A Time-Frequency Domain Detection Method for Measurement Data of Non-Stationary Signals Based on Optimized Hilbert-Huang Transform

Aiming at the phenomenon of mode mixing and information redundancy when Hilbert-Huang transform (HHT) is used for non-stationary signal measurement data processing, an optimized HHT algorithm is proposed in the study. The processing effect is improved by setting complementary ensemble empirical mode decomposition instead of empirical mode decomposition, using a frequency-domain smoothing vector to smooth and marginal spectrum feedback to optimize the time-frequency spectrum. The optimized algorithm is applied to the measurement data processing of acoustic signals of Penaeus vannamei. The duration, the range of the frequency, and the relative intensity of the frequency within 0~24 kHz of the signals are obtained. Mean-while, the optimized time-frequency spectrums obtained by processing the signals and the distribution diagrams of the number of key information points obtained under different smoothing vectors and feedback times prove that the optimized performance of the algorithm is affected by the signal quality and the selection of smoothing vectors. Besides, the primary and secondary feedback results need to be integrated when extracting signal features.