Strength prediction and optimization for microwave sintering of large-dimension lithium hydride ceramics: GA-BP-ANN modeling

IF 2.3 2区物理与天体物理 Q1 NUCLEAR SCIENCE & TECHNOLOGY

Nuclear Materials and Energy Pub Date : 2024-11-01 DOI:10.1016/j.nme.2024.101801

Hongzhou Yan , Huayan Chen , Wenyan Zhang , Maobing Shuai , Bin Huang

{"title":"Strength prediction and optimization for microwave sintering of large-dimension lithium hydride ceramics: GA-BP-ANN modeling","authors":"Hongzhou Yan , Huayan Chen , Wenyan Zhang , Maobing Shuai , Bin Huang","doi":"10.1016/j.nme.2024.101801","DOIUrl":null,"url":null,"abstract":"<div><div>Failure typically occurs during sintering due to high thermal stress and poor strength of LiH ceramics. The short sintering time has shown to be beneficial in preventing excessive grain growth and improving ceramic performance. In this study, we built a genetic algorithm back propagation artificial neural network (GA-BP-ANN) model to predict the strength margins under different work conditions. Sensitivity analysis showed that the thickness and end control time were the most relevant parameters for strength margins, and the GA-BP-ANN model demonstrated the most efficient sintering work condition for a given thickness. Through statistical analysis of the strength margin predicted by the GA-BP-ANN model, we found that the bilinear temperature control method expanded the range of safe sintering conditions by 30% compared to the linear temperature control method. The research results of this study may serve as a reference for the safe and efficient sintering of LiH ceramics.</div></div>","PeriodicalId":56004,"journal":{"name":"Nuclear Materials and Energy","volume":"41 ","pages":"Article 101801"},"PeriodicalIF":2.3000,"publicationDate":"2024-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Nuclear Materials and Energy","FirstCategoryId":"101","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2352179124002242","RegionNum":2,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"NUCLEAR SCIENCE & TECHNOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Failure typically occurs during sintering due to high thermal stress and poor strength of LiH ceramics. The short sintering time has shown to be beneficial in preventing excessive grain growth and improving ceramic performance. In this study, we built a genetic algorithm back propagation artificial neural network (GA-BP-ANN) model to predict the strength margins under different work conditions. Sensitivity analysis showed that the thickness and end control time were the most relevant parameters for strength margins, and the GA-BP-ANN model demonstrated the most efficient sintering work condition for a given thickness. Through statistical analysis of the strength margin predicted by the GA-BP-ANN model, we found that the bilinear temperature control method expanded the range of safe sintering conditions by 30% compared to the linear temperature control method. The research results of this study may serve as a reference for the safe and efficient sintering of LiH ceramics.

查看原文本刊更多论文

大尺寸氢化锂陶瓷微波烧结的强度预测与优化：GA-BP-ANN 建模

由于锂辉石陶瓷的热应力高、强度差，通常会在烧结过程中发生失效。事实证明，缩短烧结时间有利于防止晶粒过度生长和提高陶瓷性能。在这项研究中，我们建立了一个遗传算法反向传播人工神经网络（GA-BP-ANN）模型来预测不同工作条件下的强度裕度。灵敏度分析表明，厚度和终端控制时间是与强度裕度最相关的参数，GA-BP-ANN 模型证明了给定厚度下最有效的烧结工作条件。通过对 GA-BP-ANN 模型预测的强度裕度进行统计分析，我们发现与线性温度控制方法相比，双线性温度控制方法将安全烧结条件的范围扩大了 30%。该研究成果可为锂辉石陶瓷的安全高效烧结提供参考。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Nuclear Materials and Energy Materials Science-Materials Science (miscellaneous)

CiteScore

3.70

自引率

15.40%

发文量

175

审稿时长

20 weeks

期刊介绍： The open-access journal Nuclear Materials and Energy is devoted to the growing field of research for material application in the production of nuclear energy. Nuclear Materials and Energy publishes original research articles of up to 6 pages in length.