Multiobjective Optimization in Pavement Management by Using Genetic Algorithms and Efficient Surfaces

Genetic algorithms (GAs) are becoming an increasingly popular way to search huge solution spaces to find good solutions. Pavement management problems are specialized scheduling problems for which the solution space grows exponentially with the problem size so that the solution space size becomes unmanageable by “true” optimization techniques very quickly. Pavement management is thus ideally suited for directed random search heuristics such as GAs. The formulation of a typical general project-level pavement management problem for solution by GAs is described. Both the single- and the multiobjective cases are discussed, and the results of a series of tests of the performance of the formulations are presented. A very useful insight is then presented to show how the general network problem can be solved using project-efficient surfaces. It is concluded that GAs are an extremely flexible and robust approach to solving myriad forms of pavement management problems and provide a rich area for future research. It is also concluded that using efficient surfaces to break down the network problem into project subproblems holds a great deal of promise for overcoming some of the existing problems of optimization in pavement management.