Resumen de Advanced control strategies applied to power converters
Active-front-end power converters based on controllable power switches are indispensable key devices in modern power systems. They serve as intelligent interfaces between utility grid, renewable resources (including wind turbine systems and photo-voltaic systems), energy storage systems, motor drives, and micro-grids. Furthermore, they also can serve as active power filters, which cancel out the current harmonics in the utility grid and consequently provide high-quality and green power to customers. These converters have desirable advantages such as bidirectional power flow, high-quality grid currents, adjustable power index and controllable dc-link voltage. The main control tasks of the power converters are to regulate the dc-link voltage to desired value and to provide active /reactive power to grid/load. The two major factors a_ecting the control performance of power converter are: (1) load disturbance at the dc-link; (2) parametric uncertainties of the converter system. On one hand, the unknown time-varying load disturbance connected to the dc-link can cause the fluctuation of dc-link voltage. The load value is unpredictable and can vary in a broad range. The magnitude of the load and the robustness of the controller will determine whether the dc-link voltage can be restored to the reference value. To reduce the load’s negative influence to the system, one e_ective way is to employ disturbance observer, which estimates the load value and feeds the load information to the system controller. On the other hand, the system parametric uncertainties cause the di_erence between the system model and the real plant, therefore the controller designed based on the model can not achieve desirable performance. To solve these two problems, this thesis investigate the advanced control strategies applied to grid-connected power converters. On one hand, to reduce the influence caused by the dc-link load disturbance, the disturbance-observer-based control strategies have been proposed, i.e., the disturbance observer estimates the value of external disturbance in real-time and feed the estimated value to the controller. On the other hand, to improve the robustness and adaptiveness to the system uncertainties, advanced control methods have been employed, including sliding mode control, adaptive control, fuzzy control and H∞ robust control. The main research work conducted in this thesis includes: Firstly, to compare the rejecting performance of di_erent types of disturbance observers against the dc-link load disturbance, four types of disturbance observer are designed for two-level power converter in Chapter 2, which are linear observer, sliding mode observer, linear extended state observer and nonlinear extended state observer. The results reveal that the sliding mode observer, linear observer and linear extended state observer achieve good disturbance estimation.
Next, based on Chapter 2, an improved linear disturbance observer is designed, which has two parameters to adjust its performance, one is to improve the transient response meanwhile not influence the steady-state performance, the other is to maintain the steady-state performance. This observer and PI controller make up the voltage regulation loop, making it a pure linear system, which can be analyzed with classical control theories and is convenient for practical use. To verify the e_ectiveness of this strategy, series of experiments have been conducted on a real-application grid-connected 5kW power converter.
To further improve the performance of voltage regulation loop, an improved sliding mode observer is designed in Chapter 4. Comparing with the previous improved linear disturbance observer, this nonlinear sliding mode observer obtains faster convergence and stronger robustness against system uncertainty. In the meantime, super-twisting sliding mode controller is adopted in the voltage loop, which enhance the converging speed and robustness. The convergence condition of dc-link voltage is obtained via Lyapunov method, and the advantage of this strategy is verified via simulation and experiments on real-application grid-connected power converter.
To reduce the impact of parametric uncertainty, in Chapter 5, adaptive law is added to the super-twisting sliding mode controller, which adjust the control parameters along with system parameter variation. With this adaptive law, the upper bound derivative of external disturbance need not to be known a priori. Moreover, H∞ control is adopted in the voltage loop to attenuate the influence from disturbance estimation error to the controlled dc-link voltage. The convergence conditions of dc-link voltage and grid current are obtained via Lyapunov method.
Last but not the least, to expand the control target and control mode, three-level neutral-point-clamped power converter is investigated under the voltage oriented control and direct power control, respectively. Firstly, the performance improvement of using linear extended state observer is studied under voltage oriented control mode. Then the direct power control strategy is designed with linear extended state observer, H∞ control, super-twisting sliding mode control and adaptive control. The convergence conditions of dc-link voltage, instantaneous active/reactive power and capacitor voltage di_erence are obtained. Due to the resonant adaptive law, the third harmonic disturbance in capacitor voltage balance loop is eliminated.
