TY - JOUR

T1 - Evaluation of gain scheduled predictive control in a nonlinear MIMO model of a hydropower station

AU - Munoz-Hernandez, G.A.

AU - Gracios-Marin, C.A.

AU - Jones, D.I.

AU - Mansoor, S.P.

AU - Guerrero-Castellanos, J.F.

AU - Portilla-Flores, E.A.

PY - 2015/3

Y1 - 2015/3

N2 - This work deals with the evaluation of the performance of a predictive control applied to a nonlinear model of Dinorwig a pumped storage hydropower plant. The controller uses a piecewise-linear plant model for prediction and is gain-scheduled according to the number of active hydro-generation Units (ranging from 1 to 6). Simulated results are presented to evaluate the performance of the predictive controller, which is compared with a gain-scheduled PI controller that has anti-windup features; this controller was tuned using the current practical values. The results show that the response, to various changes in the plant operating conditions, obtained with the predictive controller is faster and less sensitive than the one obtained from the PI controller. The results also show how reduced-order models can be used for prediction, allowing the reduction of the computing time (or the computing cost) without compromising the closed-loop performance control signal. (C) 2014 Elsevier Ltd. All rights reserved.

AB - This work deals with the evaluation of the performance of a predictive control applied to a nonlinear model of Dinorwig a pumped storage hydropower plant. The controller uses a piecewise-linear plant model for prediction and is gain-scheduled according to the number of active hydro-generation Units (ranging from 1 to 6). Simulated results are presented to evaluate the performance of the predictive controller, which is compared with a gain-scheduled PI controller that has anti-windup features; this controller was tuned using the current practical values. The results show that the response, to various changes in the plant operating conditions, obtained with the predictive controller is faster and less sensitive than the one obtained from the PI controller. The results also show how reduced-order models can be used for prediction, allowing the reduction of the computing time (or the computing cost) without compromising the closed-loop performance control signal. (C) 2014 Elsevier Ltd. All rights reserved.

U2 - 10.1016/j.ijepes.2014.10.008

DO - 10.1016/j.ijepes.2014.10.008

M3 - Article

VL - 66

SP - 125

EP - 132

JO - INTERNATIONAL JOURNAL OF ELECTRICAL POWER and ENERGY SYSTEMS

JF - INTERNATIONAL JOURNAL OF ELECTRICAL POWER and ENERGY SYSTEMS

SN - 0142-0615

ER -