摘要
为实现大型复杂结构建筑内火情信息的快速准确感知,构建分布式火源参数反演模型,利用局部探测信息和模拟信息估计火源位置及强度等关键参数。将建筑整体划分为若干区域,各区域利用火灾模型基于本区和邻区的局部结构生成模拟信息,并在此基础上引入局部探测信息,采用贝叶斯推断估计局部范围内火源参数的概率分布。之后,通过各区域的通信协作完成火源参数在全局范围内的反演。利用某多层建筑对模型进行初步验证。结果表明,分布式火源参数反演模型能够在火灾发生50 s后计算得到与设计火灾场景相同的火源位置和强度类型。
To rapidly and accurately sense the fire in buildings with large scale and complexity,a decentralized fire inversion model was built,which can estimate fire location and intensity based on local sensor measurements and local fire model simulations. The whole building was divided into several zones. Each zone can generate simulations by fire model with local structure. Meanwhile,local sensor measurements were fed into the inversion model and Bayesian inference was utilized to estimate distribution of probabilities of fire parameters in the local region. Then through the cooperation of all building zones,the global estimation can be carried out. The inversion model was tested considering a certain multi-floor building. Results show that the decentralized fire inversion model can compute the same fire location and intensity type as design fire scenario after fire has propagated 50 s.
出处
《中国安全科学学报》
CAS
CSCD
北大核心
2014年第5期51-55,共5页
China Safety Science Journal
基金
国家自然科学基金资助(91224008
91024032
70833003)
国家科技部"973"重大基础研究发展计划(2012CB719705)
