Research on Prediction Method of Soft Soil Foundation Settlement
The current prediction methods of foundation settlement have biggish error under the condition of lesser foundation settlement observational datum. Aim at the localization of present prediction methods and the virtues of Support Vector Machine arithmetic, the method of predicting soft soil foundation settlement based on Least Square Support Vector Machine (LS-SVM) was proposed in this paper and compared with the neural network method and curve fitting method. The research results show that this proposed method is feasible and effective for predicting soft soil foundation settlement. Least Square Support Vector Machine provides a more advanced method than these conventional methods for predicting foundation settlement.
X. M. Dong "Research on Prediction Method of Soft Soil Foundation Settlement", Applied Mechanics and Materials, Vols. 97-98, pp. 36-39, 2011