Abstract: In order to seek for the effective method to extract the mechanical features of on-load tap-changer（OLTC） of converter transformer, the combination of an optimized Hilbert-Huang Transform（HHT） algorithm and Lorentz Information Measure was presented to analyze the vibration signals during contacts switch. First, an optimized masking signal obtained by Ant Colony Optimization（ACO） was added to restrict the HHT bandwidth, which has obvious superiority in aliasing reduction and decomposition accuracy of Empirical Mode Decomposition（EMD）. Then the 3D surface similarity index based on Lorentz Information Measure according to the Hilbert energy spectrums of vibration signals was defined to assess the mechanical condition of OLTC. The calculated results under normal and typical fault conditions of R type OLTC model have shown that, the energy spectrums of different conditions vary significantly so that the similarity index can measure the difference degree of energy distribution. Meanwhile the index of contact looseness is higher than the insulated panel looseness which indicates the contact looseness fault has greater influence on switch-over process of OLTC.