Real-time Signal Control Combined Model And Algorithm Based On Dynamic O-D Estimation And Multi-objective Optimization

In recent years,Intelligent Transport System(ITS)develops rapidly,Intelligent detection technique can provide short-term traffic flow data for Advanced Traffic Management System(Advanced Traffic Management System,ATMS)and create positive conditions for it to realize signal timing of real-time signal control models to change with short-term traffic flows.As traffic detectors can only obtain link flows,dynamic origin-destination(O-D)flows which are important input parameters of real-time signal control models should be estimated by dynamic O-D estimation.In addition,real-time signal timing variations in actual network will also influence dynamic O-D flows.Therefore,dynamic O-D estimation and real-time signal control both as significant basic theory of ATMS closely interact with each other.Obviously the combined model of dynamic O-D estimation and real-time signal control will help to ease traffic congestion,improve efficiency of urban road network and reduce its delay and queue length by adjusting signal timing to change with short-term traffic flows and moreover it will make traffic management and control to be more accurate.In this paper,dynamic accumulative flows in entering approaches were put forward to consider mutual influence of signal control and dynamic O-D estimation each other.Real-time signal control combined model for intersections,arterials and urban networks based on dynamic O-D estimation and multi-objective optimization were built up based on the minimum objective of dynamic accumulative flows in entering approach.The combined models could modulate signal timing according to variations of real-time dynamic O-D flows to reduce delay and stop rates and improve effective capacity.Algorithms of the models were proposed and results of case study could illustrate the validity of the model.Main works include the following 3 aspects:1.Signal control combination model for intersections.Intersection combined model belongs to parameter optimization models because dynamic O-D estimation model is a minimum square sum model constructed based on the equality relationship between entering and exiting flows and signal control model is a minimization model of dynamic accumulation flows in entering approaches.This paper applied relaxation strategy to make above bi-level optimization model to become single-level model combined dynamic O-D estimation model with signal control model by penalty function weighted method.The single-level combined model could be solved by exact penalty function algorithm.Lastly,case study compared results of single-level combined model with varied phase sequence,single-level combined model with fixed sequence and actual signal timing using average delay and average stop rates and effective capacity.2.Signal control combination model for arterials.This paper put forward computing method of travel time and dynamic O-D distribution ratios based on space and time sequence discretization for arterials.Besides,an “estimation+distribution+correction” three step dynamic O-D estimation model based on Kalman filter was proposed to obtain dynamic O-D flows influenced by signal timing.Afterwards,dynamic accumulation flows in entering approaches was regarded as a transmit variable between dynamic O-D estimation and multi-objective signal control and signal timing computing method for arterials under varied offsets was designed.Eventually,case study contrasted average delay and average stop rates and effective capacity results of arterial model with actual signal timing.3.Signal control strategy for regional road network.Signal control strategy for regional road network was based on arterial model and the solving solution and technical framework of real-time signal control strategy for regional road network based on dynamic O-D estimation and multi-objective optimization were presented.Traffic simulation was applied to achieve route choices for regional road network to be used as input data of dynamic O-D estimation and signal control.Afterwards,dynamic accumulation flows in entering approaches was regarded as a transmit variable between dynamic O-D estimation and multi-objective signal control.For one thing,a two-step dynamic O-D estimation model was put forward based on Kalman Filtering model which estimated intersection O-D flows at first and then estimated regional road network and dynamic accumulation flows in entering approaches were put into the observation equation and travel time calculation;For another thing,traffic signal timing parameters,dynamic distribution proportion and dynamic O-D flows were fed into queue dispersion model to get dynamic accumulation flows in every entering approaches in every inersection and then signal timing could be solved out.