Title :
Probabilistic training procedure for neural network based power system diagnostic software
Author :
Glinker, E.S. ; Mansour, S.Y.
Author_Institution :
Dept. of Electr. Eng., Lakehead Univ., Thunder Bay, Ont., Canada
Abstract :
This paper outlines a procedure which uses a power system description language to train a neural network to generate conditional probabilities for device failures. The proposed method minimizes the need for the use of explicit conditional statements
Keywords :
failure analysis; learning (artificial intelligence); neural nets; power distribution protection; power system analysis computing; probability; conditional probabilities; device failure probability; distribution system protective devices; neural network based power system diagnostic software; power system description language; probabilistic training procedure; Artificial neural networks; Bayesian methods; Neural networks; Power distribution; Power generation; Power system protection; Power systems; Probability; Software systems; Switches;
Conference_Titel :
Electrical and Computer Engineering, 1998. IEEE Canadian Conference on
Conference_Location :
Waterloo, Ont.
Print_ISBN :
0-7803-4314-X
DOI :
10.1109/CCECE.1998.685562
