Optimal Control of Traction Motor Drives under Electrothermal Constraints (Optimale controle van tractie aandrijvingen binnen elektrothermische limieten)

Permanent magnet synchronous motor (PMSM) drives combine an interesting set of characteristics including a high efficiency, a high torque-per-ampere ratio and a wide speed operating range. This makes them particularly suitable for application in (hybrid) electric vehicles. Future requirements on electric traction drives are getting increasingly stringent. A reduction of cost, weight and volume is required, while an increase of efficiency, power density and reliability is necessary as well. To achieve these goals, research and development mainly focuses on improved designs and materials. However, performance figures are not only determined by the properties of individual drivetrain components, but also in the way traction motors, power electronic converters and energy storage interact as a system. Because this interaction is (for a large part) determined by the applied control strategy, a large potential for improvement is situated in this area. This thesis investigates how advanced control algorithms can contribute to a better utilization of existing systems without change of the hardware, in order to meet the aforementioned requirements. This comes down to maximizing the drive's performance figures (torque, power and efficiency), while taking into account the electrical and thermal constraints.First, an enhanced current vector control (CVC) strategy for PMSM drives is elaborated. In each operating point, it generates optimal dq-current references to maximize the speed-torque envelope given the voltage and current constraints of the motor and/or inverter. The proposed algorithm is able to seamlessly switch between constant torque and flux-weakening control, allowing high-speed operation with a high degree of robustness to parameter variations and a fast transient response. Furthermore, a maximum-efficiency-per-Nm algorithm is included in the CVC-strategy to minimize overall motor and inverter losses.Because most failure mechanisms in motor drives are related to temperature, adequate thermal management is indispensable in meeting the conflicting future requirements on power density and reliability. Conventionally, the motor and inverter are rated assuming worst-case operating conditions, to safeguard switching devices and stator windings from excessive temperature amplitudes and variations. However, a conservative rating benefits lifetime on the one hand, but implies an (often unnecessary) restriction of performance on the other. As a solution, this thesis proposes an active approach to thermal management. Based on real-time estimates of switching devices and motor temperatures, losses are actively regulated by means of a dynamic switching frequency and current limit. In contrast to the conventional approach of applying a fixed (rated) current control limit and switching frequency, the deliverable (peak) torque output directly depends on the actual thermal state of the components. Torque is only curtailed in case the thermal constraints are effectively reached. Hence, active thermal management allows a better utilization of the drivetrain hardware, without jeopardizing reliability.The potential of dynamic DC-link voltage adaptation regarding thermal management of PMSM drives is investigated as well. With an additional converter, the bus voltage level can be adjusted to the required PMSM terminal voltage in each operating point. Doing so, switching losses can be reduced at low speed by lowering the bus voltage. At high speed, the voltage level is boosted and field-weakening operation with the associated additional losses is avoided. Because this implies a reduction of heat build-up in the switching devices, a higher torque and power output is allowed.Within the thesis, the different parts of the control algorithms are elaborated step by step. The emphasis is on the implementation aspects, with extensive experimental validation on a Matlab/Simulink-based rapid-prototyping platform. The experimental setup mimics a series-hybrid drivetrain, consisting of an 11 kW interior PMSM, an inverter and an active front-end converter. The PMSM is coupled to a dynamic load machine to test the algorithms under realistic driving conditions.nrpages: 232status: publishe