Abstract: Displacement-time curves represent the status and relative stability of landslide deformation. They are vital to predict landslide deformation monitoring. Currently, landslide deformation monitoring data on the whole process is rare. Most monitoring data is short-term before sliding event. The deformation monitoring data of some unexpected landslide is even more difficult to collect. Therefore, studies on the type and distribution of curve clusters are difficult due to the lack of data. Some of the landslide warning theories are derived from inferential and theoretical analysis and lack data support. With a physical modeling approach, a physic simulation framework is designed for variable angle. The displacement data automatic acquisition system is developed. During the testing for more than one year, the equipments and model materials are continuously adjusted and optimized. Eventually, seven groups of twelve displacement-time curves are collected. The landslide deformation and failure of displacement-time curve clusters under different loading conditions are obtained. They are scientific data for further theoretical research.