Research on trip generation forecasting based on BP neural network

Forecasting trip generation and attraction is the first component of the four-stage method in transportation planning, which determines the urban layout and construction of traffic facilities. To improve the accuracy of trip generation forecasting, K-means cluster analysis was used to divide traffic zones into several groups according to the population and employment. Principal component analysis was conducted to calculate the loading rate to principal components, providing the basis for choosing the influence factor. Finally, BP(Back Propagation) neural networks were set up to forecast trip generation; the input included land-use and population of each traffic zone; and the output was the trip generation. The methods were testified with the traffic survey data from city of Dalian, Liaoning province. Moreover, the results were compared with those obtained from multiple regression model. It is indicated that the BP neural network based on data pre-process produces better results in trip generation forecasting.