Control schemes of voltage source converters in more-electronics power systems

• Main Author: He, Jinsong
• Other Authors: Wen Changyun
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/160995
• Institution: Nanyang Technological University

Control schemes of voltage sources converter (VSC) are directly linked to the stability, reliability, multi-functionality, and implementation cost of the more-electronics power systems. Existing control methods are mainly linear approaches that merely ensure the local stability of the closed-loop system relating to specific operating points. Once exposed to large perturbations, such as renewable energy intermittency, sudden load change, and plant parameter variations, the system may display a bifurcation phenomenon, due to operating point migration, or instability. As a promising alternative, Lyapunov-based nonlinear control can ensure the large-signal stability of the VSC system. However, the existing Lyapunov-based approach for VSC application is inherently a single-loop control scheme. When applied to the UPS system, it leads to steady-state errors and sluggish dynamic response; If applied to LCL-filtered VSC, it fails to damp the resonance. A recently modified dual-loop approach solves the above two problems via artificial importation of voltage feedback, but the negative definiteness of the derivative of the Lyapunov function can be merely ensured under certain premises. Motivated by the above discussion, an adaptive dual-loop Lyapunov-based control strategy is proposed for a single-phase uninterruptible power supply (UPS) inverter. The control law is derived via an all-in-one Lyapunov function that considers both dual-loop requirement and global large-signal stability demand. Besides, the load disturbance is suppressed adaptively, and the load-current sensors get saved. Implementation of the proposed control does not need to calculate the derivatives of any signals. The load voltage is regulated with good steady-state and dynamic performance, with great robustness against the parametric variations. Both simulation and experiments verify the effectiveness of the proposed control scheme. The single-phase control scheme can be also extended for a three-phase application. Similarly, the control scheme inherently has dual control loops without any artificial modification. Stability analysis rigorously proves that the global large-signal stability of the system can be guaranteed unconditionally, which is valid both for linear load and nonlinear load. The proposed approach inherently has integrators, which can theoretically achieve zero steady-state-error regulation according to the internal model principle. The load disturbance is compensated using adaptive laws, saving three load-current sensors. The closed-loop system is d-q decoupled. Three controller gains are tuned via explicit formulas based on pole-placement strategy. Nevertheless, the above two control strategies are all based on an all-in-one Lyapunov function. There is no canonical form of such kind of Lyapunov functions. As a better alternative, the backstepping approach provides a systematic way to augment the Lyapunov functions step by step recursively. A backstepping-based control strategy is devised for point-of-load (POL) inverters in a shipboard power system. Firstly, the backstepping algorithm is harnessed to derive the pseudo-inductor-current-loop reference and decoupled switching functions in the d-q frame, which can rigorously guarantee the global large-signal stability of the system. Then, a state estimator is designed to estimate the load current and feedforward it to the backstepping controller for load disturbance rejection, saving three current sensors. The controller gains are selected to achieve the optimal system damping and maximized dynamic response, which can be intuitively interpreted via an ellipse-based strategy from a geometrical point of view. Stability proof and robustness analysis of the system is also provided. Although Lyapunov-based control displays good performance in terms of large-signal stability, it is hard to realize multiple control objectives. However, multiple control objectives typically require to be satisfied simultaneously in control of VSCs. Tightly regulated downstream load inverters serve as a typical type of constant power load (CPL), which risks triggering dc-link voltage oscillation when cascaded with LC input filters. In this scenario, both dc-link voltage stabilization and load voltage tracking need to be achieved at the same time. To this end, a composite bisection predictive control (CB-PC) is devised. Firstly, an improved general method is proposed to stabilize the dc-link voltage oscillation based on instantaneous power theory, which is valid for finite control set (FCS) MPC with/without a modulator. Secondly, deadbeat control is utilized to enhance the transient response of dc-link stabilization. Load voltage control is realized indirectly by tracking an offline-derived inductor-current vector. A droop-akin strategy is devised to strike a trade-off between the two control objectives. Then, a modified bisection algorithm is presented with a tunable iteration number (n) to improve steady-state control performance. It enables the exploitation of surplus computing resources to achieve lower total harmonic distortion (THD) of load voltage. Nonetheless, the selection of weighting factors (WFs) is a common obstacle for the finite control set model predictive control (FCS MPC) of power converters. To this end, a generic approach is proposed to update the WFs via reinforcement learning (RL). The WFs’ selection is self-taught online with full consideration of user-defined requirements. The trained policy is deployed to update the WFs in real-time. The self-taught process can be reactivated anytime in case of parametric variations or load changes. This idea is verified on FCS MPC-regulated stand-alone inverters cascaded with LC input filters. Simulation results demonstrate that RL significantly improves the load-voltage tracking accuracy without sacrificing dc-link voltage stabilization. Finally, Chapter 7 concludes this thesis and summarizes three recommended research directions, followed by the list of the author’s publications and the bibliography of this thesis.