Theses Doctoral

Model Predictive Critical Soft-Switching Enabling High-Performance Software-Defined Power Electronics: Converter Configuration, Efficiency, and Redundancy

Zhou, Liwei

Advanced power electronic techniques are crucial to enable high-performance energy conversion systems for the applications of various load and source interfaces, e.g., electric vehicle battery charger, solar power, wind power, motor traction, grid-connection. Also, the improvements on electrification for energy conversion contributes to the Carbon Neutrality with the reduction of fuel combustion. The control and design of the power conversion systems largely determine the efficiency, power density and system cost which typically need specialized design procedures. Since the types of interfaced energy sources may vary, the corresponding control algorithms and hardware configurations will be different. Thus, the power electronics system design is conventionally a specific routine based on the desired source and load requirements.

Generally speaking, two main perspectives need to be considered when designing a power conversion system: (1) the power converter circuitry topology with the corresponding hardware components, e.g., low/high power circuits design, passive components design; (2) control algorithms and functions design, e.g., voltage/current control techniques, active/reactive power balancing and adjustment. However, the repetitive and specific power electronics design procedures for different load/source requirements are time-consuming and costly.

This thesis proposes a software-defined power electronics concept to develop a generalized auto-converter module (ACM) by leveraging variable-frequency critical-soft-switching, model predictive control techniques and high-performance litz-PCB inductors. The software-defined power electronics techniques can be applied to various types of electrified load/source applications without the need of repetitive hardware components and software algorithms designing procedures. The fundamental unit for the generalized concept, auto-converter module, is a type of MPC-based power module. A hierarchical control architecture is designed to manage the local ACMs and satisfy different load/source energy conversion requirements with high efficiency, high power-density and high-reconfigurability.

To achieve high-performance for the software-defined power electronics system, several advanced technologies are developed and integrated including variable-frequency critical-soft-switching, modular model predictive control, litz-PCB inductor design. Firstly, a variable-frequency critical-soft-switching technique is developed to adjust the switching frequency for the zero-voltage soft-switching. Doing so, the switching losses can be largely reduced with high efficiency. Secondly, the critical-soft-switching inductor is designed based on litz-PCB winding structure and neural network model to optimize the inductor losses and reduce the volume for the application of high frequency and large current ripple. Thirdly, a modular model predictive control method is designed for each of the local ACM to improve the dynamic performance and attenuate the oscillation caused by the variable frequency operation.

Lastly, a hierarchical control architecture is developed to generalize the software-defined power electronics with multi-layer structure, central control layer, local module control layer and application layer. The hierarchical control architecture can be widely applied to different types of load/source interfaces, e.g., single/three-phase grid-connected inverters, motor traction inverter, battery charger, solar energy and so on. Leveraging the hierarchical control architecture and software-defined power electronics, the repetitive power converter hardware components and software algorithms design procedures can be simplified and standardized. Also, for different power converter applications, the efficiency and power density are both improved with better dynamic performance.


  • thumnail for Zhou_columbia_0054D_17283.pdf Zhou_columbia_0054D_17283.pdf application/pdf 11.2 MB Download File

More About This Work

Academic Units
Electrical Engineering
Thesis Advisors
Preindl, Matthias
Ph.D., Columbia University
Published Here
June 22, 2022