Research On Diagnosis Technology Of Ice Accretion On Wind Turbine Blades Based On BP Neural Network

Wind turbines installed in wet and cold areas in winter are prone to blade icing under low temperature weather conditions.At present,the existing detection and diagnosis methods and techniques of wind turbine blade icing state have certain limitations,and it is difficult to accurately and effectively detect and diagnose the actual icing state of blade.Therefore,in this paper,we adopt the research method of combining theoretical analysis with engineering data processing to carry out the research on detection and diagnosis method and technology of blade icing status of wind turbine,which aims to provide new and more effective technology for real-time monitoring of blade ice faults at the engineering site.Firstly,under the working principle of wind turbines and the principle of meteorology,the influencing factors and variation characteristics of blade icing are analyzed,and wind speed,power and pitch angle which can reflect blade icing state,are selected as state characterization parameters.By using the method of nuclear density-mean,the data of normal state and icing state have been processed based on the MATLAB software platform,by using the method of normalization and mean,the benchmark values of icing states under different wind speeds are obtained,and the quantitative expression was obtained by data fitting.On the basis of the above research,the degree of blade icing is divided into four categories,and the quantitative index intervals of these four degrees are given,the classification diagnostic criterion of blade icing state was put forward,and the quantitative interval of classification diagnosis was established.Then,On the basis of the conclusions and SCADA monitoring data,five characteristic indices of icing state have been put forward.A neural network diagnosis model of ice accretion on blade is constructed based on the MATLAB software platform.By combining the data in typical condition and actual condition,a data sample set have been set up.Test results show that the maximum relative error of the model prediction results is 5.401%.Finally,a wind turbine blade icing condition monitoring and diagnosis system software has been designed and developed based on Lab VIEW software development platform,which integrates data display,analysis,diagnosis and results preservation,the accuracy of the system is verified according to the actual SCADA data.The data of four different icing states are selected,verification results show that the monitoring and diagnosis system can diagnose the icing status of wind turbine blades accurately,which has present application value in engineering.