Abstract： The prediction of borehole collapse pressure plays a key role in drilling safety, reducing construction cost and realizing efficient drilling. The fracture development in complex ultra-deep geological conditions has a great influence on the prediction of borehole collapse pressure. The conventional methods are mostly based on finite element simulation for 3D geomechanical modeling and 3D collapse stress prediction. Although the method is highly accurate, it requires huge computing power resources. In order to solve this problem, an efficient and fast in-situ stress modeling method flow driven by seismic data is proposed in this paper, which is then used for 3D collapse pressure prediction. Firstly, combined with multi-scale data of pre-stack seismic and rock mechanics logging, a combined spring model with curvature properties is established to complete the efficient and rapid modeling of three-dimensional in-situ stress field, and is used to calculate threedimensional borehole stress. Secondly, based on the maximum likelihood attribute, the fracture development is obtained from 3D seismic data to provide 3D weak surface attribute parameters for the study area. Finally, the collapse model of sliding along fracture plane is calculated by using Mohr-Coulomb criterion, and the collapse pressure prediction of fractured formation is realized from one-dimensional logging data to three-dimensional working area. The method is applied in the woodworking area of Tari, and the results show that the prediction results of the model are in good agreement with the measured data, reaching 93.79%. The prediction results of collapse pressure are in good agreement with the interpretation results of formation microresistivity scanning imaging, which verifies the feasibility of this method in predicting borehole wall collapse events. This study can realize the rapid modeling of collapse pressure with high precision, and effectively provide an integrated solution of geological engineering for drilling construction in ultra-deep and complex areas.