JP7228599B2 - X-ray analysis device, X-ray analysis system, analysis method, and program - Google Patents

Description

The present invention provides an apparatus for analyzing an object to be measured based on a fluorescent X-ray spectrum generated from the object to be measured, an X-ray analysis system configured by the apparatus, an analysis method for analyzing an object to be measured, and the analysis. It relates to a program that implements the method.

2. Description of the Related Art Conventionally, there is known a method of analyzing elements contained in an object to be measured based on a fluorescent X-ray spectrum generated from the object to be measured. For example, the peaks contained in the fluorescent X-ray spectrum are separated using a "peak separation" technique, and the elements contained in the measurement target are specified based on the separated peaks (see, for example, Patent Document 1).

JP 2014-41065 A

When analyzing the element to be measured based on the fluorescent X-ray spectrum, various parameters related to the fluorescent X-ray spectrum are used. In order to obtain accurate elemental analysis results from the obtained fluorescent X-ray spectrum, it is necessary to appropriately set this parameter.
Traditionally, this parameter has been set empirically. When performing elemental analysis based on fluorescent X-ray spectra, empirically derived parameters may not necessarily be optimal.

In addition to the spectrum derived from the fluorescent X-rays generated from the element to be measured, the fluorescent X-ray spectrum measured by the measuring device includes effects derived from the measuring device (for example, sum peak, escape peak), fluorescence Spectra derived from measurement objects other than X-rays (for example, diffraction spectra from measurement objects) are included.
However, when the above parameters are determined based only on experience, it is extremely difficult to consider these influences and set the optimum parameters.

SUMMARY OF THE INVENTION An object of the present invention is to perform analysis in consideration of various influences contained in a fluorescent X-ray spectrum in an X-ray analysis system for analyzing elements contained in a measurement object based on the fluorescent X-ray spectrum.

A plurality of aspects will be described below as means for solving the problem. These aspects can be arbitrarily combined as needed.
An X-ray analysis apparatus according to one aspect of the present invention includes a measurement device and an analysis section.
The measuring device measures a fluorescent X-ray spectrum generated from a sample to be measured.
The analysis unit applies a learning model trained using teacher data including a first parameter related to the first fluorescent X-ray spectrum and first result information related to the analysis result when using the first parameter to the learning model with the measuring device The sample for which the second fluorescent X-ray spectrum is generated is analyzed based on the second result information obtained by inputting the second parameters related to the measured second fluorescent X-ray spectrum.
A learning model is generated using teacher data including a first parameter regarding a first fluorescent X-ray spectrum and first result information regarding an analysis result when using the first parameter, and analysis is performed using the learning model. By doing so, it is possible to perform qualitative analysis in consideration of various effects included in the fluorescent X-ray spectrum.

The X-ray analysis system described above may further include an acquisition unit. The acquisition unit acquires a third parameter related to the third fluorescent X-ray spectrum measured by the measuring device as the first parameter, and converts the third result information related to the analysis result obtained using the third parameter into the first result information. to get as
As a result, the data acquired by the measuring device included in the X-ray analysis device itself can be used as teacher data.

When the teacher data includes the third result information that has been properly analyzed among the third result information as the first result information, and this third result information (the third result information that has properly qualified the sample) is obtained may include the third parameter used in as the first parameter.
As a result, the learning model can be learned so as to obtain an appropriate target quantitative result using the X-ray analysis device, and the sample can be appropriately qualitatively obtained by the learning model.

The analysis unit may perform the analysis based on whether or not the correlation coefficient representing the degree of matching between the fluorescent X-ray spectrum and the theoretical spectrum of the element presumed to be contained in the sample is equal to or greater than a predetermined threshold. . In this case, the result information about the analysis result is the threshold used for comparison with the correlation coefficient.
As a result, parameters for elemental analysis can be optimized through learning, and accurate qualitative analysis can be performed.

If the correlation coefficient between the fluorescent X-ray spectrum and the theoretical spectrum is equal to or greater than a predetermined threshold, the analysis unit may identify the element that produces the theoretical spectrum as the element contained in the sample.
This makes it possible to specify the elements contained in the sample by simple calculation.

The first parameter and the second parameter may be fluorescent X-ray spectra. This allows the sample to be properly analyzed.

The teacher data may include information on the fluorescent X-ray spectrum obtained using the calibration sample, and elements contained in the calibration sample and composition ratios of the elements. This allows proper learning to be performed using data on calibration samples that are known to be accurate data.

An X-ray analysis system according to another aspect of the present invention comprises the X-ray analysis device described above and a server connected to this X-ray analysis device. This X-ray analysis system can perform more flexible analysis.

