704 应用基础与工程科学学报 Vol.18 Safety Science Joumal,2003,13(9):74-77 [11]Liu X.Zhai Z.Location identification for indoor instantaneous point contaminant source by probability-based invere Computational Fluid Dynamics modeling [J].Indoor Air,2008,18(1):2-11 [12] 郭少冬，杨锐，翁文国.基于MCMC方法的城区有毒气体扩散源反演[J].清华大学学报（自然科学版），2009， 49(5):629634 Guo Shaodong,Yang Rui,Weng Wenguo.Source inversion of toxic gas dispersion in urban areas based on the MCMC method []]Journal of Tainghua University(Science and Technology),2009,49(5):629-634 Investigation of the Inversion Modeling for Indoor Contaminant Source Based on the Adjoint Equation and MCMC Method GUO Shaodong'2,YANG Rui',SU Guofeng',ZHANG Hui (1.Centre for Public Safety Research,Department of Engineering Physics,Tsinghua University,Beijing 100084,China; 2.Institute of Applied Physisc and Computational Mathematics,Beijing 100084,China) Abstract Source inversion and identification method of contaminant dispersion in buildings was studied.An adjoint transportation equation was introduced and solved numerically,and a likelihood function was computed using predicted concentration database.Markov Chain Monte Carlo(MCMC)sampling based on Bayesian inference was used to inverse the parameters, including the source location and intensity.The agreement of the predicted results with the actual source location and intensity indicated that the method could be used to estimate indoor source parameters effectively.The studies also showed that the proposed procedure was more efficient when using the adjoint transportation equation with MCMC methods.The effects of the error probability distribution and detection limits of the sensors on the sensitivity and reliability of the method were then discussed.Analyses show that flatter and broader probability distribution of the sensor error leads to inversion results with larger uncertainty,while the lower sensor sensitivity leads to insufficient information and ill-posed characteristics of the inversion results with several clusters of possible source locations. Keywords:indoor contaminant source inversion;Bayesian inference;adjoint equation; computational fluid dynamics 万方数据704 应用基础与工程科学学报 V01．18 Safety Science Journal，2003，13(9)：74j77 [11]Liu X，Zhai z．Location identification for indoor instantaneous point contaminant floulw*by probability．based inverse Computational Fluid Dyllafflic$modeling[J]．Indoor Air，2008，18(1)：2-1 1 [12]郭少冬，杨锐，翁文国．基于MCMC方法的城区有毒气体扩散源反演[J]．清华大学学报(自然科学版)，2009， 49(5)：629-634 Guo Shaodong，Yang Rui，Weng Wenguo．Source inversion of toxic gas dispersion in urban glma8 based on the MCMC method[J]．Journal ofTsinshua Unlvemity(Science and Technology)。2009，49(5)：629-634 Investigation of the Inversion Modeling for Indoor Contaminant Source Based on the Adjoint Equation and MCMC Method GUO Shaodon91’。，YANG Ruil，SU Guofen91，ZHANG Huil (1．Centre for Public Safety Research，Department of Ensineering Physics，Tsinghua University，Beijing 100084，China； 2．Institute of Apphed Physisc and Computational Mathematics，Beijing 100084，China) Abstract Source inversion and identification method of contaminant dispersion in buildings was studied．An adjoint transportation equation was introduced and solved numerically，and a likelihood function WflS computed using predicted concentration database．Markov Chin Monte Carlo(MCMC)sampling based on Bayesian inference Was used to inverse the parameters， including the source location and intensity．The agreement of the predicted results with the actual source location and intensity indicated that the method could be used to estimate indoor source parameters effectively．The studies also showed that the proposed procedure Was more efficient when using the adjoint transportation equation with MCMC methods．The effects of the error probability distribution and detection limits of the sensors on the sensitivity and reliability of the method were then discussed．Analyses show that flatter and broader probability distribution of the sensor en'or leads to inversion results with larger uncertainty，while the lower sensor sensitivity leads to insufficient information and ill—posed characteristics of the inversion results with several clusters of possible source locations． Keywords： indoor contaminant source inversion；Bayesian inference；adjoint equation； computational fluid dynamics 万方数据