基于BP神经网络的油气长输管道土壤腐蚀性预测

Soil Corrosion Prediction of Oil and Gas Pipeline Based on BP Neural Network

基于BP神经网络原理,综合考虑六类土壤腐蚀指标(土壤电阻率、土壤含水量、氧化还原电位、氯离子含量、硫酸根离子含量和p H值),建立了一种土壤腐蚀速率的预测方法。基于这种方法,依据某油田的现场土壤数据,借助MATLAB神经网络工具箱建立了这一地区的土壤腐蚀性预测的BP神经网络模型。训练和预测结果表明:训练的腐蚀速率最大误差为-1.5%,预测的腐蚀速率最大误差为8%。由此可见,基于BP神经网络的土壤腐蚀性预测方法具有较好的预测精度,对油气管道的安全运行具有重要的意义。

Considering six soil erosion indexes of oilfields（soil resistivity,soil moisture,redox potential,chloride content,sulfate ion content and p H）,a kind of soil corrosion rate prediction method based on BP neural network theory was established.Based on a certain oil field soil data,training and prediction were performed by using MATLAB neural network toolbox.The results show that the maximum relative error of training is-1.5% and the maximum relative error of prediction is 8%.Therefore,the prediction method of soil corrosion of oil and gas pipeline based on BP neural network has high accuracy,which has the vital significance to the safe operation of oil and gas pipelines.