Abstract:
The growing requirement of load supply is countered by the introduction of microgrid in an Electrical Power System. Introduction of microgrid to the electrical utility has given many advantages but along with it, has come a challenge to be dealt with. One such challenge is the islanding detection in a microgrid. Islanding, when being intentional, possess no threat to the system nor does it is a threat to personnel working in the utility. But when islanding occurs unintentional, it may be a threat to both utility equipment and personnel. That is why islanding detection is an important issue to be considered. IEEE Standard Series 1547 specifies essential conditions to be fulfilled when unintentional islanding occurs in a system.
In this thesis, an intelligent based islanding detection technique is implemented in a sample microgrid model to visualize the accurateness of the concerned technique. The implemented method is based on detecting islanding using SVM. However, due to a huge set of data, complexity of implementing SVM increases. Thus PCA based feature extraction is done to extract new features using which, SVM is trained. The SVM Model is trained with data of reduced dimensionality, and is tested against other set of data. The classifier model is seen have a very high accuracy.
