Abstarct: To provide the required power in different situations of load, it is necessary to control energy sources and combine with each other in Microgrids. In this thesis, the main focus is the control voltage in Autonomous Microgrid to provide the required power of loads that separated from maingrid. In this thesis, a system consists of two parallel inverter is designed and simulated. Generally, in grid-connected mode, voltage and frequency of the microgrid are controlled by the maingrid and microgrid DGs injects predefined powers.Microgrids can operate independently of the maingrid. In this operation mode, DGs should be able to control voltage and frequency of the microgrid and supply the loads. Controlling microgrid DGs can be accomplished in two methods of centralized and decentralized. In centralized method, central controller unit and communication lixnk between DGs are needed. But for controlling DGs power and coordination between them in decentralized method, no communication is needed. One of the more common decentralized control methods is droop control. In this method only output measurement quantities are used for controlling DGs output powers and no communication is used for coordination them. In this thesis, first discusses the Droop baxsed control method, and then for improved control voltage, optimize the coefficients of voltage controller using genetic algorithm.