabili y, ene gy e iciency gains an addi ional dimension, p omo ing a educ ion
in he en i onmen al impac o ope a ions. This in ol es using echnologies ha no
Appl. Sci. 2025,15, 854 24 o 47
only educe ene gy consump ion bu also a o enewable sou ces and p ac ices ha
minimize he was e o esou ces. Thus, ene gy e iciency and sus ainabili y a e di ec ly
associa ed wi h he implemen a ion o solu ions ha ensu e bo h sys em pe o mance and
en i onmen al esponsibili y, aligning wi h esponsible business p ac ices and he adop ion
o g een echnologies in he au oma ion p ocess.
The main ad an ages o he SIRPM model in ela ion o exis ing models a e as ollows:
1.
Ad anced echnology in eg a ion (RPA, machine lea ning): i o e s a obus and
comp ehensi e combina ion o echnologies, p o iding g ea e e ec i eness and a
con inuous low o da a be ween senso s, analysis pla o ms, and ope a ional eams.
2.
In elligen au oma ion: i minimizes human in e en ion, educing e o s and making
he p ocess mo e e icien and secu e.
3.
Con inuous lea ning: he machine lea ning sys em con inuously adap s based on
ope a ional eedback, enhancing he accu acy o p edic ions and decision-making
o e ime.
4.
Flexibili y and scalabili y: he model is modula and can be implemen ed g adually
acco ding o he speci ic needs o he company, allowing o expansion in o o he
sec o s o a eas o he o ganiza ion.
5.
Real- ime moni o ing: immedia e anomaly de ec ion and eal- ime ale gene a ion
acili a e a quick esponse o ailu es, educing unplanned down ime.
6.
Ene gy e iciency and sus ainabili y: he model p omo es esou ce op imiza ion and
was e educ ion, aligning wi h en i onmen al goals and co po a e sus ainabili y
egula ions.
These ea u es no only make he SIRPM mo e e ec i e bu also ensu e i is a long-
e m solu ion o companies seeking inno a ion, cos educ ion, and sus ainabili y in he
managemen o hei indus ial asse s.
5. Case S udy
5.1. Case S udy P esen a ion
Alpha is a small amily-owned business loca ed in an indus ial a ea o a Po uguese
ci y. I s p ima y ac i i y in ol es he p oduc ion o ca dboa d packaging o he ood
sec o , wi h a pa icula ocus on manu ac u ing boxes o he anspo a ion and s o age
o esh ood. The company ope a es a p oduc ion line consis ing o se e al au oma ed
machines, ye i has aced ecu ing issues wi h machine b eakdowns, esul ing in equen
p oduc ion s oppages.
Alpha employs a o al o 11 s a membe s, dis ibu ed as ollows:
•
Managemen (one pe son): The owne o he company is esponsible o o e all
managemen and supe ision. While ac i ely in ol ed in he ac o y ope a ions, he
makes s a egic decisions and handles adminis a i e ma e s.
Adminis a i e Depa men ( h ee people):
•One inancial manage o e sees he company’s inances and cash low, ensu ing ha
p oduc ion emains wi hin a con olled budge .
•
One p oduc ion planning enginee is esponsible o planning p oduc ion sched-
ules, ensu ing o de s a e me on ime and ha machines ope a e wi hin he desig-
na ed hou s.
•
One ma ke ing manage is ocused on b and p omo ion and cus ome ela ions,
ensu ing clien e en ion and he acquisi ion o new business.
P oduc ion Depa men (six people):
•
One p oduc ion manage (p oduc ion supe iso ) o e sees he p oduc ion line, ensu -
ing smoo h ope a ions du ing he wo k shi .
Appl. Sci. 2025,15, 854 25 o 47
•
Two machine ope a o s ope a e he machines ha p oduce he packaging, ensu ing
he p oduc ion p ocess ollows echnical speci ica ions and ha ma e ials a e a ailable.
•
One main enance echnician ( esponsible o basic main enance) is esponsible o
pe o ming mino epai s and adjus men s o he machines when simple ailu es occu ,
such as pa eplacemen s o calib a ion adjus men s. Howe e , hey did no pe o m
sys ema ic p edic i e o p e en i e main enance.
•
Two p oduc ion suppo wo ke s (packe s and auxilia y ope a o s) assis wi h packag-
ing inished p oduc s and keeping he p oduc ion p ocess o ganized, mo ing boxes
and ma e ials.
Logis ics Depa men (one pe son):
•
One d i e /logis ics manage is esponsible o he dis ibu ion o packaging o
cus ome s and supplie s, as well as managing he low o ma e ials wi hin he ac o y.
Al hough Alpha had a unc ional s uc u e and smoo h ope a ions, i aced signi ican
challenges ela ed o machine main enance. B eakdowns we e equen , pa icula ly wi h
some o he olde machines, which expe ienced unexpec ed down imes. The co e issue
was he lack o p edic i e main enance; only manda o y main enance was ca ied ou ,
scheduled by he ex e nal main enance company wi h which he business had a con ac .