The server of the above X-ray analysis system may have functions corresponding to the analysis unit. This allows the server to have its own learning model and perform analysis using the learning model.

According to still another aspect of the present invention, there is provided an X-ray analysis system in which a plurality of X-ray analysis devices are interconnected. This allows more flexible analysis such as data exchange between X-ray analyzers.

An analysis method according to still another aspect of the present invention comprises the following steps.
⊚ A step of measuring a fluorescent X-ray spectrum generated from a sample to be measured.
A step of learning a learning model using teacher data including first parameters relating to the first fluorescent X-ray spectrum and first result information relating to the analysis results of the sample when using the first parameters.
◎ A sample that generates a second fluorescent X-ray spectrum based on the second result information obtained by inputting the second parameter related to the second fluorescent X-ray spectrum obtained by measurement into the learning model that has been trained. A step that analyzes the

Generating a learning model using teacher data including a first parameter related to a fluorescent X-ray spectrum and first result information related to an analysis result when using the first parameter, and executing analysis using the learning model can perform an analysis that considers various effects contained in the X-ray fluorescence spectrum.

When analyzing the measurement target based on the fluorescent X-ray spectrum, analysis can be performed in consideration of various effects included in the fluorescent X-ray spectrum.

The figure which shows the structure of the X-ray-analysis apparatus which concerns on 1st Embodiment. 2 is a diagram showing the functional block structure of the information processing apparatus according to the first embodiment; FIG. 4 is a flow chart showing sample analysis operation using the X-ray analysis system. The figure which shows the X-ray-analysis system which concerns on 4th Embodiment. The figure which shows the X-ray-analysis system which concerns on 5th Embodiment.

1. First Embodiment (1) Overall Configuration of X-ray Analysis Apparatus Hereinafter, the configuration of the X-ray analysis apparatus according to the first embodiment will be described with reference to FIG. FIG. 1 is a diagram showing the configuration of an X-ray analysis apparatus according to the first embodiment. The X-ray analysis apparatus 100 according to the first embodiment is a stand-alone type apparatus that performs sample qualitative analysis by itself.
The X-ray analysis apparatus 100 is an apparatus for performing qualitative analysis of a sample S using a fluorescent X-ray spectrum generated from the sample S to be measured. The X-ray analysis apparatus 100 according to the first embodiment can qualitatively identify the elements contained in the sample S.
Specifically, the X-ray analysis apparatus 100 includes a measurement device 1 and an information processing device 3 .

(2) Measuring Device The measuring device 1 is a device for measuring the fluorescent X-ray spectrum generated from the sample S. As shown in FIG. The measurement apparatus 1 has a sample table 11 , an X-ray source 13 , an X-ray detector 15 , a spectrum generator 17 and a controller 19 .
The sample table 11 is a member for placing the sample S thereon. The X-ray source 13 is a device that irradiates the sample S placed on the sample table 11 with X-rays. The X-ray source 13 is, for example, an X-ray tube whose target is rhodium (Rh). The target of the X-ray source 13 is not limited to rhodium, and other targets such as copper (Cu) can also be used.

The X-ray detection unit 15 is a device that detects fluorescent X-rays generated by irradiating the sample S with X-rays from the X-ray source 13 . The X-ray detection unit 15 is, for example, a semiconductor detector using high-purity Si. The X-ray detector 15 outputs a signal proportional to the energy of the detected fluorescent X-rays.

The spectrum generation unit 17 receives the signal output by the X-ray detection unit 15, counts the signals of each value, and determines the relationship between the energy of the fluorescent X-rays detected by the X-ray detection unit 15 and the count number, that is, the fluorescent X-rays. Perform processing to generate a spectrum. The control unit 19 controls each component of the measurement apparatus 1 such as the X-ray source 13 and the spectrum generation unit 17 .

The spectrum generation unit 17 and the control unit 19 described above are, for example, a computer system configured by a CPU, a storage device (RAM, ROM, etc.), and various input/output interfaces. All or part of the functions of the spectrum generation section 17 and the control section 19 may be realized by a program stored in the storage device of the computer system.
Further, all or part of the functions of the spectrum generation unit 17 and the control unit 19 may be realized by hardware such as SoC (System on Chip).
Furthermore, the spectrum generation unit 17 and the control unit 19 may be configured as separate computer systems, or their functions may be integrated into one computer system.

