In the past,only one performance parameter was considered in the reliability estimation of micro-electro-mechanical system (MEMS) accelerometers,resulting in a one-sided reliability evaluation. Aiming at the failure c...In the past,only one performance parameter was considered in the reliability estimation of micro-electro-mechanical system (MEMS) accelerometers,resulting in a one-sided reliability evaluation. Aiming at the failure condition of large range MEMS accelerometers in high temperature environment,the corresponding accelerated degradation test is designed. According to the degradation condition of zero bias and scale factor,multiple dependent reliability estimation of large range MEMS accelerometers is carried out. The results show that the multiple dependent reliability estimation of the large range MEMS accelerometers can improve the accuracy of the estimation and get more accurate results.展开更多
文摘工作在复杂环境下的多元退化设备面临失效数据少、多源信息融合准确度低和监督学习数据不平衡等问题,对此本文提出一种基于时间序列生成对抗网络(Time-series Generative Adversarial Networks,TimeGAN)与单分类支持向量机(One-Class Support Vector Machine,OCSVM)组合模型的小子样数据增广方法.方法引入了TimeGAN模型拟合真实数据时间序列相关性,从而生成新的多元退化设备数据.本文提出了一种基于最大均值差异改进方法的可信度判据,避免强相关特征对生成数据质量评价的影响,通过使用T-分布随机邻近嵌入(T-distributed Stochastic Neighbor Embedding,T-SNE)和全局最大均值差异(Global Maximum Mean Discrepancy,GMMD)的组合方法,定性定量地评价生成数据的质量水平.基于训练后的OCSVM模型,对生成数据进行异常检测与剔除,进一步提高生成数据的质量.以航空发动机数据集C-MAPSS为例进行方法验证分析,通过与其他数据增强模型对比验证了所提方法的可行性和有效性.
文摘In the past,only one performance parameter was considered in the reliability estimation of micro-electro-mechanical system (MEMS) accelerometers,resulting in a one-sided reliability evaluation. Aiming at the failure condition of large range MEMS accelerometers in high temperature environment,the corresponding accelerated degradation test is designed. According to the degradation condition of zero bias and scale factor,multiple dependent reliability estimation of large range MEMS accelerometers is carried out. The results show that the multiple dependent reliability estimation of the large range MEMS accelerometers can improve the accuracy of the estimation and get more accurate results.