机场建设总进度计划延期分布及贝叶斯估计

Schedule Default Distributions and Bayesian Estimation in Airport Construction Master Schedule Management

摘要:

基于机场建设工程实践中的工作分解结构及跟踪数据，根据工程不同子系统延期工作发生频数分布，从总体角度、系统角度刻画总进度计划延期风险；针对机场工程总进度计划进度违约数据数量少、获取难度大的特点，采用贝叶斯方法估计分布参数，使用马尔科夫蒙特卡洛（Markov chain Monte Carlo，MCMC）模拟实现贝叶斯方法的计算。研究展示了从项目整体和宏观角度刻画机场建设工程不同子系统延期风险的途径；不同分布模型的MCMC参数估计结果显示出稳健性优势。同时，贝叶斯方法的应用能够集成定性数据和实践经验，并允许该模型的持续更新和优化。

Abstract:

Based on the WBS(work breakdown structure) in practice, a probability distribution of delayed activity frequency is proposed to depict schedule default risks in different subsystems of airport construction project from an overall and systematic perspective. Considering the lack of schedule default data, Bayesian methods are employed to estimate the distribution parameters. In particular, MCMC (Markov chain Monte Carlo) simulations are applied as computational scheme to obtain Bayesian estimates. The research process shows the way to model schedule default risks for different subsystems of airport construction project. The application of the MCMC method to estimate the parameters shows advantages in robustness according to different choices of distributions. The use of Bayesian methods makes it possible to integrate qualitative information, and constantly update the model during the construction in the future.