改进的PLSR算法在大坝安全监控中的应用  

Improved Partial Least Square Regression and Its Application to Dam Safety Monitoring

作　　者：荆凯[1] 

机构地区：[1]河海大学水资源利用与工程安全国家工程研究中心,江苏南京210098

出　　处：《水电能源科学》2010年第1期64-66,共3页Water Resources and Power

基　　金：国家自然科学基金雅砻江水电开发联合研究基金资助重点项目(50539110);高等学校博士学科点专项科研基金资助项目(20070294023);国家自然科学基金资助项目(50879024);江苏省"333高层次人才培养工程"科研基金资助项目(2017-B08037)

摘　　要：针对自变量中含有过多与因变量无关的信息时应用偏最小二乘回归算法将出现偏差的问题,引入改进的偏最小二乘回归算法对坝体水平位移建立数学模型,并以安徽陈村混凝土拱坝为例进行分析。结果表明,改进的偏最小二乘回归算法通过正交投影变换解释矩阵能有效改善噪音的影响,建立的统计模型在保证解释能力的前提下其拟合效果和预测能力较偏最小二乘回归算法均有所提高。Aiming to irrelevant information between the independent variable and the dependent variable leading to the deviation problem using partial least-squares regression, the improved partial least-squares regression algorithm was applied to establish the mathematical model of horizontal displacement of the dam. Taken the certain concrete arch dam body as an example, the results show that the improved partial least-squares regression model can effectively eliminate the impact of noise by orthogonal projection transformation of the interpretation matrix,furthermore the fitting effect and the forecast capacity obtained by the established statistic model are better than those of the partial least-squares regression model on the premise of ensuring the explanation capacity.

关 键 词：大坝安全监控 偏最小二乘回归 改进的偏最小二乘回归 正交投影 拟合 

分 类 号：TV698.1[水利工程—水利水电工程]