基于振动敏感时频特征的航天轴承寿命状态识别方法

发布时间：2017-12-31 12:08

本文关键词：基于振动敏感时频特征的航天轴承寿命状态识别方法　出处：《振动与冲击》2016年17期 　论文类型：期刊论文

【摘要】：针对非敏感特征削弱了航天轴承寿命状态特征集表征能力和识别率的问题,提出基于振动敏感时频特征的航天轴承寿命状态识别方法。设计出基于散布矩阵的寿命敏感性指标计算方法,根据该指标优选出使样本类内散度小、类间距大的敏感特征构建出寿命状态敏感特征集,增强对寿命状态的表征性。通过线性局部切空间排列算法(Linear Local Tangent Space Alignment,LLTSA)对寿命状态敏感特征集进行维数约简和特征融合,去除冗余信息,获得分类特性更好的低维寿命状态特征集,并输入最近邻分类器(K-Nearest Neighbors Classifier,KNNC)实现航天轴承不同寿命状态的识别。工程应用结果证明了所提方法的有效性和可行性。
[Abstract]:For the non sensitive feature weakens the space bearing life state feature set representation ability and recognition rate of the problem, put forward the space bearing life state recognition method based on time frequency characteristics of vibration sensitive. Design calculation method of sensitivity index based on life scatter matrix, according to the index were chosen to make small divergence within class, the characteristics of high sensitive to construct the life class distance sensitive feature set, enhance the characterization on life status. Through the linear local tangent space alignment algorithm (Linear Local Tangent Space Alignment, LLTSA) were fused to dimensionality reduction and the characteristic of the life state of sensitive feature set, remove redundant information, low dimensional life features obtained better classification feature set, and input to the nearest neighbor classifier (K-Nearest Neighbors Classifier, KNNC) identification of aerospace bearings of different life states. The application results prove the proposed The effectiveness and feasibility of the method.

【作者单位】：重庆交通大学机电与车辆工程学院;四川大学空天科学与工程学院;
【基金】：国家自然科学基金(51305471;51405048) 中国博士后科学基金(2014M560719) 重庆市基础与前沿研究计划(cstc2014jcyjA70009;cstc2015jcyjA70012) 重庆市教育委员会科学技术研究(KJ1400308) 国家留学基金资助项目(201408505081)
【分类号】：V467
【正文快照】： 航天轴承是航天飞行器中最关键的活动零部件之一。美、欧、俄、日等宇航大国和地区在宇航实践中,出现很多因航天轴承故障导致整个任务失败的事例[1-4],已充分认识到了提高航天轴承的性能、寿命和可靠性对于整个空间飞行器在轨任务完成的重要性,先后开始了航天轴承高可靠长寿命

【参考文献】