Instantaneous Frequency Identification Of Time-Varying Structures Based On Synchroextracting Transform

Civil engineering structure is a time-varying structure,and its characteristic parameters also change with time.Since the characteristic parameters of time-varying structures can accurately reflect the overall dynamic characteristics of time-varying structures,it plays an important role in structural health monitoring.Therefore,it is of great significance to carry out the identification of time-varying structural characteristic parameters.With the increasing demand for parameter identification accuracy,the existing time-frequency analysis tools cannot meet the requirements due to the Heisenberg uncertainty principle.By employing stationary-phase approximation we can see that the parameters distributed on the ridge line are closely related to the instantaneous frequency of the signal.And instantaneous frequency is typical characteristic parameters of time-varying structures.Therefore,how to accurately identify the instantaneous frequency of time-varying structure is very important.In this paper,synchroextracting transform is introduced,and a variety of methods which can accurately identify the instantaneous frequency are proposed.The effectiveness and accuracy of the proposed methods are verified by numerical examples and time-varying structural tests.The main research work and innovations of this dissertation are as follows:(1)A combination of synchroextracting short time Fourier transform and dynamic optimization are proposed to estimate instantaneous frequency of response signals from time varying structures.The frequency band is firstly refined by synchroextracting short time Fourier transform.Due to synchroextracting short time Fourier transform has good noise resistance,its refined frequency band has good noise resistance Then,the dynamic optimization method is used to extract ridges in the frequency band to enhance the time-frequency resolution.The results show that index of accuracy of synchroextracting short time Fourier transform is 0.31%-1.02% less than that of wavelet coefficient modulus local maxima.(2)Inspired by synchrosquesszing wavelet transform,synchroextracting wavelet transform is proposed.This algorithm abandons the concept that synchrosquesszing transform squeezes the wavelet coefficients at the position of instantaneous frequency,and turns to use synchroextracting transform,which is designed to only retain the wavelet coefficients that are closely related to time-varying features of signals,while other weakly related wavelet coefficients are removed simultaneously.Since synchroextracting transform does not involve timefrequency rearrangement,its anti-noise performance is better than that of synchrosquesszing wavelet transform.However,the band width obtained by synchroextracting wavelet transform is unknown and unfixed,and there are still interference points outside the frequency band.In addition,the extracted frequency band is discontinuous in the direction of time and frequency.To address these issues,the extraction operator of synchroextracting wavelet transform is changed to extract frequency band by searching the most concentrated distribution of instantaneous frequency,and this enhanced method is named enhanced synchroextracting wavelet transform.(3)A combination of enhanced synchroextracting wavelet transform and dynamic optimization method are proposed to estimate instantaneous frequency.Although enhanced synchroextracting wavelet transform can refine the frequency band,it can not refine the frequency band into instantaneous frequency curve.Therefore,enhanced synchroextracting wavelet transform and dynamic programming method are combined to extract instantaneous frequency curves.To verify the effectiveness of the proposed method,synchrosquesszing short time Fourier transform,synchroextracting short time Fourier transform and synchrosquesszing wavelet transform are respectively combined with dynamic programming method to analyze the mono-component AM FM signal,multi-component AM FM signal and a time-varying cable test.The results show that the combination of enhanced synchroextracting wavelet transform and dynamic programming method has the highest accuracy and good noise resistance.Through the comparative analysis of index of accuracy,it can be seen that the index of accuracy of the combination of enhanced synchroextracting wavelet transform and dynamic optimization method is reduced by 0.04%-0.61% compared with the combination of synchroextracting short time Fourier transform and dynamic optimization method.Compared with the combination of synchrosquesszing short time Fourier transform and dynamic optimization method and the combination of synchrosquesszing wavelet transform and dynamic optimization method,the index of accuracys are reduced by 0.33%-7.12% and 3.43%-10.65%respectively(4)Aiming at the phenomenon of non rearrangement points in multisynchrosqueezing transform,the probability of non rearrangement points is reduced by rounding off the instantaneous frequency of multisynchrosqueezing transform twice,which is named as enhanced multisynchrosqueezing transform.