Prediction of settlement based on fusion model of Kalman filter and exponential smoothing algorithm

It is difficult to predict the settlement trend and to assess the stability of the settlement due to the large fluctuation and strong discreteness of the settlement monitoring data with noise, and the parameters of the conventional prediction models cannot be updated with the latest data.In this study, a prediction method based on the fusion model of Kalman filter(KF)and exponential smoothing(ES)algorithm is proposed.The idea of this method is as follows: Firstly, the Kalman filter is used to process the original settlement data for three times of filtering and noise reduction.Secondly, the first, second and third processing values of the Kalman filter are replaced by the smoothing values of the same processing times of the exponential smoothing method, and the gain coefficient of Kalman filter is replaced by the smoothing coefficient of the third exponential smoothing method.Finally, the parameters of the cubic exponential smoothing method are calculated by using the replaced smoothing values and smoothing coefficients, and the prediction model is established and extrapolated.The test results show that the KF-ES fusion model can significantly reduce the random noise interference in settlement data and has the advantages of strong adaptability, good real-time prediction, and it is suitable for short-term dynamic prediction.