Machine learning-based model for predicting performances of pyridines-quinolines as corrosion inhibitors on iron surfaces Available to Purchase

Given the substantial ramifications of material corrosion across diverse domains such as economics, the environment, society, industry, security, and safety, it stands as a significant challenge confronting both industrial and academic realms. Presently, the prevalent approach involves employing organic compounds to manage material degradation. Pyridines and quinolines stand out among the array of organic chemicals as corrosion inhibitors due to their distinct attributes: non-toxicity, cost-effectiveness, and efficacy across various corrosive environments. Considerable resources and dedicated efforts are essential for conducting extensive experimental research to generate a spectrum of potential chemical candidates for inhibitors. Within this study, we assessed the corrosion inhibition efficacy of 41 pyridine-quinoline derivatives when employed as inhibitors on iron surfaces, employing the Nu-SVR method as our machine learning model. 41 molecules in the dataset were divided into test and train sets, following a ratio of 70:30. Our findings revealed that the Nu-SVR model, boasting a Root Mean Square Error (RMSE) of 6.21%, exhibits superior predictive performance compared to the GA-ANN model (RMSE of 8.83%) utilized in previous research with the same dataset. Additionally, the Nu-SVR model showcased enhanced predictive performance, yielding an RMSE of 12.23% and a coefficient of determination (R²) value of 0.87, surpassing the GA-ANN model’s performance metrics of RMSE at 14.98% and R² at 0.80. To evaluate these models, we conducted testing using the entire dataset of 41 molecules. Through an examination of the essential characteristics, we comprehended the correlation between inhibitory efficacy and quantum chemical descriptors. In summary, our study presents novel insights into the predictive capabilities of machine learning models for forecasting corrosion inhibitors on iron surfaces.