Extraction of important variables and mining of dependencies of atmospheric corrosion of carbon steel based on a comprehensive intelligent model

Abstract

Machine learning algorithms are widely used to predict the corrosion rate of materials in a specific environment. However, the interpretability of such black-box models is poor, which hinders their application in the field of material corrosion. Therefore, to increase algorithm transparency in practical applications, the causal relationship in the material corrosion phenomenon based on machine learning models needs to be further explored. To solve the aforementioned problems, this study analyzed the corrosion process of carbon steel in the atmosphere with many variables and complex mechanisms and proposed an important variable mining framework based on the comprehensive intelligent model. This framework can mine the important environmental variables that affect the early atmospheric corrosion of carbon steel and their influence on the corrosion galvanic current. This study collected the hour-level atmospheric corrosion data, including relative humidity, temperature, rainfall, and O3, SO2, NO2, PM2.5, and PM10 concentrations, of 45# carbon steel from five test sites in China using the atmospheric corrosion monitor of the China Meteorological Administration. To ensure the stability of the results, three machine learning models with different fitting strategies, namely, random forest, gradient boosted regression trees, and backpropagation neural network, are constructed. Then, the multimodel ensemble important variable selection (MEIVS) algorithm is used to quantify the importance of environmental variables and extract important environmental variables that severely affect the early atmospheric corrosion of carbon steel. Eventually, the partial dependence plot (PDP) of the environmental variables and corrosion galvanic current is drawn. Based on the simulation results, three significant conclusions are obtained: (1) Compared with Pearson’s and Spearman’s correlation coefficients, the important environmental variables mined using the MEIVS algorithm are more consistent with the prior law of early atmospheric corrosion of carbon steel. Relative humidity, temperature, and rainfall have the most significant impact on the early atmospheric corrosion of carbon steel, and O3 has a considerable influence on the early atmospheric corrosion of carbon steel in Sanya. Moreover, other pollutants in various regions have a weak impact on the early atmospheric corrosion of carbon steel. (2) PDP shows that, in most cases, the corrosion galvanic current of 45# carbon steel is negatively correlated with temperature and positively correlated with relative humidity. (3) PDP and MEIVS are well consistent. The simulation reveals that PDP corresponding to important environmental variables has a greater range of change, and the changing trend of PDP can reflect the influence of environmental variables on the corrosion galvanic current.

Keywords

• atmospheric corrosion
• carbon steel
• model integration
• important variable extraction
• partial dependence plot