Method and system for predictive diagnosing of system reliability problems and/or system failure in a physical system

• US 5,991,707 A
• Filed: 03/09/1998
• Issued: 11/23/1999
• Est. Priority Date: 03/09/1998
• Status: Expired due to Fees

• Current Assignee: Hydrotech Systems Incorporated
• Original Assignee: Hydrotech Systems Incorporated
• Inventors: Searles, Warren W., Searles, Brett W.
• Primary Examiner(s): Barlow, John
• Assistant Examiner(s): VO, HIEN XUAN
• Application Number:
  US09/036,711
• Time in Patent Office: 624 Days
• Field of Search: 702/179, 702/181, 702/182, 702/183, 702/185, 324/772, 324/770, 324/71.3, 395/183.02, 706/45
• US Class Current: 702/185
• CPC Class Codes: G06C 23/02 of main shaft
• CPC Class Codes: G06F 11/008 Reliability or availability analysis

Abstract

A method and system diagnoses system reliability problems and/or system failure in a physical system through the predictive identification of errant fluctuations in one or more operating parameters of said physical system. The values of at least one kind of operating parameter of physical system are determined. A data stream of the parameter values are divided into one or more sets, and the sets are associated into one or more series. Through a series of mathematical calculations, it is determined whether an N0, Nx or N1 condition exists. An N0 condition indicates that the values are probably within a good operating range. An Nx condition indicates the probability of approaching a point outside of a good operating range condition. An N1 condition indicates the entering of the range of possible system failure.

Claims

1. A method for the predictive diagnosing of system reliability problems and/or system failure in a physical system through the predictive identification of errant fluctuations in one or more operating parameters of said physical system, comprising:

determining the values of at least one kind of operating parameter of said physical system by direct or indirect measurements of the values of said operating parameters at a plurality of time intervals;

forming a data stream by arraying said parameter values in sequence;

dividing said data stream into one or more sets;

associating said sets into one or more series,

processing at least one of said sets by determining the w value of said set by Formula 2.1 of

space="preserve" listing-type="equation">w=(square root of n.sub.g)(M-m.sub.s)/(s.sub.g or s.sub.p)

wherein n_g is the number of said sets in said set'"'"'s series, m_s is the mean of said parameter values within said set, s_g is an at-large-group standard deviation, s_p is a parent-group standard deviation and M is a parent-group mean or a normalized mean, wherein said at-large-group standard deviation is determined by Formula 3 of

space="preserve" listing-type="equation">s.sub.g =square root of [(d.sub.gt.sup.2 -[d.sub.gc ].sup.2 /n.sub.g)/(n.sub.g -1)]

wherein d_gt² is the cumulative squares of the individual d_g values of said sets in said set'"'"'s series, wherein d_gc is the cumulative totals of the individual d_g values of said sets in said set'"'"'s series, wherein d_g is the difference between m_s and the mean of the m_s values of said sets in said set'"'"'s series, and wherein n_g is as defined above in Formula 2.1, and

determining for said set whether an N0, Nx or N1 condition exists, wherein an N0 condition exists if w is less than nx wherein nx is the w value consistent with x% probability of no significant difference between M and m_s on the w Table or via a computational equivalent of the w Table for n_Q number of parameter values, wherein an Nx condition exists if w is greater than nx but less than ny wherein ny is the w value consistent with y% probability of no significant difference between M and m_s on the w Table or via a computational equivalent of the w Table for n_Q number of parameter values, and wherein an N1 condition exists if w is greater than ny, wherein x and y are numbers and x is numerically greater than y and wherein n_Q number of parameter values is n_g number of parameter values when s_g is used in Formula 2.1 and wherein n_Q number of parameter values is greater than 120 parameter values when s_p is used in Formula 2.1, and

wherein said set'"'"'s series encompasses a section of said data stream comprising at least three sets each including at least one of said parameter values when s_g is used in Formula 2.1 and wherein said set'"'"'s series encompasses a section of said data stream comprising at least one set including at least one of said parameter values when s_p is used in Formula 2.1.

2. The method of claim 1 wherein x is 5 and y is 1.

3. The method of claim 1 wherein said physical system is a substantially electrical system, a substantially electromechanical system, a substantially mechanical system, a substantially chemical system, or a substantially bio-mechanical system.

4. A method for the predictive diagnosing of system reliability problems and/or system failure in a physical system through the predictive identification of errant fluctuations in one or more operating parameters of said physical system, comprising:

determining the values of at least one kind of operating parameter of said physical system by direct or indirect measurements of the values of said operating parameters at a plurality of time intervals;

forming a data stream by arraying said parameter values in sequence;

dividing said data stream into one or more sets;

associating said sets into one or more series,

processing at least one of said sets by determining the w value of said set by Formula 2 of

space="preserve" listing-type="equation">w=(square root of n.sub.g)(M-m.sub.g)/s.sub.g

wherein n_g is the number of said sets in said set'"'"'s series, m_s is the mean of said parameter values within said set, s_g is an at-large-group standard deviation, and M is a parent mean or a normalized mean, wherein said at-large-group standard deviation is determined by Formula 3 of

space="preserve" listing-type="equation">s.sub.g =square root of [(d.sub.gt.sup.2 -[d.sub.gc ].sup.2 /n.sub.g)/(n.sub.g -1)]

wherein d_gt² is the cumulative squares of the individual d_g values of said sets in said set'"'"'s series, wherein d_gc is the cumulative totals of the individual d_g values of said sets in said set'"'"'s series, wherein d_g is the difference between m_s and the mean of the m_s values of said sets in said set'"'"'s series, and wherein n_g is as defined above in Formula 2, and

determining for said set whether an N0, Nx or N1 condition exists, wherein an N0 condition exists if w is less than "nx" wherein "nx is the w value consistent with x% probability of no significant difference between M and m_s on the w Table or the computational equivalent of the w Table for n_g number of parameter values, wherein an Nx condition exists if w is greater than nx but less than ny wherein ny is the w value consistent with y% probability of no significant difference between M and m_s on the w Table for n_g number of parameter values and wherein an N1 condition exists if w is greater than ny, wherein x and y are numbers and x is numerically greater than y; and

wherein said set'"'"'s series encompasses a data stream of at least three values.