(3) Information processing device (3-1) overview The information processing device 3 is, for example, a computer system configured with a CPU, a storage device (RAM, ROM, SSD, hard disk, etc.), and various input/output interfaces. is. The information processing device 3 analyzes the sample S mainly using the fluorescent X-ray spectrum obtained by the measuring device 1 . All or part of the functions described below of the information processing device 3 may be realized by a program stored in a storage device, or may be realized by hardware.
The information processing device 3 also transmits various parameters used in the measuring device 1 and/or control signals for the measuring device 1 to the spectrum generator 17 and/or the controller 19 . Furthermore, it receives various signals such as a signal reporting an abnormality that has occurred in the measuring device 1 from the control unit 19 and/or the spectrum generation unit 17 .
The functional blocks of the information processing device 3 according to the first embodiment will be described below with reference to FIG. FIG. 2 is a diagram showing the functional block structure of the information processing apparatus according to the first embodiment. The information processing device 3 has an analysis unit 5 , a storage unit 7 and an acquisition unit 9 .

(3-2) Analyzing Section Based on the fluorescent X-ray spectrum generated by the spectrum generating section 17, the analyzing section 5 analyzes the sample S for which the fluorescent X-ray spectrum has been generated. The analysis unit 5 of the present embodiment uses an automatic qualitative algorithm as an algorithm for qualitatively analyzing the sample S based on the fluorescent X-ray spectrum. The automatic qualitative algorithm mainly includes (1) peak separation, (2) calculation of the correlation coefficient, and (3) narrowing down the final candidate elements by comparing the correlation coefficient with the threshold value. This is an algorithm that narrows down and qualitatively includes elements.

(3-3) Functional Blocks of Analysis Section The functional blocks of the analysis section 5 of this embodiment will be described below. The analysis unit 5 has a peak separation unit 51 , a correlation coefficient calculation unit 52 , an element identification unit 53 , a learning unit 54 and a quantification unit 55 .
The peak separation unit 51 narrows down the elements contained in the sample S by separating the peaks of the fluorescent X-ray spectrum measured by the measurement device 1 . Specifically, the peak separation unit 51 performs peak separation using theoretical spectra generated based on the elemental data ED stored in the storage unit 7 .
The correlation coefficient calculator 52 calculates the correlation coefficient between the fluorescent X-ray spectrum measured by the measuring device 1 and the theoretical spectrum of the element filtered through the peak separator 51 . This correlation coefficient is a coefficient representing the degree of matching between the measured fluorescent X-ray spectrum and the theoretical spectrum, and the greater the value, the higher the degree of matching. The correlation coefficient can be calculated using a known formula for calculating correlation coefficients.

The element specifying unit 53 specifies, as the elements contained in the sample S, the element that produces the theoretical spectrum for which the correlation coefficient calculated by the correlation coefficient calculating unit 52 is equal to or greater than a predetermined threshold. That is, the above threshold value is a value that serves as an index for determining whether or not a certain element exists in the sample S.

In the present embodiment, the element identifying unit 53 has a threshold manually set using an input device (for example, a keyboard, etc.) of the information processing device 3, or the fluorescent X-ray spectrum measured by the measuring device 1 in the learning unit 54 Select and use one of the thresholds output by inputting the parameters for . The manually set threshold is, for example, a threshold determined by experimental evaluation.

The learning unit 54 inputs parameters related to the fluorescent X-ray spectrum as input parameters (examples of the first parameter and the second parameter), and thresholds (examples of the first result information and the second result information) used by the element identification unit 53 ), for example, a learning model composed of a neural network (an example of a learning model) including at least an input layer and an output layer.

An input parameter that can be input to the learning unit 54 is, for example, the acquired fluorescent X-ray spectrum itself. In this case, the fluorescence X-ray spectrum of the entire measured energy range may be input, or the fluorescence X-ray spectrum of a specific narrowed energy range may be input.
In addition, as input parameters, the peak position, peak ratio, peak area of the fluorescent X-ray generated from each element, the optical image of the sample S, the measurement conditions of the fluorescent X-ray spectrum, the information of the surrounding environment of the measuring device 1 (temperature, humidity, atmospheric pressure, etc.), and information indicating whether or not the element has been properly identified using a specific fluorescent X-ray spectrum and threshold value (referred to as identification success/failure information).
The identification success/failure information is, for example, a value set to "1" when the element is properly identified, and "0" when it is not properly identified.

The quantification unit 55 calculates (quantifies) the concentrations of the elements contained in the sample S for the elements that produce theoretical spectra with correlation coefficients equal to or greater than a predetermined threshold value, and outputs the results as quantification results.

(3-4) Storage Unit The storage unit 7 is a part of the storage area of the storage device of the information processing device 3, and stores at least data necessary for the analysis unit 5 to execute the automatic qualitative algorithm. Specifically, the storage unit 7 stores element data ED and teacher data TD.
The element data ED are parameters related to the element used for calculating the theoretical spectrum of each element. This parameter may be calculated based on the fluorescent X-ray spectrum actually measured using the measuring device 1 with a substance containing only a specific element as the sample S, or may be registered in a known database or the like. parameters.

