Analysis device, analysis method, and analysis program

Abstract

To provide an analyzer capable of estimating functional groups when analyzing the components of the material to be measured based on spectral data.SOLUTION: An analyzer 20 includes: an acquisition unit 200 that acquires measurement data MD, which is spectral data of a material to be measured, and multiple reference data RD, which are the spectral data of candidate components; an extraction unit 250 that extracts functional group information related to a functional group from the acquired measurement data MD; a generation unit 260 that generates a cluster map MP by clustering the reference data RD for each extracted functional group information; and an estimation unit 230 that estimates the functional groups that the material to be measured has from the position of the measured data MD plotted against the generated cluster map MP.SELECTED DRAWING: Figure 8

Description

[Technical Field]

[0001] The present invention relates to an analysis device, an analysis method, and an analysis program. [Background Art]

[0002] Patent Document 1 discloses a component identification device capable of identifying components of a sample from spectral information. In this device, the intensities of measured spectrum data and standard spectrum data are normalized, the measured spectrum data is regressed by a linear sum of the standard spectrum data, and a standard substance corresponding to the extracted standard spectrum data is selected as the composition of the sample to be measured. [Prior Art Literature] [Patent Literature]

[0003] [Patent Document 1] Japanese Unexamined Patent Publication No. 2020-201174 [Summary of the Invention] [Problem to be Solved by the Invention]

[0004] The device of Patent Document 1 merely estimates the composition of measured spectrum data with reference to standard spectrum data, and it is difficult to estimate functional groups included in a material to be measured.

[0005] An object of the present invention is to provide an analysis device, an analysis method, and an analysis program that enable estimation of functional groups when analyzing components of a material to be measured based on spectrum data. [Means for Solving the Problem]

[0006] The analysis apparatus according to claim 1 includes an acquisition unit that acquires measurement data which is spectral data of the material to be measured and a plurality of reference data which are spectral data of candidate components; an extraction unit that extracts functional group information relating to functional groups from the acquired measurement data; a generation unit that clusters the reference data for each extracted functional group information to generate a cluster map; and an estimation unit that estimates the functional groups that the material to be measured has from the positions of the measurement data plotted on the generated cluster map.

[0007] The analytical apparatus described in claim 1 is applied to an apparatus for analyzing the components of a material to be measured from its spectral data. In this analytical apparatus, the acquisition unit acquires measurement data, which is spectral data of the material to be measured, from a measuring device such as a Fourier transform infrared spectrophotometer (FTIR), a Raman spectrometer, a gas chromatograph-mass spectrometer (GC / MS), and a nuclear magnetic resonance spectrometer (NMR). The acquisition unit also acquires multiple reference data, which are spectral data of candidate components, from a server device or storage device. The extraction unit then extracts functional group information related to functional groups from the acquired measurement data, and the generation unit generates a cluster map by clustering the reference data for each functional group information. Furthermore, the estimation unit estimates the functional groups possessed by the material to be measured from the positions of the measurement data plotted on the cluster map. According to this analytical apparatus, functional groups can be estimated when analyzing the components of a material to be measured based on spectral data.

[0008] The analysis apparatus according to claim 2 is the analysis apparatus according to claim 1, wherein the generation unit performs clustering using UMAP.

[0009] According to the analysis apparatus described in claim 2, by using UMAP as the dimensionality reduction method, computation time can be reduced and visualization can be made easier to understand compared to using other dimensionality reduction methods.

[0010] The analysis method described in claim 3 involves a computer performing the following processes: acquiring measurement data, which is spectral data of the material to be measured, and a plurality of reference data, which are spectral data of candidate components; extracting functional group information related to functional groups from the acquired measurement data; clustering the reference data for each extracted functional group information to generate a cluster map; and estimating the functional groups possessed by the material to be measured from the positions of the measurement data plotted on the generated cluster map.

