To better understand the effect of insulation oil on the vibration signals of on-load tap changer（OLTC） during the switch-over process

vibration signals of OLTC for a 35 kV power transformer are measured in three different states as follows: the insulation oil exists in both transformer tank and chamber for OLTC; the insulation oil exists in transformer tank and no oil exists in chamber for OLTC; the insulation oil exists in neither transformer tank nor chamber for OLTC. The tunable Q-factor wavelet transform（TQWT） is applied to decompose the measured vibration signals of OLTC into several time series with different complexity. The approximate entropy of each time series is calculated with the cubic spline interpolation algorithm. Calculated results show that the approximate entropy can be adopted to describe the complexity of each time series resulted from the TQWT. The complexity of vibration signals of OLTC are decreased by the damping of insulation oil in chamber for OLTC. The effect of insulation oil in transformer tank on the vibration signals of OLTC is limited and presents a feature of randomness. The results is of importance for the vibration monitoring technology of mechanical condition of OLTC.