Method and system for performing multi-objective predictive modeling, monitoring, and update for an asset

• US 20060247798A1
• Filed: 04/28/2005
• Published: 11/02/2006
• Est. Priority Date: 04/28/2005
• Status: Abandoned Application

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Yan, Weizhong, Schmid, James J., Bonissone, Piero P., Iyer, Naresh S., Eklund, Neil H., Knodle, Chad E., Subbu, Rajesh V., Shah, Rasiklal P.
• Application Number:
  US11/117,596
• Publication Number:
  US 20060247798A1
• Time in Patent Office: Days
• US Class Current: 700/44
• CPC Class Codes: G06N 3/126 Evolutionary algorithms, e.g. genetic algorithms or genetic programming

Abstract

A method and system for performing multi-objective predictive modeling, monitoring, and update for an asset is provided. The method includes determining a status of each of at least two predictive models for an asset as a result of monitoring predicted performance values. The status of each predictive model includes at least one of: acceptable performance values, validating model, and unacceptable performance values. Based upon the status of each predictive model, the method includes performing at least one of: terminating use of the at least two predictive models for the asset, generating an alert for the asset of the status of the at least two predictive models, and updating the at least two predictive models based upon the status of the at least two predictive models.

Claims

1. A method for performing multi-objective predictive modeling, monitoring, and update for an asset, comprising:

determining a status of each of at least two predictive models for an asset as a result of monitoring predicted performance values, the status of each predictive model including at least one of;

acceptable performance values;

validating model; and

unacceptable performance values; and

based upon the status of each predictive model, performing at least one of;

terminating use of the at least two predictive models for the asset;

generating an alert for the asset of the status of the at least two predictive models; and

updating the at least two predictive models based upon the status of the at least two predictive models.

2. The method of claim 1, wherein the acceptable performance values are determined by comparing the predicted performance values with actual performance values of each predictive model, wherein the predicted performance values are considered to be acceptable if they coincide with the actual performance values.

3. The method of claim 1, wherein the validating model status indicates that a validation process is ongoing for the predictive model being monitored.

4. The method of claim 1, wherein the monitoring of each predictive model is performed online.

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

providing a data set to each predictive model and performing predictive analysis on application of the data set to each predictive model; and

calculating an error resulting from the predictive analysis;

adding the data set to a training data set provided in a temporary storage location if storage space in the temporary storage location permits the adding, the temporary storage location being accessible to each predictive model; and

if the storage space does not permit the adding;

creating an other training data set by combining the data set with selected data points from a historical data set;

performing batch training on each predictive model using the other training data set resulting in an updated predictive model; and

deleting the data set from the temporary storage location.

6. The method of claim 5, wherein if results of the calculating an error exceed a specified threshold, the updating further includes:

performing incremental learning on each predictive model using the data set.

7. The method of claim 5, wherein the batch training is performed at fixed time intervals.

8. The method of claim 5, wherein the batch training is performed upon reaching a maximum capacity of the temporary storage location.

9. The method of claim 5, wherein the performing batch training includes at least one of cross-validation and model configuration optimization.

10. The method of claim 5, wherein a number of data points selected from the historical data set is a function of the number of data points stored in the temporary storage location.

11. A system for performing multi-objective predictive modeling, monitoring, and update for an asset, comprising:

at least two predictive models relating to an asset;

a monitoring module in communication with the at least two predictive models, the monitoring module performing;

monitoring predictive performance values for each predictive model and determining a status of each predictive model as a result of the monitoring, the status including at least one of;

acceptable performance values;

validating model; and

unacceptable performance values; and

based upon the status of each of the predictive models, performing at least one of;

terminating use of the at least two predictive models for the asset;

generating an alert for the asset of the status of the at least two predictive models; and

updating the at least two predictive models based upon the status of the at least two predictive models.

12. The system of claim 11, wherein the acceptable performance values are determined by comparing the predicted performance values with actual performance values of each predictive model, wherein the predictive performance values are considered to be acceptable if they coincide with the actual performance values.

13. The system of claim 11, wherein the validating model status indicates that a validation process is ongoing for each predictive model being monitored.

14. The system of claim 11, wherein the monitoring of each predictive model is performed online.

15. The system of claim 11, wherein the updating comprises:

providing a data set to each predictive model and performing predictive analysis on application of the data set to each predictive model; and

calculating an error resulting from the predictive analysis;

adding the data set to a training data set provided in a temporary storage location if storage space in the temporary storage location permits the adding, the temporary storage location being accessible to each predictive model; and

if the storage space does not permit the adding;

creating an other training data set by combining the data set with selected data points from an historical data set;

performing batch training on each predictive model using the other training data set resulting in an updated predictive model; and

deleting the data set from the temporary storage location.

16. The system of claim 15, wherein if results of the calculating an error exceed a specified threshold, the updating further includes:

performing incremental learning on each predictive model using the data set.

17. The system of claim 15, wherein the batch training is performed at fixed time intervals.

18. The system of claim 15, wherein the batch training is performed upon reaching a maximum capacity of the temporary storage location.

19. The system of claim 15, wherein the performing batch training includes at least one of cross-validation and model configuration optimization.

20. The system of claim 15, wherein a number of data points selected from the historical data set is a function of the number of data points stored in the temporary storage location.