Today’s industrial plants continuously produce log data on the references of measurements, error reports, and documented user interventions. The existing solutions show some constraints, though:
- They deal mostly with analysis and optimization of local problems.
- They are not suitable for the support of complex functions, such as showing up unknown cause-effects relationships or the prediction of events.
- They work on relatively small amounts of input data and do not provide suitable parallelization or scaling strategies.
The main focus is to use the potential that sources in the analysis of log files in relation to the system level along the entire production or process context. The identification of cause-effects relationships at the system level would allow an optimization of industrial plants and corresponding processes.
Particular challenges to be addressed are:
- Deriving appropriate analysis methods and evaluating these analysis methods
- Selecting appropriate parallelization or scaling strategies in order to be able to: (a) explore very large search areas (b) waive pruning-methods
- Evaluating the results of the analysis and the analysis models in real time
- Analyzing deployment models (e.g. on-site, cloud-based, mixed)
“The optimization of log data in the industrial sector was the goal of the joint project of SDIL and ABB” says Benjamin Klöpper, Principal Scientist at ABB Research Center. “The methods developed and tested in the SDIL project are be used to develop new service offerings for operators of production facilities.”
Data Innovation Community
ABB, IBM, KIT
Dr. Benjamin Klöpper, firstname.lastname@example.org