Method for monitoring the correspondence of a beer sample with a reference beer

Abstract

In a method for monitoring the correspondence of a beer sample with a reference beer, at least 15 reference beer samples of the reference beer are brewed with the same ingredients and the same process parameters. Measurement signals for the absorption spectrum of the reference beer samples are captured and a principal component analysis is carried out for the measurement signals, in which at least 15 principal components are ascertained. A factor loading P.sub.R(i,j) is respectively determined for each principal component for the individual reference beer samples and a reference value (I) is ascertained, where i denotes the reference beer sample and j denotes the principal component, .sub.R(j) refers to the mean value of all factor loadings of the j-th principal component and .sub.P(j) refers to the standard deviation of these factor loadings. A reference interval (II) is formed, where n denotes the number of reference beer samples, m denotes the number of principal components, .sub.R(j) denotes the standard deviation of all reference values of the j-th principal component and k denotes a constant not equal to zero. A measurement signal is captured for the absorption spectrum of the beer sample and the factor loadings P.sub.B(i) of this measurement signal are determined for the principal components ascertained for the reference beer samples and a characteristic (III) is formed and compared to the reference interval. Should the characteristic B lie outside of the reference interval, a fault during the production of the beer sample is indicated.

Claims

1. A method for monitoring the correspondence of a beer sample with a reference beer that is assigned to the same beer variety as the beer sample, characterized in that at least 15 reference beer samples of the reference beer were brewed with the same ingredients and the same process parameters, that infrared absorption spectroscopy is used to capture measurement signals for the absorption spectrum of the individual reference beer samples and a principal component analysis is performed for the measurement signals, in which at least 15 principal components are ascertained and a factor loading P.sub.R(i,j) is respectively determined for each principal component for the individual reference beer samples, wherein i denotes the reference beer sample and j denotes the principal component, that a reference value $R\left(i,j\right)=.Math.\frac{P_R\left(i,j\right)-{}_P\left(j\right)}{{}_P\left(j\right)}.Math.$ is respectively calculated from the factor loadings P.sub.R(i,j) for each reference beer sample and for each principle component, wherein .sub.p(j) refers to the mean of all factor loadings of the j.sup.th principal component and .sub.P(j) refers to the standard deviation of these factor loadings, that a reference interval $\left[\frac{k}{n}\underset{i=1}{\overset{n}{.Math.}}\underset{j=1}{\overset{m}{.Math.}}R\left(i,j\right)-k\underset{j=1}{\overset{m}{.Math.}}{}_R\left(j\right)-\frac{k}{n}\underset{i=1}{\overset{n}{.Math.}}\underset{j=1}{\overset{m}{.Math.}}R\left(i,j\right)+k\underset{j=1}{\overset{m}{.Math.}}{}_R\left(j\right)\right]$ is formed, wherein n is the number of reference beer samples, m is the number of principal components, .sub.R(j) is the standard deviation of all reference values of the j.sup.th principal component and k is a constant not equal to zero, that infrared absorption spectroscopy is used to capture a measurement signal for the absorption spectrum of the beer sample to be checked for correspondence with the reference beer and the factor loadings P.sub.B(i) of this measurement signal are determined for the principal components ascertained for the reference beer samples and from these factor loadings P.sub.B(i), from the means .sub.p(j) of the factor loadings of the reference beer samples for the individual principal components, and from the standard deviations .sub.P(j) of these factor loadings, a characteristic value $B=k\underset{j=1}{\overset{m}{.Math.}}.Math.\frac{P_B\left(j\right)-{}_P\left(j\right)}{{}_P\left(j\right)}.Math.$ is formed and compared to the reference interval, and that an error during the production of the beer sample is indicated should the characteristic value B lie outside of the reference interval.

2. The method according to claim 1, characterized in that the number n of reference beer samples is greater than or equal to the number m of principal components, in particular twice as great and preferably at least three times as great as the latter.

3. The method according to claim 1, characterized in that the number m of principal components is at least 20, optionally at least 30, in particular at least 40, and preferably at least 50.

4. The method according to claim 1, characterized in that the constant k corresponds to the reciprocal of the number m of the principal components.

5. The method according to claim 1, characterized in that during the infrared absorption spectroscopy, the reference beer samples and the beer sample are irradiated with infrared radiation, the wave number of which covers the range between 950 and 3050, in particular between 960 and 2000, and preferably between 980 and 1200.

Description

DESCRIPTION OF THE INVENTION

(1) An exemplary embodiment of the invention will be explained in more detail in the following.

(2) In the exemplary embodiment, for monitoring the correspondence of a beer sample to be tested with a reference beer, which is assigned to the same beer variety as the beer sample, 100 reference beer samples of the reference beer are brewed with the same ingredients and the same process parameters in each case.

(3) Using a QFOOD QUANTOS-type infrared absorption spectrometer, measurement signals are captured for the absorption spectra of the 100 reference beer samples in a wave number range extending from the wave number 980 to the wave number 1200. Each measurement signal respectively comprises 1000 value combinations, each of which has at least one value for the wave number and a value assigned to this wave number for the optical infrared absorption of the reference beer sample.

