Research On The Information Fusion Of Typical Faults Of Rolling Bearing Based On Rough Set Theory

Although oil debris detection and vibration detection belong to two different technologies of testing, there are correlation and complementarity between them. The favorable result will be achieved in the fault diagnosis of the machine if the two methods are combined in a proper way. It has great significance of engineering and value of theory to study the coalition diagnosis based on the two technologies.To take advantages of the vibration detection and debris detection, some researchers use the two methods synthetically according to the specific condition and get some achievements when making condition monitoring and fault diagnosis to a machine. However, the two methods are subtle to disturbances from the environment and abnormal states of the machines, what's more, their sensitivities are different. So it is difficult to make researches on the information from oil debris detection and vibration detection as well as the coalition diagnosis.To implement the coalition detection of rolling bearing, an integrative test-bed of rolling bearing is developed in this dissertation. The oil and vibration signals can be collected synchronously; the experimental bearings can be replaced conveniently and different appointed faults of the rolling bearings can be simulated accurately on the test-bed. Experiments show that the elaborate test-bed can meet the need of the researches in this dissertation.Based on the analysis of rough set (R-S) theory and its application, the data from the oil debris monitoring and vibration signal generated from three failures of the rolling bearing GB306 is processed by means of fusion analysis, then the number of judging conditions is decrease to 4 from the initial number of 5. A method coalition detection of the rolling bearing based on the R-S theory is presented in this dissertation. The feasibility of the method is proved by examples. The results of the experiments show that the accuracy of this method rises to 87.5%, while the accuracy of oil detection is only 25% and the 50% for vibration method.The results of this dissertation solve to a certain extent the long-standing disputes between the two monitoring technologies. It will play an important role in harmonizing the development of two monitoring technologies, and provides a reasonable and practical technology route in the combined application of them.The expectant results are achieved in this dissertation. The relativity of information of different failures on roll bearing has some certain guidance and basis has been laid on the information combination in essence between oil debris detection and vibration detection of the failure of roll bearing.