基于极限学习机的海洋环境中3C钢腐蚀速度预测  

Prediction of 3C steel corrosion rate in the marine environment based on extreme learning machine

作　　者：靳文博[1,2] 田震 熊小伟 杨怡飞[1,2] 樊成洋 JIN Wen-bo;TIAN Zhen;XIONG Xiao-wei;YANG Yi-fei;FAN Cheng-yang(College of Petroleum Engineering,Xi’an Shiyou University,Xi'an 710065,China;Shaanxi Key Laboratory of Advanced Stimulation Technology for Oil&Gas Reservoirs,Xi'an 710065,China;OFFSHORE Oil Engineering Co.,Ltd.,Tianjin 300461,China;Oil Production Plant NO.3 of Changqing Oilfield Company,Yinchuan 750000,China)

机构地区：[1]西安石油大学石油工程学院,西安710065 [2]陕西省油气田特种增产技术重点实验室,西安710065 [3]海洋石油工程股份有限公司,天津300461 [4]长庆油田公司第三采油厂,银川750000

出　　处：《安全与环境学报》2022年第2期778-785,共8页Journal of Safety and Environment

基　　金：陕西省自然科学基础研究计划青年项目(2019JQ-811);陕西省教育厅专项科研计划项目(20JK0844)。

摘　　要：为了准确预测海洋环境下钢材的腐蚀速度,基于极限学习机的思想建立了腐蚀速度预测模型。将极限学习机的思想引入海洋环境3C钢腐蚀速度预测中,基于3C钢在不同海水环境因素下的腐蚀速率试验数据,建立了极限学习机模型并预测了不同海水环境下3C钢的腐蚀速度,分析了隐含层节点数对预测结果的影响,探讨了极限学习机模型与多元线性回归模型预测结果的差异。结果表明:极限学习机的预测结果与试验结果吻合程度较好,平均相对误差可控制在2%以内,其预测精度高于多元线性回归模型;隐含层节点数对预测结果具有较大影响,当隐含层节点数增加时,预测所得的平均相对误差出现了先降低后升高的变化趋势;极限学习机在学习速度和泛化能力方面具有较强的优势,可将其应用于海洋环境下3C钢腐蚀速度的预测中。To accurately predict the corrosion rate of steel in the marine environment,a prediction model of corrosion rate was established based on the extreme learning machine.Based on the modeling principle and steps of the extreme learning machine,the experimental data of corrosion rate of 3 C steel under different seawater environment factors was used.Seawater temperature,dissolved oxygen content,pH value,salinity,and redox potential were taken as input variables and corrosion rate was taken as the output variable.Then,the extreme learning machine model was established and the corrosion rates under different seawater environmental factors were predicted.Besides,we discussed the difference in prediction results under different hidden layer nodes and analyzed the accuracy of the extreme learning machine model and multiple linear regression model.The results show that it is feasible to predict the corrosion rate of 3 C steel under different seawater environment factors by using the extreme learning machine.The average relative error can be controlled within 2%.When the number of hidden layer nodes is different,the prediction results of the extreme learning machine model are quite different.When the number of hidden layer nodes is 8,the average relative error of the model is the lowest,while when the number of hidden layer nodes is 10,the average relative error of the model is the highest.With the increase of the number of hidden layer nodes,the prediction accuracy of the model first increases and then decreases.In other words,the increase in the number of hidden layer nodes does not mean the improvement of the prediction accuracy.Therefore,the number of hidden layer nodes should be reasonably selected in the application.The prediction accuracy of the multiple linear regression model is poor:the maximum relative error is 9.391%,the average relative error is 5.404%,while the average relative error of the extreme learning machine is only 1.928%.Therefore,the prediction accuracy of the extreme learning machine model is higher

关 键 词：安全工程 腐蚀速度 极限学习机 隐含层节点 预测精度 

分 类 号：X93[环境科学与工程—安全科学]