The p oduc ion manage , oge he wi h he main enance echnician, handled he
simple aul s, bu machine s oppages s ill occu ed unexpec edly. When mo e se e e
b eakdowns ook place, ac o y p oduc ion would be in e up ed, esul ing in o de delays,
loss o e iciency, and inc eased cos s due o he need o u gen epai s. Wi hou a eal-
ime moni o ing sys em, he ac o y was ulne able o un o eseen ailu es, which led o
p olonged pe iods o inac i i y, a ec ing he company’s abili y o mee deli e y deadlines.
This si ua ion became unsus ainable, p omp ing he owne , along wi h he p oduc ion
manage and he p oduc ion planning enginee , o conside implemen ing a mo e ad anced
sys em o add ess he b eakdowns. The inabili y o p edic o e en an icipa e ailu es be o e
hey caused p oduc ion s oppages was nega i ely impac ing bo h in e nal ope a ions and
ela ionships wi h cus ome s. As a esul , he need a ose o seek mo e e ec i e solu ions o
op imize he ac o y’s ope a ion and educe he cos s associa ed wi h unexpec ed ailu es.
A his poin , he company began explo ing echnological solu ions ha could imp o e
p oduc ion e iciency and machine eliabili y, leading hem o in es iga e op ions such as
eal- ime moni o ing sys ems, p edic i e main enance, and he use o mode n ools o assis
in ea ly aul de ec ion and main enance op imiza ion.
This p ocess o adap a ion and echnological mode niza ion ma ked he i s s ep
o Alpha in o e coming i s main enance challenges and ensu ing a mo e e icien and
sus ainable u u e o i s ope a ions.
The implemen a ion o SIRPM in he case s udy was based on RPA p ocesses op imized
o he collec ion, p ocessing, and analysis o ope a ional da a om indus ial equipmen .
These au oma ed p ocesses we e implemen ed in i ual machines (VMs) con igu ed in a
hyb id cloud en i onmen . RPA so wa e (Uipa h 2022.4.1) licenses we e ins alled di ec ly
on he VMs, whe e he obo s execu ed he au oma ions necessa y o in e ac wi h pla o ms
o access he da abases.
IoT senso s we e s a egically posi ioned on he machines o moni o c i ical a iables
such as empe a u e, ib a ion, and p essu e.
The da a collec ed by he senso s we e ansmi ed ia LTE and Wi-Fi-based elecom-
munica ions ne wo ks o IoT ga eways. These ga eways played a c ucial ole in e icien ly
agg ega ing and ansmi ing his in o ma ion o he cen al ope a ions sys em. This connec-
i i y a chi ec u e was designed o ensu e low la ency and high eliabili y in eal- ime da a
low, essen ial cha ac e is ics o p edic i e main enance and indus ial au oma ion sys ems.
Appl. Sci. 2025,15, 854 32 o 47
Au oma ed Wo k low:
1. Da a collec ion: senso s p o ided da a s o ed in he SQL da abase.
2. Da a p ocessing: he ML model analyzed pa e ns and made p edic ions.
3.
Ale s: anomalous alues igge ed au oma ic email no i ica ions o he main enance
eam.
Technologies Used:
#UiPa h: o da a collec ion and ale au oma ion.
#
Py hon (sciki -lea n): o de eloping he machine lea ning model, le e aging lib a ies
such as Pandas, NumPy, and ma plo lib o da a analysis and isualiza ion.
The iden i ica ion o ale alues o each ype o senso in Alpha machines was based
on he ypical e e ence alues men ioned by he company, wi h he aim o moni o ing ma-
chine pe o mance and an icipa ing imminen ailu es. The e e ence alues o each senso
we e de ined o ensu e e icien ope a ion o he equipmen and minimize ailu es du ing
he manu ac u ing p ocess. Below, we explain how ale s we e iden i ied o ib a ion,
p essu e, and empe a u e senso s, based on he s anda d alues and es ablished limi s.
•
Vib a ion senso (pape cu e ): Excessi e ib a ion is one o he i s signs o mechani-
cal wea o misalignmen o machine pa s. The ideal ib a ion ange o he pape
cu e was de ined be ween 1.0 and 1.3 mm/s, conside ing no mal machine ope a ion.
When ib a ion exceeded 1.5 mm/s, his indica ed a possible imminen mechanical
ailu e, such as a misalignmen o pa s o excessi e wea , which could comp omise
machine ope a ion. On he o he hand, i he ib a ion ell below
1.0 mm/s
, his
could indica e ha he machine’s mo ing componen s we e ha ing li le in e ac ion o
con ac , which could also cause ope a ional p oblems. These alues we e used as a
basis o igge ing au oma ic ale s so ha he main enance eam could in es iga e
po en ial ailu es be o e hey occu ed.
•
P essu e senso (hyd aulic p ess): The p essu e o he hyd aulic luid in he hyd aulic
p ess is essen ial o ensu e ha he p essing p ocess occu s co ec ly and wi h quali y.
