Based On PNN Study On Methods For Forecasting Trip Mode Split

The definition of transportation mode split is one of the most important part of transportation planning and policy establishment. If the splitting of transportation mode is reasonable, the limited city space can be used effectively and the congested city transport will become less crowed. But trip mode split forecasting models that are existing have different defects, so it is necessary to study on methods for forecasting urban residents trip mode split.Then the thesis expect the complex relationship between influent factors, and which can be carried out through artificial nerve network and Matlab toolbox to establish the transportation mode choice forecast model.First, the thesis introduces trip mode split forecasting models that are used usually and issues, and researches effectiveness of applying PNN on transportation split . and against the parameters in the model calibration problems, introducting method of model calibration. Secondly, the thesis analyses survey datas of Fangshan district, aggregating factors affecting. Finally, the thesis takes Fangshan district for example, calibration parameters will improve both before and after the model predicted results contrast, the result shows that improving the calibration parameters can optimiz model. And contrasting based on PNN transportation split model and aggregate method and disaggregate method, the result shows that based on PNN transportation split model is better. So can use trained PNN to split trip mode.