JP5614620B2 - Liquid inspection method and liquid inspection apparatus - Google Patents

Description

  The present invention relates to a method and an apparatus for inspecting a liquid filled in a light transmissive container, and more particularly, to contain explosives, explosive materials, or illegal drugs in a liquid filled in a light transmissive container. The present invention relates to an inspection method and an inspection apparatus for inspecting a situation.

  In recent years, there have been many bombing attacks targeting public facilities and public transportation, including the 2005 bombings in Britain. Recently, terrorists who pretend to be passengers fill liquids that contain explosives and explosive materials, etc. or dissolve them in light-transmitting beverage containers such as plastic bottles and glass bottles, and bring them into aircraft. Cases are increasing. In addition, there are an increasing number of cases where smuggling is carried out by filling a light-transmitting container with a liquid in which illegal drugs such as narcotics and stimulants are dissolved.

  With regard to bringing explosives, explosive materials, or illegal drugs into the aircraft, baggage inspections are conducted for passengers at airports from the viewpoint of preventing incidents such as terrorism and smuggling. In order to process passengers, it is necessary to conduct a quick inspection, and it is possible to determine whether or not the liquid filled in the container contains explosives, explosive materials, or illegal drugs in a short inspection. It's not easy.

  Under such circumstances, as a method for detecting the liquid filled from the outside of the container without opening the container, a detection method for quickly determining whether or not the fuel is a flammable liquid such as gasoline has already been proposed. (Patent Document 1, Non-Patent Document 1). In the detection method shown here, there is a difference in dielectric constant between water, which is the main component of a regularly filled beverage, and liquid combustibles such as gasoline, specifically, gasoline compared to the dielectric constant of water. It is determined whether or not it is a dangerous substance by utilizing the fact that the dielectric constant is small.

  Recently, however, the use of dangerous materials whose permittivity is close to that of water is increasing. For example, in the case of the above-mentioned incident in the United Kingdom, a mixed solution of hydrogen peroxide and acetone is used as an explosive. It was. As described above, the detection method shown in Non-Patent Document 1 is not an effective method for detecting a liquid dangerous substance having a dielectric constant close to that of water.

  On the other hand, in the United States, an apparatus that can detect hydrogen peroxide solution from the outside of a light-transmitting container by using Raman spectroscopy is commercially available. However, in the case of using Raman spectroscopy, since the fluorescence of the light-transmitting container and the filled liquid is strong, almost no sensitivity can be obtained, and there is almost no practical deployment.

JP 2005-274255 A

Tokyo Gas Co., Ltd., Tokyo Gas Engineering Co., Ltd., “Release of Bottled Liquid Inspection System SLC-211D”, [online], November 1, 2004, [Search September 22, 2008], Internet, (URL: http://www.tokyo-gas.co.jp/Press/20041101.html)

  In this way, there is no appropriate method for detecting various explosives, explosive raw materials, or illegal drugs in a short time, so aircraft that are particularly important in public facilities and public transportation where terrorism countermeasures are required In reality, it is currently prohibited to bring in drinking water.

  Therefore, prior to the present invention, the present applicant promptly determines the content of explosives, explosive materials, or illegal drugs in the liquid filled in a light transmissive container such as a PET bottle or a glass bottle from the outside of the container. A patent application was filed for an inspection method and an inspection apparatus capable of reliably detecting (Japanese Patent Application No. 2008-259789).

  The outline of the patent application prior to the present invention (hereinafter referred to as “prior application invention”) will be described below.

That is, the prior application invention
A liquid inspection method for inspecting the content of explosives, explosive raw materials and / or illegal drugs in a liquid filled in a light transmissive container,
From the outside of the container, a near infrared light irradiation step of irradiating the liquid with near infrared light,
A near-infrared light receiving step for receiving the near-infrared light transmitted through the liquid or the near-infrared light scattered by the liquid;
An absorption spectrum analysis step of analyzing the absorption spectrum of the received near-infrared light,
A liquid inspection method (Claim 1) characterized by analyzing the absorption spectrum and inspecting the content of explosives and explosive materials and / or illegal drugs in the liquid filled in the container. An inspection apparatus having these steps as means (claim 8).

  That is, water shows a large absorption for light, but absorption is relatively small for near infrared light. For this reason, the content of dangerous substances can be inspected by irradiating the liquid containing the dangerous substances with near infrared light and analyzing the obtained absorption spectrum. Further, in the case of near-infrared light, there is no problem that the fluorescence of containers and liquids is strong and sensitivity cannot be obtained unlike Raman spectroscopy, and an absorption spectrum that can be sufficiently analyzed can be obtained.

  Even if it is a dangerous substance with the same characteristics as water, such as hydrogen peroxide, the content of the dangerous substance can be accurately determined by devising a method for analyzing the absorption spectrum of the obtained near-infrared light. Can be inspected.

  Also, for example, even if the liquid filled in the container is different from water, tea, juice, cola, coffee, etc., it is possible to easily inspect the contents of dangerous substances by preparing each genuine absorption spectrum. be able to. Further, the same applies to the container, and it is possible to easily inspect the contents of dangerous substances in the liquid by preparing each genuine absorption spectrum regardless of whether it is colorless or colored.

