Application of multi-base fusion generalized chirplet basis transform in vibration signal analysis of multiple rotor rotating machinery

In rotating machineries, there are many signals with multiple nonproportional fundamental frequencies such as the vibration response signal from a dual rotor aeroengine, where the speeds of the high-pressure turbine and the low-pressure turbine are not synchronized. The conventional Chirplet transform (CT) based time-frequency analysis (TFA) methods are capable of processing nonstationary signals mixed with multiple synchronous components, or proportional frequency components, but not suitable for processing the nonstationary signals with multiple nonproportional fundamental frequencies. This paper presents a novel nonstationary signal TFA method, referred to as General Chirplet Basis Transform (GCBT), which is designed for processing nonstationary signals with multiple nonproportional fundamental frequencies. In the proposed GCBT, the local maximum value search method is utilized to find out all the Chirplet basis function parameters matched with the fundamental frequencies of each harmonic group contained in the signal. Then, the time-frequency representation (TFR) fusion criterion is established to effectively combine the TFA results of all the discovered Chirplet basis functions. The effectiveness and superiority of the GCBT are demonstrated by analyzing a simulation signal, the vibration signals collected in the laboratory using two independently operated small rotors, and the vibration signals from an aeroengine test. All the results indicate that the proposed GCBT has the ability to handle multi-component signals with nonproportional fundamental frequencies, synchronous frequencies, closely-spaced frequencies, as well as the signals with sever noise backgrounds.