An ameliorated synchroextracting transform based on upgraded local instantaneous frequency approximation

This paper introduces an upgraded synchroextracting transform (SET), namely ameliorated synchroextracting transform (ASET), to deal with fast time-varying and strong frequency modulated signals for time and frequency analysis (TFA). SET is a recently developed time-frequency representation (TFR) method with an exceptional capability to enhance the readability of TFR, yet the kernel of SET is based upon the assumption of purely harmonic signal, therefore, it is more applicable to slow time-varying and weekly frequency modulated signals. However, for rotating machines in real practice, fluctuation loads and varying rotational speeds are always inevitable, the resultant vibrations will exhibit non-stationary frequency-modulated nature. To tackle this problem, in this paper, the synchroextracted operator in SET is upgraded in terms of second-order Taylor expansion to estimate the fast and strong frequency modulated signals so that a better TFR can be achieved. The advantages of ASET is demonstrated with simulated and experimental studies by comparing the existing time and frequency analysis methods. The sharpness of TFR, the ability of signal reconstruction and the robustness to noise pollution are also demonstrated. (C) 2019 Elsevier Ltd. All rights reserved.