Cloud Computing and Machine Learning-based Electrical Fault Detection in the PV System

Abstract

A dependable and sustainable method of generating electricity is the use of photovoltaic systems. Each solar panel loses 0.5%–1% of its efficiency annually. Environmental issues and electrical problems cause solar panels to degrade. Electrical faults should be diagnosed promptly and correctly to minimize damage to the panel. Machine learning has shown remarkable achievements in a variety of areas recently. The focus of the current study is on developing applications for pre-trained machine learning models that have been properly tuned. For the accurate classification of electrical faults in a photovoltaic array, a suitable algorithm is chosen and will be installed on a web server after training the dataset for various electrical problems in a photovoltaic array. By simulating the PV system in the MATLAB/Simulink environment under various operating situations, the data necessary for creating the algorithm are obtained. The experimental setting validates the proposed model with 100% training accuracy and 97.399% testing accuracy for a set of randomly divided data.