摘要
对利用Bayesian模型分析热传导反问题中的导热系数预测问题的方法进行了研究。导热系数反问题的解是其后验概率密度的数学期望,用MarkovchainMonteCarlo算法计算后验状态空间以得到未知导热系数的统计估计。方法中取导热系数的先验分布满足正态分布,似然函数中的温度数据满足稳态零均值白噪声,先验分布与似然函数相乘得到后验概率密度函数。采用Metropolis-Hasting算法进行数据采样构造Markovchain,并截取收敛后的样本进行分析。
Bayesian model was used to analyze the estimation of the thermal conductivity in the field of inverse heat conduction problems. The inverse solution of the thermal conductivity was obtained by computing the expatiation of the posterior probability density functions（PPDF）. The posterior state space was exploited using Markov chain Monte Carlo （MCMC） algorithms in order to estimate the statistics of the thermal conductivity. The prior distribution of the thermal conductivity was assumed to satisfy a normal distribution, and the PPDF was obtained by multiplying the likelihood and prior distribution. Metropolis-Hasting algorithm was used to generate samples by a Markov chain mechanism.
出处
《系统仿真学报》
EI
CAS
CSCD
北大核心
2006年第6期1462-1465,共4页
Journal of System Simulation
