PCA and CNN-based detection and classification of faults in distribution network with distributed energy resources

Distributed Energy Resources (DERs) have been preferred to fulfil load demand in the Distribution Network (DN) for the past few years because of advantages like power loss reduction, improvement in reliability and voltage regulation. Integrating DERs in the distribution network increases normal as well as fault current, alters existing protection coordination and introduces complex dynamics that necessitate modern fault detection and classification techniques to ensure reliable operation. This article proposes a hybrid approach combining Principal Component Analysis (PCA) and Convolutional Neural Networks (CNN) for Fault Detection and Classification (FDC) in Distribution Networks with DERs. PCA is employed for feature extraction that captures fault-related patterns while mitigating noise, and computational complexity and fault classification with high accuracy is performed by CNN, which leverages its deep learning capabilities. The proposed method is validated using a modified IEEE 9-Bus distribution network, while data was generated through PSCAD/EMTDC software with different cases. The developed Results demonstrate that the hybrid PCA and CNN framework-based scheme achieves superior fault detection sensitivity and classification accuracy compared to other existing methods, with robust performance under different conditions. This approach offers an efficient solution for enhancing the reliability and resilience of modern distribution networks.