Fault Detection in Smart Grids by Hybrid Generative Adversarial Networks with Neuro Fuzzy Algorithm

Inrecent years, integration of advanced communication networks and estimation techniques smart grid enables real time analysis and has control to optimize distribution of energy, improve grid resilience. Detecting and classification of faults in electricity networks is important for uninterrupted provision and maintenance of costs at minimum. By implementing data-driven approaches these systems brought many improvements like consumption of energy and quick re-establishment. In this paper, hybrid Generative adversarial networks (GAN) – neuro fuzzy algorithm is implemented for detection of faults in smart grids and done by extraction of features which are extracted from smart meters, training is done by GAN then trained model is further sent for classification which is done by neuro fuzzy algorithm if there is a fault then model sent for identification or location of fault and evaluation of accuracy is done, if there is no fault then again sent to feature extraction process. Proposed hybrid GAN-neuro fuzzy algorithm achieved 95.33 % of precision and 97.23 % of recall.