The effects of underground mining are reflected on the surface in the form of ground subsidence, which causes damage to infrastructure in the wider mining area. In recent decades, awareness of the importance of protecting the environment, and thus surface structures, has led to the development of various modeling approaches for predicting surface subsidence over underground mining excavations. These differ depending on the mining methods and the natural conditions in the area of each mineral deposit. The dissertation research was divided into three sections, and our goal was to develop a hybrid prediction model that combines several tested solutions and allows prediction of subsidence for any surface point. As part of the preliminary research, we examined common methods for monitoring surface subsidence in the first section. For reasons of practicality and accuracy, we chose the UAV photogrammetry method for monitoring ground subsidence. In the second part, we analyzed the development of the subsidence of a point and compared different prediction models. Based on the obtained results, we selected the modified sigmoid function as the most suitable model for point subsidence prognosis, which we used to interpolate and extrapolate the measured subsidence values within the required accuracy limits. The mentioned subsidence prediction is based on the assessment of the subsidence trend and the prediction of the time of the next and final measurement, when the consolidation starts and further subsidence is negligible. In the third part, the theory of the hybrid model for dynamic subsidence prediction was explained, in which the area of influence is divided into rectangular sectors, each sector containing a cloud of points and a plane that best fits these points. In this way, it is possible to implement subsidence prediction in the monitoring regime. Each plane was defined by a centroid whose heights were used as input data for calculating the parameters of the modified sigmoid function. The developed hybrid model, which includes a sigmoid function, a computational grid with sectors, and a point cloud comparison, can be used to optimize surface remediation over an underground excavation. This is achieved by identifying areas of intense subsidence and categorizing sectors of the computational grid. The hybrid model was successfully verified on actual surface subsidence data in the most active mining area of the Velenje Coal Mine (slo. Premogovnik Velenje). The proposed method is a good basis for prediction of surface subsidence in areas where conventional monitoring is not possible, and enables time- and cost-efficient adjustment of further subsidence measurements based on the predicted dynamics of terrain subsidence.
|
