基于PSO-MAs算法的产品组合问题研究

PSO based Memetic Algorithms for Product Mix Problems

针对多约束的产品组合问题,提出一种基于PSO的Memetic算法。该算法首先运用约束理论识别并剔除非瓶颈约束,然后基于伪效用比率设计了一个局部搜索算法,并将其加入到PSO算法的种群进化中,以增强PSO算法的局部学习能力。通过对算法在小规模和大规模算例中测试,表明该算法在小规模问题中优于许多已有算法,同时能在相对较短地时间内更有效地求解较大规模产品组合问题。因此本文提出的基于PSO的Memetic算法可以用来有效地求解实际中的产品组合问题。

To solve the product mix problem with multiple Constraints, a memetic algorithms is proposed based on particle swarm optimization（PSO）. Firstly, the problem is simplified by recognizing and removing the non-bottle- necks based on the Theory of Constraints（TOC）. Secondly, a pseudo utility ratio based local search is proposed to improve the exploitation ability of PSO. Both small-scale benchmark datasets and a group of randomly genera- ted large-scale examples are used to test the proposed approach on solving the product mix problems. The compu- tational results show that the proposed approach outperform some existing approaches, such as TOC, revised TOC, Tabu Search （TS） , Simulated Annealing（SA） and Genetic Algorithms（GA） , and can solve the large-scale problems more effectively. Hence, the proposed approach can be accepted as a practical approach to solve the product mix problem.