Fe-C-Cr-Mn-Si steel plays a crucial role in the iron industry, and their components significantly influence microhardness and lifespan of equipment. A data-driven model combining machine learning (ML) and firefly optimization algorithm (FA) is proposed to predict components of Fe-C-Cr-Mn-Si steel. Conditional generative adversarial networks (CGANs) and solid solution strengthening theory are introduced to increase prediction accuracy with the limited data set. Ten common ML models were constructed to predict the microhardness of the steel. Three alloys were fabricated using cladding to validate the predict accuracy of the models. It is observed that the trained support vector regression (SVR) model demonstrated the highest precision in predicting microhardness. The coefficient of determination (R²) and root mean square error (RMSE) achieved 0.89 and 0.36 through the ten-fold cross-validation and Bayesian optimization method, respectively. The experimental validation revealed a maximum error of 2.09% between the predicted and experimental values. The investigation provides a valuable method to expedite design of Fe-C-Cr-Mn-Si steel with extreme high accuracy.

Fe-C-Cr-Mn-Si steel; Machine learning; Conditional generative adversarial networks; Solid solution strengthening; Firefly optimization algorithm