(3-5) Teacher Data The teacher data TD is data for generating a learning model for the learning unit 54. FIG. The teacher data TD is composed of input parameters input to the learning section 54 and output parameters output from the learning section 54 .

Input parameters included in the teacher data TD are parameters (referred to as first parameters) relating to the fluorescent X-ray spectrum (referred to as the first fluorescent X-ray spectrum). The first fluorescent X-ray spectrum on which this first parameter is based is not only the fluorescent X-ray spectrum measured by the measuring device 1, but also the fluorescent X-ray spectrum measured by another measuring device (X-ray analyzer). It may contain a spectrum.
In this embodiment, the input parameters included in the teacher data TD are the same as the input parameters input from the learning unit 54 to output the threshold.
Specifically, the input parameters of the teaching data TD include, for example, the first fluorescent X-ray spectrum, the peak position of the fluorescent X-rays generated from each element, the peak ratio, the peak area, the optical image of the sample S, the first fluorescence X There are measurement conditions of the line spectrum, information on the surrounding environment (temperature, humidity, atmospheric pressure, etc.) when the first fluorescent X-ray spectrum was measured, information on past specific success or failure, and the like.

On the other hand, the output parameter included in the training data TD is information (referred to as first result information) regarding the analysis result of the element when using the first parameter.
In this embodiment, the output parameter included in the training data TD is the threshold used when identifying the element using the corresponding first fluorescent X-ray spectrum.

(3-6) Acquisition Unit The acquisition unit 9 is, for example, an input/output interface of the information processing device 3 and a storage medium reading device. Acquisition unit 9 is connected to measuring device 1 .
The acquisition unit 9 acquires parameters (example of the third parameter) related to the X-ray fluorescence spectrum (example of the third X-ray fluorescence spectrum) measured by the measuring device 1 and the threshold manually set in the element identification unit 53 (the 3 (an example of result information) and are acquired, and if necessary, these data are associated and stored in the storage unit 7 as teacher data TD.
Thereby, the data acquired by the measuring device 1 provided in the X-ray analysis device 100 itself can be used as the teacher data TD.

Note that the acquisition unit 9 determines the contents of various conditions set in the X-ray analysis device 100 when the sample S is analyzed, and when a certain specific condition is set, the fluorescence X measured by the measurement device 1 A parameter related to the line spectrum and a threshold manually set in the element specifying unit 53 may be acquired, and stored in the storage unit 7 as teacher data TD in association with each other.
As the specific condition, for example, it is determined that the X-ray analysis apparatus 100 analyzes a specific material/element. In this case, the set specific condition itself may be included as an input parameter of the teacher data TD. For example, when it is decided to analyze a specific material/element, various conditions set for measuring the fluorescent X-ray spectrum of the specific material/element (for example, setting conditions for the X-ray source 13, setting conditions for the X-ray detection unit 15) may be included as input parameters of the teacher data TD.
As a result, it is possible to generate the teacher data TD according to the usage status of the X-ray analysis apparatus 100 and generate the learning model optimized for the usage status in the learning unit 54 .

The acquisition unit 9 is also connected to other measurement devices (X-ray analysis devices), computer systems, and the like via networks and the like. In this case, the acquisition unit 9, for example, information about the fluorescent X-ray spectrum measured by the other measuring device 1, information about the analysis result when the elemental analysis is performed using the fluorescent X-ray spectrum, is acquired as teacher data TD.
For example, the fluorescent X-ray spectrum measured by another measuring device 1, the result of qualitative determination by an automatic qualitative algorithm using the fluorescent X-ray spectrum, and / or manually using the fluorescent X-ray spectrum As a result of the qualitative analysis, and can be acquired as teacher data TD.

In addition, the acquisition unit 9 can acquire, for example, a fluorescent X-ray spectrum acquired using a specific calibration sample, and the elements contained in the calibration sample and their composition ratios as teacher data TD. In this case, abnormal spectra, spectra not used in this system (spectra of elements not measured), etc. may not be acquired as teacher data TD.
Further, from the file name given to the analysis result stored in the storage unit 7 or the like and/or the analysis result stored in the storage device of the other measuring device 1, the analysis result is the result of which substance It is also possible to determine whether or not, and obtain only the necessary analysis results as the teacher data TD.
It is also possible to refer to the measurement results and the like written in the analysis results in the electronic file format, determine which substance the analysis results are for, and obtain only the necessary analysis results as the teaching data TD.

