A total generalized optimal velocity model and its numerical tests

A car-following model named total generalized optimal velocity model （TGOVM） was developed with a consideration of an arbitrary number of preceding vehicles before current one based on analyzing the previous models such as optimal velocity model （OVM）, generalized OVM （GOVM） and improved GOVM （IGOVM）. This model describes the physical phenomena of traffic flow more exactly and realistically than previous models. Also the performance of this model was checked out by simulating the acceleration and deceleration process for a small delay time. On a single circular lane, the evolution of the traffic congestion was studied for a different number of headways and relative velocities of the preceding vehicles being taken into account. The simulation results show that TGOVM is reasonable and correct.     

基于粒子群优化的RBF神经网络交通流预测

交通流量预测一直是实时自适应交通控制的关键问题。以城市道路网络中典型的两相邻交叉口为研究对象,提出了基于粒子群优化的RBF神经网络的信号交叉口交通流量预测模型。该模型以RB F神经网络为基础,采用分组优化策略,用粒子群优化算法对基函数的中心、方差和RBF网络权值进行优化,从而提高了网络的预测精度。通过仿真,并与其他算法对比,表明了本文方法的有效性。     

OV 模型交通流阻尼特性研究

对交通流跟驰模型施以拉氏变换（或z 变换）,转换为控制系统传递函数形式, 进而通过分析传递函数来探讨交通流系统稳定性及控制方法,这是交通流稳定性理论研 究重要方向之一.为进一步挖掘分析稳定交通流系统在稳定状态下动态特性,基于最优速 度跟驰模型,以控制理论时域分析方法,研究了交通流系统在欠阻尼和过阻尼两种情况 下阶跃响应动态特性,并以仿真算例验证了理论正确性.过阻尼较欠阻尼交通流系统稳定 性好,但调节时间更长,若将交通事件抽象为阶跃响应,通过分析交通流系统调节时间、 超调量等参数,可为计算交通事件持续时间作参考,进而制定合理有效的交通管理措施.