福建省道路交通事故预测模型及其应用

Predicting model of highway transportation accidents in Fujian and its application

以2005-2006年福建省公路运输量(旅客周转量、货物周转量)与道路交通事故4项指数(事故次数、死亡人数、受伤人数和经济损失)之间的关系为基础,应用BP神经网络技术,建立了3输入、单输出、16个中间单元层的道路交通事故预测模型,分别对福建省道路交通事故的4项指数进行预测.结果表明:2007年1-3月份交通事故4项指数预测值的最大相对误差绝对值分别为2.9513%、3.5714%、3.5469%和2.9485%.因此,该3层神经网络预测模型具有使用简便、方法实用、预测精度高的优点.

Based on the relationship between the highway kilometers （passenger-kilometers and goods-kilometers） and the four terms of highway transportation accidents （quantities of traffic accidents, traffic deaths, traffic wounds, economic losses） in Fujian, the predicting models of highway transportation accidents, which included three inputs, single output and sixteen middle floors, were established by ANN technology, and the quantities of traffic accident, the number of traffic death, the number of traffic wound, traffic losses in Fujian were predicted in the paper. The results showed that from January to March in 2007, the maximum absolute values of relative error of the estimated value of the four traffic accident indexes were 2.9513% , 3.5714% , 3.5469% and 2.9485% , respectively. It was also proved that the models were feasible and easy to use, and had high accuracy.