用引入时间参量的DGM(1,1)模型预测大坝沉降

Application of DGM( 1,1) with time parameter in dam settlement forecasting

针对传统灰色模型的预测稳定性缺陷,以及实际水利工程中监测数据采集不等时距的特点,提出了引入时间参量的离散灰色预测模型DGM(1,1)的建模方法。该方法在原离散模型的基础上,引入时间累积量修正模型参数,以此反映出数据的不等时距性以及时效性。通过对糯扎渡大坝心墙沉降观测数据的建模分析,将分析结果与另一模型的拟合和预测结果进行比较,证明新模型具有较高的预测精度。

Considering the forecasting stability problem of traditional GM（1,1） model as well as the non-equal time-span features of observing data in the real hydropower projects, we propose a discrete grey forecasting model（DGM（1,1）） that takes account of time parameter. Based on the original discrete model, the cumulative amount of time is introduced to correct the model parameter to reflect the non-equal time-span feature and aging of the data. The core-wall settlement data of Nuozhadu Dam is analyzed by the model and the fitting and forecasting results are compared with another model, which proves that the presented discrete model has better forecasting accuracy.