[0011] The analysis method described in claim 3 is applicable when analyzing the components of a material to be measured from spectral data of the material. In this analysis method, the device that provides the measurement data and reference data acquired by the computer is as described above. In this analysis method, the computer extracts functional group information related to functional groups from the acquired measurement data and generates a cluster map by clustering the reference data for each functional group information. Furthermore, the computer estimates the functional groups that the material to be measured has from the positions of the measurement data plotted on the cluster map. According to this analysis method, functional groups can be estimated when analyzing the components of a material to be measured based on spectral data.

[0012] The analysis program described in claim 4 causes a computer to perform the following processes: acquire measurement data which is spectral data of the material to be measured, and a plurality of reference data which are spectral data of candidate components; extract functional group information relating to functional groups from the acquired measurement data; cluster the reference data for each extracted functional group information to generate a cluster map; and estimate the functional groups possessed by the material to be measured from the positions of the measurement data plotted on the generated cluster map.

[0013] The analysis program described in claim 4 causes a computer to perform the process of analyzing the components of a material to be measured from its spectral data. The device that provides the measurement data and reference data acquired by the computer on which the analysis program is executed is as described above. The computer extracts functional group information related to functional groups from the acquired measurement data and generates a cluster map by clustering the reference data for each functional group information. Furthermore, the computer estimates the functional groups that the material to be measured possesses from the positions of the measurement data plotted on the cluster map. According to this analysis program, functional groups can be estimated when analyzing the components of a material to be measured based on spectral data. [Effects of the Invention]

[0014] According to the present invention, when analyzing the components of a material to be measured based on spectral data, functional groups can be estimated. [Brief explanation of the drawing]

[0015] [Figure 1] This is a block diagram showing the configuration of the analysis system according to the first embodiment. [Figure 2] This is a block diagram showing the functional configuration and data flow of the analysis device according to the first embodiment. [Figure 3] This figure illustrates the preprocessing of measurement data in the first embodiment. [Figure 4] This flowchart shows the flow of the analysis process in the first embodiment. [Figure 5] This is a block diagram showing the functional configuration and data flow of the analysis device in the second embodiment. [Figure 6] This figure illustrates the filtering of reference data in the second embodiment. [Figure 7] This is a flowchart showing the analysis process flow of the second embodiment. [Figure 8] This block diagram shows the functional configuration and data flow of the analysis device according to the third embodiment. [Figure 9]It is a diagram for explaining estimation of functional groups in a third embodiment. [Figure 10] It is a flowchart showing a flow of analysis processing of the third embodiment. MODE FOR CARRYING OUT THE INVENTION

[0016] The present invention is configured as an analysis system that automatically identifies candidate components based on spectrum data obtained when measuring a material using a measuring device such as a Fourier transform infrared spectrophotometer (FTIR).

[0017] First Embodiment (Configuration) FIG. 1 shows an analysis system 10 according to the first embodiment. The analysis system 10 of the present embodiment is configured to include a measuring device, a database, and an analysis device. The measuring device of the present embodiment is a Fourier transform infrared spectrophotometer (FTIR). The measuring device can obtain measurement data MD (see FIG. 2) which is spectrum data by measuring a material to be measured.

[0018] The database is configured as an external server device, but may also be configured as a large-capacity storage device. Reference data RD (see FIG. 2), which is spectrum data of candidate components of the material to be measured, is stored in the database. The analysis system also includes an input device for operating the analysis device and a display device for displaying information output from the analysis device.

[0019] The analysis device is configured to include a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a storage, a communication I / F (Interface) and an input / output I / F. The CPU, ROM, RAM, storage, communication I / F and input / output I / F are communicatively connected to each other via a bus.

[0020] The CPU 20A is the central processing unit, which executes various programs and controls various components. Specifically, the CPU 20A reads programs from ROM 20B or storage 20D and executes them using RAM 20C as the working area.

[0021] ROM20B stores various programs and data. RAM20C temporarily stores programs or data as a working area.

[0022] The storage 20D is composed of an HDD (Hard Disk Drive) or an SSD (Solid State Drive) and stores various programs and data. In this embodiment, the storage 20D stores the analysis program 100. The analysis program 100 is a program that executes the analysis process described later. The analysis program 100 may also be stored in the ROM 20B. Alternatively, the reference data RD may be stored in storage 20D instead of database 14.

