Research On Evaluation And Prediction Of Operation Status For Wind Turbine Pitch System Based On Scada Data

As a clean and pollution-free green energy,wind energy has become the main force in the use of renewable energy.However,the operation and maintenance costs as high as 25%to 30%have seriously restricted the development of wind power industry.Meanwhile,with the rapid development of wind power technology in the last 10 years in china,the SCADA systems of large wind farms have accumulated a great deal of historical data,and the wind turbines in operation also produce real-time data all the time.These data contain a wealth of information which related to the operation of wind turbine and the condition of equipment.How to make use of these idle SCADA data to evaluate and predict the operation status of the wind turbine undoubtedly has important practical significance and academic value.Therefore,this article makes the research on the issue,which is summarized as follows:1.The research on state feature parameters extraction method of pitch system based on SCADA data.On the basis of deep research of the structure and fault mechanism of pitch system,combined with a wind farm integrated monitoring system,ReliefF algorithm and mutual information technology are used to mine the feature parameters related to the operation status of the pitch system in SCADA data.Qualitative and quantitative comparative analysis indicates that:compared with mutual information technology and the method without data preprocessing,ReliefF algorithm has faster training speed and higher accuracy of fault classification.2.Based on the pitch system preprocessing data extracted by ReliefF algorithm,this paper discusses the optimization of SVM model and its application in the evaluation and prediction of pitch system.First,cross-validation and grid search method are adopted to optimize the model parameters.Then,the SVR regression model is established by using the power output as the decision parameter of the model.For random disturbance factors,the residual trend of the prediction model is analyzed by using the sliding window residual estimation method,and then the operation status of the variable pitch system is evaluated.Compared with the neural network,the experimental results show that the SVR model built in this paper has better ability of status evaluation and prediction.3.Research on status evaluation and prediction model of pitch system based on SWPSO-SVR.Because the CV parameter optimization method has the disadvantages of large calculation,low efficiency and low precision,the SVR model is not inconvenient to practical application because of the large amount of data especially in the case of large amounts of data.Therefore,this paper further discusses the optimization of SVR parameters,and proposes an improved SWPSO-SVR algorithm.The SWPSO-SVR algorithm has the advantages of high precision,fast convergence speed and not easy to fall into local minimum compared with the CV-SVR algorithm.Combined with the test and analysis of SCADA data of a wind farm,the effectiveness and practicability of the improved model for status evaluation and prediction of pitch system are verified.4.Research on engineering application system of monitoring operation method of pitch system.Using GUI in matlab software to develop an interactive platform which can be connected with integrated monitoring system,this can expand the function of fault prediction and alarm of pitch system,and provide decision support for the safety operation and maintenance of pitch system.In summary,the SCADA data of wind farms contain rich operational status information.By analyzing the changing rules of the data,the health status of the pitch system can be evaluated and predicted.The research of this paper is economical and practical,and can realize the early warning of the pitch system on line.