基于SCADA数据特征提取的风电机组偏航齿轮箱故障诊断方法研究

Research on Diagnosis Method of Wind Turbine Yaw Gearbox Based on SCADA Data Feature Extraction

针对风力发电机组偏航系统故障处理难度大和危害严重等问题,开发出基于数据采集与监视控制(SCADA)数据的偏航齿轮箱神经网络诊断模型。利用ReliefF算法和核密度-均值法提取能反映出偏航齿轮箱运行工况的7个SCADA参数,并提取出6种故障特征指标作为神经网络诊断模型输入量,来诊断偏航齿轮箱的正常状态、磨损故障以及断齿故障共3种运行状态。结果表明:经神经网络诊断模型训练后的误差精度满足诊断要求,能准确诊断偏航齿轮箱故障。

To solve the serious problem in a wind turbine yaw system that is difficult to deal with, a neural network diagnosis model was proposed for the yaw gearbox of a wind turbine based on SACDA monitoring data. Seven SCADA characteristic parameters reflecting operation conditions of the yaw gearbox were extracted by ReliefF and kernel density-mean algorithm, while six fault characteristic indexes were extracted and taken as the input of the diagnosis model to identify the normal state, wear state and broken tooth state of the yaw gearbox. Results show that the training accuracy of the neural network model meets the requirements of diagnosis, which can be applied in the fault diagnosis of yaw gearboxes.