Research On Discovering Transition Rules Of Geographic Cellular Automata Based On Bee Colony Optimization

Modeling and simulating geographical spatial process is one of the key issues in the study of geography. Geographical spatial system is a complex giant system influenced by many factors, which brings an extremely difficult task to explain and simulate the geographical spatial process with traditional geographical methods. Fortunately, Geographic Cellular Automata (Geo-CA) was presented as a solution to this task which is becoming an important method for simulating the geographical spatial process.Geo-CA is defined by cells, neighborhood, status and transition rules. The core of Geo-CA is transition rules, which express logistic relationships for evolutionary process and determine the conversion of cells’state. So, deriving transition rules is the key to simulating geographical spatial process by Geo-CA. This paper explore a new, intelligent approach to discover transition rules for geographical cellular automata based on Bee Colony Optimization (BCO-CA). The main work focus on the definition of the mathematical model, implementation of core operators and algorithm, analysis of simulating accuracy and validity and comparison with other methods.The main research works and conclusions are as follow:1. It is given that an intelligent algorithm of discovering transition rules for Geo-CA based on Bee Colony Optimization. Based on the basic theory of cellular automata and bee colony optimization, a intelligent algorithm to discovering transition rules for Geo-CA (BCO-CA) was researched. The mathematical model, basic idea and core operators are discussed in detail. The BCO-CA was designed and implemented.2. A utility software of discovering transition rules for Geo-CA based on bee colony optimization was developed. A utility software of automatically deriving transition rules for Geo-CA was developed using C sharp program language. The software can automatically derive transition rules from sample data with a specific data format and don’t involve specific issues.3. A empirical research of simulating dynamic urban growth in nanjing was accomplished. The BCO-CA algorithm was employed to discover transition rules for urban growth in nanjng. A urban cellular automata (Urban-CA) was constructed using the transition rules and then the simulating of urban growth was achieved in nanjing. For testing the capability of the BCO-CA in discovering rules, the particles swarm optimization (PSO) and classical method of discovering transition rules for urban cellular automata-logistic regression method were used to compare with BCO-CA. A algorithm of discovering transition rules for Geo-CA based on particles swarm optimization (PSO-CA) is implemented. The PSO-CA algorithm and logistic regression was employed to discover transition rules, and then which were applied to the simulating of urban growth in the same area. The simulating results demonstrate BCO-CA is better than PSO-CA and logistic regression method.