混凝土碳化深度的贝叶斯自回归预测分析

总结和分析了影响混凝土碳化的主要因素及碳化深度计算模型,讨论了参数随机性及不确定性对碳化深度预测计算结果的影响.根据贝叶斯分析的基本原理,研究了混凝土碳化深度预测的贝叶斯自回归方法.该方法根据马尔可夫链(Markov Chain)的概率密度演化,利用吉布斯(Gibbs)抽样及蒙特卡洛(Monte Carlo)数值模拟,建立了混凝土碳化深度的随时贝叶斯自回归模型.该模型形式简单,收敛性好,且具有较高的预测精度.利用该方法和实测的碳化深度结果,建立自回归模型,可以对混凝土碳化深度进行更新预测.

Prediction and Analysis of Carbonation Depth of Concrete with Bayesian Auto-regressive Method

This paper summarizes and analyzes the effect of some main factors on carbonation depth of concrete and the model to calculate carbonation depth. The influence of stochastic and uncertainty of the parameters on the result of carbonation depth prediction is also discussed. According to the basic principle of the Bayesian analysis, the Bayesian auto-regressive method to predict the carbonation depth of concrete is studied. The method is based on the probability density evaluation of Markov Chain, using Gibbs sampling and Monte Carlo numerical simulation, and establishing the Bayesian auto-regressive model of carbonation depth at any time. This approach is simple, resulting in good convergence and higher precision of prediction. Concrete carbonation depth can be updated and predicted in auto-regressive model established by using this method and the measured carbonation depth.