JPH0373833A - Automatic qualitative analyzer - Google Patents

Abstract

PURPOSE:To allow the exact estimation of the probability of the existence of the materials in a sample by inferring by using the intrinsic production rules set in conditions for each of the respective materials. CONSTITUTION:The criterion data of respective X-ray spectral crystals are read out of a wavelength table 7 and the peak data are read out for each of the respective line spectral crystals from a peak data file 3. An inference control section 6 judges whether the peak data exists in the respective read out peak data files or not. A control section 6 makes the backward inference while referring to the results of matching with respect to the peak data of the X-ray spectral crystals where the peak data exists. Whether there is the peak corresponding to the main line or sub-line of the element which can be analyzed in the peak data or not is then judged by comparing the peak data and the criterion data. The element names and line names which are the candidates for the respective peaks are then put on. The forward inference is executed in order to ascertain the elements enumerated for the candidates.

Description

DETAILED DESCRIPTION OF THE INVENTION A. Field of Industrial Application This invention is directed to an electron beam microanalyzer (EPMA).
, detects peak data from spectral data collected with qualitative analyzers such as fluorescent X-ray spectrometers, powder X-ray spectrometers, and infrared absorption spectrometers, and automatically identifies the substance name corresponding to each peak data. The present invention relates to an automatic qualitative analysis device that makes qualitative judgments.

B. Prior art An overview of the analytical processing using a conventional automatic qualitative analyzer is given in EP.
This will be explained by taking an MA automatic qualitative analysis device as an example.

(2) Characteristic X-rays obtained by irradiating the sample surface with an electron beam are simultaneously analyzed spectroscopically using multiple X-ray spectroscopy crystals.

(2) Spectral data for each X-ray spectroscopy crystal contains many peaks corresponding to the elements contained in the sample surface, and peak data including the wavelength and intensity of these peak portions is detected.

■ The most suitable peak for identifying the element is selected from these peak data.Normally, the peak with the highest intensity is selected as such a peak. It is determined whether the main line or subline corresponding to this peak exists in the wavelength table stored in advance in the data file, and if so, the element name and line name are identified for that peak (
At the same time, the element name and line name are similarly identified for other peaks that would appear if the element exists.

Here, the wavelength table refers to EPM as shown in Figure 3.
It is created for each X-ray spectroscopy crystal equipped in A, and associates all elements that can be qualitatively analyzed with each X-ray spectroscopy crystal with elemental lines, and calculates their peak wavelengths and peak-to-peak intensities. A table that describes judgment criteria data including ratios, etc. Further, the elemental ray is a general term for a plurality of peaks that appear when characteristic X-rays of a certain element are spectroscopically analyzed.

The element line includes a main line that is the most suitable peak for specifying the element, a sub line that is the next most suitable peak, and higher-order lines that are other peak groups.

■Next, from among the remaining peak data not identified by above ■, the most appropriate peak for identifying the element is selected, and the wavelength table is referred to in the same manner as above.
The element name and line name are identified for that peak, and other peaks related to that element are also identified in the same way.

■ Thereafter, element names and line names are similarly identified in descending order of peaks.

A series of processes as described above are performed on peak data for each X-ray spectroscopic crystal.

Problems to be Solved by the C0 Invention However, the above-mentioned conventional device has the following problems.

In other words, in qualitative analysis using conventional equipment, as mentioned above, the substance name is identified using a uniform procedure for each substance to be identified, so it is difficult to identify the substance name when there is a large number of peak data or when a special peak is detected. In such cases, there is a problem that the peak cannot be determined or an incorrect determination may be made. Therefore, according to the conventional device, there is a problem in that an expert has to add or modify the determination result after performing qualitative analysis using the device.

The present invention has been made in view of the above circumstances, and provides an automatic qualitative analysis device that can automatically perform accurate qualitative analysis based on peak data detected from collected spectrum data. It is intended to.

Means for Solving the Problems In order to achieve the above-mentioned object, the present inventors have proposed, for example, an EPM
Regarding A, by acquiring, organizing, and creating rules from experts' experiential knowledge as described below, we have realized an automatic qualitative analysis device that is more reliable than conventional devices.

(1) Acquisition of knowledge We asked experts who routinely perform qualitative analysis using EPMA to make detailed judgments on several types of data, and acquired empirical knowledge.

