Empirical end-device disturbance recognition by waveform feature learning models

Abstract

A waveform holds recognizable feature patterns. To extract such features of various equipment disturbance conditions from the waveform, we present a practical model to estimate distribution line (DL) conditions by means of a multi-label extreme learning machine. The motivation for the waveform learning is to develop device embedded models which are capable of detecting and classifying abnormal operations on the DLs. The model with the real DL waveform data holds a potential for determining additional DL conditions and improving its classification performance through the update mechanism of the learning machine. On the other hand, conditional structures for distribution networks are discovered with respect to distribution network configurations and measurement device characteristics by the time-scaled class map of obtained measurement data from the field.