Paper Info
Title
An affine fuzzy model with local and global interpretations
Abstract
Abstract: Universal approximation properties of Mamdani fuzzy model are well known. On the other hand, Takagi-Sugeno fuzzy model with affine consequent was thought to be a local approximator of the dynamics. However, it can also be tuned to be an universal approximator, but loosing its local interpretation. In this paper, an innovative affine global model with universal approximation capabilities which maintains local interpretation is introduced. This novel model can be considered a generalization of Takagi-Sugeno affine fuzzy model, and is based on decoupling the dynamic parameters of the system at the fuzzification step. We demonstrate how this new model can exactly match non-linear functions expressed either as product form or additive form. Finally, we apply all the above to model a multivariable tank, analyzing the modelling errors obtained, depending on the model used: Mamdani, Takagi-Sugeno or the affine one with decoupled dynamics.
Year
DOI
Venue
2011
10.1016/j.asoc.2011.03.018
Appl. Soft Comput.
Keywords
Field
DocType
takagi-sugeno fuzzy model,takagi-sugeno affine fuzzy model,global interpretation,universal approximation property,local approximator,mamdani fuzzy model,new model,local interpretation,innovative affine global model,fuzzy control,dynamics,fuzzy systems,universal approximation capability,universal approximators,novel model,nonlinear systems,modelling,nonlinear system,fuzzy system
Affine transformation,Neuro-fuzzy,Defuzzification,Fuzzy classification,Fuzzy set operations,Fuzzy set,Artificial intelligence,Fuzzy control system,Fuzzy number,Mathematics,Machine learning
Journal
Volume
Issue
ISSN
11
6
Applied Soft Computing Journal
Citations 
PageRank 
References 
5
0.42
27
Authors
4
Name
Order
Citations
PageRank
Fernando Matía114817.47
Basil M. Al-Hadithi28511.53
Agustín Jiménez320317.61
Pablo San Segundo421213.79