Development of car-crash accident simulator based on Petri nets and its application to anti-accident measures for dangerous points

(1)内置车祸风险的plot-type模拟器的开发：我们开发了Petri-net交通模拟器（PETS），用于交通拥堵分析及对策。在这一发展过程中，我们非常重视确保道路网络中任何地方的安全交通流，并在我们提出的交通流的Petri网表示中建立了两种类型的机制，分别称为自治型和力型。这些机制会因为所谓的“杀手网”的连接而崩溃。在此，我们将“代币突然诞生”、“代币淡出”、“代币非标记”三种网络发展为杀手网。我们还开发了一个网络来检测和规范事故情况，并允许我们继续模拟。将这两种类型的网络与汽车交通网络连接，可以构建汽车碰撞仿真网络，并可以直接由我们的PETS软件执行。(2)驾驶员动作过程子网络的发展：为了完善安全机制的崩溃，我们首先考虑了危险点的风险事件的产生，并将其描述为与驾驶员动作过程网络中的强度相对应的彩色标记。其次，我们假设驾驶员可以根据情况判断采取紧急制动，并将新开发的驾驶员网与行车子网的连接修改为将紧急制动既视为恢复损坏的安全机构，又视为后续车辆发生危险。(3)追尾事故仿真结果：通过将两组驾驶员子网连接到代表130 ~ 200 m车道的车辆运行子网，构建交通流网络。在车流量为15（辆/分钟）且事故发生率较高的情况下，对追尾事故进行了仿真，结果表明，增大最小车距和抑制加速欲望都有助于减少仿真系统中的事故数量。我们还证明，如果我们成功地扩大司机的谨慎，我们可以显著减少事故的数量。我们需要将我们的方法扩展到十字路口的正面和右转事故。我们还需要包括老年和青年司机的特点子网。少

(1) Development of plot-type simulator with built-in car-crash risks :We have developed Petri-net traffic simulator (PETS) for traffic congestion analysis and its countermeasures. In this development, we paid much attention to ensure safety traffic flows anywhere in road network and built two types of mechanisms, named as autonomous and forcec type, in our proposed Petri nets representations of traffic flows. These mechanisms will collapse by connection of called killer nets. Here we develop nets of "suddenly birth of token", "fade-out of token", and "non-marking of token"as killer nets. We also develop a net that detects and normalizes accident situation and allow s us to continue simulation. Car-crash simulation nets can be constructed by connection these two type nets with car traffic nets and can directly be executed by our PETS software.(2) Development of Driver's Action Process Sub-net :To refine collapse of safety mechanisms, we first take into consideration of birth of risky oc … More casions at dangerous points and describe them as colored tokens corresponded to their intensity in driver's action process net. Secondly, we assume that driver could take emergency breaking according to circumstantial judgement and modify the connection of newly developed driver's net to car-running sub-net as taking emergency breaking both as recovery of broken safety mechanism and as occurrence of hazard for following cars. We also here include function of driver's caution in this process net(3) Results of Rear-end Collision Accident Simulation :We construct traffic flow net by connecting two set of driver's sub-net to car-running sub-net that represents a lane of 130〜200 m. And we do some simulation of rear-end crash accident, as assuming that the traffic volume is 15 (vehicle/min) and the birth rate of occasion is high, and show that expansion of minimum distance between cars and suppression of acceleration desire both contribute to reduce the means of number of accidents in our simulation system. We also demonstrate that if we success to expand driver's caution we could reduce the number of accident as remarkably. We need expand our method to head-on and right-turn accident at intersection. We also need include characteristics of elderly and young driver's sub-net. Less