[1]郭育硕,韩小蒙,舒诗湖.金属供水管道内壁腐蚀行为与多因素预测模型[J].中国给水排水,2025,41(11):24-30.
　GUOYu-shuo,HANXiao-meng,SHUShi-hu.Corrosion Behavior of Inner Wall of Metal Water Supply Pipes and Multi-factor Prediction Model[J].China Water & Wastewater,2025,41(11):24-30.
点击复制

金属供水管道内壁腐蚀行为与多因素预测模型

中国给水排水[ISSN:1000-4062/CN:12-1073/TU] 卷: 第41卷 期数: 2025年第11期 页码: 24-30 栏目: 出版日期: 2025-06-01
Title:
Corrosion Behavior of Inner Wall of Metal Water Supply Pipes and Multi-factor Prediction Model
作者:
郭育硕， 韩小蒙， 舒诗湖
（东华大学 环境科学与工程学院，上海 201620）
Author(s):
GUO　Yu-shuo， HAN　Xiao-meng， SHU　Shi-hu
（School of Environmental Science and Engineering， Donghua University， Shanghai 201620，China）
关键词:
金属供水管道;  微生物腐蚀;  电化学腐蚀;  多因素预测模型
Keywords:
metal water supply pipe;  microbiological corrosion;  electrochemical corrosion;  multi-factor prediction model
摘要:
金属供水管道在我国城镇供水管网中占有较大比例，针对灰口铸铁管道和碳钢管道进行静态烧杯实验和动态模拟实验，采用失重法和“图片+ImageJ软件”方法探究不同条件下金属试片的腐蚀程度。结果表明，动态模拟实验组与静态烧杯实验组的腐蚀速率存在显著差异；腐蚀速率均随温度、外加微生物量和电化学电位等因素发生变化。应用统计分析和建模技术，基于Python并利用实验数据建立线性回归、随机森林回归、决策树回归和支持向量机回归等多因素预测模型。其中，随机森林回归模型在多方面表现最佳，均方误差（MSE）为74.71，决定系数（R2）为0.85；决策树回归模型性能中等，MSE为85.62，R2为0.82，这些模型有助于准确预测腐蚀速率，使得预防和维护工作更具有针对性。
Abstract:
Metal water supply pipes constitute a significant proportion of urban water supply networks in China. Static beaker experiments and dynamic simulation experiments were carried out on gray cast iron pipes and carbon steel pipes. The corrosion degree of metal test specimens under various conditions was investigated using the weight loss method and the “image + ImageJ software” analysis approach. There was a marked disparity in the corrosion rate observed between the dynamic simulation experimental group and the static beaker experimental group. The corrosion rates varied depending on factors such as temperature, additional microbial quantity, and chemical potential. By leveraging statistical analysis and modeling techniques in conjunction with Python, multi?factor prediction models including linear regression, random forest regression, decision tree regression, and support vector machine regression were developed based on experimental data. Among the models, the random forest regression model demonstrated superior performance across all metrics, achieving a mean square error (MSE) of 74.71 and a coefficient of determination (R2) of 0.85. The decision tree regression model showed moderate performance, with a MSE of 85.62 and an R2?value of 0.82. These models help accurately predict the corrosion rate, making the prevention and maintenance work more targeted.

相似文献/References:

[1]刘沙沙,张卉,常珊,等.硝酸盐还原菌对给水铸铁管道的腐蚀特性研究[J].中国给水排水,2023,39(7):43.
　LIUSha-sha,ZHANGHui,CHANGShan,et al.Corrosion Characteristics of Nitrate-reducing Bacteria on Cast Iron Pipe for Water Supply[J].China Water & Wastewater,2023,39(11):43.
[2]吴霖璟,朱延平,韩小蒙,等.供水管网中金属管材的微生物腐蚀机理及防护研究[J].中国给水排水,2023,39(24):34.
　WULin-jing,ZHUYan-ping,HAN Xiao-meng,et al.Research on Microbial Influenced Corrosion Mechanism and Protection of Metal Pipe in Water Supply Pipe Network[J].China Water & Wastewater,2023,39(11):34.

更新日期/Last Update: 2025-06-01