Method of optimizing a serial manufacturing system

• US 5,229,948 A
• Filed: 11/03/1990
• Issued: 07/20/1993
• Est. Priority Date: 11/03/1990
• Status: Expired due to Term

• Current Assignee: The Research Foundation for The State University of New York (State University of New York)
• Original Assignee: Ford Motor Company
• Inventors: Wei, Kuang C., Otto, Norman C.
• Primary Examiner(s): Smith, Jerry
• Assistant Examiner(s): BROWN, THOMAS
• Application Number:
  US07/611,215
• Time in Patent Office: 990 Days
• Field of Search: 364/468, 364/148, 364/149, 364/150, 364/151, 364/152, 364/401, 364/402, 364/403, 364/478
• US Class Current: 700/99
• CPC Class Codes: G05B 19/41885 characterised by modeling, simulation of the manufacturing system
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: Y02P 90/02 Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Abstract

A method of optimizing a multi-stage serial manufacturing system which includes: (a) providing a quantitative state-space model of said serial manufacturing system that describes processing in terms of sensitivity information and performance in terms of part production and storage/retrieval, said description using process switches in the form of buffer status indicators that express coupling between stages; (b) sensing new sensitivity information that results from simulating said manufacturing system with said model using estimated system performance information; and (c) adjusting said performance information by iteratively using said new sensitivity information in an optimization algorithm, said adjustment being carried out simultaneously with the simulation of step (b).

Claims

1. A method of optimizing a multi-stage serial manufacturing system, comprising:

(a) providing a quantitative state-space model of said serial manufacturing system that describes processing in terms of stages, sensitivity information for blocking or starving of such stages, and performance information in terms of part production and buffer storage/retrieval, said description using process switches in the form of buffer status indicators that express coupling between stages;

(b) using estimated performance information, including buffer size, to simulate said manufacturing system with said model and sensing new sensitivity information that results therefrom; and

(c) concurrently adjusting said performance information, including buffer size, by iteratively using said new sensitivity information in an optimization algorithm to estimate a gradient of the performance information with respect to buffer size and thence adjust buffer size, and repeating steps (b) and (c) until the gradient substantially stabilizes.

2. The method as in claim 1, in which said serial manufacturing system is subject to random effects or variable multiple inputs accompanied by multiple random outputs.

3. The method as in claim 1, in which said state-space model is a set of mathematical quantitative equations expressing the dynamic behavior of the serial manufacturing system.

4. The method as in claim 3, in which said state-space model expresses coupling between buffer input processes, cross-coupling between input and output processes, and embedding of an operation rule to resolve contention issues.

5. The method as in claim 1, in which in step (a) processing is characterized as a function of cycle rate, time, buffer status indicator as well as workpieces being produced.

6. The method as in claim 5, in which said processing can vary linearly or discontinuously.

7. The method as in claim 1, in which in step (a) said part production is the sum of total pieces at one previous instant added to the fractional workpieces at a current instant time.

8. The method as in claim 1, in which in step (a) said storage/retrieval is a set of cross-coupled equations describing input stock and output stock.

9. The method as in claim 8, in which the input stock is equal to the sum of whole stock from previous stages added to the fractional stock completed upstream to be entered, added to the whole stock just out, added to the other sources or sinks of whole stock.

10. The method as in claim 8, in which said output stock is the sum of the whole stocks from previous stages, less fractional stock yet to be completed by the downstream machine, but added to the whole stock just in, and added to the other sources or sinks of whole stock.

11. The method as in claim 1, in which in step (a) said buffer status indicator is equal to the value of zero, if the buffer is empty or full, and equal to the value of one, otherwise.

12. The method as in claim 1, in which in step (b) said sensitivity information is a proposed change in system performance variables in response to changes in processing parameter values.

13. The method as in claim 1, in which said selected production rate is comprised of a minimum rate defined in terms of per hour/per shift.

14. The method as in claim 1, in which in step (b) said buffer space is the space for inventory between stages and can be either of the crossover type or parallel type.

15. The method as in claim 1, in which in step (c) said gradient estimator is used, which is a derivative of the function that is being optimized.