Fuzzy Gain Scheduling of Subspace Predictive Controller

We present a Fuzzy Gain Scheduling (FGS) method to update Subspace Predictive Controller (SPC) gains in the presence of constraints. The method is denoted by FGS-SPC. Unlike existing approaches, FGS-SPC does not to adapt the system model by updating the subspace predictor matrices, instead it re-tunes existing control parameters solely based on future tracking error, its derivatives and derivatives of past control signal. The SPC gains are updated by applying fuzzy logic rules. The advantage of the approach is in not requiring any persistent excitations for updating the system model. Consequently, FGS-SPC has a faster convergence capability and better time efficiency compared to the conventional SPC approaches. Simulation results illustrate the efficiency of proposed method in presence of noisy data.