角度惩罚精英策略的绿色供应链伙伴选择研究

Research on Partner Selection in Green Supply Chain Based on Angle Penalty Distance Elite Selection Strategy

针对绿色供应链伙伴选择中的典型高维目标优化问题,以运营成本、配送时间、产品质量和绿色度为优化目标,建立三阶段绿色供应链网络优化模型,提出一种角度惩罚距离(APD)精英策略的高维目标优化算法NS-RVEA进行求解.该算法对APD机制仅使用分解策略优化种群,忽略个体间Pareto关系,易导致种群退化等不足进行改进,引入非支配排序方法,先对子种群内个体进行Pareto非支配排序,再通过APD机制对剩余个体进行筛选,在维持种群多样性的同时提高选择压力和收敛速度.仿真结果表明,NS-RVEA具有收敛性强、全局性好和计算复杂度低等优点,所得解集在IGD*指标、Pareto最优解平均个数和算法运行时间等多个方面均优于MOGA、NSGA-Ⅱ、MOEA/D和RVEA.

Aiming at the problem of the typical many-objective optimization in the partner selection of green supply chain,the three stage green supply chain network optimization model was established by taking operation cost,distribution time,product quality and green degree as the optimization goal,and the many-objective optimization algorithm based on angle penalty distance(APD)elite strategy was proposed to solve this problem.The APD mechanism only uses decomposition strategy to optimize the population,ignoring the Pareto relationship between individuals,and easier leading to population deficiencies.This paper proposed a many-objective optimization algorithm improved by introducing non-dominated sorting methods to first perform Pareto non-dominated sorting of individuals within subpopulations.Then the remaining individuals were screened through the APD mechanism,which increased selection pressure and convergence speed while maintaining population diversity.The simulation results revealed that the proposed algorithm had the advantages of strong convergence,good overall performance and low computational complexity.The set of solutions was superior to the MOGA,NSGA-II,MOEA/D and RVEA in many aspects such as the IGD*index,the average number of the Pareto optimal solutions and the running time.