Computerized method and system for automated system diagnosis detection

• US 9,274,868 B2
• Filed: 10/04/2011
• Issued: 03/01/2016
• Est. Priority Date: 10/19/2010
• Status: Expired due to Fees

• Current Assignee: Siemens AG
• Original Assignee: Siemens AG
• Inventors: Hoffmann, Werner, Roshchin, Mikhail, Hubauer, Thomas, Klssinger, Ralf, Pirker, Michael
• Primary Examiner(s): Iqbal, Nadeem
• Application Number:
  US13/880,483
• Publication Number:
  US 20130268809A1
• Time in Patent Office: 1,610 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 11/004 Error avoidance G06F11/07 and subgroups take precedence
• CPC Class Codes: G06F 11/2257 using expert systems

Abstract

In a method, system, and storage medium for obtaining system diagnosis, a diagnosis domain is described that includes a number of possible diagnoses, the number of possible diagnoses prioritized based on at least one of a number of confidence levels, a preference function, and a calculation of a relationship between a number of confirmed system symptoms and a number of possible diagnoses. A number of hypotheses for incipient system symptoms is generated, and a hypothesis is selected from the generated number of hypotheses. Information is obtained regarding a number of system observations, and a selection relevant diagnostics for the system is obtained.

Claims

1. A method for identifying a diagnosis of a system incident in a computerized system upon occurrence of said system incident, said system incident producing incipient symptoms in said computerized system before said system incident results in a failure of said computerized system, said method comprising:

providing a processor with an electronic description of said incipient symptoms and, in said processor, defining a diagnosis domain from said incipient symptoms, said diagnosis domain comprising a plurality of possible diagnoses of said system incident;

in said processor, automatically generating, for each possible diagnosis in said diagnosis domain, a confidence level that numerically expresses a probability that the respective possible diagnosis is a correct diagnosis of said incipient symptoms;

providing said processor with a first electronic input that describes a predetermined preference function for making a selection of the possible diagnoses in said diagnosis domain based on their respective confidence levels;

providing said processor with a second electronic input that describes confirmed symptoms of said computerized system;

in said processor, automatically prioritizing said possible diagnoses in said diagnosis domain according to a predetermined combination of said confidence levels, said preference function, and said confirmed symptoms;

in said processor, applying an abduction algorithm to said incipient symptoms and a selected subset of the prioritized possible diagnoses that are higher in said prioritization, to obtain, for each possible diagnosis in said subset, a corresponding hypothesis for said system incident;

providing said processor with a third electronic input that describes observed system observations regarding operation of said computerized system;

in said processor, applying said system observations to the respective hypotheses corresponding to the possible diagnoses in said subset, to select at least one of said hypotheses as an explanation for said system incident; and

at a display monitor in communication with said processor, displaying, as a diagnosis of said system incident, a description of the possible diagnosis that corresponds to said at least one selected hypothesis.

2. A diagnostic system for a computerized system in which a system incident occurs having incipient symptoms associated therewith, said diagnostic system being configured to identify a diagnosis for said system incident from said incipient systems before said system incident results in a failure of said computerized system, said diagnosis system comprising:

a processor provided with an electronic description of said incipient symptoms, said processor being configured to define a diagnosis domain from said incipient symptoms, said diagnosis domain comprising a plurality of possible diagnoses of said system incident;

said processor being configured to automatically generate, for each possible diagnosis in said diagnosis domain, a confidence level that numerically expresses a probability that the respective possible diagnosis is a correct diagnosis of said incipient symptoms;

said processor being provided with a first electronic input that describes a predetermined preference function for making a selection of the possible diagnoses in said diagnosis domain based on their respective confidence levels;

said processor being provided with a second electronic input that describes confirmed symptoms of said computerized system;

said processor being configured to automatically prioritize said possible diagnoses in said diagnosis domain according to a predetermined combination of said confidence levels, said preference function, and said confirmed symptoms;

said processor being configured to apply an abduction algorithm to said incipient symptoms and a selected subset of the prioritized possible diagnoses that are higher in said prioritization, to obtain, for each possible diagnosis in said subset, a corresponding hypothesis for said system incident;

said processor being provided with a third electronic input that describes observed system observations regarding operation of said computerized system;

said processor being configured to apply said system observations to the respective hypotheses corresponding to the possible diagnoses in said subset, to select at least one of said hypotheses as an explanation for said system incident; and

a display monitor in communication with said processor, said processor being configured to display, as a diagnosis of said system incident, a description at said display monitor of the possible diagnosis that corresponds to said at least one selected hypothesis.

3. A non-transitory, computer-readable data storage medium encoded with programming instructions, said data storage medium being loaded into a processor of a diagnostic system that is in communication with a computerized system, said computerized system having a system incident that occurs therein that has incipient symptoms associated therewith, and said programming instructions configuring said diagnostic system to identify a diagnosis for said system incident before said system incident results in a failure of said computerized system, by causing said processor to:

receive an electronic description of said incipient symptoms and, in said processor, defining a diagnosis domain from said incipient symptoms, said diagnosis domain comprising a plurality of possible diagnoses of said system incident;

generate, for each possible diagnosis in said diagnosis domain, a confidence level that numerically expresses a probability that the respective possible diagnosis is a correct diagnosis of said incipient symptoms;

receive a first electronic input that describes a predetermined preference function for making a selection of the possible diagnoses in said diagnosis domain based on their respective confidence levels;

receive a second electronic input that describes confirmed symptoms of said computerized system;

prioritize said possible diagnoses in said diagnosis domain according to a predetermined combination of said confidence levels, said preference function, and said confirmed symptoms;

apply an abduction algorithm to said incipient symptoms and a selected subset of the prioritized possible diagnoses that are higher in said prioritization, to obtain, for each possible diagnosis in said subset, a corresponding hypothesis for said system incident;

receive a third electronic input that describes observed system observations regarding operation of said computerized system;

apply said system observations to the respective hypotheses corresponding to the possible diagnoses in said subset, to select at least one of said hypotheses as an explanation for said system incident; and

at a display monitor in communication with said processor, displaying, as a diagnosis of said system incident, a description of the possible diagnosis that corresponds to said at least one selected hypothesis.