摘要
人群拥挤踩踏突发性的特点决定了现场的事故救援措施效果较差,事前预防是唯一有效的策略。对商业区人群流量进行预测,对于合理控制商业网点人口,预防人群类事故的发生具有重要的意义。本文介绍了基于BP神经网络的人群流量预测方法,利用Matlab建立了相关模型,并结合实际数据对模型进行了调整,分析了隐含层神经元个数、不同输入-输出结构、不同传递函数等因素对网络性能的影响。研究表明利用神经网络的非线性映射能力对人群流量进行预测时可行的。
The crowd crushing and trampling accidents happen all at once and the consequence is serious.The effect of rescue measures is very poor,and prevention is the only effective strategies.Forecasting crowd flow in downtown is very impotrant for reasonable control the number of commercial outlets and prevent the accidents.A method based on BP neural network for crowd flow prediction was introduced in the paper.The model was established in Matlab.Combining with actual data,the model was then adjusted.The influence of number of nodes in hidden layer,different input/output structure,different transfer function to the performance of neural network was analyzed.It was concluded that using nonlinear mapping of neural network to forecast crowd flow is feasible.Results showed good agreement with the actual value.So it has a certain practical value in crowd flow prediction.
出处
《中国安全生产科学技术》
CAS
北大核心
2010年第2期61-65,共5页
Journal of Safety Science and Technology
基金
北京市科技计划"西长安街重点区域国庆综合保障科技示范研究"项目(编号:D09060603890903)资助
