Multi-level model predictive control for the energy management of hybrid electric vehicles including thermal derating

DT Machacek, K Barhoumi… - IEEE Transactions …, 2022 - ieeexplore.ieee.org
DT Machacek, K Barhoumi, JM Ritzmann, T Huber, CH Onder
IEEE Transactions on Vehicular Technology, 2022ieeexplore.ieee.org
This paper presents an online-capable controller based on model predictive control for the
energy management system of a parallel hybrid electric vehicle, which is equipped with two
electric motors (EM). Its task is to minimize the vehicle's fuel consumption along a predicted
driving mission. If the fuel consumption is the only cost to be minimized, a frequent use of the
electric components ensues. This can result in their overheating, which is prevented in
practice by defining a maximum motor temperature, beyond which the motor's maximum …
This paper presents an online-capable controller based on model predictive control for the energy management system of a parallel hybrid electric vehicle, which is equipped with two electric motors (EM). Its task is to minimize the vehicle’s fuel consumption along a predicted driving mission. If the fuel consumption is the only cost to be minimized, a frequent use of the electric components ensues. This can result in their overheating, which is prevented in practice by defining a maximum motor temperature, beyond which the motor’s maximum capability is decreased drastically. This is referred to as thermal derating. A multi-level control structure is proposed that explicitly includes maximum temperature bounds on both EMs. The high-level controller is developed based on the model predictive control approach. The low-level controller is based on Pontryagin’s maximum principle and is an extension of the equivalent consumption minimization strategy. To validate the multi-level control structure, three test scenarios of increasing difficulty are presented in simulation. They include thermal disturbances as well as driving mission mispredictions and are provided to demonstrate the robustness of the control algorithm. The proposed controller is able to recover 70% of the loss of optimality of a state-of-the-art predictive controller. A comparison to a dynamic programming optimization reveals close to optimal results.
ieeexplore.ieee.org
以上显示的是最相近的搜索结果。 查看全部搜索结果