Inversion of permanent deformation parameters of neural network based on bee colony optimization algorithm

Abstract

In this research, a method of artificial bee colony algorithm was proposed to optimize BP neural network parameters and then introduced to parameter inversion of geotechnical engineering. The method is based on the research and analysis of permanent deformation of geotechnical earthquake engineering and aimed at addressing the shortages of BP neural network which is slow to converge and may easily get minimum value. This method not only strengthens the global optimization ability of bee colony optimization algorithm, but also reflects the high accuracy of predicting nonlinear problems of BP neural network and the strong generalization ability. The proposed method also improves the prediction accuracy This new proposed method was validated by using the earthquake data of Zipingpu concrete faced rockfill dam in WenChuan earthquack. The results showed that this method can better solve the problem of permanent deformation parameters and can be extended to inversion predict of the same kind.