Aplicación de la transformada de Stockwell para detección y clasificación de fallas en líneas de transmisión

Abstract

In this research, a method to detect different types of faults in the transmission system is proposed. For this purpose, a signal processing technique such as the Stockwell Transform (ST) and an artificial neural network (ANN) are used as an unsupervised learning technique. In this context, the ST is used for the frecuency-time analysis of the current, voltage signals resulting from the short-circuit study. These signals are used to create a database for training, validation and testing of the ANN, so that a predictor is created to identify and classify faults in real time. The performance of the proposed methodology is tested on IEEE 9-bar and 14-bar systems. Also, the obtained results are compared with the fault classification based on the wavelet transform. In this way, it was found that the TS is superior to the wavelet transform. In addition, the ANN presents an effectiveness of 98% with a 2% error in fault classification.