(4) Operation of X-ray Analysis System (4-1) Learning Operation of Learning Unit Hereinafter, the operation of analyzing the elements of the sample S using the X-ray analysis apparatus 100 having the above configuration will be described. First, the learning operation of the learning section 54 until the learning section 54 autonomously outputs the optimum threshold will be described. The following description assumes that the learning unit 54 is a neural network learning model.
First, input parameters of one teacher data TD are input to the learning unit 54 . Next, the value (threshold) output from the learning section 54 when the input parameter is input is compared with the value (threshold) of the output parameter of the teacher data TD in use. Based on the result of this comparison, the "weight" between the two nodes of the neural network is adjusted so that there is no difference between these two values (thresholds). As a method of adjusting this "weight", a known method used in a neural network can be used.

By repeatedly executing the above-described adjustment of the "weight" using other teacher data TD stored in the storage unit 7, if input parameters of the teacher data TD are input to the learning unit 54, A learning model is generated that outputs the value (threshold value) of the output parameter. Note that the learning section 54 may learn by reusing the teacher data TD that has been used once.

(4-2) Elemental Analysis Operation Next, the elemental analysis operation of the sample S using the X-ray analysis apparatus 100 will be described with reference to FIG. FIG. 3 is a flow chart showing the operation of analyzing elements of a sample using the X-ray analysis system.
First, a sample S to be measured is placed on the sample table 11, and X-rays are generated from the X-ray source 13 toward the sample S. As shown in FIG. Fluorescent X-rays are generated from the sample S by irradiating the sample S with the X-rays. The X-ray detector 15 detects fluorescent X-rays generated from the sample S and outputs them to the spectrum generator 17 . The spectrum generating unit 17 generates a fluorescent X-ray spectrum that associates the fluorescent X-ray energy value and the fluorescent X-ray intensity from the fluorescent X-ray energy value and the count number received from the X-ray detection unit 15 .
Hereinafter, the fluorescent X-ray spectrum generated from the sample S to be measured will be referred to as a "second fluorescent X-ray spectrum".

After that, the analysis unit 5 acquires the second fluorescent X-ray spectrum generated by the spectrum generation unit 17 in step S1. The acquired second fluorescent X-ray spectrum is input to the peak separation section 51 . Elemental qualification is then performed using an automatic qualification algorithm.
Specifically, the peak separating unit 51 that has acquired the second fluorescent X-ray spectrum performs peak separation of the second fluorescent X-ray spectrum in step S2.

After that, in step S3, the correlation coefficient calculation unit 52 uses the second fluorescent X-ray spectrum and the theory of the elements narrowed down as a result of performing the peak separation in step S2 (elements presumed to be contained in the sample S) Calculate the correlation coefficient with the spectrum.

After calculating the correlation coefficient, the element identification unit 53 identifies the elements contained in the sample S based on the comparison between the correlation coefficient and the threshold. At this time, in step S4, the element specifying unit 53 determines whether to manually set the threshold or use the threshold output from the learning unit 54 as a learning model.

For example, when it is determined not to calculate the threshold value by the learning unit 54 because the number of data in the teacher data TD is insufficient and the learning by the learning unit 54 is not progressing (“Yes” in step S4), the learning unit 54 corrects Therefore, it decides to set the threshold manually.
Thereafter, in step S5, if the correlation coefficient calculated in step S3 is equal to or greater than the manually set threshold value, the element specifying unit 53 determines that the element that generated the theoretical spectrum for which the correlation coefficient was calculated is the sample S to be included in

After specifying the element using the manually set threshold value, in step S6, the acquiring unit 9 acquires the manually set threshold value as an output parameter, and acquires the parameter related to the second fluorescent X-ray spectrum at that time as an input parameter. , these parameters are associated and stored in the storage unit 7 as teacher data TD.

At this time, if the result of specifying the element using the manually set threshold value is incorrect, or if the result of specifying the element is not appropriate, the acquisition unit 9 outputs the threshold when the inappropriate specifying result is output, The parameters relating to the second fluorescent X-ray spectrum at this time may not be adopted as the teacher data TD.
That is, the acquisition unit 9 acquires, as the teacher data TD, the threshold at which the elements contained in the sample S can be appropriately qualified and the parameters related to the second fluorescent X-ray spectrum used when the threshold was obtained. good too. As a result, the learning unit 54 can learn so as to obtain an appropriate target quantitative result using the X-ray analysis apparatus 100, and the learning unit 54 can appropriately perform the qualitative analysis of the sample S.

