Sprecher
Beschreibung
We consider graph modeling for a knowledge graph for vehicle development, with a focus on
crash safety. An organized schema that incorporates information from various structured and
unstructured data sources is provided, which includes relevant concepts within the domain. In
particular, we propose semantics for crash computer aided engineering (CAE) data, which enables
searchability, filtering, recommendation, and prediction for crash CAE data during the development
process. This graph modeling as an example for the overall CAE process considers the CAE data
in the context of the R&D development process and vehicle safety. Consequently, we connect
CAE data to the protocols that are used to assess vehicle safety performances. The R&D process
includes CAD engineering and safety attributes, with a focus on multidisciplinary problem-solving.
We describe previous efforts in graph modeling in comparison to our proposal, discuss its strengths
and limitations, and identify areas for future work.
Today, the Finite Element method (FEM) is the preponderant tool for automotive crash simulation [1]. The large amount of complex data confronts engineers with the challenge to explore
the simulation results sufficiently, due to lack of engineering time and limitations of data storage, processing and analysis tools. This need pushed the automotive companies to uptake preand post-processing tools to be more efficient in analysing the data, with the goal to spend the
engineers time on solving the problem instead of data processing. Nevertheless, even with all
achievements so far, data flow within the companies is still inefficient. Yet, crash scenarios studied
in the development phase are just a tiny proportion of the real crashes. The need to increase the
number of simulations and the limitation of CAE engineers’ time emphasizes the importance of an
intelligent system to capture domain knowledge as knowledge graphs (KGs) for automotive, which
we call car-graph.
The modeling of CAE data is challenging since the data is complex, and several disciplines with
different requirements interact with the CAE data. However, the flexibility of graph data modeling
reflects existing uncertainties and allows the modeling to evolve. In this work, we present an initial
attempt to define a semantic representation that stores information regarding the different crash
scenarios, the vehicle design deviations during the development process, and the quantities of
interest that measure the outcome. Consequently, we propose semantic selections that follow the
development concepts, FE-modeling terminology, crashworthiness assessment quantities, and other
relevant entities. Additionally, these can be used as input for machine learning (ML) analysis, where
the graph modelling also allows storing ML results. Our vision is to use data modeling and ML to
auto-assess the cause and effect in the development process to assist engineers and, in particular,
to assess the safety of different, uncalculated crash scenarios.
As an example, we will present a summary of an industrial implementation for pedestrian
analysis. Here, the number of simulations increases enormously for each design in pedestrian analysis. We will illustrate how CAE-web visualizes this big data and allows its intuitive and easy
exploration. In this visualization, we present the traditional CAE reporting as a dynamic web
interface and graph-ML technics on this data. We propose two groups of visualization: zoom-out
and zoom-in views. Zoom-out views consider the assessment of many simulations, for example,
development trees, status tables of safety performance, or embedded results from machine learning. However, zoom-in views contain single/multiple simulation assessments and comparisons.
Additionally, the user has a multi-view functionality to combine zoom-in and zoom-out views. In
multi-view, zoom-out views are selection inputs for updating zoom-in views.
Keywords: crash-worthiness; CAE data management; CAE knowledge; Car knowledge graph; data representation
References
[1] P. Spethmann, C. Herstatt, and S. H. Thomke, “Crash simulation evolution and its impact on
R&D in the automotive applications,” International Journal of Product Development, vol. 8,
no. 3, pp. 291–305, 2009.