砂土地震液化预测的GA_SVM_Adaboost模型

为快速准确地对砂土液化情况作出预测，选取地震烈度、地下水位、覆盖厚度、标贯击数、平均粒径、地貌单元、土质及不均匀系数为主要影响因素，运用相关性分析和因子分析模型对其进行分析和属性约减，采用遗传算法（GA）对支持向量机（SVM）的参数寻优，结合Adaboost迭代算法，建立预测砂土地震液化的GA_SVM_Adaboost模型。选用唐山地震砂土液化现场勘察资料中的329组数据对模型进行训练，利用该模型对剩余68组砂土液化数据进行预测。最后，将预测结果与GA_SVM和SVM模型预测结果进行比较。结果表明，3个模型的平均预测准确率分别为100%、98.04%、89.71%，基于因子分析的GA_SVM_Adaboost模型的预测准确性优于GA_SVM模型和SVM模型，是一种解决砂土地震液化预测问题的有效方法，具有一定的应用参考价值。 

GA_SVM_Adaboost model for prediction of earthquake-induced sandy soil liquefaction

In order to predict the liquefaction of sand, the seismic intensity, groundwater level, covering thickness, standard number, average particle size, landform, soil quality and inhomogeneity coefficient are selected as influencing factors. Genetic algorithm(GA) is used to optimize the parameters of support vector machine(SVM) by using correlation analysis and factor analysis model, and combining with Adaboost iterative algorithm, the GA_SVM_Adaboost model for predicting the liquefaction of sand is established. 329 sets of survey data of sandy liquefaction site in Tangshan earthquake were used to train the model, and 68 samples of sandy liquefaction data were predicted by using the good model. Finally, The predicted results are compared with that of GA_SVM model and SVM model. The results show that the average prediction accuracy of three models is 100%, 98.04% and 89.71% respectively. The GA_SVM_Adaboost model based on factor analysis is better than GA_SVM model and SVM model which could improve the prediction accuracy. It is an effective method to predict the liquefaction of earthquakes.