论文部分内容阅读
针对实际工程优化问题,为了提高效率和减少近似模型带来的误差,提出一种基于模型管理的多目标优化方法。利用加强径向基插值函数在整个寻优区域内构造目标和约束的近似模型,结合微型多目标遗传算法寻找当前非支配解。通过模型管理方法更新近似模型,并控制由于近似模型带来的误差和更新次数,最后将误差控制在一定范围内的多个非支配解当作实际问题的解。在测试函数中验证了此方法的效率及非支配解的精度和分布的均匀性。最后成功应用于车身薄壁构件的耐撞性优化中,表明了可用于求解复杂的工程优化问题。
In order to improve the efficiency and reduce the error caused by the approximate model, aiming at the problem of practical engineering optimization, a model-based multi-objective optimization method is proposed. By using the augmented radial basis function, an approximate model of constructing targets and constraints in the entire optimal region is proposed, and the current multi-objective genetic algorithm is used to find the current non-dominated solution. The approximate model is updated by the model management method, and the errors and update times caused by the approximate model are controlled. Finally, a plurality of non-dominated solutions whose errors are controlled within a certain range are taken as solutions to practical problems. The efficiency of this method and the accuracy and uniformity of the non-dominated solutions are verified in the test function. Finally, it is successfully applied in the optimization of crashworthiness of thin-walled body members, which shows that it can be used to solve complex engineering optimization problems.