Analytical experts integrate the spectroscopy data from multiple X-ray spectroscopy crystals and identify each peak based on the peak wavelength, peak-to-peak intensity ratio, half-width, etc. of the elements that can be spectrally analyzed by each X-ray spectroscopy crystal. The determination method is especially complicated when multiple elements correspond to a single peak, in which the element name and line name corresponding to the peak are assigned.

(2) Organizing knowledge As a result of organizing the acquired knowledge into rules, it was found that the knowledge of experts regarding judgments can be classified into ``knowledge about judgments'' and ``knowledge about procedures.''

"Knowledge regarding judgment" refers to knowledge for estimating which line of which element the peak collected as data corresponds to. Such knowledge differs for each element that can be analyzed by each X-ray spectroscopy crystal.

"Knowledge regarding the procedure" refers to the procedure for determining which peak to start with among the many sampled peaks. Judgment] The order is determined according to the intensity of the sampled peaks. In other words, among the data on the short wavelength side where there are relatively few competing elements, the determination is made starting from the one with the highest and clearest peak intensity.

(3) Creating rules for knowledge A knowledge base for the expert system was created by expressing organized specialized knowledge using production rules in the IF-THEN format.

The knowledge base is divided into multiple X-ray spectroscopy crystals, and the "judgment knowledge" of each X-ray spectroscopy crystal is created as a knowledge base for backward reasoning, and the "procedural knowledge" of each X-ray spectroscopy crystal is '' was created as a knowledge base for forward inference due to its nature.

In addition, considering the ease of writing rules and the need for different judgment conditions for each element, the production rules that make up each knowledge base are written for each element.

From the above, the automatic qualitative analysis device according to the present invention is
It has the following characteristics:

First, a criterion data storage means stores individual criterion data that serves as a standard for identifying each substance for a plurality of Class II substances that can be analyzed by the apparatus, and the spectral data a matching processing means that matches the peak data in the data with the judgment standard data, and a matching processing means that matches the peak data with reference to the matching results and uses a production rule in the rlF~THEN...1 format for each substance to match the peak data. and a backward inference means for backward inferring the name of the substance.

Second, in addition to the first feature, a peak intensity calculation means calculates a peak intensity as necessary for a candidate substance determined to have a possibility of existence by the backward inference means, and the peak intensity calculation means Forward inference that determines the existence of the candidate substance using a production rule in the form of "IF ~ THEN..." for each substance, which includes conditions for identifying the substance name with reference to the peak intensity calculated by equipped with the means.

Effect 1: According to this invention, the substance name corresponding to each peak data is retrospectively inferred while referring to the result of the matching process between the peak data detected from the spectrum data and the criterion data. , since individual production rules are used for each material, inferences are made accurately according to the unique conditions of each material.

In backward inference, for example, if multiple types of candidate elements are listed for one peak data, the intensity of the main peak of each candidate element is calculated as necessary, and the results are referred to. The existence of the candidate element is determined by performing forward inference.

F. Example Embodiments of the present invention will be described below based on the drawings.

FIG. 1 is a block diagram showing the schematic configuration of an EPMA automatic qualitative analysis device that is an embodiment of the present invention, and FIG. 2 is a block diagram showing the program modules and data flow of the expert system part of the device. .

In Fig. 1, numeral 1 is an electron beam microanalyzer (
This EPMAI is equipped with a plurality of X-ray spectroscopic crystals (not shown) that simultaneously detect characteristic X-rays from a sample. The EPMA automatic qualitative analyzer according to this example uses lithium butoxide (LiF), pentaerythritol (PET), rubidium acid phthalate (RAP), ammonium dihydrogen phosphate (ADP), and stearin as X-ray spectroscopic crystals. Lead acid (P b ST) can be used. Usually, 3.4 types of X-ray spectroscopy crystals are used as appropriate. Of course, it is also possible to use other types of X-ray spectroscopic crystals by appropriately changing and expanding the wavelength table 7, knowledge base 8, etc., which will be described later.

Spectral data obtained by each X-ray spectroscopy crystal of EPMAI is provided to a data acquisition section 2. The data collection section 2 is
Spectral data obtained by qualitative analysis (in this example, point analysis) of EPMAI is analyzed to detect peak data including the wavelength, peak intensity, half-value width, etc. of data that can be considered as a peak. The detected peak data is stored in the peak data file 3.

