Fault Detection and Classification in Interconnected System with Wind Generation Using ANN and SVM

POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 20 | NUMBER: 3 | 2022 | SEPTEMBER Fault Detection and Classification in Interconnected System with Wind Generation Using ANN and SVM Hinal SHAH 1 , Nilesh CHOTHANI 2 , Jaydeep CHAKRAVORTY 1 1 Department of Electrical Engneering, Institute of Technology and Engineering, Indus University, Rancharda, 382115 Ahmedabad, India. 2 Department of Electrical Engineering, Adani Institute of Infrastructure Engineering, Gujarat Technological University, Nr Vaishnodevi Circle, SG Highway, 382421 Ahmedabad, India [email protected], c[email protected], [email protected] DOI: 10.15598/aeee.v20i3.4483 Article history: Received Feb 20, 2022; Revised May 02, 2022; Accepted May 24, 2022; Published Sep 30, 2022. This is an open access article under the BY-CC license. Abstract. Protective relays are installed in generation, transmission, and distribution system for detection, classication, and estimation of faults. To match the future load demand and to get uninterrupted power supply, use of renewable energy sources are increasing day by day. Faults can occur in transmission lines, transformers, generators, and busbars but the nature of these faults may change many times when renewable energy sources are considered. This research paper introduce techniques to detect and classify different faults on transmission line in the presence of wind energy sources using efcient tools of articial intelligence. The main challenges of the system fault detection, in presence of wind turbine lie in their non-linearity, uncertainty and unknown disturbances. PSCAD/EMTDC software tool is used to simulate the power system model with RES which is implemented in MATLAB and Python software. Articial Neural Network (ANN) and Support Vector Machine (SVM) algorithms have been used to classify and detect faults on transmission lines connected with wind energy source. The proposed technique has been validated for internal faults on transmission line and external faults on power system. In total of 4320 internal and external fault cases with wide variation in system parameters have been used for validation of the proposed model. The proposed model gives an overall fault zone identication accuracy of more than 99 % in presence of wind energy source. The results obtained from validation show that the performance of SVM classier is better than ANN in term of efcacy and classication time. Keywords Articial Neural Network, Support Vector Machine, transmission line, fault classication, renewable generation 1. Introduction Renewable energy generation is increasing tremendously nowadays whole over the world to mitigate electricity demand. Small scale and large scale penetration of wind and solar system are creating problems of false tripping, over reach, under reach and malfunctioning of transmission line relay. To overcome above problems at transmission, distribution and micro-grid level, scientists have done enough research work. In the present era, use of renewable energy sources is signicantly increasing to generate electricity as a progressive attempt towards prospective low carbon emission system [1]. Different factors are affecting protection systems of transmission line when integrated with renewable energy sources. The variation of wind parameters signicantly affects the reach of the distance protection of transmission line. Short circuit behavior is completely different in induction types wind generators as compared to conventional synchronous generators, which is one of the important aspect to decide the characteristics of the distance protection. Distance protection characteristic is also affected by parameters like fault location, wind speed, mutual coupling, © 2022 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 225 POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 20 | NUMBER: 3 | 2022 | SEPTEMBER fault resistance etc. in the presence of wind system [1] and [2]. In last few years, many researchers have focused to solve the protection issues, using numerical relay associated with the signal processing and machine learning techniques. Fast and accurate fault identication and discrimination is a primary goal of