Applying a Generalized Predictive Control Theory to a Carding Autoleveler

Abstract

Usually, a carding process is difficult to control with classic control algorithms (such as PID) for three main reasons: strong stochastic disturbance, time delays, and parameter variation. In this paper, generalized predictive control (GPC) is introduced to control such a process. First, a CARIMA model is built to describe the carding process, then the GPC control law is derived and used to design the controller for a carding autoleveler. The simulation results show that the GPC controller can greatly reduce a sliver's standard deviation and can reject measured and unmeasured step disturbance. Although this paper mainly involves the design of a feedforward-feedback GPC controller for a carding autoleveler, the methods of modeling and GPC controller design can easily be extended to a feedback case and used to design controllers for other preparatory machines' autolev elers.