Moving Load Identification Method Based On Group Sparsity Theory And Compressive Sensing

As an important part of transportation,bridges have become the focus of national infrastructure construction.The reciprocating action of moving force and overloaded transportation seriously affects the normal service life of bridges,so the moving force identification(MFI)has become one of the main concerns of bridge safety.Moving forces are constantly changing with time and space,and it is difficult to measure directly.Therefore,it is a very realistic choice to obtain information such as vehicle time-varying loads through the bridge dynamic responses caused by the vehicles.In recent years,the research of domestic and foreign scholars has proved that the moving load identification method based on regularization and compressive sensing has the advantages of high identification accuracy and low computational complexity,but there are still some shortcomings.For example,the regularization method often brings over-fitting problems when identifying the vehicle on and off the bridge;the compressed sensing theory is rarely applied in practice and the domestic research is relatively late,so it’s of important theoretical value and practical significance for this field.In order to further improve the MFI accuracy and reduce the computational costs,the concept of group sparsity is introduced into this thesis and a novel MFI method is proposed based on group sparsity and compressed sensing,which is mainly divided into the following aspects:(1)Some commonly used MFI methods are introduced and summarized.The research methods are classified into classical method,function approximation method,regularization method,influence line method,wavelet analysis method and neural network method,and the advantages and disadvantages of various methods are compared and analyzed.(2)A novel MFI method based on group sparsity and compressed sensing is proposed.And the basic theory and related knowledge of the proposed method are expounded,including the concept of sparse regularization,dictionary principle,compressed sensing and group sparse theory.Starting from the basic equations of MFI method in time domain,a new equation for MFI based on group sparseness and compressive sensing is established through derivation,and finally solved for MFI.(3)Numerical simulation verifications are carried out for the proposed MFI method based on group sparseness and compressive sensing.Using MATLAB,the three forms of moving force including single-axis moving force,two-axis moving force and single-axis simple harmonic moving force with pulse components were identified.At the same time,parametric effects on the proposed MFI method were carried out,such as the ratio g/k of the number of groups g to the sparsity k,the ratio of the compression coefficient to the sparsity m/k,and the grouping method.Then the factors of measuring points,noise,speed,and bridge span on MFI results have been summarized according to the research results above.(4)Experimental verifications of the novel MFI method based on group sparseness and compressive sensing are carried out as well.A simply supported beam structure was built in laboratory,and the first three-order frequency information of the structure was obtained through experimental modal analysis.Then,the measured impulse data acquired by the photoelectric sensors is preliminarily analyzed to obtain the actual vehicle speed of the trolley,and the measured data by the strain sensors and accelerometers is used to evaluate the moving vehicle loads under different vehicle speeds,axle loads,wheelbases,etc.,which further verifies the effectiveness and feasibility of the proposed MFI method.