Evaluation And Prediction Of Transmission Chain State Of Wind Turbine Based On Temperature

Transmission chain is an important part of wind turbine to achieve the transition from wind energy to electric energy.Because of its variable,heavy load and high friction working environment,it belongs to the unit of high-risk areas,the operation and maintenance costs are relatively high.As a direct physical monitoring quantity,the temperature signal can reflect the deterioration information of the transmission parts.The existing monitoring system evaluates the state of the transmission chain by setting a fixed reference temperature for the parts,which is unable to accurately assess the status of the unit in real time,resulting in "sick operation" and "frequent shutdown" of the transmission chain.Aiming at this problem,on the basis of studying the recession trend and fault form of the health status of the transmission chain,three quantitative alarm thresholds are obtained by analyzing the relationship between the temperature monitoring value and the temperature reference value,to assess the operation status of the transmission chain accurately.The specific research contents are as follows:First of all,starting from the composition of the transmission chain of wind turbine,the working principle of the transmission chain is studied.Based on SCADA monitoring system and data cleaning,the monitoring data of the unit under normal operation are obtained,the monitoring quantity of the evaluation model is screened,and through the contrastive analysis of the correlation coefficient,choose working condition parameters.Secondly,to comprehensively consider the effect of different wind turbines operating conditions,endowed with different affecting factors the weight,set dynamic reference value of different temperature parameters,establish the comparison between temperature reference values and measured values of temperature.On the basis of this,quantitative indexes such as temperature residuals,residuals and temperature fluctuations are introduced,and a healthy state assessment model of transmission chain is established by analyzing the temperature field of the transmission chain of wind turbines.Finally,by comparing the BP neural network,RBF neural network,support vector machine(SVM),gray dimension GM model and Time Series model on the prediction accuracy of wind speed and temperature of the parts,analysis of absolute average error,average absolute percentage error,root mean square error and root mean square percentage error quantitative indicators,the EEMD + GM prediction model for wind speed parameters and EEMD + NARX forecasting model for the various parts temperature parameters are selected to get the forecast data,assess the health state of transmission chain.