一种组合最小二乘支持向量机的研究及其应用

摘要

This paper researched deformation analysis based on statistical learing theory（SLT） and LS-SVM.First,a new optimization algorithm is proposed based on Niche Genetic Algorithm and Cross-validation,then the combined model of LS-SVM and Chaos theory is used to analyze the deformation data and also compared with BP and RBF neural network.The results show that the prediction model has a good effect.%将统计学习理论和LS-SVM用于变形分析预报,采用小生境遗传算法与交叉验证法相结合进行LS-SVM参数的选取,并用参数优选后的LS-SVM与混沌理论相结合对变形监测数据进行建模预测,并与BP和RBF两种神经网络的预测结果进行了比较分析。实例表明,基于组合LS-SVM的变形数据预报模型具有良好的效果。