Real-time statistical process monitoring system

• US 5,339,257 A
• Filed: 05/15/1991
• Issued: 08/16/1994
• Est. Priority Date: 05/15/1991
• Status: Expired due to Fees

• Current Assignee: Applied Technologies Associates, Inc.
• Original Assignee: Applied Technologies Associates, Inc.
• Inventors: Pearson, Thomas H., Layden, David J., Layden, John E.
• Primary Examiner(s): Ramirez, Ellis B.
• Application Number:
  US07/700,548
• Time in Patent Office: 1,189 Days
• Field of Search: 364/552, 364/554, 364/551.01, 364/469, 364/474.16, 364/DIG. 1, 364/DIG. 2, 364/150, 364/188, 364/550, 395/155, 395/157, 395/158, 340/715
• US Class Current: 702/84
• CPC Class Codes: G05B 19/4184 characterised by fault tolerance, reliability of production system
• CPC Class Codes: G05B 21/02 electric
• CPC Class Codes: Y02P 90/02 Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Abstract

A system for monitoring a multiparameter manufacturing process by examining on a real-time basis a stream of data units, each data unit including a numerical characteristic indicative of a current state of one of the parameters of that manufacturing process and a tag identifying the parameter. The system employs an entity-relational database to group and evaluate all data units using statistical evaluation criteria, and to identify and generate signals relating to particular testpoints in the manufacturing process so that an indication of the significance of the statistical pattern and the manufacturing process parameter involved are given in such terms as to permit human evaluation of and, if necessary, manual or automatic intervention in the manufacturing process.

Claims

1. A system for monitoring a multiparameter manufacturing process by examining on a real-time basis a stream of data units, each data unit including a numerical characteristic indicative of a current state of one of the parameters of that manufacturing process and a tag identifying the parameter, the system comprising:

grouping means, operating on each data unit as it is received by the monitoring system, for grouping all of the data units having identical tags together into groups of numerical characteristics, the size of each group being selectively variable for each tag,

pattern recognition means, operating as each data unit is received on each group of numerical characteristics assembled by the grouping means, for recognizing statistical patterns in the numerical characteristics, each pattern recognition means recognizing a single selected statistical pattern and including an output for result values indicative of the such recognition,

a plurality of registers, each register receiving the result values from a selected pattern recognition means,

comparing means for comparing the content of each register with a chosen standard for that register as each result value is received in the register to provide a comparison result at an output thereof indicative of the significance of any recognized statistical pattern, and

signal managing means for directing to at least one controller of the manufacturing process a signal of the comparison result of any register the output of which indicates that a statistically significant pattern of numerical characteristics has been identified, the signal including an indication of the significance of the pattern and the manufacturing process parameter involved to permit human evaluation of and, if necessary, intervention in the manufacturing process.

2. The system of claim 1 wherein the pattern recognition means comprises calculating means for calculating a central tendency value for each group of numerical characteristics as the size of each group reaches that previously selected, and a plurality of testing means, each testing means receiving either the numerical characteristics or the calculated central tendency values of a single specified tag, for testing against a stipulated test value, as each characteristic or value is received, to detect statistical patterns in the numerical characteristics, the testing means including an output for said result values.

3. The system of claim 1 further comprising ranking means for ranking all of the plurality of registers associated with each specified tag in a selected order.

4. The system of claim 3 further comprising selecting means coupled to the output of the comparing means and the signal managing means for selecting the highest ranking register having a comparison result indicative of the detection of a significant statistical pattern for signalling by the signalling means.

5. The system of claim 1 further comprising recording means for recording all comparison results at the outputs of comparing means indicative of the detection of a significant statistical pattern to develop a history of all such comparison results for possible later review and analysis.

11. The system of claim 1 including an entity-relational database including a plurality of entity fields containing ordered arrays of criteria specifying statistical patterns and limits for use by said pattern recognition means, and a plurality of relationship defining fields for correlating the entries in any two other fields.

12. The system of claim 11 wherein the relationship defining fields comprise doubly-linked lists linking the entries in two other fields, the entries in each doubly-linked list being arranged in an ordered array.

13. The system of claim 11 further comprising binary searching means for searching ordered arrays in fields whereby access to any desired entry therein is achieved without reading the contents of the field.

14. The system of claim 1 wherein the signal managing means includes means for providing a non-exclusive iconic presentation of said signal of the comparison result so that said controller of the manufacturing process can quickly interpret the significance of the signal and the identity of the manufacturing process parameter involved.

