可靠性(半导体)
威布尔分布
人工神经网络
过程(计算)
机器学习
威布尔模量
计算机科学
脆性
可靠性工程
材料科学
机床
人工智能
工程类
统计
数学
复合材料
功率(物理)
物理
量子力学
冶金
操作系统
作者
Zehao Guan,Haixia Tian,Na Li,Jianzhan Long,Weibin Zhang,Yong Du
标识
DOI:10.1016/j.ceramint.2022.09.030
摘要
Materials reliability analysis is one of the most important substances in industrial manufacturing and practical application. Weibull statistics is a common-used approach to evaluate reliability, especially for brittle materials. However, such a process is limited by the insufficient number of samples and complex analysis steps. Herein, a machine learning-assisted strategy to analyze the reliability of the materials was proposed. The WC–Co-based cemented carbides were taken as the target materials. The machine learning models coupled feature engineering methods with advanced machine learning algorithms. Through an evaluation by designed experiments, the artificial neural network algorithm is determined to be the best machine learning algorithm to accurately capture the variation of property data to identify their distribution and automatically predict the Weibull modulus for reliability evaluation. This study provided a novel approach to evaluate the reliability accurately and shows the application potential to design the process parameters of other materials.
科研通智能强力驱动
Strongly Powered by AbleSci AI