(4) A principal component analysis is performed for the 100 measurement signals or spectra captured in this manner, in which 30 principal components are ascertained using a software known per se. A factor loading P.sub.R(i,j) is respectively determined for each of the 30 principal components for the 100 reference beer samples. The index i denotes the reference beer sample and the index j denotes the principal component. This gives rise to 3,000 factor loadings P.sub.R(i,j) in total, only a few of which are shown below for the sake of clarity:

(5) TABLE-US-00001 j = 1 j = 2 j = 3 . . . j = 29 j = 30 i = 1 0.1230054 0.0026305 0.0003498 . . . 0.00000534 0.00000243 i = 2 0.1242563 0.0026599 0.0003300 . . . 0.00000015 0.00000127 i = 3 0.1293215 0.0019354 0.0000951 . . . 0.00000968 0.00000739 i = 98 0.1294580 0.0003328 0.00057318 . . . 0.00000786 0.00000556 i = 99 0.1286656 0.0008917 0.00045053 . . . 0.00000487 0.00000637 i = 100 0.1309384 0.0004471 0.00042969 . . . 0.00000069 0.00000744

(6) The mean .sub.P(j) of all factor loadings and the standard deviation .sub.P(j) of these factor loadings are respectively determined for each of the 30 principal components:

(7) TABLE-US-00002 j .sub.P(j) .sub.P(j) 1 0.12897300 0.00196521 2 0.00115020 0.00124260 3 0.00052527 0.00039921 28 0.00000017708 0.0000102710 29 0.00000015864 0.0000089216 30 0.00000016190 0.0000060188

(8) For each factor loading P.sub.R(i,j), respectively, a positive reference value R(i,j) is calculated:

(9) TABLE-US-00003 i R(i, 1) R(i, 2) R(i, 3) . . . R(i, 29) R(i, 30) 1 3.03662653 1.19134758 2.19197238 . . . 0.61632769 0.43063337 2 2.40007829 1.21500130 2.14247494 . . . 0.00063217 0.23790443 3 0.17732606 0.63194497 1.07755101 . . . 1.10278655 1.20091677 98 0.24679333 0.65777029 0.12001654 . . . 0.89878767 0.95066921 99 0.15641331 0.20799952 0.18721578 . . . 0.56364666 1.08524718 100 1.00010092 0.56582482 0.23942857 . . . 0.09467341 1.26302301

(10) according to the absolute value formula

(11) $R\left(i,j\right)=.Math.\frac{P_R\left(i,j\right)-{}_P\left(j\right)}{{}_P\left(j\right)}.Math.$

(12) For each main component, respectively, the standard deviation .sub.R(i) over all 100 reference values of the relevant principal component is furthermore determined:

(13) TABLE-US-00004 i .sub.R(i) 1 0.64967043 2 0.54158537 3 0.57910532 28 0.60052400 29 0.72734766 30 0.63994051

(14) From the reference values R(i,j) and standard deviations .sub.R(i) thus obtained, a reference interval: [21.133360112.0965087 . . . 21.1333601+12.0965087]=[9.0368514 . . . 33.2298688]

(15) is formed according to the formula:

(16) $\left[\frac{1}{100.Math.30}\underset{i=1}{\overset{100}{.Math.}}\underset{j=1}{\overset{30}{.Math.}}R\left(i,j\right)-\frac{1}{30}\underset{i=1}{\overset{30}{.Math.}}{}_R\left(j\right)-\frac{1}{100.Math.30}\underset{i=1}{\overset{100}{.Math.}}\underset{j=1}{\overset{30}{.Math.}}R\left(i,j\right)+\frac{1}{30}\underset{j=1}{\overset{30}{.Math.}}{}_R\left(j\right)\right]$

(17) The beer sample to be checked for correspondence with the reference beer is provided in a further method step. Using the QFOOD QUANTOS-type infrared absorption spectrometer, a measurement signal for the absorption spectrum of the beer sample is captured in the same wave number range as the one in which the absorption spectra of the reference beer samples were measured.

(18) The factor loadings P.sub.B(i) of this measurement signal are determined for the 30 principal components ascertained for the reference beer samples:

(19) TABLE-US-00005 i .sub.R(i) 1 0.64967043 2 0.54158537 3 0.57910532 28 0.60052400 29 0.72734766 30 0.63994051

(20) The factor loadings thus obtained are normalized by subtracting the mean .sub.P(j) of all factor loadings of the reference beer samples for the relevant principal component from the relevant factor loading P.sub.B(j) and dividing the result, in absolute value, of this subtraction by the standard deviation .sub.P(j) of these factor loadings:

(21) $.Math.\frac{P_B\left(j\right)-{}_P\left(j\right)}{{}_P\left(j\right)}.Math.$

(22) The arithmetic mean is calculated from the normalized factor loadings thus obtained in order to form a characteristic value B for the beer sample:

(23) $B=\frac{1}{30}\underset{j=1}{\overset{m}{.Math.}}.Math.\frac{P_B\left(j\right)-{}_P\left(j\right)}{{}_P\left(j\right)}.Math.=52.28199576$

(24) This characteristic value B is compared to the reference interval [9.0368514 . . . 33.2298688]. The characteristic value B lies outside of the reference interval, thus indicating an error during the production of the beer sample.