Research on open-circuit fault diagnosis of three-level inverter

Abstract

In view of problems of complicated calculation and low accuracy existed in traditional open-circuit fault diagnosis methods of three-level inverter, an open-circuit fault diagnosis method of three-level inverter based on wavelet analysis and particle swarm optimization support vector machine (WT-PSO-SVM) was proposed. On the basis of analyzing the characteristics of the three-phase current signal of the three-level inverter, the current signal is decomposed by using the three-layer wavelet, and the energy of each frequency band is extracted as the fault feature. After the energy was extracted by wavelet transform, the extracted energy under partial faults is very close and cannot be distinguished effectively, and then the positive half-cycle proportional coefficient is introduced as auxiliary feature. The normalized energy and the positive half-cycle proportional coefficient are used as feature vectors to input support vector machines for classification training, and the parameters of support vector machine are optimized by particle swarm optimization algorithm to achieve the best classification effect, so as to realize fault diagnosis. The experimental results show that the WT-PSO-SVM method can effectively identify open-circuit faults of the three-level inverter, which has higher diagnostic accuracy and speed than other fault diagnosis methods, and still has a higher fault identification accuracy of 97.3% in the case of variable load and noise.

Keywords

• three-level inverter
• open-circuit fault
• fault diagnosis
• positive half-cycle proportionality coefficient
• wavelet transform
• support vector machine
• particle swarm optimization algorithm