| 引用本文: | 葛承垄,朱元昌,邸彦强,等.装备剩余寿命预测平行仿真中模型动态演化方法.[J].国防科技大学学报,2019,41(5):118-127.[点击复制] |
| GE Chenglong,ZHU Yuanchang,DI Yanqiang,et al.Model dynamic evolution method of parallel simulation for equipment remaining useful life prediction[J].Journal of National University of Defense Technology,2019,41(5):118-127[点击复制] |
|
| |
|
|
| 本文已被：浏览 6731次 下载 5096次 |
| 装备剩余寿命预测平行仿真中模型动态演化方法 |
| 葛承垄, 朱元昌, 邸彦强, 崔浩浩 |
|
(陆军工程大学石家庄校区， 河北 石家庄 050003)
|
| 摘要: |
| 装备平行仿真中的一个重要概念是实时数据驱动的模型动态演化，但是至今仍缺乏具体应用领域的实现方法。以带未知离散冲击的混合退化装备剩余寿命预测为背景，以多态Wiener状态空间模型为演化对象，提出一种装备平行仿真中模型动态演化方法，包括基于交互多模型强跟踪滤波的模型软切换和基于期望最大化算法的模型参数在线估计，并实现了基于平行仿真的装备剩余寿命实时预测。利用某轴承退化数据进行实例研究，结果表明该方法能有效提高仿真逼真度，剩余寿命预测的准确度较高、不确定性较小，具有较高工程应用价值。 |
| 关键词: 平行仿真 模型演化 交互多模型 期望最大化 |
| DOI：10.11887/j.cn.201905017 |
| 投稿日期：2018-06-13 |
| 基金项目:国家部委基金资助项目(9140A04020115JB34011); 河北省自然科学基金资助项目(F2019506029) |
|
| Model dynamic evolution method of parallel simulation for equipment remaining useful life prediction |
| GE Chenglong, ZHU Yuanchang, DI Yanqiang, CUI Haohao |
|
(Shijiazhuang Campus, Army Engineering University, Shijiazhuang 050003, China)
|
| Abstract: |
| An important concept in equipment parallel simulation is the model evolution driven by real time data, but there is still a lack of model evolution method for specific application areas. Against the background of remaining useful life prediction for hybrid degradation equipment with unknown discrete shock, the polymorphic Wiener state space model was regarded as the evolutionary object and a model dynamic evolution method of equipment parallel simulation was put forward, including the interactive multiple model strong tracking filtering based model soft switch and the expectation maximum algorithm based online estimation of model parameters. Furthermore, the parallel simulation based real time prediction of equipment remaining useful life was realized. A case study was conducted by using a bearing degradation data. The results show that the method can effectively improve the simulation fidelity and the remaining useful life obtained by the proposed method has higher prediction accuracy and smaller uncertainty, implying a high practical engineering value. |
| Keywords: parallel simulation model evolution interactive multiple model expectation maximum |
|
|
