Model predictive control apparatus and methods for motion and/or pressure control of injection molding machines

• US 6,682,669 B2
• Filed: 09/29/2001
• Issued: 01/27/2004
• Est. Priority Date: 09/29/2001
• Status: Expired due to Fees

• Current Assignee: Van Dorn Demag Corp. (Sumitomo Heavy Industries Limited)
• Original Assignee: Van Dorn Demag Corp. (Sumitomo Heavy Industries Limited)
• Inventors: Bulgrin, Thomas C., Schuplin, Andrew
• Primary Examiner(s): HEITBRINK, JILL LYNNE
• Application Number:
  US09/967,731
• Publication Number:
  US 20030080452A1
• Time in Patent Office: 850 Days
• Field of Search: 264/40.1, 264/40.5, 264/328.1, 425/145, 425/149
• US Class Current: 264/40.1
• CPC Class Codes: B29C 2043/5808 pressure or compressing force
• CPC Class Codes: B29C 2945/76033 Electric current or voltage
• CPC Class Codes: B29C 2945/76525 Electric current or voltage
• CPC Class Codes: B29C 45/76 Measuring, controlling or regulating measuring in general G01; controlling or regulating in general G05
• CPC Class Codes: B29C 45/766 the setting or resetting of moulding conditions, e.g. before starting a cycle

Abstract

Methods and apparatus are disclosed for controlling an injection molding machine, wherein a control law estimates future machine states according to a proposed control output vector, the current machine state and a model of the machine, and recursively refines the control output vector using an error function and an adjustment rule in order to reduce the error between the estimated future machine states and desired future machine states. One or more control outputs are provided to actuators associated with the machine according to the refined control output vector.

Claims

1. A method for controlling motion or pressure in a non-linear injection molding machine, comprising:

obtaining a current state vector having a plurality of state values representative of a current state of said injection molding machine;

providing a proposed control output vector having a plurality of control output values representative of control outputs at a current time and at future times;

simulating a plurality of estimated future states of said injection molding machine according to said proposed control output vector using said current state vector and a model with a plurality of state equations representative of behavior of said injection molding machine;

determining an error between said plurality of estimated future states and a plurality of desired future states;

recursively refining said proposed control output vector according to an error between estimated future states and said plurality of desired future states, simulating refined estimated future states of said injection molding machine according to said refined proposed control output vector using said current state vector and said model, determining a refined error between said refined estimated future states and said desired future states until a termination condition occurs; and

providing a control output to at least one actuator in said injection molding machine according to said most recently refined proposed control output vector.

2. The method of claim 1:

wherein providing a proposed control output vector comprises providing a first proposed control output vector having a plurality of control output values representative of control outputs at said current time and at future times;

wherein simulating a plurality of estimated future states comprises simulating a first plurality of estimated future states of said injection molding machine according to said first proposed control output vector using said current state vector and said model;

wherein determining an error between said plurality of estimated future states and a plurality of desired future states comprises determining a first error between said first plurality of estimated future states and a plurality of desired future states;

wherein recursively refining said proposed control output vector, simulating refined estimated future states and, determining a refined error comprises;

providing a first refined proposed control output vector according to said first error so as to reduce said first error;

simulating a first refined plurality of estimated future states of said injection molding machine according to said first refined proposed control output vector using said current state vector and said model;

determining a first refined error between said first refined plurality of estimated future states and said plurality of desired future states; and

wherein providing said control output comprises providing a control output to said at least one actuator according to said first refined proposed control output vector.

3. The method of claim 2, further comprising:

providing a second refined proposed control output vector according to said first refined error so as to reduce said first refined error;

simulating a second refined plurality of estimated future states of said injection molding machine according to said second refined proposed control output vector using said current state vector and said model;

determining a second refined error between said second refined plurality of estimated future states and said plurality of desired future states; and

providing said control output to said at least one actuator according to said second refined proposed control output vector.

