Signalized intersection traffic safety is one of the relatively weak link in road traffic safety. With the rapid growth of the number of drivers in our country, the difference between the drivers is increasing, and the traffic safety at the signalized intersection is facing new challenges. This challenge is mainly due to the choice of the driver’s entrance roadway and the dilemma at the intersection The diversity of behavior and the expansion of the traffic safety assessment of the contents of the signal. Based on this, this paper chooses the driver’s vehicle trajectory data in the real road and dynamic traffic environment, and carries on the quantitative analysis to the driver of the different driving tendency at the intersection entrance road entrance behavior and the dilemma selection behavior, The conflict risk prediction model and the collision risk prediction model of the dilemma collision, reveal the influence of the different driving characteristics of the driving direction of the driver at different intersections on the traffic safety of the signal intersection. The research of this paper is of great significance to understand the relationship between driver’s driving tendency and traffic safety at signalized intersection, and improve the traffic safety management level of signalized intersection.
Firstly, the driving tendency of the driver is summarized, and the questionnaire, static test and dynamic test are designed to collect various characteristic indexes of driving tendency. Among them, the questionnaire mainly obtains three factors that affect driver’s driving skills, safety awareness and driving tendency. The static test collects the driver’s response time and velocity estimation ability. The dynamic test gathers the experimenter’s lane Frequency, drive frequency and braking frequency. On the basis of this, the driving tendency of the experimenter is divided into four types: radical, robust and conservative, and the driving behavior characteristics are analyzed by using K-means clustering analysis.
Secondly, the characteristics of the traditional vehicle trajectory acquisition method (high point recording method, experimental vehicle method and simulated driving method) are analyzed. Aiming at the driver ’s lane changing behavior, the traditional method of vehicle trajectory acquisition is improved. A high - point recording method and experimental vehicle method are combined to study the trajectory acquisition method. In view of the driver’ s dilemma, the recording method to obtain the dilemma of the vehicle trajectory data. Using the Tracker software to extract the trajectory data, and use the camera imaging theory to convert the data, the conversion of the real trajectory data after the smooth processing, and then calculate the different driving tendency of the driver’s lane behavior characteristics and dilemma selection Behavior characteristic parameter.
Thirdly, the meaning of driving behavior is sorted and interpreted from both macro and micro levels, and the driving characteristics of driving behavior are determined from the micro level. For the track behavior, mainly from the beginning of the implementation of the track, the duration of the road, the road speed in the process of changing the road speed and the process of vehicle spacing and other characteristics of the characterization; for the dilemma selection behavior, mainly from the green light began to flash Vehicle to the parking line distance, speed, acceleration, deceleration, head time and other characteristics of the characterization. On this basis, the driving behavior of the radical, stable and conservative drivers and the driving characteristics of the selection behavior in the dilemma are described and analyzed respectively.
Then, the risk prediction model of the road behavior conflict is constructed, and the collision risk is measured by the collision time of the probability of the collision risk and the collision avoidance deceleration to measure the severity of the conflict risk. Four risk levels are determined using this index: zero risk level, Low risk levels, intermediate risk levels and high risk levels. The model results show that the vertical velocity, the distance between the vehicle distance and the driving distance of the target lane have a significant effect on the conflict risk caused by the vehicle lane change process. Compared with the zero risk level, the probability of the risk level and the high risk level is different when the longitudinal velocity increases and the vehicle distance decreases from the target lane and the driving tendency is increased, the conflict caused by the vehicle lane is at a low risk level Increased degree. The model can also be used to predict the probability that the track behavior is at different risk levels under certain conditions. It can provide the calculation method and theoretical basis for the traffic simulation model for the road simulation model.
Finally, based on the analysis results of the driver’s dilemma selection behavior, the model of the driver’s binary selection (continued driving or parking) including the driving tendency factors in the phase transition period is established to determine the difference between the two driving conditions, The lower boundary. On the basis of this, the risk prediction model of rear collision and the risk model of lateral collision risk are established. Monte Carlo simulation is used to show that the risk probability of lateral collision in the dilemma is higher than that of the rear collision risk. The vehicle is at the same distance to the parking line, the risk of collision will be different due to the driver’s driving tendency: the impact of the risk of collision from large to small driving in order: aggressive, conservative, steady type.