针对传统GM(1,1)模型在大坝位移预测中的缺陷,通过引入时间变量构建缓冲算子、重构背景值、优化残差等方式改进传统GM(1,1)模型。以某大坝测点径向位移样本为例,分别采用两种模型对观测值进行拟合预测。结果表明,改进的非等间距GM(1,1)模型预测精度有较大提高,更适用于大坝位移监控。研究成果可为大坝位移预测提供参考。 In order to overcome the shortcomings of the traditional GM (1,1) model in dam displacement prediction, the traditional GM (1,1) model is improved by introducing time variables to construct buffer operators, reconstruct background values ​​and optimize residuals. Take a sample of radial displacements at a measuring point of a dam as an example, two models are respectively used to predict and fit the observed values. The results show that the prediction accuracy of the improved non-equidistant GM (1,1) model is greatly improved, which is more suitable for dam displacement monitoring. The research results can provide a reference for the prediction of dam displacement.