复杂装备运维系统数字孪生应用框架  

Digital twin application framework for complex equipment operation and maintenance system

作　　者：魏喆 张凯[1] 王忠凯 许铎 黄国田 WEI Zhe;ZHANG Kai;WANG Zhongkai;XU Duo;HUANG Guotian(School of Mechanical Engineering,Shenyang University of Technology,Shenyang 110870,Liaoning,China;Institute of Computing Technology,China Academy of Railway Sciences Corporation,Beijing 100081,China;Shenyang Saibao Technology Service Co.,Ltd.,Shenyang 110179,Liaoning,China;Genertec Shenyang Machine Tool Co.,Ltd.,Shenyang 110027,Liaoning,China)

机构地区：[1]沈阳工业大学机械工程学院,辽宁沈阳110870 [2]中国铁道研究院集团有限公司电子计算技术研究所,北京100081 [3]沈阳赛宝科技服务有限公司,辽宁沈阳110179 [4]通用技术集团沈阳机床有限责任公司,辽宁沈阳110027

出　　处：《沈阳工业大学学报》2024年第5期665-675,共11页Journal of Shenyang University of Technology

基　　金：辽宁省“揭榜挂帅”科技项目(2022020630-JH1/108);中国国家铁路集团有限公司科技研究开发计划项目(N2022J014)。

摘　　要：针对当前复杂装备运维系统急需实现智能化转型的问题,提出了一种与深度学习技术相结合的复杂装备运维系统数字孪生框架,建立了基于深度学习的复杂装备运维系统数字孪生理论和应用框架。在理论层面,结合深度学习构建了多层次的运维系统数字孪生框架;在应用层面,基于“知识-数据-模型”并行驱动的形式,构建了运维系统全生命周期的数字孪生应用框架;在数据角度,提出了一种理论框架与NST模型结合的应用形式,并通过实验进行了验证。实验结果表明,针对动车组的非稳态时间序列数据,NST模型具有更好的预测效果。A digital twin framework for complex equipment operation and maintenance systems,combined with deep learning technology,was proposed to address the urgent need for intelligent transformation in current complex equipment operation and maintenance systems.This article first established a digital twin theory and application framework for complex equipment operation and maintenance systems based on deep learning.At the theoretical level,a multi-level digital twin framework for operation and maintenance systems was constructed by combining deep learning.At the application level,a digital twin application framework for the entire life cycle of operation and maintenance system was constructed according to the parallel driving form of knowledge-data-model.From the data perspective,an application form combining the theoretical framework with the NST model was proposed and validated through experiments.The experimental results indicate that the NST model has better prediction performance for non-stationary time series data of high-speed trains.

关 键 词：数字孪生 深度学习 复杂装备 理论框架 应用框架 全生命周期 时间序列 智能化 

分 类 号：TH17[机械工程—机械制造及自动化]