Improved Efficient Global Optimization Algorithms And Their Application In Lightweight Design Of Vehicle Structures

With the rapid development of the automotive industry,the problem of energy and pollution is becoming more and more serious.The lightweight and crashworthiness design has been two main challenges in the automotive industry at present and in the furture.In the field of lightweight and crashworthiness design,the optimization method based on surrogate model has become an important solution.However,the traditional optimization method based on surrogate model still has problems of high computational complexity,long optimization period,and poor optimization accuracy in solving highly nonlinear optimization problems such as vehicle collisions.At present,the efficient algorithm represented by Efficient Global Optimization(EGO)provides a new solution for solving such problems.However,due to its serial optimization,the computational efficiency is still unable to meet the requirements of most vehicle optimization problems.With the increasing abundance of computing resources,parallel computing provides new ideas for further improving the optimization efficiency of EGO.In order to solve the above problems,this paper improved on the EGO and proposed a parallel EGO algorithm,which was more suitable for solving the optimization design of vehicle lightweight and crashworthiness.This method used the multi-start local optimal strategy as a parallel strategy and controled the number of parallel points and the termination conditions according to the engineering practice.Then,the mathematical benchmark examples were employed to test the parallel EGO algorithm and EGO,which highlighted the advantage and practical application value of the improved algorithm.Finally,the parallel EGO algorithm proposed in this paper was used to optimize the design of tailor rolled blank(TRB)structure.The optimization result significantly improved the crashworthiness on the basis of no increase in mass.In addition,most of optimization problems have multiple objectives in the lightweight and crashworthiness design of vehicle.However,there are few studies on multi-objective EGO algorithms,especially parallel multi-objective EGO algorithms.In this study,two novel parallelized multi-objective efficient global optimization algorithms were proposed,which was based on the multi-objective EGO algorithm proposed by Keane.To implement parallel computing,the “Kriging Believer” and “Multiple Good Local Optima” strategies were adopted;and the non-dominated sorting genetic algorithm II(NSGA-II)was employed as a searching algorithm for generating the Pareto sets.The proposed algorithms were applied to five mathematical benchmark examples first,which demonstrated faster convergence with more diverse and uniform distribution of Pareto points,in comparison with the two other conventional algorithms.The proposed “Kriging Believer” strategy approach was then applied to two more sophisticated real life engineering case studies on the tailor-rolled blank(TRB)structures for crashworthiness design.The good optimization results were obtained.Specifically,the TRB B-pillar was reduced 10.1% in mass and 12.8% in intrusion,simultaneously.These benchmark and engineering examples demonstrated that the proposed methods are fairly promising for being an effective tool for a range of problems.