基于主成分分析法的惯性器件寿命预测  

Inertia Device Lifetime Prediction Based on Principal Component Analysis

作　　者：张会会[1] 张伟[2] 胡昌华[1] 周志杰[1] 

机构地区：[1]第二炮兵工程学院302教研室,陕西西安710025 [2]第二炮兵工程学院402教研室,陕西西安710025

出　　处：《系统仿真技术》2011年第4期249-253,260,共6页System Simulation Technology

基　　金：国家自然科学基金重点课题资助项目(60736026);国家自然科学青年基金资助项目(61004069)

摘　　要：在基于随机滤波理论的剩余寿命预测模型中,所使用的输入数据为单维,而工程实践中惯性器件有多维监测信息,因此,在现有寿命预测模型中,根据专家经验所选取的单维数据仅使用其中一部分信息。针对以上问题,提出了1种基于主成分分析的寿命预测方法。主成分分析可以将相互关联的多维数据用少数几个不相关的主成分进行代替,这样在减少有用信息损失的同时,使问题得到简化。应用实际的多维历史监测数据和主成分分析的寿命预测模型,进行了某导弹惯性平台的寿命预测实验。实验结果表明,采用基于主成分分析的预测模型能够有效地进行寿命预测,且精度较高。In the residual life prediction model based on stochastic filtering theory, the data is onedimensional, but in engineering multi - dimensional information can usually be obtained. Only some of data are selected by expert, that is, some of the useful information is missing. In order to solve this problem, a new principal component analysis （PCA） based method for residual life prediction is proposed. PCA can use a few of uncorrelated principal components to substitute original interdependent indices,so it can reduce the loss of useful information and make the model simple. The PCA based forecasting model and the basic model are used to forecast the inertia system＇ s residual life by using the really multi - dimensional monitoring data. The experimental results indicate the higher precision and validation of the algorithm.

关 键 词：寿命预测 随机滤波 主成分分析法 惯性器件 

分 类 号：TP202.1[自动化与计算机技术—检测技术与自动化装置]