Optimal Iterative Learning Control for Batch Processes with Model Parameter Variations Using Strong Tracking Filter

Abstract

An optimal iterative learning control (ILC) strategy is proposed to track product quality trajectories of batch processes by updating a linear time-varying perturbation (LTVP) model. To address the problem of model parameter variations from batch to batch, the LTVP model is renewed by using strong tracking filter (STF) algorithm. Comparing recursive least squares (RLS), STF can capture the changing dynamics of the process more accurately. The tracking error transition models can be built, and the ILC law with direct error feedback is explicitly obtained. Sufficient conditions of convergence are derived for the optimal ILC based on the LTVP model. It has also been proved that the tracking error will converge to a small constant but depend on the accuracy of the LTVP model error. If there is no model error, the tracking error can converge to zero. By using STF to update the LTVP model, the model accuracy is improved and the tracking control performance is also enhanced. The proposed strategy is illustrated on a typical batch reactor, and the results demonstrate that the performance of tracking product qualities can be improved under the proposed strategy when model parameter variations occur with respect to the batch index.