About the author:

China Railway 18th Bureau Group Third Engineering Co., Ltd., Zhuozhou, Hebei 072750, China

Abstract:

An improved support vector machine model is proposed to predict settlement to improve theaccuracy and speed of settlement prediction. The sparrow search algorithm was introduced to optimize thepenalty parameters and kernel function parameters of the support vector machine to improve its predictioneffect. Taking the foundation excavation of Shangshagou Station of Nanchang Metro Line 4 as anexample, the empirical analysis of the model was carried out. The improved support vector machine wasused to predict the settlement of surrounding buildings caused by foundation excavation excavation andcompared with the traditional support vector machine model. The MSE of the support vector machinemodel improved by the sparrow search algorithm is reduced by 74%. The RMSE and the MAPE arereduced by 49% and 27%, respectively,which verifies that the improved support vector machine hasbetter prediction accuracy and prediction speed. The generalization ability of the improved support vectormachine has been verified and proven to be excellent.