Application of Self-Organized Neuro-Fuzzy Identifier in Intelligent Predictive Control of a Power Plant

H. Ghezelayagh and K. Y. Lee are with the Department of Electrical Engineering, The Pennsylvania State University, University Park, PA16802, USA (e-mail: [email protected]). Abstract In an intelligent predictive controller, a neuro-fuzzy identifier predicts the response of the plant in future time interval, and provides a non-model based control approach. This identifier generates fuzzy rules and tunes membership functions automatically. A Genetic Algorithm (GA) method is implemented to encode and alter the fuzzy rules to minimize the identifier response error to the plant’s logged data. Furthermore, the membership functions are tuned by error back-propagation method. These training methods dedicate self-organization to the neuro-fuzzy identifier. The identifier is used to model a power plant in predictive control approach. The experiments are conducted to achieve a reliable training data set.