Research On Parameter Leaning For Fuzzy Neural Network Based On Immune Genetic Algorithm

Fuzzy neural network is the product of combining fuzzy logic inference with neural network, it is the inevitable trend of the development of intelligence compound control. The learning algorithm of fuzzy neural network is vital important for the research of its theory and applications, simultaneously it is very worth paying more attention to parameters learning, since the structure learning can be changed into parameters learning. ANFIS is taken as the object of research, applying immune genetic algorithm (IGA) studies its parameter. Study content mainly as follows:1. The basic theory of ANFIS has been studied. Firstly, the adaptive network is described. Secondly the structure, the algorithm and the control mode of ANFIS are researched. Finally, the realization of ANFIS in the MATLAB is introduced.2. Immune genetic algorithm is one kind of genetic algorithm improving. Firstly, the basic theory of genetic algorithm is described. Secondly, the basic theory of immunology is described. Finally, the characteristic of immune genetic algorithm is introduced.3. The thesis has studied immune genetic algorithm (IGA) and has carried out improvement on the person, The parameter applying immune genetic algorithm to ANFIS has carried out an optimization, and carried out the simulated experiment, simulated test result has indicated immune use genetic algorithm brought forward by the main body of a book being in progress to the parameter mixing up neural networks studying , fuzzy neural networks has had a messenger approaching an ability more well , has been able to resolve the optimizing problem very good , has got ideal optimization result, The simulated experiment has indicated validity and superiority using immunity to inherit the algorithm optimization mixing up neural networks.