Method and system for condition monitoring

• US 10,317,895 B2
• Filed: 07/24/2014
• Issued: 06/11/2019
• Est. Priority Date: 07/24/2013
• Status: Active Grant

• Current Assignee: Uniper Technologies Limited (E.ON SE)
• Original Assignee: Uniper Technologies Limited (E.ON SE)
• Inventors: Burridge-Oakland, Ty
• Primary Examiner(s): Tran, Dzung
• Application Number:
  US14/900,674
• Publication Number:
  US 20160147221A1
• Time in Patent Office: 1,783 Days
• Field of Search: 702 60, 702182, 702185, 702188, 702 34
• CPC Class Codes: G05B 23/0235 based on a comparison with predetermined threshold or range, e.g. "classical methods", carried out during normal operation; threshold adaptation or choice; when or how to compare with the threshold

Abstract

The method may include providing a plurality of historical data vectors, each historical data vector including a value for each of a plurality of empirical variables associated with an apparatus or process. The method may include providing a current data vector including a current value for each of the plurality of empirical variables. The method may include determining a plurality of distance values, each distance value being determined from the current data vector and a different one of the historical data vectors and identifying a minimum distance value from the determined plurality of distance values. The method may include using the identified minimum distance value to determine a condition of the apparatus or process.

Claims

1. A method for use in monitoring a condition of an apparatus comprising at least one of an industrial plant, an industrial line, machinery, equipment, a power plant, an apparatus configured for power generation, a turbine or a generator, or for use in monitoring a condition of a process comprising at least one of an industrial process, a process for manufacture of a product, a process for manufacture of a commodity, a process for generation of power or a process for generation of electricity, the method comprising:

using a plurality of sensors to measure a plurality of empirical variables associated with the apparatus or process and thereby provide a plurality of historical data vectors, each historical data vector of the plurality of historical data vectors comprising a scalar value for each empirical variable of the plurality of empirical variables;

using the plurality of sensors to measure the plurality of empirical variables associated with the apparatus or process and thereby provide a current data vector comprising a current scalar value for each empirical variable of the plurality of empirical variables;

determining a plurality of scalar distance values, each scalar distance value of the plurality of scalar distance values being determined from the current data vector and a different one of the plurality of historical data vectors;

identifying a minimum one of the determined plurality of scalar distance values;

identifying a historical data vector among the plurality of historical data vectors corresponding to the minimum one of the determined plurality of scalar distance values as a closest historical data vector;

determining a vector of scalar residual values from the current data vector and the closest historical data vector, wherein each scalar residual value among the vector of scalar residual values is determined by subtraction of a current scalar value of the current data vector corresponding to a particular empirical variable among the plurality of empirical variables and a scalar value of the closest historical data vector which corresponds to the particular empirical variable;

comparing each respective scalar residual values of the vector of scalar residual values to a corresponding scalar residual range among a plurality of scalar residual ranges;

raising or recording a corresponding alarm when the respective scalar residual value lies outside the corresponding scalar residual range; and

reconfiguring the apparatus or process depending on a result of the comparing.

2. The method according to claim 1, comprising sampling a sensor signal to provide a value of an empirical variable among the plurality of empirical variables.

3. The method according to claim 1, comprising repeatedly sampling a plurality of sensor signals to provide values of the plurality of empirical variables for each historical data vector or the current data vector.

4. The method according to claim 1, comprising deriving a value of specific empirical variable among the plurality of empirical variables from one or more measured variable values using a known relationship between the value of the specific empirical variable and the one or more measured variable values.

5. The method according to claim 1, comprising selecting the plurality of historical data vectors so that each different historical data vector among the plurality of historical data vectors corresponds to a different region of operation of the apparatus or process.

6. The method according to claim 1, comprising selecting the plurality of historical data vectors such that the plurality of historical data vectors corresponds to, or is representative of, every different region of operation of the apparatus or process.

7. The method according to claim 6, wherein the different regions of operation of the apparatus or process include a region of operation in which the apparatus or process is non-operational, non-productive, disabled or switched-off.

8. The method according to claim 1, comprising discarding or excluding historical data vectors among the plurality of historical data vectors which include one or more scalar values that are non-physical, abnormal, uncharacteristically high or uncharacteristically low.

9. The method according to claim 1, comprising using a distance function to determine each scalar distance value from the current data vector and the different one of the historical data vectors.

10. The method according to claim 9, wherein the distance function is a Euclidean distance function, a L²-norm distance function, or a L¹-norm distance function.

11. The method according to claim 1, comprising selecting each scalar residual range among the plurality of scalar residual ranges so as to correspond to normal, acceptable, safe or optimum operation of the apparatus or process.

12. The method according to claim 1, wherein the raising or recording raises or records the corresponding alarm when the respective scalar residual value remains outside the corresponding scalar residual range for a duration of an alert period.

13. The method according to claim 1, wherein the corresponding scalar residual range is associated with a positive or negative scalar residual alarm limit.

14. The method according to claim 1, comprising identifying a change in the condition of the apparatus or process according to a result of the comparing.

15. The method according to claim 14, comprising at least one of performing remedial work, repairing a fault, replacing one or more worn-out parts, scheduling preventative maintenance operations, or interrupting or shutting-down the operation of the apparatus or the process, depending on the identified change in the condition of the apparatus or process.