numerical protection relay. In article [3], comprehensive review of different fault detection, classication and location has been presented. This paper serves as a guideline for the researchers who are working in this domain. Over the years, many machine learning and classication techniques have been developed, tested, and implemented in the electrical power system. Few of them are mentioned in the research article [3]. In [4], Arti- cial Neural Network (ANN) based back propagation technique has been implemented. Syntactic pattern recognition function model has been efciently used for detection of fault at transmission line. Moreover, VHSIC Hardware Description Language (VHDL) has been implemented on power system model for measurement of system parameters [5]. In research article [6], Deep Neural Network has been applied for fault detection and classication. In the case of fault detection, researchers have investigated the effects of two hyper parameters, number of hidden layer and number of neurons in the last hidden layer on the performance of networks. The author concluded that by increasing the network size, the fault detection accuracy did not improve above certain level. Authors in [7] implemented Support Vector Machine (SVM) technique for fault detection, and ANN technique for fault location and classication in 400 kV three phase double circuit transmission line with linear and non-linear load at better accuracy. Other authors also implemented SVM classier on 400 kV transmission line and has achieved fault classication accuracy of 99.5 % [8]. The data can be analyzed and classied based on Articial Neural Network [9]. Fuzzy interfaced scheme has been proposed in [10], which gives 99 % accuracy for detection of fault. Decision Tree method has been introduced in [11]. This method uses data from one side of the protected line and the decision is performed in less than a quarter cycle. The ANN and SVM-based approach to real-time fault classication with high accuracy and high speed implementation are discussed in research article [12], [13], [14] and [15]. Modied multi-class SVM approach has been implemented and discussed in article [13] for distribution system fault detection. In [14], fault prediction in presences of wind DG using python algorithm is proposed. Proposed method also reduced the time require to clear the fault in wind based power system network. In [16], authors presented adaptive reach of numerical distance relay by considering various system parameters. In article [15], a modied multiclass SVM technique has been used to detect and classify fault in distribution system. The Radial Basis Function (RBF) kernel function has been used to develop MMC-SVM model. To improve impedance reach of the numerical relay by adaptive setting of the quadrilateral characteristics was proposed in research paper [16]. In research article [17], SVM technique has been used to detect and classify fault, whereas ANN based classication has been shown in [19] and [26]. Multiple SVM model based hybrid classication has been introduced in [20] and [21]. Classication and location of fault in distribution network with renewable source has been implemented in [22]. Dynamic and static model comparison to classify faults in power system network has been given in [23]. Multi-resolution analysis using stockwell's transform has been implemented for detection of LG, LL, LLG and LLLG faults in power system network integrated with wind energy system. S-contour, amplitude plot and variance graph has been used to recognize the fault [25]. Decision Tree and concurrent neuro fuzzy AI techniques has been applied in [32] for fault classication and detection on nine phase transmission line system. However, as stated in [32], the complexity will be increased with the increase in the level of phases and will reduce the accuracy of program execution. The performance of the power system has been investigated during a noisy condition in [29], [30] and [31], in which white Gaussian noise has been contaminated with the recorded fault signals measured at the relaying point. The results show that the fault index is higher than threshold with noise signal. Therefore, the proposed protection scheme is not affected by the distorted signal in the presence of recorded signal as