Identification Method Of Structural Physical Parameters Based On Extended Kalman Filter

With more and more complex large buildings put into use,the simple consideration of design and construction process safety is no longer sufficient,but also the need for real-time evaluation of the structural health of the service.Therefore,structural health monitoring technology has become a research hotspot in the field of civil engineering in recent years,at present,structural health monitoring system has been applied in many practical engineering projects,playing a real-time warning,to ensure the role of safety.The core technology of health monitoring is to identify the damage condition of structure,and structural parameter identification technology is the basis of structural damage diagnosis,so the identification of structural parameters has broad application prospects.Compared with modal parameter recognition,the method of direct identification of physical parameters is more intuitive,and the most commonly used method of structural physical parameter recognition is extended Kalman filtering method.The traditional extended Kalman filter has poor tracking performance when identifying time-varying structural parameters,in order to solve this problem,the forgetting factor is introduced,but the selection of the value of the constant forgetting factor has a great influence on the recognition result,and the adaptive method has strong tracking performance,but the calculation is cumbersome.In this paper,on the basis of summing up the research results of scholars in the past,a strong tracking extended Kalman filter method is introduced to identify time-varying structural physical parameters,and the following work is mainly carried out:(1)The basic principle of Kalman filter is introduced in detail and the recursive formula is deduced.For the identification of physical parameters,An extended Kalman filtering method for nonlinear systems is proposed,the algorithm flow is given,the method of weighted integral iteration is introduced to ensure the stable convergence of recognition results,and the identification method program is written,and the single degree of freedom,multi-degree of freedom and linear and nonlinear structure are numerically simulated respectively.The nonlinear hysteresis system uses the Bouc-Wen model to describe the hysteresis curve.The results of numerical simulation show that the extended Kalman filter can effectively identify the physical parameters of the structure for different structural systems.(2)In order to solve the limitation that the extended Kalman filter can not recognize the time-varying physical parameters,a strong tracking filtering method is introduced,which increases the Kalman gain by introducing an asymptotic factor to modify the prediction error covariance matrix in real time,and ensures that the filter can effectively track the change of parameters,has strong robustness,and has moderate computation.The time-varying structural physical parameters can be identified effectively.The identification method program is written,the different structural systems of structural stiffness mutation are simulated and identified by using strong tracking extended Kalman filter method.The recognition results show that the strong tracking method can effectively identify the changes of the system parameters.At the same time,according to the incomplete output information,the different working conditions are studied,and it is pointed out that the method is still practical under the condition of incomplete information.(3)In order to verify the practicability and effectiveness of the strong tracking extended Kalman filter method in practical engineering,the method is applied to the vibration table test of a 12-layer reinforced concrete frame,and the physical parameters of the structural model under different working conditions are identified by using the acceleration seismic response obtained by the test as the input and output data.The stiffness and damping identification results of each layer of the model under each working condition are given,and the test is broken with the experiment.