Research On Optimal Combination Form Of Empirical Wavelet Transform In Bridge Modal Parameter Identification

The bridge structure condition evaluation is a key step to judge the bridge service state.Based on the health monitoring system,the response of the key structure of the bridge in operation is collected,and then the bridge modal parameters are identified by the methods of time domain,frequency domain and frequency domain,which can provide a reliable basis for bridge state evaluation.In practical engineering,the monitoring signal often has the characteristics of high noise intensity and dense low-frequency modes,and the application effect of the existing operational modal analysis methods often can not achieve the ideal accuracy.Empirical wavelet transform is a new method in time and frequency domain,which has been widely studied in mechanical fault identification,signal feature extraction and signal denoising.the accuracy of spectrum segmentation and the type of wavelet basis are decisive factors for the accuracy of empirical wavelet transform.Based on the existing research results of empirical wavelet transform,its application in the field of bridge modal parameter identification is deeply studied,and the optimal combination form of empirical wavelet transform is selected according to the application effect of practical engineering.The main contents include:1.In this paper,the principle of empirical wavelet transform is expounded,the key factors affecting its analysis accuracy are analyzed,and the existing improved methods of empirical wavelet transform based on spectrum segmentation are summarized.the comparison and selection framework of improved empirical wavelet transform based on different spectrum segmentation methods is established,and the better spectrum segmentation method is selected by taking the noisy signal as the research object.2.In this paper,the wavelet basis functions commonly used in engineering are introduced,and the wavelet basis functions are compared and selected according to the characteristic parameters.Based on the idea of wavelet packet decomposition,the embedding method of wavelet bases except Meyer wavelet basis in empirical wavelet transform is established,and the application effect of this method is verified by simulation signal.3.Based on Bernoulli theorem of large numbers,a method for determining the reconstruction order in singular value reconstruction is proposed,and the application effect of this method is verified by simulation signals of different noise levels.on this basis,an improved empirical wavelet transform method based on singular value reconstruction is established,and its application effect is compared with the existing improved methods,and the calculation accuracy of the method is verified.The comparison and selection rules of the optimal spectrum segmentation method based on bridge modal parameter identification and the empirical wavelet transform method based on the combination of wavelet bases are established.4.Taking a long-span suspension bridge as a research background,the exploratory data analysis technology was used to preprocess the acceleration response collected by its health monitoring system,and the vertical acceleration response of the main beam was taken as the analysis object.Analyze the data comparison,and analyze the entropy weight method based on the modal parameter identification number,numerical distribution range,reconstructed signal SNR and root mean square error and other indicators,and then select the optimal empirical wavelet transform combination form,and finally use The measured response data of the transverse direction of the main beam,the vertical and horizontal directions of the bridge tower and the slings verify the calculation accuracy of the method.