Web tension regulation with partially known periodic disturbances in roll-to-roll manufacturing systems

This paper deals with the problem of synthesizing feedforward control to aid the regulation of web tension in the presence of partially known exogenous inputs. The problem appears in many engineering applications including Roll-to-Roll (R2R) manufacturing systems where the governing equation for tension is nonlinear. Currently known methods for the nonlinear output regulation problem either require the solution of a constrained partial differential equation or the preview information of the signal to be tracked. In this paper, we consider the problem of regulating web tension while rejecting periodic disturbances and use a novel approach to synthesize feedforward control as the solution of a system of differential-algebraic equations, which is considerably less complex and suitable for practical implementation. We assume that the disturbance signal is the output of a known exogenous system with unknown initial conditions. A parameter identification scheme to estimate the unknown initial conditions is developed. The proposed technique is successfully applied to web tension regulation in a large R2R machine which contains real-time hardware and software that is used in industrial practice. Extensive experiments were conducted to evaluate the proposed scheme under various experimental conditions, including different web speeds and materials. We will discuss a representative sample of the results with the proposed nonlinear tension regulator and provide a comparison with a well-tuned industrial PI control scheme to highlight the benefits of using the proposed scheme.