ANALYSIS AND ASSESSMENT OF EFFICIENCY FACTORS OF PORTSIDE TRANSPORT AND TECHNOLOGICAL SYSTEMS DEVELOPMENT IN A MULTI-AGENT ENVIRONMENT

100 Danish Scientific Journal No102, 2025 ANALYSIS AND ASSESSMENT OF EFFICIENCY FACTORS OF PORTSIDE TRANSPORT AND TECHNOLOGICAL SYSTEMS DEVELOPMENT IN A MULTI-AGENT ENVIRONMENT Bakalov M. Candidate of technical sciences, associate professor, associate professor Chair «Management of Maintenance Works», Rostov State Transport University (RSTU), Rostov-on-Don Kolesnikov M. Doctor of technical sciences, associate professor, professor Chair «Economics and Management», Rostov State Transport University (RSTU), Rostov-on-Don Zadorozhniy V. Candidate of technical sciences, associate professor, associate professor Chair «Stations and cargo work», Rostov State Transport University (RSTU), Rostov-on-Don Bakalova Yu. postgraduate, Rostov State Transport University (RSTU), Rostov-on-Don https://doi.org/10.5281/zenodo.17741004 Abstract The paper considers the effectiveness of portside transport and technological systems, a distinctive feature of which is multi-agency. Methods of assessing the factors of transport and technological systems are presented, and a procedure for evaluating by experts is proposed that promotes their mutual learning. Keywords: efficiency of transport systems, multi-agency, expert analysis, reliability, internal and external factors Portside transport and technological systems (PTTS) are classified as large open systems that constantly interact with and depend on the external environment. In the current conditions of political and economic changes, uncertainty, turbulence, for dynamic positive development, it is necessary to increase the internal and external efficiency of portside transport and technological systems [1]. The effectiveness of such systems is understood in two senses: in the broad and narrow senses of the word. In a narrow sense, the financial and economic indicators of the system are studied as a rule. In a broad sense, the structure of performance indicators is much more complex. Here is a possible list of them: - Production indicators (the number of loaded and/or unloaded, transported goods, the volume of manufactured products, rendered services, etc.). - Financial and economic indicators (cost of work, income, profit, profitability, etc.). - Quality of work, products, services rendered. - Work safety. - Reliability (technical, technological, organizational). Technical reliability (reliability of devices and equipment) is well developed. And technological and organizational reliability is studied in the theory of organizational and technological reliability [2]. - The survivability of the studied systems, formed through resistance to external and internal disturbances. As a rule, it includes adaptability to operating conditions changes, self-organization of the system, which can also be considered as independent indicators. Survivability is studied using Lyapunov's theory of stability [3], the theory of cenoses [4], and the theory of catastrophes [5]. These performance indicators are not independent. They correlate with each other. For example, a more reliable system is usually more survivable (positive correlation). And higher-quality products will cost the company more, which reduces its financial and economic performance (negative correlation). The key concepts of PTTS research are: 1. Agents acting as: - entities (managers, executors, experts, manufacturers of transport services and their consumers); - infrastructure facilities (compressor station for the yard, warehouse for the port, etc.); - technological processes. In other words, our agents are of a different nature and will communicate within the framework of the Internet of Things. 2. Interactions of agents, which may also have different nature (subordination, competition, partnership, etc.). Therefore, it is an important task to classify agents and relationships, which will determine their role and place in the work of the PTTS. Classifications can be carried out on the basis of various features inherent in the objects of research. For example, it is possible to classify agents participating in the transportation process by the type of interaction: 1. In relation to the cargo transfer process – the consignor, the consignee, the carriers, the chatbots. 2. In relation to the process of cargo transportation – transportation organizers (forwarders), technology owners, infrastructure owners. Thus, at the first stage, it is necessary to determine the list of features characterizing the object of research (agents or their interactions). This procedure is implemented on the basis of expert assessment methods. These methods include: - ranking and rating method; - the method of studying the strengths and weaknesses of an organization, opportunities and threats to its activities – the SWOT analysis method; - direct assessment method; - the method of developing and analyzing goals – the SMART method. Danish Scientific Journal No102, 2025 101 Below we will consider the simplest version of such a procedure with respect to the object under study. For example, we are interested in a set of features that characterize cargo delivery. 1. A group of experts is asked about the composition of the attributes. Each expert specifies his own set based on his experience and intuition. Let the first expert indicate the following features: p11 is achieving the shortest possible delivery time, p12 is maximizing profits, and p13 is reducing costs. The second one is p21 as achieving the shortest possible cargo delivery time, p22 is maximizing profits, p24 is cargo delivery safety, and p25 is gaining the maximum market share. 2. We combine these groups of features and present them to the experts again. Additionally, we ask to indicate the significance values of the total set of features in some scale accessible to this expert. It is possible to indicate a zero value if the expert continues to insist on its insignificant influence. 3. Since experts have different experience and knowledge, the scales they use will be different. At this (third) stage, we will translate all the values using scaling into one universal range of values [0; 1]. 