Abstract: The landslide of waste dump is one of the major disasters in the mine, which seriously threatens the safety production of the mine.The mine waste dump is affected by many factors, such as geology, man-made, nature and so on.It is difficult to accurately and effectively predict the landslide deformation trend and safety stability by using a single index.In order to solve this problem, an early warning model of mine waste dump landslide based on adaptive limit learning machine is proposed.By using the indexes of friction angle, slope angle, slope height, bulk density, pore water pressure coefficient and cohesion as the input unit and stability coefficient as the output unit, the existing data are trained and tested.The application results show that the theoretical calculation results of this method are basically consistent with the actual situation of the project, and it had good adaptability to improve the landslide early warning of the mine waste dump ability and accuracy.