4. The method of claim 3, further comprising saving at least a portion of said second refined proposed control output vector, wherein providing a first proposed control output vector comprises providing said first proposed control output vector according to a previously saved proposed control output vector from a previous control cycle.

5. The method of claim 3, wherein said termination condition comprises one of a predetermined time elapsing, a predetermined number of iterations occurring, said most recently refined error being less than a threshold value, and a rate of change of said most recently refined error being less than a threshold value.

6. The method of claim 3, wherein obtaining said current state vector comprises:

obtaining at least one sensor input value from a sensor associated with said injection molding machine; and

estimating said current state of said injection molding machine using said model and said at least one sensor input value.

7. The method of claim 6, wherein estimating said current state of said injection molding machine comprises computing said plurality of state values representative of a current state of said injection molding machine using said model, and forming said current state vector using said computed state values.

8. The method of claim 7, wherein said model comprises at least one differential equation, and wherein computing said plurality of state values comprises solving said at least one differential equation according to one of a Runge-Kutta method and a difference equation technique.

9. The method of claim 3, wherein said model comprises at least one differential equation, and wherein simulating said plurality of estimated future states comprises solving said at least one differential equation using said proposed control output vector and said current state vector.

10. The method of claim 9, wherein solving said at least one differential equation comprises solving said at least one differential equation according to one of a Runge-Kutta method and a difference equation technique.

11. The method of claim 1, wherein said model comprises at least one differential equation, and wherein simulating said plurality of estimated future states comprises solving said at least one differential equation using said proposed control output vector and said current state vector.

12. The method of claim 11, wherein solving said at least one differential equation comprises solving said at least one differential equation according to one of a Runge-Kutta method and a difference equation technique.

13. The method of claim 1, wherein obtaining said current state vector comprises:

obtaining at least one sensor input value from a sensor associated with said injection molding machine; and

estimating said current state of said injection molding machine using said model and said at least one sensor input value.

14. The method of claim 13, wherein estimating said current state of said injection molding machine comprises computing said plurality of state values representative of a current state of said injection molding machine using said model, and forming said current state vector using said computed state values.

15. The method of claim 14, wherein said model comprises at least one differential equation, and wherein computing said plurality of state values comprises solving said at least one differential equation according to one of a Runge-Kutta method and a difference equation technique.

16. The method of claim 1, further comprising saving at least a portion of said most recently refined proposed control output vector, wherein providing a first proposed control output vector comprises providing said first proposed control output vector according to a previously saved proposed control output vector from a previous control cycle.

17. The method of claim 1, wherein said termination condition comprises one of a predetermined time elapsing, a predetermined number of iterations occurring, said most recently refined error being less than a threshold value, and a rate of change of said most recently refined error being less than a threshold value.

18. The method of claim 1, wherein refining said proposed control output vector according to an error comprises:

correlating a plurality of coefficients from said proposed control output vector according to a control output function;

evaluating a plurality of partial differential equations for error with respect to said plurality of coefficients;

adjusting at least one of said plurality of coefficients so as to reduce said error; and

translating an adjusted plurality of coefficients according to said control output function to provide said refined proposed control output vector.

19. The method of claim 18, wherein adjusting at least one of said plurality of coefficients so as to reduce said error comprises adjusting at least one of said plurality of coefficients using a conjugate gradient method so as to reduce said error.

20. The method of claim 18, further comprising saving at least a portion of said adjusted plurality of coefficients, wherein providing a first proposed control output vector comprises providing said first proposed control output vector according to a previously saved portion of said adjusted plurality of coefficients from a previous control cycle.

21. A method for controlling motion or pressure in an injection molding machine, comprising:

obtaining at least one sensor input value from a sensor associated with said injection molding machine;

estimating a current state of said injection molding machine using said at least one sensor input value and a model with a plurality of state equations representative of behavior of said injection molding machine;

providing a proposed control output vector having a plurality of control output values representative of control outputs at a current time and at future times;

simulating a plurality of estimated future states of said injection molding machine according to said proposed control output vector using said current state and said model;

determining an error between said plurality of estimated future states and a plurality of desired future states;

recursively refining said proposed control output vector according to an error between estimated future states and said plurality of desired future states, simulating refined estimated future states of said injection molding machine according to said refined proposed control output vector using said current state and said model, and determining a refined error between said refined estimated future states and said desired future states until a termination condition occurs; and

providing a control output to at least one actuator in said injection molding machine according to said most recently refined proposed control output vector.