4. In the following steps, we will carry out the averaging of the obtained values. The formalized form of the described procedure is as follows: The list of features set by the i-th expert has the form: Pi = (pi1, pi2, …, pini) (1) The combined list of attributes has the form: P = (p1, p2, ..., pn) (2) In our case, n1 = 3, n2 = 4, n = 5. Let the first expert evaluate these combined features using the following values: (5, 2, 3, 0, 4) (3) The second one evaluated the same 5 features: (5, 6, 5, 7, 10) (4) Scaling is carried out according to the formula 𝑢𝑖=𝑥𝑖−𝑚𝑖𝑛 𝑥𝑖 𝑚𝑎𝑥 𝑥𝑖−𝑚𝑖𝑛 𝑥𝑖 . (5) Applying (5) to (3), we obtain: (1; 0,4; 0,6; 0; 0,8). (6) Similarly for (4) we obtain: (0; 0,2; 0; 0,4; 1). (7) If we take the arithmetic mean as an average, the resulting values will look like: (0,5; 0,3; 0,3; 0,2; 0,9). (8) This procedure has a positive property: it promotes mutual training of experts. By exchanging information, they take into account each other's experience. The limited competence and awareness inherent in any person causes an inaccuracy in expressing expert assessments. Let us assume that this error is the same for all experts and is characterized by variance D. It is known from probability theory that the variance of the arithmetic mean is n times less than the variance of one dimension [6]. That is, D(n) = D/n. (9) This implies the most important property of multiple measurements result: its uncertainty is less than the uncertainty of a single measurement result. Thus, an increase in the number of experts reduces the likelihood of making an erroneous decision and increases the reliability of the result. An important issue of the procedure is the coordination of the averaging procedure. The theory knows various averages: arithmetic, geometric, harmonic, etc. The choice of the average should be justified by the properties of the technological process under study. It is also necessary to characterize the definition of efficiency. The performance indicator is usually described in terms of getting the desired result. We introduce the concept of private performance indicators: achieving the shortest possible delivery time, maximizing profits, reducing costs, securing cargo delivery, gaining the maximum market share, etc. Based on them, it is difficult to make an informed decision as they can be mutually contradictory. For example, let us consider a variant with the process of cargo delivery by various modes of transport by rail and road (B1 and B2, respectively). How to choose the most efficient shipping option. To do this, we will perform the following calculation. Let us take the following designations of the selected performance indicators, where: x1 is delivery time, x2 is delivery cost, x3 is profit earned, x4 is cargo safety. Let us write down the values of the factors for delivery by various modes of transport based on expert estimates: B1 = (x11, x21, x31, x41) = (0,2; 0,4; 0,6; 0,5); B2 = (x12, x22, x32, x42) = (0,2; 0,4; 0,6; 0,5). Based on the experts' assessment, we will indicate the share of influence of the selected indicators and we will get: P1= 0.4 x1, P2=0.3 x2, P3=0.4 x3, P4=0.5 x4. It is extremely difficult to draw a conclusion about the effectiveness of a particular type of delivery based solely on the data obtained, due to the presence of several performance indicators and various expert influence shares. To overcome this contradiction, generalized criteria are often used, for example, a weighted sum of partial criteria (linear convolution), which is calculated as: Ei=P1x1i+P2x2i+P3x3i+P4x4i, where i=1,2. We get the values: E1= 0,4·0,2+0,3·0,4+0,4·0,6+0,5·0,5=0,69; E2= 0,4·0,4+0,3·0,4+0,4·0,5+0,5·0,3=0,63. Based on the results obtained we conclude that the most effective type of transportation is the 2nd type – automobile transportation. When performing further calculations, taking into account multiple selection criteria, it is necessary to evaluate the values of the indicators. Some of them are given in numerical form (volumes of transported goods, profits, etc.), others are qualitative. For example, safety can be "low", "medium", or "high". For the former, it is advisable to use statistical analysis; for the latter, there are two well-developed ways of calculating: 102 Danish Scientific Journal No102, 2025 - Expert statistical analysis. - Using fuzzy set theory [7,8]. A separate task is efficiency optimization, which requires further research. The factors of effectiveness development presented in the study and the methods of their assessment make it possible to select the most preferable options for the development of PTTS based on the analysis performed and taking into account the situation in the region and the world. Acknowledgements The research was supported by the Russian Science Foundation grant No. 24-29-00869, https://rscf.ru/project/24-29-00869/ References: 1. Chernyaev, A.G. Based on the development of infrastructure and polygon technologies / A.G. Chernyaev, V.N. Zubkov, M.V. Bakalov // Railway transport. – 2016. – No. 9. – P. 32–37. 2. Vereskun, V. D. Organizational and technological reliability and efficiency of railway transport production facilities : a monograph / V. D. Vereskun. Novosibirsk : SGUPS, 2010. 254 p. 3. Fundamentals of the theory of motion stability in relation to the tasks of space technology: a textbook / E.V. Barinova. Samara: Samara University Press, 2023. 120 p. 4. Kuzminov A.N. Methodology of the cenological analysis of economic systems : a monograph / A. N. Kuzminov ; Rostov-on-Don : RSTU, 2009. 258 p. 5. Arnold, V.I. Theory of catastrophes / V.I. Arnold. Moscow: Editorial URSS, 2004. - 126 p. 6. Akhtyamov A.M. Probability theory for socioeconomic specialties. Moscow: Fizmatlit, 2016. 304 p. 7. Chislov O. Fuzzy modelling of the transportation logistics processes / O. Chislov, N. Lyabakh, M. Kolesnikov, M. Bakalov et al. // Journal of Physics: Conference Series. 2021. Vol. 2131. Iss. 032007. DOI: 10.1088/17426596/2131/3/032007. 8. Neuro-fuzzy modeling of transport and logistics processes / O. N. Chislov, N. N. Lyabakh, M. V. Bakalov [et al.] // Transport: science, technology, management. Scientific information collection. – 2022. – No. 10. – P. 23-27.