基于神经网络的公共建筑应急疏散风险评估方法  

Risk assessment method of emergency evacuation in public buildings based on neural network

作　　者：李嘉锋 胡玉玲 李佳旭[1,2] LI Jia-feng;HU Yu-ling;LI Jia-xu(School of Electrical and Information Engineering,Bei-jing University of Civil Engineering and Architecture,Beijing 100044,China;Beijing Key Laboratory of Intelligent Pro-cessing for Building Big Data,Beijing University of Civil En-eineering and Architecture,Beijing 100044,China)

机构地区：[1]北京建筑大学电气与信息工程学院,北京100044 [2]北京建筑大学建筑大数据智能处理方法研究北京市重点实验室,北京100044

出　　处：《消防科学与技术》2022年第4期491-495,共5页Fire Science and Technology

基　　金：北京建筑大学基本科研业务基金项目(X20109);国家重点研发项目(2018YFC0807806)。

摘　　要：公共建筑空间大、人员密集、水平疏散距离长,在应急情景下的疏散本身存在一定的风险,提出了一种基于深度神经网络(DNN)的应急疏散风险评估方法。给出了DNN预测模型的建立方法,并以某高校体育馆为案例,说明了模型数据获取、模型训练,及模型测试的整个评估过程。结果表明,相较于传统评估方法,该深度学习方法克服了主观性强、对以人为核心的复杂疏散系统风险评估困难等缺点,可以实现对公共建筑应急疏散快速有效的评估。Public buildings have large spaces, densely populated people, and long horizontal evacuation distances. There are certain risks in the evacuation in emergency situations. This paper proposes an emergency evacuation risk assessment method based on deep neural network(DNN). The establishment method of DNN prediction model is given, and a university gymnasium is used as a case to illustrate the whole evaluation process of model data acquisition, model training, and model testing. The results show that compared with traditional evaluation methods,this deep learning method overcomes the shortcomings of subjectivity and difficulty in risk assessment of complex evacuation systems centered on people, and can realize rapid and effective evaluation of emergency evacuation in public buildings.

关 键 词：应急疏散 深度学习 风险评估 DNN预测模型 AnyLogic平台 

分 类 号：X932[环境科学与工程—安全科学] TP391.9[自动化与计算机技术—计算机应用技术]