Research On GP Technologies And Applications

This research is supported by the nature science foundation of Hunan province "A research on stability analysis of multicriteria decision functions based on GP theory (No. 05JJ4O113)" and Department of Education of Hunan province "A research on network coevolution algorithms and applications (No. 04C313).Nowadays Genetic Programming(GP) is an active project, and is regarded a powerful tool for solving multiobject decision questions, software reuse engineering, and CASE and complicated problems, analyzing and designing and software crisis. However, GP has been at its preliminary research phase on the whole. GP theory was established well by using mathematical theories, in order to study problem constraints, individual representation, automatically defined functions, fitness functions and choice strategy, genetic operators , and their related algorithm design, structure depiction and mathematical modelling and this will help to solve practical problems.In this paper, the research work of GP schema theorem, and algorithm performance optimizing technology, automatically defined function coevolution model and method, approach to determining fitness function weights, application of multiobject decision function modeling, stability analysis and software reuse technology had been done, the GP schema theorem and definition of macrocosmic and microcosmic concept were given, the schema creation internal mechanism and algorithm evolution conduct were deeply explained, the convergence of GP is validated by Markov chain analytical method. It reached a conclusion that the algorithm can be convergent to global optimization solution under the optimization value reserved condition.It would increase GP algorithm's convergence performance, quicken the process of searching optimization and avoid the negative action resulted from a excessive large population scale using the three modified methods such as kinds creation, optimization creation and...