[en] In the areas of measurement, instrumentation and sensing across science and engineering, the recorded signal is typically manifested as a strong non-stationary characteristic and the amplitude of each component is varied among one another. It is necessary to propose an effective signal processing algorithm that could clearly reflect the fast varying frequency fluctuations of all components. As a recently proposed time-frequency analysis algorithm, multi-synchrosqueezing transform possesses the ability to squeeze the diffused time-frequency (TF) coefficients into the correct instantaneous frequency trajectory position in a TF plane iteratively. However, the amplitude related time-frequency representation result still makes it difficult to detect the fast varying instantaneous frequency for the amplitude weak component contained in a multi-component signal. To address this issue, this study employs a combined normalized time-frequency coefficients and the multi-synchrosqueezing operation technique to make a complete exhibition of the complex TF structure of a multi-component signal with various amplitude weak components. In the proposed method, the amplitude-related TF coefficients are all normalized as a constant value and only the reassigned TF positions after conducting the multi-synchrosqueezing operation multiple times are retained. By exploiting the proposed method on multi-component signals with amplitude weak components, the TF structure for all components can be clearly observed. The effectiveness of the proposed method is evaluated both on a set of simulated data and a set of experimental rubbing fault signals, whereby the implementation results are sufficient to demonstrate the superiority of the proposed method.

Highlights

1. An amplitude weak component detection technique capable of reflecting the entire TF structure for the measured signal in the real world is proposed in this study.

2. The amplitude weak component detection technique is established on the foundation of the multi-synchrosqueezing transform with eliminating the impact of amplitude on the TF structure.

3. The TF structure for all components with fast varying frequency fluctuation could be easily observed with the assistance of our proposed method no matter the amplitudes of each component are varied or not.

4. The simulation case and real-world rubbing signal analysis result have been sufficient to demonstrate the effectiveness of the proposed method. (paper)