Fault diagnosis and life prediction of wind turbine based on site monitoring data

With the rapid increasing of total install capacity and operating time of wind turbine, the fatigue failures and the maintenance quantity are dramatically increased. It is urgently required to analyze the wind turbine condition timely and accurately to improve the reliability and reduce the maintenance frequency. So the research on reliability and residual lifetime predictive is proposed. At first, through SCADA system, the raw data is transmitted to the state database, on which the site monitoring data is analyzed, and some key parameters and the basic characteristics of site data are extracted. And then, the fault diagnosis of certain part of wind turbine is progressed by integrating possible site data characteristics. Since the fault of certain part is possibly induced by another part, the fault causal network is constructed in order to analyze the interaction of different part of wind turbine. The Fault Tree and Parsimonious Covering Theory are utilized to establish the fault causal network. After that, the Risk Priority Number Theory is utilized to assess the risk of different analysis conclusions obtained by the causal network. Finally, the possible residual life of wind turbine is studied by using accumulation theory and life prediction methods. The reliability of wind turbine could be improved by using the presented method. The rational arrangement of maintenance schedule and economy of management costs will also be improved.