支持向量回归机在尾矿坝浸润线预测中的应用  

Forecasting of infiltration route in tailing dam by Support Vector Regression

作　　者：李娟[1,2] 李翠平[2] 李春民[1] 李仲学[2] 

机构地区：[1]中国安全生产科学研究院,北京100029 [2]北京科技大学,北京100083

出　　处：《中国安全生产科学技术》2009年第1期76-79,共4页Journal of Safety Science and Technology

基　　金：国家自然科学基金项目“地矿工程系统建模新技术”(编号:50604003);教育部博士点基金项目“基于粗糙集和支持向量机的地矿体三维构模技术”(编号:20060008005);“十一五”国家科技支撑计划《矿井典型灾害模拟仿真及数据挖掘技术研究》(编号:2006BAK04B04)

摘　　要：本文提出一种基于支持向量回归机的浸润线高度预测方法,从大量尾矿库监测数据中选取有效样本,运用留一法对支持向量回归机参数进行优化选取,建立预测模型。结果表明:该方法能在小样本、高精度要求下对浸润线高度进行准确的预测。This paper presents a new prediction method for imqhration routes based on support vector Regression, According to the method, selecting valid samples of data from a large number of the monitoring data of the tailings dam is the first step; the kernel parameters for SVR （support vector Regression） are optimized automatically by u- sing the leave one out method, It not only reduces computing time, but also improve the accuracy rate for kernel parameters selection by using it, and then establishing the forecasting model. The results show that：the method can obtain higher accuracy with fewer samples for the Forecasting of infiltration route.

关 键 词：支持向量回归机 浸润线 尾矿坝 

分 类 号：X924.4[环境科学与工程—安全科学]