Using adaptive ant colony algorithm optimized BP neural network to identify the DGA fault

Author

Cai-Tian Zhang ; An-xin Zhao

Author_Institution

Dept. of Electr. Eng., He´nan Econ. & Trade Vocational Coll., Zhengzhou, China

fYear

2013

fDate

22-25 Oct. 2013

Firstpage

1

Lastpage

4

Abstract

The BP network uses the rule of local decline to slow convergence speed and easy to fall into local optimum. With the increase of BP input parameters lead to the dimension disaster. Taking into consideration the above problems, Using adaptive ant colony optimization algorithm optimized the training process of BP network referred to as AACOABP. The BP network and the BP network optimized by adaptive ant colony algorithm to (AACOABP) were used to identify the faulty type of the collecting samples about DGA fault test data, AACOABP can make the fault identification accuracy is greatly increased, the recognition accuracy from 50% ~ 75% to 87.5% ~ 100%.

Keywords

ant colony optimisation; backpropagation; neural nets; power engineering computing; power system faults; power system reliability; AACOABP algorithm; BP input parameters; DGA fault identification; adaptive ant colony algorithm; backpropagation; convergence speed; dissolved gas-in-oil analysis; optimized BP neural network; Accuracy; Adaptive systems; Algorithm design and analysis; Ant colony optimization; Biological neural networks; Fault diagnosis; Training; BP neural network; adaptive ant colony algorithm; dissolved gas-in-oil analysis (DGA); fault identification;

fLanguage

English

Publisher

ieee

Conference_Titel

TENCON 2013 - 2013 IEEE Region 10 Conference (31194)

Conference_Location

Xi´an

ISSN

2159-3442

Print_ISBN

978-1-4799-2825-5

Type

conf

DOI

10.1109/TENCON.2013.6719070

Filename

6719070