Together with the German Research Center for Artificial Intelligence (DFKI), IAV has been working for years on the use of AI for engine development. In 2018, the joint “Research Laboratory Learning from Test Data” (FLaP) was founded.

In the lab in Kaiserslautern, all participants tinkered with analysis methods that enabled the use of artificial intelligence in test procedures in vehicle development. Forward-looking machine learning technologies, such as deep learning and time series analysis, were used.

Intelligent data analysis methods made it possible to monitor and improve data, control units and test benches. The researchers and engineers used neural networks, which made it possible, for example, to train an ECU with more than 50,000 setting parameters so that it optimally adjusts the input variables.

“Together with DFKI, we are transferring the diverse application potential of AI technologies to powertrain development,” says Matthias Schultalbers, then Head of Business Area Powertrain Mechatronics at IAV.

The researchers had set themselves the task of being able to better predict wear and maintenance incidents. They used their findings from time series analysis of engine test data to develop new approaches for “predictive health monitoring”.

The research project also developed an entire toolbox of AI tools that can be used to visualize numerous measurement data obtained from the neural networks in a new way.