Non-linear modelling of drum-boiler-turbine unit using an evolving Takagi-Sugeno fuzzy model Online publication date: Sat, 21-Mar-2015
by Jinxing Lin, Jiong Shen
International Journal of Modelling, Identification and Control (IJMIC), Vol. 12, No. 1/2, 2011
Abstract: The boiler-turbine unit (BTU) is a highly non-linear, multivariable and time-varying system. The normal linear or quasi-linear modelling cannot reflect the real non-linear characteristics of the BTU, degrading control precision and operating performance. This paper deals with non-linear modelling of a drum-type BTU using an evolving Takagi-Sugeno (T-S) fuzzy model. A novel method based on fuzzy clustering, least-squares and genetic algorithms (GA) is proposed to construct a 'parsimonious' dynamic T-S fuzzy model with high generalisation ability. In this method, a self-organising fuzzy model generation strategy based on GA is proposed for selecting the optimal structure (including the number of rules and input variables) and antecedent parameters of the fuzzy model. Furthermore, the modified Akaike information criterion is introduced as the evaluation function of GA, which enables the self-organising strategy to choose an optimal fuzzy model with a good trade-off between fitting the training data and keeping the model simple. The simulation results show that the developed dynamic T-S fuzzy model can accurately approximate the global behaviour of the non-linear physical model with a low number of rules and fewer input variables. Further, based on the obtained T-S fuzzy model, valid control strategy studies such as predictive control can be developed.
Online publication date: Sat, 21-Mar-2015
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Modelling, Identification and Control (IJMIC):
Login with your Inderscience username and password:
Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.
If you still need assistance, please email firstname.lastname@example.org