<p>Structural damage identification (SDI) is critical to maintain the structural safety of oil and gas pipeline systems, where joint loosening is a major threat to the system integrity. Due to model simplification/modeling error, traditional method/model cannot accurately simulate and quantify the joint loosening damage. This study proposes a novel joint element model to simulate the mechanics behavior of pipelines with loosened joints and to identify the joint loosening. Compared with the traditional reduced stiffness method, the proposed joint element model can simulate the joint loosening damage more accurately and efficiently. To achieve efficient joint damage localization, this study adopts neural network (NN) to learn the nonlinear relationship between modal parameters and joint damage locations. A two-step strategy via integrating multilayer perceptron (MLP) and finite element (FE) model updating is proposed for joint loosening localization and quantification of pipeline structure. Performance of the proposed method is examined via a numerical case study and laboratory testing of a pipeline structure with two bolted flange joints. It is found that the proposed method provides a classification accuracy of 96.87%. Research results show that the proposed joint element model and loosening identification method can accurately localize and quantify the pipeline joint loosening. The proposed method has the potential to efficiently identify joint loosening in complex pipeline systems with multiple joints.</p>