15. The system of claim 14 wherein the iconic representation form is hierarchical permitting a single icon of a selected shape and color to represent a selected condition at a selected number of testpoints, the selection of the particular icon being based on the significance of the events detected or the number of events detected.

6. A system for monitoring a multiparameter manufacturing process by examining on a real-time basis a stream of data units, each data unit including a numerical characteristic indicative of a current state of one of the parameters of that manufacturing process and a tag identifying the parameter, the system comprising:

grouping means, operating on all data units having certain previously specified tags, for grouping all of the data units having identical tags together into groups of numerical characteristics, the size of each group being selectively variable for each tag,

pattern recognition means operating on each group of numerical characteristics assembled by the grouping means for recognizing statistical patterns in the numerical characteristics, each pattern recognition means recognizing a single selected statistical pattern and including an output for result values indicative of such recognition,

a plurality of registers, each register receiving the result values from a selected pattern recognition means,

comparing means for comparing the content of each register with a chosen standard for that register after each result value is achieved in the register to provide a comparison result at an output thereof indicative of the significance of any recognized statistical pattern, and

signal managing means for directing to at least one controller of the manufacturing process a signal of the comparison result of any register the output of which indicates that a statistically significant pattern of numerical characteristics has been identified, the signal including an indication of the significance of the pattern and the manufacturing process parameter involved to permit human evaluation of and, if necessary, intervention in the manufacturing process, the signal managing means further comprising signal comparing means for comparing each sequential pair of signals developed by the signal managing means, the signal comparing means allowing only non-repeating signals to be emitted therefrom.

7. A system for monitoring a multiparameter manufacturing process by examining on a real-time basis a stream of data units, each data unit including a numerical characteristic indicative of a current state of one of the parameters of that manufacturing process and a tag identifying the parameter, the system comprising:

grouping means, operating on all data units having certain previously specified tags, for grouping all of the data units having identical tags together into groups of numerical characteristics, the sizes of each group being selectively variable for each tag, the grouping means comprising a summing register for summing each group of numerical characteristics of the data units having identical tags together, and calculating means for determining a means value for each sum of numerical characteristics as the size of each group reaches that previously selected,

pattern recognition means operating on each group of numerical characteristics assembled by the grouping means for recognizing statistical patterns in the numerical characteristics, each pattern recognition means recognizing a single selected statistical pattern and including an output for result values indicative of such recognition,

a plurality of registers, each register receiving the result values from a selected pattern recognition means,

comparing means for comparing the content of each register with a chosen standard for that register after each result value is received in the register to provide a comparison result at an output thereof indicative of the significance of any recognized statistical pattern, and

signal managing means for directing to at least one controller of the manufacturing process a signal of the comparison result of any register the output of which indicates that a statistically significant pattern of numerical characteristics has been identified, the signal including an indication of the significance of the pattern and the manufacturing process parameter involved to permit human evaluation of and, if necessary, intervention in the manufacturing process.

8. A system for monitoring a multiparameter manufacturing process by examining on a real-time basis a stream of data units, each data unit including a numerical characteristics indicative of a current state of one of the parameters of that manufacturing process and a tag identifying the parameter, the system comprising:

grouping means, operating on all data units having certain previously specified tags, for grouping all of the data units having identical tags together into groups of numerical characteristics, the size of each group being selectively variable for each tag,

pattern recognition means operating on each group of numerical characteristics assembled by the grouping means for recognizing statistical patterns in the numerical characteristics, each pattern recognition means recognizing a single selected statistical pattern and including an output for result values indicative of such recognition,

a plurality of registers, each register receiving the result values from a selected pattern recognition means,

comparing means for comparing the content of each register with a chosen standard for that register after each result value is received in the register to provide a comparison result at an output thereof indicative of the significance of any recognized statistical pattern, and

signal managing means comprising a plurality of terminals positioned for review by controllers of the manufacturing process for directing to at least one controller of the manufacturing process a signal of the comparison result of any register the output of which indicates that a statistically significant pattern of numerical characteristics has been identified, each terminal including a video screen for displaying information selected by the controller, and means for selecting which of the plurality of terminals to which to send any particular signal, the signal including an indication of the significance of the pattern and the manufacturing process parameter involved to permit human evaluation of and, if necessary, intervention in the manufacturing process.

9. The system of claim 8 wherein the signalling means further comprises means effecting a window on the video screen of each selected terminal to indicate the statistical significance of the detected pattern and the manufacturing process parameter involved so as to permit human evaluation of and, if necessary, intervention in the manufacturing process.

