Abstract
The Hilbert-Huang transform (HHT) is widely used for time-frequency analysis of blasting seismic wave signals due to its unique adaptability. However, blasting seismic wave signals are typical non-stationary vibration signals that are susceptible to noise interference, leading to mode confusion and endpoint effects in empirical mode decomposition (EMD) in HHT, which in turn affects the accuracy of time-frequency analysis. In order to obtain accurate time-frequency characteristic parameters of blasting seismic wave signals, it is necessary to improve HHT. A time-frequency analysis algorithm called DEP- CEEMDAN-MPE-INHT was proposed. The first step of the algorithm is to perform dual endpoint processing (DEP) on the signal. The second step is to combine the advantages of complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and multi-scale permutation entropy (MPE) to obtain CEEMDAN-MPE, and perform CEEMDAN-MPE on the DEP processed signal to achieve synchronous suppression of high-frequency noise and low-frequency trend terms. The third step is to perform a normalized Hilbert transform (NHT) on the intrinsic mode function (IMF) obtained from DEP-CEEMDAN-MPE to achieve INHT. The above three steps can establish the time-frequency analysis algorithm of DEP-CEEMDAN-MPE-INHT. Through noisy simulation signal testing, the comparative study of DEP-CEEMDAN-MPE-INHT and HHT is carried out. Finally, the algorithm is applied to the time-frequency analysis of actual blasting seismic wave signals. The results show that DEP-CEEMDAN-MPE-INHT not only suppresses the EMD endpoint effect and mode confusion, but also obtains the time spectrum with high resolution in time domain and frequency domain. Through DEP-CEEMDAN-MPE-INHT time-frequency analysis, the time-frequency characteristic parameters of blasting seismic wave signal can be accurately extracted, which has important practical significance for the hazard identification and control of blasting seismic wave.