Abstract
Photovoltaic (PV) system, integral to the renewable energy landscape, occupies a pivotal role within the modern energy frameworks. With the rapid development of PV technology, timely and precise fault diagnosis of PV arrays has become particularly significant. However, the performance of fault diagnosis model is seriously affected by the data imbalance that usually exists in PV arrays. To address this, a novel fault detection and diagnosis method that integrates an improved slime mould algorithm (ISMA) with a CatBoost model based on polyloss function (PolyLoss-CatBoost, PolyCatBoost). Firstly, according to PV modeling, the multi-coupling fault features are extracted with fault type identification. Moreover, the PolyCatBoost fault diagnosis model is established for imbalance data, and the ISMA algorithm is utilized for optimizing the model parameters to achieve accurate compound faults diagnosis. Finally, through experiment comparisons, the imbalanced classification evaluation metrics, UAR, MCC, and Hamming loss, reach to 0.858, 0.817, and 0.026, respectively. It indicates that the proposed model performs exceptionally well under data imbalance conditions. Besides, the accuracy is close to 95% under noisy conditions, which verifies the superiority of the proposed method in terms of diagnosis accuracy, robustness and stability.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have