This paper presents extensions to the optimal linear regulator control problem that allow for time-varying measurable disturbances and partially measurable disturbances. The utility of these new algorithms is domonstrated by stirred-tank reactor and staged binary distillation column examples.
K E Y WORDS Linear-quadratic optimal regulators Time-varying disturbances
Measurable and partially measurable disturbances Linear control law design
subject to a quadratic performance measure 11 10 the optimal controller is given by a multivariable proportional feedback control law or a linear regulator
The optimal gain is computed via dynamic matrix Riccati equations.' A number of important extensions in optimal regulator control have appeared since Kalman's original work. These include the work of Anderson3 who presented a transformation that allowed for the inclusion of constant measurable disturbances, d, in the state model,
Anderson showed that the optimal regulator for this case is a multivaribale proportional feedback controller with a constant predictive or feedforward element,
where xe is the new equilibrium state due to the constant disturbances d. The new equilibrium