基于PCA-BPNN模型的埋地管道腐蚀速率预测研究

Research on corrosion rate prediction of buried pipeline based on PCA-BPNN model

为了更加准确可靠地预测埋地管道的腐蚀速率,融合PCA分析法和多隐层BP人工神经网络模拟方法进行研究.选取陕西省某油气公司的埋地输油管道,构建8维度外腐蚀指标体系,在PCA-多隐层BPNN模型中模拟训练得到结果.通过PCA预处理将外腐蚀指标体系降为3维,以便减少多元素信息带来的耦合影响,模拟得到隐藏层参数最优的BPNN模型,预测腐蚀速率,求出预测值精确度,统计得到改进后方法精确度大于95%的个数是单一BP方法的2.5倍.为了检验PCA-多隐层BPNN方法的鲁棒性,另取20组数据代入验证,再次证实了PCA-多隐层BPNN模型所得的误差更小,更能满足实际工程需要.

In order to predict the corrosion rate of buried pipelines more accurately and reliably,the method integrating PCA analysis and BP artificial neural network simulation is studied.A buried oil pipeline of an oil and gas company in Shaanxi Province was selected to construct an 8-dimensional external corrosion index system,and the simulation training results were obtained in the PCA-multi-hidden layer BPNN model.The external corrosion index system was reduced to three dimensions by PCA pretreatment,so as to reduce the coupling influence brought by multi-element information.The BPNN model with the optimal hidden layer parameters was trained to predict the corrosion rate.The accuracy of the predicted value was obtained,and it was found that the number of accuracy greater than 95%of the improved method was 2.5 times that of the single BP method.In order to test the robustness of the PCA-multi-hidden layer BPNN method,an additional set of 20 groups of data are substituted for verification,reaffirming that the error of PCA-multi-hidden layer BPNN model is smaller and can better meet the needs of practical engineering.