Machine Learning-Based Fault Detection in Solar Photovoltaic Panels Using Electrical and Environmental Parameters

Abstract

Solar PVs have become a significant means of energy generation in achieving clean and sustainable energy in the world. Various fault types, including partial shading, dust accumulation, open circuit faults, short circuit faults, degradation, hotspot formation, abnormality of sensors etc., however, it influences the performance and reliability of PV panels. The impact of these faults is a decreased efficiency of power generation, maintenance costs and a shorter lifetime of the solar PV systems. Hence, it is crucial to have correct and time on-scale fault detection for increasing PV installation safety, reliability and energy output. This paper deals with the use of machine learning methods for fault detection in solar PVs. The method proposed consists of data collection and pre-processing of the PV system with electrical parameters like voltage, current, power, irradiance and temperature data. Different machine learning models like Support Vector Machine, Random Forest, Decision Tree, Artificial Neural Network, Convolutional Neural Network etc. can be trained and tested for fault classification and detection after data cleaning and feature extraction. These models are assessed with the general indicators like accuracy, precision, recall, F1 score, confusion matrix, detection time etc. This research aims to bring a high accuracy intelligent, reliable, and efficient fault detection mechanism to help detect abnormal operating conditions within PV panels. The purpose of the study is to compare the machine learning models and find out which machine learning model is most effective in real-time fault diagnosis of SPS. The findings from this research could contribute to enhancing the energy efficiency, maintenance costs, and sustainability of solar energy generation systems.