Investigation To Approaches Of Online Condition Assessment And Short-Term Reliability Prediction Of Wind Turbines

Influenced by harsh natural environment and random fluctuations of wind energy, wind turbines(WTs) suffer operation conditions with complicate variation characteristics. Because operation conditions influence on WT condition parameters significantly, traditional condition assessment approaches based on thresholding are so difficult to be used in practical applications. In order to improve the operational reliability and safety of WTs, the work focuses on operational condition assessment and short-term reliability prediction of WTs, and includes main contents as follows.First, correlation among WT monitoring parameters were investigated. According to analysis on distribution of the supervisory control and data acquisition(SCADA) data in the wind speed range, an approach of SCADA data selection was studied. Correlation coefficients of Pearson, Kendall, and Spearman, among WT parameters, were comparatively anlyzed. A comprehensive correlation index of WT condtion parameters was developed with using the three corelation coefficients. The comprehensive correlation index was used for correlation analysis of WT condition parameters, and the correlation characteristics amang WT condition parameters were presented through case study.Secondly, a model for abnormal detection of WT condition parameters was investigated. Quantification of influence of input parameters on target parameters was analyzed. An automatic parameter selection method was studied for the condition parameter prediction model. A back-propagation neural network(BPNN) was used for establishment of the sub-model of automatic parameter selection. A combination prediction method of WT condition parameters, based on least squares support vector machines(LS-SVM), radial basis function neural network(RBFNN) and BPNN, was also investigated and a sub-model for anomaly analysis of condition parameters was developed through the combination prediction method and information entropy of parameter residuals. A generalized model based on the both sub-models, are capable of being used for case study of anomaly WTs with accurate anomaly identification of WT condition parameters.Moreover, an operational condition assessment model of WTs considering operation conditions was developed. Through investigation to calculation methods of deterioration degree of WT condition parameters, a BPNN based operational condition assessment model was developed with using both the deterioration degree of condition parameters and the operation parameters. Structure parameters of the BPNN model were optimized, and the BPNN were compared with a radial basis function neural network(RBFNN) based model and a LS-SVM based model. Case study showed that the BPNN model with the deterioration degree as input parameters obtained assessment results satisfying conditions of operational WTs better and wasting time less than the other two neural network based models.Finally, a short-term reliability prediction model of WTs was developed. An initial outage probability model of WTs was investigated, and influences of data sampling rate and statistical range of state transition probability on outage probability was analyzed. Influences of operating condition of WTs on outage probability were analyzed, and time-varying state transition probability matrix was investigated. Considering both the operational condition of WTs and abnormal degree of condition parameters, the short-term reliability prediction model was established. Case study justifies that the short-term reliability prediction model with time-varying state transition probability, estimates more accurately the operational reliability of WTs than traditional reliability models of WTs.The work is an active exploration for research on smart operation and maintenance of WTs and possesses value in practice application to improve operation reliability and to reduce maintenance expenses of WTs. It will also be fundamental of furthermore research on and practice applications of security assessment models of power collection systems of wind farms.