一种基于分段二极管模型的自适应四层数字孪生用于光伏组件实时故障诊断与输出特性表征

Abstract With the rapid growth of photovoltaic (PV) power generation, accurate fault diagnosis and output characterization of PV modules have become crucial challenges. Existing PV models often lack real-time adaptability, limiting their effectiveness under complex and dynamic operational conditions. To address these limitations, this paper proposes a novel four-layer digital twin (DT) model architecture specifically designed for PV modules . The framework uniquely integrates a physics-informed segmented diode model with a hybrid Fungal Growth-Differential Evolution (FG-DE) optimization algorithm, enabling high-precision modeling and robust multi-source data fusion . Unlike traditional models, the proposed segmented diode model is dynamically adjusted to varying fault conditions – both mismatch and non-mismatch – providing detailed insights into fault mechanisms and output behavior. The virtual layer continuously synchronizes with physical module data, while the decision layer employs a hierarchical machine learning strategy for accurate, real-time fault classification and performance prediction. Experimental validation across four large-scale PV power stations demonstrates that the model achieves an average fault diagnosis accuracy of 98.49%, reflecting a 2.37% improvement over traditional methods. In terms of output characteristic representation, the proposed method attains an average root mean square error (RMSE) of 0.0555 A and a coefficient of determination R 2 of 0.9715, indicating a 70.10% reduction in error and a 10.88% improvement in goodness of fit , respectively. These results underscore the model’s significant potential for deployment in intelligent PV systems , offering a scalable, data-driven solution for real-time monitoring, diagnosis, and control.