In response to the strong multi-solution problem of conventional seismic exploration methods in geological sweet spot prediction of shale， this paper fully explored one-dimensional whole rock analysis data and well logging data， two-dimensional geological data， three-dimensional post-stack seismic data， and five-dimensional offset vector tile （OVT） orientation information. In addition， the paper researched geological sweet spot identification technology of shale based on fuzzy fusion prediction to improve the accuracy of geological sweet spot prediction of shale. Firstly， the approximate distribution direction of faults in the study area and the distribution area of favorable shale lithofacies were statistically analyzed， and the sensitive azimuthal section was selected to stack the OVT data based on azimuth. Then， the fractures on the plane were analyzed based on pre-stack preferred azimuth， and the two-dimensional prediction of favorable shale lithofacies was carried out using a feedforward neural network. Finally， a fuzzy fusion technology based on an improved Sigmoid and Takagi-Sugeno （TS） function was developed， which could weigh the significance of controlling factors in geological sweet spot identification of shale according to their degree of influence and effectively integrate the predicted results of azimuthal anisotropy fractures on the plane with those of neural network-based shale lithofacies，so as to realize decision-making integration for identifying geological sweet spots of shale. This technology has been applied in the classification and grading evaluation of geological sweet spots of shale in the Lower Submember of the 3 ^rd Member of Shahejie formation （Es₃^U） in Bonan Depression. Areas with poorly developed fractures and unfavorable lithofacies that could interfere with the analysis of geological sweet spots of shale were filtered out. Based on the predictions of fractures and lithofacies， The geological sweet spots of shale in the study area were categorized into three classes. The areas characterized by the superposition of developed fractures and favorable lithofacies were classified as sweet spots of Class I， which showed a high degree of consistency with actual drilling results and achieved notable application effects. The research results indicate that the classification and grading evaluation can be achieved using the geological sweet spot identification technology of shale based on fuzzy fusion prediction， which improves prediction reliability and provides reliable technical support for shale oil exploration.