Abstract: At present, most fault diagnosis methods of rotating machinery rely on extracting eigenvalue from the vibration waveform. The analysis on vibration signal is often based on a section of waveform collected at a specific measuring point in a certain instant. Thus the vibration waveform signal is the representation of the condition. If the fault occurs, under certain condition, it is possible that the fault information is not revealed in the vibration waveform, or submerged in other information. Under such circumstance, the fault diagnosis method based on the randomly extracted information of condition cannot well differentiate these faults. However, the fault, in any case, is accompanied by some symptoms. Under a condition, the vibration information of the fault may appear decentralized and random, but when the whole process is taken into consideration, the rules stand out. In order to describe the rule of changes in the vibration process, on the basis of the information entropy, a new identification index—the information exergy is defined. Through this concept, a new information exergy diagnosis method based on information fusion of the process is proposed.

Key words: Fault diagnosis, Information entropy, Information exergy, Information fusion, Process, Rotating machinery