Abstract:Geological defects existing extensively in foundations often do harms to the safety of buildings. At the same time,construction of large-scale structures leads to variation of the natural geological condition. And in some cases,the geological defects may deteriorate with the construction of structures,which will bring adverse impacts on the normal operation of the buildings. Deformation modulus of the geological defect is a key parameter that reflects its deformation and strength character,so deformation modulus of the geological defect is used to study its evolvement rules. The theory,which is to build time series evolution model for geological defect by use of monitoring data,is presented. Then based on Elman recurrent neural networks,the detailed process to build time series evolution model for geological defect is proposed. The application shows that the proposed method is effective.