Abstract:To resolve recognition problem of road surface conditions and satisfy measure accuracy is one key of freeway meteorology monitoring study. Three kinds of,and color models extracted constitute feature vectors. A BP neural network algorithm improved by momentum and self-adaptive learning rate is presented. On the basis of these, a model of the nonlinear system is set up and applied in road surface conditions identification. Experimental results show that the model can identify road surface meteorology more accurately and effectively than that of the traditional BP neural network.

Study on Road Surface Condition Recognition based on BP Network Improved