Research on Technical Standard Alliances Risk Warning based on BP Neural Network Optimized by Hierarchical Genetic Algorithm

By : Jing Hu, Xiaomeng Su

Page No: 415-430

Abstract :
In this study, a risk early warning model combining hierarchical genetic algorithm and BP network for technical standard alliance was proposed. The existing methods for training neural networks can only train weights of BP network, and structure of the network must be pre-set by a certain method. This study tried to optimize BP network with hierarchical genetic algorithm. Specifically, it combined both from perspectives of “weight learning” and “network topology learning”, optimizing network structure and network weight. Moreover, two genetic operators, namely crossover and mutation, were improved, which increased diversity of population evolution and avoided premature convergence of the network. Finally, through empirical analysis and a comparison with standard BP network algorithm, the model is proved feasible and effective in solving the problem of “optimal network structure”, and it enhances speed and accuracy of BP network.

Authors :
Jing Hu
China Jiliang University, No. 258 Xueyuan Street, Xiasha Higher Education District, Hangzhou, 310018 P.R. China.

Xiaomeng Su
China Jiliang University, No. 258 Xueyuan Street, Xiasha Higher Education District, Hangzhou, 310018 P.R. China.
 

DOI: https://doi.org/10.32381/JCISS.2020.45.1-4.16