Real-time monitoring of natural gas pipeline network can ensure the life and health of human beings and property safety in the surrounding area of the pipeline. Therefore, how to realize the rapid identification of pipeline leakage signal when leakage occurs is extremely important. Distributed acoustic sensing (DAS) technology is widely used in the field of pipeline leakage monitoring because of its advantages of wide coverage and high sensitivity. In this paper, a pipeline leak identification method based on DAS and one dimensional DPR-net (DAS pipeline leakage recognition network) are proposed. Pipeline leakage signal samples are collected under different working conditions (leakage aperture, leakage pressure, leakage direction). In test, the recognition accuracy of this method reached over 99%. There are reasons to believe that this method will provide an important technical means for DAS to detect pipeline leakage.