Title:
異常検出方法
Document Type and Number:
Japanese Patent JP7433648
Kind Code:
B2
Abstract:
To provide an abnormality detection method for detecting an unknown abnormality of an object to be monitored.SOLUTION: An abnormality detection method decomposes, with frequency analysis, waveform data indicating vibration of an object to be monitored, calculates a state transition probability by applying a hidden Markov model to decomposed data obtained by decomposition with frequency analysis, compares the calculated state transition probability with a state transition probability at a normal condition calculated on the basis of the frequency analysis and the hidden Markov model from waveform data at the normal condition indicating vibration of the object to be monitored at the normal condition, and determines the abnormality of the object to be monitored on the basis of the comparison.SELECTED DRAWING: Figure 5
Inventors:
Setsuya Kurahashi
Isao Ono
Isao Ono
Application Number:
JP2020139284A
Publication Date:
February 20, 2024
Filing Date:
August 20, 2020
Export Citation:
Assignee:
National University Corporation University of Tsukuba
International Classes:
G01H17/00; G05B23/02; G01M99/00
Domestic Patent References:
| JP6160172A | ||||
| JP201072782A | ||||
| JP2005251185A |
Foreign References:
| CN111207926A |
Attorney, Agent or Firm:
Suzue International Patent Office