Non-centralized Control Method of DC Microgrids Based on Voltage Sensitivity Matrix

Abstract

Demand for renewable energy sources has increased due to global warming and environmental problems. As a result, electricity power generated from renewable energy sources is increasing every year. To use effectively and manage properly of generated electrical energy, a microgrid concept has been proposed. A microgrid is a power system in which loads and distributed generators are interconnected, and the system is either connected to the AC grid or operated independently depending on the situation. The microgrid can be divided into an AC microgrid and a DC microgrid according to the characteristics of the line voltage. In particular, in the case of a DC microgrid, since the power is transmitted through the DC line, there are no reactive power and frequency synchronization issues. The researches for control of DC microgrid have been focused to stabilize the DC bus voltage and supply a balanced power, and can be categorized as the centralized control method and the non-centralized control method. In the centralized control method, the energy management system monitors the input and output states of the distributed generators and loads, and delivers the commends by optimization algorithm. However, this requires a high-bandwidth communication system with a single master-multiple slave structure, and the reliability is impaired because there is a single point of failure in the communication system. On the other hand, the non-centralized control method manages DC microgrid with the autonomous control of the distributed generators without the energy management system, so the dependency of the communication system is relatively low and easily perform the plug-and-play. The non-centralized control method can be divided into the local control method and the distributed control method. The local control method autonomously performs power sharing by using the droop control. Although many papers using droop control have been published and researched, the value of droop gain has often been given as an arbitrary value, and there has not been much research on proper droop gain design method. Especially in the local control method, control performance is affected by the droop gains and the line resistances, so the droop control design method should be addressed and defined properly. To solve this problem, this paper proposes the droop control design method considering the entire line structure of the DC microgrid. To interpret the complicated line network configuration, the voltage sensitivity matrix is derived based on the power flow analysis. This voltage sensitivity matrix is used to analyze the effects of variables such as the line resistance, the droop gain, and the output power of the components on the power sharing and the bus voltage variation, which are the main performance of the local controller. The proposed local control method gives the detailed design procedure for the droop gain using the voltage sensitivity matrix, which improves the power sharing and voltage regulation performance of the bidirectional distributed generator. The second method of non-centralized control, the distributed control method, consists of the primary controller based on the droop control and the secondary controller using the low-bandwidth communication. In the conventional distributed control methods, an additional PI controller is applied to improve voltage regulation and power sharing performance. However, using an additional controller makes it difficult to design the appropriate PI gain. In addition, the performance of the secondary controller is affected by the communication speed, and in the worst case, the controller output can be oscillated or diverged. To solve this problem, the distributed control method using the voltage sensitive matrix is proposed. The proposed distributed control method is formulated based on the voltage sensitive matrix, which makes it easier to design the distributed controller than the conventional method. In addition, since the communication data used in the proposed method is relatively small and the additional PI controller is not applied, the degradation of control performance is relatively low and perform stable control even if the communication speed is slow. Stability analysis is performed based on the small signal analysis to analyze the influence of the control parameters and the communication delay on the stability of the proposed non-centralized controller. The proposed methods are applied to the 5-bus mesh type line network for the local control method and the 4-bus ring type line network for the distributed control method individually. Improved power sharing accuracy and voltage regulation performance are verified using the PSCAD simulations and the experimental system.