Strong Motion Displacement Observation Method And Technology Based On Machine Vision

Understanding ground motion is the basis of studying earthquake response and earthquake disaster of engineering structure.How to observe and acquire the information of ground motion and its temporal-spatial variation reasonably and quickly has become a concerned research issue.In this study,a set of ground motion observation technology based on machine vision-dynamic displacement measurement technology combined with numerical analysis method was developed,which provides technical support for fast acquisition of spatial distribution of ground motions after earthquakes.In view of the application status of the visual system for the strong motion observation in practice,different combinations of rod models were constructed,based on the forward and inverse analysis of the structural dynamic response,the conversion relationship between the relative displacement of visual measurement and the absolute displacement of ground motion was established,thus the observation of ground motion displacement may be realized.The observation system and method were validated by large-scale shaking table physical model test,the robustness and sensitivity of the method were analyzed.The purpose of this paper is to develop strong motion observation technology and large-scale ground motion observation technology in earthquakes.Furthermore,the ground motion displacement observation results based on machine vision are taken as the constraints for baseline correction of strong motion acceleration records,to solve the problem of serious displacement distortion of near-fault strong motion acceleration records reasonably.The main innovative results of this study are as follows.(1)The spatial displacement measurement technology based on computer vision principle was developed.Relevant dynamic displacement measurement software was compiled based on C++,Open CV script library.The feasibility and accuracy of the technique are verified by a series of shaking table tests.The effects of different observation angles,ranging and target recognition on displacement measurement results were studied.This technology can measure the related quantity in real time,synchronization and multi-point interpretation.It solves the problems of single-point measurement,insufficient sampling rate and limited measuring range of traditional displacement measurement methods.(2)A seismic displacement observation system and technology was formed by integrating the visual testing technology and the dynamic analysis method of truss structures.The large-scale shaking table test of the ground motion displacement observation system model was carried out to simulate the observation scene of the system in the actual ground motion,and the comparative analysis of the ground motion displacement obtained simultaneously by the system observation platform and other measurement methods was carried out.At the same time,the influence of the characteristics of the observation system and the additional vibration on the stability and accuracy of the observation system was studied by shaking table test.Good experimental results fully demonstrate the pertinence,integrity,continuity,real-time and accuracy of the observation system.The observation results of different observation points were verified and revised to ensure the consistency of the system observation.(3)Based on the physical model test,the key nodes in the process of numerical modeling of observation system,such as forward and backward iteration algorithm flow construction,algorithm condition restriction,parameter determination of observation system structure and feasibility study of model,were discussed in depth.The numerical model was improved by using optimization idea,and a multi-means joint analysis method was established for fast determination of ground motion in earthquakes.(4)The observation system was used in conjunction with the accelerometer of strong motion observation,and a more perfect method of strong motion observation was formed.Taking the observed displacement obtained by the observation system as the displacement constraint condition for baseline correction of strong motion acceleration records obtained simultaneously,the serious distortion of strong motion displacement records near faults can be reasonably solved.To this end,a baseline correction and verification platform based on visual observation method was established,and shaking table tests were carried out to demonstrate the feasibility and accuracy of the method.Compared with the existing GPS co-seismic displacement measurement results,the platform can provide more complete and continuous strong motion displacement information in time and space domain,improve the accuracy standard and reliability of the existing baseline correction effect evaluation system.