摘要
多智能体交通仿真系统出于减少通信量的目的,在采用虚拟交警对交通区域实施空间分割后,局部区域中的智能体不可能也无法感知处于不断动态变化的整体交通环境,也就无法适时地疏导交通,以致会引起系统中的某些控制节点出现死锁,进而会引发全局死锁效应。为此,采用基于触发器消息的智能体通信机制,通过在发生死锁时系统发送专用的死锁消息给处于死锁节点处负责调度指挥的智能体,使其在提供的死锁消息处理函数中采用基于队列结构的运动物体调度策略来解除死锁,进而实现死锁节点处运动物体之间的避让效果,以解决控制节点处的死锁现象。仿真实验结果表明,所提出的死锁处理算法可有效处理多智能体交通仿真系统中的控制节点死锁现象,能在最大程度上提高实时交通自动控制的效率和效益。
After the whole traffic region is partitioned by the virtual traffic police for the purpose of reducing traffic in the multi-agent traffic simulation system, and the traffic cannot be directed in time because the dynamic overall traffic environment is unobserved for those agents in local region. For this reason, those regional deadlocks will be triggered in some control nodes, which result in the global dead- lock effect. To solve this problem, using the communication mechanism based on trigger message, the system sends a specialized deadlock message to the agent responsible for the traffic command of the control node when deadlock occurred. By using the method of dispatching moving objects based on queue structure in the message processing function for deadlock resolution,avoidance between the moving ob- jects can be achieved to solve the problem of the deadlock. The simulation results show that the proposed algorithm can effectively settle those deadlocks at the control nodes in the multi-agent traffic simulation system, and farthest promote the efficiency and benefit of dy- namic traffic auto-control.
出处
《计算机技术与发展》
2017年第5期25-29,共5页
Computer Technology and Development
基金
四川省应用基础项目(2010JY0023)
关键词
仿真
智能体
死锁
调度
simulation
agent
deadlock
dispatching
