基于PCA-OGM(1,N)模型的管道腐蚀深度预测  

Prediction of Pipeline Corrosion Depth Based on PCA-OGM(1,N)Model

作　　者：刘志 贾占功 丁晓勇 马江华 李世明 张雷 LIU Zhi;JIA Zhangong;DING Xiaoyong;MA Jianghua;LI Shiming;ZHANG Lei(N0.4 Drilling Engineering Company,BHDC)

机构地区：[1]渤海钻探第四钻井工程分公司,河北省任丘市华062552

出　　处：《油气田地面工程》2024年第11期31-35,49,共6页Oil-Gas Field Surface Engineering

基　　金：中国石油股份有限公司重大专项“油气管道及井筒流动保障完整性技术研究”(2022E-0205)。

摘　　要：为实现多因素协同作用下管道腐蚀趋势的有效预测,针对自变量序列的非线性特征,采用主成分分析(PCA)实现数据的聚集和降维,在分析GM(1,N)模型机理缺陷和参数缺陷的基础上,利用差分方程替换原白化方程推导近似时间响应式,同时对背景值进行优化,形成PCA-OGM(1,N)的组合模型用于管道腐蚀深度的预测,并进行多模型的验证和分析。PCA方法可以凸显自变量的特征,其中硫化氢含量、流速、含水率、Ca^(2+)含量、Mg^(2+)含量、温度、溶解氧含量等7个因素对腐蚀的影响较大;优化模型解决了强假设和参数错位的现象,OGM(1,N)模型的结构更为合理;与其余灰色和神经网络模型相比,PCA-OGM(1,N)模型的效果最优,其平均相对误差和拟合度分别为2.58%、0.9669,证明了此方法具有良好的适用性和先进性。In order to effectively predict the pipeline corrosion trend under the synergistic effect of multiple factors,according to the nonlinear characteristics of independent variable series,principal component analysis(PCA)is used to achieve data aggregation and dimension reduction.On the basis of analyzing mechanism defects and parameter defects of the traditional GM(1,N)model,the difference equation is used to replace the original whitening equation,and derive the approximate time response formula.At the same time,background values are optimized to form a combined model of PCAOGM(1,N)for the prediction of pipeline corrosion depth,and multiple models are verified and analyzed.The results show that the PCA method can highlight the characteristics of independent variables,among which seven factors such as hydrogen sulfide content,flow rate,water content,Ca^(2+)content,Mg^(2+)content,temperature,and dissolved oxygen have great influence on the corrosion.The structure of OGM(1,N)model is more reasonable,which solves the phenomenon of strong hypothesis and parameter misalignment.Compared with other grey and neural network models,the PCA-OGM(1,N)model has the best effect,its average relative error and fitting degree are 2.58%and 0.9669,respectively,which proves that the method has good applicability and advancement.

关 键 词：PCA OGM 管道 腐蚀深度预测 结构优化 背景值 

分 类 号：TE988.2[石油与天然气工程—石油机械设备]