Enhancing Fault Identification, Classification and Location Accuracy in Transmission Lines: A Support Vector Machine Approach with Positive Sequence Analysis

: This research paper presents a proposed system for fault identification, classification and location in transmission lines using a Support Vector Machine (SVM) - based technique in conjunction with a Positive Sequence Analyzer. The objective is to develop an accurate and reliable method for identifying, classifying and locating different fault types in transmission lines. The proposed system leverages the capabilities of SVMs in handling high - dimensional feature spaces and the fault signature extraction capabilities of the Positive Sequence Analyzer. Experimental evaluations are conducted to assess the performance and effectiveness of the proposed system, comparing it with existing fault identification and classification methods. The results demonstrate the superior performance and robustness of the SVM - based technique utilizing the Positive Sequence Analyzer, providing a valuable contribution to fault management and system reliability in transmission line networks.


INTRODUCTION
Power transmission systems must be able to accurately diagnose and classify faults to minimize interruption and restore power supply.Fault classification and location estimation are essential tasks in power transmission systems [1].Support vector machines (SVMs) have been successfully used for fault detection and classification in transmission lines, although these techniques have limitations, including high computational burden and memory requirements for classification, that limit their real-time implementation [2,1].SVM combined with feature extraction techniques, including wavelet transform (WT), synchro-squeezing transform (ST), principal component analysis (PCA), empirical mode decomposition (EMD), and Hilbert-Huang transform (HHT), has been widely used for fault detection and classification on transmission lines [2].In addition, positive sequence analyzer can be used for fault detection and classification in power systems as it extracts features related to the positive sequence component of line currents, which can be used to identify the fault zone [1,3].The combination of SVM-based technique and positive sequence analyzer can improve the accuracy of fault detection and classification in power systems [3].One proposed algorithm includes the use of PCA and SVMs for fault diagnosis in power system transmission lines, with SVMs used for fault classification and positive sequence analyzer used to determine the location of faults.This approach has been tested on a 300 km, 400 kV series compensated transmission line for all eleven types of faults through digital simulation, with promising results after testing on more than 800 fault cases with varying parameters [3,4].These techniques and algorithms demonstrate that SVM-based technique is a popular approach for fault detection and classification in power systems and can accurately detect and classify transmission line faults.

Background and Significance of the Study
Transmission lines are critical components of electrical power systems, and their protection is of utmost importance to ensure the stability and reliability of the power grid.One of the most common causes of transmission line failure is the occurrence of faults, which can cause a cascade of failures leading to power outages and damage to the system.Fault detection and classification are crucial for the prompt and effective protection of the transmission lines.[5] The traditional methods of fault detection and classification involve the use of time-domain and frequencydomain techniques, which are computationally expensive and often require significant computational resources.In recent years, machine learning-based techniques have gained popularity due to their ability to provide faster and more accurate fault detection and classification.In this study, we propose the use of the Support Vector Machine (SVM) method to develop a positive sequence analyzer for the protection of transmission lines.[6]

Objectives and Research Questions
The main objective of this research is to develop a positive sequence analyzer using SVM for the protection of transmission lines.The specific research questions are as follows: • Can SVM be used to accurately detect and classify faults in transmission lines?• What are the performance metrics of the proposed positive sequence analyzer using SVM in comparison to existing methods?• What are the limitations of the proposed method and how can they be addressed in future work?

Methodology Overview
The proposed methodology involves the following steps is shown in Fig. 1.
The proposed methodology will be validated through simulation results, and the performance of the proposed method will be compared with existing methods.

PROPOSED SYSTEM 2.1 Simulation Model
The proposed system simulation model is shown below Fig. 2.

Detection of External Fault
The prediction of external fault is illustrated in Fig. 3.

Detection Internal Fault
The proposed fault detection scheme can accurately distinguish different fault types.For the internal short-circuit fault, the proposed SVM-based detection scheme can identify the fault inception swiftly for any position along the protected line.During the internal fault, the fault breaker is located internal of long transmission line is shown in Fig. 4 2

.4 Type and Configuration
The single line diagram of proposed simulation model is illustrated in Fig. 5.The performance of the proposed method will be compared with existing methods.
The proposed methodology will be validated through simulation results.
Interpretation of results and discussion of implications.

Comparison of results with existing methods
Performance evaluation using metrics such as accuracy, precision, and recall

Monitoring and Measurement
In this experimental setup we have considered the various conditions of fault resistance and other parameters as described in table-IV.To identify types of fault like internal or external, we have first created the different types of fault models near to bus-I and measured the 3-phase voltage and current at bus -1 and Bus -2, also we have considered positive Three phase Voltage and current at bus 1 is measured by using three phase VI Measurement model and recorded readings for all condition mentioned in above Tab.4 this recorded readings are used to train the SVM to detect the fault and classify the type of fault, Similarly Three phase Voltage and current at bus 2 is measured by using three phase VI Measurement model and recorded readings for all condition mentioned to generate training data to train SVM Positive sequence voltage current is measured at bus-2 using a Sequence analyzer, training the data for detecting the type of faults for the transmission line.
By simulating all conditions mentioned in above Tab.4 we get training data with total 14 parameters shown in below Tab. 5.

