An Adaptive Predictive Control Method Based on Recursive Closed-loop Subspace Identification

In order to deal with model uncertainty problem for linear systems with time-variability, which exists due to noise perturbances in the model, an adaptive predictive control method is proposed based on recursive closed-loop subspace identification. A closed-loop subspace predictive control algorithm is constructed by making improvements to the closed-loop subspace predictive control algorithm through the incorporation of a PID-type objective function. This proposed algorithm is implemented online using a recursive algorithm, with fixed-size input and output data, and a simple, direct update method, which replaces LQ decompositions, for improving computational efficiency. Simulations prove this closed-loop subspace control algorithm to be efficient, predictive, and adaptive.