Application of signal processing technology based on symbolic time series analysis to rotor broken fault detection

Abstract

An improved method of motor fault detection based on symbolic time series analysis is proposed, and the method adaptively partition off the region which has the most symbols in the symbolic series into two new regions. The method makes that the regions with more information are assigned more symbols relatively but those with sparse information are assigned fewer symbols, which enhances the sensitive degree of symbols to the signal. Laboratory experiments of fault diagnosis of broken rotor for inductive motor show that comparing with the uniform partition, the new method is more sensitive to the system and also owns a stronger robustness and a better reliability.