Micro-grid Islanding Detection Model Based On Feature Extraction And Classification Algorithms

In recent years, micro-grid has became a hot research of power system. The micro-grid can be controlled flexiblly and intelligently. It may be a good way to solve the problem when distributed generator connects to the power system in a large scale. The micro-grid can fully use the generating capacity of DG, also improve power supply reliability. Using renewable energy, The micro-grid is friendly to the environment and non-polluting. It ia also a useful supplement of the high voltage, long-distance, large power grid supply model. However, to make the above benefits work, the islanding detectiong problem of the micro-grid must be solved nicely. The research of micro-grid islanding detection is really important.Today’s islanding detection methods mainly focuse on the situation of a single DG, including communication based islanding detection method, active and passive islanding detection method. The major disadvantage of the communication based islanding detection method is the high cost, with the development of smart grid, it has some potential for development. The active islanding detection method can achieve a small non-detecting zone and a high sensitivity, while this method will affect power quality. The passive detecting method can be implemented easily,has no effect on the power quality, but the non-detecting zone is larger.This paper introduces the concept of micro-grid development status over the world, studis the two operating modes of micro-grid:network operation and islanding operation, analyzes the advantages of planned islanding and the disadvantages of unplanned islanding.This paper points out the two principle of islanding: anti-islanding strategy and the use of islanding operation.This paper studies the principles and characteristics of islanding detection, proposes the main criteria and indicators of islanding detection, analysis the existing islanding detection methods and their advantages and disadvantages,proposes a new islanding detection method that based on feature extraction and classification algorithms. This paper studies the basic theory of ELM, establishes a micro-grid islanding detection model that based on feature extraction and classification algorithms, which can provides offline parameter optimization and online islanding detection, builds a micro-grid model with PSCAD software,builds a model of islanding detection with Matlab software.At last, This paper verifies the validity of the islanding detection method by simulation.