Title:
METHOD FOR FAULT DIAGNOSIS USING A FAULT DIAGNOSIS MODEL TRAINED BY VIRTUAL DATA
Document Type and Number:
WIPO Patent Application WO/2022/036689
Kind Code:
A1
Abstract:
There is provided a method, an electronic device and computer readable storage medium for fault diagnosis. The method comprises: obtaining, based on a signal associated with a motion of a component of a gearbox(101), indications of motion characteristics of the component; detecting a component fault occurred in the component based on the indications using a fault diagnosis model(212), the fault diagnosis model(212) characterizing correlations between a plurality of indications of the motion characteristics of the component and related faults; and detecting a gearbox(101) fault occurred in the gearbox(101) based on the component fault. There is provided a method for fault diagnosis that can be implemented easily and reliably. A method of training a fault diagnosis model(212) based on a virtual signal(304) is provided as well. The virtual signal(304) is created by combining two existing signals such that healthy data and/or faulty data for training the fault diagnosis model(212) are expanded.
More Like This:
Inventors:
SUN HEQING (CN)
LIU ZHANCHI (CN)
LIU ZHANCHI (CN)
Application Number:
PCT/CN2020/110515
Publication Date:
February 24, 2022
Filing Date:
August 21, 2020
Export Citation:
Assignee:
ABB SCHWEIZ AG (CH)
SUN HEQING (CN)
LIU ZHANCHI (CN)
SUN HEQING (CN)
LIU ZHANCHI (CN)
International Classes:
G01M13/028; G01M13/021; G05B23/02
Domestic Patent References:
| WO2020154266A1 | 2020-07-30 |
Foreign References:
| CN103344430A | 2013-10-09 | |||
| CN111386503A | 2020-07-07 | |||
| CN102768115A | 2012-11-07 | |||
| CN102016736A | 2011-04-13 | |||
| CN110532626A | 2019-12-03 | |||
| US20190095781A1 | 2019-03-28 |
Attorney, Agent or Firm:
KING & WOOD MALLESONS (CN)
Download PDF: