Optimization of a design

• US 7,260,516 B2
• Filed: 08/31/2002
• Issued: 08/21/2007
• Est. Priority Date: 08/31/2001
• Status: Active Grant

• Current Assignee: Optimum Power Technology LP
• Original Assignee: Optimum Power Technology LP
• Inventors: Crandall, John G., Chatfield, Glen F.
• Primary Examiner(s): Phan; Thai
• Assistant Examiner(s): LUU, CUONG V
• Application Number:
  US10/231,338
• Publication Number:
  US 20030046049A1
• Time in Patent Office: 1,816 Days
• Field of Search: 702/105, 702/27, 324/239, 703/2, 342/378, 716/8, 700/44, 706/22
• US Class Current: 703/14
• CPC Class Codes: F02D 2041/1423 Identification of model or controller parameters
• CPC Class Codes: F02D 2041/1433 using a model or simulation of the system
• CPC Class Codes: F02D 41/1406 with use of a optimisation method, e.g. iteration
• CPC Class Codes: G05B 13/042 in which a parameter or coefficient is automatically adjusted to optimise the performance
• CPC Class Codes: G05B 17/02 electric
• CPC Class Codes: G06F 2111/06 Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
• CPC Class Codes: G06F 30/00 Computer-aided design [CAD]
• 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

An apparatus, system, and method for optimization. The apparatus, system, and method include a mechanism for ending an optimization when all models within one design tolerance of an optimum model have been simulated.

Claims

1. A computer readable medium containing instructions that, when executed by a processor, cause the processor to perform a method for finding at least one local optimum model, the method comprising:

a. specifying a base model that includes a variable having a value,

b. specifying a goal which identifies a characteristic of the base model that is to be optimized;

c. specifying a maximum number of optimizations to be performed;

d. specifying a minimum value and a maximum value for the variable;

e. specifying a list of different values for the variable between the minimum and maximum values for the variable;

f. creating a model for every value of the variable included in the list;

g. running each model created to determine a value of the characteristic for each model;

h. comparing each model run with its adjacent model or models;

i. identifying each model as a local model if the model has a better value of the characteristic than its adjacent model or models;

j. ranking all local models identified;

k. optimizing each local model to find a local optimum for each local model up to the maximum number of optimizations specified; and

l. identifying and storing the at least one local optimum model.

2. A computer readable medium containing instructions that, when executed by a processor, cause the processor to perform a method for finding at least one local optimum model, the method comprising:

a. specifying a base model that includes two or more variables each having a value,

b. specifying a goal which identifies a characteristic of the base model that is to be optimized;

c. specifying a maximum number of optimizations to be performed;

d. specifying for each variable, a minimum value and a maximum value;

e. specifying a list of different values for each variable between their minimum and maximum values;

f. creating a model for combinations of the values of the variables included in the list;

g. running each model created to determine a value of the characteristic for each model;

h. comparing each model run with its adjacent model or models;

i. identifying each model as a local model if the model has a better value of the characteristic than its adjacent model or models;

j. ranking all local models identified;

k. optimizing the highest ranked local models to find the local optimum model up to the maximum number of optimizations specified; and

l. storing the local optimum model for each of the highest ranked local models.

3. A computer readable medium containing instructions that, when executed by a processor, cause the processor to perform an optimization method, the method comprising:

a. specifying a characteristic to be optimized and a desired value for the characteristic;

b. specifying a variable to be varied during the optimization;

c. setting a design tolerance equal to a minimum increment by which the variable is to be varied;

d. selecting a design configuration that includes a value for the variable and all other values necessary to simulate the design configuration as a base design configuration;

e. simulating the design configuration to arrive at a value for the characteristic;

f. selecting design configurations having variable values adjacent to the base design configuration variable value;

g. simulating the selected design configurations;

h. setting the base design configuration variable value to the design configuration value of an adjacent design configuration having a characteristic value nearest to the desired characteristic value if an adjacent design configuration has a characteristic value nearer to the desired characteristic than the base design configuration;

i. selecting design configurations having variable values more nearly adjacent to the base design configuration variable value if no adjacent design configuration has a characteristic value nearer to the desired characteristic value than the base design configuration;

j. repeating steps g, h, and i until all design configurations having variable values adjacent to the base design configuration by one design tolerance have been simulated and the characteristic value of the base design configuration is nearer the desired characteristic value than any other design configuration simulated; and

k. storing the base design configuration variable value.

4. The computer readable medium of claim 3, wherein the desired value of the goal is a plurality of values.

5. The computer readable medium of claim 3, wherein a second variable is varied during optimization.

6. The computer readable medium of claim 3, wherein adjacent variable values include a variable value greater than the base design configuration variable value and a variable value less than the base design configuration variable value.

