Studies On Initial Resetting Problem And Chattering Phenomena Of Learning Control Systems

Iterative learning control systems and repetitive control systems are typical learning control systems. In industrial applications, it’s hard to meet the strict requirement of the initial resetting of iterative learning control. The system convergence would be determined if the initial resetting of systems is inconsistent, and it’s difficult to realizing perfect tracking. In repetitive control systems, the variable structure control is introduced to widen the design techniques, but the chatting phenomena is produced by this control technology. It’s significant to develop control algorithms for system convergence improvement. In this thesis, the initial resetting of iterative learning systems and the chattering problem of the repetitive control systems are studied, and new design methods are proposed.The major results are as follows:1. For the initial resetting problem of iterative learning control systems, a new controller is designed by error tracking method to solve this problem, and the unknown continuous function in the controller is approximated by neural networks schemes. In this thesis, according to the controllers whether contain dead-time control technology or not, two controllers are designed, respectively. The convergence of control algorithms is analyzed theoretically and illustrated by numerical simulation.2. For the initial resetting problem of iterative learning control systems, an error tracking controller is designed by approximating the ideal controller directly using neural networks. Designing two controllers, respectively, whether which contains dead-time control technology or not, the convergence of control algorithms is analyzed and numerical results are given to demonstrate the effectiveness of the algorithms. 3. For the chattering phenomena of the discrete-time sliding mode control technology, a new sliding mode controller is proposed based on a novel reaching law which is designed by considering hyperbolic tangent function. For characterizing the system convergence, the boundary values of monotone descending region, absolute attractive layer, and quasi-sliding mode are derived in detail. Numerical simulation is conducted to illustrate the correctness of the three convergent boundaries.4. For the chattering phenomena of repetitive systems using ideal error dynamics method, an improved ideal error dynamics method is proposed based on hyperbolic tangent function. For characterizing the convergence of systems, the three convergence boundaries are derived in detail, and the validity of the control algorithm is illustrated by conducting numerical simulation.5. An inverter experimental platform is constructed to illustrate the validity of the discrete-time repetitive controller. Repetitive controller design methods and feedback ones are involved in this experiment. From the comparative experiment, it shows that the repetitive control methods can restrain the periodic disturbances completely and prevent the system stability from the chattering.