An Innovative Deep Learning Approach Applied to Transient Stability Assessment of Power Systems

Abstract

Secure and reliable operation of a power system is exceedingly important for efficiency in power systems and everyday life. Transient stability is a large obstacle that, if assessed properly, can help maintain this secure and reliable operation. Transient stability studies have created a big data issue and recently data mining and machine learning techniques have been broadly applied to transient stability assessment (TSA). During the operation of a power system disturbances such as, cut-off loads, short-circuit faults, etc., may occur and it is crucial to be able to quickly and accurately determine if the power system is stable after these disturbances. With the growing consumer demand for reliable power, many methods have been proposed for fast and accurate transient stability assessment, but these traditional methods such as, extended area method, direct method, and time domain simulation do not provide the most optimal solutions. The goal of TSA is to fulfill the needs of speed and capacity, calculation accuracy, and easy online calculation. In this paper, an innovative deep learning method is applied for TSA. The approach presented in this paper is tested on a 19-bus system. Simulation results illustrate the effectiveness and practicability of the proposed strategy.