基于模糊神经网络的海洋管线腐蚀速率预测新方法

A Prediction Method Based on Fuzzy Neural Network for Corrosion Rate of Marine Pipelines

针对海洋环境中油气管线腐蚀速率预测的复杂问题,提出了灰色关联分析与模糊神经网络结合的新方法对管线腐蚀速率进行预测。首先使用灰色关联分析对管线腐蚀速率与环境因素进行关联度计算,优选关联度较高的若干参数,然后应用模糊神经网络寻找管线腐蚀速率与优选环境因素之间的映射关系,使得影响管线腐蚀的主要因素数量明显减少,降低了预测难度。还根据管线已有腐蚀速率统计数据对该方法进行了测试,结果表明,管线腐蚀速率预测的平均相对误差为5.96%,方法在减少环境因素数量的情况下仍具有良好的预测精度。因此,基于灰色关联分析与模糊神经网络的新方法能够根据环境因素快速准确地预测管线的腐蚀速率,对保障管线的安全运营具有指导意义。

An intelligent method based on fuzzy neural network and grey correlation analysis was proposed to predict the corrosion rate of oil and gas pipelines in marine environment. Through cor- relation analysis, the correlation between the corrosion rate of pipelines and environmental factors was built, and from which then the appropriate factors with high correlation could be picked out. Fi- nally, the mapping relationship between the corrosion rate of pipelines and environmental factors could be figured out through fuzzy neural network. The validity and reliability of the results predict- ed by the proposed method were tested with statistic data which involved different environmental factors, it follows that the average relative error of the predicted corrosion rate was 5.96%, in other words, the present method exhibited a good accuracy in prediction even that less environmental fac- tors were involved during the analysis process. Therefore, the method based on fuzzy neural net- work and grey correlation analysis, can predict the corrosion rate of pipelines rapidly and accurate- ly with the known environmental factors.