基于绿色模块划分的多目标蚁群算法

Multi objective ant colony algorithm based on green module

针对模块划分问题中基本单元本身又是一个零件库的情况,基于绿色模块划分的多目标模型,提出了一种多目标蚁群算法;以减速器为实例,建立聚合度、耦合度、绿色度3个目标函数,然后在同一个种群中,又分出3个蚂蚁群体,每个蚂蚁群体最优一个目标函数,三者共享全局信息素,设置合理启发式因子、信息素更新原则和3个外部精英解,以均值来解决拥挤距离维护外部精英集。3个蚂蚁群体中的解分别与外部精英解比较,如果是非支配的就更新全局信息素;如果是互不支配的就按照均值来更新全局信息素,其他只考虑信息素的挥发。实验证明收敛速度极快,容易找到最优解,对各个因子和参数的改变算法均表示稳定收敛,验证了方法的可行性,具有极好的实用性。

Aiming at the module partition problem of the basic unit itself is a parts library,based on the multi objective model of green module division,this paper presents a multi-objective ant colony algorithm; the reducer as an example,establish the degree of polymerization,coupling degree,green degree three objective functions,and then in the same population,and divided into three groups each ant,ant colony optimization of an objective function,the global pheromone three share,set reasonable heuristic factor,pheromone update rule and three external elite solution,to mean to solve the crowding distance to maintain external elite set. Three ant colonies solution respectively compare with external elite solution,if the non dominant update global pheromone; if it is not dominated by the mean to update global pheromone,others only consider the volatile pheromone. The experiments show that the convergence speed,to find the optimal solution easily,change the algorithm to each factor and parameter indicated stable convergence,verify the feasibility of the method,the practicability is excellent.