Features calculation and combinatorial screening are necessary and tedious in predicting the hardness of high-entropy alloys by empirical parameters. To simplify the prediction process, we propose a strategy for predicting the hardness of high-entropy alloys based on the compositions to facilitate the rapid design of high-hardness high-entropy alloys. A random forest-based hardness prediction model was constructed using the alloy compositions as input. The Pearson correlation coefficients of the model reached 0.956 and 0.954 in the training and test sets, respectively. Subsequently, optimized alloys by inverse projection and high-throughput screening were verified using experiments. The hardness of Al_1.2Cr_17.42Fe_25.42Ni_28.32Ti_27.62 high-entropy alloy is up to 869.88 HV, which is 21.15% higher than the highest hardness in the original dataset of the Al-Cr-Fe-Ni-Ti system. The interpretability of the model was improved by introducing the Shapley additive explanation (SHAP). The results showed that the elements Al, Ti, Mo, Cr, and V play a positive role in enhancing the hardness of high-entropy alloys. Ni, Cu, Co, Mn, and Hf weaken the hardness of high-entropy alloys.

利用经验参数预测高熵合金的硬度需要进行特征计算和组合筛选，过程繁琐。为了简化预测过程，我们提出了一种基于成分预测高熵合金硬度的策略，以促进高硬度高熵合金的快速设计。使用合金成分作为输入构建了基于随机森林的硬度预测模型。该模型的皮尔逊相关系数在训练集和测试集分别达到了0.956和0.954。随后，使用实验验证了通过反投影和高通量筛选优化的合金。硬度 Al _1.2 Cr _17.42 Fe _25.42 Ni _28.32 Ti_27.62高熵合金高达869.88HV，比Al-Cr-Fe-Ni-Ti系原始数据集中的最高硬度高21.15%。通过引入 Shapley 附加解释 (SHAP)，提高了模型的可解释性。结果表明，Al、Ti、Mo、Cr、V元素对提高高熵合金的硬度起着积极作用。Ni、Cu、Co、Mn和Hf削弱了高熵合金的硬度。