Research On HHT And Its Application In Structural Health Monitoring

Signal processing is important in helping the development of Structural Health Monitoring (SHM)techniques. Lamb wave based Structural Health Monitoring method is one of the most interesting hotspots for SHM. Lamb wave is a typical non-stationary signal. However, the traditional FFT analysistheory is limited for time-frequency combined analysis. Hilbert-Huang transform(HHT) is known as atheory used for non-stationary and nonlinear signal. In HHT, a signal is first decomposed into finiteIMFs by EMD, and then, with the Hilbert transform, the meaning instantaneous frequency of everyIMF that gives precise description of the varying frequency is obtained. Hence HHT shows apromising application in Lamb wave analysis. In this paper the HHT signal processing is researchedsystematically, and the research result is applied in SHM. The main research contents of this paper areas follows:Firstly, this paper compared generalized extended interpolation to Cubic Spline Interpolation andthe mean from the local maxima and the local minima to the mean of the upper and lower envelopes,then analyzed the reason of the end effects and choosing principles and researched deeply the criterionfor stoppage of three common IMF filtering. Based on the above research, the concept of thedecomposition plane was introduced and EMD realization techniques’ influence on EMDdecomposition effect was assessed. Then an IMF criterion for sifting stoppage based on excludingmargin was put forward, which verified the EMD decomposition effect based on this method.Secondly, the influence on the extreme value point by signal amplitude, frequency and phasevariation was analyzed by experiments, and then the influence on EMD decomposition effect byextreme value variation trend was researched. Combining the variation of signal amplitude, frequencyand phase with EMD decomposition, signals’ influence on EMD decomposition effect was discoveredby both experimental induction and theoretical deduction. The relationship was validated byexperiments.Thirdly, based on the research about EMD realization technique and the influence on EMDdecomposition effect by signal, the reason of mode mixing is analyzed. Three basic types of modemixing are proposed. The solution principles to the three typical mode mixing were researched deeply,and their limitations were also analyzed. EMD based on frequency bands was proposed to solve modemixing, whose mode mixing solution effect was validated by experiments.Fourthly, the instantaneous phase variation when the direct arrival wave was superimposed by damage scattering signal was researched and its result was validated by experiments. The reason whyEMD decomposition of the active Lamb wave signal is unstable is analyzed and BF-EMD wasintroduced to improve the decomposition stability. The experimental validation was followed. Basedon that, the marginal spectrum was applied to research damage’s influence on the instantaneousfrequency of direct arrival wave and the research result was compared to signal’s frequency spectrum.Fifthly, the mode feature of Lamb wave generated by impact damage was analyzed theoreticallyand validated by lead cutting experiment. The propagation velocity calculation of A0mode wasutilized to validate the extraction ability of BF-EMD about signals of narrow frequency. This methodwas introduced to realize impact source localization. During the prognosis of impact damage, IMFprovoked by structural damage was obtained by EMD self-adaptive signal decomposition and damagelocalization was realized by BF-EMD extraction method.