The ideal p essu e ange was es ablished be ween 145 and 150 ba , wi h alues ou side
his ange indica ing po en ial p oblems. I he p essu e ose abo e 155 ba , his
could signal ailu es in he p essu e con ol sys em o e en blockages in he hyd aulic
line, which would a ec he quali y o he packaging o ma ion. On he o he hand,
p essu es below 140 ba indica ed ha he hyd aulic p essu e was below wha was
necessa y, which would impai he e iciency o he p ocess and could cause ailu es in
he p essing p ocess. These ale alues helped o an icipa e he necessa y main enance,
a oiding ailu es du ing ope a ion.
•
Tempe a u e senso ( lexog aphic p in e ): The empe a u e o he p in heads is c ucial
o ensu ing p in quali y and he e iciency o he lexog aphic p in e ’s ope a ion.
The ideal empe a u e ange was de ined as be ween 75 and 80
◦
C, wi h uppe and
lowe limi s o ale o ailu es in he cooling o hea ing sys em. I he empe a u e
ose abo e 85
◦
C, his could indica e ailu es in he cooling sys em, which would
esul in o e hea ing and possible damage o he p in heads. Tempe a u es below
70 ◦C
, on he o he hand, indica ed ailu es in he hea ing sys em, comp omising p in
quali y. These ale s we e essen ial o a oid damage o he equipmen and ensu e he
con inui y o he p oduc ion p ocess.
These e e ence alues and ale s we e undamen al o he p edic i e main enance
sys em implemen ed by Alpha. Using he da a collec ed by he senso s, he sys em was
able o quickly iden i y when alues we e ou side he ideal ange, gene a ing au oma ic
ale s o he main enance eam. This allowed p e en a i e main enance o be ca ied
Appl. Sci. 2025,15, 854 33 o 47
ou be o e se ious ailu es occu ed, imp o ing ope a ional e iciency and educing ma-
chine down ime.
•Phase 5: P oac i e Ac ion and Feedback
When he model p edic ed an imminen ailu e, RPA au oma ically ale ed he espon-
sible eams, p omp ing co ec i e ac ions such as adjus men s o epai s.
Implemen ed Ac ions:
#Au oma ed email ale s o he main enance eam.
#Execu ion o co ec i e main enance and p e en i e adjus men s.
#Feedback collec ion on ale e ec i eness and p edic ion accu acy.
The p oac i e measu es’ ou comes we e moni o ed o e ine he model using in e en-
ion eedback, he eby imp o ing machine lea ning p edic ions.
•Phase 6: Sus ainabili y and Con inuous Imp o emen
To ensu e he model’s sus ainabili y and con inued e iciency, key pe o mance indi-
ca o s such as he MTTR (mean ime o epai ) and MTBF (mean ime be ween ailu es)
we e calcula ed, alongside sus ainabili y me ics ela ed o ene gy consump ion and emis-
sions educ ion.
Con inuous Moni o ing:
#The machine lea ning model’s pe o mance was pe iodically e iewed.
#The RPA sys em was adjus ed as needed o accu acy and e icien in eg a ion.
#
Con inuous s a aining ensu ed adap a ion o he new p edic i e main enance
me hodology.
The machine lea ning model implemen ed in his s udy is a pos -p ocessing app oach.
The decision ee (DT) model was execu ed pe iodically a hou ly in e als o p ocess
he da a collec ed by he RPA sys em. This p ocess was essen ial o gene a ing ailu e
p edic ions in he moni o ed sys ems based on he selec ed c i ical a iables, such as
T ( empe a u e), V ( eloci y), and P (p essu e). These a iables we e chosen o hei
signi icance o machine y pe o mance, enabling p ecise analysis and he an icipa ion o
po en ial ailu es.
The eg esso employed in he decision ee model was a eg ession algo i hm based
on decision ees. The model’s ou pu s, namely he ailu e p edic ions, di ec ly in luence
he RPA p ocesses o he machine y. When a ailu e is p edic ed, he RPA sys em igge s
ale s o no i y he main enance eam o po en ial in e en ion needs.
The in eg a ion p ocess be ween he machine lea ning model and he RPA sys em
ollows a con inuous moni o ing cycle, du ing which he pe o mance o he model is
pe iodically e iewed. Wi h each adjus men o enhancemen made o he ML model, he
RPA sys em is upda ed o ensu e he accu acy and e iciency o he in eg a ion.
The MTBF (mean ime be ween ailu es) is a me ic ha measu es he a e age ime
be ween ailu es o a piece o equipmen o sys em du ing i s ope a ion. The highe he
MTBF, he be e he sys em’s eliabili y.
The o mula o calcula ing he MTBF is
MTBF = (To al Ope a ing Time)/(Numbe o Failu es);
whe e
•
To al Ope a ing Time is he ime du ing which he sys em o equipmen is unning
wi hou ailu es.
•
Numbe o Failu es is he numbe o ailu es ha occu du ing he conside ed ime
pe iod.
Appl. Sci. 2025,15, 854 34 o 47
The MTTR (mean ime o epai ) is a me ic ha measu es he a e age ime equi ed
o epai a piece o equipmen a e a ailu e. The lowe he MTTR, he as e he equipmen
is epai ed and e u ned o ope a ion.
The o mula o calcula ing he MTTR is
MTTR = (To al Down ime)/(Numbe o Failu es);
whe e
•
To al Down ime is he o al ime du ing which he sys em o equipmen was ou o
ope a ion due o ailu es.
