Automatic test generation for model-based real-time fault diagnostic systems

• US 5,214,577 A
• Filed: 10/24/1990
• Issued: 05/25/1993
• Est. Priority Date: 10/24/1990
• Status: Expired due to Term

• Current Assignee: Osaka Gas Company Limited, Vanderbilt University
• Original Assignee: Osaka Gas Company Limited
• Inventors: Okuda, Koji, Miyasaka, Nobuji, Padalkar, Samir, Sztipanovits, Janos, Karsai, Gabor, Biegl, Csaba
• Primary Examiner(s): Smith, Jerry
• Assistant Examiner(s): GORDON, PAUL P
• Application Number:
  US07/602,133
• Time in Patent Office: 944 Days
• Field of Search: 364/149, 364/150, 364/151, 364/184, 364/185, 364/189, 364/550, 364/551.01, 364/578, 371/3, 371/23, 371/25.1, 395/904
• US Class Current: 700/79
• CPC Class Codes: G05B 23/0262 Confirmation of fault detection, e.g. extra checks to confirm that a failure has indeed occurred
• CPC Class Codes: G08B 29/14 checking the detection circuits
• CPC Class Codes: Y02P 90/02 Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Abstract

In a real-time diagnostic system, an alarm sequence generator is used to test the correctness of a fault model. The fault model describes an industrial process being monitored. The alarm sequence generator reads the fault model and generates a user interface, from which specific components can be selected for failure at specified times. The alarm sequence generator assembles all alarms that are causally downstream from the selected set of faulty components and determines which alarms should be turned on based on probabilistic and temporal information in the fault model. The timed alarm sequence can be used by an expert to measure the correctness of a particular model, or can be used as input into a diagnostic system to measure the correctness of the diagnostic system.

Claims

1. A method for testing the accuracy and completeness of a diagnostic system which analyzes the operation of a monitored process, the system being modeled as a graph model having

a plurality of on-line nodes representing functions being monitored by a first group of alarms,

a plurality off-line nodes representing functions being monitored by off-line alarms,

a plurality of nodes representing functions of the monitored process not having alarms associated therewith,

a plurality of nodes representing fault source components, and

a plurality of propagation paths, wherein each said path selectively interconnects at least two said nodes for representing allowable propagation of information from one node to another,

the method comprising the steps of specifying each node of the model at which data signals shall originate,

specifying for each originating node a time at which the data signal shall originate,

propagating said data signal through said graph model and along said propagation paths, said propagation paths being characterized by minimum and maximum propagation times, and a probability of propagation of said signals along said paths, and

determining the nodes to which said data signal is propagated and the time of its arrival.

2. The method of claim 1 wherein said step of propagating said data signal along said propagation paths further comprises the steps of:

generating a random time for each path that is not greater than said maximum propagation time and not less than said minimum propagation time for the propagation path; and

propagating said data signal along each propagation path at a time specified by said random time.

3. The method of claim 1 wherein said step of propagating said data signal along said propagation paths further comprises the steps of:

generating a random probability value of propagation; and

propagating said data signal along said propagation path only if said generated random probability value has a specified desired relationship relative to said propagation probability for a path.

4. The method of claim 3 further comprising the step of:

comparing a resultant data signal flow through said graph model with an expected data signal flow through said model for determining the accuracy of said graph model.

5. Apparatus for testing the accuracy and completeness of a diagnostic system which analyzes the operation of a monitored process, the system being modeled as a graph model having

a plurality of on-line nodes representing functions being monitored by a first group of alarms,

a plurality of off-line nodes representing functions being monitored by off-line alarms,

a plurality of nodes representing functions of the monitored process not having alarms associated therewith,

a plurality of nodes representing fault source components, and

a plurality of propagation paths, wherein each said path selectively interconnects at least two said nodes for representing allowable propagation of information from one node to another,

the apparatus comprising

means for specifying each node of the model at which data signals shall originate,

means for specifying for each originating node a time at which the data signal shall originate,

means for propagating said data signal through said graph model and along said propagation paths, said propagation paths being characterized by minimum and maximum propagation times, and a probability of propagation of said signals along said paths, and

means for determining the nodes to which said data signal is propagated and the time of its arrival.

6. The apparatus of claim 5 wherein said means for propagating said data signal along said propagation paths further comprises

means for generating a random time for each path that is not greater than said maximum propagation time and not less than said minimum propagation time for the propagation path;

and

means for propagating said data signal along each propagation path at a time specified by said random time.

7. The apparatus of claim 5 wherein said means for propagating said data signal along said propagation paths further comprises:

means for generating a random probability value of propagation; and

means for propagating said data signal along said propagation path only if said generated random probability value has a specific desired relationship relative to said propagation probability for a path.

8. The apparatus of claim 7 further comprising:

means for comparing a resultant data signal flow through said graph model with an expected data signal flow through said model for determining the accuracy of said graph model.