Design Of Fuzzy Neural Network Controller Based On Genetic Algorithms

The paper discussed GA and FNN, which has been attented by the international control academic circles in the recent years, and the paper putted forward a new kind of self-adaptive arithmetic –ISAGA(improved self-adaptive Genetic Algorithms), which can improve the quality of the solution and quicken the convergence speed of arithmetic. In the end, the paper builded up a new kind of FNN controller based on ISAGA. The main research of the paper is as follows: 1. Aimming at the inconsistency between the convergence capability and the convergence speed of GA, a kind of new ISAGA was studied, which designed a new kind of index containing a clear physical sense and smaller operations, and also realized the self-adaptive adjustment of crossover probability and mutation probability through the analysis on the shortage of evaluation index of "prematurity"degree. 2. On the basis of FNN, the paper builded up a new kind of FNN model. Compared with general model, this model not only computed simply and had faster convergence speed, but also had obvious physical sense. Furthermore, the paper demonstrated the overall approaches of the neural network, which provided the basis for use on actual modeling in theory. 3. Aimming at the shortage of traditional BP arithmetic, such as the low study efficiency, the slow convergence speed and its tending to fall into the flat area of error curve and the part infinitesimal plot, the paper discussed the study method of FNN based on ISAGA, which can improve the convergence capability of the neural network evidently. 4. The paper adopted the single inverted pendulum as the simulink object using the new FNN Controller based on ISAGA and the result proved its validity. The control theory of combining genetic algorithms with fuzzy control and network has the remarkable advantage. It indicates that the application potential in the field of intelligent control is immense. Therefore, the research in the field of artificial intelligent will be a significant research direction in the future.