Optimal Sensor Placement And Modal Analysis Of Engineering Structure

In recent decades,a large number of major projects have been built in China.With the development of economy and the progress of science and technology,engineering with larger scale and higher complexity will come out one after another.With the increase of service age and the influence of many factors,the service ability of engineering structure decreases gradually.Health monitoring systems have been added to many large structures to monitor the safety of the structures in service.Reasonable placement of sensors is the premise of effective structural health monitoring.The optimal sensor placement scheme can reduce the number of sensors and improve the coverage and sensitivity of the response information.Modal parameters and experimental models can be effectively identified according to the measured excitation and response.The modal parameters of the structure can be effectively identified according to the response.Based on the results of modal analysis,the finite element model of the structure is updated.Finite element model updating of structures can provide scientific basis for many engineering problems such as dynamic analysis,state assessment,and damage diagnosis.Based on the above background,the following interrelated studies are carried out:(1)A fish swarm algorithm with Lévy flight and firefly behavior(LFFSA)is proposed,which combines the optimization characteristics of artificial fish swarm algorithm with firefly algorithm.This algorithm introduces the individual movement strategy of firefly into the two behavior patterns of fish swarm and preying,and the search strategy of Lévy flight is introduced into the search path of fish swarm,which makes the search process more efficient.Besides,nonlinear visual and step length based on dynamic parameter are simultaneously considered for limiting the search band and self-adaptability of step size factor is enhanced.The optimization performance of Artificial Fish-Swarm Algorithm(AFSA),Firefly Algorithm(FA),Firefly Algorithm with Lévy flight(LFA),Fish-Swarm Algorithm with Firefly Behavior(FFSA),and Fish Swarm Algorithm with Lévy Flight and Firefly Behavior(LFFSA)algorithms is tested and compared by nine different test functions.The results show that the LFFSA has a better performance in convergence speed and optimization accuracy.(2)Optimal sensor placement methods based on relaxation sequence method and fish swarm algorithm(LFFSA)improved through Lévy flight and firefly behavior are proposed respectively.Firstly,the relaxation idea is incorporated into the sequence optimization method of sensor optimal placement,and the relaxation sequence method of sensor optimal placement is proposed.Secondly,the optimal sensor placement method of Lévy flight modified fish swarm algorithm(LFFSA),which combines Lévy flight and firefly behavior,is proposed.Both methods select the maximum off-diagonal elements of the modal assurancecriterion matrix(MAC)as the objective function.Finally,taking the plane truss structure and the three-dimensional bridge structure model as examples,the above two methods are compared with the traditional accumulation sequence method.The results show that the sensor placement effects of the two methods proposed in this paper are better than the cumulative sequence method,and the optimal sensor placement method based on the LFFSA makes use of the good convergence and global search ability of the LFFSA algorithm.The best sensor arrangement effect is obtained.(3)A variational mode decomposition(VMD)and random subspace method(SSI),VMD-SSI,is proposed to realize the identification of structural modal parameters.In the step of VMD decomposition,the modal repetition ratio is proposed as the evaluation criterion of the number of layers of the mode decomposition,and the decomposition layer K is optimized.At the same time,based on the modal components obtained by VMD decomposition,singular value decomposition(SVD)is introduced,and modal parameters are identified by SSI method.The error of curve fitting in VMD parameter identification is avoided,and the accuracy of modal parameter identification is improved.Finally,statistical theory is used to test the modal frequency,modal damping and modal shape of the improved VMD-SSI method respectively so as to verify the effectiveness of the proposed method.(4)Preliminary study on model updating method based on modal parameter identification is carried out.The model between structural parameters and modal parameters is established by surrogate model method,and the problem of model updating is transformed into a constrained nonlinear programming problem.The cuckoo optimization algorithm is used to optimize the modified structural parameters,and the modification effect of the structural parameters is verified.Four kinds of surrogate models — Kriging model,radial basis function,support vector machine and polynomial response surface are used to update the model of a plane truss structure.The results show that kriging model and response surface model have high accuracy and efficiency,respectively.