Condition Monitoring And Evaluation Of Wind Turbines Based On Operation Data

With the increasing awareness of environmental protection and the gradual decline of wind energy costs worldwide,the wind power industry has made great progress.The safety performance and economic cost of wind turbines restrict the further development of wind power generation.Due to the harsh natural environment and the frequent changes of the operating conditions of wind turbines,it is difficult to achieve good results in practical applications by using the threshold method to monitor the condition of wind turbines.Therefore,from the perspective of preventing the operation risk of wind turbines and reducing the operation and maintenance cost of wind power generation,it is of great significance to study the monitoring and evaluation methods of wind turbine operating conditions.In this dissertation,the wind turbine monitoring and evaluation method based on operational data is studied.The operating parameters of wind farm are analyzed in depth.The state parameter model based on Deep Belief Network(DBN)is established,and compared with the prediction model based on Back-propagation neural network(BPNN).The results show that the feature selection based on mutual information has an impact on the accuracy of the model.The DBN algorithm has better ability to process big data information,it also verifies that DBN algorithm has greater advantages for wind turbine modeling than BPNN algorithm.Based on the DBN state parameter model,this paper proposes a wind turbine evaluation method based on multi-models.The DBN model residual probability density is used to evaluate the operating state of wind turbines.The operating state of the wind turbines is reflected in the parameter data.The multi-model evaluation method can evaluate the overall operating state.The probability density analysis method is more objective than the method of setting the residual threshold.The actual case shows that compared with the misjudgment of the single model threshold evaluation method,the state evaluation method of the fusion multi-model can find the abnormal operation state earlier and realize the early warning of the failure.