At present,seismic attribute analysis technology has become an indispensable method in oil and gas exploration and development.Because of the growing variety of seismic attributes and the increasingly high demands for reservoir prediction accuracy,the seismic attribute extraction,optimization and application became the key factors.First by taking the sandbody reservoir in Guantao Formation of Wellblock Shan 56 in Shengli Oilfield as an object of study,the suitable attributes were attracted on the basis of the rock physical statistics and attributes sensitivity analysis,and then the K-L transform was used for the dimension reduction of multiple seismic attributes.Finally the target zone oil saturation parameters are predicted based on artificial neural network identification technology.The effect of prediction is good,and it provides a basis for remaining oil development in the operation area.