Abstract:

This paper proposes a method to identify different causes of commutation failures in the High-Voltage Direct Current Transmission (HVDC) system. By using the technique of wavelet multi -resolution analysis (MRA) ,the transient signals generated by the faults are decomposed into different resolution scales. Then , the features of each fault are extracted using the wavelet signal processing methods such as scale energy and scale entropy. Moreover, two auxiliary parameters are defined as the criteria for the identification , and four thresholds are set to distinguish the commutation failures from DC line faults. Simulated results indicate that there exist certain distribution regularities for the scale energy and scale entropy of transient signals with different HVDC faults , and that the criteria based on these regularities can make effective and reliable diagnosis of commutation failures.

Key words: HVDC system, commutation failure, wavelet transform, scale energy, scale entropy, fault diagnosis