基于RBF神经网络的交通生成预测模型

针对交通生成预测中传统集计预测模型缺少行为基础这一重要数据源，以及普通离散模型建立对数学推导依赖性高和预测中存在多个预测值，导致预测结果精度不理想的问题，建立了基于RBF神经网络的交通生成预测模型，该模型将以行为基础为前提的离散数据作为数据源，模型的建立不需要进行数学推导，利用输入和输出数据自动建立，再结合RBF神经网络收敛速度快和具有唯一最佳逼近点的特点，对交通生成进行预测。通过实例进行仿真和分析，结果表明，该模型最终获得的交通生成预测结果与实际值误差在允许范围内，RBF神经网络交通生成预测效果好，能在实际中应用。

Forecasting Model of Trip Generation Based on RBF Neural Network

The traditional disaggregate models lack an important data source of behavior- based foundations, as well as the discrete model dependents highly on mathematical derivation and there is a plurality of predicted value in the prediction. These factors would result in inaccurate prediction. So a trip generation forecast model was established based on RBF neural network. The model was based on the premise of behavior discrete data as a data source. The model does not require the establishment of mathematical deduction and the use of input and output data is automatically created. The characteristics of high convergence speed and unique best approximation point of RBF neural were applied to predict traffic generation. Simulation and analysis re- suits show that the model finally obtained results and the actual traffic generation forecast error is within the allowable range. RBF neural network traffic generation forecast is practical and effective.