Enhanced Fault Detection using VGG-16 and Temporal Convolutional Networks for SPV Integrated Active Distribution Network

Abstract

This study investigated that an efficient fault detection is essential in distribution systems to ensure the dependability and stability of electrical networks, especially when including renewable energy sources like solar PV.This study presents a Fault detection approach for the IEEE-33 bus system that incorporates with Solar PV. The method utilises sophisticated deep learning models, notably VGG-16 paired with Temporal Convolutional Networks (TCN). Study assess the efficacy of several models, such as VGG-16 and TCN, Hybrid CNN-LSTM, Bi-LSTM, and ANN, across numerous fault categories. The findings of this study indicate that the VGG-16 & TCN model surpasses the other designs, obtaining an outstanding accuracy of 99.8%, along with excellent precision, recall, and F1-score. The examination of the confusion matrix reveals that both VGG-16 and TCN exhibit a high level of accuracy in classifying fault types, with just a few instances of misclassification. Furthermore, the ROC curve analysis substantiates the exceptional efficacy of VGG-16 and TCN, as shown by their ROC value 99, surpassing that of other models. The exceptional performance may be ascribed to the strong feature extraction capabilities of VGG-16 and the efficient processing of sequential data by TCN. The study's findings indicate that the VGG-16 & TCN model is the most efficient for fault identification.