logo

Supporting Operators in processes with time delays

As part of a strong collaboration between Linköping University, ABB Sweden and Södra Cell in the EXPLAIN project, the preliminary design study of an ecological visual analytics interface, ideated and implemented by Elmira Zohrevandi (Linköping University), was published at the IEEE Visualization and Visual Analytics (top venue in the field of Visualization and Visual Analytics).

The data collection was conducted in a strong collaboration with Emmanuel Brorsson (ABB) and the preliminary evaluation was conducted with the domain expert, Andreas Darnell (Södra Cell). The paper was structured and written in a strong collaboration with Anders Ynnerman (Linköping University).

The topological view (A) enables interaction with a machine learning classifier which visualizes the historical data points that have been found most similar to the current situation. The slider is used to specify the desired variation criteria depending on the controlled process stage (highlighted in yellow). The circle chart to the right provides insight on model accuracy (mapped by color intensity) and performance (mapped on the radial axis). The situation view (B) (i.e. the focus view) shows the profile (solid line) of the key performance parameter (Kappa number) through process cycle and shows deviations from a selected optimal profile (dotted line). The strategy view (C) shows variation limits for controlled (H-factor and Effective Alkali) and monitored parameters (pressure and temperature) obtained from the historical data points. The control view (D) at the bottom shows the consequences (colored solid line) of an adjustment predicted by the ML model for the selected parameter. The extent to which the explored adjustment value affects Kappa recovery is also shown on the situation view (colored solid line).

 

 

The paper was presented by Emmanuel Brorsson from ABB Sweden, in October 2023 at the IEEE VIS Conference held in Melbourne, Australia.

Read more about Elmira’s paper on the IEEE Digital Library: https://ieeexplore.ieee.org/document/10360898  

Scroll to Top