Turkish Journal of Electrical Engineering and Computer Sciences




This study aims to perform fast fault diagnosis and intelligent protection in an active distribution network (ADN) with high renewable energy penetration. Several time-domain simulations are carried out in EMTP-RV to extract time-synchronized current and voltage data. The Stockwell transform (ST) was used in MATLAB/SIMULINK to preprocess these input datasets to train the adaptive fault diagnosis deep convolutional neural network (AFDDCNN) for fault location identification, fault type identification, and fault phase-detection for different penetration levels. Based on the AFDDCNN output, the intelligent protection scheme (IDOCPS) generates the signal for isolating a faulty section of the ADN. An intelligent fault diagnosis scheme that combines ST and deep learning methods aids the artificial intelligence-based protection scheme in isolating the faulty section. This study uses the PyTorch framework to build both the AFDDCNN and IDOCPS. The proposed protection technique classifies and isolates faults and coordinates protection with minimum operating time in the IEEE 13-bus ADN. It consistently gives high accuracy for fault diagnosis and minimum operating time for the IDOCPS even when the network’s topology is modified to the IEEE 34-bus ADN. The experimental results indicate that the proposed model is more accurate and provides faster fault diagnosis and isolation than state-of-the-art methods.


Distributed energy resources, Stockwell transform, deep convolutional neural network, phasor measurementunits, distributed generation, multilayer perceptron, multiobjective firefly algorithm

First Page


Last Page