Determination of steady deformation stage for creeping slope and application analysis of comprehensive early warning model

Rock and soil slopes follow the theoretical characteristics of the “three-stage creep deformation” during instability and failure. To address the key issue of accurately applying the deformation laws of each creep stage in landslide early warning and forecasting, a quantitative division method for the three creep stages of slopes and a comprehensive early warning model are proposed. The division method adopts the forward crossing method of long-term and short-term average deformation velocities to quantitatively define the end point of the steady deformation stage. Based on the characteristic that velocity data in the steady deformation stage follow a normal distribution, combined with the reverse hypothesis test method, the start point of this stage is accurately identified, thereby clarifying the boundary moments for dividing the three creep stages and the time interval of the steady deformation stage. On this basis, the representative velocity value of the steady deformation stage is determined by the weighted average method, and the tangent angle parameter of the accelerated deformation stage is dynamically derived therefrom. Finally, a comprehensive early warning and forecasting criterion integrating deformation velocity, acceleration, velocity variation, and tangent angle is constructed, forming a three-level early warning system. Two typical landslide cases show that the proposed quantitative division method can accurately identify the time nodes of the three creep stages, and combined with the comprehensive early warning criterion, it can effectively realize landslide early warning. The quantitative division method and comprehensive early warning model proposed in this study achieve accurate identification of slope creep deformation stages and dynamic landslide forecasting, providing reliable technical support for the engineering practice of landslide disaster early warning and forecasting.