When acquiring the teacher data TD including the threshold value at which the element contained in the sample S can be appropriately qualified, the acquisition unit 9 obtains only the threshold value at which the element can be appropriately qualified and the parameters related to the second fluorescent X-ray spectrum at that time. New teacher data TD may be generated individually.
Alternatively, the threshold value at which the element could be properly qualified and the parameters related to the second fluorescent X-ray spectrum at that time may be added to the existing teaching data TD including the threshold value at which the element could not be properly qualified.

On the other hand, if it is determined to have the learning unit 54 calculate the threshold (“No” in step S4), the learning unit 54 is caused to output the threshold, and the threshold is input to the element identification unit 53 .
Specifically, first, in step S7, an input parameter (referred to as a second parameter) relating to the second fluorescent X-ray spectrum is input to the learning section 54 . Thereby, the learning unit 54 as a learning model outputs a threshold for the second fluorescent X-ray spectrum. The second parameters include, for example, parameters of the same type as the first parameters described above.

In addition, when the identification success/failure information (information indicating whether or not the identification of the element could not be properly executed) is input to the learning unit 54, and the learning unit 54 is caused to output the optimum threshold value, , "1" (that is, a value indicating that the element was properly identified) is input to the input of the identification success/failure information of the learning unit 54 .
Thereby, the learning unit 54 as a learning model can be instructed to output an appropriate threshold for the second fluorescent X-ray spectrum.

After that, in step S8, if the correlation coefficient calculated in step S4 is equal to or greater than the threshold input from the learning unit 54, the element specifying unit 53 determines that the element that generated the theoretical spectrum for which the correlation coefficient was calculated is the sample S to be included in

After specifying the element using the threshold output from the learning unit 54, the acquisition unit 9 acquires the output threshold as an output parameter, and the input related to the second fluorescent X-ray spectrum input to the learning unit 54 at that time Parameters may be acquired as input parameters, and these parameters may be associated and stored in the storage unit 7 as teacher data TD.

At this time, the acquisition unit 9 acquires, as teacher data TD, the threshold value at which the elements contained in the sample S can be appropriately qualified and the parameters related to the second fluorescent X-ray spectrum used when the threshold value was obtained. may

Further, when the element cannot be appropriately specified using the threshold output from the learning unit 54, the threshold is edited by manual setting, and the element is specified again using the edited threshold. good too.
At this time, if the element can be appropriately specified using the edited threshold, the acquisition unit 9 acquires the edited threshold as an output parameter, and the second input to the learning unit 54 at that time. Input parameters relating to the fluorescent X-ray spectrum may be acquired as input parameters, and these parameters may be associated and stored in the storage unit 7 as teacher data TD.

Note that the element specifying unit 53 may quantify the element based on the result of specifying the element. In this case, it may be determined that the element is not contained in the sample S when it is recognized that there is an error in the quantification result for a certain element.

After specifying the elements, the elements specified to be contained in the sample S are output as the analysis results of the sample S. FIG.
As the analysis result of the sample S, the quantification unit 55 may output the quantification result in addition to the element specified as being contained in the sample S. In addition, this quantitative result may include information on the material of the sample S specified based on the element concentration (composition ratio). The output quantitative results are displayed on the display of the information processing device 3, for example.

Since the fluorescent X-ray spectrum acquired by the measuring device 1 may include information about the crystal structure of the sample S, the element identifying unit 53 obtains the element identification result (and the second fluorescent X-ray spectrum ), the crystal structure of the sample S may be output as the analysis result of the sample S.

(5) Summary In this way, by machine learning the threshold used for comparison with the correlation coefficient, the threshold, which has been determined based on experience so far, can be changed by experience (fluorescence X It is possible to automatically calculate the optimum threshold by machine learning of the tendency that appears in the line spectrum, etc.
In addition, in machine learning, it is also possible to acquire data trends that the user has not been able to grasp through learning, so by using a learning model to output thresholds, it is possible to output more appropriate thresholds than ever before.

(6) Modification of Analysis Section According to First Embodiment The analysis section 5 described above learns the optimum threshold value used for comparison with the correlation coefficient calculated in step S3 by machine learning. However, without being limited to this, machine learning can also be used in other processes of the automatic qualitative algorithm described above.
For example, in step S3 described above, the parameter type and its value for calculating the correlation coefficient between the second fluorescent X-ray spectrum and the theoretical spectrum may be machine-learned. In the learning model for this learning, the fluorescent X-ray spectrum, the theoretical spectrum, the elemental data ED of the element that generates the theoretical spectrum, etc. are used as input parameters, and the parameter type and its value used to calculate the correlation coefficient, the correlation The energy width of the spectrum used for calculation of the number, etc. can be used as an output parameter.
As a result, the optimum correlation coefficient can be automatically calculated without relying on experience or the like.

