A Novel Artificial Neural Network-Based Method for Fault Diagnosis Considering Unbalanced Loading and Renewable Energy Resources

Modern electrical power systems are increasingly relying on interconnection with renewable energy sources (RES), resulting in larger, more complex networks. These modifications introduce new challenges in fault identification, especially given the variability associated with RES. Traditional fault detection techniques struggle to handle these environments, prompting the exploration of intelligent approaches. Furthermore, unbalanced loading conditions complicate protection management, as they introduce additional challenges for accurate, timely fault diagnosis. To overcome these challenges and achieve reliable fault detection, classification, and location with fewer measurements, this study proposes a novel artificial neural network (ANN) framework. To evaluate the performance of the proposed work, the ANN model is trained and validated on the IEEE 13-Bus and 9-Bus systems, with evaluations conducted under various fault types and locations, including both with and without RE integration, as well as unbalanced load scenarios. Moreover, the proposed model relies solely on current-based data, eliminating the need for voltage-based data, thereby significantly simplifying data collection while preserving high accuracy. The performance of the proposed technique is compared with that of an adaptive neuro-fuzzy inference system (ANFIS). The findings show that the proposed ANN can reliably detect, locate, and classify faults in both balanced and unbalanced systems, even in the presence of RES. The method is confirmed to be data-efficient, reliable, and generalizable. Compared to ANFIS and conventional ANN methods, this ANN model offers comparable performance with higher test accuracy while requiring simpler implementation and lower costs.