For improving the accuracy of mechanical condition monitoring of on-load tap-changer (OLTC) further

the vibration signal of the OLTC during the switching period is analyzed on the basis of the optimized tunable Q-factor wavelet transform (TQWT) method. The optimization decomposition method of the TQWT is studied by using the artificial fish swarm algorithm and AFAS (based on decomposition residue and global orthogonal coefficients)

and the mechanical fault diagnosis model of OLTC is constructed based on the optimized twin support vector machine (TWSVM) and in accordance with multiple multiple subsequences of the OLTC vibration signals which is obtained by calculation. The analysis results from vibration signals of a CM-type OLTC under normal and typical mechanical faults show that the proposed optimized TQWT decomposition method can effectively improve the accuracy of decomposing vibration signals of OLTC. The mechanical fault diagnosis mode of OLTC with AFSA-TWSVM

compared to other diagnosis models

has better classification performance and fast convergence speed.