Research For Cellular Genetic Algorithm In Dynamic Environments

Initially, the Standard Genetic Algorithm (SGA) is widely used in the static optimization problems, and the solution achieved good effect, but many problems are often non-static in real word. The environments always dynamic with the time, such as urban traffic control problem, optimal combination of capital investment problem and so on. The optimal solutions of such problems are changing between different moments. The optimal solution of present moment is not necessarily the optimal solution in next moment. In order to search for the optimal solution, which requires optimization algorithm not only be able to quickly and efficiently find the optimal solution but also demand the algorithm can adapt to dynamic environment and more effectively to trace a series of dynamic problems of optimal solutions. Therefore, it is significant both in academic and application fields to investigate the dynamic problems and dynamic optimization algorithms. The algorithm will adapt to the environment efficiently and attain optimal solution for tracking dynamic problems.In the natural evolution process, the individual distribution has spatial structure and there exist interaction between all individuals in the local area. Due to the lack of resources in the local environment and deterioration of the environment, there may have many phenomena such as fierce competition for survival and interaction. Meanwhile, the local individuals in a dynamic environment occasionally experience sudden-onset natural disasters such as earthquakes, floods, plague, etc, while those disasters can affect the evolution of species in a certain range, that leading to the extinction of species or species optimization. In order to simulate dynamic nature more realistically through the way of introduce a disaster to study on the influence of catastrophic occurrence on the population evolve in the changing environment ,base on the spatial structure of cellular automata and the combination of genetic algorithm.The main works of this paper research for optimization problems based on dynamic environments, which includes the following areas:1) Genetic algorithm base on cellular automata——cellular genetic algorithm (CGA) is mainly introduced. The performances of CGA on optimization problem by moving peaks functions are analyzed. The performances of the CGA algorithm in dynamic environment are compared with SGA. 2) Only to maintain a certain degree of population diversity, population can be evolutionary. When population diversity is low, the algorithm is easy fall into local optimum. This paper research for diversity of population in dynamic environment, an improved algorithm--DDCGA is presented, through the introduction of a disaster the way to improve the process of evolution, diversity of population. There are the results of improved algorithm better than the others in the experiment simulation.