Combined electrochemical, DFT/MD-simulation and hybrid machine learning based on ANN-ANFIS models for prediction of doxorubicin drug as corrosion inhibitor for mild steel in 0.5 M H2SO4 solution.

[Display omitted] • Doxorubicin drug acted as a mixed-type corrosion inhibitor. • Surface analysis via SEM and XPS evidenced the presence of inhibitor barrier film. • The best ML prediction was made using complete FCM-ANFIS model with high degree of accuracy. • The experimental results were supported by DFT/MDS. In this work, Doxorubicin drug was used as mild steel corrosion inhibitor in 0.5 M H 2 SO 4 solution. Herein, standard techniques like gravimetric, electrochemical measurement, density functional theory via DFT/ molecular dynamic simulation (MDS), Scanning electron microscopic (SEM), were used for proper evaluation of Doxorubicin drug as anticorrosive agent. According to the experimental findings, Doxorubicin drug significantly inhibits mild steel corrosion, and its effectiveness increases with an increase in the drug concentration. Maximum inhibition efficiency at 100 ppm was 93.3 %, 91.3 % and 96.8 % for electrochemical impedance (EIS), polarization test (PDP) and gravimetric techniques respectively. The electrochemical study indicates that Doxorubicin is a mixed-type inhibitor. Close scrutiny of the corroded and inhibited metals evidenced that Doxorubicin drug produced a better and more uniform coating on the surface of mild steel. The molecular structure of the drug and its contribution to the inhibition mechanism was further understood using simulations based on density functional theory (DFT) and molecular dynamics simulation (MDS). In addition, artificial neural networks (ANN) and adaptive neuro-fuzzy inference system (ANFIS) were used to predict and model the interactive effects affecting the response. Also, statistical parameters like coefficient of determination (R²), root mean square error (RMSE), mean absolute percentage error (MAPE), and mean absolute deviation (MAD) were used to assess the models' performance. The results demonstrate that the FCM-clustered ANFIS model with 15 clusters perform better than ANN model with RMSE, MAD, MAPE, and R²values of 0.978, 0.642, 4.823, and 0.9925 at the testing phase, and 0.6435, 0.4248, 2.8151, and 0.9998 at the training phase respectively. The best prediction was made using complete FCM-ANFIS model, which had accuracy of 98.4 %. Hence, a robust system prediction can be achieved via ANN and ANFIS algorithms to predict corrosion inhibition of mild steel in acidic environment using drug based corrosion inhibitors. [ABSTRACT FROM AUTHOR]