On-line Construction and Rule Base Simplification by Replacement in Fuzzy Systems Applied to a Wastewater Treatment Plant
| Authors: | Victor Jose, Polytechnic Institute of Leiria, Portugal
Dourado Antonio, University of Coimbra, Portugal
Angelov Plamen, Lancaster University, United Kingdom |
|---|
| Topic: | 8.3 Modelling & Control of Environmental Systems |
|---|
| Session: | Modeling and Control of Wastewater Treatment Plants |
|---|
| Keywords: | On-line learning, eTS fuzzy models, recursive fuzzy clustering, rule base simplification, transparency. |
|---|
Abstract
Abstract: Evolving Takagi-Sugeno (eTS) fuzzy models are used to build a computational model for the WasteWater Treatment Plant (WWTP) in a paper mill. The fuzzy rule base is constructed on-line from data using a recursive fuzzy clustering algorithm that develops the model structure and parameters. In order to avoid some redundancy in the fuzzy rule base mechanisms for merging membership functions and fuzzy rules simplification are introduced. The rule base simplification is done by replacement allowing the preservation of the rule (cluster) centres as data points belonging to the original data set. Results for the WasteWater Treatment Plant show that it is possible to build less complex models and preserve a good balance between accuracy and transparency.
