Temperature Trend Monitoring And Warning For Pitch Motor Based On MDI-ANFIS

Wind turbine often work in harsh areas,often with strong winds,sudden gusts and other conditions,and the main function of the pitch system is to capture wind energy and maximize the use of wind energy.The environmental impact on the system is even more pronounced.Pitch system as an important part of the wind turbine,whether the normal operation can directly affect the unit's safe and stable operation.Therefore,we will formulate an effective warning plan for the fault of the pitch system.It can not only timely and accurately predict fault information,avoid serious faults,but also ensure the economic efficiency of wind turbines.In this thesis,after deeply studying the structure of the wind power pitching system and the causes of the failure,taking the temperature of the pitch motor as an example,a temperature trend warning program of the wind turbine pitch motor based on MDI-ANFIS is proposed.The program is based on the historical operating data of wind turbine.First,the characteristic parameters of the temperature of the pitch motor are found by feature selection method.Then the data is modeled and the normal curve of the pitch motor temperature is fitted.Finally,the average value and the standard deviation of the pitch motor temperature are used to determine the abnormal state of the pitch motor.The specific research content of the paper mainly includes(1)This thesis introduces the functional structure of wind power pitch system and the types of common faults.The reasons for the temperature fault of the pitch motor are analyzed.After analyzing the SCADA data of the wind turbine,the monitoring parameters of the pitching system are selected,and then the characteristic parameters of the temperature of the pitching motor are found by using the RRelifF algorithm.(2)In view of the characteristic parameters of the pitch motor temperature,there are both numerical data and categorical data.An adaptive neuro-fuzzy inference system for mixed-type data input is proposed.Based on the adaptive neuro-fuzzy inference structure of numerical data input,the influence of the categorical data on the inference results is loaded onto the premise and consequent part of the fuzzy rules by introducing the Firing-strength Transform Matrix(FTM)and the Consequent Influence Matrix(CIM).Therefore,the system can handle the mixture type input.(3)Using the temperature model of the pitch motor established by the adaptive fuzzy inference system for the mixture data input,the temperature deviation of the actual pitch motor is found,and the respective confidence intervals are found by calculating the window mean and standard deviation of the bias sequence.An abnormal warning is issued when the boundary of the confidence interval of the actual window mean or standard deviation of the pitch motor temperature exceeds its respective threshold.