化学工业与工程
 首页 |  在线投稿 |  期刊介绍 |  编 委 会 |  投稿指南 |  期刊订阅 |  下载中心 |  出版伦理 |  联系我们 |  English
化学工业与工程 2025, Vol. 42 Issue (2) :145-153    DOI: 10.13353/j.issn.1004.9533.20230124
化工模拟与计算 最新目录 | 下期目录 | 过刊浏览 | 高级检索 << | >>
基于虚拟样本生成技术的甲醇制芳烃产物预测
高暾1, 杨宸1, 于峰1, 张玮1, 张侃2
1. 太原理工大学化学工程与技术学院, 化学产品工程山西省重点实验室, 太原 030024;
2. 中国科学院山西煤炭化学研究所, 太原 030001
Prediction of methanol to aromatics products based on virtual sample generation technology
GAO Tun1, YANG Chen1, YU Feng1, ZHANG Wei1, ZHANG Kan2
1. Shanxi Key Laboratory of Chemical Product Engineering, College of Chemical Engineering and Technology, Taiyuan University of Technology, Taiyuan 030024, China;
2. Institute of Coal Chemistry, Chinese Academy of Sciences, Taiyuan 030001, China

摘要
参考文献
相关文章
Download: PDF (1845KB)   HTML ()   Export: BibTeX or EndNote (RIS)      Supporting Info
摘要 针对两段法甲醇制芳烃(MTA)工艺过程数据获取时间成本高、样本趋同、多样性差和数据信息间隔大等问题,提出一种基于虚拟样本生成技术结合遗传算法(GA)优化的极限学习机建模方法对该工艺重要产物苯、甲苯和二甲苯(BTX)的总收率进行预测。首先,以两段法甲醇制芳烃小试装置采集的数据样本为基础,利用多分布整体趋势扩散技术对模型输入数据进行扩充,再利用GA优化的极限学习机模型获取虚拟样本集合。对数据合理性进行检验后,利用原数据集合与虚拟数据集合融合建立BTX收率预测模型,采用MTA实验数据进行验证并与另外2种建模方法进行对比分析,结果表明,基于虚拟样本的建模方法拥有最优的精度表现,且该方法具有良好的稳定性,适用于甲醇制芳烃过程BTX收率预测。
Service
把本文推荐给朋友
加入我的书架
加入引用管理器
Email Alert
RSS
作者相关文章
高暾
杨宸
于峰
张玮
张侃
关键词甲醇制芳烃   产物含量预测   多分布整体趋势扩散技术   虚拟样本   极限学习机     
Abstract: Aiming at the difficulty of high data acquisition, high data repetition rate, small sample size, and large data information interval in two-stage methanol to aromatics (MTA) process, an extreme learning machine modeling method based on virtual sample generation technology and genetic algorithm (GA) optimization is proposed to predict the total yield of benzene, toluene and xylene (BTX), which are important products of the process. First, based on the data samples collected from the two-stage methanol to aromatics pilot plant, the input data for the model are expanded using the multi-distribution mega-trend-diffusion technology, and then the virtual sample set is obtained using the limit learning machine model optimized by GA. After verifying the rationality of the data, the BTX yield prediction model is established using the fusion of the original data set and the virtual data set. Through the verification of the MTA experimental data and the comparative analysis of the other two modeling methods, the results show that the modeling method based on virtual samples has the best precision performance, and the method has good stability, which is suitable for predicting BTX yield in methanol to aromatics process.
Keywordsmethanol to aromatics   product content prediction   multi-distribution mega-trend-diffusion   virtual sample   extreme learning machine     
Received 2023-03-08;
Fund:山西省重点研发计划(201903D121027);中科院关键技术人才项目(YB2021001);化学工程联合国家重点实验室开放课题(SKL-ChE-21A01);国家自然科学基金(22178241)。
Corresponding Authors: 张玮,教授,zhangwei01@tyut.edu.cn。     Email: zhangwei01@tyut.edu.cn
About author: 高暾(1997—),男,硕士研究生,现从事为化工过程建模及优化方面的研究。
引用本文:   
高暾, 杨宸, 于峰, 张玮, 张侃.基于虚拟样本生成技术的甲醇制芳烃产物预测[J].  化学工业与工程, 2025,42(2): 145-153
GAO Tun, YANG Chen, YU Feng, ZHANG Wei, ZHANG Kan.Prediction of methanol to aromatics products based on virtual sample generation technology[J].  Chemcial Industry and Engineering, 2025,42(2): 145-153
Copyright 2010 by 化学工业与工程