基于多维度SCADA参数的风电机组异常辨识

Wind Turbine Anomaly Identification Based on Multi-dimensional SCADA Parameters

风电机组所处环境恶劣,导致风电机组易出现故障。利用数据采集与监控(supervisory control and data acquisition, SCADA)数据预测与评估风电机组整体性能,对风电机组维修与维护具有重要意义。因此,通过分析风电场SCADA系统的海量数据,提取表征机组退化信息的特征参数,通过自适应核主元分析(kernel principal component analysis, KPCA)算法建立基于多维度SCADA参数的风电机组状态监测与异常辨识模型。为了避免复杂工况对评估结果的影响,该模型引入一种工况划分方法。最后,通过某风电场SCADA数据对该模型进行实验验证,并与未进行工况划分的KPCA模型、进行工况划分的PCA模型进行对比。实验结果表明,该模型不但能够准确辨识风电机组的异常状态,且辨识结果更具可靠性。

Wind turbines are prone to failure due to their harsh environment. It is of great significance for the repair and maintenance of wind turbines to use the data of supervisory control and data acquisition(SCADA) to predict and evaluate the overall performance of wind turbines. Therefore, by analyzing the massive data of the wind farm SCADA system, the characteristic parameters characterizing the degradation information of wind turbines are extracted, and the wind turbine state monitoring and anomaly identification model based on the multi-dimensional SCADA parameters is established by the adaptive kernel principal component analysis(KPCA) algorithm in this paper. In order to avoid the influence of complex operating conditions of wind turbines on the evaluation results, a method of operating condition division is introduced into the model. Finally,the model is verified by SCADA data from a wind farm, and compared with KPCA model without operating condition division and PCA model with operating condition division. The experimental results show that the model can not only accurately identify the abnormal state of wind turbines, but also has more reliable results.