短时预测下的单点交叉口无模型自适应控制方法

Model-free Adaptive Control Method for Isolated Intersection Based on Short-term Prediction

为使单点信号控制方案能够对交叉口的交通状态趋势进行有效预测,使其更好地适应交通状态的动态变化需求,提出了一种短时预测下的单点交叉口无模型自适应控制方法。首先依据历史交通流量数据对交叉口的交通状态进行界定,制定交叉口交通状态-信号周期模板;然后以15min为时间间隔对交通流的发展趋势进行预测,进而依据预测的交通状态确定未来15min的战略信号周期,同时对每个信号周期和各相位的绿信比进行小步距调整,以期逐步降低单点交叉口的饱和度。研究结果表明:提出的方法能够对未来15min的交通状态趋势进行有效预测,且在控制过程中既能够从宏观上事先调整控制方案以适应交通流的发展趋势,又可从微观上对配时参数进行小步距微调,满足相位间需求的差异性。

In order to make the isolated traffic signal control plan predict the traffic trend of isolated intersection effectively and adapt the plan to dynamic change of traffic flow, a model-free adaptive control method for the isolated intersection based on short-term prediction was proposed. Firstly, the traffic states of an isolated intersection were defined according to the historical traffic data, and the template of traffic state-signal cycle was designed; then a prediction method with an interval of 15 min was used for traffic tendency and the strategic signal cycle in next 15 min was determined with the predicted traffic states. Moreover, each signal cycle and the green split of each phase were tuned to reduce the saturation of the isolated intersection progressively. The results show that the presented method can predict the traffic states of an isolated intersection in next 15 min successfully, and can not only adjust traffic control plan to meet the predicted traffic trend, but also can tune traffic control parameters elaborately to satisfy different needs between phases.