Method and apparatus for diagnosing machines

• US 4,985,857 A
• Filed: 08/19/1988
• Issued: 01/15/1991
• Est. Priority Date: 08/19/1988
• Status: Expired due to Fees

• Current Assignee: General Motors Corporation (Motors Liquidation Co. GUC Trust)
• Original Assignee: General Motors Corporation (Motors Liquidation Co. GUC Trust)
• Inventors: Joyce, Robert H., Marczewski, Richard W., Kishore, Nanda, Ollanik, Steven J., Marek, Melissa M., Amble, Charles S., Bajpai, Atul, Shrivastava, Anil B.
• Primary Examiner(s): Black, Thomas G.
• Application Number:
  US07/234,371
• Time in Patent Office: 879 Days
• Field of Search: 364/580, 364/508, 364/513, 364/551.01, 364/200, 364/300, 364/900, 371/15, 371/16, 371/17, 371/20
• US Class Current: 702/184
• CPC Class Codes: G05B 19/4065 Monitoring tool breakage, life or condition
• CPC Class Codes: G05B 2219/33295 Fuzzy expert system for diagnostic, monitoring
• CPC Class Codes: G05B 2219/33314 Failure reason analysis, simple strategy or multiple outcome analysis
• CPC Class Codes: G05B 2219/33318 Knowledge acquisition
• CPC Class Codes: G06F 11/2257 using expert systems
• CPC Class Codes: Y10S 706/912 Manufacturing or machine, e.g. agricultural machinery, machine tool

Abstract

A general purpose expert system architecture for diagnosing faults in any one of a plurality of machines includes a machine information database containing information on characteristics of various components of the machines to be diagnosed and a sensory input database which contains vibration data taken at predetermined locations on each of the machines. The system knowledge base contains a plurality of general rules that are applicable to each of the plurality of machines. The generality of diagnosis is accomplished by focusing on components that make up the machine rather than individual machines as a whole. The system architecture also permits diagnosis of machines based on other parameters such as amperage, torques, displacement and its derivatives, forces, pressures and temperatures. The system includes an inference engine which links the rules in a backward chaining structure.

Claims

1. A general purpose expert system architecture for diagnosing component faults in any one of a plurality of machines, comprising a machine information database containing information on physical characteristics of various components of at least one of said plurality of machines, a sensory input database containing sensory data taken at predetermined locations on at least one of said plurality of machines, a knowledge base containing a plurality of general rules that are applicable to all components of interest in the plurality of machines without reference to any specific machine, and an inference engine which manipulates the rules to enable the system to identify faults based on data from said sensory database and information from said machine information database.

2. A general purpose diagnostic expert system architecture for diagnosing faults in any one of a plurality of machines, comprising a machine information database (MID) containing physical characteristics of various components of each of said plurality of machines, a sensory input database (SID) containing sensory data taken at predetermined locations on each one of said plurality of machines, and a knowledge base (KB) containing a plurality of general rules that are applicable to all of said plurality of machines and which are based on machine component performance, said system further including an inference engine which manipulates the rules, said system responding to operator identification of a specific machine to act on information obtained from the MID and the SID that relates to said specific machine in order to render a fault diagnosis.

3. The invention defined in claim 2 wherein said KB comprises a rule-base consisting of said rules, and a fact-base which contains dynamic and static facts, said dymanic facts corresponding to the information obtained from said MID and SID and said stactic facts includes at least one lookup table containing entries corresponding to predetermined constraints that must be satisfied with respect to all components of a predetermined precision level.

4. The invention defined in claim 2 wherein said KB comprises a rule-base consisting of said rules, and a fact-base which contains dynamic and static facts, said dynamic facts corresponding to the information obtained from said MID and SID and said stactic facts are included in a plurality of lookup tables containing threshold values of amplitudes for a predetermined range of frequencies which are maximum acceptable amplitude values common to all components, each lookup table corresponding to a predefined precision level.

