This article provides a method for using machine learning to predict the reliability and hardness of Al-Co-Cr-Cu-Fe-Ni system high-entropy alloys, and searches for several high-hardness and high-entropy alloys. This paper uses the most representative machine learning feature screening method to perform feature screening on 20 physical features that are highly related to the hardness and reliability prediction of Al-Co-Cr-Cu-Fe-Ni system high-entropy alloys to find the optimal feature combination and the time-consuming and accuracy of each method was evaluated and compared horizontally. In the end, it is found that genetic algorithm has the best effect on feature selection. Based on support vector regression, a highly accurate high-entropy alloy hardness and reliability prediction model is trained using the selected feature combination, and the prediction accuracy can reach 50.468672. Based on this model, this paper conducts corresponding hardness prediction and alloy search in the vast Al-Co-Cr-Cu-Fe-Ni system high-entropy alloy composition space, and successfully searches for several high-hardness high-entropy alloys with the highest specific composition. The predicted hardness value is 791.5322HV. It means high reliability for application.

Full text: IJEIR_2894_FINAL.pdf