High-Precision Prediction Method for Mine Deformation Based on GNSS RTK and Stacking Ensemble Learning

To address the demand for high-precision deformation monitoring in mine exploitation, this paper proposes a high-precision mine deformation prediction method based on stacking ensemble learning, using Global Navigation Satellite System Real-Time Kinematic (GNSS RTK) data. Firstly, a fusion filtering preprocessing module based on median filtering, Butterworth filtering, Savitzky-Golay filtering, and Adaptive Kalman Filter (AKF) is established to suppress various types of noise in the original data. Secondly, a cumulative deformation time series is constructed, from which trend, seasonal, and residual components are decomposed; meanwhile, deformation rate and acceleration are calculated, which together serve as input features for the deformation prediction model. Finally, a stacking ensemble module is constructed by integrating three time series models and four machine learning models, with Elastic Net Regression (ENR) employed as the meta-model for dynamic weight optimization, thereby achieving high-precision prediction of cumulative deformation. Experimental results demonstrate that the fusion filtering preprocessing significantly improves the quality of the original data; the Root Mean Squared Error (RMSE) of the stacking ensemble module for prediction is consistently less than 0.3 mm, and it exhibits excellent trend consistency and response capability in the 72-hour ahead prediction. In summary, the proposed method provides efficient and reliable technical support for the active prevention and control of mine deformation disasters.