US 5,396,440 A · Optimum calibration frequency determination

Abstract

One aspect of the invention is a method for determining whether systematic biases are present in an instrument. The method involves first obtaining a series of replicates produced by an instrument. Then, a Fourier Transform is performed on that data, thereby generating a secondary spectrum which indicates the frequency characteristics of the data. This secondary spectrum is then compared to secondary spectra obtained in a similar matter from a series of random number sets to determine whether variations observed are within an expected range as defined by the random number series. Another aspect of the invention is a method for determining an optimum frequency at which the instrument should be calibrated. This method comprises obtaining a set of multivariate data as before, and treating every xth data point as a calibration point and adjusting the values of the points following the calibration point accordingly. A Fourier Transform is then calculated as before. Different values for x are tried, and an optimum calibration frequency is deemed to be a value for x in a region which produces a point which is most similar to the distribution of points corresponding to the series of random number sets.

Claims16

Dependent claims

A method for evaluating a sample comprising:

using an analytical instrument selected from the group consisting of chromatography, spectroscopy, flowmeters and balances to measure a physical property of the sample;
correlating the physical property measured using the analytical instrument with results obtained from analyzing standards having known properties under similar conditions;
periodically determining whether biases are present in the analytical instrument;
and
recalibrating the analytical instrument and reanalyzing the sample if biases are determined;
wherein the step of determining whether biases are present comprises:
(a) obtaining a set of multivariate data generated by the analytical instrument;
(b) performing a Fourier Transform for the data in step (a) thereby generating a spectrum which indicates the frequency characteristics of the data;
(c) obtaining a series of random number sets having a similar number of elements to the set of multivariate data obtained in step (a);
(d) performing a Fourier Transform for each of the random number sets in step (c) thereby generating a series of spectra which indicate the frequency characteristics of the random number sets;
and
(e) identifying biases in the instrument whenever the Fourier Transforms obtained in step (b) are not within the expected range of random distributions given the Fourier Transforms obtained in step (d).

9 more elements

A method for determining an optimum calibration frequency for an analytical instrument comprising:

(a) obtaining a set of data from a series of replicate analyses by the instrument;
(b) obtaining a series of random number sets, each set having a similar number of elements as the set of data obtained in step (a);
(c) performing a Fast Fourier Transform on each of the sets in steps (a) and (b), thereby generating secondary spectra;
(d) using a pattern recognition technique to determine how different the secondary spectrum corresponding to the data obtained in step (a) is from the distribution of the secondary spectra corresponding to the random number sets generated in step (b);
(e) treating every xth element in the set of multivariate data obtained in step (a) as a calibration point and repeating step (d), where x is an integer greater than one;
(f) repeating step e for a number of values for x;
and
(g) selecting an optimum calibration frequency for the instrument from a range of values for x which produces a relatively small difference between the secondary spectrum corresponding to the data obtained in step (a) and the distribution of secondary spectra corresponding to the random number sets generated in step (b).

5 more elements

A method of operating an analytical instrument comprising:

calibrating the instrument using known standards so that the response of the instrument can be correlated to a physical property of unknown sample;
analyzing a series of unknown samples;
recalibrating the instrument at an optimum calibration frequency, wherein said optimum calibration frequency is determined by
(a) obtaining a set of data from a series of replicate analyses by the instrument;
(b) obtaining a series of random number sets, each set having a similar number of elements as the set of data obtained in step (a);
(c) performing a Fourier Transform on each of the sets in steps (a) and (b), thereby generating secondary spectra;
(d) using a pattern recognition technique to determine how different the secondary spectrum corresponding to the data obtained in step (a) is from the distribution of the secondary spectra corresponding to the random number sets generated in step (b);
(e) treating every ith element in the set of multivariate data obtained in step (a) as a calibration point and repeating step (d), where i is an integer greater than one;
(f) repeating step e for a number of values for i;
and
(g) selecting as an optimum calibration frequency for the instrument, a value of i from a range of values for i which produces a relatively small difference between the secondary spectrum corresponding to the data obtained in step (a) and the distribution of secondary spectra corresponding to the random number sets generated in step (b).

8 more elements

Dependencies are inferred from the claim text.

Optimum calibration frequency determination

Life of this patent

Optimum calibration frequency determination

Life of this patent

Abstract

Claims16