部分遮阴下最优功率采集：基于二进制灰雁优化的光伏阵列重构与基于机器学习的故障诊断

Abstract Photovoltaic (PV) systems have become a major source of energy harvesting assistance to electric grids, serving as a sustainable substitute for conventional energy sources. However, the effect of partial shading on PV reduces the effectiveness of PV-based electricity. The PV array reconfiguration approach is one of the best practices for lessening the influence of the partial shading effect. This article proposes a new PV array reconfiguration process using a Binary Greylag Goose Optimization (BGGO) approach. A 9x9 panel PV array with six shadow configurations for arrays − bottom right, top right, bottom left, top left, centre, and diagonal shading is considered to validate the effectiveness of the proposed BGGO. The proposed method achieves optimal global maximum power (GMP) in Total Cross Tied (TCT) arrangements by 14 % in the shadow pattern on the top left, 13 % in the arrangement on the top right, and 7 % in the diagonal patterns. The proposed method provides the highest fill factor and the least power loss at a maximum output of 5990 W as compared with TCT (5462 W), Modified Suduko (5822 W), Multi-Objective Gray Wolf Optimizer (5850 W) and Binary Firefly Algorithm (5801 W). The proposed BGGO technique produces 10 % more power than the TCT arrangement and other approaches, confirmed by the energy predictions and revenue generation. The Nave Bayes-based Machine Learning (ML) approach is also utilized to detect and categorize PV panel degradation. Other strategies are examined under both faulty and non-faulty to verify the performance of the proposed ML approach. The obtained results validate the proposed BGGO with ML’s superiority and capacity to reconfigure the shaded array to be optimal.