given in article [29], [30] and [31]. However, the accuracy of WT based technique is affected by high frequency noise signals penetrated during decomposition of current signals. The same is not much affected for classication technique based on NN, SVM and RVM. Different techniques investigated by several researchers for detection, classication, and localization of transmission line faults are described in [18] and [24]. MHO relay is widely used in the protection of transmission line to detect all kinds of faults. However, this relay sometimes fails to detect high resistance fault in its own zone of protection under the situation of varying system and fault parameters. In this paper, a portion of power system has been simulated in PSCAD, where 100 km long transmission line has been considered. To test the MHO relay characteristic, a line to ground fault with varying fault resistance has been created at 70 km of line length (in-zone fault). Performance of distance protection by MHO relay at fault resistance of 5 Ω , 10 Ω , and 18 Ω have been shown in Fig. 1, Fig. 2, and Fig. 3 respectively. 226 © 2022 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 20 | NUMBER: 3 | 2022 | SEPTEMBER -30 -20 -10 0 10 20 30 Resistance (R) -5 0 5 10 15 20 25 30 35 40 45 50 Reactance (X) Mho Characteristics RG YG BG RY YB BR Fig. 1: L-G fault at 70 km, fault resistance 5 Ω . -30 -20 -10 0 10 20 30 Resistance (R) -5 0 5 10 15 20 25 30 35 40 45 50 Reactance (X) Mho Characteristics RG YG BG RY YB BR Fig. 2: L-G fault at 70 km, fault resistance 10 Ω . -30 -20 -10 0 10 20 30 Resistance (R) -5 0 5 10 15 20 25 30 35 40 45 50 Reactance (X) Mho Characteristics RG YG BG RY YB BR Fig. 3: L-G fault at 70 km, fault resistance 18 Ω . The results represent the effect of fault resistance variation on distance protection characteristics. Relay is misoperating in the second or the third zone as shown in Fig. 2 and Fig. 3, respectively, even though the fault is in the rst zone due to increasing value of fault resistance. Similarly, the variation of other parameters of power system network may weaken the performance of the relay under faulty conditions specically with the penetration of renewable sources in the network. This may create a problem of under reach and overreach of protective scheme in the transmission line. The ANN and SVM techniques are presented in this article for classication of in-zone and out-of-zone faults on transmission line. Various fault resistance, fault inception angles, load angles, and fault locations are considered in the presence of wind generation system. Feasibility of the proposed algorithms has been tested on an IEEE 9 bus power system network with integration of wind system at bus 3. The system model has been developed using PSCAD/EMTDC software package. A simulation data set of 12570 cases has been generated using an automatic fault data generation model developed by the authors. Among which, 4320 simulation cases have been considered for validation of the proposed ANN and SVM technique. 2. Proposed System Modelling Figure 4 shows a single line diagram of IEEE 9 bus 230 kV electrical power system network considered for the simulation studies. IEEE 9 bus system is consisting power generators G1, G2 and wind system generator G3, six transmission lines, three transformers and three loads connected at bus 4, 5 and 8. The generators G1, G2 are modeled as an equivalent dynamic source consisting of a multi machine system connected to bus 1 and 2 respectively. Whereas generator G3 is Type 3 Wind Turbine Model used as renewable energy source (wind farm) which is intermittent in power generation. Bergeron model with distributed parameters has been used for modelling of transmission line. The system including generation system, transmission line, transformer and connected load parameters are given in the appendix. A sampling frequency of 4 kHz at 50 Hz nominal frequency has been used. A channel plot step has been taken as 250 µ s, i.e. 80 samples/cycle. Post fault data have been captured with measuring devices like CVT and CT. The same conguration is used normally in digital relay available at the market. All ten types of faults on line between bus 8 and bus 9 at various locations with different values of fault resistance, fault inception angles