Data Resampling Techniques for Improved Fault Detection in Power Transmission Systems Using Artificial Neural Network Multiplayer Perceptron and Support Vector Machines

This paper introduces the effect of class imbalance using the data resampling synthetic minority over-sampling technique (SMOTE) and edited nearest neighbors (ENN) on artificial neural network multiplayer perceptron (ANN-MLP) and support vector machines (SVM), which are used for fault detection methods for electrical power transmission lines. The method effectively identifies various fault types, including line-to-ground, line-to-line, and multi-phase faults, significantly improving power system reliability and safety. The proposed artificial neural network multiplayer perceptron (ANN-MLP) and support vector machines (SVM) models using resampling techniques achieve exceptional accuracy of 99.7 and 95%, respectively, showcasing their superior performance compared to existing techniques. This research demonstrates the significant potential of machine learning in enhancing fault detection accuracy and mitigating power system disruptions. To evaluate the method's versatility, different faults are simulated using a range of parameters. This demonstrates the method's adaptability and potential for extension to distribution networks within power systems. All simulations and signal analyses are performed using the Python environment, ensuring a thorough assessment of the method's effectiveness across various fault conditions.