Smart three-phase power converter for more electric powertrains

Abstract

In this dissertation, a smart three-phase power converter for more electric powertrains is presented. In the context of a more electric powertrain, the smartness of a power converter is tied to its ability to adapt to rapid input-frequency/voltage variations, load changes, and extreme ambient conditions fluctuations while regulating the desired variables with seamless dynamics and satisfactory power quality. Founded on these principles, this dissertation advances the state-of-the-art in power converter control schemes for more electric powertrains. More specifically, for ac-dc three-phase power converters fed by a variable-speed electric generator. At first, a model predictive step-ahead controller is developed and experimentally validated. This control scheme's uniqueness lies in its integration with an instantaneous phase-detection technique, which enables ultrafast dynamics for the dc-bus and power factor regulation when the power converter is connected to a variable-frequency/constant-amplitude input voltage supply. Despite the step-ahead controller's ultrafast dynamic performance, it is shown and experimentally verified that it suffers a performance degradation when there is a parameter mismatch between hardware and controller due to the variable-frequency and ambient operating conditions in a more electric powertrain. Accordingly, a Lyapunov-based adaptive parameter estimation algorithm that provides accurate fast-tracking of the system's parameters is developed and experimentally validated in this dissertation. The estimated parameters are fed to the step-ahead controller, minimizing the detrimental effects of parameter mismatch. However, despite its sound performance and enhanced capabilities, the step-ahead control scheme shows a reactive power steady-state tracking error proportional to the input frequency. Consequently, a direct model reference adaptive control (MRAC) scheme is developed and experimentally verified in this dissertation. The novelty of this control scheme resides in its ability to seamlessly regulate the output dc-bus voltage and input reactive power in the presence of rapid input-frequency/voltage variations, load changes, and system parameter variations. These smooth dynamics are achieved through a multi-variable direct MRAC formulation, which allows the controller's gains to automatically adjust to external perturbations without any knowledge of the system's parameters. The validity and performance effectiveness of all the developed control schemes is verified experimentally through a proof-of-concept laboratory-scaled three-phase 1.5kW 270V SiC-based two-level voltage-source-converter (2L-VSC) using a variable-frequency programmable grid emulator. Furthermore, a traditional proportional-integral (PI) controller is implemented experimentally with the same 2L-VSC as a benchmark to highlight the merits of the developed controllers. Lastly, to further validate the concepts introduced in this dissertation, all the developed controllers are experimentally tested under variable-frequency/variable-amplitude input voltage, which emulates a more electric powertrain variable-speed permanent magnet generator that is directly connected to its primer mover, e.g., internal combustion engine or jet engine. Herein, the superiority of the developed MRAC scheme is once again experimentally verified in comparison to the step-ahead and PI-based controllers, thus solidifying the developed smart converter concept for more electric powertrains.