Abstract: Geohazard is one of the major factors threating the safe operation of oil and gas pipelines. However, due to the inspection accuracy of equipment and the discreteness of inspection data, the pipeline safety cannot be analyzed quantitatively with a single monitoring and inspection technology. Thus, it is urgent to study the Digital Twin mechanism model for the structural safety of pipelines in combination with the multi-source monitoring and inspection data, so as to realize the intelligent perception and prediction of pipeline safety in geohazard areas. Herein, the process to establish a Digital Twin mechanism model for the structural safety of pipelines in geohazard areas was put forward. Meanwhile, by building a parameterized finite element model of pipelines in geohazard areas, a stress-strain database of the pipeline was established with consideration to the range of the variable parameters affecting the stress-strain distribution of pipelines, and then the highly nonlinear relationship between the variable parameters and the stress-strain state of the pipelines was fitted by BP neural network. In addition, combined with the real monitoring data of pipeline stress and strain, a mechanism model capable of accurately inverting the real stress and strain distribution along the pipeline was constructed with the particle swarm optimization algorithm. Moreover, the accuracy of the model was verified by an application case. Therefore, the research results could be applied to the quantitative safety assessment of the pipelines in geohazard areas, and core support could be provided for the construction of Digital Twin of pipelines.