  And by using the said inspection method, while specifying the kind of explosives, explosive raw material, and / or a fraudulent drug (Claim 2), the light absorbency with respect to the predetermined wavelength in the absorption spectrum analyzed by the absorption-spectrum analysis process, The explosives, explosive raw materials and / or the concentration of illegal drugs are substituted into concentration estimation formulas created based on the absorption spectra analyzed using a plurality of liquids having known concentrations, and the explosives, explosive raw materials and The concentration of the illegal drug is measured (Claim 3).

  This makes it possible to quickly and reliably detect the content of explosives, explosive materials, or illegal drugs in a liquid filled in a light transmissive container such as a PET bottle or a glass bottle from the outside of the container. .

  Thus, in the prior application invention, a concentration estimation formula is created based on an absorption spectrum that is analyzed in advance using a plurality of liquids having a known concentration of explosives, explosive material, and / or illegal drugs, We are inspecting.

  For this reason, in the prior invention, absorption spectra were obtained for all of the products. However, with the diversification of products such as the material and shape of the container, the type of contents, the concentration, and the amount of liquid, the absorption spectrum fluctuates, which may reduce the accuracy of concentration estimation. In addition, obtaining absorption spectra for all products and creating a concentration estimation formula in this way are burdensome both in terms of time and cost. In addition, it was a heavy burden at the time of inspection.

  Therefore, the present invention enables inspection corresponding to diversifying products with less data while using the above-described prior application invention, and can increase estimation accuracy in concentration estimation. It is an object to provide an inspection method and an inspection apparatus.

As a result of intensive studies, the present inventor has found that the above problems can be solved by the following technologies , and has completed the present invention. Hereinafter, each technique will be described separately.

The first technology is
A liquid inspection method for inspecting the content of explosives, explosive raw materials and / or illegal drugs in a liquid filled in a light transmissive container,
From the outside of the container, a near infrared light irradiation step of irradiating the liquid with near infrared light,
A near-infrared light receiving step for receiving the near-infrared light transmitted through the liquid or the near-infrared light scattered by the liquid;
An absorption spectrum analysis step of examining the content of explosives, explosive raw materials and / or illegal drugs in the liquid filled in the container by analyzing the absorption spectrum of the received near-infrared light. ,
After the absorption spectrum analysis step is classified according to a plurality of discriminants derived by multivariate analysis of the absorption spectrum, a plurality of explosives, explosive raw materials and / or illegal drugs in advance having a plurality of known concentrations. The liquid inspection method is a step of measuring the concentration of the explosive, the explosive material, and / or the illegal drug by substituting into a concentration estimation formula created based on the absorption spectrum of the liquid.

  By classifying the absorption spectrum into multiple discriminants derived by multivariate analysis, the test target to which the concentration estimation formula is applied can be narrowed down. The number of data for creating a concentration measurement formula (calibration curve) can also be limited. At this time, since the spectrum of each data is within the limited range of the liquid substance, the estimation error can be reduced, that is, the estimation accuracy can be increased in the concentration estimation.

The second technology is
The liquid inspection method according to the first technique , wherein the multivariate analysis is principal component analysis or discriminant analysis.

Principal component analysis and discriminant analysis can be efficiently analyzed while reducing information loss, and can be effectively applied to the present technology .

The third technology is
The liquid inspection method according to the first technique or the second technique , wherein the explosive material is hydrogen peroxide.

Hydrogen peroxide water is similar to water in physical, chemical, and optical properties, and it is difficult to inspect in a short time by the conventional inspection method, so it is easy to be filled in a beverage container. The present technology can accurately inspect such hydrogen peroxide solution, and can exhibit the effects of the present technology remarkably.

The fourth technology is
The liquid inspection according to the third technique , wherein the absorption spectrum analysis step includes a determination step of determining the presence / absence of a liquid combustible using the main component obtained by the main component analysis. Is the method.

  Through various experiments and examinations by the present inventor, it has been found that by using a main component obtained by principal component analysis, an inspection object having a liquid combustible material can be separated in a series of inspection steps. Then, by separating the inspection object having the liquid combustible material in this way, it is possible to limit the number of data of the inspection object that needs to be subjected to concentration estimation, thereby enabling efficient inspection.

The fifth technology is
In the third technique or the fourth technique , the absorption spectrum analysis step includes a determination step of determining the presence or absence of hydrogen peroxide using the main component obtained by the main component analysis. The liquid inspection method described.

  Similarly, it was found that by using the principal component obtained by principal component analysis, it is possible to reliably determine whether or not the inspection object contains hydrogen peroxide. By removing inspection objects that have been found not to contain hydrogen peroxide, it is possible to limit inspection objects that need to be subjected to concentration estimation, thereby enabling efficient inspection.

Note that it is preferable to combine the determination step in the present technology with the determination step described in the fourth technology because the inspection target that needs to be subjected to concentration estimation can be further limited. Any one of these combinations may be performed first, but it is preferable to perform the determination step described in the fourth technique first.

The sixth technology is
In the third to fifth techniques , the absorption spectrum analysis process includes a discrimination process for determining a grouping based on a liquid amount using a principal component obtained by principal component analysis. The liquid inspection method according to any one of the above.

