【摘要】 通过对道路交通事故影响因素的分析,建立了关于道路交通事故影响因素的层次结构模型,并根据此模型建立基于RBF神经网络的道路交通事故计算机仿真预测方法。结合我国1978~2007年道路交通事故次数对RBF神经网络进行训练、检验和预测,同时与BP神经网络预测方法进行比较。结果表明RBF神经网络的平均误差和收敛次数分别为1.19%和701次,而BP神经网络则为9.8%和2401次,可见RBF神经网络具有更快的运算速度和更高的精度。更多还原

【Abstract】 Through analyzing the factors influencing the road accidents,an AHP model is established.This paper suggests a method of Radial Basis Function neural network model(RBF) for road accidents computer-simulated forecasting.The factors influencing the road accidents are analyzed and an AHP model is established.Based on this model the historical statistical data from 1978 to 2007 are used to train and check the RBF neural network,comparing with BP neural network model with the same set data.The result shows that average error and convergence rate of the RBF neural network are 1.19% and 701,compared with the results of BP neural network model which are 9.8% and 2401.So the RBF neural network model converges more quickly and gives a more accurate result for prediction.更多还原