Method for associating text in automatic diagnostic system to produce recommended actions automatically

• US 5,081,598 A
• Filed: 02/21/1989
• Issued: 01/14/1992
• Est. Priority Date: 02/21/1989
• Status: Expired due to Term

• Current Assignee: Siemens Power Generation, Inc.
• Original Assignee: Westinghouse Electric Corporation (Paramount Global (f/k/a ViacomCBS Inc.))
• Inventors: Harper, Karl E., Bellows, James C.
• Primary Examiner(s): Fleming, Michael R.
• Assistant Examiner(s): Sheikh, Ayaz R.
• Application Number:
  US07/313,265
• Time in Patent Office: 1,057 Days
• Field of Search: 364/513, 364/900, 364/200, 364/550, 364/186
• US Class Current: 702/184
• CPC Class Codes: G06F 11/2257 using expert systems
• CPC Class Codes: G06N 5/04 Inference or reasoning models
• CPC Class Codes: Y10S 706/911 Nonmedical diagnostics

Abstract

A method for automatically generating recommended actions in response to a diagnosis uses a list ordered by impact of the actions. The list is evaluated using required evidence which must be supported by data and assumed evidence which must not be contradicted by the data. The method is applied to the diagnosis of malfunctions in equipment by an expert system rulebase. A correspondence is defined between text modules including the recommended actions and each of the schemata in the rulebase which produce a diagnosis. After a diagnosis is evaluated as true, the corresponding text modules are evaluated to produce a recommended action. The text modules can also be used for documentation of the rulebase.

Claims

1. A method of automatically determining a recommended action in response to diagnosis of a malfunction in equipment, comprising the steps of:

(a) defining a correspondence between each possible diagnosis and a recommended action set including at least one action

(b) producing a diagnosis, having recommended actions associated therewith, in dependence upon a first set of evidence of conditions currently existing in the equipment; and

(c) selecting, using a computerized process, one of the recommended actions in the recommended action set corresponding to the diagnosis produced in step (b) in dependence upon a second set of evidence of conditions, different from the first set of evidence, both required and assumed evidence being includable in the second set of evidence, using a computerized process including the steps of

(c1) comparing data representing actual conditions associated with the equipment being diagnosed to the second set of evidence; and

(c2) determining an appropriate action when the data indicates that the conditions corresponding to any required evidence exist and any assumed evidence is uncontradicted by the data.

2. A method as recited in claim 1,

further comprising the step of (d) designating, when the at least one action in the recommended action set includes a plurality of actions, an order for the actions determined by impact of the actions on the equipment, and

wherein said comparing in step (b1) is performed for the actions in the recommended action set, in the order designated in step (d), whereby the appropriate action determined in step (b2) has a least severe impact on the equipment among the actions in the recommended actions set capable of being selected in step (b2).

3. A method as recited in claim 1, further comprising the step of (d) outputting the one of the recommended actions selected in step (b) and a consequence of inaction.

4. A method as recited in claim 3,

wherein the recommended action set corresponding to each possible diagnosis includes a single description of the consequence of inaction, and

wherein step (d) outputs the single description of the consequence of inaction regardless of which of the actions in the recommended action set is selected in step (b).

5. A method of associating text with schemata in a rulebase of an expert system for a evaluating data using the schemata in the rulebase, comprising the steps of:

(a) storing the selection criteria, independent of the schemata in the rulebase, for selecting text modules and a correspondence between the schemata in the rulebase and the text modules,

(b) diagnosing existing conditions represented by the data in dependence upon the schemata in the rulebase;

(c) automatically outputting at least one of the text modules in dependence upon the existing conditions diagnosed in step (b) and the correspondence and selection criteria stored in step (a); and

(d) selectively outputting any of the text modules when said diagnosing in step (b) is not being performed.

6. A method as recited in claim 5, further comprising the step of (e) providing for editing of the text modules substantially concurrently with editing of the schemata in the rulebase.

7. A method as recited in claim 6,

wherein step (a) comprises storing a header in each of the text modules to identify a corresponding schema, and

wherein said method further comprises the step of (f) generating automatically a new text module, including the header therefor, whenever one of the schemata is edited without a corresponding text module in existence.

8. A method as recited in claim 6, further comprising the step of (f) providing for editing of the text modules independently of the schemata in the rulebase.

9. A method of associating text with schemata in a rulebase of an expert system for a evaluating data using the schemata in the rulebase, comprising the steps of:

(a) storing selection criteria, independent of the schemata in the rulebase, for selecting text modules and a correspondence between the schemata in the rulebase and the text modules, including a header in each of the text modules to identify a corresponding schemata;

(b) diagnosing existing conditions represented by the data in dependence upon the schemata in the rulebase;

(c) associating an evidence set with each of the text modules; and

(d) automatically outputting, when the data is being evaluated, at least one of the text modules when the header stored therein identifies the corresponding schemata having been evaluated as true and the evidence set associated therewith is evaluated as true for the data being evaluated.

10. A method as recited in claim 9,

further comprising the step of (e) storing all of the text modules corresponding to a single one of the schemata together in a hierarchical order defined by a hierarchy number stored in the header of each of the text modules, and

wherein the selection criteria used in step (c) combines the evidence set of a higher level text module with the evidence set of a lower level text module to determine whether to output the lower level text module.

11. A method as recited in claim 10, wherein the expert system is used to diagnosis malfunctions in equipment and to provide recommendations,

wherein said method further comprises the step of (f) storing recommendation text in each of the text modules corresponding to a diagnosis hypothesis, and

wherein step (c) comprises outputting, when evaluating the data, all of the text modules meeting the selection criteria and corresponding to the diagnosis hypothesis evaluated as true.

12. A method as recited in claim 11, wherein the recommendation text in each of the text modules used to provide the recommendations includes at least one of a description of the diagnosis hypothesis corresponding thereto, a recommended action and a consequence of inaction.

13. A method as recited in claim 12, wherein step (e) comprises the steps of:

(e1) storing the text modules containing the description of the diagnosis hypothesis corresponding thereto at a first hierarchy level; and

(e2) storing at least two of the text modules containing alternative recommended actions, when more than one action is possible, with different associated evidence sets at a second hierarchy level lower than the first hierarchy level.

14. A method as recited in claim 13, wherein step (c) further comprises the step (e3) storing at least one of the text modules containing the consequence of inaction after all alternative actions for the diagnosis rule corresponding thereto, using either of the first and second hierarchy levels.