System and method for production system performance prediction

• US 7,672,811 B2
• Filed: 06/17/2005
• Issued: 03/02/2010
• Est. Priority Date: 06/17/2005
• Status: Active Grant

• Current Assignee: GM Global Technology Operations LLC (General Motors Company)
• Original Assignee: GM Global Technology Operations Incorporated (Motors Liquidation Co. GUC Trust)
• Inventors: Shadid, Basel Q., Dwibhashyam, Ram, Arinez, Jorge F., Barajas, Leandro G., Bandyopadhyay, Pulak, Xiao, Guoxian
• Primary Examiner(s): Feliciano; Eliseo Ramos
• Assistant Examiner(s): Suarez; Felix E
• Application Number:
  US11/155,098
• Publication Number:
  US 20060288260A1
• Time in Patent Office: 1,719 Days
• Field of Search: 702/108, 702/179, 702183-188, 700/79, 707/203, 709/224, 714/26
• US Class Current: 702/185
• CPC Class Codes: G05B 23/0232 based on qualitative trend analysis, e.g. system evolution

Abstract

Disclosed herein are a system, method and apparatus for reporting, making alerts and predicting fault codes generated by machines in a line. Historical fault code data is received and filtered according to particular criteria to generate filtered fault code data. Classification of the filtered fault code data into physical groups and into logical groups is followed by sorting the groups to produce fault trend data. Processing the fault trend data with a plurality of analyzers generates output including reports, alerts, and predictions of future fault code occurrences.

Claims

1. A method for reporting, making alerts and predicting fault codes generated by machines in a line, comprising:

preprocessing historical fault code data comprising fault codes, wherein preprocessing comprises;

receiving stored historical fault code data generated by the machines;

filtering the historical fault code data according to particular criteria to generate filtered fault code data;

parsing text of the filtered fault code data;

classifying the filtered fault code data into physical groups, each physical group associated with a component of the machines and corresponding to the parsed text of the fault code data;

classifying the filtered fault code data into logical groups, each logical group including a plurality of the fault codes having a logical relationship associated through the parsed text of the fault code data; and

sorting the physical groups and the logical groups to produce fault trend data; and

processing the fault trend data with a plurality of analyzers to generate report output.

2. The method recited in claim 1, wherein the historical fault code data comprises a predetermined amount of data that is compiled over a predetermined time period.

3. The method recited in claim 1 wherein the machines in the line operate during production time and do not operate during non-production time, and wherein the filtering comprises:

eliminating fault codes generated during non-production time.

4. The method as recited in claim 1 wherein the received fault codes comprise duplicates and non-duplicates, and wherein filtering comprises:

eliminating duplicate fault codes.

5. The method recited in claim 1 wherein faults codes occur at specific times, and wherein filtering comprises:

combining two fault codes if the time between them is less than a predetermined period of time.

6. The method recited in claim 1 wherein the machines in the line produce product that has particular quality, and wherein classifying comprises:

categorizing the fault codes into those affecting the product quality.

7. The method as recited in claim 1 wherein classifying comprises:

categorizing the fault codes into those associated with operator behavior.

8. The method as recited in claim 1 wherein classifying comprises:

categorizing the fault codes into those requiring attention of maintenance personnel with specific skills.

9. The method as recited in claim 1 wherein particular fault codes have more impact on the machines in the line than other fault codes, and wherein processing comprises:

reporting the number of fault codes having the more impact on the machines in the line than other fault codes.

10. The method as recited in claim 1, further comprising:

generating reports comprising current state output.

11. The method as recited in claim 1, further comprising:

generating reports comprising future state output.

12. A system for reporting, making alerts and predicting fault codes generated by machines in a line, comprising:

a preprocessing module comprising;

a receiving module capable of receiving historical fault code data comprising fault codes;

a filtering module capable of filtering the historical fault code data according to particular criteria to generate filtered fault code data;

a first classifying module for classifying the filtered fault code data into physical groups, each physical group associated with a component of the machines and corresponding to the fault code data;

a second classifying module for classifying the filtered fault code data into logical groups, each logical group including a plurality of the fault codes having a logical relationship associated through the fault code data;

a sorting module for sorting the physical groups and the logical groups to produce fault trend data; and

a processing module for processing the fault trend data with a plurality of analyzers to generate report output.

13. The system as recited in claim 12, wherein the historical fault code data comprises a predetermined amount of data that is compiled over a predetermined time period.

14. The system as recited in claim 12 wherein particular fault codes have more impact on the machines in the line than other fault codes, further comprising:

a reporting module for reporting the number of fault codes having the more impact on the machines in the line than other fault codes.

15. The system as recited in claim 12, wherein the report output comprises:

current state output.

16. The system as recited in claim 12, wherein the report output comprises:

future state output.

17. An apparatus for reporting, making alerts and predicting fault codes generated by machines in a line, comprising:

a database for storing historical fault code data comprising fault codes;

a filtering unit for filtering the historical fault code data according to particular criteria to generate filtered fault code data;

a classifying unit for classifying the filtered fault code data into physical groups and into logical groups;

each physical group associated with a component of the machines and corresponding to the fault code data, and each logical group including a plurality of the fault codes having a logical relationship associated through the fault code data;

a sorting unit for sorting the physical groups and the logical groups to produce fault trend data; and

a processor for processing the fault trend data with a plurality of analyzers to generate report output.

18. The apparatus as recited in claim 17, wherein the historical fault code data comprises a predetermined amount of data that is compiled over a predetermined period of time.

19. The apparatus as recited in claim 17, wherein the report output comprises:

current state output.

20. The apparatus as recited in claim 17, wherein the report output comprises:

future state output.

21. A method for analyzing fault data associated with a plurality of production line machines, comprising:

collecting and compiling a plurality of fault codes associated with the production line machines;

parsing each fault code;

executing a keyword search of the parsed fault code;

classifying each fault code in one of a plurality of physical groups based upon the keyword search of the parsed fault code, each physical group associated with a component of the production line machines;

classifying each fault code into one of a plurality of logical groups based upon the keyword search of the parsed fault code, each logical group including a plurality of the fault codes having a logical relationship associated through the keywords of the plurality of fault codes;

sorting the physical groups and the logical groups; and

generating fault trend data associated with the plurality of fault codes associated with the production line machines classified in the physical groups and logical groups.

22. The method of claim 21, further comprising

classifying each fault code in one of a plurality of ad hoc groups based upon the keyword search of the parsed fault code, each ad hoc group associated with a specific manufacturing process of the production line machines;

and

generating fault trend data associated with the plurality of fault codes associated with the production line machines classified in the physical groups, the logical groups, and the ad hoc groups.

23. The method of claim 21, further comprising classifying each filtered fault code in one of a plurality of logical groups based upon the keyword search of the parsed fault code, each logical group comprising keywords having a logical relationship associated with one of product quality, operator behavior, a skilled trade, and a spatial relationship of the machines.