DocumentCode :
3298874
Title :
Application of Genetic LVQ Neural Network in Credit Analysis of Power Customer
Author :
Wang, Jing-min ; Wen, Yu-qian
Author_Institution :
North China Electr. Power Univ., Baoding
Volume :
2
fYear :
2008
fDate :
18-20 Oct. 2008
Firstpage :
305
Lastpage :
309
Abstract :
Difficulties in collection of electric charge have affected the regular operation and development of power supply bureau seriously. So the credit problem of power customer has become one of the focus questions that power supply bureau pays attention to. In this paper, learning vector quantization (LVQ) neural network is used to establish the credit analysis model of power customer. And an improved genetic algorithm (NGA) is adopted to set initial reference vectors in competition layer of LVQ. Then, it solves two problems of LVQ neural network. That is, the neurons are not utilized adequately and LVQ network is sensitive to the initial data. A comparison study is reported based on LVQ with random initial reference vectors and LVQ with initial reference vectors set by NGA. Simulation results have shown that the proposed method enhances the accuracy and speed of credit classification. So, it is promising to credit analysis of power customer.
Keywords :
credit transactions; customer services; electricity supply industry; genetic algorithms; learning (artificial intelligence); neural nets; vector quantisation; vectors; credit analysis; credit classification; electric charge; genetic LVQ neural network; improved genetic algorithm; learning vector quantization; power customer; power supply bureau; random initial reference vectors; Artificial intelligence; Artificial neural networks; Computer applications; Computer networks; Genetic algorithms; Intelligent networks; Neural networks; Neurons; Power measurement; Power supplies; Credit analysis; Genetic algorithm; LVQ; Power customer;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Natural Computation, 2008. ICNC '08. Fourth International Conference on
Conference_Location :
Jinan
Print_ISBN :
978-0-7695-3304-9
Type :
conf
DOI :
10.1109/ICNC.2008.158
Filename :
4667006
Link To Document :
بازگشت