Traffic Signal Timing With Multi-objective Robust Optimization And Decision-making Analysis Under Mixed Traffic Condition

Optimization design of signal timing parameters is the core focus of the traffic signal control research. It is a very complex and important problem in practice, involving priority assignments of different traffic compositions and traffic benefits. In the context of promoting travel experience, improving the optimization methods of signal timing become essentially, due to sustainable and stable operation benefits of urban roads are the urgent needs for intelligent traffic management. Therefore, it is necessary to study the robust optimization problem of signal timing, with the purpose of providing solut ions for alleviating traffic congestions, as well as to create a safe and comfortable travel environment.Mixed traffic is the typical characteristic in China, various traffic flows including vehicles, non-motorized vehicles and pedestrians move in urban road. Based on this characteristic, this paper concentrates on the robustness analysis of signal timing plans, with taking vehicle benef its, slow traffic flow benefits and environmental benefits into consideration. In order to improve robustness of signal c ontrol system, the cycle time disturbance and traffic volume fluctuation are chosen as research objects, which have obvious influences on optimal solutions.The main scientific research works and results are organized as follows:1. Research works on the signal timing at home and board are summarized, as well as the related mult i-objective robust optimization and multi-attribute decision-making analysis in the first part of this paper. Then the basic theoretical knowledge of traffic signal timing is introduced, especially the traffic performance evaluation indexes and the optimization parameters.2. Aiming at the robustness of cycle time disturbance, this paper presents a multi-objective robust optimization and decision making analys is(MRODMA). Firstly, the performance indexes are selected with the purpose of describing the signal timing problem as a multi-objective robust optimization problem. Secondly, the modified degree of robustness is chosen as the robustness measure, and the average effective function is established. Secondly, the non-dominated sorting genetic algorithm II based on improved degree of robustness(IDR-NSGA-II) is proposed to solve average effective function, and obtain the Pareto front. Thirdly, considering the preference demands, a multi-attribute decision making method called minimum deviation analys is for subjective and objective information(MDASOI) is proposed to select the satisfactory solution from the Pareto front. Numerical experiment results confirm the feasibility and validity of the MRODMA method.3. After considering the cycle time disturbance factor, the robustness of traffic volume fluctuation is analyzed. A robust optimization model for traffic delay(ROMTD) is proposed in this paper to enhance adjustment ability of signal timing plans, when traffic flow arrives randomly. This model bases on standard deviation of traffic delay, which strengths robustness of signal timing plans. After solving the multi-objective signal timing model by IDR-NSGA-II, alternative timing plans is obtained. With the intention of expressing nonlinear feature of preference information better, a multi-attribute intelligent decision making analysis based on extreme learning machine(ELM-MIDMA) is presented to select the satisfactory alternative. Simulation results show that the multi-objective robust optimization and intelligent decis ion making analys is(MROIDMA) has preferable potential in practice.