考虑路径长度与冲突的AGV停车场停车位分配方法

Parking Space Allocation in AGV Parking Lots with Path Length and Conflict Consideration

为解决自动引导小车(Automated Guided Vehicles,AGV)智能停车场中的停车位分配不合理、多AGV协同作业过程中频繁冲突的问题,针对停车位分配问题进行了研究。根据停车场环境布局,采用拓扑图法建立停车场环境地图模型,并提出一种考虑长度与冲突的停车位分配双目标优化模型,从全局层面减少多台AGV之间的冲突。鉴于模型的多目标特征,设计非支配排序遗传算法(Nondominated Sorting Genetic Algorithm-Ⅱ,NSGA-Ⅱ)对停车位分配模型进行求解。为验证模型与算法的有效性,进行停车位分配仿真实验,并将所提出的模型与传统的基于最短停车路径长度的停车位分配方法及空停车位随机分配方法进行对比。实验结果表明,NSGA-Ⅱ算法适用于求解停车位分配模型,在不同任务数量和不同AGV配置数量下,算法均能得到多样的Pareto非劣解集;与传统方法相比,所建模型的总停车路径长度指标与基于最短停车路径长度的停车位分配方法相近,路径冲突概率则比基于最短停车路径长度的停车位分配方法和随机分配方法分别降低67.44%和44.00%;在AGV智能停车场高峰期的连续停车任务中,所提出的停车位分配模型和求解算法可寻找到最优停车位分配方案并从全局角度规划AGV的停车路径。

In response to the issues of unreasonable parking space allocation and frequent conflicts during multi-AGV collaborative operations in AGV parking lots,a study was conducted on parking space allocation within the parking lot.The topological map method was used to establish a parking lot environment map model based on the layout of the parking lot environment,and a bi-objective optimal pattern function for parking space allocation considering length and conflict was proposed to mitigate conflicts between multiple AGVs on a global level.Given the multi-objective characteristics of the model,The NSGA-Ⅱalgorithm was designed to solve the parking space allocation model.In order to validate the effectiveness of the model and algorithm,a simulation experiment was conducted on parking space allocation,and the proposed model was compared with traditional parking space allocation methods based on the shortest parking path length and random allocation of empty parking spaces.The experimental results show that the NSGA-Ⅱalgorithm is suitable for solving parking space allocation models,and the algorithm can obtain a diverse set of Pareto non inferior solutions under different task quantities and AGV configurations.Compared with traditional methods,the total parking path length index of the proposed model is similar to this minimum parking path length method,while the path conflict probability is reduced by 67.44%and 44.00%,respectively,compared to the minimum parking path length method and random allocation method.In the continuous parking tasks of AGV intelligent parking lots during peak hours,the proposed parking space allocation model and solving algorithm can find the optimal parking space allocation scheme and plan the AGV parking path from a global perspective.