针对我国城市道路相邻交叉口混合交通流环境下智能信号控制中的不确定性和效率问题, 提出一种基于鲁棒多目标优化算法的优化控制方法. 设计了相邻交叉口的鲁棒多目标信号优化控制模型, 并提出一种新的鲁棒多目标进化算法 IDR-NSGA-Ⅱ, 通过对自适应抽样技术、鲁棒度定义、鲁棒偏序关系定义等多项关键技术的综合改进, 提升了算法的求解精度和运行速度. 提出新的多属性决策方法 ELM-MADMA 来选择配时方案. 上海市相邻交叉口控制的仿真实验结果表明: IDR-NSGA-Ⅱ 算法能够有效地实现周期时长扰动和交通流波动下机动车平均延误、道路通行能力、慢行交通平均延误、机动车平均停车率等多项性能指标的最优化控制; 与其他决策方法相比, ELM-MADMA 能够较好地进行决策, 提升相邻交叉口智能信号控制的效率.

To improve efficiency and deal with uncertainty of intelligent signal control for mixed traffic flow at signalized adjacent road intersections in cities, a robust multi-objective optimization control model is established to deal with uncertainty due to cycle length disturbance and traffic flow fluctuation. An intelligent optimization control method based on robust multi-objective evolutionary algorithm, IDR-NSGA-II, is proposed. Key techniques are improved, including adaptive sampling, definition of robustness degree and robust partial order relation, for better performance and running speed of the algorithm. A new multi-attribute decision making analysis method, ELM-MADMA, is presented to select a satisfactory solution from the Pareto front set. Simulation of an adjacent intersection in Shanghai shows that the proposed control method can simultaneously optimize several traffic performance indicators such as vehicle average delay, road capacity, chronic traffic average delay and average parking rate of motor vehicle under cycle length disturbance and traffic flow fluctuation. The ELM-MADMA algorithm outperforms other decision making methods, with improved system efficiency.