10. The system of claim 8 wherein the signal managing means comprises means for retaining any information selected by the controller for display during the consideration and possible intervention so as to allow proactive control of the manufacturing process at the time the process is occurring.

16. A method for monitoring a multiparameter manufacturing process by examining on a real-time basis a stream of data units, each data unit including a numerical characteristic indicative of a current state of one of the parameters of that manufacturing process and a tag identifying the parameter, the method comprising the steps of:

grouping, for all data units having certain previously specified tags, all of the data units having identical tags together into groups of numerical characteristics, the size of each group being selectively variable for each tag,

calculating a central tendency value for each group of numerical characteristics as the size of each group reaches that previously selected, and, as each subsequent data unit is received, recalculating the central tendency value for each group to which the numerical characteristic of the subsequent data unit is added,

testing either the numerical characteristics or the calculated central tendency values of a single specified tag against a stipulated test value, as each characteristic or value is received, to detect statistical patterns in the numerical characteristics, the testing including an output of test result values indicative of the result of each test,

storing the test result values in a plurality of registers, each register receiving the test result values from a selected test,

comparing the content of each register with a chosen standard for that register after each test result value is received in the register to provide a comparison result at an output thereof indicative of the significance of any detected statistical pattern, and

signalling to at least one controller of the manufacturing process the comparison result of any register the output of which indicates that a statistically significant pattern of numerical characteristics has been identified, the signalling including indicating the significance of the pattern and the manufacturing process parameter involved to permit human evaluation of and, if necessary, intervention in the manufacturing process.

17. The method of claim 16 further comprising the step of modifying, at any point during the examination of a stream of data units, the stipulated test value employed in any testing step.

18. The system of claim 16 further comprising the step of modifying, at any point during the examination of a stream of data units, any of the chosen standards employed in any comparing step to determine the significance of any detected statistical pattern.

19. The system of claim 16 further comprising the steps of defining sets of similarly situated data elements, and analyzing coordinate sets of the similarly situated data elements to detect patterns suggestive of causal relationships.

20. The method of claim 16 further comprising the step of recording all comparison results indicative of the detection of a significant statistical pattern to develop a history of all such comparison results for later review and analysis.

21. A method for monitoring a multiparameter manufacturing process by examining on a real-time basis a stream of data units, each data unit including a numerical characteristics indicative of a current state of one of the parameters of that manufacturing process and a tag identifying the parameter, the method comprising the steps of:

grouping, for data units having certain previously specified tags, all of the data units having identical tags together into groups of numerical characteristics, the size of each group being selectively variable for each tag,

calculating a central tendency value for each group of numerical characteristics as the size of each group reaches that previously selected,

testing either the numerical characteristics or the calculated central tendency values of a single specified tag against a stipulated test value, as each characteristics of value is received, to detect statistical patterns in the numerical characteristics, the testing including an output of test result values indicative of the result of each test,

constructing an entity-relational database including a plurality of entity fields containing ordered arrays of criteria specifying statistical patterns, and selecting from the entity fields the criteria related to each test to be conducted during said testing step,

storing the test result values in a plurality of registers, each resister receiving the test result values from a selected test,

comparing the content of each register with a chosen standard for that register after each test result value is received in the register to provide a comparison result at an output thereof indicative of the significance of any detected statistical pattern, and

signalling to at least one controller of the manufacturing process the comparison result of any register the output of which indicates that a statistically significant pattern of numerical characteristics has been identified, the signal including indicating of the significance of the pattern and the manufacturing process parameter involved to permit human evaluation of and, if necessary, intervention in the manufacturing process.

22. The method of claim 21 wherein the selecting step comprises conducting a binary search of at least one ordered array in a selected field whereby access to any desired entry therein is achieved without reading the contents of the field.

23. The method of claim 21 wherein the constructing step includes forming relationship defining fields having doubly-linked lists linking the entries in two other fields, the entries in each doubly-linked list being arranged in an ordered array.

24. The method of claim 21 further comprising the step of specifying the ranking of certain of the statistical pattern specifying criteria and associating the significance indication of said indicating step with the ranking of the statistical pattern.

25. The method of claim 24 further comprising selecting the highest ranking statistical pattern specifying criteria currently indicating the detection of a significant statistical pattern for signalling according to the signalling step.

