Strength Prediction of Stabilized Soil Based on Orthogonal Experiment and BP Neural Network
Author(s): WANG Hongxiao, WANG Yinmei
Pages: 27-30
Year: 2016 Issue: 7
Journal: Pearl River
Keyword: solidifying agent; combined strengthening; unconfined compressive strength; orthogonal experiment; BP Neural Networks;
Abstract: Improved loess that combined with a newly polymeric solidified material SH and lime is used in the orthogonal test of uncon-fined compressive strength. And the results show that the optimal combination is 10% of SH mixed with 4% of lime with 28-day-dr-ying-time. Range analysis and variance analysis are carried out for orthogonal test, and BP artificial neural network model is predicted for strength. The prediction result basically agrees well with the test result, and maximum relative error of the combination is 3. 16%, which means the model has higher prediction accuracy. This study can be used as references for further studies on engineering charac-teristics of improved loess.
Pages: 27-30
Year: 2016 Issue: 7
Journal: Pearl River
Keyword: solidifying agent; combined strengthening; unconfined compressive strength; orthogonal experiment; BP Neural Networks;
Abstract: Improved loess that combined with a newly polymeric solidified material SH and lime is used in the orthogonal test of uncon-fined compressive strength. And the results show that the optimal combination is 10% of SH mixed with 4% of lime with 28-day-dr-ying-time. Range analysis and variance analysis are carried out for orthogonal test, and BP artificial neural network model is predicted for strength. The prediction result basically agrees well with the test result, and maximum relative error of the combination is 3. 16%, which means the model has higher prediction accuracy. This study can be used as references for further studies on engineering charac-teristics of improved loess.
Related Articles