Control method of winding quality in shrink sleeve labels converting process

Krzysztof Krystosiak, Wojciech Werpachowski

Faculty of Production Engineering, Warsaw University of Technology.

DOI:

https://doi.org/10.7494/cmms.2015.3.0545

Abstract:

In manufacturing practices of most big printing companies there are collected data and records of process parameters. Gathering information from this data in form of developed models and rules such as data mining, which uses statistical methods or Artificial Intelligence, Artificial Neural Networks, Decision Trees, Expert Systems, and others are subjects of interdisciplinary fields of science. In shrink sleeve production, the effects of using data mining tools not only improve the quality of the shrink sleeve and winding process but also reduce manufacturing costs. This paper describes developed models of Artificial Neural Networks (ANN) to be used for predicting initial tension parameters and winding speeds for each and every new design of shrink sleeve labels. Every individual design of shrink sleeve labels has a lot of factors. Some of them are more significant, some of them less. The aim of this paper is to choose the significant factors and build a model of ANN in the learning process by using the collected data. Finally, when the ANN model is computed, it can be used for predicting key winding parameters of new shrink sleeve label designs. For the company, this will result in saved time of experimental selection during converting winding parameters like tension and speed. It will also minimize the risk of defects occurrence with incorrect winding parameters.

Cite as:

Krystosiak, K., & Werpachowski, W. (2015). Control method of winding quality in shrink sleeve labels converting process. Computer Methods in Materials Science, 15(3), 409-415. https://doi.org/10.7494/cmms.2015.3.0545

Article (PDF):

Keywords:

Shrink sleeve, Converting, Quality, Data mining, Artificial neural networks

References: