蚁群优化的简化

为了降低使用蚁群优化求解困难的组合优化问题的复杂性,将问题的启发式信息融合进信息素的初始化中,在解的构造过程中不再考虑问题的启发式信息.这样就消除了解的构造规则中平衡信息素信息和启发式信息的两个控制参数.三种蚁群优化模型在小规模的旅行商问题上的期望迭代质量表明,简化后的比经典的需要更多的迭代步到达最优解,但比不考虑启发式信息的需要少得多的迭代步;另一方面,在每个迭代步,简化后的比经典的需要更少的CPU时间.在中等规模的TSP算例上的试验结果也证实了这个结论.因此简化后的蚁群优化保持了原有的性能且降低了使用复杂性.

Simplification of ant colony optimization

To reduce the number of parameters of ant colony optimization (ACO),the heuristic information of the problem instance was integrated into the initialization of the pheromone model,and then ACO was sped up. The expected iteration quality of ACO models show that the simplified version needs more iterations to be converged than the original one. Because each iteration of the simplified version needs much less time than that of original ACO,the numerical experimental results on traveling salesman problem (TSP) instances show that the simplified ACO takes less time than the original one to obtain solutions with the same quality. Thus,the simplified ACO is easy to be used and has the same performance as the original one.