Control and Protection Systems
In this implementation we have used two three phase breaker near to Transmission line-1 and Bus-2 respectively.The three phase breaker parameters are shown in below Tab. 6.In this project we have implemented a two series compensation devices one is connected near to transmission line 1 and second is connected near to transmission line 2.
Series compensation devices are utilized in transmission lines to enhance the efficiency and reliability of power transmission.Here are some of the benefits of using series compensation devices:

Communication and Data Acquisition
To implement fault detection, location, and classification in a long transmission line using an SVM-based technique and positive sequence analyzer, we follow below steps: • Data Acquisition: Collect data from the long transmission line using sensors or measurement devices placed at various points along the line.The data may include current and voltage measurements, which are crucial for fault detection and analysis.
• Pre-processing: Pre-process the acquired data to remove noise, normalize the values, and prepare it for further analysis.This step ensures that the data is suitable for training and testing the SVM classifier.• Feature Extraction: Extract relevant features from the pre-processed data that can be used to differentiate between normal and fault conditions.Features could include the magnitudes, angles, and harmonic content of the voltage and current signals.

RESULT FOR FAULT IDENTIFICATION AND CLASSIFICATION 3.1 Effect of Fault Resistance (Rf) on Positive Sequence Voltage and Current
Separating internal from external faults and classifying the fault as symmetrical or unsymmetrical fault is one of the key goals of this research project.This is done by considering the positive sequence voltage and current value at bus-2.
Using positive sequence current and voltage values is important for fault classification and detection in transmission lines because they provide valuable information about the system's behavior during a fault condition.Here are a few reasons why positive sequence quantities are preferred:

Result for Fault Identification and Classification
The classification of fault in both Internal and external condition for LG, LL, LLG, LLLG and LLL fault is shown in Tab.6 respectively.It shows that it will identify fault with good accuracy and precision compared with the Fuzzy system as shown in Tab.7 below for the same Transmission line configuration.

Result for Fault Location
The second aim of this research is to identify the location of fault from line-1 and line2 to resolve the fault within a short time for maintaining the continuity of supply.Fig. 10 show the result for fault location which is near about 99% accurate.

Effect of Line Length Variation
The line length has been altered to 50% of its nominal value, which is 200 km, to test the validity of the proposed design under various line length conditions.All type of fault has been simulated in line connected between Bus-1 and 3 with different ranges of fault resistance i.e.Rf = 0.01, 10, 20, 30, 40, 50, 60 Ω.Fig. 11 shows accuracy of the proposed system for detecting, classifying and locating the SLG, LL, LLG, LLLG and LLL faults.The average accuracy of proposed system is shown in below Fig. 13 it proves that proposed system is having near about 84% accuracy during line length variation.

Effect of Source Impedance(X/R Ratio) Variation
In some conditions the source impedance get varied, in this proposed system the X/R ratio get increased by double of given value i.e. to 20 ohm.Below Fig. 12  accuracy of proposed system during Source impedance variation from that we can say that proposed system has 96% accuracy in internal fault and 87% in external fault.
Figure 12 Average Accuracy of SVM system during X/R Ratio Variation

Effect of Source Power Variation
Below Fig. 13 shows the accuracy of the proposed system when source power varied from 1000 MW to 1500 MW.Result shows that the proposed system has overall 97% internal and 81% external fault accuracy.

Effect of Load Variation
Practically when the load on the system gets varied, the proposed system should have greater accuracy for fault identification and classification.Accuracy of proposed system during load variation is shown below Fig. 14; it shows that proposed system has greater accuracy during load variation.

COMPARISONS OF RESULTS WITH EXISTING METHODS
Support Vector Machines (SVM) is known for their excellent classification accuracy, especially when dealing with high-dimensional datasets and complex decision boundaries.SVM can handle both linear and non-linear classification problems effectively.
ANN systems are capable of handling imprecise and uncertain data.
In some cases, the accuracy of ANN systems may be slightly lower compared to SVM.
Below Tab. 8 shows the overall accuracy of both SVM and ANN system for proposed system.It is clear that SVM has greater accuracy than ANN system.

