The characterization of web dynamics is critical for web handling and process control of roll-to-roll (R2R) manufacturing systems. One such critical characterization parameter is the tension in a moving web, which ultimately determines the efficacy of the printing process. Calculating web tension by out-of-plane vibration measurement offers a non-contacting and cost-effective solution to tension measurement during the R2R process. However, accurately modeling the tension by vibration and web moving speed is difficult, especially when the length-width ratio of a web span is low (<=3). Herein we present a learning-based prediction approach that uses the measurement of out-of- plane web vibration to derive tension in the moving web. We experimentally measured the out-of-plane vibration of short moving web spans in various scenarios by an optical fiber sensor. The sensor measures the out-of-plane displacement of the moving web wherefrom free vibration frequency can be extracted by a fast Fourier transform (FFT) algorithm. Given a set of (vibration frequency, tension, moving speed) measurement results, a learning model is trained for predicting the web tension for any pair of measured vibration frequency and moving speed. We tested the method in a R2R system. The tension prediction method is promising for measuring the tension of web spans in a R2R process.
Journal: TechConnect Briefs
Volume: TechConnect Briefs 2019
Published: June 17, 2019
Pages: 466 - 469
Industry sector: Sensors, MEMS, Electronics
Topicss: Informatics, Modeling & Simulation, Modeling & Simulation of Microsystems