Abstract: Machine learning is a major means to realize artificial intelligence (AI), which can provide support for decision making through exploration of data law and establishment of prediction model. Since the oil and gas pipeline system has the characteristics of numerous equipment, complicated structure and complex technology, machine learning is introduced to solve the problems that are hard to solve with pure mathematical models with AI technology and to replace personnel to complete some boring, tedious and dangerous tasks. Here, the application of 3 types of machine leaning, such as deep learning, reinforcement learning and transfer learning, were reviewed with reference to the production process of oil and gas pipeline system, covering the application scenarios like pipeline leakage, recognition of multi-phase flow pattern, equipment fault diagnosis and tank target detection. The application framework of AI technology in oil and gas pipeline systems was established, and it was also pointed out that the application of deep learning, reinforcement learning and transfer learning in this field was prospective. Finally, the application of machine learning in oil and gas pipelines was prospected, so as to provide references for the intelligent research and development of oil and gas pipeline systems.