Based on relativistic multichannel theory, this paper calculates the energy levels of autoionization Rydberg series 4s^2nf（n=4- 23）J^π -（7/2）° of scandium at different levels of approximation within the fram...Based on relativistic multichannel theory, this paper calculates the energy levels of autoionization Rydberg series 4s^2nf（n=4- 23）J^π -（7/2）° of scandium at different levels of approximation within the framework of multichannel quantum defect theory. The present results show that the dipole polarizations play an important role. Considering the dynamical dipole polarization effects, this paper finds that the difference between calculated and experimental quantum defects for the 4s^2nf（n=4- 23）J^π -（7/2）° series is generally about 0.01- 0.03. Furthermore, the reason that 4s^216f is obscured in experimental spectra is suggested to be the interaction with the neighbouring resonance state converged to 3d^2（^1G4） of Sc^＋.展开更多
针对滚动轴承退化性能难以评估、寿命状态难以识别的问题,提出一种基于相空间欧式距离相关性(phase Euclidean distance cross-correlation,PEDCC)指标和多通道卷积长短时记忆网络(multichannel convolutional neural long short term m...针对滚动轴承退化性能难以评估、寿命状态难以识别的问题,提出一种基于相空间欧式距离相关性(phase Euclidean distance cross-correlation,PEDCC)指标和多通道卷积长短时记忆网络(multichannel convolutional neural long short term memory network,MCRNN)的状态识别方法。首先将正常轴承样本信号进行相空间重构,计算样本内重构后相邻数据之间的欧式距离,并将样本内的所有欧氏距离构成距离向量;然后利用互相关函数计算其余样本距离向量与正常样本距离向量之间的相关性,并将其作为轴承退化指标;最后利用所建立的PEDCC退化指标对轴承状态进行划分,将其输入到MCRNN网络中进行退化状态识别。其中MCRNN网络在不同通道中分别采取了不同卷积核,不同激活函数,以便于提取轴承振动信号的多尺度特征。通过轴承全寿命数据集对所提退化指标及网络模型的实用性进行验证,试验证明所提出的方法能更精确的实现轴承的退化状态识别。展开更多
基金Project supported by the National Natural Science Foundation of China (Grant No 10574162)partial supported from the Scientific Research Fund of GUCAS (Grant No 055101BM03)
文摘Based on relativistic multichannel theory, this paper calculates the energy levels of autoionization Rydberg series 4s^2nf（n=4- 23）J^π -（7/2）° of scandium at different levels of approximation within the framework of multichannel quantum defect theory. The present results show that the dipole polarizations play an important role. Considering the dynamical dipole polarization effects, this paper finds that the difference between calculated and experimental quantum defects for the 4s^2nf（n=4- 23）J^π -（7/2）° series is generally about 0.01- 0.03. Furthermore, the reason that 4s^216f is obscured in experimental spectra is suggested to be the interaction with the neighbouring resonance state converged to 3d^2（^1G4） of Sc^＋.
文摘针对滚动轴承退化性能难以评估、寿命状态难以识别的问题,提出一种基于相空间欧式距离相关性(phase Euclidean distance cross-correlation,PEDCC)指标和多通道卷积长短时记忆网络(multichannel convolutional neural long short term memory network,MCRNN)的状态识别方法。首先将正常轴承样本信号进行相空间重构,计算样本内重构后相邻数据之间的欧式距离,并将样本内的所有欧氏距离构成距离向量;然后利用互相关函数计算其余样本距离向量与正常样本距离向量之间的相关性,并将其作为轴承退化指标;最后利用所建立的PEDCC退化指标对轴承状态进行划分,将其输入到MCRNN网络中进行退化状态识别。其中MCRNN网络在不同通道中分别采取了不同卷积核,不同激活函数,以便于提取轴承振动信号的多尺度特征。通过轴承全寿命数据集对所提退化指标及网络模型的实用性进行验证,试验证明所提出的方法能更精确的实现轴承的退化状态识别。
