The Design and Analysis of an Improved Parallel Genetic Algorithm Based on Distributed System

migration strategy; Abstract. Genetic Algorithm (GA) is a powe rful global optimization search algo rithm imitating natural selection and genetic mechanism, but it has low search efficiency in the late evolving period. Parallel genetic algorithm (PGA) can improve computational efficiency and accuracy greatly, so it has become one of the m ain research fields of GA. This paper introduces the procedure of PGA in detail, analyses the migration limitations of traditional PGA, and puts forward an im proved coarse-grained PGA based on distributed system, which adopts adaptive migration strategy to evolve. This implementation can f ully tap th e computing capability of distributed system to im prove the convergence speed and am eliorate the population diversity in order to restrain prem ature convergence. The experiments show that this al gorithm not only has faster convergent speed but also has more accurate calculation precision as well as higher parallel speedup