  According to the study by the present inventor, it was found that the spectrum greatly fluctuates depending on the amount of the liquid to be inspected, which greatly affects the concentration estimation. Therefore, when the main components obtained by the principal component analysis are used to perform grouping according to the liquid volume, the influence of spectrum fluctuations can be suppressed in the concentration estimation, and more accurate inspection can be performed. I understood.

Note that the discrimination step in the present technology may also be combined with the above-described discrimination steps.

The seventh technology
The absorption spectrum analysis step comprises
A first determination step of determining the presence or absence of a liquid combustible using the principal component obtained by the first principal component analysis;
A second discriminating step for determining the presence or absence of hydrogen peroxide using the principal component obtained by the second principal component analysis;
A liquid inspection method according to the third technique , comprising: a third determination step of determining a grouping according to the liquid amount using the principal component obtained by the third principal component analysis. It is.

By performing the discriminating step described in the fourth technology , the discriminating step described in the fifth technology , and the discriminating step described in the sixth technology in this order, an inspection object having a liquid combustible material and hydrogen peroxide are included. Since the inspection targets that have not been removed are removed in advance, the inspection targets that need to be grouped can be limited, and thereafter, highly accurate concentration estimation can be performed for each group.

The seventh technology is
The absorption spectrum analysis step is the first technology to the inspection method of the liquid in any crab described seventh technical, characterized in that said an absorption spectrum analysis step using the second derivative spectrum.

  By applying the second derivative to the absorption spectrum, overlapping absorption bands can be separated, and the background effect can be removed, so that subtle changes in the absorption spectrum can be clearly captured. . As a result, it is possible to further reduce the estimation error in the density estimation.

The ninth technology is
The concentration estimation equation is the first technology to the inspection method of the liquid in any crab described eighth technology characterized by being prepared by using a regression analysis.

  By creating a concentration estimation formula using regression analysis, it is possible to incorporate spectral fluctuations into the data based on fluctuation factors such as the material and shape of the container, the type of contents, the concentration, and the liquid volume. High inspection can be performed.

The tenth technology is
The liquid inspection method according to the ninth technique , wherein the regression analysis is a PLS regression analysis.

In this technique , since PLS regression analysis is used, wavelength selection is not necessary, and data of all wavelengths can be used to perform inspections that are resistant to spectrum fluctuations and have high prediction performance (high estimation accuracy). it can.

The eleventh technology
A liquid inspection device for inspecting the content of explosives, explosive raw materials and / or illegal drugs in a liquid filled in a light transmissive container,
Near-infrared light irradiation means for irradiating the liquid with near-infrared light from outside the container;
A near infrared light receiving means for receiving the near infrared light transmitted through the liquid or the near infrared light scattered by the liquid;
Absorption spectrum analysis means for examining the content of explosives, explosive raw materials and / or illegal drugs in the liquid filled in the container by analyzing the absorption spectrum of the received near-infrared light. ,
After the absorption spectrum analysis means separates the plurality of discriminants derived by multivariate analysis of the absorption spectrum, a plurality of explosives, explosive raw materials and / or illegal drugs have a plurality of known concentrations. A liquid inspection apparatus configured to measure a concentration of the explosive, explosive material, and / or illegal drug by substituting into a concentration estimation formula created based on the absorption spectrum of the liquid is there.

This technology is a technology that captures the first technology, which is the technology of the inspection method, from the aspect of the inspection device. By using the liquid inspection device according to this technology , explosives, explosive materials, and / or illegal drugs Can be accurately inspected. In addition, the inspection time can be shortened.

The present invention is an invention based on the above findings, and the invention according to claim 1
  A liquid inspection method for inspecting the content of explosives, explosive raw materials and / or illegal drugs in a liquid filled in a light transmissive container,
  From the outside of the container, a near infrared light irradiation step of irradiating the liquid with near infrared light,
  A near-infrared light receiving step for receiving the near-infrared light transmitted through the liquid or the near-infrared light scattered by the liquid;
  An absorption spectrum analysis step of examining the content of explosives, explosive raw materials and / or illegal drugs in the liquid filled in the container by analyzing the absorption spectrum of the received near-infrared light. ,
  After the absorption spectrum analysis step separates the liquid filled in the container according to a plurality of discriminants derived by multivariate analysis of the absorption spectrum, the explosive material, explosive material and / or Substituting the concentration estimation formula created based on the absorption spectra of a plurality of liquids having a known concentration of the illegal drug, and measuring the concentration of the explosive, the explosive material and / or the illegal drug,
  The multivariate analysis is principal component analysis or discriminant analysis;
  The absorption spectrum analysis step includes a determination step of determining grouping by liquid volume using the principal component obtained by principal component analysis.
This is a liquid inspection method.

The invention described in claim 2
2. The liquid inspection method according to claim 1, wherein the explosive material is hydrogen peroxide.

The invention according to claim 3
3. The liquid inspection method according to claim 2, wherein the absorption spectrum analysis step includes a determination step of determining the presence or absence of a liquid combustible using the main component obtained by the main component analysis. It is.

The invention according to claim 4
  The said absorption spectrum analysis process has the discrimination | determination process which determines the presence or absence of hydrogen peroxide using the main component obtained by the main component analysis, The Claim 2 or Claim 3 characterized by the above-mentioned. This is a liquid inspection method.

