Sustainable pavement maintenance and rehabilitation planning using differential evolutionary programming and coyote optimisation algorithm.

CO₂ emission reduction in large-scale pavement network maintenance planning has been an immense concern. The conventional single-objective optimisation modelsoverlook environmental issues such as CO₂ emission. However, the introduced multi-objective optimisation aims to enhance the network condition and minimise CO₂ emissions simultaneously. Two single-objective (coyote optimisation algorithm and genetic algorithm) and two multi-objective metaheuristic algorithms (multi-objective coyote optimisation algorithm and non-dominated sorting genetic algorithm) are employed to assess the effectiveness of the introduced environmental approach. Pavement maintenance planning optimisation requires the deterioration function formula and treatment improvement equation to be modelled. Hence, a new machine learning method called 'differential evolutionary programming' is introduced, which can provide the output-input formula. Differential evolutionary programming predicts the pavement deterioration value and overlay improvement with R² of 0.992 and 0.970, respectively. The results indicate that the coyote optimisation algorithm's objective function is 66% lower than that of the genetic algorithm. Likewise, the multi-objective coyote optimisation algorithm reduces the first objective function by 72% on average compared to the non-dominated sorting genetic algorithm. The grey relational analysis is performed to compare single-objective and multi-objective optimal solutions. All optimal solutions presented by multi-objective modelling dominates the single-objective optimisation optimal solution based on the grey relational grade. [ABSTRACT FROM AUTHOR]