5. An automated method of diagnosing potential and existing faults on a plurality of machine each of which have a plurality of components, using an expert system having a knowledge engineering tool, and a knowledge base containing a plurality of general rules that are applicable to all components of interest in the plurality of machines without reference to any specific machine, said knowledge base having a fact base, and a rule base containing a plurality of rules each of which have as a premise, test conditions which when satisfied will conclude on a component fault or will conclude on another test condition which then becomes the premise for one or more other rules, comprising the step of:

(1) selecting one of the machines for diagnosis,

(2) storing, in the fact base, machine data from a machine information database (MID) and associated sensor data for the selected machine from a sensory input database (SID),

(3) applying certain rule in the rule base to the stored MID and SID data to compile a list a hypothesized faults,

(4) based on the compiled list of hypothesized faults, and the stored MID and SID data, selecting a list of rules from the plurality of rules which are applicable to the current machine configuration,

(5) backward chaining through said applicable rules to confirm which if any of the hypothesized fault are the cause of the machine problem.

6. An automated method of diagnosing potential and existing faults on a plurality of machine each of which have a plurality of components, using an expert system having a knowledge engineering tool, a machine information database (MID) containing information on characteristics of various components of said plurality of machines, a sensory input database (SID) containing sensory data taken at predetermined locations on each of said plurality of machines, and a knowledge base containing a plurality of general rules that are applicable to all components of interest in the plurality of machines without reference to any specific machine, said knowledge base including a fact base, and a rule base containing a plurality of rules each of which have as a premise, test conditions which when satisfied will conclude on a component fault or will conclude on another test condition which then becomes the premise for one or more other rules, comprising the step of:

(1) selecting one of the machines for diagnosis,

(2) storing, in the fact base, the machine data from the MID and associated sensor data from the SID for the selected machine,

(3) applying certain rule in the rule base to the stored MID and SID data to compile a list a hypothesized faults,

(4) based on hypothesized faults, and the stored MID and SID data, selecting a list of rules from the plurality of rules which are applicable to the current machine configuration,

(5) backward chaining through said applicable rules to confirm which if any of the hypothesized fault are the cause of the machine problem.

7. An automated method of diagnosing potential and existing faults on any one of a plurality of machine each of which are comprised of a plurality of components, using an expert system having a knowledge engineering tool;

and a knowledge base (KB) having a rule base containing a plurality of general rules that are applicable to all components of interest in the plurality of machines without reference to any specific machine, and a fact base, comprising the step of;

(1) retrieving data relating to the machine to be diagnosed, from a computer aided design database,

(2) continuously collecting sensory data from strategic locations on the machine,

(3) processing the sensory data to identify predetermined characteristics of the sensory data,

(4) applying the rules in the KB rule base to identify faults from the processed data of step (3).

8. An automated method of diagnosing potential and existing faults on any one of a plurality of machines each of which is comprised of a plurality of components, using an expert system having a knowledge engineering tool, and a knowledge base (KB) containing a rule base and a fact base, comprising the step of:

(1) retrieving data relating to the machine to be diagnosed, from a computer aided design database

(2) continuously collecting sensory data from strategic locations on the machine,

(3) processing the sensory data to identify predetermined characteristics of the sensory data,

(4) applying the rules in the KB rule base to identify faults from the processed data of step (3).

9. An automated method of diagnosing potential and existing faults on any one of a plurality of machines each of which is comprised of a plurality of components, using an expert system having a knowledge engineering tool, and a knowledge base (KB) containing a rule base and a fact base, comprising the step of:

(1) establishing a machine information database (MID) on the various machines to be diagnosed

(2) collecting sensory data from the specific machine that is to be diagnosed

(3) storing the sensory data in a sensory input database (SID),

(4) retrieving from the MID and storing in the KB fact-base the description of a specific machine to be diagnosed,

(5) retrieving from the SID and storing in the KB fact-base the sensory data for the specific machine to be diagnosed,

(6) processing the sensory data to identify predetermined characteristics of the sensory data,

(7) applying the rules in the KB to try to confirm faults from the processed data of step (6)

(8) presenting to the user, any confirmed faults and appropriate recommendations for their repair.

