The Neural Network Training Based on A New ABC Algorithm and Its Application
LI Xiao-jing;Department of Public Infrastructure,Guangxi College for Preschool Education;
An improved bee colony algorithm which has powerful ability of global optimization was proposed firstly in the paper. And the algorithm has improved the update formula of artificial bee colony algorithm and simplified whole search strategy. The global optimization tests on multiple high-dimensional and multi-extremum benchmark functions with the algorithm indicate that the convergence precision of improved bee colony algorithm is improved greatly. Secondly,improved bee colony algorithm was applied to weight training of neural network.Finally,the neural network based on bee colony algorithm train was applied to precipitation forecast. And the test results show that the training method is feasible.