Study on token parameters for diagnosing fault of oil-cooled transformers based on statistical technique

When diagnosing inner faults in power transformer, the effects of different fault gases are different. So the diagnosing efficiency of the same diagnosis technique based on different token parameters is different. The paper proposes a new token parameter for diagnosing fault of oil-cooled transformers based statistical technique, the weight gases concentrations radios (WGCR), different from the traditional radios. In this paper, diagnosis technique based on the multilevel weight fuzzy membership degree is proposed firstly. According to this diagnosis technique and statistical technique, the significance degrees of the gases for diagnosing faults are decided. Using fuzzy classification technique based on WGCR and traditional radios, the efficiency of two types of token parameters are compared. Test results show that WGCR is better.