北京航空航天大学学报 ›› 2018, Vol. 44 ›› Issue (4): 725-736.doi: 10.13700/j.bh.1001-5965.2017.0262

• 论文 • 上一篇    下一篇

面向装备RUL预测的平行仿真技术

葛承垄, 朱元昌, 邸彦强   

  1. 军械工程学院电子与光学工程系, 石家庄 050003
  • 收稿日期:2017-04-26 出版日期:2018-04-20 发布日期:2017-09-18
  • 通讯作者: 邸彦强 E-mail:1049084176@qq.com
  • 作者简介:葛承垄,男,博士研究生。主要研究方向:装备平行仿真及其应用;朱元昌,男,博士,教授。主要研究方向:系统仿真;邸彦强,男,博士,副教授。主要研究方向:系统仿真。
  • 基金资助:
    装备预研基金重点项目(9140A04020115JB34011)

Equipment RUL prediction oriented parallel simulation technology

GE Chenglong, ZHU Yuanchang, DI Yanqiang   

  1. Department of Electronic and Optics Engineering, Ordnance Engineering College, Shijiazhuang 050003, China
  • Received:2017-04-26 Online:2018-04-20 Published:2017-09-18
  • Supported by:
    Key Project of Equipment Pre-Research Foundation of China (9140A04020115 JB34011)

摘要: 装备平行仿真是系统建模与仿真领域的新兴仿真技术,已经成为研究热点。在装备维修保障领域中,分析了装备剩余寿命(RUL)预测存在的突出问题,即模型参数固定、不具备自适应演化能力,成为阻碍实现装备剩余寿命自适应预测的首要因素。结合装备平行仿真理论,在建模分析的基础上提出了面向装备剩余寿命预测的平行仿真框架,该框架以Wiener状态空间模型为基础仿真模型,在动态注入的装备退化观测数据驱动下,利用期望最大化(EM)算法在线更新模型参数,并利用卡尔曼滤波(KF)算法实现仿真输出数据与观测数据的同化(DA),从而实现仿真模型动态演化,使得仿真输出不断逼近装备真实退化状态,为准确预测剩余寿命提供高逼真度仿真模型和数据输出。以某轴承性能退化数据为数据驱动源,对该框架进行了验证,仿真结果表明平行仿真方法能准确仿真装备性能退化过程,在提高预测精度的基础上实现了装备剩余寿命的自适应预测,有力证明了平行仿真方法的可行性和有效性。

关键词: 平行仿真, 模型演化, 剩余寿命(RUL), 数据同化(DA), 参数估计

Abstract: As an emerging simulation technology in the field of system modeling & simulation, equipment parallel simulation has become research emphasis. In the field of equipment maintenance support, the outstanding problem of equipment remaining useful life(RUL) prediction is analyzed, i.e., the stable model parameters without self-evolution ability, which has become the primary factor that hinders adaptive prediction of equipment remaining useful life. Combined with parallel systems theory, equipment remaining useful life prediction oriented parallel simulation framework is proposed on the basis of modeling analysis and Wiener state space model is taken as the basic simulation model in the framework. Driven by the dynamic implanted equipment degradation observation data, the model parameters are updated online by using expectation maximum(EM) algorithm and the data assimilation (DA) between simulation outputs and observation data is executed by using Kalman filter(KF), so as to realize dynamic evolution of the simulation model. The simulation model evolution which makes the simulation outputs close to equipment real degradation state provides high fidelity model and data for equipment remaining useful life prediction accurately. The framework is verified by the performance degradation data of a bearing. The simulation results show that the parallel simulation method can accurately simulate the equipment performance degradation process and the adaptive prediction of equipment remaining useful life is realized on the basis of the improved prediction accuracy, proving the feasibility and effectiveness of parallel simulation method.

Key words: parallel simulation, model evolution, remaining useful life(RUL), data assimilation(DA), parameter estimation

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发