Principal component analysis based fault classification

• US 8,041,539 B2
• Filed: 08/07/2008
• Issued: 10/18/2011
• Est. Priority Date: 12/31/2003
• Status: Expired due to Fees

• Current Assignee: Honeywell International Inc.
• Original Assignee: Honeywell International Inc.
• Inventors: Guralnik, Valerie, Foslien Graber, Wendy
• Primary Examiner(s): BHAT, ADITYA S
• Application Number:
  US12/187,975
• Publication Number:
  US 20080294374A1
• Time in Patent Office: 1,167 Days
• Field of Search: 702/179
• 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: G06F 18/232 Non-hierarchical techniques
• CPC Class Codes: G06F 18/2433 Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection

Abstract

Principle Component Analysis (PCA) is used to model a process, and clustering techniques are used to group excursions representative of events based on sensor residuals of the PCA model. The PCA model is trained on normal data, and then run on historical data that includes both normal data, and data that contains events. Bad actor data for the events is identified by excursions in Q (residual error) and T2 (unusual variance) statistics from the normal model, resulting in a temporal sequence of bad actor vectors. Clusters of bad actor patterns that resemble one another are formed and then associated with events.

Claims

1. A computer implemented method of identifying events in a process, the method comprising:

running, in a computer processor, a principal component analysis model on sensor data from the process;

calculating, using the computer processor, statistics related to the model;

determining, with the computer processor, if an event is occurring;

finding, using the computer processor, a nearest cluster of bad actors related to the event to identify the event;

differentiating, using the computer processor, a sensor that directly senses the event from a sensor that senses a residual of the event;

storing the found nearest cluster of bad actors in a computer storage device; and

further comprising for new bad actors;

identifying a sequence of cluster matches; and

correlating the sequence of cluster matches to known events; and

further comprising;

determining if a cluster needs to be split when new bad actors are added; and

splitting the cluster into two clusters using a goodness of fit algorithm.

2. The method of claim 1 wherein finding a nearest cluster of bad actors comprises comparing the bad actor vectors to known clusters in a library of clusters for bad actors.

3. The method of claim 1 and further comprising:

determining if a new event category is encountered; and

broadening limits for the sequence of clusters.

4. The method of claim 1 and further comprising using a feature scoring scheme to identify top contributors of bad actors.

5. The method of claim 4 wherein the feature scoring scheme is based on rank, value, and percent of contribution to a Q-residual sensor to identify a relative importance.

6. The method of claim 5, wherein the top-contributors are determined based on a majority percentage of the Q-residual.

7. The method of claim 5, where the top-contributors are determined based on only the contributors with absolute values that are drastically different from values of other contributors.

8. A system for identifying events in a process, the system comprising:

means for running a principal component analysis model on sensor data from the process;

means for calculating statistics related to the model;

means for determining if an event is occurring;

means for finding a nearest cluster of bad actors related to the event to identify the event;

means for differentiating a sensor that directly senses the event from a sensor that senses a residual of the event;

means for storing the found nearest cluster of bad actors in a storage device;

and further comprising;

means for identifying a sequence of cluster matches; and

means for correlating the sequence of cluster matches to known events;

and further comprising;

means for determining if a cluster needs to be split when new bad actor(s) are added; and

means for splitting the cluster into two clusters using a goodness of fit algorithm.

9. The system of claim 8 wherein the means for finding a nearest cluster of bad actors comprises means for comparing the bad actor vectors to known clusters in a library of clusters for new bad actors.

10. The system of claim 8 and further comprising:

means for determining if a new event category is encountered; and

means for broadening limits for the sequence of clusters.

11. The system of claim 8 and further comprising means for feature scoring to identify top contributors of bad actors in a cluster.

12. The system of claim 11 wherein the means for feature scoring is based on rank, value, and percent of contribution to a Q-residual sensor to identify a relative importance.

13. The system of claim 12, wherein the top-contributors are determined based on a majority percentage of the Q-residual or the top-contributors are determined based on only the contributors with absolute values that are drastically different from values of other contributors.

14. A computer implemented method of identifying events in a process, the method comprising:

running, in a computer processor, a principal component analysis model on sensor data representative of multiple process parameters in the process;

calculating, using the computer processor, statistics related to the model;

determining, using the computer processor, if an event is occurring in the process; and

finding, using the computer processor, a nearest cluster of bad actors related to the event to identify the event, wherein an event consists of one or more process parameters being out of a normal range in one or more parts of the process;

differentiating, using the computer processor, a sensor that directly senses the event from a sensor that senses a residual of the event;

storing the found nearest cluster of bad actors in a computer storage device; and

further comprising for new bad actors;

identifying a sequence of cluster matches; and

correlating the sequence of cluster matches to known events; and

further comprising;

determining if a cluster needs to be split when new bad actors are added; and

splitting the cluster into two clusters using a goodness of fit algorithm.

15. The method of claim 14 wherein multiple process parameters are out of normal range.

16. The method of claim 15 wherein one or more process parameters are measured by multiple sensors.