Abstract
Statistical data of P.R. China indicate that traffic accidents on curved roads take a large proportion of the total road accidents. To improve safety and driving comfort on curved roads, a mathematic speed model is developed to determine an optimised safety speed on curved roads. The model takes quantitative static road environment parameters (such as curvature, road gradient, superelevation, road surface, road width, road attributes), and dynamic environmental conditions (such as luminosity, vision and weather), and driver characteristics into account. In order to adapt to the individual driving behaviour, especially the speed selection at the curvature by taking into account the individual characteristics, the developed model is further modified by combining a BP (back-propagation) neural network model. The modified safety speed model has been verified with the tests based on a series of driver behaviour experimental data.
Keywords: speed model, curved road, BP (back-propagation) neural network, driver behaviour, road geometry, environmental condition