On Energy Management Strategies for Hybrid Electric Vehicles
Licentiate thesis, 2006
Hybrid electric vehicles are characterized by the existence of an electric energy buffer in the powertrain. Compared to a conventional vehicle the existence of the buffer means an extra degree of freedom in the powertrain. The driver's request for a specific power demand can thus be met by a combination of power from the primary power unit and power from the electric buffer. Furthermore, when the vehicle is braking, energy can be regenerated and stored in the buffer.
The subject of this thisis is the control of the load distribution between the power sources in the hybrid electric powertrain. The control problem is to choose at every moment the distribution of power from the electric buffer and primary power unit that minimizes the fuel consumption and emissions in the long run. To solve this problem the efficiency characteristics of the components in the powertrain must be considered. The problem is complicated due to that the future driving is largely unknown. This uncertainty of the future driving makes it difficult from a fuel efficiency viewpoint to compare taking energy from the buffer with taking energy from the fuel tank. In this thesis this problem is handled by using a model based information perspective. The controller is derived by Stochastic Dynamic Programming using a simple model of the powertrain and a Markov process model of the future driving. The resulting controller minimizes the expected fuel consumption with respect to the Markov process model of the future driving.
In cooperation with Volvo Cars a controller supplied by Volvo Cars is compared with a controller that is derived with Stochastic Dynamic Programming on drive data collected in Gothenburg. When simulated on the Gothenburg drive data the controller derived with Stochastic Dynamic Programming reduces the fuel consumption by 5\%.
The model based information perspective is used to study which type of information that should be supplied to the controller. This thesis investigates if information from GPS and digital maps can be used to schedule the use of the buffer so that fuel consumption reductions are achieved. A novel algorithm for predictive control of the power distribution in parallel hybrid vehicle powertrains is presented. The algorithm is based on dynamic programming and can be used when the future route is known to the controller. In simulations the algorithm achieves close to the theoretical minimum fuel consumption.
Hybrid Electric Vehicle
Stochastic Dynamic Programming