Design And Implementation Of Adaptive Learning Control Of Discrete Time-Varying Systems

Adaptive control can deal with the control problem of uncertain dynamic systems withunknown parameters effectively. Considering discrete time-varying systems, this thesisintroduces the basic idea and analysis of discrete adaptive control to design learning controller,and we give the new adaptive learning control algorithms. The results can be considered anapplication of the Key Technique Lemma in repetition domain, by which stability andconvergence of discrete adaptive repetitive control systems are established, including iterativelearning control and repetitive learning control. Finally, permanent magnet linear synchronousmotor (PMLSM) is taken as an example, and experiment results are presented to demonstrateeffectiveness of the proposed algorithms.In this dissertation, we focus on application feasibility and effectiveness of discreteadaptive learning control. Taking account of the issues mentioned above, the main work andachievements are summarized as follows:1. Based on the projection algorithm with dead-zone, adaptive controller is designed foruncertain discrete time-invariant systems. By the virtue of the Key Technical Lemma,theoretical analysis proves the convergence of the proposed algorithm and the stability of allinput and output signals. Furthermore, simulation results are shown that systems can achievethe desired trajectory tracking.2. For a class of repetitive tracking control problem over an infinite interval of perioddiscrete time-varying systems, a new adaptive repetitive learning algorithm with dead-zonemodification is presented. The extension of the Key Technical Lemma to repetition domain isused to prove that stability and convergence of adaptive repetitive learning control system. Atthe same time, we prove the linear growth condition which is required in the application of theKey Technical Lemma. Furthermore, when repetitive cycle tends to infinity, the system input and output signals are bounded, and the convergence of tracking error is able to be proved infinite time-intervals. The simulation results demonstrate the effectiveness of the proposedcontroller.3. Considering discrete-time systems with unknown time-varying parametric uncertainties,we present new adaptive iterative learning algorithms with satuation and dead-zonemodification. The linear growth condition which is required in the application of the KeyTechnical Lemma is proved. At the same time, the convergence and the stability of theproposed algorithm is proved. The input and output signals are bounded along the iteration axiswith non-identical and iterative-varying desired trajectory. Theoretical analysis shows thatsystem states can follow the desired trajectory totally. Numerical simulation results demonstratethat the control algorithms can effectively eliminate the periodic disturbance, and improve thecontrol accuracy.4. DSP board and ELMO power driver play important roles in linear servo control system.Software programming is used to implement discrete-time adaptive repetitive controller designrespectively. Experimental results are presented to verify the validity of the proposedalgorithms.