Fault Prediction Of Wind Turbines Based On A SVM Optimized By The Sparrow Search Algorithm

In recent years,the field of wind energy develops rapidly.However,in the process of wind turbine operation and maintenance,various faults frequently occur.More than half of the faults occur in the wind turbine gearbox,and the gearbox is the component of the entire wind turbine system that takes the longest to break down and repair.In the event of a gearbox failure problem,will directly cause significant economic losses.Therefore,in order to avoid unnecessary losses and reduce the cost of wind turbine operation and maintenance,the gearboxes of wind turbines are used as the object of study in this paper.The study was carried out in terms of theoretical methods and system design,and a fault prediction model for support vector machines optimized by the sparrow search algorithm was constructed,the corresponding system was designed and developed.The main research contents of this paper are as follows:(1)A Sparrow Search Algorithm(SSA)combined with Support vector regression(SVR)is proposed.For the high-dimensional and complex nature of wind turbine data,SVR is selected to address large number of dimensions and data non-linearity.The choice of key parameters in SVR has a direct impact on the algorithm’s strengths and weaknesses,to address this feature,the parameters of the SVR are optimised using the sparrow search algorithm with high global search capability and fast search speed.The superiority of this method is verified by UCI data set experiment.(2)A gearbox oil temperature fault prediction model based on sparrow search algorithm optimized SVR was established,after grey relational analysis.There is a lot of redundant information in the wind turbine data extracted from Supervisory Control And Data Acquisition,which can lead to poor prediction model accuracy.To address this issue,the grey relational analysis method is proposed to analyse the data and extract the key attributes that are highly correlated with the oil temperature of the wind turbine gearbox.Then,Combined the sparrow search algorithm to optimise the SVR method to construct a fault prediction model for gearbox oil temperature prediction.And compared with other models.The experimental results fully show that the proposed model has higher accuracy and better generalization ability,can be effectively applied to accurately predict the actual wind field of the wind turbine.The validity of the proposed model in practical problems is verified.(3)Design and develop an oil temperature prediction system for wind turbine gearbox based on SVR optimized by sparrow search algorithm.According to the model established above,The wind turbine data from actual wind farms were collected.And a wind turbine gearbox oil temperature prediction system based on SVR optimized by the sparrow search algorithm was designed and developed.It can realize the selection and uploading of data set,grey relational analysis of data and data processing,as well as the model presentation,the graph of the model prediction results and the corresponding indicators;The results are compared with SVR model in the system.The effectiveness of the gearbox oil temperature prediction system of wind turbine based on SSA-SVR is demonstrated,and a new technology application is provided for the engineering practice of wind turbine.The sparrow search algorithm which can overcome the local optimization was used to solve the problem that support vector regression is difficult to find the optimal parameters.The data were analyzed by using grey relational analysis method,and a gearbox oil temperature fault prediction model was built based on sparrow search algorithm to optimize SVR,which overcame the complex problem of wind turbine data redundancy in practical wind field application.On this basis,a gearbox oil temperature fault prediction system based on a sparrow search algorithm optimized for SVR has been developed.Providing a wind turbine fault prediction system with simple operation and high accuracy for practical engineering practice with practical engineering significance.