Nowadays and due to environmental concerns, the automotive industry is greatly involved with the reduction of on road vehicle pollutant emissions. Transport electrification is considered as a key technology to achieve this goal. However, an extensive research and development of high efficient, low cost, optimized and reliable electric propulsion systems is required in order to achieve these objectives.

Electric machines can be considered as the main elements of Electric Vehicle (EV) propulsion systems. These machines must fulfill a number of particular characteristics, such as high power and torque densities, high efficiency and fault tolerance capability. Due to these requirements, Permanent Magnets Synchronous Machines (PMSM) are the most established technologies in current hybrid and electric vehicles. However, their high power density is achieved thanks to rare-earth magnetic materials, whose use involves economic, environmental and politic issues. Therefore, alternative rare-earth free machines are being investigated, such as Permanent Magnet assisted Synchronous Reluctance Machines (PM-assisted SynRMs).

Regarding the aforementioned automotive synchronous machines, the development of efficient and reliable torque control strategies is of great importance. Due to volume optimization requirements, electric machines suffer from high non-linearities as a consequence of magnetic saturation. Therefore, an accurate machine electromagnetic model is required by the controller, where magnetic saturation phenomena cannot be neglected. Besides, mismatches between Finite Element Models (FEM) and real prototypes are usual due to manufacture tolerances, operation temperature and ageing, leading to electric parameter deviations. Conventional control approaches highly depend on machine electrical parameters, and they can eventually loss controllability when the stator voltage limit is reached, i.e., in field weakening or deep field weakening operation. Additionally, fault tolerance is considered as crucial by the EV industry, being limp-home capability a desirable feature for future electrified vehicles. In this sense, a great amount of literature focused on both topics can be found in the scientific literature.

In this thesis, a general overview of an EV powertrain system is provided. The existing electric machine technologies for EV applications and the most common torque control approaches for synchronous machines are then reviewed. Once an accurate electromagnetic model for three-phase synchronous machines including magnetic saturation is provided, Proportional Integral (PI) based Field Oriented Control (FOC) and second order Sliding Mode Control (SMC) approaches are thoroughly detailed.

Once the state of the art has been reviewed, this thesis proposes two novel control solutions which improve the robustness of current automotive synchronous machine torque control strategies in field weakening and deep field weakening operation regions, ensuring a reliable and efficient system performance under parameter uncertainties. These strategies combine a Voltage Constraint Tracking (VCT) feedback algorithm with the well established Look-up Table (LUT) set-point generation approach. The proposed strategies are conceived to drive the system into a pseudo-optimal operation point when the controllability is not guaranteed, ensuring proper field weakening operation and maintaining the optimal current set points when deviations are small and system controllability is ensured by LUT data.

On the other hand and considering the importance of fault tolerance in the automotive industry, this work also proposes a novel sensorless control algorithm to provide limp-home capability under machine encoder or resolver failures. This strategy combines a Phase Locked Loop (PLL) based position estimator for medium to high speed ranges, and a High Frequency Injection (HFI) technique for low speeds and standstill. The work also focuses on the development of a robust strategy to ensure smooth transitions between both PLL and HFI techniques. The proposed sensorless algorithm is successfully combined with the novel VCT/LUT approach, providing the required limp-home operation.

Simulation and experimental results obtained in an automotive grade 51 kW PM-assisted SynRM are provided, demonstrating the validity of the proposed strategies.