Method and apparatus for predicting a characteristic of a product attribute formed by a machining process using a model of the process

• US 6,850,874 B1
• Filed: 04/17/1998
• Issued: 02/01/2005
• Est. Priority Date: 04/17/1998
• Status: Active Grant

• Current Assignee: United Technologies Corporation (Raytheon Technologies Corporation d/b/a RTX)
• Original Assignee: United Technologies Corporation (Raytheon Technologies Corporation d/b/a RTX)
• Inventors: Higuerey, Evelitsa E., Schweizerhof, Aaron L.
• Primary Examiner(s): JONES, HUGH M
• Application Number:
  US09/061,644
• Time in Patent Office: 2,482 Days
• Field of Search: 395/500.27, 395/500.25, 703/4, 703/6
• US Class Current: 703/4
• CPC Class Codes: B23H 7/20 for programme-control, e.g. adaptive programme-control systems in general G05B19/00
• CPC Class Codes: B23H 9/10 Working turbine blades or nozzles
• CPC Class Codes: B23H 9/14 Making holes
• CPC Class Codes: G05B 13/027 using neural networks only

Abstract

The present invention includes at least one sensor that senses the characteristic of one or more parameters associated with a stochastic machining process and provides one or more measurement signals indicative thereof, and further includes a signal processor having a model indicative of a correspondence between the characteristic of the parameters associated with the stochastic machining process and the characteristic of the resulting product attribute, and uses the model to generate a signal indicative of the predicted characteristic of the attribute of the product.

Claims

1. An apparatus for predicting a characteristic of an attribute of a workpiece, the attribute having been produced as a result of a stochastic machining process, the apparatus comprising:

at least one sensor that senses a characteristic of at least one parameter associated with the stochastic machining process and provides at least one measurement signal indicative thereof; and

signal processor means, responsive to the at least one measurement signal, having a model indicative of a correspondence between the characteristic of the at least one parameter associated with the stochastic machining process and the characteristic of the resulting product attribute, and using said model to generate a signal indicative of the characteristic of the attribute of the workpiece to predict the characteristic of the attribute.

2. The apparatus of claim 1, wherein the signal processing means comprises a trained artificial neural network for generating the signal indicative of the characteristic of the attribute of the workpiece.

3. The apparatus of claim 1, wherein the signal processing means comprises

signal analyzer means, responsive to the at least one measurement signal, for generating at least one analyzer signal representative of a line fit to the magnitude of the at least one measurement signal during the stochastic machining process;

and

model means, responsive to the at least one analyzer signal, for generating the signal indicative of the characteristic of the attribute of the workpiece in accordance with a predetermined correspondence between lines fit to the magnitude of the at least one measurement signal and characteristics of attributes resulting therefrom.

4. The apparatus of claim 3, wherein the at least one analyzer signal comprises a signal indicative of a slope and a signal indicative of an intercept for the magnitude of the at least one measurement signal during the stochastic machining process.

5. The apparatus of claim 1, wherein the stochastic machining process is an electrical discharge machining process, the at least one sensor comprises at least one sensor that senses a characteristic of at least one parameter associated with the electrical discharge machining process and provides at least one measurement signal indicative thereof, and the signal processing means comprises means for generating the signal indicative of the characteristic of the attribute in accordance with a predetermined correspondence between characteristics of the at least one parameter of the electrical discharge machining process and characteristics of attributes resulting therefrom.

6. The apparatus of claim 5, wherein the at least one process parameter includes a voltage differential between an electrode and the workpiece, and further includes a current between the electrode and the workpiece.

7. The apparatus of claim 6, wherein the at least one process parameter further includes a distance between the electrode and the workpiece.

8. The apparatus of claim 7, wherein the attribute comprises at least one passage and the signal processing means comprises means for generating the signal indicative of the characteristic of the attribute in accordance with a predetermined correspondence between characteristics of the at least one parameter of the electrical discharge machining process and characteristics of at least one passage resulting therefrom.

9. The apparatus of claim 1, wherein the attribute comprises at least one passage and the signal processing means comprises means for generating the signal indicative of the characteristic of the attribute in accordance with a predetermined correspondence between characteristics of the at least one parameter of the electrical discharge machining process and characteristics of at least one passage resulting therefrom.

10. The apparatus of claim 9, wherein the signal indicative of the characteristic of the attribute comprises a signal indicative of a breakthrough condition for the at least one passage.

11. The apparatus of claim 9, wherein the signal indicative of the characteristic of the attribute comprises a signal indicative of an airflow measure for the at least one passage.

12. A method for predicting a characteristic of an attribute of a workpiece, the attribute having been produced as a result of a stochastic machining process, the method comprising the steps of:

sensing a characteristic of at least one parameter associated with the stochastic machining process and providing at least one measurement signal indicative thereof; and

generating, in response to the at least one measurement signal, a predicted signal indicative of the characteristic of the attribute of the workpiece, said predicted signal being generated by a model.

13. The method of claim 12, wherein the step of generating comprises training an artificial neural network included in said model.

14. The method of claim 12, wherein the step of generating comprises:

generating at least one analyzer signal representative of a line fit to the magnitude of the at least one measurement signal during the stochastic machining process; and

generating the signal indicative of the characteristic of the attribute of the workpiece in accordance with a predetermined correspondence between lines fit to the magnitude of the at least one measurement signal and characteristics of attributes resulting therefrom.