10. An automated method of diagnosing potential and existing faults on any one of a plurality of machines each of which is comprised of a plurality of components, using an expert system having a knowledge engineering tool, and a knowledge base (KB), the KB having a rule base containing a plurality of rules each of which have as a premise, test conditions which when satisfied will conclude on a component fault or will conclude on another test condition which then becomes the premise for one or more other rules, and a fact base which includes both static and dynamic portions, comprising the step of:

(1) establishing a machine information database (MID) on the various machines to be diagnosed

(2) collecting sensory data from the specific machine that is to be diagnosed

(3) converting the sensory data to a form acceptable by the knowledge engineering tool and storing the sensory data in a sensory input database (SID),

(4) retrieving from the MID and storing in the KB dynamic fact-base the description of a specific machine to be diagnosed,

(5) retrieving from the SID and storing in the KB dynamic fact-base the sensory data for the specific machine to be diagnosed,

(6) processing the sensory data to identify predetermined characteristics of the sensory data,

(7) applying the rules in the KB to try to confirm faults from the processed data of step (6)

(8) presenting to the user, any confirmed faults and appropriate recommendations for their repair.

11. An automated method of diagnosing potential and existing faults on any one or a plurality of machines each of which is comprised of a plurality of components, using an expert system having a knowledge engineering tool, and a knowledge base (KB), the KB having a rule base containing a plurality of rules each of which have as a premise, test conditions which when satisfied will conclude on a component fault or will conclude on another test condition which then becomes the premise for one or more other rules, and a fact base which includes both static and dynamic portions, comprising the step of:

(1) establishing a machine information database (MID) on the various machines to be diagnosed

(2) collecting sensory data from the specific machine that is to be diagnosed

(3) converting the sensory data to a form acceptable by the knowledge engineering tool and storing the sensory data in a sensory input database (SID),

(4) retrieving from the MID and storing in the KB dynamic fact-base the description of a specific machine to be diagnosed,

(5) retrieving from the SID and storing in the KB dynamic fact-base the sensory data for the specific machine to be diagnosed,

(6) processing the sensory data to identify and to qualitatively classify the values of the sensory data which exceed predetermined thresholds,

(7) applying the rules in the KB to try to confirm faults from the data that exceeds the threshold,

(8) presenting to the user, any confirmed faults and appropriate recommendations for their repair.

12. An automated method of diagnosing potential and existing faults on any one of a plurality of machines each of which is comprised of a plurality of components, using an expert system having a knowledge engineering tool, and a knowledge base (KB), the KB having a rule base containing a plurality of rules each of which have as a premise, test conditions which when satisfied will conclude on a component fault or will conclude on another test condition which then becomes the premise for one or more other rules, and a fact base which includes both static and dynamic portions, comprising the step of:

(1) establishing a machine information database (MID) on the various machines to be diagnosed

(2) collecting sensory data from the specific machine that is to be diagnosed

(3) converting the sensory data to a form acceptable by the knowledge engineering tool and storing the sensory data in a sensory input database (SID),

(4) retrieving from the MID and storing in the KB dynamic fact-base the description of a specific machine to be diagnosed,

(5) retrieving from the SID and storing in the KB dynamic fact-base the sensory data for the specific machine to be diagnosed,

(6) processing the sensory data to identify and to qualitatively classify the values of the sensory data which exceed predetermined thresholds,

(7) generating a hypothesized fault list based on the sensory data that exceeds the thresholds,

(8) applying the rules in the KB to try to confirm faults from the list,

(9) presenting to the user, any confirmed faults and appropriate recommendations for their repair.

13. An automated method of diagnosing potential and existing faults on any one of a plurality of machines each of which is comprised of a plurality of components, using an expert system having a knowledge engineering tool, and a knowledge base (KB), the KB having a rule base containing a plurality of rules each of which have as a premise, test conditions which when satisfied will conclude on a component fault or will conclude on another test condition which then becomes the premise for one or more other rules, and a fact base which includes both static and dynamic portions, comprising the step of:

() establishing a machine information database (MID) on the various machines to be diagnosed

(2) collecting sensory data from the specific machine that is to be diagnosed

(3) converting the sensory data to a form acceptable by the knowledge engineering tool and storing the sensory data in a sensory input database (SID),

(4) retrieving from the MID and storing in the KB dynamic fact-base the description of a specific machine to be diagnosed,

(5) retrieving from the SID and storing in the KB dynamic fact-base the sensory data for the specific machine to be diagnosed,

(6) processing the sensory data to identify predetermined characteristics of the sensory data,

(7) grouping the processed sensory data to isolate data representing potential faults of the machine on a priority basis,

(8) selecting one of the groups for analysis,

(9) generating a hypothesized fault list for the selected group,

(10 ) applying the rules in the KB to try to confirm faults from the list,

(11) presenting to the user, any confirmed faults and appropriate recommendations for their repair.

