Developing an atmospheric aging evaluation model of acrylic coatings: A semi-supervised machine learning algorithm

To study the atmospheric aging of acrylic coatings, a two-year aging exposure experiment was conducted in 13 representative climatic environments in China. An atmospheric aging evaluation model of acrylic coatings was developed based on aging data including 11 environmental factors from 567 cities. A hybrid method of random forest and Spearman correlation analysis was used to reduce the redundancy and multicollinearity of the data set by dimensionality reduction. A semi-supervised collaborative trained regression model was developed with the environmental factors as input and the low-frequency impedance modulus values of the electrochemical impedance spectra of acrylic coatings in 3.5wt% NaCl solution as output. The model improves accuracy compared to supervised learning algorithms model (support vector machines model). The model provides a new method for the rapid evaluation of the aging performance of acrylic coatings, and may also serve as a reference to evaluate the aging performance of other organic coatings.