Study On Security Of Freewayway Entrance And Exit Ramp Based On Traffic Conflict Technique And Neural Network Model

Freeway entrance ramp is an important part of the highway and is accident-prone areas.But because of its own particularities and influences of many factors,the ramp traffic accidents not only can result in casualties,but also seriously affect the speed Highway traffic flow.In this paper,starting from the system,the use of traffic conflict techniques and neural networks,considering driving the vehicle,the driver and the highway alignment and so on,researches on freeway entrance ramp safety.Firstly,this paper analyzes its scope,by research on development oftraffic conflict technique safety analysis method based on traffic conflict technology.The entrance ramp traffic flow characteristics were analyzed and the entrance ramp traffic traveling are divided into sub-confluent,free travel,and with the car travel,and based on traffic conflict technique,it analyzes the entrance ramp traffic operation.Through analysis showed: increase of the vehicle speed in the free state is relatively slow,while speed decreases slowly,so to achieve the purpose of protection of the safe operation of the vehicle;when the target vehicle in a highway entrance ramp travels with the car,the speed of which meet certain requirements,operational state of the target vehicle is a safe state.Secondly,it analyses the basic situation of the entrance rampand researches the entrance ramp safety,detailedly analyses the relationship between geometric road alignment and traffic conflicts,the relationship between traffic flow and traffic factors and speed indicators of conflict,the relationship between driving behavior and traffic conflicts and the relationship between factors and ramp traffic safety,which provides a theoretical basis for the establishment of traffic safety analytical prediction model.Finally,this paper studies artificial neuron,artificial neural network applications field in freeway entrance ramp and use the BP neural network model of freeway entrance ramp to predict traffic conflicts.The minimum relative error in the entrance and the traffic conflict predictive value was 0.167%.That the maximum relative error is-14.8%,can be seen traffic conflict with a certain chance and randomness.The mean square error is5.375% percentage,then the BP neural network model may reflect variation freeway ramp merging area as a whole.The minimum relative error in the exit and the traffic conflict predictive value is 0.00%,That.the maximum relative error is-19.83%can be seen traffic conflict has a certain chance and randomness,the mean square error is 8.181% percentage,then the BP neural network model as a whole may reflect changes in the law on motorway exit triage area.