Journal article
Application of a general regression neural network for health index calculation of power transformers
International Journal of Electrical Power & Energy Systems, Vol.93, pp.308-315
12/2017
Abstract
A power transformer is one of the most important components in a transmission network. To assess the overall condition of this valuable asset, health index calculations are recently gaining more attention from the utility companies that operate networks. Only limited research has been conducted on health index calculations of transformers. Most of the past approaches are based on the linear combination of weighted scores of measurements following the industry standards such as IEEE, IEC and CIGRE. A few previous methods based on artificial intelligence and statistical approaches such as fuzzy logic, multivariate analysis and binary logistic regression have been published in recent years. In this paper, a General Regression Neural Network (GRNN) which has a nice nonlinear property and can work with measurements without quantization has been evaluated. The GRNN allows multi-dimensional measurements to be combined through an optimal weighting and scoring system to compute a quantitative health index of power transformers. The weighting of each test was assigned based on a smoothly interpolated continuous function. The efficacy of the model has been validated against expert classifications and data sets published in the literature. The comparative results demonstrate that, the proposed method is reliable and very effective for condition assessment of transformers through an automated health index calculation.
Details
- Title
- Application of a general regression neural network for health index calculation of power transformers
- Authors/Creators
- M.M. Islam (Author/Creator) - Murdoch UniversityG. Lee (Author/Creator) - Murdoch UniversityS.N. Hettiwatte (Author/Creator) - SRI
- Publication Details
- International Journal of Electrical Power & Energy Systems, Vol.93, pp.308-315
- Publisher
- Elsevier Limited
- Identifiers
- 991005542617007891
- Copyright
- © Elsevier Ltd
- Murdoch Affiliation
- School of Engineering and Information Technology
- Language
- English
- Resource Type
- Journal article
Metrics
128 Record Views
InCites Highlights
These are selected metrics from InCites Benchmarking & Analytics tool, related to this output
- Collaboration types
- International collaboration
- Citation topics
- 7 Engineering & Materials Science
- 7.251 Electrical - Harvesting & Discharging
- 7.251.1052 Partial Discharge
- Web Of Science research areas
- Engineering, Electrical & Electronic
- ESI research areas
- Engineering