Time-Frequency Feature Extraction without Cross-Terms Based on Acoustic Signal in Rotor Malfunctions

Abstract

An effective approach is presented to eliminate the cross-terms in Wigner distribution by ICA (independent component analysis) and EMD (empirical mode decomposition), through which the cross-terms caused by the uncorrelated mixing signals can be removed successfully. This method is used for time-varying signal analysis and is powerful in signal feature extraction, especially for joint time frequency resolution, which is demonstrated by numerical examples. To further understand the method and its application, a detailed analysis about abrupt unbalance experimental example is shown to explain the cause of malfunction as well as its occurrence and phenomenon. In addition, the proposed approach based upon independent component analysis, empirical mode decomposition method and wigner distribution allows the separation and analysis of the sources with nonlinear and non-stationary properties. In this method, the main conceptual innovations are the associated introduction of ‘source separation’ and ‘intrinsic mode functions’ based on the local properties of the mixed signals, which makes the instantaneous frequency meaningful; the method serves to illustrate the roles played by the nonlinear and non-stationary effects in the energy-time-frequency distribution. At the same time, the method can also be expanded and applied in other fields.