ABSTRACT

Appropriate pavement maintenance is of great importance due to the increasing deterioration of pavements and limited resources. Nowadays, highway agencies face large-scale networks. The management of large-scale networks is a challenge concerning the computational complexity, which typically increases exponentially with the dimension of the network. Water cycle and genetic algorithms are utilised in this paper to find the optimal maintenance schedule of large-scale pavement networks. A new practical constraint is introduced in the mathematical formulation that the total annual costs should not fluctuate more than a predefined limit over the planning horizon. Moreover, a novel index is developed to calculate the equity level in pavement maintenance scheduling, and the outcomes of the algorithms are compared based on this equation. A real road network with 103 pavement sections is the case study of this paper. The results show that ‘Equity index’ is reduced by 94% and 48% during the analysis period by WCA and GA, respectively. Drawing on WCA and GA optimal solutions, the average international roughness index of the network is decreased by 35% and 31% respectively in a 5-year horizon. Moreover, the variance of the maximum and minimum allocated budget in the analysis period is less than 15%.

摘要

由于路面日益恶化和资源有限，适当的路面维护非常重要。如今，高速公路机构面临着大规模的网络。大规模网络的管理是一项涉及计算复杂性的挑战，计算复杂性通常随着网络的维度呈指数增长。本文利用水循环和遗传算法来寻找大型路面网络的最佳维护计划。在数学公式中引入了一个新的实际约束，即年度总成本的波动不应超过计划范围内的预定义限制。此外，开发了一种新的指标来计算路面维护调度中的公平水平，并基于该方程比较算法的结果。本文的案例研究是一个具有 103 个路面部分的真实道路网络。结果表明，在 WCA 和 GA 的分析期间，“股票指数”分别下降了 94% 和 48%。借鉴 WCA 和 GA 最优解，该网络的国际平均粗糙度指数在 5 年内分别下降了 35% 和 31%。此外，分析期间最大和最小分配预算的方差小于15%。