الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 65 |  |  |
| | | from | 65 | | |
Abstract
Due to the fossil fuel depletion problem in recent years, employing clean, renewable energy resources has become a necessity. The idea of microgrids is presented to smooth the way for implementing renewable energy sources in power systems. Microgrids can be divided into two types: AC and DC. Privileges of DC microgrids have arisen throughout the past few years, as most of the distributed generators and energy storage systems have output in DC voltage. Also, DC systems are getting a standard solution for many residential and industrial usage types. Moreover, the control of a DC microgrid is simpler and more flexible than that of an AC microgrid. On the contrary, DC microgrids suffer from errors in the load sharing between different DGs as well as a voltage deviation at the point of common coupling.
Many studies have been produced to investigate and examine all DC microgrid topics concerning DC microgrid control. One of the most widely used methods for controlling a DC microgrid is droop resistance. Based on local information, it makes decisions.
In this thesis, the optimal droop resistances are obtained using the Local Unimodal Sampling (LUS) optimization algorithm at a specific load condition to guarantee the best current sharing among different DGs according to their ratings. A case study is implemented to show the superiority of using the optimal values rather than the conventional droop resistances. The results show that the current sharing error between different DGs becomes almost zero when using the optimal droop resistances. Furthermore, the performance of DC microgrids is tested using the optimal droop resistances at different load conditions, which contain heavy and light load conditions.
The results show the superiority of using the obtained optimal droop resistances to solve the current sharing issue under heavy and light load conditions. Another case study is conducted using a fractional order PI (FOPI) controller with optimal droop resistances. The parameters of FOPI control are tuned using two optimization techniques: the local uniform sampling (LUS) optimization algorithm and the driving training-based optimization (DTBO) algorithm. Moreover, the performance of the system under study with FOPI, whose parameters were gained from driving training-based optimization (DTBO), is compared with the same system based on the conventional PI that represents conventional methods in the literature. Results show that the FOPI had a lower maximum overshoot and a shorter settling time than the conventional system with PI. Also, FOPI presents an outstanding performance and reaches the exact nominal voltage of the grid faster.