[0023] The communication interface 20E is an interface for connecting to external devices. The communication interface 20E is capable of communication via Ethernet (registered trademark), for example, and is connected to the database 14 via an internet connection.

[0024] The input / output interface 20F is an interface for communication with the measuring device 12, the input device 22, and the display device 24. The measuring device 12, the input device 22, and the display device 24 may also be directly connected to the bus 20G.

[0025] The input device 22 may be a mouse, keyboard, or other device. The display device 24 may be an example of a liquid crystal monitor. Alternatively, a touch panel that also functions as the input device 22 may be used as the display device 24.

[0026] As shown in Figure 2, in the analysis device 20 of this embodiment, the CPU 20A functions as an acquisition unit 200, a detection unit 210, a configuration unit 220, and an estimation unit 230 by executing the analysis program 100.

[0027] The acquisition unit 200 has the function of acquiring measurement data MD and multiple reference data RD. Specifically, the acquisition unit 200 acquires measurement data MD when it is transferred from the measuring device 12. The acquisition unit 200 also acquires reference data RD by reading it from the database 14.

[0028] The detection unit 210 has the function of detecting the index of the peak value for the measurement data MD and the reference data RD. When detecting the index of the measurement data MD, the detection unit 210 first uses a known module called convolve() to calculate the moving average of the measurement data MD and remove noise. Then, using the noise-removed measurement data MD, it obtains the index of the peak value using a known module called signal.argrelmax(). The index is obtained as a value when the horizontal axis of the spectrum is divided into intervals from 0 to 3400. The detection unit 210 also detects the index of the reference data RD in the same way as the measurement data MD.

[0029] The component 220 has the function of reconstructing the measurement data MD based on the index of the measurement data MD and the reference data RD. When reconstructing the measurement data MD, the component 220 reconstructs the measurement data MD from the index and peak value of the measurement data MD using a Gaussian function. Furthermore, the component 220 normalizes the intensity, which is the vertical axis, within the range of 0 to 100. As a result, as shown in Figure 3, measurement data MD with the effects of noise removed can be obtained from the measurement data MD acquired from the measuring device 12. The component 220 also reconstructs the reference data RD in the same way as the measurement data MD.

[0030] As shown in Figure 2, the estimation unit 230 has the function of estimating the components of the material being measured. Specifically, the estimation unit 230 estimates the components of the material being measured by performing regression on the reconstructed measurement data MD for each reference data RD using Lasso regression, which is a non-negative linear least squares method.

[0031] Here, in the regression performed by the estimation unit 230, the regression is carried out using the non-negative linear least squares method such that L in Equation 1 below is minimized. In Equation 1, the L1 norm is added as a regularization term, and the regression is carried out by controlling the strength of regularization.

[0032]

number

[0033] The estimation unit 230 then calculates the degree of agreement between the measurement data MD and each reference data RD based on the results of the Lasso regression, and estimates the components whose degree of agreement exceeds a predetermined value as components of the material being measured.

[0034] (Control flow) The flow of the analysis process as an analysis method executed in the analysis device 20 of this embodiment will be explained using the flowchart in Figure 4. The processing in the analysis device 20 is realized by the CPU 20A functioning as the acquisition unit 200, detection unit 210, configuration unit 220, and estimation unit 230 described above.

[0035] In step S100 of Figure 4, the CPU 20A acquires measurement data MD from the measuring device 12. In step S102, CPU 20A retrieves reference data RD from database 14.

[0036] In step S104, the CPU 20A performs peak detection processing on the measurement data MD and the reference data RD. The content of the peak detection processing is the same as the processing performed in the detection unit 210 described above.

[0037] In step S106, the CPU 20A performs spectral reconstruction processing on the measurement data MD and the reference data RD. The details of the spectral reconstruction processing are the same as the processing performed in the component 220 described above.

[0038] In step S108, the CPU 20A performs a screening process. The content of the screening process is the same as the process performed by the estimation unit 230 described above.

