Detección de patrones para identificación de variaciones de voltaje en redes de distribución implementando redes neuronales basadas en la transformada de Hilbert-Huang

Abstract

The article presented below aims to detect patterns that occur in a distribution system, for which it incorporates an algorithm based on neural networks based on the Hilbert Huang transform. It seeks to detect patterns that alter the system, obtaining the detection of the system of 33 IEEE bar. The training of the neural network was executed in Matlab software, which allows the algorithm to be able to provide detections in the event of future variations that may occur in a system. The input data of the neural network is obtained by implementing the Hilbert Huang transform, using the bus voltages, each node present in the system was modified to simulate the sags, interruption and swell disturbances provided by the IEEE 1159.The output data generated from the neural network corresponds to the classification of four categories, by implementing label allows to effectively interpret the detection of the signal, when disturbances are present. The classification model achieved a percentage of 94.09% accuracy, which reflects the efficiency of detection of signals that are associated with disturbances.