In the process of shale gas development in the Sichuan Basin, casing deformation accidents seriously affect the engineering efficiency and gas production, restricting the development efficiency of shale gas. Using seismic data to predict natural fractures and provide early warning before fracturing is an important direction for reducing the casing deformation rate. This article proposes a method to improve the accuracy of seismic fracture prediction and apply it to predict the risk of casing deformation. Firstly, a high-precision pre-stack processing technology, based on well constrained quantitative quality control, is implemented to preserve the valid signal of high frequency, followed by the logging-seismic post-stack high resolution processing technology (LSHR) based on neural network algorithm, helping to improve the seismic resolution significantly. Then, multiple high-precision fracture seismic attributes are calculated, selected and combined, to evaluate the accident risk of the platform wells to be fractured. The study has shown that, the pre-stack and the post-stack frequency expanding method proposed in this article, both contains well-constraining technique, which ensures a high correlation between the processing result and the synthetic record, and meanwhile enhances the effective high-frequency signals, increasing the dominant frequency by 30 Hz, frequency bandwidth expanded over 100 Hz. On the basis of that, the accuracy of fracture attribute generated by high resolution data, is significantly improved compared with that of the conventional data. Moreover, the combination of these multiple high-precision fracture attributes has a strong correlation with the locations where casing deformations take place. The risk locations of casing deformation early-warned before fracturing have a coincidence rate over 70%, providing a reliable reference for preventing casing deformation from the perspective of seismic interpretation.