The invention described in claim 5
  The absorption spectrum analysis step comprises
  A first determination step of determining the presence or absence of a liquid combustible using the principal component obtained by the first principal component analysis;
  A second discriminating step for determining the presence or absence of hydrogen peroxide using the principal component obtained by the second principal component analysis;
  A third discriminating step for determining a grouping according to the amount of liquid using the principal component obtained by the third principal component analysis;
The liquid inspection method according to claim 2, further comprising:

The invention described in claim 6
  6. The liquid inspection method according to claim 1, wherein the absorption spectrum analysis step is an analysis step using the absorption spectrum and its second derivative spectrum.

The invention described in claim 7
  The liquid concentration inspection method according to any one of claims 1 to 6, wherein the concentration estimation formula is created by using regression analysis.

The invention according to claim 8 provides:
  8. The liquid inspection method according to claim 7, wherein the regression analysis is a PLS regression analysis.

The invention according to claim 9 is:
  A liquid inspection device for inspecting the content of explosives, explosive raw materials and / or illegal drugs in a liquid filled in a light transmissive container,
  Near-infrared light irradiation means for irradiating the liquid with near-infrared light from outside the container;
  A near infrared light receiving means for receiving the near infrared light transmitted through the liquid or the near infrared light scattered by the liquid;
  Absorption spectrum analysis means for examining the content of explosives, explosive raw materials and / or illegal drugs in the liquid filled in the container by analyzing the absorption spectrum of the received near-infrared light. ,
  After the absorption spectrum analysis means separates the liquid filled in the container according to a plurality of discriminants derived by multivariate analysis of the absorption spectrum, the explosive material, explosive material and / or It is configured to measure the concentration of the explosive, the explosive material, and / or the illegal drug by substituting the concentration estimation formula created based on the absorption spectra of a plurality of liquids having known concentrations of the illegal drug. And
  The multivariate analysis is principal component analysis or discriminant analysis;
  The absorption spectrum analyzing means has a discriminating means for determining grouping according to the liquid amount using the principal component obtained by principal component analysis.
This is a liquid inspection apparatus.

  According to the present invention, it is possible to quickly and surely detect the contents of explosives, explosive materials, or illegal drugs in a liquid filled in a light transmissive container such as a PET bottle or a glass bottle from the outside of the container. To provide an inspection method and an inspection apparatus capable of performing inspection corresponding to diversified products with less data and capable of increasing estimation accuracy in concentration estimation. it can.

It is a figure which shows the structure of the principal part of the liquid test | inspection apparatus of one embodiment of this invention. It is a flowchart explaining the algorithm of dangerous goods detection in one embodiment of this invention. It is a figure explaining the procedure which calculates | requires the density | concentration estimation formula of hydrogen peroxide by PLS regression analysis. It is a figure which shows the secondary differential absorption spectrum of various samples, and its principal component analysis result. It is a figure which shows the score plot and loading plot of the 2nd main component of an absorbance spectrum, and the 5th main component of a secondary differential spectrum about the sample which excluded the liquid combustible material. It is a figure which shows the principal component analysis result of the 2nd derivative absorption spectrum of the sample which excluded black drinking water and a part of drinking water. It is a figure which shows the regression vector in a class A PLS regression analysis, and a calibration curve. It is a figure which shows the regression vector in a class B PLS regression analysis, and a calibration curve. It is a figure which shows the hydrogen peroxide concentration measurement result of class A and class B. It is a regression model in PLS regression analysis.

  Hereinafter, the present invention will be described based on embodiments. Note that the present invention is not limited to the following embodiments. Various modifications can be made to the following embodiments within the same and equivalent scope as the present invention.

1. Liquid Inspection Device First, the configuration of the liquid inspection device according to the present embodiment will be described with reference to FIG. FIG. 1 is a diagram schematically showing a main part of the liquid inspection apparatus according to the present embodiment. In FIG. 1, 10 is a light source, 11 is a condenser lens, 12 is a shutter, 21 is an irradiation optical fiber, 22 is a light receiving optical fiber, and 23 is a calibration optical fiber. , 30 is a container mounting table, 31 is a ring-shaped irradiation unit, 32 is a light receiving unit, 40 is a light receiving diffraction unit, 41 is a filter, 42 is a slit, and 43 is an optical grating. , 44 is a photoelectric conversion sensor for spectrum, 50 is a personal computer (PC), and 70 is a plastic bottle. A broken line in the PET bottle 70 is light incident on the light receiving unit 32 due to reflection and scattering.

  Light including near-infrared light emitted from the light source 10 (hereinafter also simply referred to as “light”) passes through the shutter 12, and the ring on the upper part of the container mounting table 30 is efficiently collected by the condenser lens 11 and the irradiation optical fiber 21. It is guided to the shape irradiation unit 31 and irradiated from the bottom surface of the plastic bottle 70 to the inside. A part of the light scattered therein enters the light receiving part 32 at the center of the upper part of the container mounting table 30 and is guided to the light receiving / diffractive part 40 through the light receiving optical fiber 22, and the filter 41 and the slit 42. After reaching the optical grating 43 through spectrum decomposition, the spectrum is decomposed and converted into a corresponding electric signal by the photoelectric conversion sensor 44 for spectrum, and the electric signal is input to the PC 50. The PC 50 performs a mathematical process according to a program and data input in advance, and inspects the liquid in the PET bottle 70.

