Method and system of monitoring and prognostics

• US 20070073518A1
• Filed: 09/26/2005
• Published: 03/29/2007
• Est. Priority Date: 09/26/2005
• Status: Active Grant

• Current Assignee: Lockheed Martin Corporation (Martin Marietta Corporation)
• Original Assignee: Lockheed Martin Corporation (Martin Marietta Corporation)
• Inventors: Turischev, Alexey, Harrison, Gregory
• Granted Patent: US 7,277,823 B2
• Time in Patent Office: Days
• US Class Current: 702/183
• CPC Class Codes: G05B 23/024 Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
• CPC Class Codes: G06N 3/02 Neural networks

Abstract

A neural network learns the operating modes of a system being monitored under normal operating conditions. Anomalies can be automatically detected and learned. A control command can be issued or an alert can be issued in response thereto.

Claims

1. An analysis system for monitoring an operating device comprising:

at least one sensor of a device related parameter;

at least one neural network for receipt and processing of representations of sensor output information, including automatically detecting device anomalies to generate selected output information, the neural network including an input port for receipt of time indicating indicia.

2. A system as in claim 1 which includes transforming circuitry coupled to the sensor for transforming sensor outputs to frequency spectrum information.

3. A system as in claim 2 where the transformed circuitry comprises a fast Fourier transform.

4. A system as in claim 1 which includes an expert system for automatically analyzing the output information.

5. A system as in claim 1 where the neural network comprises a programmed processor and a plurality of network implementing instructions.

6. A system as in claim 1 wherein parameters of the network are automatically adjusted in response to data associated with at least one detected anomaly.

7. A system as in claim 6 which includes control circuitry, responsive to at least one detected anomaly to produce an output control signal.

8. A system as in claim 1 which includes a plurality of neural networks.

9. A system as in claim 8 where at least some members of the plurality sense a common predetermined set of inputs.

10. A system as in claim 9 where the members of the plurality communicate with a common location.

11. A system as in claim 10 which includes circuitry at the common location that processes information received from at least the members of the plurality and determines the presence of an alarm condition.

12. A system as in claim 10 where the members of the plurality each include a wireless transmitter.

13. A system as in claim 10 where the members of the plurality are substantially identical.

14. A system as in claim 8 where at least some members of the plurality sense different inputs than do other members of the plurality.

15. A system as in claim 14 where outputs from at least some members of the plurality are received by common control circuitry.

16. A system as in claim 15 which includes a plurality of transducers coupled to the control circuitry for adjusting operational characteristics of an object.

17. A system as in claim 16 where the object is selected from a class which includes at least, a land vehicle, a waterborne vehicle and an airborne vehicle.

34. A system method as in claim 1 which includes at least one of an input port for information from a displaced computer network, or, an output port to transmit information to a displaced computer network.

35. A system as in claim 1 where the at least one neural network responds to a plurality of time based representations of sensor output information indicative of normal operation by creating stored internal representations thereof, and, where a device anomaly comprises at least one time based representation which deviates from the stored internal representation in accordance with at least one pre-established criterion.

18-24. -24. (canceled)

25. A method comprising:

sensing a plurality of conditions associated with operation of a selected system;

sensing periodic time based indicia;

generating representations of the sensed conditions;

processing the representations along with at least some of the time based indicia, with at least one neural network;

combining results of the processing, and responsive to the combining,

producing at least one of an alarm indicator, or, a control signal.

26. A method as in claim 25 which includes distributing the members of the plurality in a plurality in a geographical region.

27. A method as in claim 25 which includes establishing a common set of sensed conditions, and coupling corresponding representations to each member of the plurality.

28. A method as in claim 25 which includes coupling some representation to one member of the plurality and coupling other representations to a different member of the plurality.

29. A method as in claim 25 which includes training members of the plurality to establish an expected range of behavior as a function of time.

30. A method as in claim 29 where the members of the plurality are trained to establish a common expected range of behavior as a function of time.

31. A method as in claim 30 where the common expected range of behavior includes at least one of an atmospheric weather condition, an oceanic condition, or, a geological condition.

32. A method as in claim 29 where the range of behavior includes at least one of operating a land vehicle, operating a watercraft, operating an aircraft, monitoring a process, or monitoring operation of an apparatus.

33. A method as in claim 25 where processing includes comparing time evolution of sensed conditions with learned time evolution of operation of the selected system.