A new machine learning model has been proposed for identifying the source of pollution on photovoltaic panels. The model, which is based on stacking ensemble classifiers, aims to improve the maintenance and management of solar panels by determining the cause of power output decrease. Pollution on the surface of the solar panels can result in reduced solar radiation and PV efficiency, which can decrease the performance of the panel and potentially void the warranty. It is therefore crucial to identify the source of pollution on the panels and implement regular cleaning to maximize output and maintain the warranty.
Solar power is a promising renewable energy source that has the potential to advance sustainable growth. The utilization of solar power ranks among the largest renewable natural energy sources in the world and can be harnessed for distribution beyond the demands of local consumption. Furthermore, solar power does not require additional costs, is clean, and does not pose a risk of contamination. However, there are adjustable and unadjustable aspects that can affect a PV module’s efficiency, one of which is pollution on the surface of the solar panel.
To demonstrate the efficiency of PV systems and create cost-effective mitigation, soil impact assessments were recommended at different locations and times. This study proposes using machine learning for pollution source classification on PV panels. The model considers various weather features and uses metrics such as accuracy, precision, and F1 score to evaluate the performance of the model compared to state-of-the-art machine learning models.
The case study involved six solar panels, each exposed to different sources of pollution, and their power generation was recorded. The proposed model’s performance was tested and evaluated based on the collected data. The results showed that the stacking ensemble classifiers-based machine learning model outperformed other state-of-the-art machine learning models in identifying the source of pollution on the PV panels.
In conclusion, the proposed machine learning model has significant potential for improving the maintenance and management of PV panels. By identifying the source of pollution on the panels, solar panel owners and maintenance staff can implement regular cleaning to maximize output, maintain the warranty, and prevent permanent module staining. Future research on the application of the model to real-world scenarios can further improve the proposed model’s effectiveness, leading to more efficient and sustainable use of renewable energy sources.
