Systems and methods for selecting training data and generating fault models for use in use sensor-based monitoring

• US 7,035,763 B2
• Filed: 09/02/2004
• Issued: 04/25/2006
• Est. Priority Date: 09/03/2003
• Status: Expired due to Fees

• Current Assignee: Siemens Energy Incorporated (Siemens AG)
• Original Assignee: Siemens Westinghouse Power Corporation
• Inventors: Fang, Ming, Neubauer, Claus, Yuan, Chao, Cataltepe, Zehra, Brummel, Hans-Gerd
• Primary Examiner(s): Bui, Bryan
• Assistant Examiner(s): Le, Toan M.
• Application Number:
  US10/932,577
• Publication Number:
  US 20050049827A1
• Time in Patent Office: 600 Days
• Field of Search: 702/179, 702/185, 702/199, 706/907, 700/108
• US Class Current: 702/179
• CPC Class Codes: G05B 23/0254 based on a quantitative model, e.g. mathematical relationships between inputs and outputs; functions: observer, Kalman filter, residual calculation, Neural Networks
• CPC Class Codes: G05B 9/02 electric

Abstract

A system for generating a sensor model for use in sensor-based monitoring is provided. The system includes a segmenting module for segmenting a collection of sensor vectors into a plurality of bins comprising distinct sensor vectors. The system also includes a set-generating module for generating a set of statistically significant sensor vectors for each bin. The system further includes a consistency determination module for generating at least one consistent set of sensor vectors from the sets of statistically significant sensor vectors. Additionally, the system includes a model-generating module for generating a sensor model based upon the at least one consistent set.

Claims

1. A system for generating a sensor model for use in sensor-based monitoring, the system comprising:

a segmenting module for segmenting a collection of sensor vectors into a plurality of bins comprising distinct sensor vectors;

a set-generating module for generating a set of statistically significant sensor vectors for each bin;

a consistency determination module for generating at least one consistent set of sensor vectors from the sets of statistically significant sensor vectors; and

a model-generating module for generating a sensor model based upon the at least one consistent set.

2. The system as in claim 1, wherein the set-generating module generates a set of statistically significant sensors by determining, for each sensor vector in each bin, a likelihood that the sensor vector has a predefined probability distribution.

3. The system as in claim 2, wherein the likelihood is based upon a chi-square statistic.

4. The system as in claim 1, wherein the consistency determination module combines two sets of statistically significant vectors if the two sets are consistent.

5. The system as in claim 4, wherein the two sets are consistent if a squared difference between a first vector mean computed for the sensor vectors of one of the two sets and a second vector mean computed for the sensor vectors of the other of the two sets is less than a pre-selected threshold.

6. The system as in claim 1, wherein the model-generating module:

computes a minimum residual for at least one other consistent set using the sensor model, if the at least one consistent set comprises two or more consistent sets; and

,

if the at least one consistent set comprises two or more consistent sets, then

combines the at least one consistent set with at least one other consistent set and replaces the sensor model with a revised sensor model based upon the combination, if the minimum residual is less than a pre-selected residual threshold; and

constructs an additional sensor model based upon the at least one other consistent set, if the minimum residual is not less than the pre-selected threshold.

7. A system for generating statistically significant and consistent sets of training data usable in training a statistical model for purposes of sensor-based monitoring, the system comprising:

a segmenting module for segmenting a plurality of sensor vectors into at least two different bins, each bin containing distinct sensor vectors;

a set-generating module for

generating a first set by selecting and including in the first set at least one sensor vector from the first bin if the sensor vector from the first bin is statistically significant, and

generating a second set by selecting and including in the second set at least one sensor vector from the second bin if the sensor vector from the second bin is statistically significant; and

a consistency determination module for adding the second set to the first set if the second set is consistent with the first set.

8. The system as in claim 7, wherein the set-generating module determines if a sensor vector is statistically significant based upon a likelihood that the sensor vector has a predefined probability distribution.

9. The system as in claim 8, wherein the likelihood is based upon a chi-squared test.

10. The system as in claim 7, wherein the consistency determination module combines the second set with the first set only if a squared difference between a first vector mean based upon the sensor vectors of the first set and a second vector mean based upon the sensor vectors of the second set is less than a pre-selected threshold.

11. A method for generating a sensor model for use in sensor-based monitoring, the method comprising the steps of:

segmenting a collection of sensor vectors into a plurality of bins comprising distinct sensor vectors;

generating a set of statistically significant sensor vectors for each bin;

generating at least one consistent set of sensor vectors from the sets of statistically significant sensor vectors; and

generating a sensor model based upon the at least one consistent set.

12. The method as in claim 11, wherein the step of generating a set of statistically significant sensor vectors comprises determining, for each sensor vector in a bin, a likelihood that the sensor vector has a predefined probability distribution.

13. The method as in claim 12, wherein the likelihood is based upon a chi-square statistic.

14. The method as in claim 11, wherein the step of generating at least one consistent set comprises combining at least two sets of statistically significant vectors if one of the two sets is consistent with the other.

15. The method as in claim 14, wherein two sets are consistent if a squared difference between a first mean vector computed for the sensor vectors of one of the two sets and a second mean vector computed for the sensor vectors of the other of the two sets is less than a pre-selected threshold.

16. The method of claim 11, wherein the step of generating a sensor model based upon the at least one consistent set comprises:

computing a minimum residual for at least one other consistent set using the sensor model, if the at least one consistent set comprises two or more consistent sets; and

,

if the at least one consistent set comprises two or more consistent sets, then

combining the at least one consistent set with at least one other consistent set and replacing the sensor model with a revised sensor model based upon the combination, if the minimum residual is less than a pre-selected residual threshold, and

constructing an additional sensor model based upon the at least one other consistent set, if the minimum residual is not less than the pre-selected threshold.

17. A method of selecting training data usable in generating a statistical model for purposes of sensor-based monitoring, the method comprising the steps of:

segmenting a plurality of sensor vectors into at least two distinct bins, each bin containing distinct sensor vectors;

generating a first set by selecting and including in the first set a sensor vector from the first bin if the sensor vector from the first bin is statistically significant;

generating a second set by selecting and including in the second set a sensor vector from the second bin if the sensor vector from the second bin is statistically significant; and

combining the second set with the first set if the second training-data set is consistent with the first set.

18. The method as in claim 17, wherein the step of generating a training data set comprises determining whether a sensor vector is statistically significant based upon a likelihood that the sensor vector has a predefined probability distribution.

19. The method as in claim 18, wherein the likelihood is based upon a chi-square test of normality.

20. The method as in claim 17, wherein the step of combining comprises combining the second set with the first set only if a squared difference between a first vector mean based upon the first set and second vector mean based upon the second set is less than a pre-selected threshold.

21. A computer-readable storage medium for use in sensor-based monitoring, the storage medium comprising computer instructions for:

segmenting training data comprising a plurality of sensor vectors into at least two distinct bins, each bin containing distinct sensor vectors;

generating a first set by selecting and including in the first set a sensor vector from the first bin if the sensor vector from the first bin is statistically significant;

generating a second set by selecting and including in the second set a sensor vector from the second bin if the sensor vector from the second bin is statistically significant; and

forming a consistent set by combining the second training-data set with the first training-data set if the second training-data set is consistent with the first training-data set.

22. The computer-readable storage medium as in claim 21, further comprising a computer instruction for generating a sensor model based upon the consistent set.