Research On A Self-Organizing Fuzzy Neural Networks Algorithm

Fuzzy logic has the ability of mimicking human reasoning capabilities, and it is widely used in pattern identification, expert systems,fault diagnosis,system identification and in the control of nonlinear systems. Neural networks have a few advantages, such as adaptive learning, parallelism, fault tolerance, and generalization. Fuzzy neural networks combines the advantages of both, overcoming the"black-box"nonlinear mapping from input to output, and also, the subjectivity of selecting fuzzy rules by human. Fuzzy neural networks attract more and more attentions from academic circle. It is predicted by many experts that fuzzy neural networks would become the core technique in the region of intelligent control in 21 century.To overcome the uncertainties of the structure of fuzzy neural networks, a new cluster algorithm is proposed to identify the structure of fuzzy neural networks, and after that, back propopagation algorithm is used to identify the parameter of fuzzy neural networks. This clustering algorithm can on-line partition the input data, pointwise update the clusters, and self-organize the fuzzy neural structure. No priori knowledge of the input data distribution is needed for initialization. All rules are self-created, and they grow automatically with more incoming data. There are no conflicting rules in the created fuzzy neural networks.To test the effectiveness of this fuzzy neural networks algorithm, we use it to identify nonlinear Single-Input-Single-Output, Multiple-Input-Single-Output, Multiple-Input-Multiple-Output dynamic systems. We also use it to predict time-series datas. It is demonstrated by the simulation that this fuzzy neural networks can effectively identify nonlinear systems.Based on the identification of nonlinear model, self-organizing fuzzy neural networks is also used to the inverse control of inverted pendulum. The inverse model of inverted pendulum is identified by fuzzy neural networks and its balance control is implemented. It is showed by the simulation that the fuzzy neural networks can effectively control inverted pendulum.