Self-optimizing method and machine

• US 4,472,770 A
• Filed: 09/29/1982
• Issued: 09/18/1984
• Est. Priority Date: 06/22/1981
• Status: Expired due to Term

• Current Assignee: Chou H. Li
• Original Assignee: Chou H. Li
• Inventors: Li, Chou H.
• Primary Examiner(s): Ruggiero, Joseph F.
• Application Number:
  US06/428,277
• Time in Patent Office: 720 Days
• Field of Search: 364/148-166, 364/178, 364/179, 364/425, 364/442, 364/479, 364/497, 364/431.04, 364/560, 364/501, 364/502
• US Class Current: 700/47
• CPC Class Codes: F02B 2075/027 four
• CPC Class Codes: G05B 13/0205 not using a model or a simulator of the controlled system
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202

Abstract

The invention discloses method for self-optimizing, dynamically and in real time, a machine relative to a specific performance characteristic in response to variations on a prescribed number m of variables. This method comprises statistically designing a matrix of operating conditions of the machine; sequentially operating the machine according to the designed matrix of operating conditions; testing the performance characteristic of the thus-operated machine; analyzing the tested performance characteristic against a given optimizing criterion; and feeding back for or returning to the designing step for a new, improved operating condition. A machine in the form of programmed microprocessor embodying this self-optimizing method is also disclosed.

Claims

1. A method for self-optimizing, in real time, a machine relative to a specific performance characteristic in response to variations on a prescribed number m of variables or parameters comprising:

programming the machine to cause it to plan, by itself and without human control and intervention, a statistical design which prescribes the number n of tests and the exact design matrix in which the test level of each of the m variables is specified for all the n tests;

according to the design matrix substantially continuously performing said n designed tests on the very particular machine without relying on extrapolations based on sampled test results obtained on other similar but often different or even irrelevant machines so as to eliminate errors due to sampling and extrapolation from machine to machine, part to part, and time instant to time instant;

substantially continuously collecting and analyzing the resultant n sets of test data on the test levels of the m variables and the associated performance characteristic to determine the functional relationship between said variables and said specific performance characteristic;

from the functional relationship determining the combination of the conditions of said m variables that gives optimum value for the specific performance characteristic for the particular machine, environment, and time; and

setting the conditions or levels of said m variables at their respective, thusdetermined conditions.

2. The method as in claim 1 wherein the programming is done to cause the machine itself to plan said statistical design which reduces many fold the number of combinationally required tests, the calculating steps, and the computer time thereby enabling the experimental range to be increased many fold to handle a multitude of variables and, more importantly, to achieve real-time self-optimization.

3. The method as in claim 2 wherein said machine is a machine for manufacturing a prespecified product and including substantially continuously manufacturing said product on said machine, said collecting and analyzing step comprising collecting and analyzing quality test data on the thus-manufactured product.

4. The method as in claim 1 wherein all said machine statistical planning, performing, collecting and analyzing, determining, and setting steps are done within a minute.

5. The method as in claim 4 wherein all said steps are done within a second and yet m exceeds 25.

6. The method as in claim 2 wherein said machine is a machine for manufacturing a prespecified product and including checking the quality of the resultant product around the clock and reself-optimizing the machine when the quality is not optimal.

7. The method as in claim 2 including repeating all said performing, collecting and analyzing, determining, and setting steps within every minute so that the machine conditions are always maintained optimal.

8. The method as in claim 2 including providing a second self-optimizing machine similar to said machine;

and telecommunicating the optimizing procedures including said functional relationship and said combination of conditions, from said second machine to said machine, thereby reducing the time required for said machine to achieve self-optimization.

14. The method as in claim 1 for practice on the machine in conjunction with a second, physically independent but telecommunicable machine having similar but not exactly identical performance characteristics under the same operation condition as the machine;

and including the additional step of telecommunicating from the machine to the second machine as to the calculated singular combination for starting up the self-optimizing procedure on the second machine thereby reducing the time required to achieve the similar, singular combination on the second machine.

16. The product of method claim 1.

9. A machine for self-optimizing, in real time, a specific performance characteristic in response to variations on a prescribed number m of variables comprising:

means for planning a statistical design which prescribes the number n of tests and the exact design matrix in which the test level of each of the m variables is specified for all the n tests;

means for substantially continuously performing said n designed tests on the very particular machine without relying on extrapolations based on sampled test results obtained on other similar but often different or even irrelevant machines so as to eliminate errors due to sampling and extrapolation from machine to machine, part to part, and time instant to time instant;

means for substantially continuously collecting and analyzing the resultant n sets of test data on the test levels of the m variables and the associated performance characteristic to determine the functional relationship between said variables and said performance characteristic;

means for determining the combination of the conditions of said m variables that gives optimum value for the specific performance characteristic for the particular machine, environment, and time; and

means for setting the conditions or levels of said m variables at their respective, thus-determined conditions.

10. The machine as in claim 9 wherein said planning means plans said statistical design on the machine itself without human control and intervention.

11. The machine as in claim 10 wherein said planning means plans said statistical design which reduces many fold the number of combinationally required tests, the calculating steps, and the computer time thereby enabling the experimental range to be increased many fold to handle a multitude of variables and, most importantly, to achieve real-time optimization.

12. The machine as in claim 11 in the form of a manufacturing machine for a perspecified product and said collecting and analyzing means comprising means for collecting and analyzing quality test data on the thus-manufactured product.

13. The machine as in claim 11 wherein said planning, performing, collecting and analyzing, determining, and setting means jointly perform their tasks with a cycle time of less than a minute and yet m exceeds 25;

and including means for checking the product quality around the clock to determine if additional self-optimizing cycles are needed.

15. The machine as in claim 9 including a second, physically independent but telecommunicable machine having similar but not exactly identical performance characteristics under the same operation condition as the machine;

and including also means for telecommunicating from the machine to the second machine as to the optimizing combination of the conditions of said m variables for starting up the self-optimizing method on the second machine thereby reducing the time required to achieve the similar, optimizing combination on the second machine.