Computer-implemented system and method for constructing a system

• US 5,793,933 A
• Filed: 07/17/1997
• Issued: 08/11/1998
• Est. Priority Date: 09/13/1993
• Status: Expired due to Fees

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation), Toshiba Corporation
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Iwamasa, Mikito
• Primary Examiner(s): Downs, Robert W.
• Assistant Examiner(s): SHAH, SANJIV
• Application Number:
  US08/895,997
• Time in Patent Office: 390 Days
• Field of Search: 395/50, 395/60
• US Class Current: 706/45
• CPC Class Codes: G06F 11/2257 using expert systems

Abstract

A computer-implemented system for helping to construct a model-based diagnostic system which allows a user to customize the model-based diagnostic system according to the nature of various diagnosis targets. Unit procedure data including the operation of the deducing unit 4, hypothesizing unit 5, and verification unit 6 is stored in the unit procedure data storage unit 13. When the user uses the customization unit 14 to select unit procedure data in a desired sequence, the customize unit 14 stores user-selected unit procedure data or data to identify the unit procedure data in the deductive procedure data storage unit 8, hypothesize procedure data storage unit 9, and verification procedure data storage unit 10, respectively. The deductive unit 4, hypothesize unit 5, and verification unit 6 each refer to procedure data during operation to know processing function names that are set.

Claims

1. A computer-implemented system for constructing a diagnostic system which diagnoses on a behavior of a diagnosis target comprising a plurality of elements by using a model, wherein said computer-implemented system comprises:

a deducing means comprising a function calculation means for calculating expected output data corresponding to input data for each of said elements based on said model, and a structural propagation means for propagating the output data of each of said elements to other elements associated with the element based on said model;

a hypothesizing means for comparing the output data of one element acquired by said function calculation means with an observed value of the element acquired from said diagnosis target for each of said elements, and presuming a candidate for an abnormality factor;

a hypothesis verification means comprising a function backward-calculation means for backward-calculating the input data from which output data is produced for each of said elements based on said model, and a structural propagation means for propagating the input data of each of said elements to other elements associated with the element based on said model, wherein said hypothesis verification means compares the input data of each of said elements acquired by said backward-calculation means with an observed value acquired from said diagnosis target and decides all elements each of which is free from abnormality;

a customizing means for setting up separately each of deductive procedure data representing how said deducing means operates, hypothesis procedure data representing how said hypothesizing means operates and hypothesis verification procedure data representing how said hypothesis verification means operates, into desired content.

3. A computer-implemented system as claimed in claim 1 or 2, further comprising:

a reasoning control means for controlling coordinative relationships among said deducing means, said hypothesizing means, and said hypothesis verification means; and

wherein said customizing means customizes reasoning control data representing how said reasoning control means operates.

2. A computer-implemented system for constructing a diagnostic system, wherein said diagnostic system uses a model representing a diagnosis target comprising a plurality of elements in view of a structural aspect and a functional aspect for each of said elements and diagnoses an abnormality of said diagnosis target based on said model, and wherein said computer-implemented system comprises:

a deducing means comprising a function calculation means for calculating expected output data corresponding to input data for each of said elements based on said model, and a structural propagation means for propagating the output data of each of said elements to other elements associated with the element based on said model;

a hypothesizing means for comparing the output data of one element acquired by said function calculation means with an observed value of the element acquired from said diagnosis target for each of elements located on the data-output side of and associated with a predetermined element, detecting an element having a different output data and observed value, and presuming an existence of a candidate for an abnormality factor within a group of the detected element and all elements located on the data-input side in relation to the detected element;

a hypothesis verification means comprising a function backward-calculation means for backward-calculating the input data from which output data is produced for each of said elements based on said model, and a structural propagation means for propagating the input data of each of said elements to other elements associated with the element based on said model, wherein said hypothesis verification means compares the input data of one element acquired by said backward-calculation means with an observed value of the element acquired from said diagnosis target for each of elements located on the data-input side of and associated with a predetermined element, for each of which its input data and observed value correspond with each other; and

a customizing means for setting up separately each of deductive procedure data representing how said deducing means operates, hypothesis procedure data representing how said hypothesizing means operates and hypothesis verification procedure data representing how said hypothesis verification means operates, into desired content.

