Intelligent cleanup scheme for soiled photovoltaic modules. (15th February 2023)

• Record Type: Journal Article
• Authors: Po-Ching Hwang, Humble; Ku, Cooper Cheng-Yuan; Chao-Yang Huang, Mason
• Is Part Of: Energy. Volume 265(2023)
• Publication Date: 2023-02-15
• Subjects: Photovoltaic cleaning policy -- Image processing -- Statistics -- Machine learning -- Soiling detection; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2022.126293
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: In recent years, solar energy systems have increased significantly worldwide. However, over time, the efficiency of photovoltaic (PV) systems is always affected primarily by soiling deposits on the surfaces of PV modules. The soiling deposits lower the intensity of the irradiation transmittance, and the performance of the PV system is also reduced. Therefore, cleaning PV modules is a very routine and critical task. To reduce the efficiency loss caused by soiling deposits and increase lifetime revenue as much as possible, we propose an intelligent method for monitoring soiling status with a statistical approach, an image processing (IP) scheme, and a machine learning (ML) algorithm. Based on the experimental result, the accuracy of our method is 98.39% which indicates that it classifies the soiling status of solar panels excellently. Therefore, we believe the proposed method can assist maintenance personnel in determining the near-optimal policy of cleaning schedules for PV systems. This also decreases power loss and saves labor and time for long-term maintenance. Highlights: An intelligent method for monitoring the soiling status of PV modules is proposed. A revenue function of PV modules is formulated to classify the soiling levels. The proposed method can help determine a superior cleanup schedule for PV modules.