and power ow angles have been simulated, including large numbers of internal faults. For each case, the voltage and current values are measured and saved as a data le from PSCAD software. In the similar way, external faults have been also simulated outside the line between bus 8 and bus 9 including location on bus 8, bus 9, line between bus 7 and 8, line between bus 6 and 9 along with all above mentioned internal fault parameters. © 2022 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 227 POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 20 | NUMBER: 3 | 2022 | SEPTEMBER 270 MW 18.45 KV G2 27 8 300 MVA 18.45/230 KV Y/Y L1 5 170.338 km 89.93 km 76.176 km L3 106.646 km 93 GT3 GT2 150 MVA 230/13.8 KV Y/Y 120 MW 13.8 KV G3 L2 6 97.336 km 4 1 GT1 550 MVA 17.16/230 KV ∆/Y 512 MW 17.16 KV G1 179.86 km Fig. 4: Single line diagram of IEEE 9 bus power system network. Tab. 1: Internal fault cases generated for fault on transmission line bus 89. Power system parameter ANN/SVM training patterns ANN/SVM unseen testing patterns Variation in parameter Numbers of variation Variation in parameter Numbers of variation Fault type L-G, LL, LL-G, LLL-G 10 L-G, LL, LL-G, LLL-G 10 Fault location FL (km) 10 %, 20 %, 30 %, 50 %, 75 % 515 %, 40 %, 60 %, 80 % 4 Fault resistance RF ( Ω )1 Ω , 5 Ω , 10 Ω , 15 Ω , 20 Ω 5 0 Ω , 8 Ω , 15 Ω , 25 Ω 4 Power ow angle of G2 ( δ )0 ◦ , 5 ◦ , 10 ◦ 3 4 ◦ , 8 ◦ , 12 ◦ 3 Fault inception angle FIA ( ◦ )0 ◦ , 45 ◦ , 90 ◦ , 135 ◦ , 180 ◦ 5 0 ◦ , 30 ◦ , 80 ◦ , 135 ◦ 4 Total (5670) Total training patterns for fault 3750 Total testing patterns for fault 1920 Tab. 2: External fault cases created for fault outside of transmission line bus 89. Power system parameter ANN/SVM training patterns ANN/SVM unseen testing patterns Variation in parameter Numbers of variation Variation in parameter Numbers of variation Fault type L-G, LL, LL-G, LLL-G 10 L-G, LL, LL-G, LLL-G 10 Fault location FL (km) On bus-8, On bus-9, Line 78 (2 location, 40 %, 70 %), Line 9-6 (2 location, 40 %, 70 %) 6 On bus-8, On bus-9, Line 7-8 (1 location, 50 %), Line 9-6 (2 location, 30 %, 60 %) 5 Fault resistance RF ( Ω )1 Ω , 5 Ω , 10 Ω , 15 Ω , 20 Ω 5 0 Ω , 8 Ω , 15 Ω , 25 Ω 4 Power ow angle of G2 ( δ )0 ◦ , 5 ◦ , 10 ◦ 3 4 ◦ , 8 ◦ , 12 ◦ 3 Fault inception angle FIA ( ◦ )0 ◦ ,45 ◦ , 90 ◦ , 135 ◦ , 180 ◦ 5 0 ◦ , 30 ◦ , 80 ◦ , 135 ◦ 4 Total (6900) Total training patterns for fault 4500 Total testing patterns for fault 2400 228 © 2022 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 20 | NUMBER: 3 | 2022 | SEPTEMBER Table 1 and Tab. 2 shows different cases of 5670 internal faults and 6900 external faults created, respectively. It is observed from Tab. 1 that out of 5670 internal faults and 6900 external faults, 3750 (66.16 % of total internal cases) and 4500 (65.22 % of total external cases) have been utilized as training process of ANN and SVM. The remaining 1920 (33.84 % of total internal cases) and 2400 (34.78 % of total external cases) have been utilized for testing and validation of the proposed algorithm. The trained ANN and SVM based fault classier models are then extensively used for testing of unseen fault data. The ANN and SVM fault detection technique have been veried for all symmetrical and asymmetrical faults (L-G, LL, LL-G, LLL-G) at different locations. These algorithms are tested with wide variation in fault resistance, Fault Inception Angle (FIA) (0180 ◦ ) and also load ow angle are evaluated for internal and external faults in the system. Input Hidden nodes Output u1 u2 u3 u4 uk w11 w12 wik 1 2 3 4 k 1 2 3 i 1 j w11 wj1 wji Fig. 5: Feed forward neural network topology. 3. Articial Neural Network Technique (ANN) 3.1. Evaluation of Training model and Testing Before applying the ANN, the training and testing data sets are normalized column wise using Eq. (1) to avoid under tting issues, as it may destroy accuracy of the model. The model generally does not perform well for given data set, so now pre-processing of the data points, removal of noise from the data is the prime requirement [6]. The training and testing input values are required to re-scale using Eq. (1). Input values ui are normalized as shown in Eq. (1) to improve the accuracy of algorithm to detect and classify faults of power system network. ui=(ui−umin) (umax −umin), (1) where, ui