Method for correcting a signal delivered by a measuring instrument

• US 5,708,593 A
• Filed: 05/20/1996
• Issued: 01/13/1998
• Est. Priority Date: 05/19/1995
• Status: Expired due to Fees

• Current Assignee: ELF Antar France (Totalenergies Se)
• Original Assignee: ELF Antar France (Totalenergies Se)
• Inventors: Saby, Claude-Alain, Ricoux, Philippe
• Primary Examiner(s): Voeltz, Emanuel T.
• Assistant Examiner(s): Bui, Bryan
• Application Number:
  US08/650,715
• Time in Patent Office: 603 Days
• Field of Search: 364/571.04, 364/571.02, 364/481, 364/498, 364/550, 364/571.01, 324/601, 324/307, 324/76.11, 324/130, 73/1 R
• US Class Current: 702/85
• CPC Class Codes: G01N 21/274 Calibration, base line adjustment, drift correction

Abstract

A method for correcting a signal delivered by a slave measuring instrument with respect to a reference signal delivered by a master measuring instrument of the same type involved in the analyzing of the same product includes the provision of standardization signals delivered by both the slave instrument and the master instrument as they analyze calibration products, each calibration product having a particular known calibration characteristic. These standardization signals are subjected to series decomposition so that they are simplified prior to their being formed into a correction coefficient matrix useful in correcting a signal derived from a slave measuring instrument performing an actual analysis of the characteristic of a test product like the calibration products. When a test product undergoes analysis, it is analyzed by the slave measuring instrument to determine a signal which is then subject to a further series decomposition and correction with the aid of the stored correction matrix. Once corrected, the signal is subject to recomposing and a final determination of the characteristic of concern using a calibration model established relative to the master measuring instrument and the known characteristics of calibration products.

Claims

1. Method for correcting a signal delivered as a result of an analysis of a characteristic of a test product by a slave measuring instrument, said method comprising:

establishing a calibration model for a master measuring instrument of a same type as the slave measuring instrument on the basis of a plurality of calibration products each being of a same type as the test product and having a known calibration characteristic;

using said slave measuring instrument and said master measuring instrument of the same type to deliver slave standardization signals and master standardization signals as to each of the calibration products;

performing a series decomposition of each of the master instrument and slave instrument standardization signals delivered respectively by the master and slave instruments for each of the calibration products;

establishing mathematical relationships between parameters resulting from the series decompositions of the master and slave standardization signals;

using the established mathematical relationship to determine correction coefficients for the parameters resulting from the series decomposition of the slave standardization signals delivered from the slave instrument with respect to similar parameters resulting from the series decomposition of the master standardization signals delivered from the master instrument for each calibration product;

storing the correction coefficients in the form of a correction matrix;

using the slave instrument to analyze a test product;

performing a series decomposition of a resulting test product signal delivered by the slave instrument;

correcting coefficients resulting from the series decomposition of the signal delivered by the slave instrument using the stored correction matrix;

using the corrected coefficients in recomposing a corrected version of the signal delivered by the slave instrument analyzing the characteristic of the test product; and

determining the desired characteristic of the test product analyzed by the slave instrument by applying the calibration model to the corrected version of the signal delivered by the slave instrument.

2. Method according to claim 1, comprising:

verifying the validity of the correction coefficients by executing the following steps,

performing a validation series decomposition of at least one validation signal delivered by the slave instrument, said at least one validation signal being obtained from the analysis of at least one validation product,

correcting the parameters of the validation series decomposition by using the correction matrix;

recomposing the validation signal delivered by the slave instrument in order to obtain the corrected signal,

determining the value of a characteristic of the validation product using the calibration model applied to the corrected signal,

verifying that the characteristic of the validation product value is substantially equal to the value of the characteristic determined for the same validation product from the signals delivered by the master instrument using the same calibration model.

3. Method according to claim 2, wherein the slave instrument and the master instrument are the same instrument, used at different periods of time.

4. Method according to claim 2, wherein the series decompositions are carried out by using Fourier transforms and the recomposing steps are carried out by using inverse Fourier transforms.

5. Method according to claim 2, wherein the series decomposition and the recomposing steps are carried out by wavelet decomposition and reconstruction algorithms.

6. Method according to claim 2, wherein the mathematical relationships between parameters resulting from the series decompositions are linear relationships whose coefficients are estimated by a regression algorithm.

7. Method according to claim 2, wherein the mathematical relationships between the parameters resulting from the series decompositions are polynomial relationships, of order at least equal to 2, whose coefficients are estimated by a regression algorithm.

8. Method according to claim 1, wherein the slave instrument and the master instrument are the same instrument, used at different periods of time.

9. Method according to claim 8, wherein the series decompositions are carried out by using Fourier transforms and the recomposing step is carried out by using an inverse Fourier transform.

10. Method according to claim 8, wherein the series decomposition and the recomposition steps are carried out by wavelet decomposition and reconstruction algorithms.

11. Method according to claim 8, wherein the mathematical relationships between parameters resulting from the series decompositions are linear relationships whose coefficients are estimated by a regression algorithm.

12. Method according to claim 8, wherein the mathematical relationships between the parameters resulting from the series decompositions are polynomial relationships, of order at least equal to 2, whose coefficients are estimated by a regression algorithm.

13. Method according to claim 1, wherein series decompositions are carried out by using Fourier transforms and the recomposing step is carried out by using an inverse Fourier transform.

14. Method according to claim 13, wherein the mathematical relationships between parameters resulting from the series decompositions are linear relationships whose coefficients are estimated by a regression algorithm.

15. Method according to claim 1, wherein series decompositions and the recomposing steps are carried out by wavelet decomposition and reconstruction algorithms, respectively.

16. Method according to claim 1, wherein the mathematical relationships between the parameters resulting from the series decompositions are linear relationships whose coefficients are estimated by a regression algorithm.

17. Method according to claim 16, wherein the regression algorithm is a linear or nonlinear algorithm.

18. Method according to claim 1, wherein the mathematical relationships between the parameters resulting from the series decompositions are polynomial relationships, of order at least equal to 2, whose coefficients are estimated by a regression algorithm.

19. Method according to claim 1, wherein the mathematical relationships between the parameters resulting from the series decompositions are relationships chosen from the relationships representing growth curves whose coefficients are estimated by a regression algorithm.

20. Method according to claim 1, wherein the mathematical relationships between the parameters resulting from the series decompositions are represented by neural networks.