Research On Cell Series Multi-level High Voltage Inverter Fault Diagnosis Based On Fast-ICA And Support Vector Machines

With the rapid development of modern industry and science and technology, thehigh voltage inverter is widely used in industrial production. As the Importantequipment of industrial transmission control, The reliability and stability of high voltageinverter are taken more seriously. So current research in international automatic controlfield focus on fault diagnosis technology of high voltage inverter. How to establish ahighly efficient and accurate high voltage inverter fault diagnosis system is a veryimportant project.First,This paper aimed at problem on fault diagnosis of high voltage inverter, tookcell series multi-level high voltage inverter as the research object, established thesimulation model of IGBT open circuit fault in power cell of high voltage inverter basedon Simulink tool, then,researched the whole method in diagnosing IGBT open circuitfault of high voltage inverter.Second, the ICA algorithm of BSS has been studied, the Fast ICA algorithm hasbeen mainly discussed, and simulation experiment has been done on four random signalwith Fast ICA, and separation result is good. Then used Fast ICA to extract featurevalue from IGBT open circuit fault signal of high voltage inverter, and establishedSimulation platform to do experiments on it.Finally, studied SVM, and applied OVO SVMs algorithm to high voltage inverterIGBT open circuit fault diagnosis. used feature values which extracted by Fast ICAalgorithm as OVO SVM’s input eigenvector, trained OVO SVMs,and establishedcorresponding relationship between fault condition and eigenvector, and then finishedsimulation on it.After tested, combined with Fast ICA algorithm and SVM algorithm are moreaccurate and robustness in diagnosing the IGBT open circuit fault in power cell of highvoltage inverter, so it has good application prospect.