Abstract: Baochu slope zone is one of the main battlefield of oil and gas exploration in Xihu Sag, and K area is the main target area of rolling potential exploration in Baochu slope at present. However, due to the complex tectonic background and the development of faults in K area, it is difficult to accurately characterize reservoir spatial distribution of Pinghu Formation. In this article, the lithology combination model of K area is predicted by automatic lithofacies recognition, and shear wave prediction is carried out by neural network algorithm, which has a good correlation with the measured shear wave. On this basis, high-resolution geostatistical random inversion has been carried out, P-wave impedance and post-stack seismic statistical lithology volume of K area are obtained, and the sandstone thickness and reservoir distribution characteristics of K area are precisely predicted. The spatial distribution of sandstone at the level of single sand group is realized, the accuracy of reservoir prediction of K area is improved. This article provides the basis for further well location research and reservoir expansion.