A DL-Based Approach for Evaluating Mine Slope Instability

Open pit mining is the most important technique for extracting mineral resources from the earth's crust. However, steep slope angles lead to an increased risk of slope collapse, the consequences of which could affect mining operations. Therefore, continuous assessment of excavation slope stability is an important part of open pit design and operation. The scientists used the Rock Engineering System (RES) paradigm, the limit equilibrium method, the artificial neural network (ANN), respectively and other methods to quantitatively analyze mine slope instability. However, these methods have problems such as difficulty in obtaining prior information, simple calculation model, and weak fitting ability, which make it difficult to apply to actual layered soil slopes and rock slopes. This research utilizes the powerful fitting ability of Deep Learning (DL) to construct a workflow of dataset construction - DL network training - DL network testing. In the process, the DL network is trained using the dataset of slope expert features. We provide a DL-based Classification Network model that can efficiently fit and evaluate slope states. Local environmental protection agencies and governments can use it to monitor slope instability in mines and mountainous areas, and effectively improve mine production efficiency.