[0039] In step S110, the CPU 20A outputs the estimated result ER. That is, the CPU 20A displays information about one or more components estimated for the material under test on the display device 24. Then the analysis process ends.

[0040] (Summary of the embodiments) In the analysis device 20 of this embodiment, the acquisition unit 200 acquires measurement data MD, which is spectral data of the material to be measured, from the measuring device 12, and acquires a plurality of reference data RD, which are spectral data of candidate components, from the database 14. Then, the detection unit 210 detects the index of the peak value of the spectrum in the measurement data MD, and the configuration unit 220 reconstructs the measurement data MD based on the detected index. Furthermore, the estimation unit 230 regresses the reconstructed measurement data MD on each reference data RD using Lasso regression, which is a non-negative linear least squares method, and estimates the components of the material to be measured based on the degree of agreement between the measurement data MD and the reference data RD. According to the analysis device 20 of this embodiment, when analyzing the components of a material to be measured based on spectral data, a preprocessing is performed in which the index of the peak value is detected in the measurement data MD and the measurement data MD is reconstructed based on the index. This makes it possible to reduce noise in the measurement data MD before performing Lasso regression and improve the accuracy of the analysis.

[0041] Furthermore, in this embodiment, the index of the spectral peak value is detected not only for the measurement data MD but also for the reference data RD, and the reference data RD is reconstructed based on the index of the reference data RD before estimating the components of the material being measured. Therefore, according to this embodiment, the accuracy of the analysis can be further improved by performing pre-processing to reduce noise by the detection unit 210 and the component unit 220 not only on the measurement data MD but also on the reference data RD.

[0042] [Second Embodiment] The second embodiment differs from the first embodiment in the method of preprocessing the reference data RD before screening. The differences from the first embodiment will be described below. Note that the other configurations are the same as in the first embodiment, and a detailed explanation will be omitted.

[0043] As shown in Figure 5, the analysis device 20 of this embodiment includes, in addition to the acquisition unit 200, detection unit 210, configuration unit 220, and estimation unit 230, a narrowing unit 240 as a function of the CPU 20A executed by the analysis program 100.

[0044] The filtering unit 240 has the function of narrowing down candidate components of the material to be measured by clustering each reference data RD. Specifically, the filtering unit 240 reduces the dimensionality of the intensity at each index of the measurement data MD and reference data RD using UMAP (Uniform Manifold Approximation and Projection).

[0045] Then, as shown in Figure 6, the filtering unit 240 plots the two-dimensional reference data RD together with the two-dimensional measurement data MD on the cluster map MP. The filtering unit 240 then narrows down the reference data RD that falls within a predetermined range AR relative to the measurement data MD as candidate components of the material to be measured.

[0046] (Control flow) The flow of the analysis process performed in the analysis device 20 of this embodiment will be explained using the flowchart in Figure 7. The processing in the analysis device 20 is realized by the CPU 20A functioning as the acquisition unit 200, detection unit 210, configuration unit 220, estimation unit 230, and narrowing unit 240 described above. The same reference numerals are used for the same steps as in the first embodiment, and detailed explanations are omitted.

[0047] The peak detection process in step S104 and the spectral reconstruction process in step S106 in Figure 7 are performed only on the measurement data MD (see Figure 5).

[0048] In step S107, the CPU 20A performs a candidate narrowing process using clustering on the reference data RD. The content of the candidate narrowing process using clustering is the same as the process performed in the narrowing unit 240 described above.

[0049] The screening process in step S108 is performed using the measurement data MD from which the processes in steps S104 and S106 have been performed, and the reference data RD from which the process in step S107 has been performed.

[0050] In step S110, the CPU 20A outputs the estimated result ER. Then the analysis process ends.

[0051] (Summary of the embodiments) In the analysis device 20 of this embodiment, the narrowing unit 240 narrows down the candidate components of the material to be measured by clustering each reference data RD. According to this embodiment, clustering makes it possible to narrow down the reference data RD of components that are close to the measurement data MD, thereby shortening the reference data RD reading time and analysis time.

