Research On Abnormal Condition Monitoring And Critical Parameters Prediction Of Wind Turbines

Wind power industry has seen the rapid developments in the recent years. Meanwhile in order to improve the economic benefits of wind power better, there are higher requirements on the power generation process and operation. The abnormal condition monitoring, early failure and critical parameters prediction of wind turbines have already become the current research focus.Large scale wind turbines are equipped with a more complete monitoring and data acquisition system—Supervisory Control And Data Acquisition System. For the operating turbines, SCADA system can perform many tasks, such as remote monitoring, failures’ statistics and analysis, and so on. Generally SCADA system has recorded a large number of wind turbines’ operational status and fault information. However, how to use the large information to analyze the running state, improve operation and maintenance(O&M) and perform fault prognostics has important significance for wind turbines. In this paper, the relevant work will be carried out from two aspects—abnormal condition monitoring and critical parameters prediction of wind turbines based on the SCADA system recorded data.The specific research contents are as follows:Firstly, the SCADA parameter selection approach based on Grey Relational Analysis is studied. The operation and monitoring principle and a large number of monitoring items of SCADA system are summarized and investigated. Considering the correlation and uncertainty between the SCADA variables, the Grey Relational Analysis method is used to analyze and select the proper and relevant variables for the certain modeling and monitoring purpose. Thus it also could reduce the modeling difficulty and the training time.Secondly, the abnormal condition monitoring approach of wind turbines based on Support Vector Data Description(SVDD) is proposed. The kernel function parameter selection of SVDD is discussed. Furthermore, anomaly detection model based SVDD is built after selecting the proper subsystems or components monitoring parameters based on Grey Relational Analysis. Compared to the randomly selected parameters for gearboxsignal modeling, the proposed approach achieves more high reliability and accuracy of fault detection.Finally, the integrated prediction method for the short-term wind speed based on Ensemble Empirical Mode Decomposition(EEMD) and Auto-Regressive(AR) model is investigated. The EEMD method is employed to decompose the wind speed time-series into several intrinsic mode functions(IMFs), then each IMF component is modeled through Auto-Regressive(AR) model separately for prediction. Meanwhile, the weights are obtained by the least squares method and finally wind speed prediction can be realized.Compared with the single time-series AR modeling prediction method and EMD-AR modeling prediction method, case studies through different sampling intervals of wind speed signals verify the reliability and accuracy of the proposed EEMD-AR modeling prediction method.