Data Analysis And Early Warning Of Wind Farm Motor Operation State

Since the reform and opening up,China’s economy has been developing rapidly,and people’s living standards have been improving day by day.However,China is still in the state of resource consumption and environmental pollution,so it is urgent to develop clean renewable energy.As an emerging clean energy industry,wind power generation is of great significance to improve energy structure,promote ecological and environmental construction,and realize the sustainable use of energy.However,wind farms are generally located in remote and open areas with harsh environment.Coupled with the long-term operation of wind turbine equipment,the frequency of wind turbine failure is gradually increasing.The failure of wind turbines will reduce the power generation efficiency of wind farms,and even lead to downtime in serious cases,resulting in production accidents,which will bring huge risks and losses to the safety and benefits of wind farms.Under the background of current wind farm production,this thesis takes the monitoring data of wind farm operation status as the breakthrough point,combines with the historical data of SCADA,designs and realizes the online early warning system of wind farm equipment operation through data analysis,machine learning and other methods,carries out real-time monitoring on the measuring points of wind power equipment,and uses the historical data to predict,so as to achieve the purpose of early warning,Reduce the cost of operation and maintenance,improve the production efficiency of wind farm,and achieve a virtuous cycle.Taking the generator stator temperature as an example,this thesis completes the following work1.The Pearson correlation coefficient method is used to select the characteristic parameters of the generator stator temperature.The regression prediction model of the wind turbine stator temperature is established by using these parameters and the historical data of the generator stator temperature under normal working conditions,and the reasonable residual between the actual value and the predicted value of the generator stator temperature under stable operation is set.The characteristic parameters that need to be predicted are input into the trained regression prediction model,and the predicted value of stator temperature is compared with the actual value to get the degree of deviation from the normal state of the generator operation state.According to the set residual,the fault warning is decided.2.The actual data of wind field is used for modeling,and four methods are used: gradient boosting regression tree(GBRT),random forest(RF),support vector machine(SVM)and K nearest neighbor algorithm(KNN).The results show that the regression prediction model using gradient lifting regression tree method has the highest accuracy,with the mean absolute error(MAE)of 0.21 and root mean square error(RMSE)of 0.26.It has high accuracy,and can give fault early warning 4.17 hours in advance,which meets the experimental expectation.3.The online warning system is designed and implemented,including the functions of real-time monitoring and fault warning of wind farm equipment operation data.Users can view the operation of wind power equipment in real time on the web,and obtain warning information in the warning box.After using the system,the wind farm can greatly reduce the operation and maintenance costs and improve the production efficiency.