文献类型：期刊文献

中文题名：大型活动拥挤踩踏事故BP神经网络安全评估方法应用分析

英文题名：Application of BP Neural Networks Assessment Method in the Stampede Accidents of Large-scale Activities

作者：王起全[1];郑乐[2]

机构：[1]中国劳动关系学院安全工程系;[2]中国精密机械进出口总公司

年份：2009

卷号：19

期号：4

起止页码：127

中文期刊名：中国安全科学学报

外文期刊名：China Safety Science Journal(CSSJ)

收录：CSTPCD、、CSCD2011_2012、BDHX2008、BDHX、CSCD

语种：中文

中文关键词：大型活动;拥挤踩踏;人工神经网络;评估;分析

外文关键词：large-scale activities ; stampede ; artificial neural networks ; assessment; analysis

中文摘要：根据大型活动拥挤踩踏事故是一个极其复杂的多因素、多变量、多层次的人-机-环境系统,利用数学建模的思路,引入人工神经网络评估方法;以沈阳世博会开幕式为例,通过前馈型BP网络训练和测试,将开幕式举办过程中的人员拥挤踩踏事故风险系统划分为人的因素、物的因素、环境因素、管理因素4个子系统和拥挤踩踏事故综合评估总系统,分别对其进行拟合分析,在各个系统的拟合残差满足条件的情况下,得出量化的评估预测值及BP神经网络的使用范围。研究实例表明,BP神经网络评估结果直观易懂,可操作性强,是值得推广运用的一种评估方法。

外文摘要：In view of the fact that the stampede accidents in large-scale activities is a multi-factor, multivariable and multi-layer man-machine-environment system, the artificial neural network method is introduced to assess the risk of the stampede accidents. With the opening ceremony of the Shenyang Expo as an example, through the training and testing of the feed-forward BP network, The risk system of the stampede accident is divided into a comprehensive assessment total system and four subsystems, which includes human factors subsystem, physical factors subsystem, environmental factors subsystem and management factors. Fitting analysis is respectively made on these systems, and then the quantitative assessment result and the application range of the BP network can be obtained when the fitting residuals meet the conditions. The application example shows the BP neural network assessment method is feasible and practical.