一种改进的目标级联多学科设计优化方法  

AN IMPROVED TARGET CASCADE MULTIDISCIPLINARYDESIGN OPTIMIZATION ALGORITHM

作　　者：董明刚[1] 郑伟程 Dong Minggang;Zheng Weicheng(College of Information Science and Engineering,Guilin University of Technology,Guilin 541000,Guangxi,China)

机构地区：[1]桂林理工大学信息科学与工程学院,广西桂林541000

出　　处：《计算机应用与软件》2023年第11期36-42,79,共8页Computer Applications and Software

基　　金：国家自然科学基金项目(61563012,61203109);广西自然科学基金项目(2014GXNSFAA118371);广西嵌入式技术与智能系统重点实验室基金项目(2018A-04)。

摘　　要：在实际大规模复杂工程任务的处理中,因目标级联分析法(ATC)在初始设计变量的选择上具有一定随机性,增加了优化过程中变量协调的复杂度,导致解的质量欠佳。为此,提出一种改进的OLHD-SS-ATC方法。通过引入最优拉丁超立方设计方法选择合理的初始设计变量值,降低优化系统中变量之间协调的复杂度,并将模拟退火算法和序列二次规划算法分别应用到ATC方法的系统和子系统层次优化中,通过系统级和子系统级之间的循环优化,最终获得满足系统一致性约束条件的最优解。实验结果表明,该方法在齿轮减速器最小质量的优化中发挥了明显效果,在满足整体性约束的条件下找到了最优解,表明该方法对于解决复杂工程问题的有效性和可行性。In the actual processing of large-scale complex engineering tasks,the target cascade analysis method(ATC)has a certain randomness in the selection of initial design variables,which increases the complexity of variable coordination in the optimization process,resulting in poor solution quality.For this reason,an improved OLHD-SS-ATC method is proposed.By introducing the optimal Latin hypercube design method to select reasonable initial design variable values,the complexity of coordination between variables in the optimization system was reduced,and the simulated annealing algorithm and sequential quadratic programming algorithm were respectively applied to the system and subsystem levels of the ATC method in optimization.Through the loop optimization between the system level and the subsystem level,the optimal solution that met the system consistency constraints was finally obtained.The experimental results show that the method has a significant effect in the optimization of the minimum mass of the gear reducer,and the optimal solution is found under the condition of satisfying the overall constraints,which proves the effectiveness and feasibility of the method for solving complex engineering problems.

关 键 词：目标级联分析法 最优拉丁超立方 层次优化 齿轮减速器 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]