22. The method of claim 21, wherein estimating said current state of said injection molding machine comprises:

computing a plurality of state values representative of said current state of said injection molding machine using said model; and

forming a current state vector using said computed state values.

23. The method of claim 22, wherein said model comprises at least one differential equation, and wherein computing said plurality of state values comprises solving said at least one differential equation according to one of a Runge-Kutta method and a difference equation technique.

24. The method of claim 21:

wherein providing a proposed control output vector comprises providing a first proposed control output vector having a plurality of control output values representative of control outputs at said current time and at future times;

wherein simulating a plurality of estimated future states comprises simulating a first plurality of estimated future states of said injection molding machine according to said first proposed control output vector using said current state and said model;

wherein determining an error between said plurality of estimated future states and a plurality of desired future states comprises determining a first error between said first plurality of estimated future states and a plurality of desired future states;

wherein recursively refining said proposed control output vector, simulating refined estimated future states and, determining a refined error comprises;

providing a first refined proposed control output vector according to said first error so as to reduce said first error;

simulating a first refined plurality of estimated future states of said injection molding machine according to said first refined proposed control output vector using said current state and said model;

determining a first refined error between said first refined plurality of estimated future states and said plurality of desired future states; and

wherein providing said control output comprises providing a control output to said at least one actuator according to said first refined proposed control output vector.

25. The method of claim 24, further comprising:

providing a second refined proposed control output vector according to said first refined error so as to reduce said first refined error;

simulating a second refined plurality of estimated future states of said injection molding machine according to said second refined proposed control output vector using said current state and said model;

determining a second refined error between said second refined plurality of estimated future states and said plurality of desired future states; and

providing said control output to said at least one actuator according to said second refined proposed control output vector.

26. The method of claim 25, further comprising saving at least a portion of said second refined proposed control output vector, wherein providing a first proposed control output vector comprises providing said first proposed control output vector according to a previously saved proposed control output vector from a previous control cycle.

27. The method of claim 25, wherein said termination condition comprises one of a predetermined time elapsing, a predetermined number of iterations occurring, said most recently refined error being less than a threshold value, and a rate of change of said most recently refined error being less than a threshold value.

28. The method of claim 25, wherein said model comprises at least one differential equation, and wherein simulating said plurality of estimated future states comprises solving said at least one differential equation using said proposed control output vector and said current state.

29. The method of claim 28, wherein solving said at least one differential equation comprises solving said at least one differential equation according to one of a Runge-Kutta method and a difference equation technique.

30. The method of claim 21, wherein said model comprises at least one differential equation, and wherein simulating said plurality of estimated future states comprises solving said at least one differential equation using said proposed control output vector and said current state.

31. The method of claim 30, wherein solving said at least one differential equation comprises solving said at least one differential equation according to one of a Runge-Kutta method and a difference equation technique.

32. The method of claim 21, further comprising saving at least a portion of said most recently refined proposed control output vector, wherein providing a first proposed control output vector comprises providing said first proposed control output vector according to a previously saved proposed control output vector from a previous control cycle.

33. The method of claim 21, wherein said termination condition comprises one of a predetermined time elapsing, a predetermined number of iterations occurring, said most recently refined error being less than a threshold value, and a rate of change of said most recently refined error being less than a threshold value.

34. The method of claim 25, wherein estimating said current state of said injection molding machine comprises:

computing a plurality of state values representative of said current state of said injection molding machine using said model; and

forming a current state vector using said computed state values.

35. The method of claim 34, wherein said model comprises at least one differential equation, and wherein computing said plurality of state values comprises solving said at least one differential equation according to one of a Runge-Kutta method and a difference equation technique.

