基于特征正交分解的土坝变形安全监控模型

为解决性态变化剧烈的大坝施工期或固结期实测资料建模的非平稳、小样本和多重共线性等问题,以某均质土坝坝顶上下游水平位移为研究对象,采用基于特征正交分解的逐步回归分析方法,对土坝坝顶下游侧实测值进行建模,分析了模型的可预报时间尺度和精度。结果表明,对于性态变化比较剧烈的土坝,基于特征正交分解回归方法,可有效解决非平稳、小样本和多重共线性问题,但该新建模型的预测精度在接近建模时间尺度的1/3范围内有效,后期实测资料与预报模型预测的差异反映大坝性态已发生了变化。

Safety Monitoring Model of Earth Dam Based on Proper Orthogonal Decomposition

In order to solve the problem of non stationary, small samples and multicollinearity of the measured data modeling of homogeneous earth dam which the state changes dramatically during construction period or consolidation period, this paper takes the downstream deformation of dam crest in one trial earth dam as the research object. The analytical method of stepwise regression based proper orthogonal decomposition is used to make the modeling analysis of the downstream displacement in trial earth dam. The prediction time scale and accuracy of the model are analyzed. The results prove that the regression method based proper orthogonal decomposition can solve the problem of small samples and multicollinearity for the dam which the state changes dramatically. The prediction accuracy of the new model is effective close to 1/3 scale range of the modeling time. The difference between the latter measured data and prediction model can reflect the abnormal state of the dam in a certain degree.