Abstract: In order to improve fracture identification for tight sandstone reservoirs with strong heterogeneity,the dynamic elastic mechanics theoretical model has been used with loggin data to predict fractures in tight reservoirs.First,the optimization algorithm and the array acoustic data were applied to obtain various modulus parameters of rocks,and then to calculate the rock pore aspect ratio (α) with the differential equivalent medium (DEM) theoretical model,so as to quantitatively characterize the fracture.Predictions of fractures in single wells achieved an accuracy level of 90% when predicted fractures were compared with actual.Such a high rate of coincidence between predicted and actual confirmed the reliability of DEM model in predicting fractures in coal-bearing tight sandstone reservoirs.Based on those results,the DEM model was used to predict the fracture plane distribution of the sand group Ⅱ of Shanxi Formation in the southern part of the Qinshui Basin based on well log data.Further,the fracture prediction results were consistent with fracture properties,structural locations and core fracture observations.The research showed that the dynamic elastic mechanics model was applicable to the quantitative characterization of fractures in coal-bearing sandstone reservoirs.