Chapter 4 - Machine learning techniques for power system application: fault analysis

Abstract

When a fault occurs in the power system, it is essential to identify the fault types and their locations for the protection and operation of the electrical system. A common format for transient data exchange for power systems (COMTRADE) file is used in this process to record the transient voltage and current information. By analyzing the COMTRADE file, protection engineers can locate the lines and determine the fault types. This study suggests power system fault location and line detection techniques based on convolutional neural networks (CNN). In addition, a coloring method has been applied to increase classification accuracy. The proposed CNN, trained on a large number of transient signals under various fault conditions, converts the three-phase fault information from the COMTRADE file into an image file and extracts it adaptively. The test results demonstrate that the proposed CNN-based analyzer can categorize the fault types and locations under different circumstances.