DocumentCode :
2622181
Title :
A method to obtain prior knowledge in the fault diagnosis based on probability model
Author :
Zeng, Yongguo ; Shen, Shanhong ; Wang, Dayong ; Gao, Zhipeng
Author_Institution :
State Key Lab. of Networking & Switching Technol., Beijing Univ. of Posts & Telecommun., Beijing, China
fYear :
2009
fDate :
16-18 Oct. 2009
Firstpage :
216
Lastpage :
220
Abstract :
Network has uncertain characteristic as its appearances, which made the reasoning based on probability model become a hotspot in faults diagnosis field. For uncertainty reasoning methods, how to obtain the prior knowledge is very important. In the past, the knowledge usually obtained through artificial experience or random hypothesis, which is very random and not reasonable. In order to obtain prior knowledge effectively, a method using Bayesian networks learning to learn knowledge dynamically from previous sample has been put forward, which made the diagnosis system be self-adaptive. Experiment turns out that Bayesian networks learning can obtain prior knowledge effectively and increase the diagnosis accuracy.
Keywords :
belief networks; fault diagnosis; telecommunication network reliability; Bayesian networks learning; fault diagnosis; incertainty reasoning; node failure; prior knowledge; probability model; Approximation algorithms; Bayesian methods; Communication networks; Fault diagnosis; Iterative algorithms; Knowledge management; Learning systems; Network topology; Probes; Uncertainty; Bayesian Networks Learning; fault diagnosis; prior knowledge; uncertainty reasoning;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Communications Technology and Applications, 2009. ICCTA '09. IEEE International Conference on
Conference_Location :
Beijing
Print_ISBN :
978-1-4244-4816-6
Electronic_ISBN :
978-1-4244-4817-3
Type :
conf
DOI :
10.1109/ICCOMTA.2009.5349206
Filename :
5349206
Link To Document :
بازگشت