Journal of System Simulation ›› 2017, Vol. 29 ›› Issue (7): 1447-1457.doi: 10.16182/j.issn1004731x.joss.201707008

Previous Articles     Next Articles

Multiple Faults Identification of Three-level Inverter

Shen Yanxia, Wu Juan, Zhao Zhipu, Ji Zhicheng   

  1. Institute of Electrical Automation, College of Internet of things Engineering, Jiangnan University, Wuxi 214122, China
  • Received:2015-09-07 Published:2020-06-01

Abstract: A three-level neutral point clamped (NPC) inverter was taken as an example, phase current and bridge voltage in the fault states of single switch open and several switches open at the same time were analyzed, and a method based on reconstructive phase space (RPS) and wavelet packet analysis was proposed to identify three-level inverter faults. Based on RPS method, totally different reconstructed current trajectories were obtained, which showed the features of the inverter under different fault conditions. With the help of image processing technology, all kinds of faults with different phase currents were identified. The wavelet packet analysis was adopted to extract the features of frequency-spectrum for different faults with the same phase currents and all typical faults were identified. Simulations and experiments show that the method can accurately extract the feature vectors for each fault condition and various failure modes can be identified.

Key words: three-level inverter, fault identification, phase current, bridge voltage, reconstructive phase space, wavelet packet analysis

CLC Number: