A hybrid genetic algorithm to maintain road networks using reliability theory

• Published in: Structure and infrastructure engineering Vol. 19; no. 6; pp. 810 - 823
• Main Authors: Altarabsheh, Ahmad, Altarabsheh, Ibrahim, Ventresca, Mario
• Format: Journal Article
• Language: English
• Published: Taylor & Francis 03.06.2023
• Subjects: dynamic programming; genetic algorithms; knapsack algorithm; linear multi-state consecutively connected system; reliability theory; Roads maintenance
• ISSN: 1573-2479; 1744-8980
• DOI: 10.1080/15732479.2021.1981400

Maintaining road pavement is a challenging task, due to the vast size of the roads network and the limited maintenance budget available to road agencies. Therefore, most agencies aim at selecting the most deteriorated road segments for maintenance, within their available budget. However, the majority of the current developed asset management tools aim at improving the average performance of the roads network, regardless of the condition of the maintained segments. To overcome this limitation, this study first proposes a novel measure for roads network performance using reliability theory, then it proposes a hybrid optimization approach that is integrated with the proposed performance measure to enable road agencies to focus on the most deteriorated segments in the network while maximizing the network performance. The hybrid optimization approach combines dynamic programming with a traditional genetic algorithm. The proposed algorithm was applied to two major road systems in Jordan and validated by comparing its performance with that of a traditional genetic algorithm. The results show that the proposed algorithm outperforms the traditional genetic algorithm, because of its ability to select the most deteriorated segments in the network, while achieving better network condition at lower cost.