Fault diagnosis for HVDC systems based on consensus filter and SVM

A fault diagnosis scheme for HVDC (High Voltage Direct Current Transmission) system is proposed based on the consensus filter and support vector machine theory. The measured DC voltages in the DC line can not be used to detect system falut on account of the random noise effect. Firstly, consensus filter is used to filter the volts d.c. measured by multisensors. Then a fault observer is constructed using the output of the consensus filter to detect system faults. In order to classify the detected faults, DC voltage signals are selected to be transformed by using S transformation method and then fault samples composed of effective features are extracted. On this basis, we establish SVM fault diagnosis models, and compare performance of different models. Simulation results show the efficientcy of the proposed method. © 2008 IEEE.