Visual analysis of multi-dimensional metamodels for manufacturing processes

• Visuelle Analyse multidimensionaler Metamodelle für Fertigungsprozesse

Gebhardt, Sascha; Kuhlen, Torsten (Thesis advisor); Weyers, Benjamin (Thesis advisor)

Aachen (2020)
Dissertation / PhD Thesis

Dissertation, RWTH Aachen University, 2020

Abstract

Computer simulations have become vital in the design and configuration of manufacturing processes, as they often significantly reduce development time and costs. Still, complex simulations have run times in the magnitude of hours to weeks. To speed up analysis processes, surrogate models are constructed, which represent a further abstraction of the original simulation models. These are also referred to as metamodels, which can be represented as multi-dimensional functions f: R^n -> R^m. In this thesis, I investigate how the overall understanding of a dedicated production process can be improved by visualizing such metamodels. To this end, I created an interactive visualization approach based on Coordinated Multiple Views (CMV) that comprises different visualization techniques, all dedicated to support users in understanding different aspects of metamodels. This approach is prototypically implemented in the application memoSlice. This application aims at scaling from desktop environments to CAVE Automated Virtual Environment (CAVE) Virtual Reality (VR) systems. Thus, further research was done in interaction in VR to provide the means for a consistent interaction concept over all systems. Furthermore, various means of increasing the responsiveness of the whole system were investigated to allow for lag-free interaction in all target environments. I demonstrate the usefulness of my approaches by illustrating several case studies, comprising the illustration of common workflows, the presentation of the results of an external user study, and description of the integration of memoSlice into a VR factory planning application.