Research On Wind Turbine Impeller Fault Monitoring And Prediction Based On NSET And DDN

In recent years,with the intensification of the energy crisis and the enhancement of people's awareness of environmental protection,the wind energy,as a widely distributed and well-stored clean renewable energy,has gradually attracted extensive attention from the international community.Therefore,the wind power has become an important way of energy supply.However,due to the adverse operating environment,the wind power system has a high frequency of the fault,long downtime after the fault,and large maintenance costs.Therefore,the construction of the wind power system operation and maintenance has become an important focus for future development.As a main component in the process of wind power generation,the wind turbine is usually composed of the three parts: impeller,generator and tower.As an important part of the energy conversion of the wind turbine,the impeller has a decisive influence on the power generation efficiency of the entire wind turbine.The statistics show that the impeller fault is an important factor that causes wind turbine shutdown and reduces the power generation efficiency of the wind farms.The research of the operation and maintenance methods,such as monitoring and predicting the operation state of the wind turbine impeller,has important economic value and engineering application demand.(1)Based on the Nonlinear State Estimate Technology(NSET),the wind turbine impeller fault monitoring method was studied,and an improved NSET model for the impeller fault monitoring of the wind turbine was formed.Aiming at the problems of the method of determining the memory matrix based on fixed steps in the traditional NSET model,an improved NSET monitoring model was proposed.Firstly,the DBSCAN(Density Based Spatial Clustering of Application with Noise)algorithm was used to cluster the historical optimal working condition data,then each type of the data was selected based on the Mahalanobis distance,and finally the selection results of each type were merged as the final memory matrix.Based on the principle of Statistical Process Control(SPC),the upper and lower limits of alarms and warnings as well as the criteria of exception determination were determined.The experiment results show that the improved monitoring model has smaller memory matrix and higher fault monitoring accuracy.(2)The Deep Belief Network(DBN)was introduced into the impeller fault prediction of the wind turbine,and a new method of fault prediction combining DBN and Back Propagation(BP)neural network was proposed.Based on the advantages of DBN in feature extraction of the high-dimensional data,the intrinsic relation between the degraded data attributes of wind turbine impeller was abstracted and modeled,and the low-dimensional intrinsic representation of the high-dimensional data characteristics was obtained.Based on the advantages of BP neural network in time series prediction,it was used as the top decision layer of DBN to predict the change trend of the speed of the wind turbine blade(BLASPE1).The experiment results show that the prediction accuracy of the model is higher than that of the traditional prediction model.(3)The prototype system was designed and developed so that the wind turbine impeller fault monitoring and prediction model studied in this paper could be applied in the future.The prototype system mainly includes modules such as the system configuration,data management,model management,fault monitoring,fault prediction and the other modules.The system configuration module mainly realizes the initialization of each functional subsystem.The data management module mainly realizes the acquisition and preprocessing of the SCADA(Supervisory Control and Data Acquisition)monitoring data of the wind turbine impeller.The model management module mainly realizes the construction,update and preservation of the fault monitoring and prediction model.The fault monitoring module mainly realizes the realtime monitoring and analysis of the wind turbine impeller SCADA data.The fault prediction module mainly realizes the prediction of the changed trend of BLASPE1 in the wind turbine impeller degradation data obtained by monitoring.