埋地纯氢管道多变参数泄漏规律及预测模型研究

打开文本图片集

中图分类号：TB9;TE88;TH48 文献标志码：A 文章编号：1674-5124（2026）02-0040-12

Abstract： To investigate the gas leakage difusion pattrns and prediction techniques of buried pure hydrogen pipelines under the influence of multiple factors， this paper employs multidimensional numerical analysis and an SSA-GA-BP neural network to focus on analyzing the effects of leakage aperture size，location，burial depth，and soil permeability characteristics on hydrogen diffusion.The results show that hydrogen diffsion velocity is positively correlated with leakage aperture size and negatively correlated with pipeline burial depth and soil resistance coeficient; the leakage point location significantlyaffects diffusionbehavior， with diffsion rates exhibiting the following order： above > upper right > horizontal > below. When leakage occurs at a 45∘ angle to the upper right，the vertical upward rate is significantly greater than the horizontal direction.Difusion rates under different soil conditions follow the order： pure sandy soil > pure loamy soil > pure clay soil. A combination of small pore size and clay soil can lead to high-pressure accumulation and slow hydrogen release. The diffusion concentration prediction error based on the SSA-GA-BP neural network is below 1.2% . The study indicates that soil permeability characteristics have the most significant impact on the diffusion of leaks from buried pure hydrogen pipelines. Low-permeability soils can reduce the influence of leak pore size and burial depth; shallow burial depth + clay + small pore size constitutes a high surface concentration risk scenario.This neural network model demonstrates excellent accuracy，providing theoretical guidance for predicting hydrogen leak concentrations and the time to reach the lower explosive limit in engineering applications.

Keywords： buried pure hydrogen pipeline; leak difusion; leak location; pipeline burial depth;soil properties; neural network

0 引言

氢能被誉为21世纪最具发展潜力的清洁能源[1-3]。（剩余18294字）