To support the transition to climate neutral mobility, the complexity of future powertrains will increase due to electrification and the introduction of waste heat recovery and highly efficient combustion concepts.
The majority of goods are transported by trucks and ships, which are propelled by heavy-duty powertrains. These powertrains are mainly powered by internal combustion engines. In the coming decades, internal combustion engines will remain the primary power source. To support the transition to climate neutral mobility, the complexity of future powertrains will increase due to electrification and the introduction of waste heat recovery and highly efficient combustion concepts.
Control systems are the brain of the powertrain. These systems play an essential role in minimizing fuel consumption, in making vehicle performance robust for real-world driving conditions, and in enabling the use of a wide range of sustainable fuels. As powertrain development time and costs will reach unacceptable levels with conventional control methods, industry is facing a turning point in the near future.
Development of self-learning powertrains is crucial to deal with the complexity and diversity of future ultra-clean and efficient vehicles and to minimize development time and costs. This requires integration of energy and emission management strategies at system level. My research concentrates on the development of self-learning control concepts, in which the energy efficiency of the total powertrain is optimized online by the application of smart sensors and route information.
In 2021, a novel self-learning control strategy for Plug-in Hybrid Electric Vehicles (PHEVs) is developed for operation in future zero emission zones. With this adaptive energy management strategy, PHEVs can already meet post-2025 zero emission requirements in cities now. We demonstrated increased full electric driving range and compliance with real-world Euro-VI emission limits outside the cities.
In addition, AI-based methods are applied to dramatically reduce development time and costs. For a Diesel engine use case, it has been demonstrated that application of supervised learning methods results in 97% reduction of calibration parameters compared to the benchmark production controller. Identical reduction levels in calibration effort were obtained for a thermal control system of an electric vehicle.
- Maarten Vlaswinkel (TU/e, self-learning control for highly efficient engines)
- Prasoon Garg (TU/e, AI-based powertrain control)
- Florian Meier (JKU Linz, Integrated HEV control for geofencing)
- F. Willems, F. Kupper, S. Ramesh, A. Indrajuana, E. Doosje, Coordinated air-fuel path control in a Diesel-E85 RCCI engine. SAE World Congress Experience (WCX 2019), Detroit, United States, SAE Technical Paper 2019-01-1175, DOI:
- F. Willems (2018). Is cylinder pressure-based control required to meet future HD legislation? IFAC-PapersOnLine, 51(31), 111-118.
- E. Feru, N. Murgovski, B. de Jager, F. Willems, Supervisory control of a heavy-duty diesel engine with an electrified waste heat recovery system. Control Engineering Practice, 54, 190-201.