Research on Intelligent Diagnosis Method of Oil Temperature Defect in Distribution Transformer Based on Machine Learning

Thumbnail Image

Files

CIRED 2019 - 676.pdf (273.67 KB)

Paper number

676

Working Group Number

Conference name

CIRED 2019

Conference date

3-6 June 2019

Conference location

Madrid, Spain

Peer-reviewed

Yes

Short title

Convener

Authors

Xiao, Fei, State Grid ShangHai Municipal Electric Power Company, China
Yang, Guo-jian, State Grid ShangHai Municipal Electric Power Company, China
Hu, Wei, Tellhow Software Co. Ltd., China

Abstract

Long-time operation under abnormal oil temperature is one of the critical factors causing short service life and limit capacity of distribution transformers, and even tripping accident due to insulator failure. At present, only a simple static alarm function is available to detect the over-limit of oil temperature in distribution transformer. This function cannot disclose the defect evolvement. A prior alarm function to early identify abnormal oil temperature is needed in order to stop the development of defects. This paper introduces a novel intelligent prediction method based on machine learning methods for real-time abnormal oil temperature detection. This method establishes a dynamic association learning model using decision forests algorithm to predict oil temperatures based on transformer parameter, power load, as well as weather condition under the normal operation state. Comparing the predicted oil temperature with online measurement, we can find the abnormal oil temperature state and develop prior alarm function to detect the defect of distribution transformer. The decision making procedure based on this method has been applied into distribution transformers of Shanghai. The results validate the accuracy of the method and show efficiency when applying on the maintenance planning of distribution transformer.

Table of content

Keywords

Publisher

AIM

Date

2019-06-03

Published in

CIRED 2019 Conference

Permanent link to this record

https://cired-repository.org/handle/20.500.12455/110
http://dx.doi.org/10.34890/226

ISSN

2032-9644

ISBN

978-2-9602415-0-1