An innovative Hilbert–Huang transform (HHT) based frequency identification approach designated as adaptive magnitude spectrum algorithm(AMSA) is proposed in this paper. Characterized by the a posteriori property, the AMSA does not need the a priori information about the modalfrequencies to be identified, and the situations that a modal frequency may be contained along specific segments of the whole time duration ofone or more intrinsic mode functions (IMFs) are allowed for automatically in the resulting adaptive magnitude spectrum (AMS). The algorithmintroduces a banded frequency sweep procedure, during which a series of digital filters are designed to process the original signal. Then uponapplying HHT to the filtered signals, the forward weighted averages and the backward weighted averages are computed to construct the AMS,based on which the frequencies can be clearly identified. Two numerically simulated examples, i.e. the free vibration signal from a concreteslab subjected to impact loading and the random vibration signal generated by the Phase I IASC-ASCE structural health monitoring analyticalbenchmark problem, and one experimental example, i.e. the free vibration signal based on the Phase II IASC-ASCE structural heath monitoringexperimental benchmark problem, are used to demonstrate the efficacy of the algorithm. The results indicate that the AMSA is an effectivefrequency identification technique
An innovative Hilbert–Huang transform (HHT) based frequency identification approach designated as adaptive magnitude spectrum algorithm(AMSA) is proposed in this paper. Characterized by the a posteriori property, the AMSA does not need the a priori information about the modalfrequencies to be identified, and the situations that a modal frequency may be contained along specific segments of the whole time duration ofone or more intrinsic mode functions (IMFs) are allowed for automatically in the resulting adaptive magnitude spectrum (AMS). The algorithmintroduces a banded frequency sweep procedure, during which a series of digital filters are designed to process the original signal. Then uponapplying HHT to the filtered signals, the forward weighted averages and the backward weighted averages are computed to construct the AMS,based on which the frequencies can be clearly identified. Two numerically simulated examples, i.e. the free vibration signal from a concreteslab subjected to impact loading and the random vibration signal generated by the Phase I IASC-ASCE structural health monitoring analyticalbenchmark problem, and one experimental example, i.e. the free vibration signal based on the Phase II IASC-ASCE structural heath monitoringexperimental benchmark problem, are used to demonstrate the efficacy of the algorithm. The results indicate that the AMSA is an effectivefrequency identification technique