1.College of Artificial Intelligence and Automation, Hohai University, Nanjing 210098, China;2.School of Electrical and Power Engineering, Hohai University, Nanjing 210098, China;3.Department of Electrical and Electronic Engineering, The Hong Kong Polytechnic University, Hong Kong 999077, China

With the proposal of the “carbon peaking and carbon neutrality” goals, the penetration rate of photovoltaic power generation in the power grid continues to increase. However, photovoltaic power generation may be affected by various environmental factors. Among them, local obstruction caused by photovoltaic panel pollution is an important factor that causes power loss and affects the efficiency of photovoltaic power generation. In response to the traditional pollution detection relying on the construction of large datasets, and the problems of low forecasting accuracy and single data form in loss forecasting, a forecasting method of photovoltaic output loss based on image correction and reconstruction is proposed, which uses image correction and reconstruction to detect photovoltaic panel pollution and estimate power loss. This method first detects pollution through image correction and image reconstruction, and converts image data into text data. Then, features are extracted from the corrected and reconstructed image data. Finally, multi-modal feature data containing temporal information is constructed for loss forecasting. The test results show that the proposed method has improved performance compared with traditional methods.

This work is supported by National Natural Science Foundation of China (No. 62073121).