改进BP神经网络在交通事故预测中的研究

发布时间：2018-01-09 05:23

本文关键词：改进BP神经网络在交通事故预测中的研究　出处：《华东师范大学学报(自然科学版)》2017年02期 　论文类型：期刊论文

【摘要】：交通事故严重程度受多种因素的影响,适合用人工神经网络来建模预测.因为标准BP(Back Propagation,BP)神经网络具有收敛较慢的缺陷,所以在自适应学习和附加动量因子改进BP神经网络~([1])基础上做了进一步的优化改进,使附加动量因子也具有自学习性.利用改进后的BP神经网络算法,选取英国利兹市的公开交通事故数据集,用影响交通事故严重程度的多种影响因素和事故严重程度构建并训练神经网络,并对最新数据进行预测.通过大量的实验对比收敛速度和预测结果,验证了改进后的算法具有更快的收敛速度和更高的预测准确率.
[Abstract]:The severity of traffic accidents is affected by many factors, so it is suitable to use artificial neural network to model and predict traffic accidents, because of the standard BP(Back Propagation. BP- neural network has the defect of slow convergence, so the BP neural network is improved by adaptive learning and additional momentum factor. [1) on the basis of further optimization and improvement, the additional momentum factor is also self-learning. Using the improved BP neural network algorithm, the open traffic accident data set in Leeds, England is selected. The neural network is constructed and trained by many factors which affect the severity of traffic accidents, and the latest data are forecasted. The convergence rate and prediction results are compared by a large number of experiments. The improved algorithm has faster convergence speed and higher prediction accuracy.
【作者单位】：华东师范大学计算中心;华东师范大学信息化办公室;
【分类号】：U491.31;TP183
【正文快照】： o 引言交通事故会引发不同程度的灾难性后果,对人的生命和财产造成严重影响.交通事故预测建模对预测交通事故严重程度具有重大预防意义,以便在事故可能发生前及时对驾驶员发出提醒,使其采取相应的应对措施,这对减少交通事故的发生有相当大的作用. 时间序列法、回归模型

【相似文献】