Abstract: In the digital workshop production,the product variety is diverse,the production cycle is strict,and the logistics demand is complex.It is a difficult problem to make the large-scale Automated Guided Vehicle(AGV)system quickly and accurately carry out path planning.Therefore,an adaptive neighborhood search particle swarm optimization algorithm based on electronic map was proposed.The dynamic obstacle matrix mostly based on AGV position was established,and the improved fitness function was designed.The solution speed in early stage was accelerated by using particle swarm initialization method based on map prior knowledge to overcome the defect that large-scale AGV path planning was easy to fall into local optimal value.A real-time obstacle avoidance strategy was proposed to improve the efficiency of AGV transportation.Python was used to develop AGV path planning platform based on improved particle swarm optimization algorithm.The algorithm was applied to AGV dispatching system of tens of millions of electric meter verification workshop of metrology center of a provincial electric power science research institute,which solved the large-scale AGV path planning problem in complex environment efficiently.

Key words: automated guided vehicle system, path planning, improved particle swarm optimization, neighborhood search, obstacle avoidance

摘要： 数字化车间生产中,产品种类多样,生产节拍严格,物流需求复杂,对大规模自动导引车(AGV)系统快速精确地进行路径规划是一个难题,因此提出一种基于电子地图的自适应邻域搜索粒子群算法。首先建立了以自动导引车实时位置为主的动态障碍物矩阵,设计了改进的适应度函数,并通过基于地图先验知识的粒子群初始化加快前期求解速度,以克服大规模自动导引车路径规划易陷入局部最优值的缺陷;其次提出了实时避障策略以提高自动导引车运输效率;最后采用Python开发出基于改进粒子群算法的自动导引车路径规划平台。将该算法应用于某省电力科学研究院计量中心自动化千万级电表检定车间的自动导引车调度系统,有效地解决了复杂环境中大规模自动导引车路径规划问题。

关键词: 自动导引车系统, 路径规划, 改进粒子群算法, 邻域搜索, 避障