Optimal iterative learning control for product qualities in batch processes based on generalized hinging hyperplanes

A generalized hinging hyperplanes (GHH) based iterative learning control (ILC) strategy is proposed to improve product qualities of batch processes. The optimal ILC for batch process is usually based on linear model, but GHH is introduced here to construct the nonlinear dynamic model of batch process to improve the model accuracy. Because GHH is a kind of piecewise linear model, its gradient information can be obtained explicitly. With a quadratic objective function in the optimal ILC, the input of the next batch can be calculated analytically based on the linearization of GHH model. The output tracking error can be gradually reduced under the GHH-ILC method. This proposed scheme is illustrated on a typical batch polymerization reactor, and simulation results show that the GHH-ILC method can obtain better tracking performance.