机电工程技术2025，Vol.54Issue(22)：40-44,5.DOI:10.3969/j.issn.1009-9492.2025.22.008
水轮发电机组水导轴瓦温度SE注意力机制-CNN预测模型
SE Attention Mechanism Based CNN Prediction Model for Temperature of Guide Bearing in Hydroelectric Generator Set
摘要
Abstract
The temperature of the turbine guide bearing is an important parameter for measuring the operating status of the turbine,and its accurate prediction can provide a guarantee for the safety detection and fault prediction of the turbine.To solve the problem of predicting the temperature of the guide bearing of a hydroelectric generator unit,convolutional neural network(CNN)and SE(Squeeze and Excitation)attention mechanisms are integrated to construct an SE-CNN prediction model for the temperature of the guide bearing of the hydroelectric generator unit.CNN is used to extract local features from raw time series data,thereby capturing the dependency relationships in temperature data.Combined with SE attention mechanisms,the feature channels useful for the task can be enhanced and those not useful for the current task can be suppressed,so as to more accurately predict the temperature of the turbine's guide bearing.Based on actual power generation data from pumped storage power stations,the experimental results show that the proposed SE-CNN prediction model outperforms GRU,LSTM,and CNN prediction methods in MAE,MAPE,and RMSE evaluation indicators of 1.569 2,3.608 1%,and 2.028 6℃,respectively.This indicates that the proposed method has high prediction accuracy and robustness,providing an effective technical means for fault diagnosis and early warning of pumped storage power stations.关键词
水轮机/水导轴瓦温度/时序预测/CNN/SE注意力机制Key words
water turbine/temperature of water guide bearing shell/time series prediction/CNN/SE attention mechanisms分类
建筑与水利引用本文复制引用
刘轩,巩宇,吴昊,雷俊雄,熊江翱,王卓艺..水轮发电机组水导轴瓦温度SE注意力机制-CNN预测模型[J].机电工程技术,2025,54(22):40-44,5.基金项目
南方电网公司研发项目(022200KK52222006) （022200KK52222006）
