Comparison of the Metaheuristic Algorithms Used in Road Maintenance Decision Making

When it comes to road network maintenance and rehabilitation (M&R) work, a lack of funds is the main challenge faced by decision makers. At present, how to develop a scientific and reasonable M&R program to maximize the effects of road network maintenance with limited maintenance funds has been the focus of research in the field of road maintenance. In this regard, this study establishes a hierarchical maintenance decision-making (DM) model based on bi-level optimization to enhance the pavement performance of the road network as the maintenance objective. It divides the large-scale road network into sub-networks according to the road network characteristics and maintenance needs to realize the scientific allocation of maintenance resources and accurate M&R of the road network. To demonstrate the effectiveness of the model in maintaining the road network, four population-based metaheuristic algorithms, namely the genetic algorithm (GA), particle swarm optimization (PSO), the seagull optimization algorithm (SOA), and the spotted hyena optimizer (SHO), are selected to compute the real road network. The results show that SHO performed the best. Based on the initial road network, the objective function growth rate of SHO is improved by 10.13%, 2.45%, and 5.22% compared with GA, PSO, and SOA. Meanwhile, when compared with the traditional DM model without sub-network delineation, this model presents obvious hierarchical maintenance effects on different sub-networks, and the total pavement quality index (PQI) and the average PQI during the road network maintenance planning period are improved by 14.0% and 134%, respectively.