Rock Mass Deformation Analysis based on Immune BP Network Model with Partial Least Squares Regression
Aiming at the correlation among variables, the non-linearity between independent variables and dependent variables and the deficiency of traditional BP neural network in monitoring data analysis, a new model is proposed based on partial least squares regression method, BP neural network and immune clone algorithm. The model deals with the correlation according to least squares regression, settles the non-linearity with BP neural network, and moreover it applies immune clone algorithm to search of BP neural network. Take rock mass deformation as the dependent variable and six factors of rock mass as the independent variables, a practical project are analyzed based on the model. The analysis result shows that the model is effective in overcoming the effects of correlation and nonlinearities with a speedy and stable convergence. Therefore it has superiority in practical applications.
monitoring data analysis partial least squares regression BP neural network immune clone algorithm rock mass deformation
JIN Yongqiang Zhao Erfeng
College of Water Conservancy & Hydropower Eng HoHai University Nanjing, China National Eng Research Center of Water Resource Efficient Utilization & Eng Safety, HoHai University
国际会议
大连
英文
1-4
2008-10-12（万方平台首次上网日期，不代表论文的发表时间）