CONCLUSIONS
In this study, a novel technique for fault classification and defective phase identification is introduced, utilizing a single-ended mixed Support Vector Machine (SVM).The technique focuses on analyzing the approximation coefficients of current signals, which are exclusively measured at one end of the line.The proposed SVM-based approach offers several advantages, including the ability to identify faults in both primary and backup protection systems, covering up to 92% of the entire line length.Furthermore, the suggested SVM-based relay demonstrates remarkable performance with minimal training patterns.
The proposed method exhibits a high level of accuracy in locating various types of shunt faults, achieving a success rate of 92% across different fault locations.Extensive testing confirms the reliability and selectivity of the approach, providing satisfactory performance for three-phase transmission lines.Although the training process is performed offline, it should be noted that the training time for constructing the SVM network increases with larger training data sizes resulting from system configuration changes.
The effectiveness of the proposed scheme is demonstrated through successful detection and classification of different types of faults, including symmetrical and unsymmetrical faults, as well as unique cases involving High Impedance Faults (HIF), evolving faults, current transformer (CT) saturation, capacitive voltage transformer (CVT) transients, close-in faults, swing conditions, and source strength variations.A comparative analysis conducted against recently proposed techniques highlights the scheme's potential and robustness.

Figure 1
Figure 1 Methodology steps of proposed system

Figure 2
Figure 2 Simulation model for 400 kV 300 km double fed transmission line with SVM

Figure 3 Figure 4 Figure 5
Figure 3 Simulation output of transmission line during external fault [0.9337e-3 4.1264e-3] Capacitance per unit length (F/km) [N×N matrix] or [c1 c0 c0m]: [12.74e-9 7.751e-9] Training and testing the Fuzzy /SVM model Data preprocessing and feature selection Simulation setup and data collection using MATLAB software Development of the positive sequence analyzer using Fuzzy/SVM Description of the transmission line under consideration

1 )
sequence analyzer to increase the accuracy and precision of fault detection technique.By utilizing a positive sequence analyzer we have recorded the positive sequence voltage and current at Bus-2 by generating training data we are able to train the SVM for detecting internal faults.By simulating each type of fault condition model in Matlab and recording the following six values to identify both Internal and External fault.Voltage at Bus1 (V1) 2) Current at Bus1 (I1) 3) Voltage at Bus 2 (V2) 4) Current at Bus 2 (I2) 5) Positive Sequence Voltage 6) Positive sequence Current.

Figure 6 Figure 7 Figure 8 Figure 9
Symmetrical Faults: Positive sequence quantities represent the symmetrical component of the fault current and voltage.During a balanced fault, where the fault impedance is purely resistive, the fault current and voltage have a positive sequence component only.By analyzing the positive sequence values, it becomes easier to identify and classify symmetrical faults such as lineto-line and line-to-ground faults.• Simplified Analysis: Positive sequence analysis simplifies fault calculations by considering only the symmetrical component.This simplification reduces computational complexity and allows for efficient fault detection algorithms.• Fault Discrimination: Positive sequence quantities help in distinguishing between internal and external faults.Since internal faults predominantly affect the positive sequence values, they exhibit significant changes compared to the healthy system.On the other hand, external faults may cause minor perturbations in positive sequence values but have a more pronounced impact on negative and zero sequence components.Analyzing positive sequence values aids in distinguishing between different fault types and their location within the transmission line.• Fault Localization: Positive sequence information can be utilized to determine the location of the fault within the transmission line.By comparing the positive sequence voltage and current phasors at different locations along the line, engineers can estimate the fault position based on the phase angle and magnitude differences.• Protection System Design: The design of protective relaying systems relies on positive sequence quantities.Protective relays are responsible for detecting faults and isolating faulted sections of the transmission line.By focusing on positive sequence values, relays can make quick and accurate decisions, improving the selectivity and speed of fault detection.[15, 16] Overall, positive sequence current and voltage values play a crucial role in fault classification and detection in transmission lines due to their simplicity, discriminative power, and ability to aid in fault localization and protection system design.Below Figs. 6 and 7 shows a graph for behavior of positive sequence current and voltage for LG, LL, LLG, LLLG faults for internal condition and Figs. 8 and 9 shows a graph for behavior of positive sequence current and voltage for LG, LL, LLG, LLLG faults for external condition respectively.Variation of positive sequence current for Internal (a) SLG fault, (b) LL fault, (c) LLG fault, (d) LLLG fault and (e) LLL fault Variation of positive sequence voltage for Internal (a) SLG fault, (b) LL fault, (c) LLG fault, (d) LLLG fault and (e) LLL fault Variation of positive sequence current for external (a) SLG fault, (b) LL fault, (c) LLG fault, (d) LLLG fault and (e) LLL fault TECHNICAL JOURNAL 18, 2(2024), 183-190 Variation of positive sequence voltage for external (a) SLG fault, (b) LL fault, (c) LLG fault, (d) LLLG fault and (e) LLL fault

Figure 10
Fault location (a) from Line -1 is 100 km and Line-2 is 50 km and (b) from Line-1 is 200 km and Line-2 is 100 km

Figure 11
Figure 11 Average Accuracy of SVM system during Line Length Variation

Figure 13 Figure 14
Figure13 Average accuracy of SVM system during source power variation

Table 1
Specification of transmission line

Table 4
Fault scenarios detail

Table 5
Sample Data to train SVM

Table 6
Three phase breaker parameter

Table 7
Fault classification result result shows

Table 8
Accuracy of SVM and ANN technique for proposed system