基于粒子群遗传算法的泊车系统路径规划研究  

Research on path planning of parking system based on PSO-genetic hybrid algorithm

作　　者：王辉[1] 朱龙彪[1] 朱天成 陈红艳[1] 邵小江 朱志慧 

机构地区：[1]南通大学机械工程学院,江苏南通226019 [2]中国联合通信网络有限公司江苏省分公司,江苏南京210024 [3]江苏金冠立体停车股份有限公司,江苏南通226003

出　　处：《工程设计学报》2016年第2期195-200,共6页Chinese Journal of Engineering Design

基　　金：国家自然科学基金资助项目(51405246);江苏省产学研联合创新基金资助项目(BY2014081-07)

摘　　要：针对智能停车库自动导引运输车(automated guided vehicle,AGV)存取车路径规划问题,提出了一种基于粒子群和遗传算法的动态自适应混合算法.在标准粒子群算法和遗传算法的基础上,通过引入动态自适应调整策略分别对惯性权重系数、学习因子以及交叉变异概率公式进行了优化.在进化初期,通过在惯性权重系数和学习因子之间建立动态联动关系来实现对粒子速度和位置的实时有效更新;在进化后期,通过引入自适应遗传算法的交叉、变异操作来增强混合算法的全局搜索能力,提高算法的进化速度和收敛精度.为验证混合算法的可行性和有效性,选用MATLAB软件对其进行仿真测试.仿真测试结果显示,与禁忌搜索算法、蚁群算法以及遗传算法相比,混合算法表现出较强的全局搜索能力和较好的收敛性能,表明混合算法可行和有效.Aiming at path planning problem of AGV accessing cars, the dynamic adaptive hybrid algorithm was proposed by combining particle swarm optimization （PSO） with genetic algorithm （GA）. Based on the standard PSO and GA, inertia weight coefficient, learning factor and the formula of crossover and mutation probability were optimized and improved with the dynamic adaptive adjustment strategy. In the initial stage of algorithm evolution, the dynamic linkage relation built between inertia weight coefficient and learning factor were utilized to deal with real-time updates of particle＇s velocity and position. In the later stage of evolution, in order to strengthen global search ability of hybrid algorithm and improve evolution speed and convergence precision of the algorithm, crossover and mutation operators of adaptive genetic algorithm （AGA） were introduced. Finally, MATLAB was used to verify the feasibility and effectiveness of hybrid algorithm. The simulation results showed the global search ability and convergence performance of hybrid algorithm were optimal by being compared with tabu search algorithm （TSA）, ant colonyalgorithm （ACO） and GA. The conclusion indicates that the hybrid algorithm is feasible and effective.

关 键 词：粒子群算法 遗传算法 泊车系统 AGV 路径规划 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]