Data in Dialogue

How are you using data and AI?

Koch:

A prerequisite for automation is often that machines need to be able to »see«. We teach them this skill – for example through using AI-based image processing. Thanks to deep learning, robots and machines can recognize increasingly complex patterns and solve ever more difficult tasks. The key to successful learning is optimally prepared data.

Bösing:

Data is the foundation for networked production. We analyze data down to the smallest detail to optimize machines, processes and maintenance. Smart maintenance is no longer a trend, but a necessity. And AI is also driving forward the evaluation of machine data.

From process analysis to AI projects: What role does data quality play?

Bösing:

The challenge lies in identifying the data that is most meaningful for well-founded analyses. This is not always easy. Companies benefit from specific use cases that make it possible to grasp the potential of data solutions.

Koch:

Smart data before big data, I agree. By filtering large amounts of data with the help of special algorithms, we can significantly reduce the amount of training data. The results are more precise AI models and a better energy footprint.

What are you currently working on and to what extent does it involve data?

Bösing:

We need a sophisticated network to allow all devices on the shop floor to communicate with each other. As part of the MRO 2.0 project, we are developing a multilayer infrastructure consisting of edge, fog and cloud. Each layer is responsible for one aspect of data processing. We also want to create a truly end-to-end digital component twin using a gas turbine blade as an example.

Koch:

In the project KIKERP we are developing an AI assistant that evaluates old electrical appliances for refurbishing. Until enough image data is available, we are generating artificial training data that resembles the real data – also using AI. At the same time, we are researching how purely synthetic data can be used to train for an optical inspection – even before the very first product has been manufactured. This reduces costs and ensures high quality right from the start.