| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 409254 | Neurocomputing | 2008 | 10 Pages |
Abstract
This paper presents a new neuro-fuzzy system based model, which is useful for the modelling of nonlinear dynamic systems. The new proposed model constitutes a soft computing method, namely, reasoning with a fuzzy inference system (FIS) and an optimisation by the neural-network learning algorithm. A structure, named the decomposed neuro-fuzzy ARX model is proposed. This structure is based on decomposition of the FIS. An evolution of a learning algorithm for the decomposed fuzzy model is suggested. A comparative study of dynamic system identification using conventional FIS models and the proposed neuro-fuzzy ARX model is presented for Box–Jenkins data set.
Keywords
Related Topics
Physical Sciences and Engineering
Computer Science
Artificial Intelligence
Authors
Marjan Golob, Boris Tovornik,