Method and system for knowledge discovery using non-linear statistical analysis and a 1st and 2nd tier computer program

• US 7,979,372 B2
• Filed: 03/26/2008
• Issued: 07/12/2011
• Est. Priority Date: 03/30/2007
• Status: Expired due to Fees

• Current Assignee: UT-Battelle LLC
• Original Assignee: UT-Battelle LLC
• Inventors: Hively, Lee M.
• Primary Examiner(s): VINCENT, DAVID ROBERT
• Application Number:
  US12/055,515
• Publication Number:
  US 20090150323A1
• Time in Patent Office: 1,203 Days
• Field of Search: 703/2, 702/34, 702/35, 702/41, 702/42, 702/56, 700/44
• US Class Current: 706/45
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation

Abstract

The invention relates to a method and apparatus for simultaneously processing different sources of test data into informational data and then processing different categories of informational data into knowledge-based data. The knowledge-based data can then be communicated between nodes in a system of multiple computers according to rules for a type of complex, hierarchical computer system modeled on a human brain.

Claims

1. A method for simultaneously processing different categories of source data into informational data, and then processing different groups of informational data into knowledge-based data in a first computer or in a plurality of computers, the method comprising:

acquiring a plurality of sets of source data within a first category of data for at least one test subject or process;

processing the sets of source data within the first category utilizing a first occurrence of a first-tier computer program executing on a first computer in the first computer system to produce a first group of informational data;

processing the sets of source data within a second category of data utilizing a second occurrence of a first-tier computer program executing on either the first computer or another computer in the first computer system to produce a second group of informational data;

wherein the processing of the sets of source data within the first category and within the second category includes a non-linear statistical analysis of the source data further comprising computing at least two distribution functions from connected phase space data, computing at least one measure of dissimilarity between at least two of distribution functions, and comparing said measure of dissimilarity to a threshold for a number of occurrences to indicate a condition change in said measure of dissimilarity to provide a forewarning of a condition change;

processing the first group and the second group of informational data with a second-tier computer program executing on either the first computer or another computer to produce knowledge-based data; and

transmitting at least one of the source data, the informational data and the knowledge-based data from the first computer to a second computer having other source data, informational data or knowledge-based data for evaluation with the data from the first computer.

2. The method of claim 1, wherein the knowledge-based data provides a perceptible indication to a human observer of a forewarning of at least one of:

a machine failure and a biomedical event.

3. The method of claim 1, wherein one category of source data relates to signals from a patient'"'"'s brain, another category of source data relates to signals from a patient'"'"'s heart, and a third category of source data relates to the patient'"'"'s lungs.

4. The method of claim 1, wherein the first-tier computer program includes a program sequence for evaluating a group of source data and determining a probability of an occurrence of an event representing information based on said source data, and wherein said first-tier computer program provides an output with the informational data to the second-level computer program.

5. The method of claim 1, wherein the evaluation by the second computer system of the knowledge-based data from the first computer is carried out according to rules for a complex, hierarchical system modeled on a human brain.

6. The method of claim 5, wherein the rules are at least one of propositional rules, first-order rules, decision trees, fuzzy logic, clustering, neural networking rules or Bayesion networking rules.

7. The method of claim 1, wherein knowledge-based data is transmitted from the first computer to a second computer for evaluation with other knowledge-based data in the second computer.

8. The method of claim 1, wherein the first-tier computer program and the second-tier computer program are software agent programs.

9. The method of claim 1, wherein the knowledge-based data provides a forewarning of gear failure resulting from analysis of torque data, vibration data, and motor power data.

10. The method of claim 1, wherein the knowledge-based data provides a forewarning of bearing failure from vibration data and motor power data.

11. The method of claim 1, wherein the knowledge-based data provides a measure of imbalance and misalignment in a motor-driven pump from electrical power data.

12. The method of claim 1, wherein the knowledge-based data includes a detection of motor faults from vibration data and electrical power data.

13. The method of claim 1, wherein the knowledge-based data includes a detection of a progressively larger crack in a rotating blade from vibration data and electrical power data.

14. The method of claim 1, wherein the knowledge-based data include a detection of progressively larger drill bit wear from spindle-motor current data.

15. The method of claim 1, wherein the knowledge-based data include a detection of different states of balance and unbalance for a centrifugal pump from motor power data.

16. The method of claim 1, wherein the knowledge-based data include a forewarning indication of structural failure due to cracking from time-serial stress data and strain data.

17. A knowledge-based computer system for simultaneously processing different categories of source data into informational data, and then processing different groups of informational data into knowledge-based data, the computer comprising:

a memory for storing a plurality of sets of source data within a first category of data for at least one test subject or process;

a first occurrence of a first-tier computer program executing on a first computer in the system for processing the sets of source data within the first category utilizing executing on a first computer in the system to produce a first group of informational data;

wherein the processing of the sets of source data includes a non-linear statistical analysis of the data further comprising computing at least two distribution functions from connected phase space source data, computing at least one measure of dissimilarity between at least two of distribution functions, and comparing said measure of dissimilarity to a threshold for a number of occurrences to indicate a condition change in said measure of dissimilarity to provide a forewarning of a condition change;

a second occurrence of a first-tier computer program executing on either the first computer or another computer in the computer system for processing the sets of source data within a second category to produce a second group of informational data;

a second-tier computer program executing on either the first computer or another computer for processing the first and second groups of information to produce knowledge-based data;

a network communication device for transmitting at least one of the source data, the informational data and knowledge-based data from the first computer to a second computer having other source data, informational data or knowledge-based data for evaluation with the data from the first computer.

18. The system of claim 17, wherein the knowledge-based data provides a perceptible indication to a human observer of a forewarning of at least one of:

a machine failure and a biomedical event.

19. The system of claim 17, wherein one category of source data relates to signals from a patient'"'"'s brain, another category of source data relates to signals from a patient'"'"'s heart, and a third category of source data relates to the patient'"'"'s lungs.

20. The system of claim 17, wherein the first-tier computer program includes a program sequence for evaluating a group of source data and determining a probability of an occurrence of an event representing information based on said source data, and wherein said first-tier computer program provides an output with the information to the second-level computer program.

21. The system of claim 17, wherein the evaluation by the second computer system of the knowledge-based data from the first computer system is carried out according to rules for complex, hierarchical system modeled on a human brain.

22. The system of claim 17, wherein the rules are at least one of propositional rules, first-order rules, decision trees, fuzzy logic, clustering, neural networking rules or Bayesion networking rules.

23. The system of claim 17, wherein the first-tier computer program and the second-tier computer program are software agent programs.