Research On Fault Diagnosis Of Converter In Wind Energy Conversion Systems Based On Data-Dirven

With the rapid development of wind power industry, the reliability and stabilityrequirements of all components have become higher. The PWM converter is taken as theinterface of the power generation system and the grid. It is controlled to maintain the stabilityof the DC link voltage for achieving the maximum utilization rate of wind power, and is alsoone of the most easily faulted components. Once any fault occurs, it would reduce the qualityof power supply, which threats the safety and quality of the grid. Therefore, locating anddiagnosing the converter faults are of great significance to improve the security and reliabilityand reduce the cost of system.In this paper, firstly, the PWM converter’s characteristic, work principle are introduced,and the classification of IGBT open circuit fault is given. The output voltage features of PWMconverter in rectifier mode are analyzed and the fault effects are discussed. Based onsimulation experiment, the accuracy of fault classification is verified.Secondly, take advantage of the wavelet packet analysis, with the processing of outputvoltage, the different fault features are obtained and the preliminary identification of the faultis realized. The simulation results show the effectiveness of this method.Thirdly, for the inverter state, fault influence on the DC current is discussed. Using thewavelet packet transform, the fault feature information of different fault types are obtained,and with the help of support vector machine, fault information is classified, and IGBT opencircuit fault diagnosis is realized. The simulation results show the accuracy of this method.Finally, when the IGBT open circuit fault occurs, in order to make the system to restorethe normal operation quickly, a fault tolerant topology is presented according to the structurecharacteristics of the back-to-back PWM converter. Simulation analysis verifies therationality of this fault tolerant strategy.