14. An automated method as defined in claim 13 wherein said sensory data is vibration data and the predetermined characteristics identified in step (6) are peaks and humps.

15. An automated method of diagnosing potential and existing faults on any one of a plurality of machines each of which is comprised of a plurality of components, using an expert system having a knowledge engineering tool, and a knowledge base (KB), the KB having a rule base containing a plurality of rules each of which have as a premise, test conditions which when satisfied will conclude on a component fault or will conclude on another test condition which then becomes the premise for one or more other rules, and a fact base which includes both static and dynamic portions, comprising the step of:

(1) establishing a machine information database (MID) on the various machines to be diagnosed

(2) collecting sensory data from the specific machine that is to be diagnosed

(3) converting the sensory data to a form acceptable by the knowledge engineering tool and storing the sensory data in a sensory input database (SID),

(4) retrieving from the MID and storing in the KB dynamic fact-base the description of a specific machine to be diagnosed,

(5) retrieving from the SID and storing in the KB dynamic fact-base the sensory data for the specific machine to be diagnosed,

(6) processing the sensory data to identify predetermined characteristics of the sensory data,

(7) grouping the processed sensory data to isolate data representing potential faults of the machine on a priority basis,

(8) selecting one of the groups for analysis,

(9) generating a hypothesized fault list for the selected group,

(10 ) applying the rules in the KB to a selected group of data in an attempt to reach conclusions regarding which of the potential faults can be confirmed,

(11) if no fault is confirmed, then selecting tests which will further assist in confirming potential faults,

(12) if tests have been selected, then ordering the tests in a prioritized sequence based on predetermined criteria,

(13) performing at least one of the ordered tests and storing the collected data in the fact base,

(14) repeating steps (10 )-(13) until a fault can be confirmed or until no more test remain, and

(15) presenting to the user, any confirmed faults and appropriate recommendations for their repair.

16. An automated method as defined in claim 15 wherein said sensory data is vibration data and the predetermined characteristics identified in step (6) are based on values of the data that significantly deviate from normal peaks and humps.

17. An automated method as defined in claim 15 wherein said sensory data is vibration data and the predetermined characteristics identified in step (6) are peaks and humps.

18. An automated method of diagnosing potential and existing faults on any one of a plurality of machines each of which is comprised of a plurality of components, using an expert system having a knowledge engineering tool, and a knowledge base (KB), the KB having a rule base containing a plurality of rules each of which have as a premise, test conditions which when satisfied will conclude on a component fault or will conclude on another test condition which then becomes the premise for one or more other rules, and a fact base which includes both static and dynamic portions, comprising the step of:

(1) establishing a machine information database (MID) on the various machines to be diagnosed

(2) collecting vibration data from the specific machine that is to be diagnosed

(3) converting the vibration data to a form acceptable by the knowledge engineering tool and storing the vibration data in a sensory input database (SID),

(4) retrieving from the MID and storing in the KB dynamic fact-base the description of a specific machine to be diagnosed,

(5) retrieving from the SID and storing in the KB dynamic fact-base the vibration data for the specific machine to be diagnosed,

(6) processing the vibration data to identify the peaks and humps in the vibration data,

(7) grouping the peak and humps to isolate data representing potential faults of the machine on a priority basis,

(8) selecting one or more of the groups for analysis,

(9) generating a hypothesized fault list for the selected group,

(10 ) applying the rules in the KB to a selected one of groups of data in an attempt to reach conclusions regarding which of the potential fault can be confirmed,

(11) if no fault is confirmed, then selecting tests which will further assist in confirming potential faults,

(12) if tests have been selected, then ordering the tests in a prioritized sequence based on (a) time required for test, (b) equipment needed, (c) ease of performing the test, and (d) number of potential faults which are identifiable by running the particular test,

(13) performing at least one of the ordered tests and storing the collected data in the fact base,

(14) repeating steps (10)-(13) until a fault can be confirmed or until no more test remain,

(15) repeating step (10) if step (8) selected more than one group and repeating step (11)-(14) for the next selected group until no more groups remain to be analyzed,

(16) presenting to the user, any confirmed faults and appropriate recommendations for their repair.