中文摘要：

交通事故的发生因受随机因素的影响而呈现出不确定性和非线性的特点。在分析交通事故与人口、车辆、道路、经济发展等因素关系的基础上,综合考虑影响交通事故的多种因素,建立了BP神经网络。进而,选取总人口、机动车驾驶员人数、公路密度、民用车辆、人均GDP作为交通事故预测模型的输入向量,以交通事故的四项指标作为输出向量,利用LM算法或GALM算法优化的BP神经网络模型对交通事故进行预测。实验表明,GALM算法优化的BP神经网络模型与BP神经网络或LM算法优化的BP神经网络相比,具有较高的精度和较快的收敛速度,能更好地适用于交通事故预测。

英文摘要：

The phenomenon of traffic accidents appears its uncertainty and nonlinearity due to the fact that it is affected by various random factors. Starting from the analysis of the relation between traffic accidents and factors, which includes population, vehicle, road, economic development and so on, the forecasting model based on BP neural network is proposed by combining the factors affecting traffic. Then the model is improved by LM algorithm and GALM algorithm. A traffic accident prediction model uses the amount of population, the number of drivers, road network density, civilian vehicles and per capita GDP as the input vectors. Meanwhile,its output vectors are the four figures of traffic accidents. The results show that the BP neural network improved by GALM algorithm can be better-suited for the forecasting of road traffic accident with higher precision and faster convergence rate than those algorithm we mentioned before.