2. Second Embodiment Machine learning can be applied not only to the automatic qualitative algorithm described in the first embodiment, but also to other algorithms for analyzing the sample S based on the fluorescent X-ray spectrum.
For example, (1) for all elements, the degree of coincidence between the peak position of the theoretical spectrum of each element and the peak position of the measured fluorescent X-ray spectrum is calculated as a value of 0 to 100, (2 ) Machine learning can also be applied to the qualitative analysis of elements by an algorithm that specifies whether or not the element is contained in the sample S depending on the range of the degree of coincidence.

Note that the basic functions and configuration of the X-ray analysis apparatus 100 are substantially the same as those described in the first embodiment even when qualitatively determining elements using other algorithms such as those described above. Therefore, description of the configuration of the X-ray analysis apparatus according to the second embodiment is omitted here.

When using the above-described qualitative algorithm, for example, the range of matching values for determining whether a specific element is contained in the sample S can be learned by machine learning.
In this case, a learning model for machine learning can receive, for example, parameters used for calculating the degree of coincidence as an input, and can output a range of value of the degree of coincidence. Parameters for calculating the degree of coincidence include, for example, the peak position, peak ratio, and peak area of the measured fluorescent X-ray spectrum, and the peak position, peak ratio, and peak area of the theoretical spectrum.

3. Third Embodiment In the first and second embodiments described above, the analysis (qualitative) of the elements contained in the sample S is performed by a specific algorithm. However, without being limited to this, the sample S may be analyzed directly from the measured fluorescent X-ray spectrum using machine learning.
Even when the sample S is analyzed directly from the measured fluorescent X-ray spectrum, the basic functions and configuration of the X-ray analysis apparatus 100 are substantially the same as those described in the first embodiment. be. Therefore, description of the configuration of the X-ray analysis apparatus according to the third embodiment is omitted here.

In this case, the analysis unit 5 can be a learning model that receives the measured fluorescent X-ray spectrum and outputs the presence or absence of each element contained in the sample S. For example, the analysis unit 5 can be a learning model configured by a neural network such as deep learning.
In such an analysis unit 5, for example, other information about the sample S, such as the content of each element contained in the sample S, the material of the sample S, and the crystal structure, may be included in the output.
In the analysis unit 5, for example, the fluorescent X-ray spectrum is used as an input parameter for the teaching data TD, and the elemental analysis results (for example, automatic qualitative results, manual qualitative results, etc.) are used as output parameters for the teaching data TD. .

4. Fourth Embodiment In the first to third embodiments described above, the X-ray analysis apparatus 100 was a stand-alone analysis apparatus. However, not limited to this, the X-ray analysis system 200 according to the fourth embodiment, as shown in FIG. You may be connected so that communication is possible.
FIG. 4 is a diagram showing an X-ray analysis system according to a fourth embodiment.

In this case, the X-ray analysis apparatuses 100 may exchange data with each other on a peer-to-peer (P2P) basis. For example, the teacher data TD, the actually measured fluorescent X-ray spectrum, the learning model of the learning unit 54, and the like, which are owned by each other, can be shared with each other.
Also, some X-ray analysis devices 100 may be set to restrict communication with other X-ray analysis devices 100 .

5. Fifth Embodiment Further, as shown in the fourth embodiment, not limited to the case where a plurality of X-ray analysis apparatuses 100 are connected to each other, as shown in FIG. In the analysis system 300, (a plurality of) X-ray analysis apparatuses 100 may be collectively connected to a server 10 connected to a network.
FIG. 5 is a diagram showing an X-ray analysis system according to a fifth embodiment.

The server 10 is, for example, a large-scale computer system composed of a CPU, large-capacity storage devices (RAM, ROM, SSD, hard disk, etc.), and various input/output interfaces.

In the fifth embodiment, the server 10 downloads fluorescent X-ray spectra measured by each of the X-ray analyzers 100 from a plurality of X-ray analyzers 100 connected to the server 10, may be stored in the storage device of

Also, the server 10 may have a function corresponding to the analysis unit 5 described in the first to third embodiments above. A function corresponding to this analysis unit 5 is realized by a program executed by the server 10 .
In this case, the server 10 may use the fluorescent X-ray spectrum downloaded from each X-ray analysis device 100 as the training data TD to allow the analysis unit to learn the corresponding functional part. Furthermore, the server 10 itself may automatically analyze the sample S using the analysis unit as a learning model and the fluorescent X-ray spectra collected as the teacher data TD.
Furthermore, the server 10 itself may generate a virtual X-ray fluorescence spectrum and automatically perform elemental analysis using the X-ray fluorescence spectrum. Also, the results of elemental analysis conducted by the user himself/herself may be used as the training data TD.

