Interactive data visualization of chatter conditions in a cold rolling mill

Abstract

Rolling of flat steel products is an industrial process in the field of metalworking where two or more pairs of rolls reduce the thickness of a steel strip to produce a uniform thickness material. Despite it has been studied for many years, there are still unpredictable problems that can affect the final quality of the product. One of them is the so-called chatter, that is a powerful self-excited vibration that appears suddenly and limits the productivity of the process. In this paper, a visual analytics approach is considered for exploratory analysis in order to discover and understand the factors and conditions under which chatter appears. An interactive web-based interface is presented here which allows the user to explore a map of dynamical conditions and visualize relevant details of each chatter onset. A validation case is performed using real data where normal/fault conditions have been identified automatically. By means of interactive exploration, the tool allows to refine an automatic chatter detection method. Moreover, it is shown to reveal correlations between variables, providing in some expected cases data-based confirmation, but also revealing less obvious relationships. Finally, it provides context, allowing to carry out comparative analysis, both qualitative and quantitative, for different subsets of coils (e.g. different years) as well as for different working conditions