考虑大坝蓄水期监测资料平稳性较差、坝体性态随水位升高而有所改变等特征,在原有传统统计模型的参数求解过程中引入遗忘因子,通过建立遗忘矩阵突出现在及近期资料对模型的贡献,尝试建立了大坝蓄水期监测资料的时变分析模型。实例分析表明,由于模型参数根据大坝蓄水期实际情况实时进行了阶段性更新,其拟合及预测精度均优于传统统计模型,更适用于蓄水期监测资料分析。
Considering the unstable monitoring data and the changing dam behavior in the storage period,the forgetting factor is introduced into the traditional statistical model for solving the model parameters,and a forgotten matrix is established to give prominence to the contributions of recent data.Then the time-varying analysis model is set up for monitoring data for the dam storage period.Case analysis shows that the model parameters can be periodically updated with the changes of the storage state,and the accuracy of fitting and forecasting is better than that of the traditional statistical model,so it is more suitable for analyzing the monitoring data for the storage period.