最小二乘支持向量机模型在大坝监测中的应用  

Application of Least Squares Support Vector Machine Model in Bam Monitoring

作　　者：路亮[1] 张爱军[1] 

机构地区：[1]西北农林科技大学水利与建筑工程学院,陕西杨凌712100

出　　处：《人民黄河》2013年第11期99-100,103,共3页Yellow River

摘　　要：提出了一种基于混沌时间序列的最小二乘支持向量机预测方法,该方法的依据是相空间重构技术以及最小二乘支持向量机模型(LS_SVM)。阐述了基于混沌时间序列的最小二乘支持向量机模型的算法步骤,并指出该模型的评价指标为平均绝对误差(MAE)以及预测均方误差(PMSE)。利用该模型对某混凝土大坝5#坝段102号测点的垂直位移进行了预测,结果表明:基于混沌时间序列的LS_SVM模型的预测性能较好,能够很好地体现出模型的实际应用能力;模型的拟合及预报结果能够满足精度要求,与回归模型相比具有预测结果精度较高的优点。It put forward a kind of chaotic time series using least square support vector machine prediction method,which was based on phase space reconstruction and least squares support vector machine model （LS_SVM）. It also described the steps of the least square algorithm for chaotic time series based on support vector machine model and pointed out that the evaluation index of the model for the mean absolute error （MAE）and the mean square error of prediction （PMSE）. It predicted the vertical displacement of No. 102 measuring point in No. 5 dam section of a concrete dam by using the model. The results show that LS_SVM model of performance prediction of chaotic time series based is better. It can well reflect the model’s practical application ability,the fitting and prediction results of the model can meet the precision requirement,compared with the regres-sion model have high precision of prediction and can provide a new solution for dam monitoring.

关 键 词：相空间重构技术 最小二乘支持向量机模型 大坝位移 

分 类 号：TV698.1[水利工程—水利水电工程] TV642.1