A Method Of Modal Parameter Identification For Large Bridges

Because of the high cost of long-span bridges,many kinds of sensors are usually installed to monitor the working state of bridges in real time,and modal parameter identification is the premise and foundation of evaluating the state of bridges.At the same time,The long-span bridge has the characteristics of low natural frequency,dense mode and easy to be interfered by noise.The general time-frequency and frequency-domain methods can not extract the non-stationary and non-linear parts;In the time-frequency-domain methods,wavelet transform and HHT transform have the limitations of wavelet basis selection and weak theoretical basis.This paper explores the highly adaptive EWT time-frequency analysis method,and improves and proposes a modal parameter identification method that can be used for large-span bridges.According to the EWT algorithm,the improvement is put forward: the selection of spectrum division method is one of the key steps of EWT decomposition.According to the characteristics of large-scale bridge modal density,the scale space method which can automatically estimate the modal number is selected.Then,by dividing the spectrum of multiple simulation signals and the measured bridge vibration signals,the three threshold calculation methods corresponding to the scale space method are discussed.And according to the division results The maximum class variance method suitable for bridge vibration signal is selected.According to the characteristics of large-scale bridge vibration signal susceptible to noise interference,this paper proposes to use morphological closed operation filter to denoise the frequency spectrum,and verify the effectiveness of the method through simulation signal and measured signal.Based on the improved EWT algorithm,the identification process of bridge modal parameters is proposed.The correlation coefficient is used to select the effective components from the decomposed modal components,and the random decrement method and Hilbert transform are used to identify the modal parameters from the effective components.The validity of this method is verified by simply supported beam model.Firstly,the vibration response of each node of simply supported beam excited by multiple vertical white noise is obtained,then the modal parameters of simply supported beam are obtained according to the above identification process,then the modal parameters of the finite element model of simply supported beam are obtained through modal analysis,and finally the comparative analysis is carried out;Then through the actual vibration signals of Nanjing Yangtze River Bridge to verify: the collected vibration signals are decomposed and modal parameters are identified,then a finite element model of Nanjing Yangtze River Bridge is established to obtain modal parameters through eigenvalue analysis,and finally the results are compared and analyzed.The final results show that: the reference of correlation coefficient can screen out the false modes,and the noise reduction of morphological closed operation filter can smooth the peak of the spectrum caused by the noise,so that the phenomenon of mode over decomposition can be suppressed,and the result of spectrum division will be more accurate.Whether it is the bridge response signal obtained by simulation or the bridge response signal collected by the vibration test of the bridge under environmental excitation,the modal parameter identification process based on the improved EWT algorithm proposed in this paper can be effectively decomposed and accurately identified The modal parameters of the bridge.