Abstract:Open-pit coal mines typically maximize slope excavation angles to improve economic efficiency, but this also significantly increases the risk of slope instability. Unmanned aerial vehicle (UAV) oblique photogrammetry, as an emerging technology, can acquire high-spatial resolution images, providing rapid and accurate data for slope stability analysis. The image data acquired through UAV aerial surveys is processed to generate a digital elevation model (DEM). Slope instability parameters such as slope and aspect, as well as relevant geological information, are extracted and used as input for an artificial neural network (ANN). ANNs have the ability to learn and generalize from unknown data. This study used a random sampling method to select open-pit coal mines from different regions as training samples. A feedforward backpropagation algorithm was used to conduct a slope sensitivity analysis, categorizing the study area into four disaster susceptibility levels. Four input parameters, slope, aspect, drainage density, and geological structure, were used in model training. Each factor was ranked according to its importance in inducing slope instability. Twenty percent of the samples were used for testing and 20% for validation. The resulting hazard risk assessment zoning map provides a scientific reference for future open-pit coal mine slope instability risk warning and mitigation planning.