Reference numeral 4 is an expert system for inferring element names and line names corresponding to each peak data stored in the peak data file 3, and a data processing unit 5. Inference control unit 6. It consists of a wavelength table 7 and a knowledge base 8.

As shown in FIG. 2, the data processing unit 5 reads the peak data of each X-ray spectroscopy crystal from the peak data file 3, and also identifies the main lines of elements that can be analyzed for each x1 spectroscopy crystal from the wavelength table 7. A data reading program module 5a for reading the judgment standard data of the sub-line, a matching program module 5b for matching the read judgment standard data and the peak data, and a backward inference for the peak data. A naming program module 5C that assigns element names and line names to elements that have been selected as candidates; a peak intensity calculation program module 5d that calculates the intensity of peak data; and a writing program module 5e for writing the determined result of the inference into the peak data file 3.

Knowledge base 8 is meta-knowledge 8 that controls the entire flow of inference.
a, module knowledge 8b for checking the presence or absence of data for each X-ray spectroscopy crystal, and the five types of X-ray spectroscopy crystals (LiF, P
ET, RAP, ADP, Pb5T) module knowledge 8c for backward reasoning expressing "knowledge regarding judgment"
, 8e, 8g8i, 8, forward inference module knowledge 8d, 8f, 8h8 expressing "procedural knowledge"
j, 8ffi.

Each of the above-mentioned knowledge 88 to 8Q is organized expert knowledge expressed in a production rule in the form of "IF~THEN...". The inference control unit 6 uses the knowledge base 8
Inferences as described below are made according to the module knowledge 8a to 82 stored in the module.

As shown in FIG. 3, the wavelength table 7, which serves as a criterion data storage means, contains information about each X-ray spectroscopic crystal (LiF, P
For each element line that can be spectroscopically analyzed by ET, RAP, ADP, Pb5T), the peak wavelength
Peak-to-peak intensity ratios, half-widths, and the like are stored as judgment standard data.

Returning to FIG. 1, the inference results obtained by the expert system 4 are written to the peak data file 3. The display control unit 9 reads the inference results written in the peak data file 3, displays the element name and line name for each peak in the peak waveform of each X-ray spectroscopy crystal displayed on the CRTlo, and also displays the identification. Necessary data such as wavelength, intensity ratio, and half-width are displayed for the detected peak.

Next, with reference to the flowchart shown in FIG. 4, the inference processing executed in the expert system 4 of the apparatus according to the above-described embodiment will be specifically explained.

Step S1: The data reading program module 5a of the data processing unit 5 is activated, the judgment standard data for each X-ray spectroscopic crystal is read from the wavelength table 7, and the peak data for each X-ray spectroscopic crystal is read from the peak data file 3. is read out.

The read criterion data is given to the data processing section 5, and the peak data is given to the data processing section 5 and the inference control section 6, respectively.

The step Sl peak data file 3 contains the above-mentioned 5
There are files for individually storing the peak data of each type of X-ray spectroscopy crystal.

The inference control unit 6 determines whether peak data exists in each read peak data file based on the module knowledge 1118b for crystal data confirmation of the knowledge base 8,
Perform inference processing on data that exists.

Step Sl inference control unit 6 extracts meta knowledge 8a from knowledge base 8, module knowledge 8c, 8e for backward inference,
Read the module knowledge about the X-ray spectroscopy crystal for which peak data exists among 8g, 8i, and 8, and read out the module knowledge for the X-ray spectroscopy crystal for which peak data exists.
For all the elements in the wavelength table of the X-ray spectroscopy crystal, the matching program module 5b of the data processing section 5 is activated as needed, and backward inference is performed while referring to the matching results in step S4.

The rules as knowledge for determining the possibility of the existence of elements are:
They differ for each X-ray spectroscopic crystal and even for each element, and there are hundreds of them, so examples are shown below.

For example, in a production rule for determining whether a peak of element A1 appears in the data of X-ray spectroscopy crystal C1, With crystal C2, there is no possibility that the peak of element A1 does not exist in the data of X-ray spectroscopy crystal C2. ■ The peak most suitable for determining element A1 in the data of X-ray spectroscopy crystal C1 (main ■ If there is a possibility that the main line has not been detected due to interference from other element lines, a peak (sub line) similar to the main line has been detected. ■ If the half-width and peak shape are similar to those of element A1, etc., it is assumed that there is a possibility of the existence of element AI, and element AI is selected as a candidate element. .

