Industrial asset health model update

• US 10,534,361 B2
• Filed: 11/26/2013
• Issued: 01/14/2020
• Est. Priority Date: 06/10/2013
• Status: Active Grant

• Current Assignee: Hitachi Energy Limited (Hitachi, Ltd.)
• Original Assignee: ABB Schweiz AG (ABB Limited)
• Inventors: Smiley, Karen J., Zyglowicz, Steven Thomas, Mahate, Shakeel M., Hou, Chihhung
• Primary Examiner(s): Knox, Tyler W.
• Application Number:
  US14/090,154
• Publication Number:
  US 20140365271A1
• Time in Patent Office: 2,240 Days
• CPC Class Codes: G05B 23/0283 Predictive maintenance, e.g. involving the monitoring of a system and, based on the monitoring results, taking decisions on the maintenance schedule of the monitored system; Estimating remaining useful life [RUL] preventive maintenance, i.e. planning maintenance according to the available resources without monitoring the system G06Q10/06
• CPC Class Codes: G06Q 10/0635 Risk analysis of enterprise or organisation activities

Abstract

Among other things, one or more techniques and/or systems are provided for generating a health profile of an industrial asset based upon data pertaining to such an industrial asset. The health profile describes an expected condition of the industrial asset during a prediction period, a predicted cause of the expected condition, and/or a predicted impact of the expected condition. In some embodiments, the health profile is generated using a model that is configured to be periodically and/or intermittently updated. Further, in some embodiments, a diagnostic profile may be generated describing diagnostic actions that may be taken to improve predictions included within a health profile and/or to improve a confidence in one or more of those predictions.

Claims

1. A method, comprising:

performing a first diagnostic analysis of an industrial asset during a first time period to obtain first data;

generating, using a model, a health profile for the industrial asset based upon the first data, the health profile describing a predicted condition of the industrial asset during a second time period subsequent to the first time period;

performing a subsequent diagnostic analysis of the industrial asset during the second time period to obtain second data;

evaluating the health profile for the industrial asset based upon the second data obtained via the diagnostic analysis of the industrial asset to determine whether there is a discrepancy between an actual condition of the industrial asset during a second time period and the predicted condition;

responsive to a determination that there is a discrepancy between the actual condition and the predicted condition, obtaining operational record data indicative of operations that have been performed with the industrial asset prior to the prediction period and determining from the operational record data, whether a preemptive operation occurred that prevented the predicted condition from occurring;

determining, in response to a determination that the preemptive operation did occur, not to update the model to account for the discrepancy or updating, in response to a determination that the preemptive operation did not occur, the model to create an updated model; and

responsive to predicting that embedding a sensor configured to measure at least one of structural fatigue, vibrations or wall thickness will increase an accuracy of the generating the health profile by more than a threshold corresponding to a level of interference with operational performance of the industrial asset associated with the embedding of the sensor, facilitating an embedding of the sensor within the industrial asset based upon the updated model and generating an updated health profile for the industrial asset using the sensor.

2. The method of claim 1, wherein the updating the model comprises:

selecting a second model, different than the model, to use as the updated model.

3. The method of claim 2, wherein the industrial asset is sorted into a first subset of industrial assets based upon a first criterion associated with the model, and the method comprises:

re-sorting the industrial asset into a second subset of industrial assets based upon a second criterion associated with the second model.

4. The method of claim 1, comprising:

selecting the model, but not a second model, for generating the health profile based upon a safety criterion.

5. The method of claim 1, wherein the predicted condition is associated with structural fatigue, vibrations and wall thickness, the method comprising:

facilitating an embedding of a second sensor configured to measure vibrations; and

facilitating an embedding of a third sensor configured to measure a wall thickness.

6. The method of claim 1, wherein the sensor is configured to measure at least one of an internal temperature or an ambient air temperature associated with the industrial asset.

7. The method of claim 1, comprising:

responsive to predicting that selection of a second model for generating the health profile would result in more than a threshold number of industrial assets being reclassified from having a first level of health to having a second level of health and that the reclassification would increase at least one of a cost factor or a resource burden beyond a threshold, selecting the model, but not the second model, for generating the health profile.

8. The method of claim 1, wherein the generating the health profile is based upon a combination of two or more of structural fatigue data for the industrial asset, gas concentration data for the industrial asset or temperature data for the industrial asset.