  In order to obtain an accurate absorption spectrum of the light reflected inside the PET bottle 70, the reference light from the light source 10 can also be input to the light receiving diffraction unit 40 and further to the PC 50 via the calibration optical fiber 23. .

2. Inspection Method Next, an inspection procedure using the liquid inspection apparatus will be described. A plastic bottle is used as the container.

(1) Placement of Sample When the PET bottle 70 to be inspected is placed on the container mounting table 30, centering or the like is automatically performed, the shutter 12 is opened, and the irradiation optical fiber 21 moves to the bottom of the PET bottle 70. Light is irradiated.

(2) Light Irradiation and Scattered Light Reception As described above, the light emitted from the light source 10 is collected by the condenser lens 11 and is provided on the container mounting table 30 via the irradiation optical fiber 21. The near-infrared light is directed to the liquid in the pet bottle 70 and the PET bottle 70 by being guided to the shape irradiation unit 31. The irradiated near infrared light is scattered by the liquid and becomes scattered near infrared light indicated by a broken line. The scattered near-infrared light is received by the light receiving unit 32 provided on the container mounting table 30 and guided to the light receiving diffraction unit 40 via the light receiving optical fiber 22. The scattered near-infrared light guided to the light receiving diffraction unit 40 is sent to the optical grating 43 through the filter 41 and the slit 42, and is dispersed to obtain an absorption spectrum.

  Some of the near infrared light passes through the liquid without being scattered, and it is also possible to analyze the transmitted light. In this case, the light receiving unit is opposed to the irradiation unit. By providing, the same processing can be performed.

(3) Analysis of absorption spectrum The absorbance at a plurality of predetermined wavelengths of the acquired absorption spectrum is second-order differentiated to obtain a second-order differential absorption spectrum. By applying the second derivative, it is possible to separate the overlapping absorption bands and to remove the background effect, and to capture a minute change occurring in the collected absorption spectrum. Then, the absorption spectrum and the second-order differential absorption spectrum are subjected to multivariate analysis, and inspected according to the detection algorithm described below, and the content status such as the type and concentration of the dangerous substance to be inspected is specified. The inspection result is transmitted by the PC 50, other display means, or sound means.

  Each process based on this detection algorithm can be recorded as a program in the PC 50. In this case, the inspection is performed simply by placing the PET bottle 70 on the container mounting table 30 and irradiating near infrared light. The result can be displayed.

3. Detection Algorithm Next, the contents of the inspection will be described. In the detection algorithm below, principal component analysis is adopted as multivariate analysis.
(1) Dangerous Goods Detection Algorithm First, the dangerous goods detection algorithm will be described with reference to the flowchart shown in FIG. Note that in this embodiment, the object of inspection is liquid combustible material, water, hydrogen peroxide water, and drinking water other than water that does not contain hydrogen peroxide (hereinafter also simply referred to as “drinking water”). .

(2) Discrimination of liquid combustible material First, it is determined whether or not the inspection object is a liquid combustible material using the discriminant 1, and the inspection object determined to be a liquid combustible material is notified as a liquid dangerous material.

(3) Discrimination of drinking water Next, it is discriminated whether or not the inspection object contains hydrogen peroxide by the two-stage inspection using the discriminants 2 and 3 for the inspection object not notified in the discriminant 1. Thus, it is based on the property of drinking water that the two-stage inspection is performed. The inspection object determined not to contain hydrogen peroxide is notified as non-hazardous drinking water.

(4) Discrimination of groups provided according to the liquid amount Next, the inspection object that has not been discriminated using the discriminants 2 and 3 is selected using the discriminant 4, and the inspection target is any group according to the liquid amount It is determined whether it corresponds to.

(5) Measurement of hydrogen peroxide concentration Next, the concentration of hydrogen peroxide to be inspected is measured using the concentration estimation formula of the corresponding group. If the measured value is greater than or equal to the reference value, it is reported as a dangerous substance. .

  Next, a method for creating a concentration estimation formula will be described. The discriminant creation method will be described in detail in the embodiment.

4). Preparation of concentration estimation formula The above concentration estimation formula uses PLS regression analysis, multiple regression analysis, etc. for each group using a large number of measurement data for various (known) hydrogen peroxide solutions in various PET bottles. A concentration estimation formula is created for each group.

(1) Preparation of concentration estimation formula by PLS regression analysis and evaluation of measurement accuracy a. Creation of concentration estimation formula by PLS regression analysis The creation of a calibration curve for the concentration of hydrogen peroxide does not require wavelength selection, uses information on all wavelengths, is resistant to spectral fluctuations, and has high prediction performance. Analysis is preferably used. A regression model in PLS regression analysis is shown in FIG.

  FIG. 3 is a diagram for explaining a procedure for obtaining a hydrogen peroxide concentration estimation formula by PLS regression analysis. Specifically, for example, each sample is placed in four types of PET bottles with hydrogen peroxide concentrations of 0%, 8%, 26%, and 31% and liquid amounts of 100 ml, 200 ml, 300 ml, 400 ml, and 500 ml. The second derivative spectrum in the range of 690 to 980 nm shown in FIG. 3 (a) is obtained, and β (regression vector) shown in FIG. 3 (b) is obtained from the obtained result by the above regression model. Is obtained. FIG. 3C shows the correlation between the actual concentration and the concentration (estimated concentration) calculated using the concentration estimation formula by adding the evaluation data to the calibration data used for creating the regression model.

