水电能源科学

定子绕组是水轮发电机的主要发热部件，其温度变化直接关系到机组运行安全。传统水电站多采用温度超限报警方式进行监测，但该方式响应滞后，难以及时预警，进而影响电力系统运行稳定性。为此，以抽水蓄能电站水轮机定子绕组为例，通过分析定子绕组温升的影响因素，选取关键特征变量，并开展相关性分析，提出了结合SMOTE与ENN的新型组合采样方法，以提高数据采样的精度，同时通过多策略改进方式对HHO算法和SVM模型进行优化，建立了基于组合采样-CAHHO-LSSVM的定子绕组最高温度预测模型，最后，利用该模型对现场监测数据进行试验验证。结果表明，所提模型能够准确预测定子绕组的最高温度变化情况。

The stator winding is the main heat-generating component of a hydroelectric generator, and its temperature variation is directly related to the safe operation of the unit. Most traditional hydropower stations adopt the temperature over-limit alarm method for monitoring, but this method has a lagging response and is difficult to issue timely warnings, which in turn affects the stability of the power system operation. Therefore, the stator winding of the turbine in a pumped storage power station was taken as an example. By analyzing the influencing factors of the stator winding temperature rise, the key characteristic variables were selected and the correlation analysis was carried out. Then, an new combined sampling method combining SMOTE and ENN was proposed to improve the accuracy of data sampling. Meanwhile, the HHO algorithm and the SVM model were optimized through a multi-strategy improvement approach. A stator winding maximum temperature prediction model based on combined sampling-CAHHO-LSSVM was established. Finally, the model was used to validate the on-site monitoring data.The results show that the proposed model can accurately predict the maximum temperature change of the stator winding.