Multi-Objective Optimization of Torque Distribution inHybrid Vehicles

Sammanfattning: Electrification is one of the mega-trends in the transportation and automotive industry today. Boththe alarming environmental conditions and the ever decreasing fuel reserves are driving the shifttowards hybrid, all electric and alternative fuel source vehicles. This thesis work is another smallstep towards studying, addressing and handling this issue while also laying the groundwork for developingand moving towards more efficient and commercially viable vehicles.This thesis work aims at investigating the trade-off offered by optimal control techniques betweenenergy consumption and reference tracking for torque allocation to the various actuators available topropel a hybrid electric vehicle. The particular vehicle under consideration has two electric motorsat the rear wheels and an internal combustion engine along with an integrator starter generatordriving the front wheels. The torque allocation problem is originally solved by proposing a one stageoptimization strategy (OSOS) that takes into account actuator limits, losses, and objectives throughconstraints. The performance of this formulation is presented over two simulated test tracks on apareto front where the advantage on relaxing complete reference tracking becomes visible. Next,two new formulations each as a two stage optimization strategies (TSOS) are proposed, the mainobjective being to split the original formulation into two parts. One addressing energy optimalityand the other addressing reference tracking of total wheel torque and yaw moment request fulfilment.These formulations are then similarly investigated and presented in comparison with the originalformulation. In developing the formulations, an assumption about the loss models is made andthe problem size of the second stage quadratic program is significantly reduced. The problems areappropriately scaled and made mathematically robust to handle the constraints and inputs in theoperating range. As reference tracking for the vehicle is split into lateral and longitudinal torquerequests from the vehicle, this becomes a multi-objective optimization problem. To further studythe behaviour of these formulations, they are given constant inputs and simulated over a single timestep. The effect of changing hybridization level, i.e, the amount of electrical energy used comparedto fuel energy on the behaviour of these formulations is also explored. One of the effects of the twostage formulations was the confinement of solutions within a reasonable error for the majority ofchosen weights due to the energy considerations in the first stage. The proposed formulations wereable to generate results close but not equal to the original formulation on the pareto front. Anotherfinding was that due to the implementation of two actuators at the rear of the vehicle, a desired yawrate could be achieved at no additional energy cost because of regenerative and propulsive torquesgenerated respectively on either side of rear axle for torque vectoring. Furthermore with a dedicatedsolver, the TSOS could present an interesting alternative to enhance independent development invehicle dynamics control and energy management of the vehicle.