A problem governments commonly facing is the lack of maintenance and rehabilitation (M&R) for the large scale of road network due to the shortage of fund. The unreasonable M&R strategy accelerates the deterioration of the road network in the life cycle, which in turn aggravates the fund gap. Therefore, it is necessary to improve the utilization efficiency of limited M&R fund regarding the goal of excellent and good road. This article proposes a spatial-temporal optimization model based on Markov process and dynamic programming, which can provide optimal M&R strategy for the road network in the multi-year planning horizon. It maximizes the average proportion of excellent and good road (EGP) of the road network in the designed time horizon under a fixed and limited budget. Considering the dynamic and random changes of the road network condition over time, a dynamic solution algorithm based on genetic algorithm (GA) is proposed to solve the combinatorial problem of multi-year network-level M&R programming. The proposed method is calibrated and verified by a case study of the road network of Hanjiang County, Jiangsu province, China. The results indicate that the multi-year road network M&R scheme obtained by spatial-temporal optimization model is better than the single-year optimization model in terms of M&R fund and EGP of the road network. Comparing to the single-year model, our model saves up to 17.80% of M&R fund under insufficient M&R investment and improves the EGP of the road network by 17.49% at the end of the planning period. In addition, it is found that there is a quadratic function relationship between investment amount and EGP of the road network, which provides evidence for decision-making of reasonable M&R fund investment to road networks. Keywords: Maintenance decision-making; Road network; Multi-year; Dynamic programming; Genetic algorithm.

CICTP