5. The method of claim 4 wherein x is 5 and y is 1.

6. The method of claim 4 wherein said physical system is a substantially electrical system, a substantially electromechanical system, a substantially mechanical system, a substantially chemical system, or a substantially bio-mechanical system.

7. The method of claim 4 wherein said operating parameter is an electrical performance or condition parameter.

8. The method of claim 4 wherein each of said sets encompasses a plurality of said parameter values.

9. The method of claim 4 wherein a plurality of sets that are associated into a plurality of series are processed.

10. The method of claim 4 wherein a plurality of sets that are associated into a plurality of cumulative series are processed.

11. The method of claim 4 wherein a plurality of sets that are associated into a plurality of series are processed, and wherein each of the series of said plurality of series contains the same number of sequential sets.

12. The method of claim 4 wherein said set contains at least three values.

13. The method of claim 4 wherein said set'"'"'s series contains a plurality of sets, and wherein each of the sets in said plurality of sets contains the same number of parameter values.

14. The method of claim 4 wherein M is a parent mean seen during normal operating conditions of said physical system.

15. The method of claim 4 further including signaling a warning when an Nx or an N1 condition is seen.

16. The method of claim 4 further including displaying the N0, Nx or N1 condition of at least one set.

17. The method of claim 4 further including a plurality of layers of measurements, a plurality of parent and/or normalized mean selections, a plurality of layers of set and/or series definitions, and/or a plurality of layers of condition display elections, and/or combinations thereof.

18. The method of claim 4 further including a superimposed method for warning if one or more of said parameter values exceeds a prescribed limit.

19. A system for the predictive diagnosing of system reliability problems and/or system failure in a physical system through the predictive identification of errant fluctuations in one or more operating parameters of said physical system, comprising:

a means for determining the values of at least one kind of operating parameter of said physical system by direct or indirect measurements of the values of said operating parameters at a plurality of time intervals;

a means for forming a data stream by arraying said parameter values in sequence;

a means for dividing said data stream into one or more sets each containing a plurality of said parameter values and an equal number of said parameter values;

a means for associating said sets into one or more series,

a means for processing a plurality of said sets separately by Formula 2.1 of

space="preserve" listing-type="equation">w=(square root of n.sub.g)(M-m.sub.s)/(s.sub.g or s.sub.p)

wherein n_g is the number of said sets in said set'"'"'s series, m_s is the mean of said parameter values within said set, s_g is an at-large-group standard deviation, s_p is a parent-group standard deviation and M is a parent-group mean or a normalized mean, wherein said at-large-group standard deviation is determined by Formula 3 of

space="preserve" listing-type="equation">s.sub.g -square root of [(d.sub.gt.sup.2 -[d.sub.gc ].sup.2 /n.sub.g)/(n.sub.g -1)]

wherein d_gt² is the cumulative squares of the individual d_g values of said sets in said set'"'"'s series, wherein d_gc is the cumulative totals of the individual d_g values of said sets in said set'"'"'s series, wherein d_g is the difference between m_s and the mean of the m_s values of said sets in said set'"'"'s series, and wherein n_g is as defined above in Formula 2.1, and

a means for determining for said set whether an N0, Nx or N1 condition exists, wherein an N0 condition exists if w is less than "nx" wherein "nx is the w value consistent with x% probability of no significant difference between M and m_s on the w Table or via a computational equivalent of the w Table for n_Q number of parameter values, wherein an Nx condition exists if w is greater than nx but less than ny wherein ny is the w value consistent with y% probability of no significant difference between M and m_s on the w Table or via a computational equivalent of the w Table for n_Q number of parameter values, and wherein an N1 condition exists if w is greater than ny, wherein x and y are numbers and x is numerically greater than y and wherein n_Q number of parameter values is n_g number of parameter values when s_g is used in Formula 2.1 and wherein n_Q number of parameter values is greater than 120 parameter values when s_p is used in Formula 2.1, and

wherein said set'"'"'s series encompasses a section of said data stream comprising at least three sets each including at least one of said parameter values when s_g is used in Formula 2.1 and wherein said set'"'"'s series encompasses a section of said data stream comprising at least one set including at least one of said parameter values when s_p is used in Formula 2.1.

20. The system of claim 19 wherein x is 5 and y is 1.

21. The system of claim 19 wherein said means for dividing said data stream into one or more sets, said means for associating said sets into one or more series, said means for processing a plurality of said sets separately by Formula 2.1, and/or said means for determining for each of said sets processed whether an N0, Nx or N1 condition exists comprise one or more computer software programs operating within sufficient computer hardware.

22. The system of claim 19 wherein said means for dividing said data stream into one or more sets, said means for associating said sets into one or more series, said means for processing a plurality of said sets separately by Formula 2.1, and/or said means for determining for each of said sets processed whether an N0, Nx or N1 condition exists comprise one or more computer software programs operating within sufficient computer hardware, and wherein said computer hardware further includes a means for displaying at least one N0, Nx or N1 condition determined for at least one of said sets.