变形监测分析和预警对大坝健康运行起着关键的作用, 但是监测数据都不可避免地存在随机误差。卡尔曼滤 波法可以有效地剔除测量数据中的噪声, 然而, 利用其对大坝未来的趋势位移做出预测时与实际情况吻合度不高。 因此提出了马尔可夫链- 卡尔曼滤波法, 既能剔除测量数据噪声又可以准确的预测未来位移。利用浙江某拱坝位 移实测资料, 拟合并预测了该大坝的变形, 验证结果表明该方法的拟合预测效果良好, 可用于拱坝变形预测和安全 监控。

Deformatio n mo nitor ing and ea rly war ning analysis of the dam play a key ro le in the healthy operatio n of dam. H ow ev2 er , there ar e unavo idable random erro rs in the monito ring data. Kalman filter ing met ho d can eliminate the no ise fr om t he ob2 serv ed data effect ively; howev er, the predictio n o f the displacement trend o f the dam does not ag ree w ell w ith the actua l condi2 tions using this method. In this paper , Mar ko v chain2Kalman filter metho d w as pro posed, w hich can not only remov e the r andom erro r but a lso pr edict the displacement accur ately . Based o n the observed displacement data o f an ar ch dam in Zhejiang Pr ovince, the metho d show ed an accurate prediction of dam deformatio n. Therefo re, Mar kov chain2Kalman filter met ho d can be used in the pr ediction of ar ch dam defo rmatio n and security monit or ing .