Machine Learning-Based Intelligent Fault Diagnosis and Signal Forecasting for Photovoltaic Systems Using Performance Indicators

Abstract

Reliable fault diagnosis and predictive monitoring are essential for improving the performance, reliability, and operational safety of photovoltaic (PV) systems. This paper presents an intelligent framework for fault diagnosis and signal prediction based on performance indicators and machine learning techniques. First, a set of representative electrical and performance indicators is extracted from the PV system under normal and faulty operating conditions. These indicators are then used to identify different fault scenarios and to predict the evolution of key electrical signals, enabling early detection of abnormal behavior and supporting predictive maintenance. Several machine learning algorithms are investigated and evaluated according to their diagnostic accuracy, prediction capability, and computational efficiency. The proposed approach demonstrates high fault classification accuracy and reliable signal prediction under varying environmental conditions, while reducing false alarms and improving the robustness of the monitoring process. The results confirm that the integration of performance indicators with machine learning provides an effective solution for intelligent condition monitoring, fault diagnosis, and predictive maintenance of photovoltaic systems.