Surveillance system and method having parameter estimation and operating mode partitioning

• US 6,609,036 B1
• Filed: 06/09/2000
• Issued: 08/19/2003
• Est. Priority Date: 06/09/2000
• Status: Active Grant

• Current Assignee: Intellectual Assets LLC
• Original Assignee: Intellectual Assets LLC
• Inventors: Bickford, Randall L.
• Primary Examiner(s): PATEL, RAMESH B
• Application Number:
  US09/591,140
• Time in Patent Office: 1,166 Days
• Field of Search: 700/28, 700/29, 700/30, 700/31, 700/32, 700/40, 700/44, 700/45, 700/46, 700/47, 700/48, 700/49, 700/50, 700/51, 701/1, 701/2, 701/3, 701/4, 701/11, 701/14, 701/23, 706/14-17
• US Class Current: 700/30
• CPC Class Codes: G05B 23/021 adopting a different treatment of each operating region or a different mode of the monitored system, e.g. transient modes; different operating configurations of monitored system
• CPC Class Codes: G05B 23/024 Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
• 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

Abstract

A system and method for monitoring an apparatus or process asset including partitioning an unpartitioned training data set into a plurality of training data subsets each having an operating mode associated thereto; creating a process model comprised of a plurality of process submodels each trained as a function of at least one of the training data subsets; acquiring a current set of observed signal data values from the asset; determining an operating mode of the asset for the current set of observed signal data values; selecting a process submodel from the process model as a function of the determined operating mode of the asset; calculating a current set of estimated signal data values from the selected process submodel for the determined operating mode; and outputting the calculated current set of estimated signal data values for providing asset surveillance and/or control.

Claims

1. A surveillance system for monitoring an asset, said system comprising in combination:

a) a data acquisition means for acquiring a current set of signals engendered from said asset correlative to asset status;

b) a digitizing means for digitizing said current set of signals for defining a current set of digitized signals;

c) a process model array comprised of an array of process submodels, each said process submodel trained with at least one of a plurality of training data subsets partitioned from a unpartitioned training data set for providing training of each submodel and wherein each of said plurality of training data subsets has at least one defined operating mode associated therewith;

d) an operating mode determination means for determining a current operating mode of said asset as a function of both said current set of digitized signals and said defined operating modes associated with said plurality of training data subsets;

e) a process model selection means for selecting at least one of said process submodels as a function of said determined current operating mode wherein said selected process submodel is representative of said determined current operating mode of said asset;

f) a parameter estimation means for producing a set of estimated signal values from said selected process submodel;

g) a fault detection means for detecting the presence of a fault as a function of said set of estimated signal values from said selected process submodel and said current set of digitized signals;

h) a communication means for outputting detected faults for providing asset surveillance.

2. The system of claim 1 wherein each of said plurality of process submodels are of a type individually selected from the group comprised of a multivariate state estimation technique model, a neural network model, a mathematical model, an autoregressive moving average model, and a Kalman filter model.

3. The system of claim 1 wherein said operating mode determination means is selected from the group comprised of a logic sequence method, a mathematical model method, a neural network method, and a expert system method.

4. The system of claim 1 wherein said parameter estimation means is different for producing individual estimated signal values in said set of estimated signal values.

5. The system of claim 1 wherein said parameter estimation means for producing said set of estimated signal values from said selected operating mode specific model is of a type individually selected from the group comprised of a multivariate state estimation technique method, a neural network method, a mathematical model method, an autoregressive moving average method, and a Kalman filter method.

6. The system of claim 1 wherein said fault detection means is of a type individually selected from the group comprised of a threshold limit test, a statistical hypothesis test, a sequential probability ratio test, and a conditional probability test.

7. A method for training a process model array using historical data, the steps including:

acquiring historical operating data of an asset;

generating observation vectors from said historical operating data such that each observation vector represents an operating state of said asset;

forming a training data set comprised of said observation vectors;

partitioning said training data set into a plurality of training data subsets;

associating at least one of a plurality of predefined operating modes to each of said plurality of training data subsets;

determining a plurality of operating modes;

iteratively creating a process submodel for at least one of said plurality of determined operating modes to be included in said process model array by determining which determined operating mode is enabled and extracting from said partitioned training data set at least one of said plurality of training data subsets having an associated predefined operating mode substantially matching said enabled determined operating mode and training said process submodel with said training data subset; and

storing each created process submodel as a new element in said process model array for subsequent use in surveillance of said asset.

8. The method of claim 7 wherein said operating mode determination means is selected from the group comprised of a logic sequence method, a mathematical model method, a neural network method, and a expert system method.

9. The method of claim 7 wherein said process model training means for training a operating mode specific process model is of a type individually selected from the group comprised of a multivariate state estimation technique method, a neural network method, a mathematical model method, an autoregressive moving average method, and a Kalman filter method.

10. The method of claim 7 wherein said plurality of process submodels are stored as elements in an array of model elements indexed by operating mode type.