•
Numbe o Failu es is he numbe o ailu es ha occu du ing he conside ed ime
pe iod.
In he p esen s udy, calcula ions based on speci ic o mulas we e used o e alua e
pe cen age educ ions in di e en cos ca ego ies ela ed o main enance and ope a ions.
These calcula ions we e pe o med based on he alues p o ided by he company, wi h he
ollowing o mulas applied:
•The o mula o calcula ing REDUCTION_A e age Repai Cos s is
REDUCTION_A e age Repai Cos s % = [1 −[(AFTER_ A e age
Repai Cos s)/(BEFORE_ A e age Repai Cos s)]] ×100;
•The o mula o calcula ing REDUCTION_ To al Main enance Cos s is
REDUCTION_ To al Main enance Cos s % = [1 −[(AFTER_ To al
Main enance Cos s)/(BEFORE_ To al Main enance Cos s)]] ×100;
•The o mula o calcula ing REDUCTION_ Unplanned Down ime Cos s is
REDUCTION_ Unplanned Down ime Cos s % = [1 −[(AFTER_Unplanned
Down ime Cos s)/(BEFORE_ Unplanned Down ime Cos s)]] ×100;
•The o mula o calcula ing REDUCTION_ To al Ope a ional Cos s is
REDUCTION_ To al Ope a ional Cos s % = [1 −[(AFTER_ To al
Ope a ional Cos s)/(BEFORE_ To al Ope a ional Cos s)]] ×100;
The i s o mula measu es he pe cen age educ ion in o al ope a ional cos s o e he
analyzed pe iod, e lec ing he o e all impac o he implemen ed s a egies. The second
o mula calcula es he educ ion in cos s associa ed wi h unplanned down ime, highligh -
ing imp o emen s in ope a ional e iciency. The hi d o mula e alua es he pe cen age
dec ease in o al main enance cos s, o e ing a clea iew o he cos -sa ing measu es im-
plemen ed. Las ly, he o mula o a e age epai cos s measu es he pe cen age educ ion
by compa ing he alues be o e and a e he imp o emen s, showcasing he e ec i eness
o he in e en ions.
Rega ding sus ainabili y me ics, he company iden i ied and moni o ed i e main
c i e ia, namely ene gy consump ion, CO
2
emissions, was e p oduced, use o ma e ials,
and ecycling a e. Fo each o hese c i e ia, his o ical alues we e p o ided be o e
and a e he model was implemen ed. The pe cen age educ ion was calcula ed based
on he di e ence be ween he alues be o e and a e implemen a ion, whe e he alue
ob ained a e implemen a ion was di ided by he p e ious alue, sub ac ed om one
and mul iplied by 100. These calcula ions allow us o accu a ely measu e he imp o emen
in e ms o sus ainabili y. Ene gy consump ion and CO
2
emissions a e c ucial o assessing
a company’s en i onmen al impac , and educing hese indica o s is di ec ly ela ed o
educing he ca bon oo p in . The was e p oduced and he use o ma e ials e lec he
e iciency in he use o na u al esou ces, while he ecycling a e indica es he company’s
Appl. Sci. 2025,15, 854 35 o 47
commi men o sus ainabili y in was e managemen . These alues we e acked by he
company, allowing o a de ailed analysis o he en i onmen al gains esul ing om he
implemen a ion o he model.
REDUCTION % = [1 −[(AFTER Implemen a ion)/(BEFORE Implemen a ion)]] ×100;
6. Analysis o Resul s and Discussion
6.1. Analysis o Resul s
In his chap e , he esul s ob ained a e implemen ing he model will be p esen ed
and analyzed. To his end, we will p esen da a om one mon h o implemen a ion, which
was du ing mon h 8 o ope a ion. Du ing his pe iod, he ac o y was in ope a ion o
22 business days, and on each o hese days, 16 daily p edic ions we e gene a ed o
each senso ins alled on he moni o ed machines. These p edic ions we e made based on
da a om he las 24 h o ope a ion using a machine lea ning model, speci ically decision
ees, which was execu ed e e y hou . The model conside ed c i ical a iables such as he
ib a ion, p essu e, and empe a u e o each machine.
Fo each senso , he execu ion o he model gene a ed a single p edic ion o he nex
hou o ope a ion, based on he mos ecen his o ical da a. In p ac ical e ms, he p edic ion
gene a ed o he “nex hou ” was based on in o ma ion om he las 24 h and he model’s
lea ning his o y.
O e he 22 business days o ope a ion, he model gene a ed a o al o 352 p edic ions,
dis ibu ed among he h ee ypes o senso s, namely ib a ion, p essu e, and empe a u e
senso s. To acili a e analysis, he esul s a e p esen ed h ough h ee dis inc g aphs, each
co esponding o a ype o senso as ollows: ib a ion (Figu e 4), p essu e (Figu e 5),
and empe a u e (Figu e 6), allowing o a clea isualiza ion o he luc ua ions in he
p edic ions and he compa ison o he p edic ed ailu es h oughou he mon h.