[0052] [Third Embodiment] In the analysis process of the first embodiment, one or more components were estimated for the material being measured. However, the third embodiment differs from the first embodiment in that, in addition to estimating components, it also estimates information about the functional groups present in the material being measured. The differences from the first embodiment will be explained below. Note that the other configurations are the same as in the first embodiment, and a detailed explanation will be omitted.

[0053] As shown in Figure 8, the analysis device 20 of this embodiment includes, in addition to the acquisition unit 200, detection unit 210, component unit 220, and estimation unit 230, an extraction unit 250 and a generation unit 260 as functions of the CPU 20A executed by the analysis program 100.

[0054] The extraction unit 250 has the function of extracting functional group information from the reference data RD. For example, the extraction unit 250 extracts functional groups such as hydroxy(OH) groups, ketones, thiol groups, and nitro groups from the reference data RD.

[0055] As shown in Figure 9, the generation unit 260 has the function of clustering the reference data RD for each extracted functional group information and generating a cluster map MP. Specifically, the generation unit 260 reduces the dimensionality of the intensities at each index of the measurement data MD and the reference data RD using UMAP and then generates the cluster map MP.

[0056] In this embodiment, the estimation unit 230 estimates one or more components of the material under test, and also estimates the functional groups of the material under test from the positions of the measurement data MD plotted on the generated cluster map MP. The estimation unit 230 plots the measurement data MD, which has been dimensionally reduced by UMAP, on the cluster map generated by the generation unit 260. The estimation unit 230 then estimates the functional groups in the vicinity of the plotted measurement data MD as the functional groups of the material under test.

[0057] (Control flow) The flow of the analysis process performed in the analysis device 20 of this embodiment will be explained using the flowchart in Figure 10. The processing in the analysis device 20 is realized by the CPU 20A functioning as the acquisition unit 200, detection unit 210, configuration unit 220, estimation unit 230, extraction unit 250, and generation unit 260 described above. The same reference numerals are used for the same steps as in the first embodiment, and detailed explanations are omitted.

[0058] In step S120 of Figure 10, the CPU 20A performs a functional group information extraction process. The content of the functional group information extraction process is the same as the process performed in the extraction unit 250 described above.

[0059] In step S122, the CPU 20A executes the map generation process. The contents of the map generation process are the same as the process executed in the generation unit 260 described above.

[0060] In step S124, the CPU 20A performs functional group estimation processing. Specifically, in the generated cluster map MP, the functional groups for which measurement data MD exists are estimated to be the functional groups possessed by the material being measured.

[0061] In step S126, the CPU 20A outputs the estimated result ER, which includes the functional groups. That is, the CPU 20A displays the information of one or more components estimated for the material under test, along with the information of the estimated functional groups, on the display device 24. The analysis process then ends.

[0062] (Summary of the embodiments) In the analysis device 20 of this embodiment, the extraction unit 250 clusters the reference data RD for each functional group information extracted, and the generation unit 260 generates a cluster map MP. Then, the estimation unit 230 estimates the functional groups present in the material being measured from the positions of the measurement data MD plotted on the cluster map MP. According to this embodiment, not only the components of the material being measured but also the functional groups can be estimated. In particular, by clustering the measurement data MD and the reference data RD, functional groups can be estimated from spectral data, from which it is difficult to extract the characteristics of functional groups.

[0063] [remarks] In this embodiment, the analysis device 20 is connected to the measurement device 12 as a Fourier transform infrared spectrophotometer (FTIR) and applied to the analysis of components in the infrared absorption spectrum of the material being measured, but this is not limited to this. For example, it can be applied to the analysis of Raman scattering spectra when a Raman spectrometer is used as the measurement device 12, the analysis of mass spectra when a gas chromatograph-mass spectrometer (GC / MS) is used as the measurement device 12, and the analysis of NMR spectra when a nuclear magnetic resonance spectrometer (NMR) is used as the measurement device 12.

