ABSTRACT

This study approaches a multi-stage stochastic mixed-integer programming model for the high-level complexity of large-scale pavement maintenance scheduling problems. The substance of some parameters in the mentioned problems is uncertain. Ignoring the uncertainty of these parameters in the pavement maintenance scheduling problems may lead to suboptimal solutions and unstable pavement conditions. In this study, annual budget and pavement deterioration rate are considered uncertain parameters. On the other hand, pavement agencies generally face large-scale pavement networks. The complexity of the proposed stochastic model increases exponentially with the number of network sections and scenarios. The problem is solved using the Progressive Hedging Algorithm (PHA), which is suitable for large-scale stochastic programming problems, by achieving an effective decomposition over scenarios. A modified adaptive strategy for choosing the penalty parameter value is applied that aims to improve the solution process. A pavement network including 251 sections is considered the case study for this investigation, and the current study seeks optimal maintenance scheduling over a finite analysis period. The performance of the stochastic model is compared with that of the deterministic model. The results indicate that the introduced approach is competent to address uncertainty in maintenance and rehabilitation problems.

摘要

本研究采用多阶段随机混合整数规划模型来解决大规模路面维护调度问题的高度复杂性。上述问题中某些参数的实质是不确定的。在路面维护调度问题中忽略这些参数的不确定性可能导致次优解和不稳定的路面条件。在这项研究中，年度预算和路面恶化率被认为是不确定的参数。另一方面，路面机构通常面临大规模的路面网络。所提出的随机模型的复杂性随着网络部分和场景的数量呈指数增长。使用适用于大规模随机规划问题的渐进式对冲算法 (PHA) 解决了该问题，通过实现对场景的有效分解。应用改进的自适应策略来选择惩罚参数值，旨在改进求解过程。包括 251 个路段的路面网络被认为是本次调查的案例研究，目前的研究旨在在有限的分析期内寻求最佳的维护计划。将随机模型的性能与确定性模型的性能进行比较。结果表明，引入的方法能够解决维护和修复问题中的不确定性。目前的研究是在有限的分析期内寻求最佳的维护计划。将随机模型的性能与确定性模型的性能进行比较。结果表明，引入的方法能够解决维护和修复问题中的不确定性。目前的研究是在有限的分析期内寻求最佳的维护计划。将随机模型的性能与确定性模型的性能进行比较。结果表明，引入的方法能够解决维护和修复问题中的不确定性。