Figu e 4. P edic ion— ib a ion (mm/s).
When he p edic ed alues o he ib a ion, p essu e, and empe a u e senso s ex-
ceeded he es ablished uppe o lowe limi s, an au oma ic ale was gene a ed o he
main enance eam. These limi s we e de ined as ollows: o ib a ion, he maximum and
minimum alues we e 1.5 mm/s and 1 mm/s, espec i ely; o p essu e, he limi s we e
155 ba (maximum) and 140 ba (minimum); and o empe a u e, he es ablished limi s
we e 85 ◦C (maximum) and 70 ◦C (minimum).
Appl. Sci. 2025,15, 854 36 o 47
Figu e 5. P edic ion—p essu e (ba ).
Figu e 6. P edic ion— empe a u e (◦C).
Whene e he p edic ed alues o any o hese pa ame e s exceeded he limi s, an ale
was igge ed, allowing he main enance eam o be immedia ely no i ied. This app oach
was c ucial in p e en ing unexpec ed ailu es, as i allowed he eam o moni o he machine
p oac i ely, iden i ying po en ial p oblems and ca ying ou he necessa y main enance
be o e hey could esul in un o eseen down ime o se ious damage o he machines.
The implemen a ion o he SIRPM model a Alpha esul ed in signi ican imp o e-
men s in bo h ope a ional e iciency and sus ainabili y. The analysis p esen s he key
pe o mance indica o s (KPIs) be o e and a e he adop ion o he model du ing he
6-mon h ope a ing pe iod be ween mon hs 7 and 12. This analysis highligh s he e ec-
i eness o he p edic i e main enance sys em and i s impac on cos educ ion, esou ce
op imiza ion, and en i onmen al sus ainabili y.
The e ec s o he SIRPM model on main enance pe o mance a e e iden when com-
pa ing he key indica o s be o e and a e i s implemen a ion. These indica o s clea ly
demons a e he ope a ional bene i s achie ed wi h he p edic i e main enance sys em,
which combines RPA and machine lea ning.
Fo he analysis, he alues we e compa ed wi h he his o ical main enance da a om
he p e ious yea , which we e p o ided by he company.
Appl. Sci. 2025,15, 854 37 o 47
The impac o he SIRPM model on main enance pe o mance can be clea ly seen
in he compa ison o key me ics be o e and a e i s adop ion. These me ics show he
ope a ional bene i s achie ed h ough he p edic i e main enance sys em, which in eg a es
RPA and machine lea ning (Table 14).
Table 14. Main enance pe o mance imp o emen s.
Me ic Be o e
Implemen a ion
A e
Implemen a ion Imp o emen
Mean Time
Be ween Failu es (MTBF) 23.47 h/e en s 70.4 h/e en s +100%
Mean Time o
Repai (MTTR) 12 h 4 h −67%
Unplanned
Down ime E en s 15 e en s/mon h 5 e en s/mon h −66.67%
Failu e P edic ion
Accu acy +60% +90% +50%
The o mula o calcula ing MTBF is
MTBF = (To al Ope a ing Time)/(Numbe o Failu es);
Be o e he implemen a ion o imp o emen s, he company ope a ed wi h an a e age
o 15 ailu es pe mon h. Conside ing ha he mon hly ope a ion is 352 h (22 wo king days
×16 wo king hou s pe day), he MTBF calcula ion be o e implemen a ion was
MTBF (be o e) = (352 h)/(15 e en s/mon h) = 23.47 h/e en s
This means ha on a e age, ailu es occu ed e e y 23.47 h o ope a ion.
A e implemen ing he imp o emen s, he numbe o ailu es was educed o i e
ailu es pe mon h. The MTBF calcula ion a e implemen a ion was
MTBF (a e ) = (352 h)/(5 e en s/mon h) = 70.4 h/e en s
In o he wo ds, wi h he educ ion in he numbe o ailu es, he sys em began o
ope a e o an a e age o 70.4 h be ween ailu es, ep esen ing a +100% imp o emen in
he MTBF.
The MTTR is
MTTR = (To al Down ime)/(Numbe o Failu es);
Be o e he imp o emen s we e implemen ed, he a e age ime o epai a ailu e was
12 h. Wi h 15 e en s pe mon h, he o al mon hly down ime was
To al Idle Time (be o e) = 12 h/e en s ×15 h/e en s = 180 h/mon h
A e implemen a ion, he a e age epai ime was educed o 4 h pe ailu e, wi h
i e ailu es pe mon h. The o al mon hly down ime was
To al Idle Time (a e ) = 4 h/e en s ×5 h/e en s = 20 h/mon h
This esul s in a signi ican imp o emen in epai ime, wi h a
−
67% educ ion in
he MTTR.