7. The computer readable medium of claim 6, wherein the variable value greater than the base design configuration variable value is equal to the base design configuration variable value plus an increment and the variable value less than the base design configuration variable value is equal to the base design configuration variable value less the increment.

8. The computer readable medium of claim 7, wherein the increment is reduced each time step i is performed.

9. A computer readable medium containing instructions that, when executed by a processor, cause the processor to perform a method for selecting a best value for a variable in a model, the method comprising:

a. specifying a model that includes an input variable having a tolerance and a step having an initial magnitude greater than the tolerance, and specifying an output characteristic having an objective;

b. running the model with the input variable set to a selected value to arrive at a resulting output characteristic;

c. running the model with the input variable set to the selected value changed by the step to arrive at a second resulting output characteristic;

d. selecting the input variable value that results in an output characteristic that more nearly conforms to the objective;

e. reducing the step if the selected input variable value is unchanged at d;

f. setting the step equal to the tolerance if the step would otherwise be reduced to less than the tolerance;

g. repeating c through f until the step reaches the tolerance; and

h. storing the selected input variable value.

10. A computer readable medium containing instructions that, when executed by a processor, cause the processor to perform a method for selecting a best value for a variable in a model, the method comprising:

a. specifying a model that includes an input variable having a tolerance and a step having an initial magnitude greater than the tolerance, and specifying an output characteristic having an objective;

b. running the model with the input variable set to a selected value to arrive at a resulting output characteristic;

c. running the model with the input variable set to the selected value changed by the step to arrive at a second resulting output characteristic;

d. selecting the input variable value that results in an output characteristic that more nearly conforms to the objective;

e. selecting the input variable value changed by a step if the output characteristic resulting when the selected input variable is set to the selected value and the output characteristic resulting when the selected input variable is set to the selected value changed by the step are the same;

f. reducing the step if the selected input variable value is unchanged at d and e;

g. repeating c through f until the step reaches the tolerance; and

h. storing the selected input variable value.

11. A computer readable medium containing instructions that, when executed by a processor, cause the processor to perform a method for selecting a best value for a variable in a model, the method comprising:

a. specifying a model that includes an input variable having a tolerance and a step having an initial magnitude greater than the tolerance, and specifying an output characteristic having an objective;

b. running the model with the input variable set to a selected value to arrive at a resulting output characteristic;

c. running the model with the input variable set to the selected value changed by the step to arrive at a second resulting output characteristic;

d. selecting the input variable value that results in an output characteristic that more nearly conforms to the objective, wherein the output characteristic that more nearly conforms to the objective is the output characteristic having the largest value;

e. reducing the step if the selected input variable value is unchanged at d;

f. repeating c through e until the step reaches the tolerance; and

g. storing the selected input variable value.

12. A computer readable medium containing instructions that, when executed by a processor, cause the processor to perform a method for selecting a best value for a variable in a model, the method comprising:

a. specifying a model that includes an input variable having a tolerance and a step that is an integer multiple of the tolerance and has an initial magnitude greater than the tolerance, and specifying an output characteristic having an objective;

b. running the model with the input variable set to a selected value to arrive at a resulting output characteristic;

c. running the model with the input variable set to the selected value changed by the step to arrive at a second resulting output characteristic;

d. selecting the input variable value that results in an output characteristic that more nearly conforms to the objective;

e. reducing the step if the selected input variable value is unchanged at d;

f. repeating c through e until the step reaches the tolerance; and

g. storing the selected input variable value.

13. A computer readable medium containing instructions that, when executed by a processor, cause the processor to perform a method for selecting a best value for a variable in a model, the method comprising:

a. specifying a model that includes a first input variable having a first tolerance and a first step having an initial magnitude greater than the first tolerance, specifying a second input variable having a second tolerance equal to the first tolerance and a second step equal to the first step, the second step having an initial magnitude equal to the first initial magnitude and greater than the second tolerance, and specifying an output characteristic having an objective;

b. running the model with the first input variable set to a first selected value and the second input variable set to a second selected value to arrive at a resulting output characteristic;

c. running the model with the first input variable set to the first selected value changed by the first step and the second input variable set to the second selected value changed by the second step to arrive at a second resulting output characteristic;

d. selecting the first input variable value and the second input variable value that result in an output characteristic that more nearly conforms to the objective;

e. reducing the first step and the second step if the first selected input variable value and the second selected input variable value are both unchanged at d;

f. repeating c through e until the first step reaches the first tolerance and the second step reaches the second tolerance; and

g. storing the selected first input variable value and the selected second input variable value.

