Automated controller elicitation and refinement for power trains of hybrid electric vehicles

Hybrid electric vehicles (HEV) incorporate a combination of internal combustion devices and electrical generators as sources for the energy to drive the vehicle and to operate on-board accessories. The efficient management of the power train of HEV requires the careful coordination of these devices and of ancillary power reservoirs such as batteries or flywheels. The authors present initial results of investigations leading to rule-based controllers for various HEV power train architectures. Two major type of methods for controller design are presented and described. The first class of techniques is based on approximation and refinement of existing controllers by automated learning techniques while the second directly produces controllers by analysis of approximate models of the plant and of logical representations of system goals and their interactions. The authors discuss both approaches presenting also preliminary results of application of the first class of techniques to the approximation and refinement of an existing generalized proportional integral controller