Identification of Gear Conditions Using Nonlinearity Measures by Delay Vector Variance

Abstract

Vibrations in different running stages of a gear generally demonstrate different dynamical behavior. As a result, nonlinearity of vibration data can seemingly serve as a measure for describing running conditions of a gear. Nevertheless, measuring nonlinearity of vibration data seems to be a tough task because of complexities of data. To resolve this problem, this paper exploited delay vector variance (DVV), a time-delay method for quantifying nonlinearity and randomization of data, to investigate gear vibration data. Along this path, this paper proposed a novel method for identification of gear conditions using nonlinearity measures by DVV. Following this, the performance of the proposed method was benchmarked against Approximate Entropy and Sample Entropy using gear vibration data containing similar gear faults. The comparison indicated that the proposed method in this paper has the edge over Approximate Entropy and Sample Entropy in identifying gear fault conditions.