26. A system for monitoring a multiparameter manufacturing process by examining on a real-time basis a stream of data units, each data unit including a numerical characteristic indicative of a current state of one of the parameters of that manufacturing process and a tag identifying the parameter and testpoint, the system comprising:

an entity-relational database including a plurality of entity fields for containing ordered arrays of information of the type including said numerical characteristics, standards and criteria specifying statistical patterns and operational limits, and a plurality of relationship defining fields for correlating the entries in any two other fields,

grouping means, operating on all data units having certain previously specified tags, for grouping all of the data units having identical tags together into groups of numerical characteristics located in at least one of the plurality of entity fields, the size of each group being selectively variable for each field,

pattern recognition means operating on each field containing a group of numerical characteristics assembled by the grouping means for recognizing statistical patterns in the numerical characteristics, each pattern recognition means employing criteria taken from at least one of the fields specifying statistical patterns to recognize a single selected statistical pattern, and including an output for result values indicative of the such recognition,

a plurality of output receiving fields, each output receiving field receiving the result values from a selected pattern recognition means,

comparing means for comparing the content of each output receiving field with a chosen standard related to that output receiving field after each result value is received to provide a comparison result at an output of the comparing means indicative of the significance of any recognized statistical pattern, and

signal managing means for directing to at least one controller of the manufacturing process a signal of said comparison result including means for providing a non-exclusive iconic presentation of said signal of the comparison result so that said controller of the manufacturing process can quickly interpret the significance of the signal and the identity of the manufacturing process parameter involved to permit human evaluation of and, if necessary, intervention in the manufacturing process.

27. The system of claim 26 wherein the signal managing means further comprises means effecting a window on a video screen, the iconic presentation form being hierarchical to permit a single icon of a selected shape and color to represent a selected condition at a selected number of testpoints.

28. The system of claim 27 wherein the window effecting means comprises means for retaining any information selected by the controller for display during the consideration and possible intervention so as to allow proactive control of the manufacturing process at the time the process is occuring.

29. The system of claim 27 wherein the relationship defining fields comprise doubly-linked lists linking the entries in two other fields, the entries in each doubly-linked list being arranged in an ordered array.

30. The system of claim 27 further comprising binary searching means for searching ordered arrays in fields whereby access to any desired entry therein is achieved without reading the contents of the field.

31. The system of claim 27 wherein the iconic representation form is hierarchical permitting a single icon of a selected shape and color to represent a selected condition at a selected number of testpoints, the selection of the particular icon being based on the significance of the events detected or the number of events detected.

32. A system for monitoring a multiparameter manufacturing process by examining on a real-time basis a stream of data units, each data unit including a numerical characteristic indicative of a current state of one of the parameters of that manufacturing process and a tag identifying the parameter and testpoint, the system comprising:

a plurality of first testing means, each first testing means receiving the numerical characteristics of a single specified tag for testing against a stipulated test value as each data unit is received, and including an output for result values indicative of the current state of one of the parameters of the manufacturing process being monitored,

grouping means, operating on each data unit as it is received by the monitoring system, for grouping all of the data units having identical tags together into groups of numerical characteristics, the size of each group being selectively variable for each tag,

calculating means for calculating a central tendency value for each group of numerical characteristics as the size of each group reaches that previously selected,

a plurality of second testing means, each second testing means receiving the calculated central tendency values of a single specified tag as each numerical characteristic is received to detect statistical patterns in the numerical characteristics, the testing means including an output for result values indicative of a recognized statistical pattern,

a plurality of registers, each register receiving the result values from either a first testing means or a second testing means,

comparing means for comparing the content of each register with a chosen standard for that register as each result value is received in the register to provide a comparison result at an output thereof indicative of the significance of the result value, and

signal managing means for directing to at least one controller of the manufacturing process a signal of the comparison result of any register the output of which indicates that a significant result value has been identified, the signal including an indication of the significance of the value and the manufacturing process parameter involved to permit human evaluation of and, if necessary, intervention in the manufacturing process.

33. The system of claim 32 further comprising means for modifying the stipulated test value employed by any of the first testing means at any point during the examination of a stream of data units.

34. The system of claim 32 further comprising means for modifying any of the second testing means at any point during the examination of a stream of data units to change the detection criterion indicative of a recognized statistical pattern.

35. The system of claim 32 further comprising means for modifying any of the chosen standards employed by any comparing means to determine the significance of result values received by the comparing means.

36. The system of claim 32 further comprising means for defining sets of similarly situated data elements, and means for analyzing coordinate sets of the similarly situated data elements to detect patterns suggestive of causal relationships.