主成分-BP组合模型在坝顶位移监控中的应用  

Application of Principal Component-BP Combined Model in Dam Crest Displacement Monitoring

作　　者：张伟[1] 游艇[2] 李双艳[3] 张诗悦[1] 

机构地区：[1]河海大学水利水电工程学院,江苏南京210098 [2]辽宁省水利水电勘测设计研究院,辽宁沈阳110006 [3]河北省水利水电勘测设计研究院,天津300250

出　　处：《人民黄河》2012年第2期115-117,共3页Yellow River

基　　金：水利部公益性行业科研专项目(201101013);国家自然科学基金资助项目(51079046;50909041;508090255;0879024);"十一五"国家科技支撑计划项目(2006BAC14B032;008BAB29B03)

摘　　要：在大坝位移监控中,主成分分析和BP神经网络方法运用较广,建立两种方法的组合模型,可以结合两种方法的优势,利用主成分分析消除坝顶位移预测中多重相关性给模型带来的不利影响,利用BP网络可通过自组织、自学习和高容错性等功能解决复杂非线性问题。通过实例计算可知,组合模型精度能够达到要求,可以推广应用。Principal component analysis and the BP neural network are two methods which are widely used in dam displacement monitoring model. In this paper, the two methods are combined together. The principal component analysis is used to erase the disadvantages in crest displacement prediction created by multiple correlation, while the BP neural network is used to solve complex nonlinear problems by self-organization, self-learn- ing and high fault tolerance features. This model combines the advantages of both methods, from an example calculation, the accuracy can meet the requirement and so the model can be widely used.

关 键 词：大坝观测 位移监控 主成分-BP组合模型 精度 

分 类 号：TU196[建筑科学—建筑理论]