摘要
提出自适应径向基函数代理模型,并结合微型多目标遗传算法对整车耐撞性进行优化。在每个迭代步中,以最优拉丁方进行样本点设计,以遗传最优拉丁方进行测试点设计,通过隔代映射遗传算法对径向基函数代理模型的误差进行评价并获得最优光滑参数。将测试点不断添加到样本空间,直至耐撞性各个目标代理模型在测试点的误差都达到要求,再采用贪婪算法将最后迭代步的测试点筛选到样本空间以进一步提高精度。最后采用微型多目标遗传算法对达到许可误差的各个自适应径向基函数模型进行优化,获得Pareto前沿面,根据工程要求或工程人员的经验权衡耐撞性,优化各个目标之间的关系以获得不同最优妥协解。
An ARBF model method was suggested and combined with micro multi-objective genetic algorithm（μMOGA） to solve vehicle crashworthiness.In each iterative,sampling points were obtained by the optimal Latin hypercube design,while testing points were obtained by the inherit optimal Latin hypercube design,this method regarded the errors of testing points as fitness of intergeneration projection genetic algorithm（IP-GA）,assessed the model systematically and got the optimal smooth parameters to maximize model accuracy,testing points added to sample space until reaching errors allowable of each crashworthiness objective.Then greed algorithm was adopted to filter the testing points from the last iterative to sampling space to increase accuracy.At last,μMOGA was applied to optimize the ARBF,and got Pareto and balanced each objective to get different best compromise solutions according to engineer experiments or engineerring requirements.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2011年第4期488-493,共6页
China Mechanical Engineering
基金
国家自然科学基金资助项目(10725208)
国家重点基础研究发展计划(973计划)资助项目(2010CB832700)
高等学校博士学科点专项科研基金资助项目(200805321034)
国家科技重大专项项目(2010ZX04017-013-005)
关键词
自适应径向基函数
光滑参数
多目标优化
整车耐撞性
adaptive radial basis function（ARBF）
smooth parameter
multi-objective optimization
vehicle crashworthiness