As described above, the production rule conditions differ for each element; for example, which X-ray spectroscopy crystal data is prioritized differs for each element.

Also, depending on the element, the peak line may not be affected by interference, so when determining such elements, the determination may be made based only on the peak line; conversely, the peak line may not be affected by interference. If the element is easily sensitive, the presence or absence of sub-lines may be determined preferentially. Furthermore, the presence of many elements can be estimated by paying attention to their peak wavelengths, so the above conditions (■) and (■) are often sufficient, but for elements that are difficult to determine based only on the peak wavelength. The above condition ■ is added.

Step S4: The matching program module 5b of the data processing section 5 is activated by the inference control section 6 as needed during the backward inference of step S3, and the matching program module 5b of the data processing section 5 is detected as an element that can be analyzed by the corresponding X-ray spectrometer crystal in the peak data. It is determined whether there is a peak corresponding to the main line or sub-line by comparing the peak data with judgment reference data (wavelength, inter-peak intensity ratio, half-width of each main line/sub-line).

Step S52: When candidate elements are listed for the peaks of each X-ray spectroscopic crystal through the above backward inference, the naming program module 5b of the data processing section 5 is activated, and the naming program module 5b of the data processing section 5 is activated to assign the candidate element name to each peak. A line name can be added. Once the main line or sub-line of a certain element is identified, the wavelength table of the X-ray spectroscopy crystal can be used to determine the wavelength positions of other higher-order lines, so if there is a peak in that area, the same peak can also be detected. be named.

Step S6: According to the backward inference in step S3, two or more element names may be listed as candidates for the same peak. In such a case, all the element names listed as candidates may be correct, but there are also cases where one candidate element is correct and the other candidate elements are incorrect. Therefore, in the following processing, forward inference is performed in order to determine the elements that have been tentatively listed as candidates by backward inference. For this forward reasoning, module knowledge 8d, 8
f, 8h, 8j, and 81 are read out.

Similar to the rules for backward inference, there are many rules for forward inference for each element that can be analyzed by each X-ray spectroscopic crystal, and examples thereof are shown below.

For example, in the production rule for determining element A1, which is estimated to exist in the data of X-ray spectroscopic crystal C1 by the backward inference, ■ element A1 is listed as a candidate element, ■ peak corresponding to element A1. The element lines (peaks) of other elements do not overlap, ■ If the peaks of other elements overlap, calculate the peak intensity of each element, and as a result, confirm that the peak of element A1 overlaps. If the conditions such as being able to be determined are satisfied, the element name and line name identified as candidate element A1 are determined to be correct. On the other hand, if the existence of the candidate element is denied in the forward inference, the naming of the candidate element performed in step S5 is deleted.

In such forward inference, by referring to the results of the peak intensity calculation program module 5d of the data processing unit 5 and proceeding with inference in the order of the candidate elements that can be determined, the presence or absence of all candidate elements is determined. will be finalized.

As mentioned above, the production rules used in forward inference are also determined for each element, and they generally have the characteristics described in ■ or ■ above. However, for some elements, peaks cannot overlap, so peak intensities are not referenced in production rules for determining such elements.

Step S7: As described above, when the existence of candidate elements is confirmed by forward inference for all the elements listed as candidates for each X-ray spectroscopic crystal, the results are written in the data processing unit 5. is written to the peak data file 3 by the program module 5e.

As described above, lt by expert system 4
When the theoretical processing is completed, the display control section B 9 is activated as described above.
reads the inference result from the peak data file 3 and displays it on the CRTIO or outputs it to a printer (not shown).

The following table shows the accuracy rate of the determination results made in the above embodiment in comparison with the conventional device.

(Hereinafter, blank space) Table 1. In addition, the production rules explained in the above example,
If knowledge about the material of the sample (for example, whether the sample is alloy-based, mineral-based, compound-based, ceramic-based, etc.) is added as a condition,
It is advantageous in the following points.

(2) In other words, the peak wavelength for each element is stored with a certain width (range) in the wavelength table 7, which serves as a judgment criterion data storage means. In some cases, the peak wavelength may shift and deviate from the range of the peak wavelength as the criterion data. In such a case, if the operator provides knowledge about the material of the sample as a production rule condition in advance, and performs the matching process using criteria data with the peak wavelength range changed accordingly, the above problem can be solved. Even if there is a shift in the peak wavelength, the presence of the element can be estimated, making it possible to make a highly accurate determination.

