Abstract Modern automotive test beds do not only produce the measurement data they are operated for, but also a lot of auxiliary data which is generated by sensors and software. This data is a byproduct of the testing process and not directly related to the unit under test. Nevertheless, its size can be enormous. A prototypical example of this data are software log-files that contain not only information about connected hardware devices and their current statuses, but also records of generated warnings and errors. If the information contained in these logs is used at all, then mostly by manual inspection of the files by engineers to track down certain errors and their root causes. Currently this inspection and service process is hardly supported by computational approaches. In our contribution we demonstrate how new methods of industrial big data analytics, data-mining, and visualization can be put to use in order to generate knowledge and new insights from the available data. This can help machine vendors, test bed equipment vendors and OEMs to avoid expensive downtimes. We will shine a light on the whole analytics process: How the data looks like, what information is contained, how it needs to be prepared to enable analyses and what difficulties can arise at each step. This will provide a deep view on how industrial data analytics projects should be set up and conducted while also mentioning potential obstacles and how to overcome them.
Applying industrial data analytics in automotive test bed environments
2018-01-01
13 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Machine Learning in Automotive Data Potential, Analytics Power
Springer Verlag | 2015
|Data Analytics Applied to Automotive Wiring Harness (Electrical Distribution Systems)
SAE Technical Papers | 2024
|