Natural gas pipeline network leakage early warning method based on machine learning method

• Main Authors: ZHU SONGQIANG, CAI KUN, SHEN GUOLIANG, SHEN JIAYUAN, XU YATING, CHEN JIALE, YAN JUNCHI, QIAN JIREN
• Format: Patent
• Language: Chinese; English
• Published: 28.01.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention belongs to the technical field of pipeline leakage detection, and particularly relates to a natural gas pipeline network leakage early warning method based on a machine learning method. The method the following steps: S1, collecting historical data indexes during normal work and leakage of a natural gas pipeline as a training set, and constructing a topological structure diagram; S2, establishing a fault detection and small probability early warning model and a fault reason analysis model, and training; S3, collecting working data of the natural gas pipeline in real time by utilizing the model, and judging whether leakage occurs or not; S4, when judging that leakage occurs, using the model for obtaining the cause of leakage; and S5, collecting a data sample, and performing fine adjustment and updating on the model. The invention has the characteristics that the leakage of the natural gas pipeline can be automatically pre-warned, the empirical updating model is established, the current model is q