is the input values of post fault sending end and receiving end voltages and currents, umax and umin are the maximum and minimum values of the input column, respectively. b ui ftj uj Input Adder Activation function Output Weight adjustment bias Compare with target output ∑ Fig. 6: Feedback Supervised Learning ANN structure. Human brain has millions of neurons which do many sensitive tasks. It takes signals from different parts of the body and using the brain, it generates appropriate action naturally. The ANN works similarly but it is articial in nature. The ANN has capability of parallel processing, nonlinear mapping, online and ofine learning approach. Neurons are known as nodes in articial system. The ANN has input layer, hidden layer and output layer. ANN process depends on network topology like, feed forward single or multi-layer as shown in Fig. 5 [4], and feedback network (weight updating or learning) as given in Fig. 6. The ANN basically classies three types of learning methods, supervised learning, unsupervised learning and reinforcement learning. Here, supervised learning method has been used as shown in Fig. 6 [4]. The estimated output has been compared with the desired output; the error signal is generated as the difference between the predicted values and the actual values. Based on the error signal, weights are modied to minimize the error so that desire output matches with the calculated output. ANN algorithm can be applied as a feed forward and feedback neural network. Here in this paper, the back propagation method has been applied for detection and classication of faults. Back propagation algorithm eventually corrects the weights among the different layers, according to the difference between the targeted output and calculated output. An activation function makes back propagation achievable since the gradient are passed with error to update weight and bias. Linear activation function and nonlinear activation function such as Sigmoid, Tanh, ReLU, Softmax activation functions are used to achieve the accurate output. Output of the hidden layer is calculated from the activation function. Activation value of the connected node depends on the summation of bias and weights sum of all inputs connected to it as given by Eq. (2) and Eq. (3). uj=f( n X i=0 uiwji +bi), (2) where uiwji =u1wj1+u2wj2+u3wj3+· · · +unwjn. (3) © 2022 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 229 POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 20 | NUMBER: 3 | 2022 | SEPTEMBER In normal practice Rectied Linear Unit (ReLU) activation function is preferred because of less computation, faster in operation and easy to reach at desire output. But for binary classication sigmoid function is widely used. Sigmoid non linear transformation is used to detect and classify faults as shown in Eq. (5). Equation (5) is computed from Eq. (2), Eq. (3) and Eq. (4). Netj= n X i=0 uiwji +bi, (4) uj=f(Netj) = 1 1 + e−Netj=1 1 + e−Pn iuiwji+bi, (5) where, Netj = Net Input of the jth Layer, bi = Bias of hidden layer, uj Bias is the degree of sensitivity, with which the hidden layer uj answer to the perturbation it receives by the net input. Equation (5) represents the feed forward algorithm of neural network. Error factor is calculated by taking square of actual outputs subtracted from target outputs summation [27] as shown in Eq. (6) and Eq. (7). Error signal is dened as: Ex=1 2X k (txk −uxk)2=1 2X k (txk −fk(Netxk))2, (6) Ex=1 2X k (txk −fk(X j (wkjuxj +bk))2, (7) where, E is the error, x is the model, tk is the target, uk is the ouput To correct the weight for achieving desire output, the back propagation delta rule has been applied. The coefcient of error in delta rule is calculated by difference between the actual output and the predicted output and relating this difference to the derivative between the activation state of the actual output and the net input of that output as shown in Eq. (8) and Eq. (9). ∂uj ∂Netj =uj(1 −uj), (8) ∆outj= (tj−uj)∂uj ∂Netj = (tj−uj)uj(1 −uj), (9) where, tj is the target output, uj is the actual output, uj(1−uj) is the derivative between actual output and net input of jth layer as given in Eq. (8). The error coefcient of back propagation method is indicated in Eq. (9). Weight correction is calculated using Eq. (10) with