Driver pre-accident behavior pattern recognition based on dynamic radial basis function neural network

In this article, driver´s pre-accident behavior mode is systematically studied by means of combining the use of regional mapping error function with conditions of the resource distribution network and utilizing dynamic radial primary function neural network and its training method for pattern recognition. As is proved in this study, this method can not only improve the velocity of network training, reduce network structure, but also improve the properties of network generalization and the precision rate of pattern recognition. Simulated result preferably coincides with the measured result, which improves the adoptive method and the established model in this study to be right.