A nonlinear learning control approach for a cement milling process

In this paper, a novel parameter tuning strategy for a class of controllers is discussed. The mathematical background of the algorithm is based on variable structure systems theory, which is well known with its robustness to systems having uncertainties and imprecision in the representative model. Since the course of modeling is concerned primarily with the dominant behavior, the nonlinearities existing in the system dynamics, external disturbances and plant-model mismatches require the control engineers to design controllers, which adapt the design parameters to alleviate the stated difficulties. The method introduced in this study is applicable to controller structures, whose outputs are linear in the adjustable parameters. In order to demonstrate the performance of the proposed technique, control of a cement milling circuit is studied with time-varying set values, time-varying plant parameters and a considerable amount of observation noise.