Case-based knowledge source for artificial intelligence software shell

• US 5,402,524 A
• Filed: 12/22/1992
• Issued: 03/28/1995
• Est. Priority Date: 12/22/1992
• Status: Expired due to Fees

• Current Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation), Sumitomo Rubber Industries Limited
• Original Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation)
• Inventors: Wada, Nobuyoshi, Schultz, Brian A., Lowenfeld, Simon, Thompson, Robert W. Jr., Bauman, Douglas A.
• Primary Examiner(s): Downs, Robert W.
• Application Number:
  US07/994,664
• Time in Patent Office: 826 Days
• Field of Search: 395/53, 395/77, 395/68, 395/50, 395/61
• US Class Current: 706/45
• CPC Class Codes: G06N 5/043 Distributed expert systems; Blackboards

Abstract

An artificial intelligence software shell for plant operation simulation includes a blackboard module including a database having objects representing plant elements and concepts. A rule-based knowledge source module and a case-based knowledge source module, in communication with the blackboard module, operate on specific predefined blackboard objects. The case-based knowledge source module includes cases having past data points on a time scale. Present data is compared with the past data in the cases both on a time scale and on a magnitude scale for determining levels of closeness in time and magnitude. An input data module, in communication with the blackboard module, enables a user to input data to the shell. A control module, in communication with the knowledge source modules and the input data module, receives all input data and controls operation of the knowledge source modules in accordance with a predetermined knowledge source priority scheme.

Claims

1. An artificial intelligence software shell for plant operation simulation comprising:

a blackboard module including a database having objects representing plant elements and concepts, the blackboard module storing data values corresponding to the objects, each stored data value having an assigned time variable;

a case-based knowledge source module including a time-magnitude scoring artificial intelligence operation scheme, in communication with the blackboard module, operating on new case data points for comparison with stored data values;

an input data module, in communication with the blackboard module and the at least one knowledge source module, enabling data to be inputted to the shell; and

a control module, in communication with the input data module and the case-based knowledge source module, receiving all input data and controlling operation of the case-based knowledge source;

wherein, in accordance with the time-magnitude scoring artificial intelligence operation scheme of the case-based knowledge source each new case data point is assigned a time window within which a stored data value is selected for scoring the new case data point based on a level of closeness in time and magnitude between the new case data point and the stored data value.

2. A software shell as claimed in claim 1 wherein the control module includes:

an event detector module, in communication with the input data module, receiving all input data and determining when the case-based knowledge source module should execute; and

an activation/agenda manager module, in communication with the case-based knowledge source module and the event detector module, executing the case-based knowledge source module in accordance with a predetermined knowledge source priority scheme.

3. A software shell as claimed in claim 2 further including:

a rule-based knowledge source module having a forward-chaining belief propagation scheme including rules in if-then-else form with associated levels of belief.

4. A software shell as claimed in claim 3 wherein the case-based knowledge source module further includes:

a data comparison scheme including predefined patterns and conditions, whereupon execution of the case-based knowledge source, the conditions are inferred to be true if a certain level of closeness is found between received data and the patterns.

5. A software shell as claimed in claim 3 wherein the blackboard module database includes a hierarchy of objects having vertical hierarchial connections and horizontal links.

6. A software shell as claimed in claim 5 wherein each blackboard object has attributes and each attribute has at least one value, the database including a hash table and plurality of pointers for linking the objects and corresponding attributes and values.

7. A software shell as claimed in claim 5 wherein the vertical hierarchical connections include IS-A and PART-OF hierarchical connections.

8. A software shell as claimed in claim 3 further including:

a blackboard structure description file including object names and attribute/value pairs for each object;

a rule-based description file including rules forming the rule-based knowledge source and a description of execution preconditions for the rule-based knowledge source; and

a case-based description file including cases forming the case-based knowledge source and a description of execution preconditions for the case-based knowledge source;

wherein the blackboard structure description file, rule-based description file, case-based description file, and input data file are user created.

9. A software shell as claimed in claim 8 further including:

a diagnostics file, created by the shell for each input data file cycle completed, including diagnoses made by the rule-based knowledge source and case matches found by the case-based knowledge source; and

a blackboard status file, created by the shell, including blackboard information as modified by any knowledge sources after completion of each input data file cycle.

10. A software shell as claimed in claim 2 wherein the case-based knowledge source module further includes:

a data comparison scheme including predefined patterns and conditions, whereupon execution of the case-based knowledge source, the conditions are inferred to be true if a certain level of closeness is found between received data and the patterns.

11. A software shell as claimed in claim 2 wherein the blackboard module database includes a hierarchy of objects having vertical hierarchial connections and horizontal links.

