Stochastic Methods for Handling Uncertainties in Automotive Applications
It is no secret that the market share of electric vehicles is growing rapidly. The next generation of electric vehicles will have faster charging capabilities, have greater range and so on. In order to achieve those goals, new electric components are being developed. And to test those new electric components, high-performance test systems are needed. With high-performance we don’t just mean cutting-edge hardware, but also advanced control concepts, which allow for enhanced testing capabilities.
State of the Art Calibration
- typical automotive calibration workflows do not account for varying operation conditions, different drive cycles and other uncertainties
- higher than expected consumption and emissions
- excessive real-world testing for real driving emissions (RDE) and safety in advanced driver assistance system (ADAS)
Robustness under operating uncertainty can be achieved by using stochastic optimisation methods, instead of traditional formulations. The solution of these stochastic problems enables safe and low-emission operation of powertrains under uncertain and unpredictable conditions such as those encountered in real-world driving. This is also highly relevant for the calibration of combustion engines due to current real driving emissions (RDE) regulations. In order to calibrate a powertrain for general real-world RDE drive cycles, it is important to consider the possible variability of real-world driving in the powertrain optimisation.
For that purpose, concepts for generating realistic RDE drive cycles, which fulfil all regulatory constraints as required for real-world testing, were investigated. The goal is to generate a large number of driving cycles that cover the operating conditions occurring in reality (and also their statistical distribution) as closely as possible. Taking uncertainties into consideration during the development and calibration of powertrain systems, the expensive real-world testing efforts can thus be reduced considerably.
Goals and Contribution
Stochastic Optimisation for Calibration
- inclusion of uncertainties in optimization formulation
- adequate optimization goals depending on risk-aversion
- efficient approximation methods for statistical quantities
Interested in more information?
Watch our project video on this exciting topic!
Wasserburger, Alexander, Alexander Schirrer, and Christoph Hametner. "Stochastic Optimisation for the Design of Energy-Efficient Controllers for Cooperative Truck Platoons, opens an external URL in a new window." International Journal of Intelligent Transportation Systems Research (2022): 1-11.
Wasserburger, A., N. Didcock, and C. Hametner. "Efficient real driving emissions calibration of automotive powertrains under operating uncertainties, opens an external URL in a new window." Engineering Optimization (2021): 1-18.
Wasserburger, Alexander, Nico Didcock, Stefan Jakubek, and Christoph Hametner. "6.2 Risk Averse Real Driving Emissions Calibration under Uncertainties, opens an external URL in a new window." In International Conference on Calibration Methods and Automotive Data Analytics, p. 211. Expert Verlag, 2019.
Wasserburger, Alexander, and Christoph Hametner. "Automated generation of real driving emissions compliant drive cycles using conditional probability modeling, opens an external URL in a new window." In 2020 IEEE Vehicle Power and Propulsion Conference (VPPC), pp. 1-6. IEEE, 2020.
Wasserburger, Alexander, Christoph Hametner, and Nico Didcock. "Risk-averse real driving emissions optimization considering stochastic influences, opens an external URL in a new window." Engineering Optimization (2019).
Wasserburger, Alexander, Alexander Schirrer, and Christoph Hametner. "Stochastic optimization for energy-efficient cooperative platooning, opens an external URL in a new window." In 2019 IEEE Vehicle Power and Propulsion Conference (VPPC), pp. 1-6. IEEE, 2019.
- February 2017 - January 2024