Översikt
- Datum:Startar 29 augusti 2025, 10:00Slutar 29 augusti 2025, 13:00
- Plats:Lecture hall HC3, Hörsalsvägen 14, Chalmers University of Technology, 412 58 Göteborg, Sweden.
- Opponent:Tijs Donkers, Associate Professor, Department of Electrical Engineering, Eindhoven University of Technology, Eindhoven, Netherlands.
- AvhandlingLäs avhandlingen (Öppnas i ny flik)
Modern electric vehicles, particularly plug-in hybrid electric vehicles (PHEVs) and battery electric vehicles (BEVs), often feature over-actuated powertrains with modular architectures that offer high degree of control freedom. Efficient energy management is essential to maximize the operational efficiency (driving range) of these EVs, without compromising performance.
This thesis presents an efficient model-based supervisory energy management framework that co-optimizes torque allocation and discrete decisions online, in over-actuated EVs. Control models capturing powertrain hybrid dynamics are explicitly incorporated into the optimization problem to minimize energy consumption and reduce frequent discrete transitions that degrade performance. Time-scale separation in the supervisory control structure is leveraged to ensure model tractability. To solve the resulting mixed-integer nonlinear problems, customized solution strategies are proposed that exploit their problem structures: relaxation-based methods for PHEVs and bilevel programming approach for BEVs. The framework is implemented using model predictive control and validated with high-fidelity simulations.
The results demonstrate that explicit inclusion of engine dynamics in power-split optimization yields up to 10 % energy savings over a rule-based baseline in PHEVs. At least an additional 3.6 % energy savings is achieved by co-optimizing torque allocation and discrete decisions in both EVs with only a marginal increase in discrete transitions.
Finally, this work also investigates the integration of torque vectoring mechanisms in dual-motor BEVs through a comprehensive torque distribution strategy. This proposed approach enhances energy efficiency, steering performance and dynamic handling, illustrating the potential in advancing the performance envelope of multi-motor EVs.
This thesis presents an efficient model-based supervisory energy management framework that co-optimizes torque allocation and discrete decisions online, in over-actuated EVs. Control models capturing powertrain hybrid dynamics are explicitly incorporated into the optimization problem to minimize energy consumption and reduce frequent discrete transitions that degrade performance. Time-scale separation in the supervisory control structure is leveraged to ensure model tractability. To solve the resulting mixed-integer nonlinear problems, customized solution strategies are proposed that exploit their problem structures: relaxation-based methods for PHEVs and bilevel programming approach for BEVs. The framework is implemented using model predictive control and validated with high-fidelity simulations.
The results demonstrate that explicit inclusion of engine dynamics in power-split optimization yields up to 10 % energy savings over a rule-based baseline in PHEVs. At least an additional 3.6 % energy savings is achieved by co-optimizing torque allocation and discrete decisions in both EVs with only a marginal increase in discrete transitions.
Finally, this work also investigates the integration of torque vectoring mechanisms in dual-motor BEVs through a comprehensive torque distribution strategy. This proposed approach enhances energy efficiency, steering performance and dynamic handling, illustrating the potential in advancing the performance envelope of multi-motor EVs.
