Theoretic Study On Granular Computing And Artificial Selection Algorithm

The study of this paper belongs to the cross disciplines of Computer Science, Intelligent Information Process, and Evolutionary Computing.The research fruits of this paper mainly including 4 parts: Firstly, proposed a GrC-based rough set model; secondly, proposed a GrC-based knowledge discovery algorithm; the third, proposed a GrC-based engineering modeling algorithm; and the last, the GrC-based Artificial Selection Algorithm (ASA).The main innovations of this paper are listed as follows:Build up GrC-based Rough Set model by definition of Bit Granular Matrix (BGrM), which unites the traditional Pawlak Model, Probability Model, Inclusion Degree Model of Rough Set Theory (RST) into GrC model. Propose Rough Relation Matrix (RRM). Prove the algebraic definition and rough membership function definition given by professor Z.Pawlak for rough inclusion and rough equality are same in nature. Prove the sufficient and necessary condition for rough inclusion and rough equivalence defined by rough membership function doesn't hold water, it is just a necessary condition.Propose GrC-based knowledge discovery algorithm, which makes knowledge reduction become computation of BGrM. Besides, the proposed algorithm can deal with not only the consistent decision table but also the inconsistent decision table, it can realize reduction for both attribute and attribute value. The new algorithm greatly improves the knowledge discovery efficiency.Modeling for superheated steam temperature system and traditional furnace problem by GrC-based knowledge discovery method.Propose Evolution Information System (EIS), Granular Evolution Space, Artificial Seed Cultivation Space (ASCA), Artificial Selection Operator (ASO), and ASA. By employ GrC-based knowledge discovery algorithm, ASA simulates the artificial selection process of human being, which greatly improves the efficiency of conventional Genetic Algorithm.