DIAGNOSIS OF EQUIPMENT FAILURES USING AN INTEGRATED APPROACH OF CASE BASED REASONING AND RELIABILITY ANALYSIS

• US 20070288795A1
• Filed: 08/23/2007
• Published: 12/13/2007
• Est. Priority Date: 09/17/2003
• Status: Active Grant

• Current Assignee: GlobalFoundries, Inc.
• Original Assignee: Gyana Parija, Samir Sharma, Ying Leung
• Inventors: Sharma, Samir, Parija, Gyana, Leung, Ying
• Granted Patent: US 7,730,020 B2
• Time in Patent Office: Days
• US Class Current: 714/26
• CPC Class Codes: G05B 23/0275 Fault isolation and identification, e.g. classify fault; estimate cause or root of failure
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing
• CPC Class Codes: Y10S 707/99948 Application of database or data structure, e.g. distributed, multimedia, or image

Abstract

Equipment failures are diagnosed using an integrated approach of case-based reasoning (CBR) and statistical reliability analysis. The method outputs a single list of suggested failed components, ranked by an overall probability of failure, and their associated past solutions from the case base. The overall probability of failure is calculated using the combined logic of case based reasoning and statistical reliability analysis. The method is typically used in a real-time decision support system to aid equipment diagnosis by a maintenance technician working in the field.

Claims

1-10. -10. (canceled)

11. A method to diagnose failures for a plurality of given machines, each machine having a corresponding plurality given hardware components, comprising:

collecting a statistical reliability data for a plurality of the given machines and for a plurality of the given hardware components corresponding to each machine;

constructing a case base database for each machine, having a plurality of case based database records, each of said records including a failed machine identifier field, a failed component identifier field, a failure description text field, and a solution record field;

receiving an equipment problem description from a user, said description including a problem machine identifier and a problem description text;

generating a statistical reliability based conditional statistical probability, for each of a plurality of the hardware components corresponding to the machine identified by said problem machine identifier, of said component having a failed state given that said machine has a failed state, based on said collected statistical reliability data;

generating a case-based probability, for each of a plurality of the hardware components corresponding to the machine identified by said problem machine identifier, of said component having a failed state given that said machine has a failed state, based on comparing the problem description text received from the user to the failure description text field of a plurality of the case base database records;

identifying a most probable failed hardware component from among the hardware components corresponding to the machine identified by said problem machine identifier, said identifying based on a combination of the generated statistical reliability based conditional probability of said identified component having a failed state given that said machine has a failed state, and said generated case-based probability of said component having a failed state.

12. The method according to claim 11, wherein said collecting includes calculating a statistical average time between two successive failures of each of said machines, calculating a statistical average time between successive two failures for each of said hardware components, calculating a statistical average time to repair failures of each machine, and calculating a statistical average time to repair failures of each of the hardware components corresponding to said machine.

13. The method according to claim 11, wherein said identifying includes identifying a plurality of probable failed hardware components from among the hardware components corresponding to the machine identified by said problem machine identifier, said identifying a plurality based on a combination of the generated statistical reliability based conditional probability of each the identified probable failed hardware components having a failed state given that said machine has a failed state, and said generated case-based probability of each of said identified probable failed components having a failed state.

14. The method according to claim 13, wherein each said generated case-based probability of each of said identified probable failed components having a failed state is generated based on comparing the problem description text received from the user to the failure description text field of a plurality of the case base database records having a failed machine identifier field corresponding to the machine identified by said problem machine identifier and having a failed component identifier field corresponding to at least one of said identified probable failed components.

15. The method according to claim 13, wherein said collecting includes calculating a statistical average time between two successive failures of each of said machines, calculating a statistical average time between successive two failures for each of said hardware components, calculating a statistical average time to repair failures of each machine, and calculating a statistical average time to repair failures of each of the hardware components corresponding to said machine.

16. The method according to claim 14, wherein said collecting includes calculating a statistical average time between two successive failures of each of said machines, calculating a statistical average time between successive two failures for each of said hardware components, calculating a statistical average time to repair failures of each machine, and calculating a statistical average time to repair failures of each of the hardware components corresponding to said machine.

17. A decision support system to diagnose failures for a plurality of given machines, each machine having a corresponding plurality given hardware components, comprising:

a statistical reliability database storing a statistical reliability data for a plurality of the given machines and for a plurality of the given hardware components corresponding to each machine;

a case base database storing a plurality of case based database records, each of said records including a failed machine identifier field, a failed component identifier field, a failure description text field, and a solution record field;

a decision support system client for receiving an equipment problem description from a user, said description including a problem machine identifier and a problem description text;

a real-time decision support system engine for calculating failure probability for each hardware component in the equipment, wherein said engine is arranged

to generate a statistical reliability based conditional statistical probability, for each of a plurality of the hardware components corresponding to the machine identified by said problem machine identifier, of said component having a failed state given that said machine has a failed state, based on said collected statistical reliability data,

to generate a case-based probability, for each of a plurality of the hardware components corresponding to the machine identified by said problem machine identifier, of said component having a failed state given that said machine has a failed state, based on comparing the problem description text received from the user to the failure description text field of a plurality of the case base database records, and

to identify a most probable failed hardware component from among the hardware components corresponding to the machine identified by said problem machine identifier, said identifying based on a combination of the generated statistical reliability based conditional probability of said identified component having a failed state given that said machine has a failed state, and said generated case-based probability of said component having a failed state.

18. The system according to claim 17, wherein said statistical reliability database includes collecting includes a statistical average time between two successive failures of each of said machines, a statistical average time between successive two failures for each of said hardware components, a statistical average time to repair failures of each machine, and a statistical average time to repair failures of each of the hardware components corresponding to said machine.

19. The system according to claim 17, wherein said real-time decision support system engine is further arranged to identify a plurality of probable failed hardware components from among the hardware components corresponding to the machine identified by said problem machine identifier, said arranged to identify said plurality based on a combination of the generated statistical reliability based conditional probability of each the identified probable failed hardware components having a failed state given that said machine has a failed state, and said generated case-based probability of each of said identified probable failed components having a failed state.