Towards Smart Solar Panel System Development: Case Study from Cleaning System Design and Algorithm Development Using Data-Driven Method
DOI:
https://doi.org/10.37798/2026752743Keywords:
Solar panel, cleaning, data-driven control algorithm, hand-engineered features, dark channel, datasetAbstract
Solar energy is one of the most widely used renewable energy resources. In general, solar power is generated by solar panels, but their efficiency depends on panel conditions, which are affected by environmental factors. For example, their performance can be severely hindered by the soiling of dust, snow, and organic dirt.
The listed vary across regions by climate and landform, and different sensing methodologies will be critical to the design of a solar panel cleaning system.
In this work, we used a data-driven approach to develop a control algorithm that accounts for the vastly different dirty-surface conditions of solar panels. As another essential design space is the budget, various feature development methods were considered for algorithm development. Two cleaning systems - a wiper-based and a rail-based system - were designed and used to evaluate the developed algorithms. With our experiment setup, the wiper-based system outperformed the rail-based system in terms of cleaning speed and power efficiency. The algorithms based on hand-engineered features could reduce the size of the algorithm and inference time while maintaining competitive performance. However, the algorithms with learned features showed consistently decent performance and development time.
Downloads
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Journal of Energy - Energija

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.






