BP neural network prediction model of highway network land in urbanization process

To collocate the land for the highway network reasonably, a prediction model using the BP(back propagation) neural network has been created. Based on the analysis of the effect of highway network in the urbanization process, the importance of planning reasonable lands of the road network in the process of urbanization is clarified. In the modeling process using BP neural network, the urbanization rate and the proportion of urban land area are used as the neurons of the input layer, and the proportion of highway land area is used as the neuron of the output layer. The data of seven western countries are chosen as No.1 training sample according to the situation of urbanization level and highway development, and the historical data from Japan are chosen as No.2 training sample according to the situation of land resources. The historical data from the region can be used as No.3 training sample in the prediction model. The operation procedure is designed by MATLAB. Finally, the model is proved by using the data from Anhui province. The model considers the constraint of land resources to the land for highway network and overcome the deficiencies of the traditional linear model. The model presented in this paper can be used to predict the land for highway network under different levels of urbanization.