Combined Prediction of Foundation Pit Deformation under the Condition of Data Filtering

In order to evaluate the deformation law of foundation pit construction process reasonably, based on the deformation monitoring results, the filtering of deformation data was realized by double tree complex wavelet, and then the sub-combination prediction was realized by GWO RVM model, ARIMA model and chaos theory. The analysis of the example shows that the deformation data of foundation pit can be effectively decomposed into trend and error components by double tree complex wavelet. The decomposition effect can be further improved by optimizing the model parameters, which is more powerful than traditional wavelet. At the same time, the applicability of various sub item prediction models in different deformation components is also strong. The average relative error of the combined prediction results is about 2%, which is obviously better than the traditional prediction model. It verifies the applicability of the combined prediction method in the deformation prediction of foundation pit, and provides a new way for the research of deformation development law of foundation pit.