16. The method according to claim 1, comprising a historical data scaling step which comprises:

selecting first empirical variable among the plurality of empirical variables; and

scaling each of a first plurality of scalar values for the first empirical variable by a scaling constant, each of the plurality of historical data vectors including one of the first plurality of scalar values.

17. The method according to claim 16, comprising determining the scaling constant to be an inverse of a normalisation constant for the first empirical variable.

18. The method according to claim 17, comprising selecting the normalisation constant for the first empirical variable according to observed variations in the first empirical variable.

19. The method according to claim 17, comprising selecting the normalisation constant for the first empirical variable according to a function of at least one of a maximum variation, a minimum variation, a mean variation, a median variation or a mode variation of the first empirical variable.

20. The method according to claim 16, comprising repeating the historical data scaling step for each of the plurality of empirical variables to thereby provide a plurality of scaled historical data vectors.

21. The method according to claim 20, comprising a current data scaling step which comprises:

selecting a specific current scalar value for specific empirical variable among the plurality of empirical variables from the current data vector; and

scaling the specific current scalar value for the specific empirical variable selected from the current data vector by the same scaling constant used to perform the historical data scaling step for the specific empirical variable.

22. The method according to claim 21, comprising repeating the current data scaling step for each current scalar value of the current data vector to thereby provide a scaled current data vector.

23. The method according to claim 22, wherein the determining the plurality of scalar distance values determines the plurality of scalar distance values from the plurality of scaled historical data vectors and the scaled current data vector.

24. The method according to claim 23, wherein the identifying the historical data vector includes identifying a scaled historical data vector among the plurality of scaled historical data vectors corresponding to the minimum one of the determined plurality of scalar distance values as the closest historical data vector.

25. The method according to claim 24, wherein the determining the vector of scalar residual values determines the vector of scalar residual values according to a difference between the scaled current data vector and the closest historical data vector.

26. The method according to claim 25, comprising a residual value de-scaling step comprising dividing a particular scalar residual value among the vector of scalar residual values corresponding to a respective empirical variable among the plurality of empirical variables by a corresponding scaling constant used to perform the historical data scaling step for the respective empirical variable to thereby determine a de-scaled scalar residual value.

27. The method according to claim 26, comprising:

repeating the residual value de-scaling step for each scalar residual value among the vector of scalar residual values to thereby provide a vector of de-scaled scalar residual values,

wherein the comparing compares each respective de-scaled scalar residual value among the vector of de-scaled scalar residual values to a corresponding unsealed scalar residual range among a plurality of unsealed scalar residual ranges.

28. The method according to claim 25, comprising:

determining a scaled scalar residual range for each respective empirical variable among the plurality of empirical variables by scaling an unsealed scalar residual range for the respective empirical variable by the same scaling constant used to perform the historical data scaling step for the respective empirical variable to thereby provide a plurality of scaled scalar residual ranges; and

identifying a change in the monitored condition of the apparatus or process according to a result of the comparing,

wherein the comparing compares each respective scalar residual value of the vector of scalar residual values to a corresponding scaled scalar residual range among the plurality of scaled scalar residual ranges.

30. The method according to claim 1, comprising determining a condition of the apparatus or process according to a result of the comparing.

31. The method according to claim 1, comprising at least one of:

performing remedial work;

repairing a fault;

replacing one or more worn-out parts;

scheduling preventative maintenance operations;

or

interrupting or shutting-down the operation of the apparatus or the process,

depending on a result of the comparing.

29. A monitoring apparatus for use in monitoring a condition of a monitored apparatus or process, the monitored apparatus comprising at least one of an industrial plant, an industrial line, machinery, equipment, a power plant, an apparatus configured for power generation, a turbine or a generator, and the monitored process comprising at least one of an industrial process, a process for manufacture of a product, a process for manufacture of a commodity, a process for generation of power or a process for generation of electricity, and the monitoring apparatus comprising:

a plurality of sensors for measuring a plurality of empirical variables associated with the monitored apparatus or process to thereby provide a plurality of historical data vectors, each historical data vector of the plurality of historical data vectors comprising a scalar value for each empirical variable of the plurality of empirical variables, and to thereby provide a current data vector comprising a current scalar value for each empirical variable of the plurality of empirical variables;

at least one processor; and

a memory,

wherein the memory stores instructions which, when executed by the at least one processor, cause the monitoring apparatus to,

determine a plurality of scalar distance values, each scalar distance value of the plurality of scalar distance values being determined from the current data vector and a different one of the plurality of historical data vectors,

identify a minimum one of the determined plurality of scalar distance values,

identify a historical data vector among the plurality of historical data vectors corresponding to the minimum one of the determined plurality of scalar distance values as a closest historical data vector,

determine a vector of scalar residual values from the current data vector and the closest historical data vector, wherein each scalar residual value among the vector of scalar residual values is determined by subtraction of a current scalar value of the current data vector corresponding to a particular empirical variable among the plurality of empirical variables and a scalar value of the closest historical data vector which corresponds to the particular empirical variable,

compare each respective scalar residual values of the vector of scalar residual values to a corresponding scalar residual range among a plurality of scalar residual ranges,

raise or record a corresponding alarm when the respective scalar residual value lies outside the corresponding scalar residual range, and

reconfigure the monitored apparatus or process depending on a result of the comparison.