Abstract:

Gathering pipelines are an integral part of oil and gas gathering and transportation systems, and their safe and stable operation is essential to the reliability of the oil and gas supply chain. Leakage monitoring plays a critical role in maintaining operational safety by identifying potential leaks and providing early warning of associated risks. In practice, however, fluctuations in equipment operating conditions, changes in weather conditions, and human activities near the pipeline can introduce strong interference into monitoring systems. Such interference can lead to frequent false alarms and undermine alarm credibility, thereby compromising subsequent risk assessments. An alarm therefore needs to be interpreted in light of the operating conditions and external influences present when it is generated.To address this challenge, this study proposes a leakage risk monitoring method based on multisource information fusion for gathering pipelines subject to strong interference. Operational scenarios are first analyzed to identify relevant interference factors. A Bayesian network is then developed to integrate information on equipment operating conditions, weather conditions, human activities, and alarms from multiple monitoring sources. Using the available evidence, the network estimates the posterior probability of actual leakage given the observed alarms and interference conditions. Leakage is then defined as the initiating event in an event tree model for quantitative consequence assessment. Its occurrence probability is set to the posterior leakage probability inferred from the Bayesian network. A relative risk index combines the inferred leakage probability with the assessed consequences to provide an integrated measure of leakage risk.The method is evaluated through a case study of a pipeline in Zichang, Yan’an, Shaanxi Province, China, using regional land-use information and weather statistics. Different interference scenarios are compared using the same observed alarm evidence. Specifically, only an acoustic monitoring alarm is observed, while the assumed leak orifice diameter and the conditions used for consequence calculations remain unchanged across the scenarios. As interference factors are progressively combined, the inferred leakage probability decreases from 93.63% to 75.49%. The corresponding relative risk index decreases from 1.4606 to 1.1776, representing a reduction of 19.376% relative to the baseline scenario.The case study illustrates how the method assesses the credibility of leakage alarms under strong interference and incorporates the resulting probability estimates into consequence-based risk assessment. By explicitly accounting for uncertainty about whether an alarm indicates an actual leak, the method reduces the influence of false alarms on risk assessment and provides risk estimates that better reflect the effects of interference under practical operating conditions. The proposed approach provides a basis for operational safety management, early warning of leakage risks, and emergency decision-making for gathering pipelines operating under strong interference conditions.

Key words: gathering pipelines, multi-source information fusion, risk monitoring, bayesian network, interference suppression, leakage risk assessment

摘要：

集输管道作为油气储运系统的重要组成部分，其安全稳定的运行对产业链至关重要，而泄漏监测正是保障其安全运行、识别泄漏事件并进行风险预警的关键技术手段。然而，实际运行中设备工况波动、天气环境变化及人员活动等强干扰因素，易导致泄漏监测结果误报率高、警报可信度不足，从而削弱后续风险评估的可靠性。针对这一问题，本文提出一种基于多源信息融合的强干扰集输管道泄漏风险监测方法。首先，以集输管道运行场景分析为基础，构建融合设备工况、天气环境、人员活动及多源监测报警信息的贝叶斯网络模型，实现对监测警报真实性的概率推理。在此基础上，将泄漏事件作为事件树的初始事件输入，并将贝叶斯网络推理得到的真实泄漏概率作为该事件的发生概率，定量评估泄漏后果，并采用相对风险指数对泄漏风险进行综合表征。以陕西省延安市子长市某管线为例，结合该区域土地利用、天气统计等数据开展实例分析。结果表明：在仅观测到声波监测报警、泄漏孔径和后果计算条件保持一致的情况下，随着干扰因素逐步叠加，集输管道泄漏概率由93.63%下降至75.49%，相对风险指数由1.4606下降至1.1776，较基准场景下降19.376%。实例验证结果表明，该方法能够在强干扰场景下有效判别泄漏报警的真实性，并将其与风险后果评估相衔接，显著降低误报对风险评估结果的干扰，使泄漏风险监测结果更贴近工程实际。研究成果可为强干扰环境下集输管道的运行安全管理、风险预警及应急决策提供科学依据。

关键词: 集输管道, 多源信息融合, 风险监测, 贝叶斯网络, 干扰抑制, 泄漏风险评估

CLC Number: 

• TE88

• X937