On-load tap-changer based on TQWT and SVM algorithms Research on live detection technology

The vibration signal of on-load tap-changer (OLTC) can reflect the operating state of the transformer when it is energized. The extraction and analysis of vibration signal characteristic parameters and fault diagnosis can help to realize transformer health state early warning. In this paper, the optimized wavelet packet algorithm: quality factor tunable wavelet transform algorithm (TQWT) is used to extract the effective information of vibration signal obtained by the detection device, overcome the frequency confusion caused by wavelet filter, eliminate frequency folding, and effectively improve the accuracy of feature parameters. Support vector machine (SVM) algorithm is used to identify the fault type of vibration characteristic signal. On this basis, the on-load tap-changer live detection device is developed, which realizes the field detection and classification of the on-load tap-changer running state, greatly reduces the workload of field maintenance personnel, and improves the working efficiency of maintenance personnel.

Key words: on-load tap-changer; Feature extraction; Fault diagnosis; TQWT; SVM