In this way, the server 10 collects fluorescent X-ray spectra as training data TD from each X-ray analysis apparatus 100, and the server 10 itself analyzes the elements to generate training data TD. Data TD can be held.

In addition, the server 10 may provide each X-ray analysis apparatus 100 with part or all of the vast amount of teacher data TD it owns in accordance with a request from each X-ray analysis apparatus 100 . Furthermore, the server 10 may provide each X-ray analysis apparatus 100 with a learning model generated using a huge amount of teacher data TD according to a request from each X-ray analysis apparatus 100 .

6. Other Embodiments Although a plurality of embodiments of the present invention have been described above, the present invention is not limited to the above embodiments, and various modifications are possible without departing from the gist of the invention. In particular, multiple embodiments and modifications described herein can be arbitrarily combined as needed.
In the analysis operation described using the flowchart of FIG. 3, the processing contents of each step and/or the order of processing of each step can be changed without departing from the gist of the invention.

INDUSTRIAL APPLICABILITY The present invention can be widely applied to devices and/or systems for qualitatively determining a measurement target based on the fluorescent X-ray spectrum generated from the measurement target.

100 X-ray analysis device 200, 300 X-ray analysis system 1 Measurement device 11 Sample table 13 X-ray source 15 X-ray detection unit 17 Spectrum generation unit 19 Control unit 3 Information processing unit 5 Analysis unit 51 Peak separation unit 52 Correlation coefficient calculation unit 53 element identification unit 54 learning unit 55 quantification unit 7 storage unit 9 acquisition unit 10 server ED element data S sample TD teacher data

Claims (10)

a measuring device for measuring a fluorescent X-ray spectrum generated from a sample to be measured;
A learning model trained using teacher data including a first parameter related to a first fluorescent X-ray spectrum and first result information related to an analysis result when using the first parameter was measured by the measuring device. an analysis unit that analyzes the sample that generated the second fluorescent X-ray spectrum based on second result information obtained by inputting a second parameter related to the second fluorescent X-ray spectrum;
with
The analysis unit analyzes based on whether or not a correlation coefficient representing a degree of matching between the second fluorescent X-ray spectrum and a theoretical spectrum of an element presumed to be contained in the sample is equal to or greater than a predetermined threshold. and
The result information about the analysis result is the threshold used for comparison with the correlation coefficient and serving as an index for determining whether the sample contains the element.
X-ray analyzer. A third parameter related to the third fluorescent X-ray spectrum measured by the measuring device is obtained as the first parameter, and third result information related to the analysis result obtained using the third parameter is the first result information. The X-ray analysis apparatus according to claim 1 , further comprising an acquisition unit that acquires as . The teacher data includes, as the first result information, third result information that was appropriately analyzed among the third result information, and the third parameter used when obtaining the third result information is the 3. An X-ray analysis apparatus as claimed in claim 2 , comprising as a first parameter. If the correlation coefficient between the fluorescent X-ray spectrum and the theoretical spectrum is equal to or greater than a predetermined threshold, the analysis unit identifies the element that produces the theoretical spectrum as an element contained in the sample . 4. The X-ray analysis device according to any one of 3 . 5. The teaching data according to any one of claims 1 to 4 , wherein the teacher data includes information on a fluorescent X-ray spectrum obtained using a calibration sample, and an element contained in the calibration sample and a composition ratio of the element. X-ray analyzer. an X-ray analysis apparatus according to any one of claims 1 to 5 ;
a server connected to the X-ray analysis device;
An X-ray analysis system, comprising: 7. The X-ray analysis system according to claim 6 , wherein said server has a function corresponding to said analysis unit. An X-ray analysis system in which a plurality of X-ray analysis devices according to any one of claims 1 to 5 are interconnected. a step of measuring a fluorescent X-ray spectrum generated from a sample to be measured;
a step of learning a learning model using teacher data including a first parameter regarding a first fluorescent X-ray spectrum and first result information regarding an analysis result when using the first parameter;
Generate the second fluorescent X-ray spectrum based on second result information obtained by inputting a second parameter related to the second fluorescent X-ray spectrum obtained by measurement into the learning model that has been trained. analyzing the sample;
with
The step of analyzing the sample is based on whether or not a correlation coefficient representing a degree of matching between the fluorescent X-ray spectrum and a theoretical spectrum of an element presumed to be contained in the sample is equal to or greater than a predetermined threshold. , including the step of performing the analysis,
The result information about the analysis result is the threshold used for comparison with the correlation coefficient and serving as an index for determining whether the sample contains the element.
Analysis method. A program that causes a computer to execute the analysis method according to claim 9 .

Images