We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Bayesian Optimized CNN Model for Fault Classification in a Distribution System.
- Authors
Tiwari, Garima; Saini, Sanju
- Abstract
A fault in a power system is an anomalous state that must be recognized as soon as feasible. To minimize the repercussions of the fault, such as damages occurred to the device, loss of tangible assets and loss of human resources, it is critical to notice the problem promptly. In a power distribution system, there are several approaches for detecting different types of faults. In this paper, a neoteric approach using Bayesian optimized Convolutional Neural Network is used to detect and classify different symmetrical as well as unsymmetrical faults in power distribution systems. The effectiveness of the proposed CNN model is validated for an IEEE 13 bus radial distribution system grid modeled (and simulated) in PSCAD. Time series of the measured 3-phase fault currents (for eleven different categories of faults) are used to create training & testing data. This data has been imported in MATLAB software to develop a CNN classifier (whose hyper-parameters are optimized by using a Bayesian optimizer) for faults in power distribution system, under distinct fault situations by varying fault resistance, faulty node and fault inception angles. Findings of simulation clearly indicate that proposed model has very high categorizing accuracy and is superior and competitive to other techniques available in literature.
- Subjects
CONVOLUTIONAL neural networks; FAULT currents; GRIDS (Cartography); CLASSIFICATION; HUMAN capital
- Publication
Journal of Engineering Science & Technology Review, 2024, Vol 17, Issue 2, p35
- ISSN
1791-2377
- Publication type
Article
- DOI
10.25103/jestr.172.05