利用拉丁超立方与相关性采样的通风系统仿真不确定研究  

Uncertainty analysis of ventilation system simulation using Latin hypercube and correlation sampling methods

作　　者：曹鹏[1,2] 刘剑[1,2] CAO Peng;LIU Jian(College of Safety Science&Engineering,Liaoning Technical University,Huludao 125105,Liaoning,China;Key Laboratory of Mine Thermo-Motive Disaster&Prevention,Ministry of Education,Huludao 125105,Liaoning,China)

机构地区：[1]辽宁工程技术大学安全科学与工程学院,辽宁葫芦岛125105 [2]矿山热动力灾害与防治教育部重点实验室,辽宁葫芦岛125105

出　　处：《安全与环境学报》2025年第3期1016-1025,共10页Journal of Safety and Environment

基　　金：国家自然科学基金项目(51574142,51774169,52404212);辽宁省教育厅科研基金项目(LJ212410147072)。

摘　　要：为降低数字通风系统仿真不确定性给智能通风决策带来的安全隐患,基于巷道风阻系数获取过程中存在的不确定性,同时考虑相同支护巷道间的相关性,提出了一种结合全局模拟的拉丁超立方采样(Latin Hypercube Sampling,LHS)技术和关注局部相关性的多维变量精确采样技术的数字通风系统不确定性分析方法。LHS随机模拟生成多维独立标准正态分布初始变量;利用乔莱斯基分解法分解风阻系数间协方差矩阵求得相关性变换矩阵;线性组合初始变量与变换矩阵生成具有相关性的风阻系数随机变量;随机变量作为数字通风系统的输入参数;解算风量和风压,统计仿真结果。结合实际生产矿井通风系统进行分析,结果表明:风阻系数的不确定性对数字通风系统仿真结果有显著影响;当不考虑相关性且风阻系数的变异系数为15%时,仿真风量和风压波动范围分别为0.62~73.88 m^(3)/s和0.06~522.89 Pa,平均波动率分别为83.04%和243.85%;考虑相关性时,风量和风压波动率分别下降了75.58%和84.97%,个别分支风量的波动率减少了95.90%;进风区分支和角联分支仿真风量、回风分支和角联分支仿真风压波动性最大;角联分支风量波动率最大为1115.09%,回风分支风压波动值最大为539.75 Pa。To mitigate the safety risks associated with simulation uncertainty in digital ventilation systems for intelligent decision-making,this paper proposes an uncertainty analysis method that integrates Latin Hypercube Sampling(LHS)with multidimensional correlated variable sampling techniques.The focus is on addressing uncertainties in the determination of roadway airflow resistance coefficients.First,the uncertainty distribution of the roadway resistance coefficients is quantified using a normal distribution.Correlation coefficients are established among the resistance coefficients of roadways that share the same support,and a covariance matrix is computed.Next,LHS is employed to randomly generate initial variables following a multidimensional independent standard normal distribution.Cholesky decomposition is then applied to derive the correlation transformation matrix from the covariance matrix of the airflow resistance coefficients.Subsequently,the correlated random variables for the resistance coefficients are obtained through a linear combination of the initial variables and the transformation matrix.Finally,these random variables are used as input parameters for the digital ventilation system to simulate airflow and wind pressure calculations.The fluctuations in the simulation results are subjected to statistical analysis.The proposed method has been validated using the ventilation system of an actual production mine.It is found that uncertainty in the roadway resistance coefficients significantly impacts the accuracy of the simulation outcomes.When correlation is not accounted for,and the coefficient of variation of the airflow resistance is 15%,the simulated airflow and wind pressure fluctuate in 0.62-73.88 m^(3)/s and 0.06-522.89 Pa,respectively.The average fluctuation rates are 83.04%for airflow and 243.85%for wind pressure.The maximum fluctuation rate of the diagonal branch airflow reaches 1115.09%,while the maximum fluctuation in wind pressure for the return airway branch is 539.75 Pa.When correlation is c

关 键 词：安全工程 矿井通风系统 不确定性分析 拉丁超立方采样 参数相关性 

分 类 号：X936[环境科学与工程—安全科学]