Low Complexity Direct Power Control Methods for Active Front end Rectifiers and Applications

Abstract

In modern power systems, Low Voltage Direct Current (LVDC) distribution system is an emerging technology that enables increased penetration of renewable energy sources and provides scope for a number of applications like Electric Vehicle charging, DC lighting and illuminations, Electrical Drives, data centers, etc. As LVDC systems are located at the load end and operate at low power, electrical energy is taken from the AC grid during shortage and given to the AC grid during surplus. A bidirectional flow of AC to DC conversion is achieved by using a power electronic converter and advanced control algorithms. Direct Power Control (DPC) has evolved as one of the well-established control methods which was initially demonstrated for AC to DC conversion but in later years applied to DC to AC conversion as well. The literature describes many advanced and complex methods of DPC having high computational complexity. However, there are only a few works that are intended to reduce the computational complexity and focus on low-complexity switching vector selection methods. Realizing the importance of the lower computational complexity, this dissertation attempts to develop the low complexity DPC methods that can be implemented using low-cost microcontrollers. Classical DPC operates with 12 sectors and it is based on a predefined table. However, a table that changes according to the load requirements and system parameters is always preferred. It is also investigated that with the increase in the number of sectors, the resolution of sector detection will increase and provides a better response. Hence a dynamically adjustable switching algorithm method is developed which is based upon an 18-sectored division of the phase angle of the grid voltage vector. This dissertation also attempts to hybridize the highly accurate predictive group of the controllers which have high computational complexity with the simplest control of switching table-based DPC (ST-DPC). It is achieved by providing compensation for the unavoidable control delay present in the real-time environment by the delay compensated multifold switching table DPC (DCMST-DPC). The aim of such a controller is to substantially improve the performance without significant additional computation so that it can be implemented with a lower-cost microcontroller. The practical power system does not always have a balanced three-phase supply system. Hence adaptive controllers are required for the PWM rectifier so that the converter performance does not deteriorate even during grid unbalance. This dissertation proposes one of such controllers which is able to accurately estimate the instantaneous powers without grid voltage sensors even in an unbalanced three-phase system. A simpler algorithm deadbeat DPC (DB-DPC) has been chosen for the switching vector selection, which follows a more detailed system modeling yet has a lower computational requirement. Most of the articles in the literature are related to one application of the DPC technique, which is the PWM rectifier control. Apart from that, the other potential applications of DPC have been less explored. Hence, this dissertation presents a detailed study of new sights of the control technique in different fields of application. Additionally, this dissertation also focuses on DPC applications in the shunt active power filter (SAPF) and battery charging systems. In this work, a SAPF control method is developed which can handle different non-ideal characteristics of the grid and operate at a constant switching frequency. On the other hand, in the field of battery charging applications, this work proposes a single-stage converter, unlike the classical methods, which employ a two-stage power converter. The proposed battery charging system uses the dynamic dc link to abolish the dc-dc converter, which constitutes the second stage of the two-stage power converter in the existing methods. To evaluate the feasibility and performance of the developed techniques, the respective system models have been tested in MATLAB Simulink, RT-LAB, and a real-time environment with an extensive analysis of the obtained results.