9. The method of claim 1, comprising:

sorting the industrial asset into a subset of industrial assets associated with the model based upon at least one of a voltage class, an operating environment, a manufacturer, an output production, or a loading capacity of the industrial asset.

10. The method of claim 1, comprising:

randomly selecting the model from a set of available models to use for generating the health profile.

11. The method of claim 1, wherein the updating the model comprises:

obtaining a set of data from a plurality of industrial assets;

pooling the set of data, using a machine learning algorithm, to identify a trend; and

updating the model to create the updated model based upon the trend.

12. The method of claim 1, wherein the updating the model comprises:

obtaining a set of data from a plurality of industrial assets;

pooling the set of data, using a machine learning algorithm to identify a trend; and

selecting a second model, different than the model, to use as the updated model based upon the trend.

13. The method of claim 1, wherein the sensor is configured to measure a structural fatigue of the industrial asset.

14. The method of claim 1, comprising:

performing forecasting upon the first data to generate forecasted operational record data for the industrial asset; and

the generating comprising generating the health profile based upon the forecasted operational record data.

19. The method of claim 1, wherein generating the health profile is based upon a combination of two or more of structural fatigue data for the industrial asset, gas concentration for the industrial asset or temperature data for the industrial asset.

20. The method of claim 1 further comprising:

determining whether a probability of the predicted condition occurring satisfies a reference threshold and, in response to a determination that the probability satisfies the reference threshold, adding additional parameters to the model to reevaluate the probability of the predicted condition.

15. A system, comprising:

a processor; and

memory comprising processor-executable instructions that when executed by the processor cause performance of operations, the operations comprising;

performing a first diagnostic analysis of an industrial asset during a first time period to obtain first data;

generating, using a model, a health profile for an industrial asset based upon the first data, the health profile describing a predicted condition of the industrial asset during a second time period subsequent to the first time period;

performing a subsequent diagnostic analysis of the industrial asset during the second time period to obtain second data;

responsive to a determination that there is a discrepancy between the actual condition and the predicted condition, obtaining operational record data indicative of operations that have been performed with the industrial asset prior to the prediction period and determining, from the operational record data, whether a preemptive operation occurred that prevented the predicted condition from occurring;

determining, in response to a determination that the preemptive operation did occur, not to update the model to account for the discrepancy or updating, in response to a determination that the preemptive operation did not occur, the model to create an updated model; and

responsive to predicting that replacing a first sensor configured to measure at least one of structural fatigue, vibrations or wall thickness with a second sensor configured to measure at least one of structural fatigue, vibrations or wall thickness will increase an accuracy of the generating the health profile by more than a threshold, facilitating replacement of the first sensor with the second sensor.

16. The system of claim 15, wherein the industrial asset is sorted into a first subset of industrial assets based upon a first criterion associated with the model, and the operations comprise:

re-sorting the industrial asset into a second subset of industrial assets based upon a second criterion associated with a second model.

17. The system of claim 15, the operations comprising:

performing forecasting upon the first data to generate forecasted operational record data for the industrial asset; and

the generating comprising generating the health profile based upon the forecasted operational record data.

18. A non-transitory computer readable medium comprising processor-executable instructions that when executed cause performance of operations, the operations comprising:

performing a first diagnostic analysis of an industrial asset during a first time period to obtain first data;

generating, using a model, a health profile for the industrial asset based upon the first data, the health profile describing a predicted condition of the industrial asset during a second time period subsequent the first time period;

performing a subsequent diagnostic analysis of the industrial asset during the second time period to obtain second data;

responsive to a determination that there is a discrepancy between the actual condition and the predicted condition, obtaining operational record data indicative of operations that have been performed with the industrial asset prior to the prediction period and determining, from the operational record data, whether a preemptive operation occurred that prevented the predicted condition from occurring;

determining, in response to a determination that the preemptive operation did occur, not to update the model to account for the discrepancy or updating, in response to a determination that the preemptive operation did not occur, the model to create an updated model; and

responsive to predicting that embedding a sensor configured to measure at least one of structural fatigue, vibrations or wall thickness will increase an accuracy of the generating the health profile by more than a threshold corresponding to a level of interference with operational performance of the industrial asset associated with the embedding of the sensor, facilitating an embedding of the sensor within the industrial asset based upon the updated model.