Abstract An effective method for evaluating the distribution system reliability using an artificial neural network (ANN) is proposed in this paper. The ANN is constructed according to the back-propagation learning rule which is an iterative gradient algorithm designed to adjust the interconnection weights among processing elements. Therefore, the developed ANN is used to predict the distribution system reliability using historical data. At the same time, the system average interruption frequency index (SAIFI) and the system average interruption duration index (SAIDI) of a real distribution system are computed and compared with results generated by the network method. It was found that the deviation of the results computed by the proposed approach is below 1% and the required running time on a SUN network environment is less than 2 s. In addition, for handling the distribution system configuration changes induced by overloading or faults, the ANN approach demonstrates an advantage over the network method.
[1]
Richard P. Lippmann,et al.
An introduction to computing with neural nets
,
1987
.
[2]
J.E.D. Northcote-Green,et al.
Development of distribution system reliability and risk analysis models. Final report
,
1981
.
[3]
Geoffrey E. Hinton,et al.
Learning representations by back-propagating errors
,
1986,
Nature.
[4]
Geoffrey E. Hinton,et al.
Learning internal representations by error propagation
,
1986
.
[5]
R. Billinton,et al.
An Efficient Algorithm for Deducing the Minimal Cuts and Reliability Indices of a General Network Configuration
,
1976,
IEEE Transactions on Reliability.
[6]
Paul A. Jensen,et al.
An Algorithm to Determine the Reliability of a Complex System
,
1969
.
[7]
David Bailey,et al.
How to develop neural-network applications
,
1990
.