15. The method of claim 14, wherein the step of generating the predicted signal indicative of the characteristic of the attribute comprises generating a signal indicative of a slope and a signal indicative of an intercept for the magnitude of the at least one measurement signal during the stochastic machining process.

16. The method of claim 12, wherein the stochastic machining process is an electrical discharge machining process, the step of sensing comprises sensing a characteristic of at least one parameter associated with the electrical discharge machining process and providing at least one measurement signal indicative thereof, and the step of generating comprises generating the signal indicative of the characteristic of the attribute in accordance with a predetermined correspondence between characteristics of the at least one parameter of the electrical discharge machining process and characteristics of attributes resulting therefrom.

17. The method of claim 16, wherein the step of sensing includes sensing a voltage differential between an electrode and the workpiece and further includes sensing a current between the electrode and the workpiece.

18. The method of claim 17, wherein the step of sensing further includes sensing a distance between the electrode and the workpiece.

19. The method of claim 18, wherein the attribute comprises at least one passage and the step of generating comprises generating the signal indicative of the characteristic of the attribute in accordance with a predetermined correspondence between characteristics of the at least one parameter of the electrical discharge machining process and characteristics of at least one passage resulting therefrom.

20. The method of claim 12, wherein the attribute comprises at least one-passage and the step of generating comprises generating the signal indicative of the characteristic of the attribute in accordance with a predetermined correspondence between characteristics of the at least one parameter of the electrical discharge machining process and characteristics of at least one passage resulting therefrom.

21. The method of claim 20, wherein the step of generating the signal indicative of the characteristic of the attribute comprises generating a signal indicative of a breakthrough condition for the at least one passage.

22. The method of claim 20, wherein the step of generating the signal indicative of the characteristic of the attribute comprises generating a signal indicative of an airflow measure for the at least one passage.

23. An apparatus for predicting a characteristic of an attribute of a product, said product being fabricated by a stochastic machining process, said apparatus comprising:

a plurality of sensors for sensing a plurality of parameters associated with said stochastic machining process, said plurality of sensors generating a plurality of measurement signals indicative of said plurality of parameters; and

a signal processor receiving said plurality of measurement signals from said plurality of sensors, said signal processor including a process model that based on information from said plurality of measurement signals predicts a characteristic of an attribute of said product.

24. The apparatus according to claim 23, wherein said apparatus predicts airflow characteristics of cooling holes of an airfoil.

25. The apparatus according to claim 23, wherein said apparatus predicts break-through characteristics of cooling holes of an airfoil.

26. The apparatus according to claim 23, wherein said stochastic machining process is an electrical discharge machining (EDM) process.

27. The apparatus according to claim 23, wherein said apparatus predicts said characteristic of said attribute prior to a defect being machined during said stochastic machining process.

28. The apparatus according to claim 23, wherein said plurality of parameters are voltage, current and displacement.

29. The apparatus according to claim 23, wherein one of said plurality of parameters is voltage.

30. The apparatus according to claim 23, wherein one of said plurality of parameters is current.

31. The apparatus according to claim 23, wherein one of said plurality of parameters is displacement.

32. The apparatus according to claim 23, wherein one of said plurality of sensors is a voltage sensor.

33. The apparatus according to claim 23, wherein one of said plurality of sensors is a current sensor.

34. The apparatus according to claim 23, wherein one of said plurality of sensors is a displacement sensor.

35. The apparatus according to claim 23, wherein said model includes at least one neural network for predicting said characteristic of said attribute.

36. The apparatus according to claim 35, wherein said neural network is a break-through neural network for predicting a break-through for cooling holes of an airfoil.

37. The apparatus according to claim 35, wherein said neural network is an airflow neural network for predicting an airflow for cooling holes of an airfoil.

38. The apparatus according to claim 23, wherein said model includes a first neural network for predicting break-through for cooling holes of an airfoil and a second neural network for predicting an airflow for said cooling holes of said airfoil.

39. The apparatus according to claim 23, wherein said model has been trained.

40. The apparatus according to claim 23, wherein said signal processor comprises:

an input portion for conditioning said plurality of parameters and generating a plurality of conditioned signals; and

a processing portion including a signal analyzer for analyzing said conditioned signals received from said input portion and generating analyzed signals, said analyzed signals being communicated to said model.

41. The apparatus according to claim 40, wherein said signal analyzer provides a slope signal and an intercept signal for each of said conditioned signals.

42. A method for predicting a characteristic of an attribute of a workpiece during a stochastic machining process of said workpiece, said method comprising the steps of:

sensing at least one parameter associated with said stochastic machining process;

communicating said at least one parameter to a model;

processing said at least one parameter within said model; and

generating a predicted characteristic of an attribute of said workpiece.

43. The method of claim 42 further comprising a preceding step of:

training said model.

44. The method of claim 42 further comprising a preceding step of:

creating said model using a plurality of data sets that represent actual stochastic machining process inputs and outputs.

45. The method of claim 42 wherein said model includes at least one neural network.

46. The method of claim 42 further comprising subsequent steps of:

determining said predicted characteristic of said attribute in real-time; and

preventing potential defect from being manufactured into said workpiece.

47. The method of claim 42 further comprising an intermediate step of:

conditioning said at least one parameter prior to communicating said at least one parameter to said model.

48. The method of claim 47 further comprising an intermediate step of:

analyzing a signal subsequent conditioning said at least one parameter and prior to communicating said at least one parameter to said model.

49. The method of claim 47 further comprising an intermediate step of:

providing a slope signal and an intercept signal to said model subsequent to said step of conditioning said at least one parameter.