[0064] In the second embodiment, UMAP was applied as a dimensionality reduction method when clustering the reference data RD in the filtering unit 240. Similarly, in the third embodiment, UMAP was applied as a dimensionality reduction method when generating the cluster map MP in the generation unit 260. However, other dimensionality reduction methods such as t-SNE (t-distribution Stochastic Neighbor Embedding) and Isomap can be applied. In the second and third embodiments, using UMAP as a dimensionality reduction method reduces computation time and improves the clarity of the visualization compared to using other dimensionality reduction methods.

[0065] In addition, the various processes that the CPU 20A reads and executes in each of the above embodiments may be executed by various processors other than the CPU. Examples of such processors include PLDs (Programmable Logic Devices) such as FPGAs (Field-Programmable Gate Arrays) whose circuit configuration can be changed after manufacturing, and dedicated electrical circuits that are processors with circuit configurations specifically designed to execute specific processes, such as ASICs (Application Specific Integrated Circuits). Furthermore, each of the above processes may be executed by one of these various processors, or by a combination of two or more processors of the same or different types (for example, multiple FPGAs, and a combination of a CPU and an FPGA). More specifically, the hardware structure of these various processors is an electrical circuit that combines circuit elements such as semiconductor elements.

[0066] Furthermore, in each of the above embodiments, the program was described as being pre-stored (installed) on a computer-readable non-temporary recording medium. For example, the analysis program 100 in the analysis device 20 is pre-stored on storage 20D. However, the program is not limited to this, and may be provided in a form recorded on a non-temporary recording medium such as a CD-ROM (Compact Disc Read Only Memory), DVD-ROM (Digital Versatile Disc Read Only Memory), or USB (Universal Serial Bus) memory. Alternatively, the program may be downloaded from an external device via a network.

[0067] The processing flow described in the above embodiment is just one example, and unnecessary steps may be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose. [Explanation of symbols]

[0068] 20 Analyzer 100 Analysis Programs 200 Acquisition Department 230 Estimation Department 250 Extraction part 260 Generation part MD measurement data RD Reference Data

Claims

1. An acquisition unit that acquires measurement data, which is spectral data of the material to be measured, and multiple reference data, which are spectral data of candidate components. A detection unit detects the index and peak value of the spectrum for the acquired measurement data and multiple reference data, respectively. A component that reconstructs the measurement data and multiple reference data using a Gaussian function based on the index and peak value detected by the detection unit, An extraction unit that extracts functional group information relating to functional groups from the aforementioned reference data, A generation unit clusters the reference data for each extracted functional group information to generate a cluster map containing clusters for each functional group, An estimation unit that projects the reconstructed measurement data onto the generated cluster map with dimensionality reduction, and estimates the functional groups of the material being measured based on the functional groups near the projection position, An analytical device equipped with the following features.

2. The generation unit is The analysis apparatus according to claim 1, which performs clustering using UMAP.

3. We obtain measurement data, which is the spectral data of the material being measured, and multiple reference data, which are the spectral data of candidate components. For the acquired measurement data and multiple reference data, the index and peak value of the spectrum are detected, respectively. Based on the detected index and peak value, the measurement data and multiple reference data are reconstructed using a Gaussian function. From the aforementioned reference data, functional group information relating to the functional group is extracted, The reference data is clustered for each extracted functional group information to generate a cluster map. The reconstructed measurement data is projected onto the generated cluster map with dimensionality reduction, and the functional groups of the material being measured are estimated based on the functional groups near the projection position. An analysis method in which a computer performs the processing.

4. We obtain measurement data, which is the spectral data of the material being measured, and multiple reference data, which are the spectral data of candidate components. For the acquired measurement data and multiple reference data, the index and peak value of the spectrum are detected, respectively. Based on the detected index and peak value, the measurement data and multiple reference data are reconstructed using a Gaussian function. From the aforementioned reference data, functional group information relating to the functional group is extracted, The reference data is clustered for each extracted functional group information to generate a cluster map. The reconstructed measurement data is projected onto the generated cluster map with dimensionality reduction, and the functional groups of the material being measured are estimated based on the functional groups near the projection position. An analysis program that causes a computer to perform a process.

Images