  Note that the measurement error due to the difference in the shape of the PET bottle is reduced when the number of types of PET bottles used in creating the concentration estimation formula is increased. However, the measurement error is not reduced even when eight or more types are used. Therefore, in order to derive the concentration estimation formula, it is better that the number of types of PET bottles is larger. However, in practice, it is sufficient to use about 8 types of PET bottles.

B. Evaluation of measurement accuracy Predictive standard error (SEP: Standard Error of Prediction) is used for evaluation. The SEP can be expressed by the following equation using an error (residual) vector f, its transposed vector f ^T , and the number n of evaluation samples, and the SEP is obtained using the measurement result of a sample having a known hydrogen peroxide concentration. Thus, measurement accuracy can be evaluated.

However, SEP: Standard Error of Prediction
(Predicted standard error)
f: error (residual) vector f ^T : transposed vector of f n: number of evaluation samples

(2) Preparation of concentration estimation formula by multiple regression analysis When multiple regression analysis is used, a multiple regression formula is established to express the relationship between the estimated concentration and the absorbance at each wavelength shown in the following general formula (1). , (1) create the required number of equations by substituting the concentration of liquids of various concentrations and absorbance at a plurality of selected wavelengths (λ _i ) of the absorption spectrum, and use the least squares method to calculate the regression constant of equation (1) β ₀ and partial regression coefficient β _i are determined.
y = β ₀ + β ₁ x ₁ + β ₂ x ₂ + β ₃ x ₃ +... + β _p x _p (1)
x: Absorbance at each selected wavelength λ _i (i = 1 to p)
y: concentration
β ₀ : regression constant (determined by liquid volume, container, solution, etc.)
β _i : Partial regression coefficient (determined by liquid volume, container, solution, etc.)

Next, the multiple regression equation (1) is evaluated by substituting the absorbance measurement result of a liquid having a known dangerous substance concentration prepared separately into the multiple regression equation (1) in which β ₀ and β _i are determined. Accordingly, correction is performed and finally a density estimation formula shown in the following formula (2) is created.
_{c = K 0 + K 1 E} 1 + K 2 E 2 + K 3 E 3 + ··· + K p E p (2)
c: Concentration (estimated value)
E _n : Absorbance at each predetermined wavelength
K ₀ , K _i (i = 1 to p): constant

  Next, discrimination of liquid dangerous materials and the measurement of hydrogen peroxide concentration will be specifically described with reference to examples. In this example, a sample in a plastic bottle having a liquid volume of 100 to 500 ml was used as an inspection target.

1. Discrimination of liquid dangerous goods by principal component analysis Data of unopened drinking water, water and hydrogen peroxide (0%, 8%, 26%, 31%, 45.5%) 100ml, 200ml, 300ml, Data of samples that were changed into 400 ml and 500 ml and put into 8 types of PET bottles with different capacities of 500 ml, and data of samples that changed the amount of liquid combustibles (alcohol, acetone, gasoline, salad oil) in three stages Used to identify liquid dangerous goods by principal component analysis.

(1) Discrimination of whether or not it is a liquid flammable material Drinking water, water, 45.5% hydrogen peroxide water, liquid flammable material (alcohol, acetone, gasoline, salad oil) as sampling samples, and from their scattered / reflected light The obtained absorption spectrum was analyzed by principal component analysis. As a result, it was found that discrimination was possible by using the first principal component and the second principal component obtained by principal component analysis.

  The analysis results are shown in FIG. FIG. 4 is a diagram showing secondary differential absorption spectra and principal component analysis results of various samples, (a) is a diagram showing secondary differential absorption spectra, and (b) is a graph showing secondary differential absorption spectra. It is a loading (discriminant) plot of the first principal component and the second principal component obtained by component analysis, and (c) is a score plot of the first principal component and the second principal component. In (c), the horizontal axis is the feature amount (score) of the first principal component, and the vertical axis is the score of the second principal component. Also, ◇ is drinking water or water, ■ is hydrogen peroxide solution, ▲ is alcohol, × is gasoline, □ is salad oil, and ◆ is acetone.

  As shown in (c), it was found that if the score of the first main component is 0 or more, it is a liquid combustible material (liquid dangerous material), and if it is 0 or less, it can be determined that it is not a liquid combustible material. That is, it was found that the analysis result shown in FIG. 4 is preferable as the discriminant 1 in FIG.

  The first main component is considered to reflect the characteristics of water. Thereby, the absorption band of water near 950 nm and the absorption band slightly on the short wavelength side are discriminated. For this reason, a spectrum having a water absorption band has a negative value, and a spectrum having an absorption band on a slightly short wavelength (near 930 nm) side like a liquid combustible has a positive value. Thus, whether or not the inspection object is a liquid combustible can be determined only by the score of the first principal component of the second-order differential absorption spectrum.

  Further, in (c), in the case of a liquid combustible material, the change in the liquid amount appears linearly for each type, and it can be seen that the type of the liquid combustible material can be determined by this inclination.

(2) Discrimination of drinking water If it is determined that it is not a liquid flammable material, it will be separated from “water, hydrogen peroxide solution” and “drinking water” such as tea or cola. As described above, it is divided into two stages using the second discriminant and the third discriminant.

