Abstract: The evaluation for reliability of long-distance oil pipelines is urgent requirement for pipeline's owner.The traditional methods for evaluating the reliability of oil pipeline are based on the theory of reliability and supposal of failure rate,regarding the pipeline as a system of series connection or parallel connection.The failure rate is used to represent pipeline's reliability.The reliability calculated using statistic data of failure rates is unpractical for design,inspection and maintenance of pipeline's.A new concept of integrated reliability is put forward in consideration of more factors related with design, operation and maintenance of pipeline.A new model is programmed with artificial neural network.The new method for evaluating reliability is more practical than the traditional ones.The approach and steps to establish analysis model are given.

Key words: oil pipeline, reliability analysis, integrated reliability, neural network, modeling

摘要： 长输管道可靠性评估的传统方法是把管道结构视作一个串联或并联的系统,运用可靠性的理论概念和各种失效概率假设,以故障率的统计资料为计算依据,进行可靠度的计算分析,可靠度也采用故障率等参数来表达.该方法表达简单、含义通俗易懂、计算过程简便.但是,该可靠度的计算结果实用性不强,不便于将其结论直接用于管道设计、在线检测评估、管线维护等实际工作.从6个方面对此问题予以分析,在此基础上提出了构建综合可靠度的理论设想.综合可靠度,充分考虑了管道设计各个特征参数的影响以及它在管道设计、运行和维护中的变化规律,概念内涵较以前丰富、实用.还运用神经网络的基本原理和模型进行了分析建模,并提供了相关的分析思路和步骤.

关键词: 管道, 可靠度, 综合可靠度, 神经网络, 方法