A model predictive control algorithm with soft constraints is presented. Th
e technique, finite number of weights model predictive control (FNWMPC), is
based on the selection of a proper combination of weights when constraints
are violated. Robust stability of the FNWMPC technique is studied using mu
-structured singular value techniques. The algorithm is tested through simu
lations of a lime kiln control problem and is compared to other existing mo
del predictive control techniques. (C) 1999 Elsevier Science Ltd. All right
s reserved.