基于马尔可夫模型的交叉口两难区自适应控制

提出一种基于马尔可夫模型的信控交叉口两难区自适应控制方法.根据实时监测的两难区内车辆数据,采用马尔可夫模型主动预测陷入两难区内车辆数的概率分布,提出采用基于n近邻的状态转移矩阵的更新框架,并综合考虑相位时长,建立相位延长和切换时两难区内当量车辆数的计算模型,进而以相位切换的风险概率为准则,采用即时决策的交通控制自适应调整相位时长.以广州市某交叉口进行VISSIM仿真实验,在不同强度的交通条件下,验证提出方法的效用并进行参数敏感性分析.仿真结果表明经过模型参数校准后,提出的控制方法在有效减少陷入两难区内的车辆数的同时,可减小交叉口的平均延误.

Markov Model Based Adaptive Traffic Signal Control for Dilemma Zone at Signalized Intersections

This paper presents a markov model based adaptive traffic signal controller for dilemma zone(DZ) at signalized intersections. With the real time traffic data of vehicles trapped in the DZ, the probability distribution of vehicles in the DZ is predicted using markov model, while the state transition matrixes is rolling updated using the n nearest neighbors algorithm. Taking the phase time and the predicted trapped vehicles into account, the model of equivalent number of vehicles in the DZ is developed, and then under the real time decision signal control strategy, the green phase time is adjusted accroding to the defined risk probability of switching phase. Extensive experiments were conducted on a typical isolated intersection in Guangzhou via online VISSIM simulation under different traffic conditions, and the sensitive analysis of model parameters were analyzed in detail. The simulation results have demonstrated that with the calibriation of model parameter, the developed controllers has the great potential in the reduction of vehicles trapped in the DZ, as well as the average traffic delay.