When a required task is repetitive and detailed knowledge about the plant is not much available, as in many industrial processes, the iterative learning control is found to have a good performance. To apply the technique to a plant with time-delay, we must know the delay-time in advance. Otherwise, the control input may be divergent due to uncertainty of the delay-time.
In this thesis, we propose a new type of iterative learning algorithm for the plant with time-delay in which delay-time is not exactly measurable. By using a holding mechanism, it is found that the output of the plant with time-delay can be convergent even if delay-time estimation error exists, and track the discrete points of a given desired trajectory with the robust property for estimation error of delay-time.
Also, we investigate some important effects of error in the initial conditions as the proposed learning control algorithm is applied at each iteration and propose an advanced algorithm which is robust for estimation error and initial error at the same time.