Fast algorithm of constrained dynamic matrix control

Superfluous model parameters of dynamic matrix control (DMC) lead to a heavy computational burden in the computation of process predictive output and online receding horizon optimization. To overcome the above demerits of DMC. a fast algorithm of constrained DMC is presented in this paper on the basis of step response, by employing a model with fewer redundant parameters to approach step response of a plant. Because of far fewer parameters of the redundant-parameter model than non-parameterized models, the computational burden of multi-step prediction and online receding horizon optimization is greatly lightened. The effectiveness of the presented algorithm is demonstrated by computer simulations by employing step response data collecting from a heat exchanger of a hot-water network.