Multi-Controller Adaptive Control For Discrete-Time Systems

Multi-Controller Adaptive Control (MCAC) for linear discrete-time SISO systems with high uncertainty is proposed, based on the Unfalsified Control theory. MCAC needs the least prior information of the plant to be controlled. The scale-independent hysteresis switching logic guarantees the convergence to the stabilizing controller and the input-output stability of the closed-loop in the sense ofL_2e. In this thesis, we discuss the following aspects:1. When the controllers are SCLI, we Firstly, construct the candidate controller set based on the One-Step Ahead controllers. The Proposed algorithm can track the reference signal well. Secondly, construct the candidate controller set based on the RST controllers with two-freedom. Modify the parameters of the controllers. The Proposed algorithm is a general method. It can track the reference signal and object the disturbance well. Thirdly, give a new parameter adaptation for the numeric PID controller. Under the circumstance that there exists at least one group of parameter in the given parameter set which satisfies the control goal, the adaptation method will identify the one quickly and lock it to the control loop.2. When the controllers are not SCLI, based on the internal model principle, MCAC is proposed. The tracking of the reference input and the objection of the deterministic disturbance are achieved. The proposed algorithm does not request that the controllers are minimum phase systems. For the above schemes, the proofs of the stability are given and simulations are done respectively. The results of the simulations confirm the feasibility of the MCAC.