Abstract:

In order to improve the traffic safety monitoring level for SCCR of a two-lane highway, a scale measurement and driving simulation experiments were used to quantitatively evaluate the driver's anxiety level on curves and describe the relationship between the driver's anxiety level and the driving safety characteristics. Through the statistics of anxious driving behaviors, spearman analysis was used to screen the key driving safety characteristics that affect the anxiety level. A multi-factor driver anxiety level prediction model was established by using a radial basis function neural network(RBFNN), which integrates some driving safety characteristic factors consisting of road alignment conditions, individual characteristics of drivers and driving behavior characteristics. The results show that the anxiety level is significantly negatively correlated with driving age and driver's age, and significantly negatively correlated with vehicle speeds, lateral offset, corner variation coefficient and curve radius (p<0.01). Through verification, the prediction model of driver's anxiety level based on radial basis function neural network has a high discrimination accuracy, up to 73.7%. The angle variation coefficient, age and driving age are the important factors affecting the driver's anxiety level, and their importance is 100%, 93.3% and 90.7%, respectively. The research results can provide theoretical support for monitoring driving anxiety levels and the maintenance of traffic safety in SCCR of two-lane highways.

Key words: sharp circular curve road(SCCR), drivers, anxiety level, prediction model, neural network