the help of Eq. (8) and Eq. (9). ∆wji =− ∂Ex ∂wji , (10) ∂Netxk ∂wkj =∂(Pjwkj uxj +bk) ∂wkj =uxj. (11) By substituting Eq. (11) into Eq. (12), we obtain: − ∂Ex ∂wkj = (txk −uxk)f′ k(Netxk)uxj . (12) By substituting Eq. (12) and Eq. (9) into Eq. (10), we obtain: ∆wji =r∆outjuj. (13) Quantity of the value added or subtracted from the weight wji depends on δoutj with respect to the activation state of layer ui the activation with which uj is connected to weight wji and in relation to coefcient r as shown in Eq. (13). The δwji can be negative or positive. The value can be added or subtracted from the previous value of weight wji as shown in Eq. (14). wji(n+1) =wji(n)∆wji. (14) In Eq. (14), each arriving layer of weight has an actual value which is comparable with an ideal value as mentioned in the articles [14] and [27]. Figure 7 shows owchart of ANN training model. 3.2. ANN Technique Result Discussion The ANN Back propagation model is trained using MATLAB functions and Python coding. Both the software, MATLAB and Python are giving satisfactory results of faults classication as shown in Tab. 3 to Tab. 6. Table 3 shows overall classication accuracy of in-zone faults (line between buses 89) and out-ofzone faults on transmission line using Python coding. Table 4 shows the fault type wise classication accuracy using Python coding. Table 5 shows the fault classication accuracy with 6 hidden layers at different training functions in MATLAB. Training function depends on many factors, such as complexity of the problem, the number of data points in the training set, the number of weights and biases in the network, no of hidden layers, the error goal, and whether the network is being used for pattern recognition regression. Table 5 shows the fault classication accuracy of internal and external faults achieved by Resilient, Gradient descent and Levenberg-Marquardt (LM) back propagation training functions. Levenberg-Marquardt is the fastest back propagation algorithm and gives better accuracy compared to the other two back propagation algorithms. At the same time, Levenberg-Marquardt requires more memory for the execution of the program. However, resilient is slow in convergence but is a memory efcient algorithm. Accuracy of the various internal and external faults identication with back propagation training function Levenberg-Marquardt 230 © 2022 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 20 | NUMBER: 3 | 2022 | SEPTEMBER Tab. 3: Classication accuracy for internal and external faults using Python. Condition No of test cases Total classication data % Accuracy ( η ) TP TN Internal faults 1920 1899 21 98.90 % External faults 2400 2360 40 98.33 % Total 4320 4259 61 98.58 % TP = Test Positive (true) and TN = Test Negative (false) Tab. 4: Fault zone identication accuracy for various types of In-zone and Out-of-zone faults in Python. Fault type In zone fault cases Out-of-zone fault cases No of test cases Total classication data % Accuracy No of test cases Total classication data % Accuracy TP TN TP TN LG 576 570 6 98.95 % 720 712 8 98.88 % LLG 576 570 6 98.95 % 720 709 11 98.47 % LL 576 568 8 98.61 % 720 701 19 97.36 % LLLG 192 191 1 99.48 % 240 238 2 99.16 % Tab. 5: Classication regression accuracy for internal and external faults using MATLAB. Hidden layer = 6 Condition Back propagation training function % Accuracy ( η ) train % Accuracy ( η ) test % Accuracy ( η ) valid Internal fault (1920 cases) Resilient 97.39 96.77 96.25 Gradient descent 96.41 95.05 94.73 Levenberg-Marquardt 98.12 98.1 98.25 External fault (2400 cases) Resilient 92.13 91.56 90.83 Gradient descent 92.03 91.09 90.72 Levenberg-Marquardt 98.28 97.13 96.92 Tab. 6: Fault zone identication accuracy for internal faults at different hidden layers using MATLAB. Fault type No of test cases Hidden layer = 6 Hidden layer = 10 Total classication data % Accuracy ( η ) Total classication data % Accuracy ( η ) TP TN TP TN LG 576 576 0 100 % 576 0 100 % LLG 576 576 0 100 % 574 02 99.65 % LL 576 549 27 95.3 % 576 0 100 % LLLG 192 183 09 95.3 % 192 0 100 % Total 1920 1884 36 98.1 % 1918 02 99.89 % Tab. 7: Fault zone identication accuracy for external faults at different hidden layers using MATLAB. Fault type No of test cases Hidden layer = 6 Hidden layer = 10 Total classication data % Accuracy ( η ) Total classication data % Accuracy ( η ) TP TN TP TN LG 720 717 3 99.58 % 720 0 100 % LLG 720 689 31 95.69 % 686 34 95.3 % LL 720 704 16 97.78 % 712 8 98.89 % LLLG 240 232 8 96.66 % 240 0 100 % Total 2400 2342 58 97.6 % 2358 42 98.3 % It is observed that the overall fault classication accuracy of the ANN algorithm is 98.58 % using Python programming performed in Python 3.7.1 software tool. Whereas, using MATLAB programming performed in MATLAB Version 9.4 (R2018a) software tool, LM function gives test accuracy more than 97 % as stated in Tab. 5. Table 4, Tab. 6 and Tab. 7 shows the classication accuracy for different types of faults on considered line and outside it. increases with increasing hidden layer from 6 to 10 as given in Tab. 6 and Tab. 7, respectively. The confusion matrix plots have been plotted in Fig. 8 and Fig. 9 of all symmetrical and asymmetrical internal faults with hidden layer 6 and 10, respectively. The confusion matrix represents the total number of fault detection observations in each cell. The rows of the confusion matrix correspond © 2022 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 231 POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 20 | NUMBER: 3 | 2022 | SEPTEMBER Start learning Import and initialize training and testing data base Normalize the training and testing data set Initialize the random weights for hidden and output layer Read input and target output as pair of the network Calculate the actual output of the network Calculate error of each output node Error = (target output – actual output) Error < threshold? End learning Adjust the weight of neural Network to minimize the error Less than threshold N Y Fig. 7: Flowchart of the back propagation ANN Training Model. to the output class (actual value) and the columns of the confusion matrix correspond to the target class (predicted values). In the confusion matrix, all ten types of faults have been mentioned, i.e. rst three faults are L-G faults for R-Y-B 3-phase, respectively. Diagonal and off-diagonal cells show correctly and incorrectly classied fault observations, respectively. As shown in Fig. 8, out of 1920 fault cases, 192 fault cases that are taken for each fault types. The rst column represent 192 fault cases are correctly classi- ed as R-G fault type, therefore the column accuracy for R-G fault types are 100 %. Also rst row indicates 192 R-G faults along with 4 YB (LL). faults are misclassied as R-G (L-G) faults, so row accuracy was reduced to 98 %. It has been observed from Fig. 8 and Fig. 9 that the accuracy of the classication increases with the increases in the number of hidden layers. Similarly, the accuracy of the identication of external faults increases with the increment of hidden 1 2 3 4 5 6 7 8 9 10 Target Class 1 2 3 4 5 6 7 8 9 10 Output Class Confusion Matrix 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 100% 0.0% 4 0.2% 0 0.0% 0 0.0% 0 0.0% 23 1.2% 0 0.0% 0 0.0% 0 0.0% 165 8.6% 0 0.0% 85.9% 14.1% 0 0.0% 0 0.0% 9 0.5% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 183 9.5% 95.3% 4.7% 98.0% 2.0% 100% 0.0% 95.5% 4.5% 100% 0.0% 89.3% 10.7% 100% 0.0% 100% 0.0% 100% 0.0% 100% 0.0% 100% 0.0% 98.1% 1.9% Fig. 8: Confusion matrix internal faults HL = 6. 1 2 3 4 5 6 7 8 9 10 Target Class 1 2 3 4 5 6 7 8 9 10 Output Class Confusion Matrix 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 2 0.1% 0 0.0% 0 0.0% 190 9.9% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 99.0% 1.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 0 0.0% 100% 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 0 0.0% 192 10.0% 100% 0.0% 99.0% 1.0% 100% 0.0% 100% 0.0% 100% 0.0% 100% 0.0% 100% 0.0% 100% 0.0% 100% 0.0% 100% 0.0% 100% 0.0% 99.9% 0.1% Fig. 9: Confusion matrix internal faults HL = 10. layers from 6 to 10. With further increase of the number of hidden layers, algorithm increases data classi- cation accuracy but simultaneously the convergence time also increases and this slows down the learning process to achieve the target. The performance curves of training, validation and test data for internal fault with 10 hidden layers is shown in Fig. 10. It has been observed that all three curves are similarly formed, this means that the network responds similarly to learning data as well as to the validation and test data by reducing the probability of over-tting [21]. Over trained or over-tting occurs if the validation error increases at the same epoch, the training error slope decreases. 