Infrastructure Management with Stochastic Dynamic Programming

本研究提出了一种利用随机动态规划技术确定基础设施维护/更换策略的方法。基础设施在较长的使用寿命期间可能出现退化，其退化过程本身是随机的。然后导出策略，使退化过程本身是随机的。然后，将退化建模为随机过程，推导出使基础设施预期经济价值最大化的策略。本研究首先展示了一个总体框架，即退化用随机微分方程建模，目标函数定义为项目生命周期内净效益流的折现总和。考察了该框架的数学特征，并讨论了其经济解释。在此基础上，提出了公路桥疲劳退化的具体模型。退化用马尔可夫链建模。将该模型应用于实际案例，给出了准确的基础设施管理策略。

This study proposes a methodology for determining the maintenance/replacement strategy of infrastructure by using techniques of stochastic dynamic programming. Infrastructures may be degraded during a long service life and the process of degradation itself is stochastic. Then strategy should be derived so as to maximize the process of degradation itself is stochastic. Then strategy should be derived so as to maximize the expected economic value of infrastructure with modeling of degradation as a stochastic process.This study first shows a general framework that the degradation is modeled by stochastic differential equation and the objective function is defined as the discounted sum of net benefit flow over the project life. The mathematical features in the framework are examined and economic interpretations are discussed.Then the study presents the particular model for a road bridge which may be degraded by fatigue. The degradation is modeled as Markov Chain. The model is applied to the real case and to show the exact strategy of infrastructure management.