A novel technique for multi-objective sustainable decisions for pavement maintenance and rehabilitation

To maintain pavement in good condition while considering financial costs and sustainability, it is necessary to develop a comprehensive pavement management plan. Pavement Maintenance and Rehabilitation (M&R) consists of two essential components: firstly, predicting the pavement condition within...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Case Studies in Construction Materials Ročník 20; s. e03037
Hlavní autoři: Naseri, Hamed, Aliakbari, Amirreza, Javadian, Mahdie Asl, Aliakbari, Alireza, Waygood, E.O.D.
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier Ltd 01.07.2024
Elsevier
Témata:
Ensemble learning
Maintenance and rehabilitation
Multi-objective optimization
Pavement condition prediction
Sustainability
Pavement condition prediction
Maintenance and rehabilitation
Multi-objective optimization
Ensemble learning
Sustainability
ISSN:2214-5095, 2214-5095
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:To maintain pavement in good condition while considering financial costs and sustainability, it is necessary to develop a comprehensive pavement management plan. Pavement Maintenance and Rehabilitation (M&R) consists of two essential components: firstly, predicting the pavement condition within a specified timeframe, and secondly, employing an appropriate optimization algorithm. This study utilized three ensemble learning techniques including extreme gradient boosting, categorical boosting, and light gradient boosting machine to develop accurate predictions about the pavement condition. Subsequently, the most accurate prediction technique, which was extreme gradient boosting, was combined with non-dominated sorting genetic algorithm III which is a multi-objective metaheuristic optimization algorithm, resulting in a hybrid technique that offers highly accurate multi-objective maintenance and rehabilitation planning. Although previous studies neglected important criteria such as road closure in the optimization process, this study takes into account four objective functions including greenhouse gas emission, M&R cost, pavement condition, and road closure to be minimized over a 5-year program. This process generated 52 non-dominated optimal solutions known as the Pareto front. To compare and rank various optimal maintenance and rehabilitation plans, grey relational analysis was employed. The results suggested that there is a direct correlation between M&R costs and GHG emissions. Minimizing only pavement conditions in the planning can significantly increase GHG emissions, M&R costs, and road closure. Implementing preventive M&R actions can reduce M&R costs and overall road closure while light and medium rehabilitation actions are recommended to optimize the condition of pavements. •Extreme gradient boosting was the most accurate technique of IRI prediction.•A new hybrid prediction-optimization algorithm is introduced to optimize M&R plans.•Implementing preventive M&R actions can reduce GHG emissions.•Single-objective M&R planning can deteriorate other criteria, such as GHG emissions.
ISSN:2214-5095
2214-5095
DOI:10.1016/j.cscm.2024.e03037