Cellular Genetic Algorithm with Density Dependence for Dynamic Optimization Problems

Abstract

For dynamic optimization problems, the aim of an effective optimization algorithm is both to find the optimal solutions and to track the optima over time. In this paper, we advanced two kinds of cellular genetic algorithms inspired by the density dependence scheme in ecological system to solving dynamic optimization problems. Two kinds of improved evolution rules are proposed to replace the rule in regular cellular genetic algorithm, in which null cells are considered to the foods of individuals in population and the maximum of living individuals in cellular space is limited by their food. Moreover, in the second proposed rule, the competition scheme of the best individuals within the neighborhoods of one individual is also introduced. The performance of proposed cellular genetic algorithms is examined under three dynamic optimization problems with different change severities. The computation results indicate that new algorithms demonstrate their superiority respectively on both convergence and diversity.