36. The method of claim 21, wherein refining said proposed control output vector according to an error comprises:

correlating a plurality of coefficients from said proposed control output vector according to a control output function;

evaluating a plurality of partial differential equations for error with respect to said plurality of coefficients;

adjusting at least one of said plurality of coefficients so as to reduce said error; and

translating an adjusted plurality of coefficients according to said control output function to provide said refined proposed control output vector.

37. The method of claim 36, wherein adjusting at least one of said plurality of coefficients so as to reduce said error comprises adjusting at least one of said plurality of coefficients using a conjugate gradient method so as to reduce said error.

38. A system for controlling motion or pressure in a non-linear injection molding machine, comprising:

a model comprising a plurality of state equations representative of behavior of said injection molding machine;

a current state vector comprising a plurality of state values representative of a current state of said injection molding machine; and

a control law, comprising;

a simulator component operative to simulate a plurality of estimated future states of said injection molding machine according to a proposed control output vector having a plurality of control output values representative of control outputs at a current time and at future times using said current state vector and said model;

an error function evaluator operative to determine an error between said plurality of estimated future states and a plurality of desired future states; and

an adjustment rule operative to adjust said proposed control output vector according to said error to provide a refined proposed control output vector;

wherein said control law is operative to recursively refine said proposed control output vector using said adjustment rule, to simulate refined estimated future states using said simulator, and to determine a refined error between said refined estimated future states and said desired future states until a termination condition occurs, and wherein said control law is operative to provide a control output to at least one actuator in said injection molding machine according to a most recently refined proposed control output vector.

39. The system of claim 38, further comprising an observer receiving at least one sensor input value from a sensor associated with said injection molding machine, wherein said observer is operative to estimate said current state of said injection molding machine using said model and said at least one sensor input value.

40. The system of claim 39, wherein said model comprises at least one differential equation, and wherein said observer is operative to solve said at least one differential equation using said proposed control output vector and said current state vector.

41. The system of claim 38, wherein said model comprises at least one differential equation, and wherein said simulator is operative to solve said at least one differential equation using said proposed control output vector and said current state vector.

42. The system of claim 41, wherein said simulator is operative to solve said at least one differential equation according to one of a Runge-Kutta method and a difference equation technique.

43. The system of claim 38, wherein said control law is operative to save at least a portion of said most recently refined proposed control output vector, and wherein said control law is operative to provide a first proposed control output vector according to a previously saved proposed control output vector from a previous control cycle.

44. The system of claim 38, wherein said termination condition comprises one of a predetermined time elapsing, a predetermined number of iterations occurring, said most recently refined error being less than a threshold value, and a rate of change of said most recently refined error being less than a threshold value.

45. The system of claim 38, wherein said control law further comprises a control output function and a control output parameter translator, wherein said adjustment rule is operative to correlate a plurality of coefficients from said proposed control output vector according to said control output function, to evaluate a plurality of partial differential equations for error with respect to said plurality of coefficients, and to adjust at least one of said plurality of coefficients so as to reduce said error, and wherein said control output parameter translator is operative to translate an adjusted plurality of coefficients according to said control output function to provide said refined proposed control output vector.

46. The system of claim 45, wherein said adjustment rule is operative to adjust at least one of said plurality of coefficients using a conjugate gradient method.

47. The system of claim 38, wherein said control law provides said control output to at least one electric motor to control a motion or a pressure in said injection molding machine according to said most recently refined proposed control output vector.