■ Also, depending on the sample, elements that can never exist may be known in advance, so in such cases,
By excluding that element from the matching process etc. from the beginning, it is possible to speed up the determination.

Furthermore, although the above-described embodiments have been explained using EPMA as an example, the present invention can also be applied to the following qualitative analysis apparatus.

(1) Fluorescent X-ray analyzer A fluorescent X-ray analyzer is a device that irradiates a sample with X-rays and spectrally analyzes the emitted secondary X-rays, but the detected spectral data is EPMA. Therefore, the same element determination method as in the above-mentioned embodiment can be applied.

(2) Powder X-ray spectrometer A powder X-ray spectrometer is a device that detects the angle at which X-rays irradiated onto a powder sample are reflected. The detected reflection angle is unique depending on the compound in the sample (for example, quartz, iron oxide, etc.), and the detected peak data is matched with pre-stored judgment standard data for multiple types of compounds. By doing so, the compounds in the sample are identified. In this identification process, compounds in the sample can be inferred using production rules created for each compound that can be analyzed.

(3) Infrared absorption spectrum analyzer An infrared absorption spectrum analyzer is a device that irradiates a sample with infrared light whose wavelength is continuously changed and measures the spectrum of the infrared light that is absorbed at that time. Infrared absorption is
Since it shows a spectrum specific to the bonding group in the sample molecule,
Spectral data of a plurality of types of substances that are expected to exist are stored in a database in advance as criterion data, and molecules in a sample can be identified by matching each criterion data with detection data. With this analysis method, multiple spectra are obtained for one molecule, and these data may overlap with data of other molecules, so the present invention can also be applied to the determination of such data. be able to.

Effects of the G6 Invention As is clear from the above explanation, according to this invention, peak data obtained from collected spectral data and determination for each substance that serves as a standard for identifying the substance to be analyzed. When inferring the substance name of the peak data by referring to the results of matching with reference data, the inference is made using a unique production rule with conditions set for each substance. The possibility of the presence of a substance in a sample can be estimated more accurately than when each substance is determined using a uniform rule, such as in the following.

In addition, after selecting elements that may exist as candidates through backward inference, we use the production rules for each substance to determine the candidate elements, and if necessary, we use the production rules for each substance. Since the process determines whether the peak intensity of a candidate substance is confirmed, even if two or more substance names are listed as candidates for one peak data in backward inference, each candidate substance's peak intensity is determined. The possibility of existence can be determined with high accuracy.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the schematic configuration of an EPMA automatic qualitative analysis device according to an embodiment of the present invention, FIG. 2 is a block diagram showing the program modules and data flow of the expert system part of the device, and FIG. The figure is an explanatory diagram of the wavelength table, and FIG. 4 is a flowchart of inference processing performed by the expert system. 1...Electron beam microanalyzer (EPMA) 2...
・Data collection unit 3...Peak data file 4...Expert system 5...Data processing unit 5b...Matching program module 5d...
- Program module for peak intensity calculation 6...Inference control unit 7...Wavelength table 8...Knowledge bases 8c, 8e, 8g, 8i, 8k...Module knowledge for backward inference 8d, 8f, 8h, 8j ,BN...Module knowledge for forward reasoning

Claims (2)

[Claims] (1) In an automatic qualitative analyzer that detects peak data from collected spectral data and identifies the substance name corresponding to each peak data, each of the multiple types of substances that can be analyzed with the equipment is a criterion data storage means that stores individual criterion data serving as a reference for identifying a substance; a matching processing means that matches peak data in the spectrum data with the criterion data; and a result of the matching. and a backward inference means for backward inferring the name of a substance corresponding to the peak data using an "IF~THEN..." format production rule for each substance while referring to Analysis equipment. (2) In the automatic qualitative analysis apparatus according to claim (1), a peak intensity calculation means for calculating a peak intensity, if necessary, for a candidate substance determined to have a possibility of existence by the backward inference means; "IF~THEN..." for each substance, including conditions for identifying the substance name by referring to the peak intensity calculated by the intensity calculation means.
and a forward inference means for determining the existence of the candidate substance using production rules of the form "."