A Performance Evaluation and Two New Implementations of Evolutionary Algorithms for Land Partitioning Problem

Many bio-inspired techniques are proposed and implemented to solve real-world applications. The number of these techniques is increasing day by day, so the researchers (especially out of computer sciences) have difficulty in deciding which technique to select for the problem. In this study, two new implementations to solve land partitioning problem and also a performance analysis of three evolutionary algorithms were carried out on this real-world engineering problem. Land partitioning is a discrete optimization problem that cannot be solved in linear time with conventional techniques. Two new implementations of automated land partitioning (ALP-DE and ALP-SS) were carried out by using differential evolution algorithm (DE) and scatter search (SS) methods. The algorithms were adapted to the land partitioning problem by being discretized with permutation coding. These two proposed methods were compared with a similar study in the published literature and a designer's plan for a project area that contains 18 blocks using a mathematical model. These proposed automatic methods (ALP-DE and ALP-SS) resulted in more successful and more appropriate partitioning plans than those of a designer in accordance with land partitioning criteria. When the comparison of these three different evolutionary algorithms was examined, the ALP-SS method showed superior performance in all blocks. The low standard deviation values of the proposed methods indicated that both methods are robust and successful tools for the land partitioning problem.