48. A system for controlling motion or pressure in an injection molding machine, comprising:

a model comprising a plurality of state equations representative of behavior of said injection molding machine;

an observer receiving at least one sensor input value from a sensor associated with said injection molding machine, wherein said observer is operative to estimate said current state of said injection molding machine using said model and said at least one sensor input value to provide a current state vector comprising a plurality of state values representative of a current state of said injection molding machine; and

a control law, comprising;

a simulator component operative to simulate a plurality of estimated future states of said injection molding machine according to a proposed control output vector having a plurality of control output values representative of control outputs at a current time and at future times using said current state vector and said model;

an error function evaluator operative to determine an error between said plurality of estimated future states and a plurality of desired future states; and

an adjustment rule operative to adjust said proposed control output vector according to said error to provide a refined proposed control output vector;

wherein said control law is operative to recursively refine said proposed control output vector using said adjustment rule, to simulate refined estimated future states using said simulator, and to determine a refined error between said refined estimated future states and said desired future states until a termination condition occurs, and wherein said control law is operative to provide a control output to at least one actuator in said injection molding machine according to a most recently refined proposed control output vector.

49. The system of claim 48, wherein said model comprises at least one differential equation, and wherein said observer is operative to solve said at least one differential equation using said proposed control output vector and said current state vector.

50. The system of claim 48, wherein said model comprises at least one differential equation, and wherein said simulator is operative to solve said at least one differential equation using said proposed control output vector and said current state vector.

51. The system of claim 50, wherein said simulator is operative to solve said at least one differential equation according to one of a Runge-Kutta method and a difference equation technique.

52. The system of claim 48, wherein said control law is operative to save at least a portion of said most recently refined proposed control output vector, and wherein said control law is operative to provide a first proposed control output vector according to a previously saved proposed control output vector from a previous control cycle.

53. The system of claim 48, wherein said termination condition comprises one of a predetermined time elapsing, a predetermined number of iterations occurring, said most recently refined error being less than a threshold value, and a rate of change of said most recently refined error being less than a threshold value.

54. The system of claim 48, wherein said control law further comprises a control output function and a control output parameter translator, wherein said adjustment rule is operative to correlate a plurality of coefficients from said proposed control output vector according to said control output function, to evaluate a plurality of partial differential equations for error with respect to said plurality of coefficients, and to adjust at least one of said plurality of coefficients so as to reduce said error, and wherein said control output parameter translator is operative to translate an adjusted plurality of coefficients according to said control output function to provide said refined proposed control output vector.

55. The system of claim 54, wherein said adjustment rule is operative to adjust at least one of said plurality of coefficients using a conjugate gradient method.

56. The system of claim 48, wherein said control law provides said control output to at least one electric motor to control a motion or a pressure in said injection molding machine according to said most recently refined proposed control output vector.

57. A method for controlling motion or pressure in an injection molding machine, comprising:

providing a proposed control output vector having a plurality of control output values representative of control outputs at a current time and at future times;

simulating a plurality of estimated future states of said injection molding machine according to said proposed control output vector using a model;

determining an error between said plurality of estimated future states and a plurality of desired future states;

refining said proposed control output vector according to an error between estimated future states and said plurality of desired future states; and

providing a control output to at least one actuator in said injection molding machine according to a refined proposed control output vector.

58. The method of claim 57, wherein refining said proposed control output vector comprises recursively simulating refined estimated future states of said injection molding machine according to said refined proposed control output vector using a current state and said model, and determining a refined error between said refined estimated future states and said desired future states until a termination condition occurs.

59. A computer readable medium having computer-executable instructions for controlling motion or pressure in an injection molding machine, comprising computer-executable instructions for:

providing a proposed control output vector having a plurality of control output values representative of control outputs at a current time and at future times;

simulating a plurality of estimated future states of said injection molding machine according to said proposed control output vector using a model;

determining an error between said plurality of estimated future states and a plurality of desired future states;

refining said proposed control output vector according to an error between estimated future states and said plurality of desired future states; and

providing a control output to at least one actuator in said injection molding machine according to a refined proposed control output vector.

60. The computer readable medium of claim 59, wherein the computer-executable instructions for refining said proposed control output vector comprise computer-executable instructions for recursively simulating refined estimated future states of said injection molding machine according to said refined proposed control output vector using a current state and said model, and computer-executable instructions for determining a refined error between said refined estimated future states and said desired future states until a termination condition occurs.