Prioritising and scheduling road projects by genetic algorithm

The optimum timetable for implementing a set of road projects so as to achieve maximum investment effectiveness can be found by ranking or Goal Programming (GP) under the assumption that the payoffs of all the projects are divisible and proportional to their proportions undertaken. This assumption is valid for upgrading projects (Type 1) but not for new ones (Type 2). When this difference is taken into account, neither ranking nor GP are effective methods to find the optimum timetable. This paper develops a genetic algorithm (GA) to address this problem. The GA uses the ranking vector of the projects as a GA individual and then transforms it into a project proportion matrix by imposing the budget constraint. Experiments show that the GA can find the optimum solution with an acceptable accuracy, and when the projects are differentiated between Type 1 and 2, the GA finds the optimum timetable that is different from that in the case of all Type 1 projects.