12. A software shell as claimed in claim 11 wherein each blackboard object has attributes and each attribute has at least one value, the database including a hash table and plurality of pointers for linking the objects and corresponding attributes and values.

13. A software shell as claimed in claim 11 wherein the vertical hierarchical connections include IS-A and PART-OF hierarchical connections.

14. A software shell as claimed in claim 2 wherein the activation/agenda manager module includes a mechanism for interrupting execution of a knowledge source and executing a higher priority knowledge source in accordance with the knowledge source priority scheme.

15. A software shell as claimed in claim 14 wherein the activation/agenda manager module includes a mechanism for determining whether an interrupted knowledge source should be rescheduled for execution.

16. A software shell as claimed in claim 2 wherein the event detector module includes a data structure implementing an algorithm such that when data is received, only a knowledge source which operates on the received data is checked for execution availability.

17. A software shell as claimed in claim 1 wherein the case-based knowledge source module further includes:

stored cases including previously encountered data; and

a similarity function for ranking present data by closeness to data in a stored case.

18. A software shell as claimed in claim 17 wherein the case-based knowledge source module further includes:

a table having columns and rows;

a plurality of case frames including at least one variable, each case frame stored in a row of the table; and

a plurality of cases stored in the columns of the table and including the at least one variable.

19. A software shell as claimed in claim 1 wherein each new case data point includes an associated scoring function indicating the type of comparison between the new case data point and the stored data value.

20. A software shell as claimed in claim 19 wherein scoring function includes a minimum data value scoring function.

21. A software shell as claimed in claim 19 wherein scoring function includes a maximum data value scoring function.

22. A software shell as claimed in claim 19 wherein scoring function includes closest in magnitude data value scoring function.

23. An artificial intelligence software shell for plant operation simulation comprising:

blackboard means for storing objects representing plant elements and concepts and storing data values corresponding to the objects, each stored data value having an assigned time variable;

a case-based knowledge source module including a time-magnitude scoring artificial intelligence operation scheme, in communication with the means for storing, operating on new case data points for comparison with stored data values;

means, in communication with the means for storing and the case-based knowledge source module, for enabling data to be inputted to the shell; and

means, in communication with the means for enabling and the case-based knowledge source module, for receiving all input data and controlling operation of the case-based knowledge source module;

wherein, in accordance with the time-magnitude scoring artificial intelligence operation scheme of the case-based knowledge source each new case data point is assigned a time window within which a stored data value is selected for scoring the new case data point based on a level of closeness in time and magnitude between the new case data point and the stored data value.

24. A software shell as claimed in claim 23 wherein the means for receiving and controlling include:

means, in communication with the means for enabling, for determining when the case-based knowledge source module should execute; and

means, in communication with the at least one knowledge source module and the means for determining, for executing the case-based knowledge source module in accordance with a predetermined knowledge source priority scheme.

25. A software shell as claimed in claim 24 further including:

a rule-based knowledge source module having a forward-chaining belief propagation scheme including rules in if-then-else form with associated levels of belief.

26. A software shell as claimed in claim 25 wherein the means for storing includes a database having a hierarchy of objects having vertical hierarchical connections and horizontal links.

27. A software shell as claimed in claim 26 wherein each object has attributes and each attribute has at least one value, the database including a hash table and a plurality of pointers for linking object and corresponding attributes and values.

28. A software shell as claimed in claim 27 wherein the means for executing includes means for interrupting execution of a lower priority knowledge source and executing a higher priority knowledge source in accordance with the knowledge source priority scheme.

29. A software shell as claimed in claim 28 wherein the means for determining includes means for receiving input data, determining when knowledge source execution preconditions are met, and checking only a knowledge source whose execution preconditions are met for execution availability.

30. A software shell as claimed in claim 29 further including:

a blackboard structure description file including object names and attribute/value pairs for each object;

a rule-based description file including rules forming the rule-based knowledge source and a description of execution preconditions for the rule-based knowledge source;

a case-based description file including cases forming the case-based knowledge source and a description of execution preconditions for the case-based knowledge source;

wherein the blackboard structure description file, rule-based description file, case-based description file, and input data file are user created.

31. A software shell as claimed in claim 30 further including:

a diagnostics file, created by the shell for each input data file cycle completed, including diagnoses made by the rule-based knowledge source and case matches found by the case-based knowledge source; and

a blackboard status file, created by the shell, including blackboard information as modified by any knowledge sources after completion of each input data file cycle.

32. A software shell as claimed in claim 31 further including stream sockets for intercommunication, the sockets communicating messages in the form of a request-response pair.

33. A software shell as claimed in claim 23 wherein the case-based knowledge module source further includes:

a data comparison scheme including predefined patterns and conditions, whereupon execution of the case-based knowledge source, the conditions are inferred to be true if a certain level of closeness is found between received data and the patterns.

