Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
408886 | Neurocomputing | 2008 | 6 Pages |
This paper presents a grey–fuzzy predictive controller that is based on fuzzy theory, grey prediction and on-line switching algorithms. The grey predictor is applied to extract key information and reduce the randomness of the measured non-stationary time-series signals from sensors, and send the prediction information to the fuzzy controller. The complete mathematical model is derived and the sufficient condition for convergence is given. To achieve better transient performance and steady-state responses, an on-line switching mechanism is adopted to regulate appropriately the forecasting step size of the grey predictor, according to the error feedback from different periods of the system response. Experimental results obtained from a plant show that the control accuracy and robustness are much improved when the proposed new method is applied.