Design And Implementation Of Repetitive Controller And Iterative Learning Controller

In industrial control, the plants often are required to do periodic tasks. The exact mathematical models of these systems are high-order and complicated. So doing research on these systems is of practical value. Based on a new reaching law, this thesis proposes the repetitive sliding mode control and the terminal sliding mode control which can eliminate the periodic disturbance and improve the control quality. A sampled data characteristic model is presented to reduce the order of the high-order non-linear models which maintains the dynamic characteristics of the original systems. The adaptive iterative learning control is applied to deal with the time-varying parameter and control of the first-order sampled data characteristic model. A permanent magnet synchronous motor is used to confirm the effectiveness of the proposed control method.The main work of this thesis includes the following aspects:1. Instead of the switching function, a continuous function is used to in the conventional reaching law to form reaching law.This paper certificats that the new reaching law makes the sliding surface reach zero in a finite time. An index reaching term is applied to accelerate the convergence rate.2. Based on the new reaching law, design the continuous-time and discrete-time sliding mode controller, respectively. This paper embeds a repetitive control active into the sliding mode to eliminate the periodic disturbance. Monotone descending region, absolute attractive layer and steady state error are introduced to describe the convergence of the switching surface.3. A new terminal sliding mode switching surface is presented according to the new reaching law. The terminal surface is improved to become a nonsingular terminal surface. A repetitive controller is bedded into the nonsingular terminal sliding mode controller.4. The first-order sampled data characteristic model is proposed on the basis of first-order time-varying system. The high-order non-linear system’s first-order sampled data characteristic model is also presented. The parameters of the first-order sampled data characteristic model often change rapidly or even unexpectedly. We use the least squares learning algorithm and the gradient learning algorithm to estimate the first-order sampled data characteristic parameters of the system, respectively. LQ optimal control method is applied to design the adaptive iterative learning controller.5. By using the permanent magnet synchronous motor, the repetitive sliding mode controller and the adaptive iterative learning controller are applied to this system. The frequency domain method and least squares method are used to build the discrete-time mathematical model of the system. Sliding mode repetitive controller is implemented on a constant second-order discrete model and adaptive iterative learning controller estimates the parameters of the first-order sampled data characteristic mode and controls the motor position.