Convergence Analysis of a Reduced Model Estimator Applied to an Industrial Plastic Film Extrusion Test Bed

A computationally efficient reduced model estimator is described for determining the effective position of actuators used for cross-directional control in web processes. The estimator is represented as a Radial Basis Function Network, where the basis functions in the hidden layer are defined using discrete Chebyshev polynomials. The problem is considered as a nonlinear optimization problem and solved online using a recursive prediction error method. Convergence analysis is used to compare the performance of both full and reduced model estimators, with the advantages of the reduced model being demonstrated using results obtained from an industrial plastic film extrusion test bed.