NC (numerical control) table is a complex dynamic system. By collecting vibration acceleration signals from NC table, the signals with SVD (singular value decomposition) method is analyzed, the singular spectrum is acquired and the singular entropy of the signals is calculated. The dynamic characteristics and their regulations of NC table are revealed via the characteristic quantities such as principal component, singular spectrum, singular entropy, etc. The results of SVD show that the first one or two principal components can be used to analyze the signal characteristics of the total time series in order to simplify characteristic extraction. The steep degrees of singular spectrums can be used to discriminate complex degrees of signals. The results show that the signals in direction of driving axes are the simplest and the signals in perpendicular direction are the most complex. The singular entropy values can be used to study the ind etermination of signals. The results show that the signals of NC table are not simple signal nor white noise, the entropy values in direction of driving axes are lower, the entropy values increase a little bit along with the increment of driving speed and the entropy values at the classic conditions, such as resonance or creep, decrease obviously.