Structural Displacement And Modal Identification Methods Based On Phase Video Motion Amplification

The deflection of long-span bridge structures is an intuitive basis to reflect the structure’s overall state,and it is necessary to monitor the deflection of the structure during operation.Modal parameters(mode shape,frequency,damping ratio)of structures are critical indexes for understanding structural dynamic characteristics,detecting structural damage,and evaluating safety.Due to the limitation of the number of sensors,the traditional vibration test methods are difficult to achieve full field measurement.Non-contact measurement method based on computer vision has become a research hotspot in recent years because of its high precision,full field measurement,convenience and high efficiency.Therefore,this paper combines the machine vision theory to study the identification method of structural vibration displacement and modal parameters.The main research contents of this paper include:First,the vibration displacement extraction method using phase-based video motion amplification is studied.The thesis discusses the video micro motion amplification algorithm based on phase,and introduces the Euler method’s related principles,optical flow method,image pyramid,and phase.In this algorithm,the video sequence is decomposed into phase spectrum and amplitude spectrum through a pyramid of multi-scale and multi-directional decomposition,and the phase spectrum is filtered in the time domain to select the frequency band of interest.Finally,the extracted signal is amplified and added with the original video sequence signal to obtain the amplified structure motion.On this basis,the relationship between displacement and phase is constructed by combining displacement and phase.The reference video frame is introduced to obtain the central phase,that is,the relative displacement of the structure.The experimental model verifies the accuracy of the recognition results.Then,a structural modal identification method based on an improved Fast ICA algorithm is proposed.Based on the Fast ICA algorithm of the blind source separation theory and Newton downhill method,a new Fast ICA algorithm based on the Newton downhill method is proposed.By introducing downhill factor,the limitation of Newton method on the initial value is weakened and the convergence speed is increased.The experimental and numerical examples are employed to verify the accuracy of the improved algorithm.Finally,the field test and verification of monitoring data of the real bridge are carried out.The phase-based video motion amplification algorithm and the improved Fast ICA algorithm are applied to the analysis of structural vibration video data to extract the structural displacement information and identify the structural modal parameters.The displacement and modal parameters obtained by the BSS algorithm are compared with the proposed method’s identification results to further verify the application effects of the proposed method in practical engineering.