基于正反馈机制的蚁群算法,在进行全局搜索时,具有很强的全局收敛能力;遗传算法则具有快速的全局搜索能力.为了充分利用两种算法在寻优过程中的优势,提出一种带有参数自适应调节能力的混合算法.该算法利用灰预测对最大最小蚁群策略中的信息素上(下)界进行估计,以达到实时控制信息素限界、避免算法陷入局部最优的目的.同时,通过云模型建立了一系列的关联规则,利用算法在迭代过程中的反馈信息,可实现算法参数的自适应控制,有效减小算法对参数初始设置的依赖.最后,对车间调度问题(JSP)和旅行商问题(TSP)算例的仿真结果证明了算法的有效性.
Ant colony algorithm with positive feedback has a good capability of global convergence;while the genetic algorithm(GA) is with a fast performance in global search.A hybrid algorithm with adaptive parameters is proposed to take advantages of the above two optimization algorithm.Using the grey prediction,we obtain in the ant colony strategy the estimates of the maximum(minimum) trail limits which are controlled for avoiding the immature convergence.Meanwhile,we employ the cloud models to build a set of association rules which are used to adaptively adjust algorithm parameters by information feedback during the iterative process,thus reducing the reliance on initial parameters.Simulation results for job-shop scheduling problem(JSP) and traveling salesman problem(TSP) validate the algorithm.