Device and method for assessing a quality class of an object to be tested

• US 7,873,518 B2
• Filed: 11/10/2006
• Issued: 01/18/2011
• Est. Priority Date: 05/13/2004
• Status: Expired due to Fees

• Current Assignee: Fraunhofer Gesellschaft Zur Foerderung Der Angewandten Forsching E.V.
• Original Assignee: Fraunhofer Gesellschaft Zur Foerderung Der Angewandten Forsching E.V., Technische Universitat Munchen
• Inventors: Eichner, Matthias, Tschoepe, Constanze, Wolff, Matthias, Hoffmann, Ruediger, Hentschel, Dieter
• Primary Examiner(s): Jackson; Jakieda R
• Application Number:
  US11/558,816
• Publication Number:
  US 20070100623A1
• Time in Patent Office: 1,530 Days
• Field of Search: 704256-2565
• US Class Current: 704/256.1
• CPC Class Codes: G01N 29/045 by imparting shocks to the workpiece and detecting the vibrations or the acoustic waves caused by the shocks measuring resonant frequency G01H13/00; measuring strength properties by application of mechanical stress G01N3/00
• CPC Class Codes: G01N 29/14 using acoustic emission techniques echo of particles G01N29/046; measuring mechanical vibrations or acoustic waves in solids in general G01H1/00
• CPC Class Codes: G01N 29/30 Arrangements for calibrating or comparing, e.g. with standard objects
• CPC Class Codes: G01N 29/4445 Classification of defects
• CPC Class Codes: G01N 29/4472 Mathematical theories or simulation
• CPC Class Codes: G01N 29/449 Statistical methods not provided for in G01N29/4409, e.g. averaging, smoothing and interpolation

Abstract

A device for assessing a quality class of an object to be tested includes a unit for detecting a test signal from the object to be tested. Furthermore, the device for assessing includes a unit for providing a stochastic Markov model including states and transitions between states on the basis of reference measurements of objects of known quality classes, and a unit for evaluating the test signal using the stochastic Markov model. In addition, the device for assessing includes a unit for associating the object to be tested with a quality class based on the evaluation of the test signal. Such a device has the advantage to be able to perform a more precise association of an object to be tested with a quality class as compared to prior art.

Claims

1. A device for assessing a quality class of a device under test, comprising:

a unit for detecting an acoustic non-voice test signal from the device under test;

a unit for providing a stochastic Markov model including states and transitions between states on the basis of reference measurements of devices of known quality classes, wherein, in the stochastic Markov model, a single probability density distribution is associated with each state, and wherein several states have transitions from a common predecessor state;

a unit for evaluating the acoustic non-voice test signal using the stochastic Markov model; and

a unit for associating the device under test with a quality class based on the stochastic Markov model evaluation of the acoustic non-voice test signal.

2. The device for assessing according to claim 1, further comprising a unit for applying an acoustic excitation signal to the device under test, wherein the unit for detecting is designed to detect the test signal in response to the acoustic excitation signal.

3. The device for assessing according to claim 1, wherein the unit for providing is designed to receive reference measurement data of devices of known quality classes to form states and transitions between the states in the stochastic Markov model based on the reference measurement data.

4. The device for assessing according to claim 3, wherein the unit for providing is designed to associate one transition probability each with the transitions between the states, wherein the unit for providing is further designed to prevent a transition between a first state and a second state if the transition probability from the first state to the second state is less than a predetermined threshold.

5. The device for assessing according to claim 3, wherein the unit for providing is designed to split a state into a plurality of states.

6. The device for assessing according to claim 1, wherein the unit for detecting is designed to extract, when detecting the test signal, the same into a primary n-th dimension feature space and to transform the primary n-th dimension feature space to a secondary m-th dimension feature space, wherein m and n are natural numbers meeting the condition that m is larger than n.

7. The device for assessing according to claim 1, wherein the unit for providing is designed to assign a probability density function and a feature vector formed from signal values of the test signal to each state of the stochastic Markov model.

8. The device for assessing according to claim 7, wherein the unit for evaluating is designed to form a log-likelihood between the probability density function of the state and the feature vector assigned to the state.

9. The device for assessing according to claim 7, wherein the unit for evaluating is designed to perform the stochastic Markov model evaluation on the basis of the Viterbi algorithm.

10. The device for assessing according to claim 1, wherein the unit for evaluating is designed to output a classification value for determining a quality class, and the unit for associating is designed to use the classification value for associating the device under test with the quality class, wherein the unit for associating is further designed to use a deviation between the classification value and a value characteristic for the quality class as confidence measure for a similarity of the device under test with respect to a quality class.

11. The device for assessing according to claim 1, wherein the unit for providing is designed to provide a first Markov model associated with a first quality class and to provide a second stochastic Markov model associated with a second quality class including second states and second transitions between the second states on the basis of reference measurements of devices of known quality classes, wherein the unit for evaluating the test signal is designed to use the second stochastic Markov model for an evaluation of the test signal into the first quality class or the second quality class.

12. The device for assessing according to claim 1, wherein the unit for associating is designed to perform the association of the device under test with the quality class on the basis of a decision threshold.

13. The device for assessing according to claim 1, wherein the acoustic non-voice test signal comprises an ultrasonic signal.

14. The device for assessing according to claim 1, wherein the known quality classes include good and bad.

15. The device for assessing according to claim 1, wherein the unit for providing is configured to generate a stochastic Markov graph for each quality class in an automatic manner based on reference measurement data of devices of known quality classes, wherein generating the stochastic Markov graph comprises changing the number and arrangement of states and transitions thereof during a training phase.

16. A method for assessing a quality class of a device under test, comprising:

detecting an acoustic non-voice test signal from the device under test;

providing a stochastic Markov model including states and transitions between states on the basis of reference measurements of devices of known quality classes, wherein, in the stochastic Markov model, a single probability density distribution is associated with each state, and wherein several states have transitions from a common predecessor state;

evaluating the acoustic non-voice test signal using the stochastic Markov model; and

associating the device under test with a quality class based on the stochastic Markov model evaluation of the acoustic non-voice test signal.

17. The method according to claim 16, wherein the acoustic non-voice test signal comprises an ultrasonic signal.

18. The method according to claim 16, wherein the known quality classes include good and bad.

19. The method for assessing according to claim 16, wherein providing a stochastic Markov model comprises generating a stochastic Markov graph for each quality class in an automatic manner based on reference measurement data of devices of known quality classes, wherein generating the stochastic Markov graph comprises changing the number and arrangement of states and transitions thereof during a training phase.

20. A non-transitory computer readable medium encoded with a computer program for assessing a quality class of a device under test , when the program runs on a computer, comprising:

detecting an acoustic non-voice test signal from the device under test;

providing a stochastic Markov model including states and transitions between states on the basis of reference measurements of devices of known quality classes, wherein, in the stochastic Markov model, a single probability density distribution is associated with each state, and wherein several states have transitions from a common predecessor state;

evaluating the acoustic non-voice test signal using the stochastic Markov model; and

associating the device under test with a quality class based on the stochastic Markov model evaluation of the acoustic non-voice test signal.

21. The computer readable medium according to claim 20, wherein the acoustic non-voice test signal comprises an ultrasonic signal.

22. The computer readable medium according to claim 20, wherein the known quality classes include good and bad.

23. The computer readable medium according to claim 20, wherein providing a stochastic Markov model comprises generating a stochastic Markov graph for each quality class in an automatic manner based on reference measurement data of devices of known quality classes, wherein generating the stochastic Markov graph comprises changing the number and arrangement of states and transitions thereof during a training phase.