基于PSO-ABC的混合算法求解复杂约束优化问题

Hybrid algorithm for solving complex constrained optimization problems based on PSO and ABC

为了改善粒子群优化(particle swarm optimization,PSO)算法在处理复杂约束优化问题时的求解效果,提出了一种基于粒子群和人工蜂群的混合优化(particle swarm optimization-artificial bee colony,PSO-ABC)算法。在采用可行性规则进行约束处理的基础上,将PSO种群分为可行子群和不可行子群,并在ABC算法从粒子种群中选择蜜源时,保留部分较优的可行解信息和约束违反程度较低的不可行解信息,弥补了联赛选择算子在处理最优点位于约束边界附近的问题时存在的不足。同时,使用禁忌表存储局部极值,减小了PSO算法陷入局部最优的危险。针对4个标准测试实例的实验结果表明,该算法能够寻得更优的约束最优化解,且稳健性更强。

In order to improve the performance of particle swarm optimization （PSO） in complex constrained optimization problems, a hybrid method combining PSO and artificial bee colony （ABC） is proposed. A feasibility-based rule is used to solve constrained problems, and the particle swarm is divided into feasible subpopulation and infeasible subpopulation. Some PSO particles containing the information of better feasible solutions and smaller constraint violation infeasible solutions are selected as food sources for ABC algorithm, which can make up for the tournament selection operator being invalid when the optimum is close to the boundary of constraint conditions. And the tabu table is used to save the local optimization results so as to avoid PSO trapping into local optimum. The algorithm is validated using four well-studied benchmark problems, and the results indicate that the PSO-ABC algorithm can find out better optimum and has a stronger solidity.