Time Series Analysis in Vacuum Roll to Roll Coating Technology

State of the art vacuum web coaters collect an enormous amount of data during their regular operation. This applies for machine related parameters like web tension or base pressure as well as for technology related data like power or discharge voltage for plasma processes, gas flow, gas composition etc. This information is often complemented by results of the inline monitoring setups for optical or electrical properties of the coated material.
The actual operation of the machine is normally exploiting only a minor part of the available information. This conference contribution presents a method of time series analysis. The algorithm uses data which are collected by the machine. An adapted fitting procedure is predicting the trend of various parameters. This prediction is compared to the actual values during the operation of the process. The algorithm calculates a sum of square deviations which is constantly decreasing for stable processes. Any deviation from this behavior indicates immediately an unexpected event or instability. The monitoring of several parameters in parallel reveals the interdependence and can support quick failure analysis.
The application of the algorithm is demonstrated for roll-to-roll deposition of transparent conductive oxides.