  For example, carbonated beverages such as cola and cider, tea beverages such as green tea, oolong tea and black tea, sports beverages, water such as fruit juice and mineral water, and various beverages such as dairy beverages, 100 ml, 200 ml, 300 ml, 400 ml, 500 ml (Measured twice for samples with a capacity of less than 500 ml). Furthermore, water and 30% hydrogen peroxide solution were measured in the same manner in each container (excluding the same type of container) to obtain a total of 952 spectrum data.

  Next, a principal component analysis was performed on the data. FIG. 5 is a diagram showing score plots and loading plots of the second principal component of the absorbance spectrum and the fifth principal component of the second derivative spectrum for samples of water, hydrogen peroxide solution, and drinking water excluding liquid combustibles. Yes, the score plot is shown in FIG. 5 (a), and the loading plot is shown in FIG. 5 (b). In FIG. 5A, the horizontal axis is the second principal component of the absorbance spectrum, and the vertical axis is the fifth principal component of the secondary differential spectrum. ◇ is drinking water, and □ is water and hydrogen peroxide. In FIG. 5, it was found that if the value of the fifth principal component of the second derivative spectrum is 0.001 or more, or the value of the second principal component of the absorbance spectrum is 2.5 or more, it can be determined that it is drinking water. Thus, FIG. 5 is changed to discriminants 2 and 3 in FIG. From the above, whether or not the test object is drinking water can be determined by simply measuring the score of the second principal component of the absorbance spectrum and the score of the fifth principal component of the second derivative spectrum.

  In addition, it turns out that it isolate | separated using the wavelength of a visible region from FIG.5 (b). This makes it possible to separate black drinking water, tea, etc., together with hydrogen peroxide water, water, various drinking water (sports drink, white liquid such as calpis soda, 100% fruit juice drink, cider, mineral water, liquid volume A total of 824 types of data remained.

(3) Grouping by liquid volume As a result of various investigations on the measurement of hydrogen peroxide concentration, the spectral fluctuation due to the liquid volume is large, and this greatly affects the measurement of hydrogen peroxide concentration in hydrogen peroxide water. It was found that it is desirable to divide the samples into groups according to the amount of liquid (band intensity) and perform concentration measurement by applying a concentration estimation formula corresponding to each group. And it discovered that it could group by the liquid quantity by the principal component analysis of a secondary differential spectrum.

  Specifically, the wavelength range of the squeezed sample was set to 690 to 900 nm so that the change in the band intensity depending on the liquid amount was easily reflected as the main component. FIG. 6 is a diagram showing a principal component analysis result of a second derivative absorption spectrum of a sample from which black drinking water and a part of drinking water are excluded, and the first principal component and the second principal component of the second derivative absorption spectrum. The component score plot is shown in FIG. 6A, and the loading plot is shown in FIG. 6B. This was designated as discriminant 4 in FIG. In FIG. 6A, the horizontal axis is the first principal component, and the vertical axis is the second principal component. Also, ◇ is water, □ is hydrogen peroxide water, and △ is drinking water.

  In FIG. 6A, the discrimination reference value of the first principal component is set to 0.0025. Thus, the samples were grouped into two groups of class A of about 100 to 200 ml and class B of about 200 to 500 ml. As described above, the grouping of which group the inspection object corresponds to can be performed only by measuring the score of the first principal component of the second-order differential absorption spectrum in the wavelength range of 690 to 900 nm.

  All the measurements necessary for the above-described discrimination and grouping can be performed in a short time. And since the inspection object which requires the density | concentration measurement of hydrogen peroxide in such a short time can be narrowed down, many inspection objects can be inspected in a short time. In addition, as a result of narrowing the inspection target for concentration measurement to water or hydrogen peroxide solution, factors affecting the measurement accuracy such as the type of liquid and the shape of the PET bottle are reduced, so that the measurement accuracy is improved.

2. Preparation of concentration estimation formula of hydrogen peroxide solution by regression analysis A concentration estimation formula (calibration curve) was prepared by PLS regression analysis for each of class A and class B described above. Specifically, a concentration estimation formula (calibration curve) was created by the regression model described above using one of the measurement data in the wavelength range of 690 to 980 nm of data measured twice for each of class A and class B as calibration data. In class A, the number of factors was 8, and in class B, the number of factors was 10.

(1) Class A
The regression vector and calibration curve for class A are shown in FIGS. 7 (a) and 7 (b), respectively.
(2) Class B
Class B regression vectors and calibration curves are shown in FIGS. 8A and 8B, respectively.
In FIGS. 7A and 8A, the horizontal axis represents the wavelength, and the vertical axis represents the regression vector. The horizontal axis of FIGS. 7B and 8B is the actual concentration of hydrogen peroxide in the sample used to create the calibration curve, and the vertical axis is the estimated (measured) concentration.

3. Concentration measurement result (1) Measurement result Of the measurement data measured twice, the estimated concentration obtained from the calibration curve based on the other measurement data not used for the creation of the calibration curve, that is, based on the measured value and the actual concentration And plotted. The plotted results of class A and class B are shown in FIGS. 9 (a) and 9 (b), respectively. The horizontal axis in the figure is the actual density, and the vertical axis is the estimated density. In both (a) and (b), the estimated concentrations are distributed in a narrow range centering on the actual concentrations of 0% and 30%, and it can be seen that good measurement with high accuracy was performed.

