Analysis And Design Optimization Of Vehicle Structures Based On Uncertainty Models

There are a lot of uncertainty problems in engineering structures.The potential uncertainty will have an important effect on the structural performance.Therefore,considering the influence of uncertainty in the early stage of structural design is of great significance to improve product performance.The mechanical structure designed with uncertainty not only has low performance failure risk but also has high-performance stability.It goes without saying that as a large-scale mechanical structure system,the design optimization of key vehicle structures is particularly important.Based on the interval model with limited cognition of uncertain information,this thesis studies the application method of structural analysis and design optimization based on the uncertain model,to further supplement the existing structural uncertainty analysis and design optimization problems,and put forward effective innovative design methods for vehicle parts in the initial structural design.The main research works of this thesis are as follows:(1)The structural design optimization method considering the uncertainty of design parameters is further developed.For the structural optimization considering the known deviation range of design variables,taking the structural design of the planetary gear train of electric vehicles as an example,the design variables and other design parameters are considered as interval uncertainty values,and a variety of gear design schemes are obtained by using multi-criteria decision making(MCDM)method on the premise of considering a variety of uncertainty degrees.Furthermore,for the structural optimization considering the unknown deviation range of design variables,taking the structural design of a cantilever beam as an example,the design variables are considered as interval uncertainty values,while other design parameters are considered as random uncertainty values,and the deviation level index of design variables is introduced as a new objective function.This optimization method can not only obtain the optimal design variables but also obtain the matching design deviation range.It can further provide more ideas for the design scheme of other complex vehicle structures.(2)To further analyze the structural performance changes caused by interval uncertain parameters,a structural uncertain response analysis method(RBFNNIE)using generalized radial basis function neural network(RBFNN)and interval Taylor expansion is proposed.Among them,the optimal network parameters of RBFNN are trained by the k-means++ clustering algorithm and singular value decomposition(SVD),and the dimension of multidimensional interval uncertain Taylor expansion is reduced by retaining only the data on the diagonal of the Hessian matrix.Through the test function analysis,it is found that the derivative expression of the original function derived by RBFNN can approximate different nonlinear functions.Finally,the method proposed is applied to the uncertainty analysis of key vehicle structures,including the uncertain response analysis of the modal performance of a hood,the bearing performance of a mechanical claw,the crashworthiness of a front-end anti-collision structure,and the kinematics performance of a multi-link suspension.To further improve the accuracy of RBFNNIE,multiple sub-interval expansion is implemented in RBFNNIE,and the uncertainty analysis results are compared with the genetic algorithm(GA)and Monte Carlo simulation(MCS).The advantages of RBFNNIE with multiple subintervals are verified in terms of efficiency and accuracy.(3)To solve the problem of mixed uncertainty in structural design optimization,a multi-stage surrogate model is proposed to optimize the design of mixed uncertainty.The idea of the multi-stage surrogate model is to use the global surrogate model method with an adaptive update strategy to establish the deterministic surrogate model of system response and the upper and lower bound surrogate model of interval response,and then use the polynomial chaos expansion(PCE)method considering random uncertainty to calculate the random output response on the basis of interval response surrogate model.Firstly,to verify the feasibility of this method,the interval uncertainty response analysis,and probability uncertainty response analysis models are verified by using the test functions.Finally,the design optimization framework with a multi-stage surrogate model is applied to the kinematic performance design of the new double trailing arm suspension,to improve the chances of wheel alignment parameters in the process of wheel jump.Compared with the initial design scheme and the deterministic design optimization scheme,the superiority of the uncertain design optimization method proposed in this study in improving the kinematic performance design of double trailing arm suspension is verified.(4)Considering the possible model uncertainty in the performance design of complex nonlinear structures by the auxiliary surrogate model optimization framework,an optimization method of model cognitive uncertainty is proposed in this study.The basic idea of this method is to find the maximum deviation of the polynomial response surface model(RSM)in limited sample points,take it as the interval radius in the interval model response,and then establish the upper and lower bound response surface model of specific structural performance response.Compared with the general probability model,this method can effectively wrap all finite sample points in the interval of the surrogate model without probability statistics.In addition,under the background of lightweight and passive safety,a series of honeycomb structures with new bionic cells are proposed,and the accuracy of its finite element numerical simulation model is verified by experiments and theoretical analysis.Then,the method proposed in this study is successfully applied to the crashworthiness design of the new lightweight honeycombs.In addition,combined with the vehicle collision model with high nonlinearity,the proposed method is further successfully applied to the crashworthiness design of the vehicle front-end structure.(5)Multi-objective design optimization using hybrid search algorithms(MDOHSA)is proposed,considering the problem that the current non-preference multi-objective optimization method is easy to lose uncertain information in the structural uncertain design.MDOHSA is mainly composed of modified grey wolf optimization(m GWO)algorithm and pattern search(PS)algorithm.Among them,m GWO is designed by improving the initialization population and update mechanism,which is mainly used to guide the generation of new solution sets in MDOHSA.PS algorithm is used to quickly search the upper and lower bounds of uncertain response in MDOHSA,and the interval possibility comparison method is used to implement the non-dominated strategy in MDOHSA.Finally,after the effectiveness of the relevant numerical simulation model is verified by experiments,MDOHSA is successfully applied to the multi-objective design optimization of vehicle structures such as the vehicle subframe,the anti-collision structure,and the actuator support.