Improving the diagnosis of partial shading faults by utilising artificial neural networks optimised with the whale optimisation algorithm

Abstract

This paper introduces a hybrid approach combining an artificial neural network (ANN) with the whale optimisation algorithm (WOA) to diagnose partial shading in photovoltaic (PV) systems. It features two WOA-ANN models: WOA-ANN-classification for detecting and classifying PV array states as normal or partially shaded, and WOA-ANN-localisation for pinpointing the shading location. The WOA was compared with other algorithms like grey wolf optimisation (GWO), particle swarm optimisation (PSO), and differential evolution (DE). The ANN was trained using metrics such as mean square error, CPU time, and training accuracy. Experimental results showed the WOA-ANN models outperformed others, with the classification model achieving 99.99% accuracy and the location model 99.96% accuracy. This hybrid methodology significantly enhances fault diagnosis accuracy in PV systems, supporting sustainable energy efficiency.