Rapid Design Method for Generating Power System Stability Databases in SPS for Machine Learning

摘要

Today, the electrical networks are subjected to various internal or external disturbances that are likely to affect their stability. In view of their complexities and sizes, it is important to understand and rapidly predict the behavior of power grids in reaction to these disturbances during their operations. Big data presents itself as an alternative for the fast and efficient prediction of the stable or unstable state of electrical networks. In this work, we will be presenting a simple and fast technic for generating a large amount of data stability on Matlab/Simscape Power System (SPS) for deep learning. This is to facilitate the use of machine learning in the power system, which is necessary for the extraction and classification of stability criteria. By performing a simulation from the pseudo code of $i$ fault on each $j$ lines of the IEEE 68 bus, we obtain large matrices of $n$ time response signals per generator, for each of the $p$ generation-load patterns in our database.