Optimal Planning of Public Works Using Improved Genetic Algorithm

Abstract

The purpose of this research is to establish an appropriate program for the long/middle term planning of public works. This planning needs to satisfy various constraints such as the order relations among construction projects and annual budget. Therefore, in order to solve this planning problem as a combinatorial optimization, it is necessary to search for solutions in very complicated solution space. In this optimization, Genetic Algorithm (GA) is more likely to be trapped by a local optimum due to the lack of diversity among individuals. In this study, an attempt is made to develop a decision support system for the optimal planning by using Improved GA. The proposed method adopts a local search which aims at satisfying constraint of the order relations. This local search can prevent that the search of GA tends to converge to a local optimum. Numerical examples are presented to demonstrate the applicability of the proposed method through the comparison with Simple GA computation.