Finite Element Model Updating Based On Wind Driven Optimization And Wavelet Neural Network

With the increasing complexity of large structures,it is essential to establish a precise finite element model for the structure to guarantee the normal operation and safe use.However,it is very difficult to establish precise finite element models only based on design drawings and engineering experience.To make the established finite element models precise,model updating is required.In this paper,the finite element model updating is studied,and the main contents and constructions are as follows:(1)In this paper,the finite element model updating of structure is taken as the research goal,and the finite element model updating based on the structural response and the updating method based on the optimization algorithm and the agent model are reviewed.Therefore,a finite element model updating method based on wind driven optimization algorithm and wavelet neural network combined with structural static and dynamic response information is proposed.(2)The theory and implementation process of the wind driven algorithm are introduced in detail,and the relationship between the air particles and the correction parameters is established,so that the algorithm can be applied to the structural finite element model updating.To compare with the particle swarm optimization algorithm,the function optimization test is conducted,and the experimental results show that the wind driven optimization is more advantageous than the particle swarm optimization.The algorithm converge is fast,and the optimization ability is strong.(3)The theoretical core,the training process and the operation method of the wavelet neural network are studied.Meanwhile,in order to improve the network stability,a method of selecting the network coefficient matrix by wind driven optimization is proposed.And through a function fitting test,this paper compares the function fitting performance of wavelet neural network optimized by wind driven with the traditional wavelet neural network and BP neural network and RBF neural network to verify the applicability of the optimized wavelet neural network as the surrogate model(4)The theory and realization process of model updating are elaborated in detail.And based on the wind driven optimization and wavelet neural network,combined with the static and dynamic responses,the finite element model of a numerical structure is modified.And the results of model updating show that the wavelet neural network can reflect the nonlinear relationship between the structural response and the parameters,and has an outstanding simulation performance;the wind driven optimization has excellent ability of optimization and can effectively improve theefficiency of model updating;the proposed method can effectively update the finite element model and improve the working efficiency,and it is applicable.(5)The modification of the finite element model of the bridge structure is studied,and the initial finite element model of the Ningbo WaiTan Bridge is set up.Combined with the dynamic responses of the bridge,the initial model is modified based on the driven optimization and the wavelet neural network.The bridge model updating results show that the finite element model updating method based on the wind driven optimization and wavelet neural network is applicable to the updating of multi-parameter bridge model,which has practical significance in engineering and high efficiency in finite element model updating,and the results of calculation are reliable...