Multi-axle Vehicle Load Identification Theory Based On Minimum Residual Norm Steepest Descent Method And Its Experimental Study

Moving force identification(MFI)is an important research content in the field of bridge structure health monitoring,which provides effective data support for bridge structure reliability analysis,safety performance evaluation during operation and maintenance.Identification of bridge moving force belongs to the second kind of inverse problem in structural dynamics,and the inverse problem often has ill-posed characteristics.The problem of vehiclebridge system force identification can eventually be transformed into the solution of large-scale linear equations.Different identification methods have obvious differences in noise immunity and ill-posed immunity,which further affects the accuracy of load identification.Therefore,based on the related theories and research results of traditional time domain method(TDM)and preconditioned conjugate gradient method,this paper puts forward the minimum residual norm steepest descent(MRNSD)to identify multi-axle moving vehicles on bridges.The feasibility and accuracy of the proposed moving force identification method are verified by numerical simulation and laboratory vehicle-bridge model test.The main research contents and conclusions are as follows:(1)Traditional TDM is highly sensitive to noise,and the identification load under high noise can’t accurately invert the true force.In contrast,the MRNSD method shows remarkable robustness and satisfactory ill-posed immunity.When the noise level is as high as 20%,the 12 cases set in this paper can still effectively identify the load,and the identification errors are below 6.9%.The numerical simulation results show that the effect of identifying moving force by the proposed MRNSD method is obviously better than that by TDM.(2)The vehicle speed and axle spacing are two important parameters in MFI.With different vehicle speed and different axle spacing conditions,MRNSD method shows high identification accuracy,and its identification results are less affected by vehicle speed and axle spacings.The above phenomenon shows that MRNSD method has good adaptability to vehicle speeds and axle spacing changes.In addition,compared with TDM,MRNSD method can identify impact time-varying force more accurately,which shows that it can be used to identify different types of force.(3)When MRNSD method is applied to multi-axle moving vehicle load identification,the identification results of multi-type vehicle cases show that the identification errors are less than8.6%,and its identification accuracy can meet the actual engineering requirements.Compared with TDM and preconditioned conjugate gradient method,MRNSD method is less sensitive to noise and the number of vehicle axles,and the identification result is more stable,so it can identify multi-axle moving vehicle load on bridge deck more accurately.(4)The vehicle-bridge model test is carried out in the laboratory,and the moving vehicle load on the bridge deck is identified by using the measured bridge response.The efficiency and applicability of the MRNSD method were verified by setting different vehicle speeds and different vehicle weights.The identification load of MRNSD method fluctuates up and down near the static load of the vehicle,and the identification load accords with the actual situation.Then,the identified vehicle load is used as the input solution to solve the bridge reconstruction response,and the reconstruction response is in high agreement with the measured response,which further verifies the effectiveness of the proposed method.