High-ratio distributed photovoltaic anomaly detection based on improved deep learning

Photovoltaic power plant firmware structure is complex, the operating conditions are complex and variable, and the acquired data often present complex nonlinear and correlated coupling characteristics. Based on this, this paper proposes a distributed photovoltaic anomaly detection method based on improved neural convolutional network, and utilizes Focal Loss to improve the loss function to improve the model performance. Spatial features are extracted from the original data and a nonlinear mapping relationship is established between the data and state labels to output the anomaly detection results. Real data analysis is used as an example to verify the feasibility and effectiveness of the proposed method.