On-Line Extraction of Fuzzy Rules in a Wastewater Treatment Plant
Authors
Abstract
Fuzzy systems have an important role in knowledge extraction from the huge amount of data acquired by the industrial distributed computer control systems. The paper presents a work concerning the building of a computational model of the WasteWater Treatment Plant (WWTP) in Soporcel mill (pulp and paper). Clustering of data is developed and from clusters a set of fuzzy rules describing the process behaviour is obtained, building up simple and applicable models with reasonable accuracy. Due to the time-varying dynamics of the process, on-line learning algorithms are necessary. Evolving Takagi-Sugeno (eTS) fuzzy models are used to predict the pH values in the plant. The approach is based on an on-line learning algorithm that recursively develops the model structure and parameters during the operation of the process. Results for the second stage of the effluent neutralization process are presented and, despite the complexity, non-linear characteristics and time-varying dynamics of the process, the results show that the eTS fuzzy models are computationally very efficient and have practical relevance.
Keywords
On-line learning; TS fuzzy models; eTS fuzzy models; subtractive clustering algorithm; recursive fuzzy clustering; pulp and paper industry.
Subject
On-line Learning Algorithms
Conference
IFIP World Computer Congress, WCC'04, August 2004