Hierarchické prediktivní řízení pro dynamické rozdělení výkonu hybridních vozidel s palivovými články

Abstract

In order to reduce emissions of the transport sector, fuel cell hybrid vehicles (FCHVs) constitute a promising alternative as they have zero local emissions and overcome the limited range of electric vehicles. The power management of the propulsion system poses many challenges since it is a highly nonlinear, constrained, strongly coupled, multiple-input multiple-output (MIMO) system. The control objectives aim at dynamic power delivery, minimization of hydrogen consumption and charge sustainability of the battery. This thesis presents a hierarchical model predictive control (MPC) with three levels approaching the control problem on different time scales. The high-level control (HLC) implemented as a nonlinear MPC optimizes the static power split between battery and fuel cell system. The intermediate-level control (ILC) uses static optimization to determine the optimal operating point of the air supply. The lowlevel control (LLC) is a nonlinear MPC and tracks the reference trajectories received from the higher levels. The hierarchical MPC is evaluated on a detailed model of an FCHV using the worldwide harmonized light vehicles test cycle. Utilizing predictive information about the power demand, the HLC provides a power split that assures charge sustainability of the battery and only deviates by 0.2% from the optimal solution in terms of hydrogen consumption. Due to the predictive behavior and inherent decoupling capability of an MPC, the LLC achieves dynamic power delivery while explicitly considering the system constraints caused by prevention of oxygen starvation and limited operating range of the compressor. Moreover, the actual hydrogen consumption deviates only by 1% from the hydrogen consumption that is predicted by the HLC. Even for uncertain power demand prediction, the LLC attains dynamic power delivery by deviating from the reference trajectories to relieve the fuel cell system when operating under system constraints.

In order to reduce emissions of the transport sector, fuel cell hybrid vehicles (FCHVs) constitute a promising alternative as they have zero local emissions and overcome the limited range of electric vehicles. The power management of the propulsion system poses many challenges since it is a highly nonlinear, constrained, strongly coupled, multiple-input multiple-output (MIMO) system. The control objectives aim at dynamic power delivery, minimization of hydrogen consumption and charge sustainability of the battery. This thesis presents a hierarchical model predictive control (MPC) with three levels approaching the control problem on different time scales. The high-level control (HLC) implemented as a nonlinear MPC optimizes the static power split between battery and fuel cell system. The intermediate-level control (ILC) uses static optimization to determine the optimal operating point of the air supply. The lowlevel control (LLC) is a nonlinear MPC and tracks the reference trajectories received from the higher levels. The hierarchical MPC is evaluated on a detailed model of an FCHV using the worldwide harmonized light vehicles test cycle. Utilizing predictive information about the power demand, the HLC provides a power split that assures charge sustainability of the battery and only deviates by 0.2% from the optimal solution in terms of hydrogen consumption. Due to the predictive behavior and inherent decoupling capability of an MPC, the LLC achieves dynamic power delivery while explicitly considering the system constraints caused by prevention of oxygen starvation and limited operating range of the compressor. Moreover, the actual hydrogen consumption deviates only by 1% from the hydrogen consumption that is predicted by the HLC. Even for uncertain power demand prediction, the LLC attains dynamic power delivery by deviating from the reference trajectories to relieve the fuel cell system when operating under system constraints.