个人快速交通动态任务分配问题的优化研究

Research on Optimization of Dynamic Task Allocation in Personal Rapid Transit

个人快速交通(PRT)是一种新型的公共交通工具。由于此系统中的车辆仅在乘客需要时移动,因此这种特殊的按需特性会造成运输能力的高度浪费。个人快速交通系统优化的目标是制定一个力求为所有行程请求提供服务的任务分配策略,在满足每辆车的电池容量的前提下,找到里程利用率和乘客等待时间之间的最佳权衡。首先基于NetLogo平台建立了以里程利用率和乘客等待时间为优化目标的PRT系统多智能体模型,提出了一种基于淘汰机制的粒子群算法(EBPSO)求解系统中的动态任务分配问题。所提算法在不损失里程利用率的前提下,相比标准粒子群算法使平均等待时间和最长平均等待时间分别降低了47.95%和41.31%;相比仅改进适应度函数的粒子群算法使平均等待时间和最长平均等待时间分别降低了11.17%和14.85%。仿真结果表明,该算法在解决PRT车辆动态任务分配问题上与标准粒子群算法相比使系统效能大大提高。

Personal Rapid Transit(PRT)is a new type of public transport.Because vehicles in this system move only when passengers need it,this special on-demand feature can cause a high waste of transport capacity.The goal of the system optimization is to develop a task allocation strategy to provide services for all travel requests,and find the best balance between mileage utilization and passenger waiting time on the premise of meeting the battery capacity of each vehicle.Firstly,based on NetLogo platform,a multi-agent model of PRT system with mileage utilization and passenger waiting time as optimization objectives is established,and a particle swarm optimization algorithm based on elimination mechanism(EBPSO)is proposed to solve the dynamic task allocation problem in the system.Compared with the standard particle swarm optimization algorithm,the proposed algorithm reduces the average waiting time and the longest average waiting time by 47.95%and 41.31%respectively without losing the mileage utilization rate.Compared with the particle swarm optimization algorithm which only improves the fitness function,the average waiting time and the longest average waiting time are reduced by 11.17%and 14.85%respectively.The simulation results show that the proposed algorithm greatly improves the system efficiency compared with the standard particle swarm algorithm in solving the dynamic task allocation problem of PRT vehicles.