Identification and control of an extruder using multivariable algorithms

A bench-scale single-screw extruder was identified using recursive multivariable least-squares and extended least-squares algorithms. A second-order, discrete multivariable process model having nonminimum-phase characteristics was obtained. The extruder was then controlled using two multivariable self-tuning regulators: a quadratic algorithm and a state-space eigenvalue-placement algorithm.