Fault Classification for High Impedance Faults Using Back Propagation Based Artificial Neural Network Technique

This paper presents an Artificial Neural Networks (ANN)-based fault analysis approach for a power distribution system which is an IEEE-9 bus system. It is designed for specifically targeting the identification and classification of high impedance faults (HIFs). By training ANN s on historical fault data, these models can achieve high accuracy in distinguishing HIFs from non-fault events. This results in enhancing the reliability of fault classification in complex, real-world distribution environments. The study evaluates the effectiveness of various ANN architectures in HIF classification, considering factors like accuracy, training time, and computational efficiency. The root mean square error (RMSE) algorithm is used for training and testing to test the effectiveness of the technique for detection, classification, and location of fault. From results, it is seen the proposed ANN model improve fault detection and classification capabilities, reducing false positives and improving response times compared to traditional methods like fuzzy-based methods. This improvement enables quicker identification and localization of faults, supporting operational reliability and paving the way for predictive maintenance and real-time monitoring in smart grid systems.