Appl. Sci. 2025,15, 854 38 o 47
Failu e p edic ion accu acy was p o ided di ec ly as a business pe o mance me ic,
wi h no addi ional calcula ion equi ed. P io o he implemen a ion o he imp o emen s,
ailu e p edic ion accu acy was 60%, meaning ha 60% o ailu es we e co ec ly an ic-
ipa ed. A e he implemen a ion o he imp o emen s, his accu acy inc eased o 90%,
ep esen ing a signi ican inc ease in he abili y o p edic ailu es be o e hey occu . The
+50% imp o emen is a alue p o ided by he business, e lec ing imp o emen s in ail-
u e moni o ing and p edic ion p ac ices and echnologies, allowing o a mo e e ec i e
an icipa ion o ailu e e en s and, consequen ly, mo e eliable ope a ion.
The p edic i e capabili ies o he SIRPM model no only imp o ed main enance pe o -
mance bu also esul ed in conside able cos sa ings (Table 15). By p e en ing ca as ophic
ailu es and educing he need o eme gency epai s, he model led o subs an ial educ-
ions in epai and ope a ional cos s.
Table 15. Cos educ ion.
Cos Ca ego y Be o e Implemen a ion A e Implemen a ion Reduc ion
A e age Repai Cos s EUR 2500 pe inciden EUR 1200 pe inciden −52%
To al Main enance Cos s EUR 4000/mon h EUR 2500/mon h −37.5%
Unplanned
Down ime Cos s EUR 3500/mon h EUR 1000/mon h −71.4%
To al Ope a ional Cos s EUR 30,000/mon h EUR 25,000/mon h −16.67%
As illus a ed, he a e age epai cos s dec eased by 52%, om EUR 2500 o EUR
1200 pe inciden . This educ ion e lec s he abili y o he p edic i e main enance sys em
o iden i y po en ial issues ea ly, allowing o less cos ly epai s. The o al main enance
cos s also saw a 37.5% educ ion, om EUR 4000 o EUR 2500 pe mon h. Fu he mo e,
unplanned down ime cos s we e signi ican ly lowe ed by 71.4%, om EUR 3500 o EUR
1000 pe mon h. This esul ed in an o e all dec ease in o al ope a ional cos s o 16.67%,
om EUR 30,000 o EUR 25,000 pe mon h.
The cos educ ion igu es p esen ed in Table 15 we e p o ided by he company and
ep esen a e age cos s o he pe iod be o e and a e he implemen a ion o he p edic i e
main enance model based on a s udy conduc ed o e a pe iod o 6 mon hs. These cos s
e lec he a e age expendi u e on epai s and main enance du ing his pe iod, bo h be o e
and a e he implemen a ion o he new sys em. Wi h he abili y o p edic ailu es be o e
hey occu , in e en ions could be planned in ad ance, esul ing in a signi ican educ ion
in unexpec ed ailu es and he cos s associa ed wi h hese ailu es. This con ibu ed o a
dec ease in epai cos s, main enance cos s, and unplanned down ime cos s, as shown in he
able, e idencing he e ec i eness o he sys em in imp o ing ope a ional cos managemen .
In addi ion o ope a ional and cos - ela ed imp o emen s, he SIRPM model con-
ibu ed o signi ican sus ainabili y gains. By imp o ing ope a ional e iciency and e-
ducing was e, ene gy consump ion, and emissions, he model helped Alpha achie e i s
en i onmen al goals (Table 16).
Appl. Sci. 2025,15, 854 39 o 47
Table 16. En i onmen al and sus ainabili y imp o emen s.
En i onmen al Me ic Be o e
Implemen a ion
A e
Implemen a ion Imp o emen
Ene gy
Consump ion 10,000 kWh/mon h 7500 kWh/mon h −25%
CO2Emissions 5000 kg/mon h 3000 kg/mon h −40%
Was e P oduced 500 kg/mon h 350 kg/mon h −30%
Ma e ial Usage 8000 kg/mon h 6500 kg/mon h −18.75%
Recycling Ra e 70% 85% +21.43%
The ene gy consump ion was educed by 25%, om 10,000 kWh o 7500 kWh pe
mon h, as a esul o a mo e e icien use o equipmen and ewe unplanned down imes.
CO
2
emissions saw a educ ion o 40%, om 5000 kg o 3000 kg pe mon h, con ibu ing o
a lowe en i onmen al oo p in . Simila ly, was e p oduced dec eased by 30%, om 500 kg
o 350 kg pe mon h, while ma e ial usage was educed by 18.75%, om 8000 kg o 6500 kg
pe mon h. The ecycling a e also imp o ed, ising by 21.43%, om 70% o 85%, e lec ing
be e esou ce managemen and sus ainabili y p ac ices.
The alues p esen ed in Table 16 ega ding en i onmen al and sus ainabili y imp o e-
men s we e p o ided by he company, based on da a collec ed du ing a pe iod o 6 mon hs,
be o e and a e he implemen a ion o he p edic i e main enance model (SIRPM). These
alues e lec he a e age o he en i onmen al indica o s du ing his s udy pe iod and
we e used o compa e he esul s be o e and a e he applica ion o he model.
Ene gy and CO
2
emissions educ ion can be analyzed agains ecognized s anda ds
such as hose de ined by ISO 50001 (ene gy managemen sys em) [
67
], which p o ide
essen ial guidelines o companies o manage ene gy e icien ly and educe en i onmen-
al impac s. These s anda ds a e widely adop ed by companies wo ldwide o p omo e
con inuous imp o emen in ene gy and en i onmen al pe o mance.
