Electric Power Client Credit Assessment Based on GA Optimized BP Neural Network
Judging from the characteristics of electrical products and power supply enterprises, this paper combines quantitative and qualitative analysis together, and establishes a electric power client credit assessment indicator system on the basis of analysis o
- PDF / 274,536 Bytes
- 7 Pages / 439.37 x 666.142 pts Page_size
- 89 Downloads / 184 Views
Electric Power Client Credit Assessment Based on GA Optimized BP Neural Network Xinli Wang
Abstract Judging from the characteristics of electrical products and power supply enterprises, this paper combines quantitative and qualitative analysis together, and establishes a electric power client credit assessment indicator system on the basis of analysis on factors influencing electric power client credit. The method of using genetic algorithm to optimize connection weight and threshold value of BP overcomes the defects of falling into the regional minutiae and slow velocity of convergence, builds the electric power client credit assessment model, and conducts empirical study. The research result indicates that BP neural network optimized by genetic algorithm plays a scientifically practical role in assessment of electric power client credit risk, supplying a reference for risk elusion of electric power clients. Keywords Genetic algorithm
BP neural network Credit assessment
11.1 Introduction At present, the status is that there are few researches on power client credit assessment, and the assessment indicator system is not established soundly yet. There are mainly two methods to study power client credit assessment, quantitative assessment and qualitative assessment [1]. Applying hierarchical analysis method, . This paper uses genetic neural network to build assessment model, and conducts assessment on power clients via machine learning [2]. The empirical analysis X. Wang (&) Fundamental Research Funds for the Central Universities Baoding, Baoding, 071003 Hebei, People’s Republic of China e-mail: [email protected] X. Wang School of Economy and Management North China Electric Power University Baoding, Baoding, People’s Republic of China
Y. Yang and M. Ma (eds.), Proceedings of the 2nd International Conference on Green Communications and Networks 2012 (GCN 2012): Volume 3, Lecture Notes in Electrical Engineering 225, DOI: 10.1007/978-3-642-35470-0_11, Ó Springer-Verlag Berlin Heidelberg 2013
85
86
X. Wang
indicates the model presented in this paper can determine accurately power client conditions, and serve as a solution to eluding client risk for power supply enterprises [3]. However, grounded on weight modification philosophy of error gradient descent, inevitably the training of BP neural network has the problem of falling into regional minutiae; while genetic algorithm has its origin in global parallel search algorithm of natural selection and genetic rules, and so features in strong macro search ability and global optimization [4]. Combining , initial weight and threshold value of BP network optimized by genetic algorithm together can conduct weight values in a large scope and solve the problem of falling into regional minutiae easily. It is rare for the combination of genetic algorithm and neural network to be applied in power client credit assessment. Based on the empirical research on power clients in Baoding City, neural network model optimized by genetic algorithm can realize effective credit assess
Data Loading...