14. A computer readable medium containing instructions that, when executed by a processor, cause the processor to perform a method for selecting a best value for a variable in a model, the method comprising:

a. specifying a model that includes an input variable having a tolerance and a step having an initial magnitude greater than the tolerance, wherein the input variable is a dimension of a component of a machine and the model simulates the machine including that component, and specifying an output characteristic having an objective;

b. running the model with the input variable set to a selected value to arrive at a resulting output characteristic;

c. running the model with the input variable set to the selected value changed by the step to arrive at a second resulting output characteristic;

d. selecting the input variable value that results in an output characteristic that more nearly conforms to the objective;

e. reducing the step if the selected input variable value is unchanged at d;

f. repeating c through e until the step reaches the tolerance; and

g. storing the selected input variable value.

15. A computer readable medium containing instructions that, when executed by a processor, cause the processor to perform a method for finding a variable value that results in an optimum value for an output characteristic by varying the magnitude of the variable value, the method comprising:

a. setting the variable to a base value;

b. setting a tolerance to a minimum step;

c. determining a first output characteristic value with the variable set to the base value;

d. determining a second output characteristic value with the variable set to the base value plus a whole number multiple of the tolerance;

e. choosing one of the first output characteristic value and the second output characteristic value, whichever is nearest a desired output characteristic value;

dimension of a component of a machine and the model simulates the machine including that component, and specifying an output characteristic having an objective;

b. running the model with the input variable set to a selected value to arrive at a resulting output characteristic;

c. running the model with the input variable set to the selected value changed by the step to arrive at a second resulting output characteristic;

d. selecting the input variable value that results in an output characteristic that more nearly conforms to the objective;

e. reducing the step if the selected input variable value is unchanged at d;

f. repeating c through e until the step reaches the tolerance;

g. storing the selected input variable value.

f. setting the base value to the variable value providing the chosen output characteristic value;

g. reducing the whole number multiple if the base value is unchanged;

h. repeating d through g until the whole number multiple is one; and

i. storing the variable value.

16. The computer readable medium of claim 15, wherein the whole number is a negative number.

17. The computer readable medium of claim 15, wherein the whole number is a positive number and further comprising determining a third goal value with the variable set to the base value minus the whole number multiplied by the tolerance.

18. The computer readable medium of claim 15, wherein setting the base value to the variable value providing the chosen output characteristic value includes leaving the base value at the current value if the base value was previously set to the chosen output characteristic value.

19. The computer readable medium of claim 15, further comprising limiting the number of values determined for the goal to a predefined number of goal values.

20. The computer readable medium of claim 15, wherein the variables for which goal values are to be calculated are selected randomly.

21. The computer readable medium of claim 20, wherein the variables for which goal values are to be calculated are selected based on a Monte Carlo form of selection.

22. The computer readable medium of claim 20, wherein the selection utilizes a seed.

23. The computer readable medium of claim 15, further comprising setting the variable to the base value plus a reduced whole number multiple of the tolerance if the first goal value is nearer to the desired goal value than the second goal value.

24. The computer readable medium of claim 15, further comprising setting the variable to the second goal value plus the whole number multiple of the tolerance if the second goal value is nearer to the desired goal value than the first goal value.

25. The computer readable medium of claim 15, wherein determining an output characteristic value includes simulating the output characteristic with the value to which the variable is set.

26. The computer readable medium of claim 15, further comprising a display depicting the proximity of the output characteristic value nearest to the desired output characteristic value.

27. The computer readable medium of claim 26, wherein a user may terminate the optimization method.

28. The computer readable medium of claim 15, further comprising:

setting a second variable to a second base value; and

wherein the first output characteristic value is determined with the variable set to the base value and the second variable set to the second base value.

29. The computer readable medium of claim 28, wherein the tolerance for the variable is different than a tolerance utilized with the second variable.

30. The computer readable medium of claim 15, further comprising setting a minimum boundary for the variable below which optimization will not be performed.

31. The computer readable medium of claim 15, further comprising setting a maximum boundary for the variable above which optimization will not be performed.

32. The computer readable medium of claim 15, wherein at least a second output characteristic is optimized simultaneously.

33. The computer readable medium of claim 15, wherein the variable value is a dimension of a component of a machine and the output characteristic is an operating characteristic of the machine.

34. The computer readable medium of claim 15, wherein the desired output characteristic value is a predetermined value.

35. The computer readable medium of claim 15, wherein the desired output characteristic value is a predetermined set of values.

36. The computer readable medium of claim 15, wherein the optimum output characteristic value is one that nearly matches a predefined value for the output characteristic.

37. The computer readable medium of claim 15, wherein the desired output characteristic value is a greatest value determined for the output characteristic.

38. The computer readable medium of claim 15, wherein the desired output characteristic value is a lowest value determined for the output characteristic.

39. The computer readable medium of claim 15, wherein the step is a minimum amount the variable is desired to be varied.

40. The computer readable medium of claim 15, wherein the step is a predetermined value.