中文摘要：

基于贝叶斯理论,以马尔可夫链蒙特卡罗方法（Markov chain Monte Carlo Simulation,MCMC法）的自适应差分演化Metropolis算法为参数后验分布抽样计算方法,建立利用时变测试数据的参数随机反分析及模型预测方法。以香港东涌某天然坡地降雨入渗测试为算例,采用自适应差分演化Metropolis算法对时变降雨条件下非饱和土一维渗流模型参数进行随机反分析,研究参数后验分布的统计特性,并分别对校准期和验证期内模型预测孔压和实测值进行比较。研究结果表明,DREAM算法得到的各随机变量后验分布标准差较先验分布均显著减小;经过实测孔压数据的校准,模型计算精度很高,校准期内95%总置信区间的覆盖率达到0.964;验证期第2～4个阶段95%总置信区间的覆盖率分别为0.52、0.79和0.79,模型预测结果与实测值吻合程度较高。

英文摘要：

Based on the Bayesian theory, a probabilistic back analysis method using time-varying measurement data is established. The back calculated posterior distributions are determined using the Markov chain Monte Carlo method （MCMC） with the differential evolution adaptive Metropolis algorithm. In this paper, a case study of a well instrumented natural terrain is presented. The deterministic model for pore-water pressure evaluation is an analytical model. Field measurements of pore-water pressure are used to calibrate the unsaturated parameters of the deterministic model. Statistical properties of the posterior distributions are presented and discussed. It is found that the posterior standard deviations of the six parameters are all greatly reduced. The predicted and measured pore-water pressures during the calibration period and the validation period are compared. The coverage of the 95% total uncertainty bounds is estimated to be 0.964 for the calibration period, during which the field measured pore pressures are used to back analyze the input parameters. For periods 2 to 4 of the validation period, the coverage by the 95% total uncertainty bounds are 0.52, 0.79 and 0.79, respectively. These results indicate an overall good performance by the calibrated model.