4. A computer-implemented method for constructing a diagnostic system which diagnoses on a behavior of a diagnosis target comprising a plurality of elements by using a model, wherein said method comprises:

deducing process comprising a function calculation step for calculating expected output data corresponding to input data for each of said elements based on said model, and a structural propagation step for propagating the output data of each of said elements to other elements associated with the element based on said model;

hypothesizing process for comparing the output data of one element acquired by said function calculation step with an observed value of the element acquired from said diagnosis target for each of said elements, and presuming a candidate for an abnormality factor;

hypothesis verification process comprising a function backward-calculation step for backward-calculating the input data from which output data is produced for each of said elements based on said model, and a structural propagation step for propagating the input data of each of said elements to other elements associated with the element based on said model, wherein said hypothesis verification process compares the input data of each of said elements acquired by said backward-calculation step with an observed value acquired by said backward-calculation step with an observed value acquired from said diagnosis target and decides all elements each of which is free from abnormality; and

a customizing process for setting up separately each of deductive procedure data representing how said deducing process operates, hypothesis procedure data representing how said hypothesizing process operates and hypothesis verification procedure data representing how said hypothesis verification process operates, into desired content.

5. A computer-implemented method for constructing a diagnostic system which diagnoses a diagnosis target by using a model representing said diagnosis target comprising a plurality of elements in view of a structural aspect and a functional aspect for each of said elements, wherein said method comprises:

deducing process comprising a function calculation step for calculating expected output data corresponding to input data for each of said elements based on said model, and a structural propagation step for propagating the output data of each of said elements to other elements associated with the element based on said model;

hypothesizing process for comparing the output data of one element acquired by said function calculation step with an observed value of the element acquired from said diagnosis target for each of elements located on the data-output side of and associated with a predetermined element, detecting an element having a different output data and observed value, and presuming an existence of a candidate for an abnormality factor within a group of the detected element and all elements located on the data-input side in relation to the detected element;

hypothesis verification process comprising a function backward-calculation step for backward calculating the input data from which output data is produced for each of said elements based on said model, and a structural propagation step for propagating the input data of each of said elements to other elements associated with the element based on said model, wherein said hypothesis verification process starts from said predetermined element, compares the input data of one element acquired by said backward-calculation means with an observed value of the element acquired from said diagnosis target for each of elements located on the data-input side of and associated with said predetermined element, and decides an absence of an abnormality for all elements, for each of which its input data and observed value correspond with each other; and

customizing process for setting up separately each of deductive procedure data representing how said deducing process operates, hypothesis procedure data representing how said hypothesizing process operates and hypothesis verification procedure data representing how said hypothesis verification process operates, into desired content.

6. A computer-implemented system for constructing a system by using a model representing a system comprising a plurality of elements in view of a structural aspect and a functional aspect for each element of said plurality of elements, wherein said computer-implemented system comprises:

a plurality of basic tasks, corresponding to a deducing means, a hypothesizing means and a hypothesis verification means, for executing each process requested for constructing a system by using said model;

procedure data representing how each of said basic tasks operates, wherein said procedure data, corresponding to the features of said model, includes first procedure data representing functional operation, second procedure data representing structural operation, and third procedure data for controlling of how each of said basic tasks operates;

each of said basic tasks including a first executing means for executing a process related to function based on said first procedure data, a second executing means for executing a process related to structural propagation based on said second procedure data, and a third executing means for controlling of how said first and second executing means execute based on the third procedure data; and

a customizing means for setting up at least one of said first, second and third procedure data into desired content.

7. A computer-implemented system as claimed in claim 6, wherein:

said customizing means sets up fourth procedure data representing how orderly said plurality of basic tasks operate;

said system further comprising controlling means for controlling of how orderly said plurality of basic tasks operate based on the fourth data set up by said customizing means.

8. A computer-implemented method for constructing a system by using a model representing a system comprising a plurality of elements in view of a structural aspect and a functional aspect for each of said elements, the method comprises the steps of:

providing a plurality of basic tasks, corresponding to a deducing means, a hypothesizing means and a hypothesis verification means, for executing each process requested for constructing a system by using said model;

providing procedure data representing how each of said basic tasks operates, wherein said procedure data, corresponding to features of said model, includes first procedure data representing a functional operation, second procedure data representing a structural operation, and third procedure data for controlling of how each of said basic tasks operates; and

executing processes by each of said basic tasks, wherein said processes includes a first executing step for executing a process related to function based on the first procedure data set up by a customizing process, a second executing step for executing a process related to structural propagation based on said second procedure data set up by the customizing process, and a third executing step for controlling how said first and second executing steps execute based on the third procedure data set up by the customizing process.

9. A computer-implemented method as claimed in claim 8, wherein how orderly said plurality of basic tasks operate is controlled based on fourth data set up by the customizing process.