PID self-tuning controller for aluminum rolling mill

An adaptive self-tuning gauge control system has been implemented on a cold-rolling aluminum mill, using an 80286-based microcomputer system. The mill process is modeled as a second-order system with one input (unwind tension reference) and one output (gauge error). The model parameters are estimated online using a recursive-least-squares estimation algorithm. The inputs to the estimator are correlated in time so that the estimator is not affected by the gauge measurement transport delay. Based on the estimated model parameters and a pole placement design, a control signal is calculated. The regulator structure is a filtered version of a discrete PID controller. Digital filters are used on the inputs to the controller and estimator and as part of an energy detection scheme used to ensure that estimation only takes place when the system is excited in the appropriate frequency range. Implementation details and results are described