摘要
在对蚁群优化算法(ACO)和粒子群优化算法(PSO)进行分析的基础上,提出一种解决函数连续优化的群智能混合策略——CA-PSO。在求解过程中,首先对解空间进行区域划分,进而利用ACO在优化初期具备的快速收敛性能,在整个解空间内搜索最优解的敏感区域。然后利用蚁群的搜索结果初始化PSO粒子,利用PSO快速和全局收敛性进行所在小区域内的搜索。种群更新时根据蚁群的拓扑结构和小区域间的阶跃规则,蚁群不断向最优解敏感区域聚集,使得敏感区域内粒子数增加,则局部的PSO搜索策略可以更细密的搜索最优。实例结果表明,CA-PSO既能保证解的分布性与多样性,又避免了在多峰值函数寻优过程中陷入局部最优解而停止运算,最终将收敛到全局最优解。
On the basis of the analyses of ant colony optimization (ACO) and particle swarm optimization (PSO), continuous antparticle swarm optimization (CA-PSO) applied in continuous function optimization is proposed. After the space partition is properly employed, ACO is applied to search the sensitive areas through the whole solution space. And then PSO is initialized according to ACO search results and applied to search the optimal solution in every area. After each iteration, swarm renovation is processed following ant-colony topological structure and transfer roles. As a result, the ants gradually assemble in the areas where optimal solution is mostly likely to be. With the number growth of particles in the sensitive areas, local PSO can search more meticulously until find the optimal solution. In conclusion, the CA-PSO can not only ensure the distributing diversity of the solution, but also avoid suboptimal solution when it is applied in the Multimodal function optimization.
出处
《计算机工程与设计》
CSCD
北大核心
2009年第8期1969-1973,共5页
Computer Engineering and Design
基金
国家自然科学基金项目(NO6047200)
关键词
连续优化
蚁群优化算法
粒子群优化算法
群集智能
空间划分
拓扑结构
continuous optimization
ant colony optimization (ACO)
particle swarm optimization (PSO)
swarm-intelligence
space partition
topological structure