Real-Time Device-level Modeling of Power Converters for Advanced Transportation Application

• Author / Creator

  Liang, Tian

• The advance of the transportation system has helped human beings to change their standard of living dramatically through the facilitation of trade, travel and exchange in both time and energy-efficient way. The modern transportation system consists of a complex multi-domain system: electrical, mechanical, hydraulic, thermal, multi-body dynamics, etc. Among these systems, the electrical system is considered as the most critical one, that can provide conditioned and reliable power to essential loads for the other physical systems. Thus, emulating the detailed electrical system accurately in real-time can help engineers and scientists to predict and control the whole transportation system dynamics efficiently, which can be extended to the application of repeatedly evaluating the newly designed power switch, protection circuit and algorithm, power converter and controller for design optimization and tuning in a non-destructive emulation environment. There is a demanding need for accurate detailed device-level modeling in real-time hardware-in-the-loop (HIL) emulation, especially under extreme and extensive range of conditions for advanced transportation application. These electrical system devices come in many shapes and sizes and often have to connect and correlate with many other devices inside the electrified transportation system. Discrete-time solution of a complete topology of the equivalent represented nonlinear circuit model can provide the detailed device dynamics with iteration calculation. However, these developed analytical and numerical models lead to the heavy computational burden of the real-time capable hardware, which introduces extra execution time-delay and makes real-time problem-solving infeasible.
  Introducing model order reduction methods and implementing these methods in dedicated real-time capable hardware are common techniques for reducing the computational complexity of mathematical models and shortening the model execution delay time to the nanosecond-level timing margin. Based on the common nonlinear system identification method, block-structured behavioral models, such as Hammerstein, Wiener-Hammerstein configuration, are introduced as the first model order reduction methods for the simplicity of their real-time implementation in the following three chapters (3, 4, 5). Then, hybrid kNN-RNN neural networks are introduced as the second model order reduction method for their real-time adaptive features of deviated parasitic parameters in Chapter 6. These methods require both highly sequential clocking and massive parallel structure in problem-solving. By utilizing state-of-the-art hardware platforms of multi-processing system-on-chip (MPSoC) or field programmable gate array (FPGA), both system-level and device-level transients were captured in real-time, which is also the first time of deploying the block-structured and neural network models on state-of-the-art System-on-Chip (SoC) and FPGA hardware platform for nanosecond-level real-time device-level power electronic converter emulation.
  Practical advanced transportation applications have also been investigated in the thesis, including three-phase self-balancing traction power system, massive existing Beijing-Shanghai AC high speed rail (HSR) traction system, future medium voltage DC traction network, and electromagnetic rail gun (EMRG) in all electrical ship (AES) system. With the anticipated increasing demand placed on the land, sea, and air, these above mentioned novel transportation model can be installed, constructed, and testified repeatedly by the proposed real-time device-level modeling scheme.

• Subjects / Keywords

  • Real-Time
  • Transportation
  • Device-Level

• Graduation date

  Spring 2020

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/r3-850v-0241

• License

  Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.