The DMC(Dynamic Matrix Control) algorithm evolved from a technique of representing process dynamics with a least square formulation to minimize the integral of the error/time curve. DMC makes it possible to solve complex control problems on a digital computer which are not solvable with traditional PID control concepts. Incorporation of the process dynamics into the synthesis of the design of the DMC makes it possible to maintain an awareness of deadtime and unusual dynamic behavior. And the combination of quadratic programming and DMC can handle constraints on manipulated variables and controlled variables explicitly.
In this work tuning guidelines for Dynamic Matrix Controller design for both single-input-single-output systems and multi-input-multi-output systems have been developed considering performance and robustness. For open loop stable systems these tuning rules provide good performance and robustness under the presence of model/plant mismatch. But the original DMC algorithm cannot be applied directly to open loop unstable system. So we combine Linear quadratic Regulator (LQR) with DMC resulting in a cascade control structure. This control structure shows better control performance than the others such as Multiloop Single Variable Control, Multivariable Control with Compensators, PI Controllers with LQR, while remaining robust in the face of modelling errors.