Detecting broken strands in transmission line - Part 2: Quantitative identification based on S-transform and SVM

Abstract

SUMMARY Strikes of lightning, corrosion of chemical contaminants, ice shedding, wind vibration of conductors, lines galloping and damage due to external forces may induce some fatal accidents such as broken transmission lines. The common means to be used for latent faults inspection in transmission lines is manually and periodically examined by workers from commercial electrical power company. With the development of both the artificial intelligence technologies and the smart grid, the method of detecting broken strands in transmission line by inspection robot with detectors is a method with good prospect. In this paper, a method with eddy current transducer carried by robot for detecting broken strands in transmission line is developed and the broken strands in transmission line identification approach based on S-transform is proposed. The proposed approach utilizes S-transform to extract the module and phase information at each frequency point from detection signals. Through module phase and comparison, the characteristic frequency points are ascertained, and the fault information of the detection signal is constructed. The broken strands identification confidence degree is defined with Shannon fuzzy entropy (SFE-BSICD). The proposed approach combines module information while utilizing phase information, SFE-BSICD and the energy, so the reliability is greatly improved. These characteristic qualities of broken strands in transmission line are the input of support vector machine with multi-classification, and then the number of broken strands can be determined. Through field experimental verification, it can be concluded that the proposed approach shows high accuracy and the SFE-BSICD is defined reasonably. Copyright © 2012 John Wiley & Sons, Ltd.