School of Electronics and Control Engineering, North China Institute of Aerospace Engineering
China Suntien Green Energy Corporation Limited

发电机是风电系统中重要的核心部件,为了提高风电机组的稳定、高效运行,对风电机组发电机的故障预测十分必要。文章围绕风电系统发电机机侧轴承温度超限故障预测的问题,考虑到所采集的故障特征信号具有较大噪声的特点,引入自适应完备噪声经验模态分解(CEEMDAN)联合自适应小波阈值去噪的方法实现信号有效去噪,同时结合GA-BP神经网络建立故障预测模型。通过与BP神经网络、GA-BP神经网络对比预测指标、误差指标和预测效果图形,验证了所提算法可以获得较好的预测效果。误差指标和预测效果均有提升,对提前15d风电系统发电机故障预测的准确率达到了92.98%。

Generator is an important core component in wind power system,in order to improvethe stable and efficient operation of wind turbine, the fault prediction of wind turbine generator isnecessary. Focusing on the problem of generator machine-side bearing temperature overrun faultprediction in wind power system, this paper takes into account that the collected faultcharacteristic signal is characterized by large noise, introduces CEEMDAN joint adaptive waveletthreshold denoising method to realize effective denoising of the signal, and at the same timeestablishes a fault prediction model by combining GA -BP neural network. By comparing theprediction indexes, error indexes and prediction effect graphs with BP neural network and GA-BPneural network, it is verified that the proposed algorithm can obtain better prediction effect. Theerror index and prediction effect are improved, and the accuracy of the prediction of generatorfailure of wind power system days in advance reaches%.

河北省教育厅重点项目(ZD2022089);北华航天工业学院博士基金项目(BKY-2023-03);北华航天工业学院校重点项目(ZD-2022-03)