Discovery of transition rules for geographical cellular automata by using ant colony optimization

Abstract

A new intelligent algorithm of geographical cellular automata (CA) based on ant colony optimization (ACO) is proposed in this paper. CA is capable of simulating the evolution of complex geographical phenomena, and the core of CA models is how to define transition rules. However, most of the transition rules are defined by mathematical equations, and are hence not explicit. When the study area is complicated, it is much more difficult to extract parameters for geographical CA. As a result, ACO is applied to geographical CA to automatically and intelligently obtain transition rules in this paper. The transition rules extracted by ACO are defined as logical expressions rather than implicit mathematical equations to describe the complex relationships of the nature, and easy for people to understand. The ACO-CA model was applied to simulating rural-urban land conversions in Guangzhou City, China, and appropriate simulation results were generated. Compared with See5.0 decision tree model, ACO-CA is more suitable to discovering transition rules for geographical CA.