摘要
火灾是地下空间出现频次最高的灾害,构建有效的地下空间火灾火源定位方法对于提升逃生疏散效率、降低火灾后果具有重要意义。依托重庆轨道交通10号线红土地站,基于FDS火灾模拟结果构建火灾数据库,采用BP神经网络分别建立判断着火层及确定火源平面坐标的神经网络,通过前者判断着火层数再利用对应层的坐标定位神经网络确定火源的平面坐标,从而实现具有复杂内部构造的深埋地铁车站火灾火源的三维空间定位。分析了火灾进程、传感器间距及输入数据类型和数量对定位精度的影响。研究表明:建立的火源定位方法具有较高的定位精度;存在一个最优的传感器间距能在保证定位精度的前提下,减少监控成本;基于温度驱动的定位模型优于基于CO浓度驱动的定位模型。研究成果可为未来深埋地铁车站火源定位方法的建立提供参考,并指导火灾时救援疏散策略的制定。
Fire accident is the most frequent underground accidents.Development of an effective fire source estimation method for underground fire accidents is of great significance for enhancing evacuation efficiency and mitigating fire accident consequences.Taking the Hongtudi Station of Chongqing Metro Line 10 as the background project,a fire accident database was established through FDS fire accident simulations.A floor location BP neural network and a coordinate determination BP neural network were established separately.The fire accident floor was firstly identified using the floor location BP neural network.Then the fire source’s planar coordinates were determined using the coordinate determination BP neural network of the corresponding floor.Thereby,the 3D estimation of fire source for deep-buried metro station with complex internal structures can be realized.The influences of fire process,sensor spacing and the number and type of sensors on the accuracy were investigated.It is shown that the established fire source estimation method has a high accuracy.There exists an optimal sensor spacing which could reduce the monitoring cost with a satisfactory accuracy.The temperature data-driven fire source estimation model is superior to the CO concentration data-driven model.The paper can provide reference for future development of fire source estimation for deep-buried metro station and guide on making the rescue and evacuation strategies when fire accidents occur.
作者
靳健
黄昕
许祺航
JIN Jian;HUANG Xin;XU Qihang(Department of Geotechnical Engineering,Tongji University,Shanghai 200092;Chengdu Environment Group,Chengdu 610041)
出处
《现代隧道技术》
CSCD
北大核心
2022年第S01期322-331,共10页
Modern Tunnelling Technology
基金
国家重点研发计划(2019YFC0605105)
关键词
深埋地铁车站
火灾
火源定位
BP神经网络
Deep-buried metro station
Fire accident
Fire source estimation
BP neural network
作者简介
靳健(1996-),男,硕士,助理工程师,主要从事地下工程安全控制方面的研究工作,E-mail:1932299@tongji.edu.cn;通讯作者:黄昕(1985-),男,博士,副教授,主要从事地下工程方面的研究工作,E-mail:xhuang@tongji.edu.cn.