Evaluation of soil Liquefaction based on GRNN model and empirical procedure
When evaluating the potential seismic soil liquefaction based on the theory of neural networks, the relatively precise training samples are needed, and some improvement methods should be combined with the process of application to enhance the precision of networks calculation. In this paper, emended empirical evaluation liquefaction equations are used to evaluate soil liquefaction, and then the results are divided into two parts. They are training and predicting samples of General Regression Neural Network (GRNN). We adopt different data-pretreatment methods when using GRNN to assess soil liquefaction potential and find that proportion-united pretreatment can be more simulative in enhancing the precision of calculations. Research shows that GRNN has great application prospect in evaluating the potential of soil liquefaction. As the improvement of laboratory conditions and the accumulation of training samples from different seismic fields, the method will procure more precise evaluation results.
Biao LI Naiqi SHEN Yefan LI
School of Engineering & Technology,China University of Geosciences,Beijing,China
国际会议
International Symposium and The 7th Asian Regional Conference of IAEG(国际工程地质与环境协会年会暨第七届亚洲工程地质会议)
成都
英文
442-445
2009-09-09（万方平台首次上网日期，不代表论文的发表时间）