232 © 2022 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING POWER ENGINEERING AND ELECTRICAL ENGINEERING VOLUME: 20 | NUMBER: 3 | 2022 | SEPTEMBER The regression plot representing the regression analysis between the network output and the corresponding target was carried out. Figure 11 shows the regression plot of the In-zone fault with 10 hidden layers. It shows a good t of ANN predicted values to actual output data for training (70 %), testing (15 %), and validation data sets (15 %). The data set model includes all training, testing and validation data sets. In Fig. 11, `R' show the regression factor. R represents the slope of linear tting. Output equation used in this method for regression plot is given in Eq. (15). Output =w·T arget +b, (15) where, w is the weight, b is the bias. 0123456789 9 Epochs 10-4 10-3 10-2 10-1 100 Mean Squared Error (mse) Best Validation Performance is 0.00017288 at epoch 9 Train Validation Test Best Goal Fig. 10: Best validation performance error vs. epochs. 4. Support Vector Machine (SVM) Classication Technique SVM is a statistical technique used for the purpose of computational learning which overcomes the drawback of ANN by giving a global solution rather than a local minima [16]. SVM classiers offer great accuracy and work well with high dimensional space. SVM classiers basically use a subset of training points hence very less memory required in validation. SVM classiers can be used either in single layer as binary classier which has two possible states in-zone faults (+1) and out-of-zone ( −1 ) fault or multi-layer classi- er, which is a discrete classier that mainly focused on regression problems. The inputs of the SVM classiers provide maximum amount of margin between different class labels. Boundary between the In-zone and Out-zone fault class is known as hyperplane [8]. It is represented by Eq. (16). f(x) = wT·x+b= 0, (16) where, w is weight vector and b is bias term to determine position of hyper-plane. The separation distance can be increased by considering minimum value of w . For linear separation, SVM can be realized by support vector as shown in Eq. (17). Labels of the output class are given as shown in Eq. (18) and Eq. (19). ϕ(w) = 1 2wTw, (17) IfwT+b⩾1,target class = 1, (18) wT+b⩽1,target class = −1. (19) Here, the output function f(x) is equal to ' +1 ' which indicates one class of label (In-zone fault) and ' −1 ' indicates second class of label (Out-of-zone fault). The ow chart of the SVM Classier algorithm is shown in Fig. 12. Cost (C) and Gamma ( γ ) are hyperparameters, which are set before the training model as given in SVM ow chart. Hyper-parameters are used to control error and also indicate curvature weight of the decision boundary respectively. When C is small, margin will be wide. So, there will be many support vectors and many mis-classied observations. When C is wide, margin will be small. So, there will be less support vectors and less mis-classied values. However low value of cost (C) will give better test data sets performance and also will prevent over tting. Accuracy of support vector machine learning algorithm is shown in Eq. (20) [7] and [20]. %Accuracy =Accurate classified samples Total no of samples ·100. (20) Table 8 indicates the internal and external faults detection and classication accuracy of the test data set which is not the part of the trained data set. Table 9 and Tab. 10 shows the accuracy of internal and external faults identication respectively in Python. Similarly tabulated accuracy of Tab. 8 has been veried in MATLAB and Python (SVM training and SVM model t functions) programming. 5. Wind Farm Impact on Transmission Line Protection The variation in wind parameter signicantly affects the distance measurement problems in transmission line protection. Fluctuation in wind speed causes variation in voltage level connected to power grid and this leads to the change in impedance measure by protective relays [2]. The impact of a 3-phase short circuit on the transmission line connected with DFIG is more © 2022 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 233