The KIRA project (AI methods for optimized control of electric traction drives) focuses on the holistic optimization of the operation of electric traction drives by using methods based on artificial intelligence (AI). The focus is the development of novel actuation and control concepts, which require a fundamental revision of current methods and models. With its optimization methodology, KIRA addresses the key aspects of electric drive systems for vehicles: increasing efficiency, increasing power density, reducing noise and increasing torque accuracy. The project is funded by the German Federal Ministry for Economic Affairs and Energy (BMWi).
The key is a more precise AI-based model description of various physical domains of an e-drive. They are investigated in the project in combination with classical and AI-based control methods, which will enable precise optimization of the efficiency, utilization, noise development, and torque accuracy of the e-drive. Within the project, the focus is on pure electric vehicles. However, the solutions are in principle applicable to hybrid vehicles and all other types of electric drives. To achieve these goals, KIRA focuses on an interlinked approach across all relevant drive levels.
On the one hand, ISEA's goal in KIRA is to enable the control of acoustic emissions through AI-based acoustic modelling of the electrical machine. This offers the possibility to represent the complex multi-physical relationships from excitation to radiation of acoustic emissions in a simplified way. This allows real-time capability of the acoustic model while maintaining sufficiently high accuracy. On the other hand, the thermal behaviour of the electrical machine is investigated. Again, AI-based approaches such as neural networks offer the potential to significantly reduce the computational complexity of the models without negatively affecting the accuracy. Here, ISEA demonstrates that AI-based models provide comparable results to classical modelling approaches. Another goal of ISEA is to find the optimal generation of training data for the considered modelling approaches. Both model-based and empirical data are investigated.
01st August, 2021 – 31st July, 2024
Federal Ministry for Economic Affairs and Energy (BMWi) – Funding Code 19l21030G