泄漏监测是石油产品在长距离输管道送中的安全保障措施。泄漏监测软件在系统测试时,虽然可以通过在管道上放油来模拟泄漏,但受客观条件限制,在测试内容与测试次数两方面都很难达到充分检验这一类软件功能的目的。设计并实现了一种基于深度学习的管线泄漏仿真与测试平台(以下简称平台)，使用管道的实际运行数据训练出管段模型,同时将典型的管道操作流程制成情景模板,结合在项目中开发的仿真指令,可以较为真实地仿真长输管线的各种运行状态和测试所需的泄漏事件，并将数据以通讯协议的方式提供待测试系统使用。平台的建成丰富了评价泄漏监测软件的试验手段,具有安全、高效、低成本、不限测试次数的优点,为管道行业测试泄漏监测软件提供一种新的解决方案。

Leakage monitoring is a safety measure for petroleum products in long-distance pipeline transportation. When testing the leakage monitoring software, although the leakage can be simulated by discharging from the pipeline, it is difficult to fully test the functions of this kind of software in terms of the content and cycle times due to the limitation of site conditions. A pipeline leakage simulation testing platform based on deep learning is designed and implemented. The actual operation data of the pipeline is used to train a pipeline segment model, and the typical pipeline operation procedure is made into a set of scenario templates. Combined with the simulation instructions developed in the project, various operation states of the long-distance pipeline, including leakage events, can be realistically simulated. The testing platform delivers the data to the system to be tested via standard communication protocols. The establishment of simulation testing platform enriches the testing methods of leakage monitoring software evaluation, which is safe and efficient, and allows unlimited repetition of testing cycles with lower costs. This study provides a new solution for testing leakage monitoring software in the pipeline industry.