多样性保持离散差分进化算法及齿轮传动优化应用

发布时间：2018-03-18 00:03

本文选题：离散约束优化　切入点：差分进化算法　出处：《机械工程学报》2016年21期 　论文类型：期刊论文

【摘要】：将离散约束优化问题转化为等效约束整数规划问题(Constrained integer programming problem,CIPP),并应用离散差分进化(Discrete differential evolution,DDE)算法求解这类问题。提出DDE算法种群维分量一致的概念,并揭示种群维分量呈现一致状态将导致进化停滞的缺陷,数值算例仿真结果验证了理论分析的正确性。在引入定量刻画DDE算法种群多样性指标——种群多样度Γ的基础上,提出一种基于种群连续进化停滞代数和Γ监测的种群多样性保持策略,以克服种群维分量一致进化缺陷,进而增强算法全局优化能力。将多样性保持策略嵌入DDE算法,并结合可行性规则约束处理技术,形成求解CIPP的多样性保持离散差分进化算法(DDE algorithm with diversity maintenance strategies,DMDDE)。建立某刮板输送机所用二级斜齿圆柱齿轮传动体积最小化设计的离散约束优化模型,再采用DMDDE算法求解该问题。数值试验表明,DMDDE算法的鲁棒性和收敛精度优于4种对比算法。优化结果明显好于原设计方案,齿轮体积较原设计下降40.8%。
[Abstract]:The discrete constrained optimization problem into an equivalent constrained integer programming problem (Constrained integer programming problem, CIPP), and the application of discrete differential evolution (Discrete differential evolution DDE) algorithm to solve this kind of problem. The DDE algorithm proposed dimensional components consistent with the concept of population, and to reveal the population dimensional component showed a consistent state will lead to evolutionary stagnation defects a numerical example and simulation results verify the correctness of the theoretical analysis. Based on the introduction of quantitative characterizations of DDE algorithm the population diversity index of population diversity, presents a continuous population evolutionary stagnation algebra and gamma monitoring of population diversity maintenance strategy based on the population to overcome the heft of the same evolutionary defects, thus enhancing the algorithm the global optimization ability. The diversity maintenance strategy into DDE algorithm, and combining feasible rules processing technology, the formation of diversity for CIPP Keep the discrete differential evolution algorithm (DDE algorithm with diversity maintenance strategies, DMDDE). Set up two grade helical cylindrical gear discrete optimization model of minimum volume design used for a scraper conveyor, then use DMDDE algorithm to solve the problem. Numerical experiments show that the 4 kinds of contrast DMDDE algorithm robustness and convergence precision. The optimization results are significantly better than the original design, gear volume compared to the original design by 40.8%.

【作者单位】：重庆工商大学制造装备机构设计与控制重庆市重点实验室;重庆工程职业技术学院机械工程学院;
【基金】：重庆市教育委员会科学技术研究计划(KJ1403201) 制造装备机构设计与控制重庆市重点实验室(重庆工商大学)开放基金(611115006) 重庆市基础科学与前沿技术研究专项(cstc2015jcyjA 70006)资助项目
【分类号】：TH132.41

【相似文献】