(2) Evaluation of measurement accuracy The SEP described above was calculated based on the results shown in FIGS. 9A and 9B, and the measurement accuracy was quantitatively evaluated. In (a), the SEP is 2.81%, and in (b), the SEP is 2.78%, which confirms that the measurement accuracy is greatly improved compared to, for example, 4.25% when not grouped. It was.

  In actual inspection, when the measured concentration of hydrogen peroxide is 15% or more as shown in FIG. 2, for example, it is determined as highly concentrated hydrogen oxide water (dangerous material), that is, detected as a dangerous material. If it is less than 15%, it should not be detected as a dangerous substance.

  As described above, in this embodiment, a large number of inspection objects can be inspected in a short time, and the concentration of hydrogen peroxide can be accurately measured.

  As described above, according to the present embodiment, it is possible to narrow down samples for which concentration estimation is required by efficiently performing discrimination by performing systematic analysis even with a large number of samples. And estimation accuracy can also be improved.

  The above-described technology for inspecting a liquid in a beverage container according to the present invention is particularly effective for inspecting a beverage container to be brought in by a passenger at an airport.

DESCRIPTION OF SYMBOLS 10 Light source 11 Condensing lens 12 Shutter 21 Irradiation optical fiber 22 Light reception optical fiber 23 Calibration optical fiber 30 Container mounting base 31 Ring-shaped irradiation part 32 Light reception part 40 Light reception diffraction part 41 Filter 42 Slit 43 Optical grating 44 Spectrum Photoelectric conversion sensor 50 Personal computer 70 PET bottle

Claims (9)

A liquid inspection method for inspecting the content of explosives, explosive raw materials and / or illegal drugs in a liquid filled in a light transmissive container,
From the outside of the container, a near infrared light irradiation step of irradiating the liquid with near infrared light,
A near-infrared light receiving step for receiving the near-infrared light transmitted through the liquid or the near-infrared light scattered by the liquid;
An absorption spectrum analysis step of examining the content of explosives, explosive raw materials and / or illegal drugs in the liquid filled in the container by analyzing the absorption spectrum of the received near-infrared light. ,
After the absorption spectrum analysis step separates the liquid filled in the container according to a plurality of discriminants derived by multivariate analysis of the absorption spectrum, the explosive material, explosive material and / or by substituting the concentrations estimated equation generated based concentration illegal drug absorption spectra of a plurality of liquids of known concentration, the explosives, Ri step der measuring the concentration of explosive material and / or illicit drugs,
The multivariate analysis is principal component analysis or discriminant analysis;
The liquid inspection method, wherein the absorption spectrum analysis step includes a determination step of determining a grouping based on a liquid amount using a main component obtained by the main component analysis . The liquid inspection method according to claim 1, wherein the explosive material is hydrogen peroxide. 3. The liquid inspection method according to claim 2 , wherein the absorption spectrum analysis step includes a determination step of determining the presence or absence of a liquid combustible using the main component obtained by the main component analysis. . The said absorption spectrum analysis process has the discrimination | determination process which determines the presence or absence of hydrogen peroxide using the main component obtained by the main component analysis, The Claim 2 or Claim 3 characterized by the above-mentioned. Liquid inspection method. The absorption spectrum analysis step comprises
A first determination step of determining the presence or absence of a liquid combustible using the principal component obtained by the first principal component analysis;
A second discriminating step for determining the presence or absence of hydrogen peroxide using the principal component obtained by the second principal component analysis;
The liquid inspection method according to claim 2 , further comprising a third determination step of determining a grouping according to the liquid amount using the principal component obtained by the third principal component analysis. The absorption spectrum analysis step, the inspection method of a liquid according to any one of claims 1 to 5, wherein the the absorption spectrum analysis step using the second derivative spectrum. The concentration estimation equation, the inspection method of a liquid according to any one of claims 1 to 6, characterized in that it is prepared using a regression analysis. The liquid inspection method according to claim 7 , wherein the regression analysis is a PLS regression analysis. A liquid inspection device for inspecting the content of explosives, explosive raw materials and / or illegal drugs in a liquid filled in a light transmissive container,
Near-infrared light irradiation means for irradiating the liquid with near-infrared light from outside the container;
A near infrared light receiving means for receiving the near infrared light transmitted through the liquid or the near infrared light scattered by the liquid;
Absorption spectrum analysis means for examining the content of explosives, explosive raw materials and / or illegal drugs in the liquid filled in the container by analyzing the absorption spectrum of the received near-infrared light. ,
After the absorption spectrum analysis means separates the liquid filled in the container according to a plurality of discriminants derived by multivariate analysis of the absorption spectrum, the explosive material, explosive material and / or It is configured to measure the concentration of the explosive, the explosive material, and / or the illegal drug by substituting the concentration estimation formula created based on the absorption spectra of a plurality of liquids having known concentrations of the illegal drug. And
The multivariate analysis is principal component analysis or discriminant analysis;
The liquid inspection apparatus, wherein the absorption spectrum analysis means includes a determination means for determining a grouping based on a liquid amount using a main component obtained by principal component analysis .

Images