US 11,853,047 B2
Sensor-agnostic mechanical machine fault identification
Ori Negri, Haifa (IL); Christopher Bethel, New York, NY (US); Daniel Barsky, Haifa (IL); Gal Ben-Haim, Haifa (IL); Gal Shaul, Haifa (IL); and Saar Yoskovitz, Haworth, NJ (US)
Assigned to AUGURY SYSTEMS LTD., Haifa (IL)
Appl. No. 17/639,795
Filed by AUGURY SYSTEMS LTD., Haifa (IL)
PCT Filed Sep. 3, 2020, PCT No. PCT/IL2020/050958
§ 371(c)(1), (2) Date Mar. 2, 2022,
PCT Pub. No. WO2021/044418, PCT Pub. Date Mar. 11, 2021.
Claims priority of provisional application 62/895,247, filed on Sep. 3, 2019.
Prior Publication US 2022/0334573 A1, Oct. 20, 2022
Int. Cl. G05B 23/02 (2006.01)
CPC G05B 23/024 (2013.01) [G05B 23/0281 (2013.01); G05B 23/0283 (2013.01)] 22 Claims
OG exemplary drawing
 
1. A method for identifying a fault of at least one mechanical machine, comprising:
causing a first plurality of sensors coupled to a corresponding first plurality of mechanical machines to acquire a first plurality of sets of signals emanating from said first plurality of mechanical machines, said first plurality of mechanical machines sharing at least one characteristic;
supplying at least said first plurality of sets of signals of said first plurality of mechanical machines to a pre-existing fault classifier previously trained to automatically identify faults of a second plurality of mechanical machines based on signals emanating therefrom and previously acquired by a second plurality of sensors, said second plurality of sensors being of a different type than said first plurality of sensors, said second plurality of mechanical machines sharing said at least one characteristic;
modifying said pre-existing fault classifier by employing transfer learning, based at least on said first plurality of sets of signals of said first plurality of mechanical machines, thereby providing a modified fault classifier, wherein said pre-existing fault classifier comprises a neural network including a data layer and an input layer for receiving data from said data layer, and said modifying said pre-existing fault classifier comprises adding at least one mapping layer to said neural network, said at least one mapping layer being added between said data layer and said input layer, whereby said at least one mapping layer is configured to receive said data from said data layer in said modified fault classifier;
applying said modified fault classifier to at least one additional set of signals acquired by at least one sensor of said first plurality of sensors and emanating from at least one given mechanical machine sharing said at least one characteristic, said modified fault classifier being configured to automatically identify at least one fault of said at least one given mechanical machine based on said at least one additional set of signals; and
providing a human sensible output, by an output device, including at least identification of said fault of said at least one given mechanical machine, at least one of a repair or maintenance operation being performed based on said human sensible output.