In the process of open-pit mine stripping,the surface and slop that influenced by open-pit mining stripping, rock mass structure and lithology, blasting vibration, slope angle, groundwater and precipitation start to move and deform,which threatens the safe use of industrial and civil buildings adjacent to the open pit mining .In this paper, through the establishment of increasing the momentum of adaptive BP neural network,with the aid of existing sample data,we carried out on the network training, testing, and compared with the target, its precision is higher.The result showed that network is effective and reliable,and the trained network is applied to the observation point prediction,so we get the value of its settlement deformation in 2013.The method is expected to have the important meaning on the multi-factor coupling surface subsidence .