Adaptive separation method for mixed pulse sequences based on time-frequency analysis

Abstract

Existing signal separation methods such as Fourier transform, wavelet transform, empirical mode decomposition, and variational mode decomposition are ineffective in separating mixed pulse sequences. This paper proposes an adaptive decomposition method for mixed pulse sequences by analyzing the generation mechanism and time-frequency characteristics of mixed laser pulse sequences in an accurate laser positioning system (ALPS). This method can adaptively decompose mixed pulse sequences into steady-state or non-steady-state pulse trains, with decomposed results having clear physical meanings. The effectiveness and robustness of the proposed adaptive decomposition method for mixed pulse sequences are validated using the ALPS platform. Experimental results demonstrate that this method can adaptively decompose ALPS mixed pulse measurement sequences and has strong interference resistance. This paper provides new insights into adaptive decomposition methods for signals with many discontinuous points, offering new tools for diagnosing faults in rotating machinery and monitoring sub-cycle speed fluctuations. Additionally, based on this method, a coordinate online calculation model for ALPS measurement nodes is designed, which increases the coordinate data refresh rate of ALPS by more than ten times, effectively improving the system's dynamic measurement performance.