IoT based solar panel fault and maintenance detection using decision tree with light gradient boosting

Abstract

Solar PV (photovoltaic) technology has advanced greatly in recent years due to advantages such as renewability, environmental friendliness, simple maintenance, and dependability. Nevertheless, a number of PV faults may appear and result in degradation, a decrease in output power, or even a storm surge at different levels, depending on the outside working conditions and regular weather changes that might cause harm to the production, distribution, or setup, it is critical to monitor PVSs (PV systems) for their power generation efficiencies. IoT (Internet of Things) are evolving technologies that have been studied for enhanced fault detection and predictive analysis in the maintenance and environmental monitoring of solar power plants. This research work suggests a method based on MLTs (machine learning techniques) to analyze power data and predict faults for the maintenance of solar power plants. Input data from solar power plants consist of plant power generation and weather data which are first pre-processed and then trained using the suggested DT-LGB (Decision Trees with Light Gradient Boosting) algorithm to predict errors. The trained model was able to identify major/minor faults or anomalies present in input data. Conventionally these identifications require more effort in detection and maintenance. The results of this work showed that the suggested model obtained 8.74 MSEs ((Mean Square Errors), 2.96 RMSEs (Root Mean Square Errors), and R2 values of 0.9939 which is 12.8%, 6.8%, and 11.08% improved than the existing method respectively.