11. A method for performing surveillance of an asset, the steps including:

acquiring historical operating data of an asset;

generating observation vectors from said historical operating data such that each observation vector represents an operating state of the asset;

forming a training data set comprised of said observation vectors;

partitioning said training data set into a plurality of training data subsets each having at least one of a plurality of predefined operating modes associated therewith by partitioning said observation vectors into said plurality of training data subsets such that each of said observation vectors is associated with at least one of said plurality of predefined operating modes;

determining a plurality of operating modes;

iteratively creating a process submodel for at least one of said plurality of determined operating modes to be included in said process model array by determining which determined operating mode is enabled and extracting from said partitioned training data set at least one of said plurality of training data subsets having an associated predefined operating mode substantially matching said enabled determined operating mode and training said process submodel with said training data subset; and

storing each created process submodel as a new element in said process model array for subsequent use in surveillance of the asset;

acquiring a current set of observed signal data values from the asset;

determining an operating mode of the asset for the current set of observed signal data values as a function of both said current set of observed signal data values and said predefined operating modes associated with said plurality of training data subsets;

selecting at least one said process submodel from said process model array such that said selected process submodel has an associated operating mode substantially matching said determined operating mode of the asset for the current set of observed signal data values;

calculating a current set of estimated signal data values from said selected process submodel for said determined operating mode;

comparing the calculated current set of estimated signal data values with the current set of observed signal data values for detecting a fault condition;

outputting a signal correlative to each detected fault condition for providing asset surveillance.

12. A system for controlling an asset, said system comprising in combination:

a) a data acquisition means for acquiring a current set of signals engendered from the asset correlative to asset status;

b) a digitizing means for digitizing said current set of signals for defining a current set of digitized signals;

c) a process model array comprised of an array of process submodels, each said process submodel trained with at least one of a plurality of training data subsets partitioned from an unpartitioned training data set for providing training of each submodel and wherein each of said plurality of training data subsets has at least one defined operating mode associated therewith;

d) an operating mode determination means for determining a current operating mode of the asset as a function of both said current set of digitized signals and said defined operating modes associated with said plurality of training data subsets;

e) a process model selection means for selecting at least one of said plurality of process submodels as a function of said determined current operating mode wherein said selected process submodel is representative of said determined current operating mode of the asset;

f) a parameter estimation means for producing a set of estimated signal values from said selected process submodel;

g) a communication means for communicating the estimated signal values to a control means for using one or more of the estimated signal values for providing asset control.

13. The system of claim 12 wherein each of said plurality of process submodels are of a type individually selected from a group comprised of a multivariate state estimation technique model, a neural network model, a mathematical model, an autoregressive moving average model, and a Kalman filter model.

14. The system of claim 12 wherein said operating mode determination means is selected from a group comprised of a logic sequence method, a mathematical model method, a neural network method, and a expert system method.

15. The system of claim 12 wherein said parameter estimation means for producing said set of estimated signal values from said selected operating mode specific model is of a type individually selected from a group comprised of a multivariate state estimation technique method, a neural network method, a mathematical model method, an autoregressive moving average method, and a Kalman filter method.

16. A method for controlling an asset, the steps including:

providing a training data set partitioned into a plurality of training data subsets having a plurality of predefined operating modes associated therewith such that each of said plurality of training data subsets has at least one of said predefined operating modes associated thereto;

acquiring a current set of observed signal data values from the asset;

determining an operating mode of the asset for the current set of observed signal data values by comparing the current set of observed signal data values to said plurality of training data subsets;

selecting a process submodel as a function of said determined operating mode of the asset;

calculating a current set of estimated signal data values from said selected process submodel for said determined operating mode;

outputting the calculated current set of estimated signal data values for providing asset control.

17. A method for monitoring an asset, the steps including:

a) acquiring a current set of signals engendered from said asset correlative to asset status;

b) digitizing said current set of signals for defining a current set of digitized signals;

c) providing a process model array comprised of an array of process submodels, each said process submodel trained with at least one of a plurality of training data subsets partitioned from a unpartitioned training data set for providing training of each submodel and wherein each of said plurality of training data subsets has at least one defined operating mode associated therewith;

d) determining a current operating mode of said asset as a function of both said current set of digitized signals and said defined operating modes associated with said plurality of training data subsets;

e) selecting at least one of said process submodels as a function of said determined current operating mode wherein said selected process submodel is representative of said determined current operating mode of said asset;

f) producing a set of estimated signal values from said selected process submodel;

g) detecting the presence of a fault as a function of said set of estimated signal values from said selected process submodel and said current set of digitized signals;

h) communicating detected faults for providing asset surveillance.

