基于BP神经网络的油气长输管道失效概率预测

油气长输管道风险评价是保障油气管道安全的重要技术之一，而油气长输管道风险评价的核心内容是定量确定管道的失效概率，因此，油气管道失效概率值的准确性会极大地影响到定量风险评价结果的合理性和适用性。目前，国内外均没有建立管道失效概率计算的数学模型，其预测多是依靠现代数学分析中的预测方法如：层次分析法、故障树分析方法和概率统计方法等。人工神经网络是以大量的具有相同结构的简单单元的连接来模拟人类大脑的结构和思维方式的一种可实现的物理系统或计算机模拟系统。本文将神经网络技术中常用的BP神经网络技术运用到油气长输管道失效概率预测中，依靠其强大的非线性映射关系，在输入、输出关系完全未知的情况下映射出输入、输出的非线性关系，从而建立基于故障树失效因素作为输入而管道失效概率作为输出的预测模型。

Long-Distance Oil/Gas Pipeline Failure Rate Prediction Using BP Artificial Neural Network Model

Risk assessment is one of most important technologies to ensure the safety of the long-distance oil/gas pipeline, and calculating the failure probability is the important part of the risk assessment, then the veracity of the failure probability impact the rationality and applicability of the result of the risk assessment. At present, there is no mathematical model of pipeline failure probability calculation both at home and abroad, their forecasts are mostly relied on modem mathematical analysis in forecasting methodologies such as： AHP, fault tree analysis method and statistical probability methods. Artificial neural networks are a lot of the simple structure with the same unit connected to simulate the human brain＇s structure and ways of thinking, which can be realized in a physical system or a computer simulation system. In this paper, BP neural network technology used in long-distance oil and gas pipeline failure probability forecast, under completely unknown circumstances, and rely on its powerful non-linear to map between the input and output non - linear relationship, so as to establish the forecast model., which based on the fault tree failure factors as input and pipeline failure probability as output.