High-Order Synchroextracting Chirplet Transform for Accurate Instantaneous Frequency Estimation and Its Application in Fault Diagnosis of Rotary Machinery

Abstract

Time-frequency analysis (TFA) technology is an effective tool for extracting time-varying characteristics of a signal, but for strong AM-FM signals, time-frequency representation (TFR) with high time-frequency (TF) resolution is often difficult to generate. The measured vibration signals of mechanical equipment often contain a large number of non-stationary features and are polluted by noise, which hinders the use of TFA technologies. In this paper, we propose a TFA method termed high-order synchroextracting chirplet transform (HSECT) to effectively process strong AM-FM signals. On the basis of Chirplet transform (CT), we first determine the best linear demodulation operator at each moment through the local maximum amplitude of the generated TFR. Subsequently, the amplitude and phase of the signal are subjected to high-order Taylor expansion, and a new frequency estimation operator termed high-order synchroextracting chirplet operator (HSECO) is obtained by the high-order approximation of the time-varying laws of the signal. HSECO can accurately extract the instantaneous frequency of strong AM-FM signal and the construction of HSECT is completed by Dirichlet function at last. HSECT can generate TFR with high energy concentration from which the time-varying law of strong AM-FM signal can be extracted accurately. Meanwhile, the signal reconstruction operation can be completed concisely. In addition, the proposed method is applied to the fault diagnosis of the rotor rub impact and variable speed equipment and compared with other TFA methods. Simulation and experiments can verify the effectiveness and competitiveness of the proposed method.