Fault Diagnosis Of Permanent Magnet Direct-Drive Wind Turbine Converter

With the increasing capacity of direct-drive permanent magnet synchronous generator,its importance to the power grid has become increasingly evident so that it would directly affect the stability of grid operation and power quality.Therefore,it is significant to improve the reliability of wind turbine.The permanent magnet direct-drive wind power system uses back-to-back full power converter.As the main components and weak part of the permanent magnet direct-drive wind power system,the converter has a highest failure-rate.In order to keep the system running uninterrupted under converter faults,it is urgent to study the fault diagnosis technology of the converter.The topology of permanent magnet direct-drive wind power system is introduced.Based on the PWM mathematical model and the circuit topology after the converter fault,the operating characteristics of the system under the failure of a single power tube and two power pipes in the converter are analyzed theoretically.Under the PSCAD/EMTDC simulation environment,the permanent magnet direct-drive wind power system model is built up and simulation is done to validate.A fault diagnosis method based on current mean method is proposed.The mean value of the normalized current of the machine side and the network side is selected as the fault detection parameter of the converter.The power of the fault is determined according to the positive and negative current.This method can only diagnose the failure of a single power tube,so on the basis to improve.The mean value of the normalized current and the normalized current mean are chosen as the fault diagnosis variables of the converter.By comparing the diagnostic variables with the threshold,a single power tube fault and two power pipe faults in the converter are realized accurate diagnosis.The feasibility and reliability of the two fault diagnosis methods are verified by simulation.Considering the influence of system input,system running condition and control mode on the threshold selection,a fault diagnosis method based on current multi-scale entropy mean method is proposed.The fault diagnosis method uses local mean decomposition to decompose the current to obtain a series of product function components.The first three product function components with the highest correlation are selected to obtain the mean and standard of multi-scale entropy,and they are input to the back propagation neural network neural network.The faulty power tube is positioned by the output of the neural network.The feasibility and accuracy of the fault diagnosis method is verified by simulation.Aiming at the problem that the fault diagnosis method of converter based on current multi-scale entropy mean method for longer time,a fault diagnosis method based on current mean and harmonic distortion rate is proposed.The current signal is fourier transformed to calculate the total harmonic distortion of the current.The normalized current mean and total harmonic distortion are used as fault characteristics,and input to the radical basis function neural network.The faulty power tube is positioned by the output of the neural network.The simulation verifies the accuracy and rapidity of the fault diagnosis method.