Fault diagnosis of commutation failures in the HVDC system based on a method of wavelet energy statistics

This paper proposes a method to identify different causes leading to commutation failures in the HVDC system based on wavelet transform. By using the technique of wavelet multi-resolution analysis (MRA), the transient signals generated by faults are decomposed into different resolution levels, with the features of each extracted. Based on the wavelet energy statistics, two auxiliary parameters are defined as the criteria for the identification; four thresholds are then set to distinguish between different faults. Simulation results indicate that the proposed approach makes it possible to perform definite identification of commutation failures from DC line faults and normal operations. Furthermore, discrimination in AC short circuit faults and missing firing faults, which cause commutation failures, can also be made.