Detection classification and location identification of short circuit faults in power system network

Abstract

A microgrid is a network consisting of one or several loads and distributed generation (DG)sources that operate as a single aggregate load or source. Microgrids help to manage the DGsources more efficiently. The penetration of distributed renewable energy sources degrades theprotection of microgrids, which leads to incorrect data flow in the energy systems. It is critical todetect faults , types of defects and location of faults in order to improve the protection and securityof microgrids. To cater this issue in hybrid renewable energy system, a novel fault detectionscheme is adopted using artificial intelligence. The deep learning techniques are applied foridentification, classification and location identification of short circuit faults. Advanced featureextraction methods like empirical mode decomposition are carried out for fault detection intransmission lines and found to be more efficient compared to DWT. Fault detection andclassification are carried out in an IEEE standard 30 bus network using Wavelet Transform (WT)and Bees Optimization Algorithm (BOA) based Improved Convolution Neural Network (ICNN).Matlab/Simulink is used to model a simple microgrid which is divided into four zones. Faults aresimulated to generate training data for the neural network model. To improve the accuracy of faultdetection, the features are extracted from the time series data using empirical wavelet transform(EWT). First, EWT evaluates the frequency components in the signal, then calculates the boundsand gets the basis of the oscillating components. The obtained samples are classified using a HybridConvolutional Recurrent Neural Network (HCRNN) and optimized by the Pelican OptimizationAlgorithm. Eleven types of faults are identified along with the location of faults using the proposedsystem. The results are compared with the existingmethods and found that the proposed methodhas improved the fault sample detection accuracy by 1.56%. The proposed approach significantlyreduce the response time to faults, thereby improving the reliability and efficiency of powersystems.