Hybrid modelling and control for switched-mode power converters

Abstract

Switched-mode power converters are some of the most widely used power electronics circuits due to their advantages of high conversion efficiency, flexible output voltage, light weight. A variety of control methods have been developed for the switched-mode power converters. However, in many practical situation, additional constraints need to be considered, e.g., safety measurement, current limiting or soft-starting, gross changes of operation point with guaranteed system stability, which has not been fully addressed in the available research works. On the other hand, the majority of the control design for power converters are based on the state-space averaged approach which involves considerable approximation in analysis and synthesis. Hence, advanced control techniques are in demand, which should be more constraints friendly and based on more precise models.

In this thesis, much attention has been spent on designing controllers for both DC-DC converters and DC-AC inverters based on hybrid modelling and Lyapunov stability theory. Due to the existence of the power switches, switched-mode power converters are hybrid systems with both continuous dynamics and discrete transition events. Instead of linearizing the converter model around a specific operating point, hybrid modelling captures both dynamics, which results in more accurate models.

Firstly, a novel sampled-data control approach is proposed for DC-DC converters. DC-DC converters are modeled as sampled-data switched affine systems according to the status of the power switch. In order to avoid the delay of the switching signal, an on-line prediction method is adopted to estimate the system state at the next switching instant. Based on the switched affine model and the predicted system state, a novel switching control algorithm is synthesized by using the switched Lyapunov theory. The proposed approach is able to not only drive the output to a prescribed set point from any initial condition, but also track a varying reference signal, and the switching frequency can be adjusted online with guaranteed stability. In addition, with this approach, Continuous Conduction Mode (CCM) and Discontinuous Conduction Mode (DCM) operations can be treated in a unified way. Experimental verification has been carried out to test the effectiveness and merits of the proposed method.

Furthermore, to compensate the information loss due to limited access to the state, a multiple sampling scheme is employed to derive a discrete-time switched affine model with an augmented measurement output for DC-DC converters. Based on the model, an output-feedback switching control law, which drives the system state to a set of attainable switched equilibria, is synthesized by using a quadratic state-space partition. The multiple sampling scheme not only facilitates the controller synthesis, but also improves the energy efficiency of the converter by allowing a lower switching frequency.

In addition, hybrid modelling techniques have been extended to more complicated cases – DC-AC inverters as the increasing number of power switches and the time-variant nature of the references. A current controller based on the hybrid model of the three-phase two-level inverter has been developed, which can drive the inverter currents tracking the desired power references in realtime and keep a unity power factor at the same time. This method has been extended to three-phase NPC inverters later on. However, in order to solve the neutral point balancing issue, a capacitor voltages prediction algorithm, modified from model predictive control, has been adopted. It should also be mentioned that a novel hybrid model for a grid-connected single-phase NPC inverter also has been presented, which models not only the dynamic of the inverter but also the dynamic of the current reference. An experimental test platform including a three-phase NPC inverter and a FPGA control board has been designed to demonstrate the implementation of the proposed control scheme in practice.

More information

Identifiers

ORCID for Zhan Shu: orcid.org/0000-0002-5933-254X

Catalogue record

Export record