18. A system for training a process model array using historical data, said system comprising in combination:

means for acquiring historical operating data of an asset;

means for generating observation vectors from said historical operating data such that each observation vector represents an operating state of said asset;

means for forming a training data set comprised of said observation vectors;

means for partitioning said training data set into a plurality of training data subsets;

means for associating at least one of a plurality of predefined operating modes to each of said plurality of training data subsets;

means for determining a plurality of operating modes for defining a plurality of determined operating modes to be included in a process model array;

means for iteratively creating a process submodel for at least one of said plurality of determined operating modes to be included in said process model array by determining which determined operating mode is enabled and extracting from said partitioned training data set at least one of said plurality of training data subsets having an associated predefined operating mode substantially matching said enabled determined operating mode and training said process submodel with said training data subset; and

means for storing each created process submodel as a new element in said process model array for subsequent use in surveillance of said asset.

19. A system for performing surveillance of an asset, said system comprising in combination:

means for acquiring historical operating data of an asset;

means for generating observation vectors from said historical operating data such that each observation vector represents an operating state of the asset;

means for forming a training data set comprised of said observation vectors;

means for partitioning said training data set into a plurality of training data subsets each having at least one of a plurality of predefined operating modes associated therewith by partitioning said observation vectors into said plurality of training data subsets such that each of said observation vectors is associated with at least one of said plurality of predefined operating modes;

means for determining a plurality of operating modes for defining a plurality of determined operating modes to be included in a process model array;

means for iteratively creating a process submodel for at least one of said plurality of determined operating modes to be included in said process model array by determining which determined operating mode is enabled and extracting from said partitioned training data set at least one of said plurality of training data subsets having an associated predefined operating mode substantially matching said enabled determined operating mode and training said process submodel with said training data subset; and

means for storing each created process submodel as a new element in said process model array for subsequent use in surveillance of the asset;

means for acquiring a current set of observed signal data values from the asset;

means for determining an operating mode of the asset for the current set of observed signal data values as a function of both said current set of observed signal data values and said predefined operating modes associated with said plurality of training data subsets;

means for selecting at least one said process submodel from said process model array such that said selected process submodel has an associated operating mode substantially matching said determined operating mode of the asset for the current set of observed signal data values;

means for calculating a current set of estimated signal data values from said selected process submodel for said determined operating mode;

means for comparing the calculated current set of estimated signal data values with the current set of observed signal data values for detecting a fault condition;

means for outputting a signal correlative to at least one detected fault condition for providing asset surveillance.

20. A method for controlling an asset, the steps including:

a) acquiring a current set of signals engendered from the asset correlative to asset status;

b) digitizing said current set of signals for defining a current set of digitized signals;

c) providing a process model array comprised of an array of process submodels, each said process submodel trained with at least one of a plurality of training data subsets partitioned from an unpartitioned training data set for providing training of each submodel and wherein each of said plurality of training data subsets has at least one defined operating mode associated therewith;

d) determining a current operating mode of the asset as a function of both said current set of digitized signals and said defined operating modes associated with said plurality of training data subsets;

e) selecting at least one of said plurality of process submodels as a function of said determined current operating mode wherein said selected process submodel is representative of said determined current operating mode of the asset;

f) producing a set of estimated signal values from said selected process submodel;

g) communicating the estimated signal values for providing asset control.

21. A system for controlling an asset, said system comprising in combination:

a training data set partitioned into a plurality of training data subsets having a plurality of predefined operating modes associated therewith such that each of said plurality of training data subsets includes at least one of said predefined operating modes associated thereto;

means for acquiring a current set of observed signal data values from the asset;

means for determining an operating mode of the asset for the current set of observed signal data values by comparing the current set of observed signal data values to said plurality of training data subsets and using at least one of said plurality of predefined operating modes associated with at least one of said training data subsets which is found to be correlative to the current set of observed signal data values based on said comparison as a determined operating mode;

means for selecting a process submodel as a function of said determined operating mode of the asset;

means for calculating a current set of estimated signal data values from said selected process submodel for said determined operating mode;

means for outputting the calculated current set of estimated signal data values for providing asset control.

22. A method for performing surveillance of an asset, the steps including:

partitioning an unpartitioned training data set into a plurality of training data subsets each having an operating mode associated thereto;

creating a process model comprised of a plurality of process submodels each trained as a function of at least one of the training data subsets;

acquiring a current set of observed signal data values from the asset;

determining an operating mode of the asset for the current set of observed signal data values;

selecting a process submodel from the process model as a function of the determined operating mode of the asset;

calculating a current set of estimated signal data values from the selected process submodel for the determined operating mode; and

outputting the calculated current set of estimated signal data values for providing asset surveillance.

23. A method for performing control of an asset, the steps including:

partitioning an unpartitioned training data set into a plurality of training data subsets each having an operating mode associated thereto;

creating a process model comprised of a plurality of process submodels each trained as a function of at least one of the training data subsets;

acquiring a current set of observed signal data values from the asset;

determining an operating mode of the asset for the current set of observed signal data values;

selecting a process submodel from the process model as a function of the determined operating mode of the asset;

calculating a current set of estimated signal data values from the selected process submodel for the determined operating mode; and

outputting the calculated current set of estimated signal data values for providing asset control.