文章信息/Info

Title:

Research Progress of Machine Learning Aided High Entropy Alloy Design

文章編號:

1674-3962(2021)07-0508-10

作者:

趙鼎祺; 喬珺威; 吳玉程

（太原理工大學(xué)材料科學(xué)與工程學(xué)院，山西 太原 030024）

Author(s):

ZHAO Dingqi; QIAO Junwei; WU Yucheng

(College of Materials Science and Engineering, Taiyuan University of Technology, Taiyuan 030024, China)

關(guān)鍵詞:

高熵合金; 復(fù)雜成分合金; 機(jī)器學(xué)習(xí); 人工神經(jīng)元網(wǎng)絡(luò); 人工智能

Keywords:

High entropy alloy;  Complex concentrated alloys;  Machine learning;  Artificial neural network;  Artificial intelligence

分類號:

TG139;TP181

DOI:

10.7502/j.issn.1674-3962.202011011

文獻(xiàn)標(biāo)志碼:

A

摘要:

近年來，高熵合金因其優(yōu)異的性能和廣闊的發(fā)展前景吸引了越來越多的關(guān)注，成為材料科學(xué)中的熱門領(lǐng)域。由于高熵合金復(fù)雜的元素組成，使用傳統(tǒng)方法對高熵合金進(jìn)行計(jì)算不僅困難而且代價高昂，影響因素的多樣性也為高熵合金的設(shè)計(jì)增加了困難，開發(fā)新方法加速對高熵合金成分空間的探索是當(dāng)務(wù)之急。隨著對高熵合金研究的不斷深入，實(shí)驗(yàn)數(shù)據(jù)不斷積累，人們嘗試從數(shù)據(jù)的角度尋求解決方案。與此同時，人工智能的興起極大改變了我們的生活方式，以數(shù)據(jù)為驅(qū)動的機(jī)器學(xué)習(xí)與高熵合金領(lǐng)域交叉融合，二者相得益彰并取得了一系列成果。人工神經(jīng)元網(wǎng)絡(luò)、支持向量機(jī)、主成分分析等方法被應(yīng)用于高熵合金的分析和預(yù)測。除此之外，機(jī)器學(xué)習(xí)還與從頭算和基于熱力學(xué)數(shù)據(jù)庫的方法相結(jié)合，在挖掘數(shù)據(jù)價值與指導(dǎo)實(shí)驗(yàn)設(shè)計(jì)方面展現(xiàn)出了優(yōu)勢。首先對材料科學(xué)中的機(jī)器學(xué)習(xí)和高熵合金兩個領(lǐng)域做了簡述，介紹了近年利用機(jī)器學(xué)習(xí)輔助高熵合金設(shè)計(jì)的典型研究成果。并對未來機(jī)器學(xué)習(xí)在高熵合金中的應(yīng)用提出一些展望與建議。

Abstract:

In recent years, high entropy alloys have increasingly attracted attention due to their excellent properties and broad development prospects, become a hot field in materials science. Due to the complex element composition of high entropy alloy, it is difficult and expensive to use the traditional methods to calculate. The diversity of the influencing factors also makes the design of high entropy alloy more difficult. It is urgent to develop new strategies to accelerate the exploration of high entropy alloy composition space. With the development of research on high entropy alloys and the accumulation of experimental data, researchers try to find solutions from data. At the same time, the rise of artificial intelligence has dramatically changed our life. Machine learning and high entropy alloy field cross each other, and a series of achievements have been achieved. Artificial neural networks, support vector machine, principal component analysis, and other methods have been applied to analyzing and predicting high entropy alloys. Besides, machine learning combined with ab initio and thermodynamic library-based methods, has shown advantages in mining data value and guiding experimental design. In this paper, machine learning and high entropy alloys in materials science are briefly introduced, and recent researches on high entropy alloy design aided by machine learning are reviewed. Some prospects and suggestions for the application of machine learning in high entropy alloys in the future are put forward.