Research on fault diagnosis method of canned motor pump based on random forests

For the typical faults that happened during the operation of canned motor pump in nuclear power system, the principle and implementation process of the diagnosis method based on random forests were studied in this paper, and the vibration signal data that characterize the faults of canned motor pump were obtained through the faults simulation test. On the basis of the above, first of all, the vibration features extraction and analysis were carried out, and then the faults diagnosis model for canned motor pump was established by using random forests algorithm to learn the part of the sample data. Finally, the diagnostic model was validated by using the remaining sample data, and the results showed that the fault diagnosis method based on random forests was good. (authors)