摘要
针对路基沉降分析和预测中的正则回归模型算法,分析了正则参数的选取方法。根据矩阵条件数理论和最小二乘原理,提出了以控制信息矩阵条件数和残差平方和大小的正则参数选取新策略。结合二元回归模型的计算结果表明:提出的正则参数选取策略不仅降低了矩阵条件数,也能获得较小的残差平方和值。
Regularized polynomial regression model is used to analysis the monitoring data of subgrade settlement. Based on matrix condition number theory and least square principle, a new regularization parameter selecting strategy was proposed which aim to control the number of matrix condition and residual sum of squares. The calculation results of binary regression model show that the new regularization parameter selecting strategy not only decreases the matrix condition number, but also obtains small residual sum of squares value.
出处
《铁道科学与工程学报》
CAS
CSCD
2010年第4期125-127,共3页
Journal of Railway Science and Engineering
基金
湖南省科技计划资助项目(2008SK3054)
关键词
路基沉降
正则回归模型
正则参数
条件数
残差平方和
subgrade settlement
regularized polynomial regression model
regularization parameter
condition number
residual sum of squares
