الفهرس | Only 14 pages are availabe for public view |
Abstract Integrating distributed generator into the existing distribution network is predicted to play an important role in the near future. Distributed generators, specifically renewable energy technologies, such as wind turbine, photovoltaic and fuel cells are entering a stage of fast expansion. A Microgrid is a group of energy sources located in the same local area that is in turn connected into the national grid while also being able to disconnect from it and operate independently, for example in the event of an electricity outage. Microgrids usually consist of distributed generation sources, particularly renewable energy generators such as wind turbines and solar panels, usually accompanied by some form of energy storage device, invariably a battery or bank of batteries. Connecting distributed generator to the distribution network has many benefits such as improving local energy delivery, increasing reliability, saving money, generating revenue, aiding economic growth, making the grid more resilient, helping to counter climate change and enhancing the power quality. However, it gives rise to many problems. Hence, integrating distributed generators into the existing distribution network is not problem free. Unintentional islanding is one of the encountered problems. Islanding is the situation where the distribution system containing both distributed generators and loads is separated from the main grid as a result of many reasons such as electrical faults and their subsequent switching incidents, equipment failure, or pre-planned switching events like maintenance. In this thesis passive based microgrid islanding detection techniques have been developed. The proposed methods have two stages the first stage is a feature extraction, in this stage Fourier Transform analysis is applied to extract the second harmonic from voltage and current wave forms and then calculate the symmetrical components which used as a features. The second stage is a classification, using three artificial intelligent techniques (Logistic Regression - K-Nearest Neighbor – Ensemble) to detect microgrid islanding. Features are then fed to a trained pattern recognition model which is if well trained capable of differentiating between islanding event and any other transient events such as switching or temporary fault. The trained classifier was then tested using novel waveforms. The three methods were applied to simulation data as well as experimental data and results indicated that this approach can detect islanding events with high degree of accuracy. KEYWORDS Microgrid – Islanding Detection – Logistic Regression – K-Nearest Neighbor – Ensemble Classifier – Distributed Generation – Distribution Systems. |