•
ISO 50001 ocuses on he con inuous imp o emen o ene gy pe o mance, wi h he
main goal o educing ene gy consump ion. The s anda d sugges s a educ ion o 5% o
10% pe yea in companies ha implemen ene gy managemen sys ems [
68
]. The 25%
educ ion in ene gy consump ion obse ed a Alpha signi ican ly exceeds he s anda d’s
objec i es, demons a ing excellen pe o mance in e ms o ene gy e iciency.
The cos educ ions and e iciency imp o emen s also ansla ed in o s onge inan-
cial pe o mance o Alpha, demons a ing he long- e m economic bene i s o adop ing
p edic i e main enance echnologies (Table 17).
Table 17. Financial impac .
Financial Me ic Be o e
Implemen a ion
A e
Implemen a ion Imp o emen
P o i Ma gin 5% 12% +7%
As seen in he able, he p o i ma gin inc eased om 5% o 12%, e lec ing he o e all
inancial bene i o educed main enance and ope a ional cos s. This inc ease in p o i abili y
highligh s he e ec i eness o he SIRPM model in no only enhancing ope a ional e iciency
bu also con ibu ing o he company’s inancial g ow h.
The calcula ions and su eys pe o med o measu e imp o emen s in main enance
pe o mance, ene gy consump ion, and emissions we e based on manual eco ds om he
main enance eam be o e and a e he implemen a ion o he SIRPM model. Du ing he
pe iod in which he model was no applied, he main enance eam eco ded all ailu es,
Appl. Sci. 2025,15, 854 40 o 47
epai imes, down ime e en s, and ene gy consump ion manually, using sp eadshee s and
daily epo s o moni o equipmen pe o mance and ope a ions. Wi h he implemen a ion
o he p edic i e model, he eam began o compa e he da a om each mon h wi h he
eco ds om he p e ious yea , now wi h he adop ion o he p edic i e sys em, which
included g ea e accu acy in ailu e p edic ions, op imiza ion o epai ime, and be e
ene gy managemen p ac ices. The compa ison be ween he pe iods be o e and a e he
applica ion o he model allowed o he quan i ica ion o imp o emen s in e ms o educed
ope a ing cos s, main enance e iciency, and sus ainabili y, such as ene gy consump ion
and CO2emissions, alida ing he posi i e impac s o he adop ed solu ion.
6.2. Discussion
The implemen a ion o he SIRPM (sma in eg a ed p edic i e main enance) model a
Alpha ep esen s a signi ican ad ancemen in bo h ope a ional e iciency and sus ainabili y.
The esul s clea ly demons a e he e ec i eness o he p edic i e main enance sys em,
which in eg a es RPA ( obo ic p ocess au oma ion) and machine lea ning, con ibu ing o
imp o emen s ac oss mul iple dimensions, such as main enance pe o mance, cos e i-
ciency, sus ainabili y, and o e all inancial ou comes. These indings a e no only ele an o
Alpha bu also o e aluable pe cep ions o b oade applica ions ac oss a ious indus ies.
The analysis o he key pe o mance indica o s (KPIs) e ealed subs an ial imp o e-
men s in ope a ional pe o mance. No ably, he mean ime be ween ailu es (MTBF) mo e
han doubled, om 23.47 h pe e en o 70.4 h pe e en , ep esen ing a 100% imp o emen .
This inc ease in he MTBF highligh s he enhanced eliabili y o he machine y, allowing
Alpha o educe unplanned down ime and imp o e o e all p oduc i i y. Fu he mo e, he
mean ime o epai (MTTR) saw a ema kable educ ion o 67%, om 12 h o 4 h, e lec ing
mo e e icien and as e esponse imes om he main enance eam.
These imp o emen s a e di ec ly a ibu able o he p edic i e capabili ies o he
SIRPM model, which allowed Alpha o an icipa e and add ess ailu es be o e hey occu ed.
The inc eased p edic ion accu acy, om 60% o 90%, demons a es he e ec i eness o
machine lea ning in imp o ing ailu e de ec ion and p oac i e main enance s a egies. This
educed down ime and epai ime, ul ima ely enhancing ope a ional pe o mance and
educing cos s.
The inancial bene i s o implemen ing he SIRPM model a e equally imp essi e. The
educ ion in a e age epai cos s by 52%, om EUR 2500 o EUR 1200 pe inciden , and he
37.5% educ ion in o al main enance cos s, om EUR 4000 o EUR 2500 pe mon h, clea ly
demons a e he cos -sa ing po en ial o p edic i e main enance. The model enabled he
company o educe unplanned down ime cos s by 71.4%, om EUR 3500 o EUR 1000 pe
mon h, leading o a subs an ial dec ease in o al ope a ional cos s, down by 16.67%, om
EUR 30,000 o EUR 25,000 pe mon h. These educ ions we e d i en by he model’s abili y
o iden i y po en ial ailu es ea ly, allowing o less expensi e and mo e planned epai s
a he han eme gency in e en ions.
