Investigation of heat treatment effects on microstructure and mechanical properties of AlCuCrFeNiTi high-entropy alloys employing hybrid model and feature-adaptive machine learning

This research investigates the microstructure and mechanical properties of AlCuCrFeNiTi high-entropy alloys (HEAs) synthesized by using vacuum arc melting. Microstructural analysis revealed a multiphase system comprising FCC, BCC, and Laves phases, with different heat treatments of 800 °C, 950 °C, and 1100 °C for 6 h, along with the annealing process, significantly influencing grain size and phase distribution. The untreated alloy exhibited an average grain size of 21.703μm, which increased to 56.393μm after heat treatment at 1100 °C. The highest hardness of 501.8 HV was achieved in the as-cast condition, attributed to solid solution strengthening and lattice distortion induced by Ti. Machine learning (ML) models, including the novel Feature-Adaptive KNN-SVR (FA-KNN-SVR) ensemble, were employed to predict thermodynamic and mechanical properties. Hardness and yield strength predictions achieved R2 values of 0.985 and 0.972, respectively, with RMSE values of 1.73 for hardness and 4.96 for yield strength. The Tree-Neural Ensemble Classifier (TNEC) achieved 93% accuracy in phase classification, outperforming traditional models like Random Forest (83% ), Support Vector Machine (77%) and k-Nearest Neighbors (74%).