A Study on Vehicle Trip Distribution Forecasting Based on BP Neural Network

摘要

Trip distribution is an important step of the four-phase method inrntransportation demand forecasting. It is affected by many factors. First, factorsrninfluencing vehicle trip distribution are analyzed and the results show that locationrnpotential, traffic impedance, trip production, and trip attraction play an importantrnrole. Second, an aggregation scale factor of the traffic zone is evaluated by fuzzyrnalgorithm and the zone accessibility is calculated, and then, the location potentialrnthat is determined by the aggregation scale factor and zone accessibility is obtained.rnThird, traffic impedance based on travel time is determined based on analyzing therninfluences of traffic flow discontinuation, bicycles, pedestrians, and width of lanes.rnFinally, the trip distribution forecasting model based on the BP neural networkrnwhich takes location potential, traffic impedance, trip production, and trip attractionrnas input parameters and the result of trip distribution as output parameters isrnestablished. The test shows that the forecasting result fits well with the survey data.rnThus, the BP neural network model can be used for trip distribution forecasting withrnhigh prediction accuracy.