Abstract: In the view of information fusion, the methods how to evaluate the vibration state of rotating machinery with single and multiple channel signals systematically is studied. The several spectrum entropy features in different domains are proposed. Multi-channel singular spectrum entropy reflects the indetermination degree of vibration energy distribution of spatially spreading multi-sensor signals in time domain. Whirling state feature entropy of multi-section rotor considers the complication of energy and state varying simultaneously in frequency domain. Wavelet space feature entropy characterizes complexity of feature patterns distribution in time-frequency domain. The analysis and calculation of the fault signals show that these information entropy are capable of classifying and indicating complication degree of faulty signals authentically, so they can form the characteristic index of effectively and synthetically evaluating the rotor vibrating state.

Key words: Information entropy, Information fusion, Rotating machinery, State evaluation