Investigating the Dynamic Security of Power System to Detect System Stability or Instability by Using Neural Network

Abstract

Modern power systems are very complex due to constant variations of the load. These systems are constantly exposed to internal and external disturbances that can cause system instability. The process of determining the stability of a system under turbulence is called security assessment. In other words, the security assessment of the power system is performed to determine the stability or instability of the system. The security assessment of the power system is a combination of static and dynamic security analysis. One of the ways to determine the dynamic security is to find the critical time to fix the fault. This time is a combination of functions with many variables, so its acquisition is relatively difficult. In addition, finding and evaluating the critical time of fault correction requires detailed and timely computations. Therefore, data classification can be used as the best option for assessing the security of a power system. Data classification, sampled data and computational time reduces security assessment. In this paper, three methods are used for classifying data. These methods include: least squares (correlation), Kohonen neural network and wavelet transform. The use of these methods eliminates the problems and issues that traditional methods have. If the classification of data is correct with the methods mentioned for input patterns and the critical times to correct the existing fault, then these methods can be used to determine the critical lines of the new input patterns without performing detailed calculations of transient stability.

Keywords: Power System - Neural Network - Dynamic Security - Critical Fault Time

DOI: 10.7176/NCS/10-02

Publication date:July 31^st 2019