Addi ionally, he p edic i e main enance model led o a mo e e icien alloca ion
o esou ces. By educing he need o eme gency epai s and minimizing unplanned
down ime, Alpha was able o be e manage i s wo k o ce, epai ma e ials, and ene gy
esou ces, ul ima ely leading o educed ope a ional cos s and imp o ed p o i abili y.
The en i onmen al impac o he SIRPM model was also subs an ial. The educ ion in
ene gy consump ion by 25%, om 10,000 kWh o 7500 kWh pe mon h, can be a ibu ed
o he mo e e icien use o machine y and educed down ime. Simila ly, CO
2
emissions
dec eased by 40%, om 5000 kg o 3000 kg pe mon h, and was e p oduc ion dec eased
by 30%, om 500 kg o 350 kg pe mon h. The imp o emen in ma e ial usage, which
d opped by 18.75%, om 8000 kg o 6500 kg pe mon h, e lec s be e esou ce managemen
Appl. Sci. 2025,15, 854 41 o 47
p ac ices. The ecycling a e also saw an imp o emen , ising by 21.43%, om 70% o 85%,
indica ing a s onge commi men o sus ainabili y and e icien esou ce use.
These en i onmen al imp o emen s align wi h global sus ainabili y s anda ds, such
as ISO 50001, which ocuses on con inuous ene gy pe o mance imp o emen . The 25%
educ ion in ene gy consump ion exceeds he s anda d’s ypical a ge s o 5–10% annual
educ ions, demons a ing Alpha’s leade ship in ene gy e iciency.
The implemen a ion o he SIRPM model had a posi i e e ec on Alpha’s wo k o ce.
Employees we e p o ided wi h aining o ope a e new echnologies, which enhanced
hei echnical skills and empowe ed hem o make mo e da a-d i en decisions. This
upskilling no only inc eased employees’ engagemen wi h he sys em bu also os e ed
a cul u e o con inuous lea ning and imp o emen . As employees became mo e adep a
using he p edic i e main enance sys em, hey we e able o con ibu e mo e e ec i ely o
p oac i e main enance e o s, esul ing in g ea e ope a ional owne ship and imp o ed
eam collabo a ion.
The wo k o ce’s enhanced echnical capabili ies also con ibu ed o g ea e job sa is ac-
ion, as employees we e mo e in ol ed in he decision-making p ocess and be e equipped
o handle eme ging challenges. The shi owa ds a mo e da a-d i en and p oac i e main-
enance cul u e s eng hened he o e all pe o mance o he main enance eam, ensu ing a
mo e esponsi e and e icien ope a ion.
One o he key s eng hs o he SIRPM model lies in i s b oad applicabili y ac oss
a ious indus ies. While he esul s p esen ed he e a e speci ic o Alpha, he co e p inciples
o he model—p edic i e main enance, RPA in eg a ion, and machine lea ning—can be
adap ed o sui he needs o o he sec o s. Fo example, indus ies such as manu ac u ing,
ene gy, anspo a ion, and heal hca e can bene i om implemen ing simila p edic i e
main enance s a egies. In manu ac u ing, he model could enhance machine y up ime
and educe p oduc ion delays. In he ene gy sec o , i could help op imize he pe o mance
o powe plan s and educe he isk o equipmen ailu es ha lead o cos ly ou ages.
Addi ionally, in he anspo a ion indus y, p edic i e main enance could imp o e
lee managemen by iden i ying po en ial issues in ehicles o ai c a be o e hey lead
o ope a ional dis up ions. In heal hca e, he model could be used o p edic equipmen
ailu es in c i ical sys ems, such as medical imaging de ices o pa ien moni o ing sys ems,
ensu ing ha equipmen is always a ailable and ope a ional when needed.
The scalabili y and lexibili y o he SIRPM model make i a highly p ac ical solu ion
o businesses ac oss di e en sec o s. The abili y o cus omize he model based on he
unique needs o each indus y ensu es ha he sys em can deli e op imal esul s ega dless
o he con ex .
The esul s om he implemen a ion o he SIRPM model a Alpha demons a e he
p o ound impac ha p edic i e main enance can ha e on ope a ional e iciency, cos
sa ings, en i onmen al sus ainabili y, and wo k o ce de elopmen . By in eg a ing RPA
and machine lea ning, Alpha was able o achie e signi ican imp o emen s in machine
eliabili y, down ime educ ion, and esou ce op imiza ion, all o which con ibu ed o
enhanced inancial pe o mance and sus ainabili y goals.
The model’s applicabili y o o he indus ies u he emphasizes i s po en ial o d i e
simila imp o emen s in di e se sec o s, o e ing a cos -e ec i e and scalable solu ion o
companies looking o enhance ope a ional e iciency and educe hei en i onmen al oo -
p in . As indus ies con inue o emb ace he digi al ans o ma ion, p edic i e main enance
sys ems like he SIRPM model will play a pi o al ole in shaping he u u e o ope a ions
managemen , sus ainabili y p ac ices, and o e all business pe o mance.
While he esul s p esen ed in his s udy a e speci ic o Alpha, he SIRPM model
demons a es b oad applicabili y ac oss a ious indus ies, gi en i s modula and adap -
