|
下载链接:https://www.sciencedirect.com/science/article/pii/S2950576324000023
01研究亮点
1. 基于CFD模型搭建了地铁隧道火灾仿真数据库
2. 提出一种基于改进TCNN的火灾温度场预测代理模型
3. 该模型可以高效准确预测火灾时空温度场
02 摘要
地铁隧道火灾产生的烟雾温度和有毒气体对被困人员构成威胁,阻碍救援工作。为了快速获取烟气扩散情况来了解地铁隧道火灾的未来发展态势,进而指导应急救援,本研究提出了一种基于转置卷积神经网络(TCNN)的深度学习代理模型来预测时空温度场。
采用计算流体动力学(CFD)方法,建立了涵盖了不同火灾位置、热释放率(HRRs)和通风速度的地铁隧道火灾数值模拟数据库来收集火灾数据。集成自适应上投影机制、通道注意力机制、轴向注意力机制以及深浅层特征融合策略来改进TCNN,提升模型对复杂火灾场景的识别与预测能力。
研究结果表明,该模型在输入火灾位置、HRR和通风速度后,能在0.034 s内以高达94%的准确率重现火灾烟气的时空温度场。所提出的深度学习代理模型为地铁隧道火灾的应急控制和救援工作提供一种高效精准的技术手段。
03 期刊简介
Journal of Dynamic Disasters致力于刊发动力灾害相关文章。期刊特邀请十多位院士担任期刊编委。
期刊主页:
https://www.keaipublishing.com/en/journals/journal-of-dynamic-disasters/
投稿链接:
https://www.editorialmanager.com/dds
The Journal of Dynamic Disasters is an international journal dedicated to publishing authoritative articles on dynamic disasters. Topics of interest include earthquake, wind, wave, explosion, shock, vehicle and environmental vibrations. The journal features original research and case studies focusing on the dynamical analysis, disaster mechanisms, disaster prevention, disaster monitoring, disaster assessment and post-disaster restoration for engineering structures (such as civil engineering structures, mechanical engineering structures, aerospace structures, marine structures). The journal welcomes interdisciplinary studies, covering topics such as sensing, signal processing, intelligent management and control of dynamic disasters.
Dynamical analysis of structures
Disaster mechanism of structures
Analysis on generation and propagation of vibration waves
Disaster prevention of structures
Health monitoring of structures
Disaster assessment of structures
Post-disaster restoration of structures
Sensing and signal processing of structures or dynamic disasters
Post-disaster restoration of structures
Intelligent management and control of dynamic disasters
期刊为优秀稿件特别评选优秀论文,欢迎广大师生、学者同仁积极投稿!
推荐阅读
Archiver|手机版|科学网 ( 京ICP备07017567号-12 )
GMT+8, 2024-10-5 23:49
Powered by ScienceNet.cn
Copyright © 2007- 中国科学报社