2016 24 th Iranian Conference on Electrical Engineering (ICEE) 978-1-4673-8789-7/16/$31.00 ©2016 IEEE Decentralized Voltage and Frequency Control in an Autonomous ac Microgrid using Gain Scheduling Tuning Approach Hossein Karimi IEEE Student Member Electrical and Computer Engineering Tarbiat Modares University Tehran, Iran Hoseinkarimi.ee@gmail.com Mohammad T.H. Beheshti Faculty member Electrical and Computer Engineering Tarbiat Modares University Tehran, Iran mbehesht@modares.ac.ir Amin Ramezani Faculty member Electrical and Computer Engineering Tarbiat Modares University Tehran, Iran ramezani@modares.ac.ir Abstract— The renewable energy resources provide us with green energy and flexible integration into power grids, but due to their variable nature, they cause significant frequency, voltage and power fluctuation for power system so that a sophisticated control strategy is needed to cope with these challenges. In this paper, the proposed control strategy consists of inner voltage and current controller and outer power loop. In this study, at first, coefficients of conventional PI controllers are obtained by a proposed objective function, and then it is shown that this PI controller with constant parameters doesn’t work properly over wide range of load variation, and the system is likely to be unstable. To obtain better performance, gain scheduling tuning approach is proposed to adjust PI coefficients for resistive, inductive and capacitive loads. This approach is applied to an autonomous three bus ac microgrid using Particle Swarm Optimization. Keywords; Decentralized Control, Inverter, Microgrid, power controller, Gain Scheduling. I. INTRODUCTION The advent of microgrids and Distributed Generations (DG) have improved power system reliability, environmental and economical issues. Nowadays, Renewable Energy Sources (RES) are in focus and many countries have been trying to employ them in their energy system since they are cheap and pollution free. DGs are mainly connected to grids through power electronic devices like Voltage Source Inverter (VSI). VSIs convert DC voltage to AC voltage through high frequency switching devices and their response is so fast in comparison to power systems dynamics [1]-[3]. In general, microgrids can operate in two different islanded and grid-connected mode. In stand-alone mode, micro-sources regulate frequency and voltage of grids besides providing power for loads. In grid connected mode, microgrids should just exchange power and its voltage and frequency is dictated by main power grid due to the relatively small size of migrogrid [4]-[5]. In spite of the numerous advantageous of this new concept, it causes important technical challenges that a proper control methodology should be designed to deal with its important problems. Voltage and frequency regulation, uncertainty in power generation, and DGs protections are some new challenges which RESs have brought with itself. To address these new challenges in microgrids, two centralized and decentralized control method are widely used. In centralized method, all information about loads and sources are gathered by a central unit, and this unit determines control strategy for entire microgrid [6]-[8]. In this strategy, all control actions are defined by a single unit, and communication link is needed to transfer data and control commands. However, its main disadvantages is its high cost [9]. Decentralized control method works base on local measurement and is cheaper and more reliable than central method [10]. In this method, system is regulated by local voltage, current and droop controller [11]-[12]. Droop controller is implemented in microgrids in order to mimic power system behavior in case of load changes that is decrease frequency as demand increases. Droop controller regulates active power by frequency and reactive power by voltage; therefore, P and Q can be adjusted independently so voltage and frequency are determined consequently [13]-[15]. There are some works in this regard. Reference [16] tries to joint thermal comfort optimization and demand response in microgrids equipped with energy storage unites and RESs. Reference [17] presents a stochastic model for day-ahead microgrid management. The model applies probabilistic reconfiguration and unit commitment simultaneously to achieve the optimal set points of the microgird’s units. Voltage and frequency in microgrids are very susceptible to load changes. On the other hand, power generated by DGs are not reliable so much; therefor, if any factor in regulating microgrid is ignored, it is likely to make the whole system unstable. To fulfill this goal, in this study, by considering main factors, an objective function is proposed and PI coefficients are obtained for a given load. Particle Swarm Optimization (PSO) is applied to find objective function minimum. At first, limits for PI coefficients is determined, and then PSO runs Matlab/Simulink each time to find optimal value. Also, it has been noticed that this obtained parameters for PI controller not only doesn’t regulate system properly for other loads but also sometime makes the system unstable, especially when severe load changes occurs in the grid; therefore, a simple gain scheduling technique is proposed to tune PI parameters in accordance to load changes.