Operation Status Evaluation of Pumped Storage Units Based on CNN-GRU Attention

Based on deep learning methods such as Convolutional Neural Networks (CNN), Gated Recurrent Units (GRU), and Attention mechanism, this study aims to intelligently extract the intrinsic distribution patterns among operational data of power units. CNN can efficiently capture local features in data, GRU helps to handle long-term dependencies in time series data, and the Attention mechanism further improves the accuracy and interpretability of the model by assigning different weights to different data points. Through these advanced machine learning algorithms, a comprehensive evaluation model for the operating status of pumped storage units can be constructed to accurately predict the operating status of the units under different working conditions. To verify the effectiveness of the proposed model, the study will construct a comprehensive residual sequence by calculating the residuals between the predicted and actual values of each operating parameter. This sequence can comprehensively reflect the deviation of the unit under different operating conditions, providing a basis for judging the overall operating status of the unit. By verifying the monitoring data of the lower frame of Unit 3 of a pumped storage power station in Shandong, the accuracy and reliability of the evaluation model in practical applications are further verified, ensuring that the model can provide effective operational status evaluation results under complex actual working conditions.