Inverter Circuit Fault Diagnosis Based On Fractional Fourier Signal Processing And Pattern Recognition

With the rapid development of power electronic technology and the complexity of powerelectronic equipment, research about power electronic circuit fault diagnosis has become more andmore important. As a typical power electronic circuit, inverter circuit has always attracted extensiveattention from researchers. How to implement the testing and diagnosis of the inverter circuit, at thelevel of components based on signal processing and pattern recognition, is studied in this paper.Firstly, this paper introduces the current research situation and difficulties of inverter circuit faultdiagnosis. Then the simulation of a three-phase brushless DC motor inverter circuit is built withMATLAB. The signals of the output phase current in normal and fault modes are collected andfractional Fourier transform(FRFT) is used to extract the features that are suitable for the subsequentfault classification, compared with other ways like Fourier transform and wavelet decomposition. Andthe support vector machine(SVM) is utilized to identify the inverter circuit fault. With the study ofcommonly used multiclass classifier, the choice of kernel function and optimizing the parameters withgenetic algorithm are illustrated.To testify the practical validity of the feature extraction method based on fractional Fouriertransform, a brushless direct current motor control system based on STM8S903K3microcontroller isbuilt in this paper. The fault mode of the inverter circuit is set with the software and the relevantoutput phase current signals are collected. Then feature extraction and fault identification areconducted. The simulation and practical results show that the inverter circuit fault feature extractionmethod based on fractional Fourier transform is valid and even better than others. Meanwhile, theexperimental data and results prove the inverter circuit fault identification based on support vectormachine is feasible.