Abstract: A hybrid predictor for predicting the output of a process, and a hybrid prediction method using the hybrid predictor. Also, a system for and a method of controlling a process using the hybrid predictor and hybrid prediction method are provided. In order to obtain a shifted prediction vector and step response coefficients, the hybrid predictor uses a parameter model which can be updated in accordance with a variation in the process. The control method includes the steps of predicting process output signals generated in a prediction horizon, based on input signals applied to the process, correcting the predicted values, based on output signal values measured at the next scan, comparing the corrected prediction value with a reference value to derive an error vector, calculating a control signal to minimize the error vector, and applying the control signal to the process. The hybrid predictor is characterized by an addition of a model parameter estimator and a model response converter.
Type:
Grant
Filed:
September 9, 1997
Date of Patent:
May 16, 2000
Assignee:
Sunkyung Engineering & Construction Limited