Research On Diversion Method At Congested Expressway Off-ramp Areas

When the traffic volume of the urban expressway main line entering the off-ramp exceeds the capacity of the off-ramp,vehicle will queue within the off-ramp.When the queuing gradually spreads and overflows to the main line,traffic congestion will form in the area where the main line and the off-ramp are connected.This paper intends to research in traffic congestion optimization methods for this congested area in order to relieve regional traffic congestion and prevent queue overflow on off-ramps.Based on the theory of cell transmission model,the traffic flow model of the connected area is established from the three aspects of the ordinary cellular connection combination,the splitting cell connection combination,and the confluent cell connection combination.The parameters such as free flow speed,capacity,and congestion are calibrated to complete the traffic characteristic analysis of regional road network.According to the application conditions of the user-optimal model,road network traffic condition is classified.Set State 1 is: before the queued vehicles overflow the main line,the road network has reached the user-optimal model;State 2 is: after the queued vehicles overflow the main line,the road network can reach the user-optimal model;State 3 is: the road network cannot reach the user-optimal model.According to the different parameters of diversion,the diversion strategy is divided into the diversion strategy based on the real-time travel time estimation and the diversion strategy based on the number of queuing vehicles on the off-ramp.Based on the different traffic conditions of the road network,the application functions of the traditional bang-bang splitting method are researched separately,and the corresponding induction or control manner and corresponding diversion strategies are given.Since bang-bang method can only set the splitting rate to 0 or 1,there is a problem that it is easy to increase the delay of the road network.Therefore,the paper applies the theory of fuzzy neural network to establish fuzzy neural network diverter.The diverter takes the queuing vehicles on off-ramp and the rate of change of the queuing vehicles on off-ramp as input,takes the splitting rate as the output,and then establishes the fuzzy relation of input and output,and completes the parameters setting by neural network,making the splitting rate between 0 and 1,which reduces the network delay.Take Wufengkou expressway network of Zhenjiang as the example,according to the actual traffic demand,select total travel time and total disbenefit as evaluation index.Implement the regional road network traffic simulation by using the fuzzy neural network splitting method proposed in the paper.Total disbenefit is: when there are several routes from starting point to the end,the time lost because of some drivers choosing the larger time route resulting from unknown of the road.Simulation results show that: if the road network traffic condition is State 1,if take the strategy based on real-time travel time estimation,compared with bang-bang method,fuzzy neural network method can reduce total travel time by 5.3% and total disbenefit by 86.7%,if take the strategy based on the queuing vehicles on off-ramp,the fuzzy neural network method can reduce total travel time by 3.4% and total disbenefit by 56.4%;If the road network traffic condition is State 2,compared with the bang-bang method,fuzzy neural network method can reduce total travel time by 4.6% and the total disbenefit by 30.9%.If the road network traffic condition is State 3,the system can only adopt the control manner through bang-bang method,compared with no diversion method,bang-bang method can reduce total travel time by 15.2%,but can increase total disbenefit by 13.9veh·h.