34. A software shell as claimed in claim 33 wherein the data comparison scheme compares both magnitude and timing of the data upon execution of the case-based knowledge source.

35. A software shell as claimed in claim 34 wherein the case-based knowledge source module further includes:

stored cases including previously encountered data; and

a similarity function for ranking present data by closeness to data in a stored case.

36. A software shell as claimed in claim 35 wherein the case-based knowledge source module further includes:

a table having columns and rows;

a plurality of case frames including at least one variable, each case frame stored in a row of the table; and

a plurality of cases stored in the columns of the table and including the at least one variable.

37. A method of simulating plant operation with an artificial intelligence software shell comprising the steps of:

storing, in a database, objects representing plant elements and concepts and data values corresponding to the objects each data value having an assigned time variable;

reading input data including data points from an input data source, each data point having an associated time window variable;

determining, from the input data read, when at least a case-based artificial intelligence knowledge source should execute in accordance with a predetermined knowledge source priority scheme; and

executing at least the case-based knowledge source, in accordance with the determination;

wherein the step of executing includes selecting one of said stored data values from within a time window defined by the time window variable for scoring a data point based on a level of closeness between the data point and the data value in time and magnitude closeness to data in a stored case.

38. A method as claimed in claim 37 wherein the step of executing at least the case-based knowledge source includes the step of updating object values.

39. A method as claimed in claim 38 wherein the step of executing at least the case-based knowledge source further includes the steps of:

executing rule-based knowledge sources having a forward-chaining belief propagation scheme including rules in if-then-else form with associated levels of belief;

and executing case-based knowledge sources having a data comparison scheme including predefined patterns and conditions.

40. A method as claimed in claim 39 wherein the step of executing case-based knowledge sources includes the step of inferring to be true the conditions if a certain level of closeness is found between received data and the patterns.

41. A method as claimed in claim 37 wherein the step of storing includes the step of storing objects, attributes and values, each object having an attribute and each attribute having at least one value.

42. A method as claimed in claim 41 wherein the step of executing at least the case-based knowledge source includes the step of interrupting execution of a lower priority knowledge source and executing a higher priority knowledge source in accordance with the knowledge source priority scheme.

43. A method as claimed in claim 42 wherein the step of determining further includes the steps of determining when knowledge source execution preconditions are met and checking only a knowledge source whose execution preconditions are met, for execution availability.

44. A method as claimed in claim 43 further including the steps of:

generating a blackboard structure description file including object names and attribute/value pairs for each object;

generating a rule-based description file including rules forming the rule-based knowledge source and execution preconditions; and

generating a case-based description file including cases and a description of execution preconditions; and

45. A method as claimed in claim 44 further including the steps of:

generating a diagnostics file including diagnoses made by the rule-based knowledge sources and case matches found by the case-based knowledge sources; and

generating a blackboard status file including blackboard information as modified by any knowledge sources during execution.

46. A method as claimed in claim 45 further including the steps of communicating messages through sockets between modules in the form of request-response pairs.

47. A case-based knowledge source module for an artificial intelligence software shell for plant operation simulation, the shell including a blackboard module database having objects representing plant elements and concepts, and a control module for receiving data and controlling operation of the case-based knowledge source module on the blackboard objects, the case-based knowledge source module comprising:

a table including a plurality of stored cases, each case having a plurality of stored data points and an associated time value for each stored data point; and

a time magnitude scoring scheme, whereupon if a certain level of closeness is found between data received and the stored data points in both magnitude and time, a condition is inferred to be true;

wherein the data received includes new data points, each new data point having an assigned time window value, and a stored data point is selected from within the time window to score the new data point based on a level of closeness in time and magnitude between the stored data point and the new data point.

48. An artificial intelligence software shell comprising:

a blackboard module including a database storing past data points associated with objects representing plant elements and concepts, each past data point having an associated time variable; and

a case-based knowledge source module, in communication with the blackboard module, receiving new data points, each new data point having an associated time window variable defining a time window, the case-based knowledge source module, for each new data point received, selecting a past data point from within the time window based on a level of closeness between the past data point and the new data point for scoring the new data point.

49. A case-based knowledge source module for an artificial intelligence software shell for plant operation simulation, the shell including a blackboard module database having objects representing plant elements and concepts, and a control module for receiving data and controlling operation of the case-based knowledge source module on the blackboard objects, the case-based knowledge source module comprising:

a table including a plurality of stored cases, each case having a plurality of stored data points and an associated time value for each stored data point; and

a time-magnitude scoring scheme in which received data points have an associated time window variable, the time window variable defining a time window within which a stored data point is selected to score the received data point based on a level of closeness in time and magnitude between the stored data point and the received data point.