WO2023100854A1

WO2023100854A1 - Analysis system, server, analysis method, and program

Info

Publication number: WO2023100854A1
Application number: PCT/JP2022/043916
Authority: WO
Inventors: 佳津紀山口; 翔太山渡; 絵里佳松本; 恭兵西澤
Original assignee: 株式会社堀場製作所
Priority date: 2021-12-03
Filing date: 2022-11-29
Publication date: 2023-06-08

Abstract

In the present invention, extraction of features included in data pertaining to particulate matter is performed efficiently and accurately. An analysis system (100) comprises a data acquisition unit (1), a feature extraction unit (35), and an output unit (37). The data acquisition unit (1) acquires related data (RD) related to the particulate matter. The feature extraction unit (35) executes prescribed feature extraction processing that takes the related data (RD) as input to extract features included in the related data (RD). The output unit (37) outputs information relating to the features extracted by the feature extraction unit (35).

Description

Analysis system, server, analysis method, and program

The present invention relates to an analysis system for analyzing particulate matter, a server for analyzing particulate matter, an analysis method for particulate matter, and a program for causing a computer to execute the analysis method.

In recent years, various particulate matter (eg PM2.5) has become a major environmental problem. For this reason, data on particulate matter such as the concentration of particulate matter contained in a predetermined area and information on elements contained in particulate matter (e.g., elements contained in particulate matter and the content of such elements) are acquired. Then, a system is known that analyzes particulate matter based on the obtained data (see Patent Document 1, for example).

WO2018/117146

In order to perform an accurate analysis of particulate matter, it is necessary to extract the features contained in the data on particulate matter, but to do so, it is necessary to analyze the large amount of data obtained. In conventional systems, users analyze a large amount of data to extract features, so it is difficult to efficiently and accurately extract features contained in data on particulate matter.

The object of the present invention is to efficiently and accurately extract features contained in data on particulate matter.

A plurality of aspects will be described below as means for solving the problem. These aspects can be arbitrarily combined as needed.
An analysis system according to one aspect of the present invention is a system for analyzing particulate matter. The analysis system includes a data acquisition section, a feature extraction section, and an output section. The data acquisition unit acquires relevant data regarding particulate matter. The feature extraction unit extracts features included in the related data by executing a predetermined feature extraction process with the related data as input. The output unit outputs information related to the features extracted by the feature extraction unit.

In the above analysis system, the feature extraction unit automatically executes a predetermined feature extraction process with input of relevant data related to particulate matter obtained by the data acquisition unit, and automatically extracts features included in the relevant data. are extracted explicitly. In this manner, the features included in the related data are automatically extracted by the feature extraction unit, so that the features included in the related data can be efficiently and accurately extracted. Since the features included in the relevant data characterize the particulate matter to be analyzed, efficient and accurate extraction of the features included in the associated data enables efficient analysis of the particulate matter. Further, the output unit outputs information related to the features extracted by the feature extraction unit, so that the user can be shown what features have been extracted from the related data.

In the above analysis system, the feature extraction unit may extract related data whose instantaneous value exceeds the first threshold. In this case, the output unit may display a list of information related to related data whose instantaneous value exceeds the first threshold. As a result, it is possible to automatically extract relevant data whose instantaneous value exceeds the first threshold value and include anomaly, and to present to the user in which relevant data an anomaly has occurred.

In the above analysis system, the feature extraction unit may extract related data whose average value or median value exceeds the second threshold. In this case, the output unit may display a list of information related to related data whose average value or median value exceeds the second threshold. Thereby, it is possible to automatically extract related data whose average value or median value exceeds the second threshold value and include anomaly, and to present to the user in which related data an anomaly has occurred.

In the above analysis system, the output unit may graphically display related data corresponding to the specified information among the above listed information. This makes it possible to visually confirm changes in related data including anomalies.

In the analysis system described above, the feature extraction unit may calculate the correlation of a plurality of related data and extract information related to the related data for which the calculated correlation is equal to or greater than the third threshold. This eliminates the need for the user to calculate and analyze correlations for various combinations of multiple pieces of related data, so that related data with high correlation can be efficiently and accurately extracted.

In the above analysis system, the output unit may display a scatter diagram of a plurality of related data whose correlation is equal to or greater than the third threshold. This makes it possible to visually confirm the magnitude of the correlation of the extracted related data.

In the above analysis system, the output unit may display a plurality of scatter diagrams of the plurality of related data and highlight the scatter diagrams of the plurality of related data whose correlation is equal to or greater than the third threshold. Thereby, it is possible to visually recognize which of all related data has a large correlation.

In the above analysis system, the relevant data may be data that changes over time. In this case, the feature extraction unit may calculate the correlation of multiple pieces of related data in a predetermined time interval. This allows calculation of the correlation of multiple pieces of related data in a specific time interval. Correlation of relevant data in a particular time interval provides information about events that occurred during that time interval.

In the above analysis system, the above predetermined time interval may be variable. This makes it possible to flexibly set the target time interval for calculating the correlation of a plurality of pieces of related data.

In the analysis system described above, the feature extraction unit may calculate the correlation of the plurality of related data in each of the plurality of small intervals included in the predetermined time interval. This makes it possible to obtain more detailed information about an event that occurred in a specific period by correlating related data in a specific small interval.

The output unit may graphically display chronological changes in a plurality of related data whose correlation is equal to or greater than the third threshold. Thereby, it is possible to visually confirm changes over time in a plurality of related data.

In the above analysis system, the data acquisition unit may acquire the mass concentration of the particulate matter and information related to the elements contained in the particulate matter as related data. This allows analysis of the particulate matter based on features extracted from information about the mass concentration of the particulate matter and/or the elements contained in the particulate matter.

In the above analysis system, the data acquisition unit may acquire the wind direction at the location where the particulate matter was collected as related data. In this case, the feature extraction unit determines whether the instantaneous value of the content of the element contained in the particulate matter in the related data in the specific wind direction exceeds the first threshold, or the content of the element included in the particulate matter Relevant data in which the mean or median of exceeds a second threshold may be extracted. As a result, it is possible to extract related data indicating anomalies regarding particulate matter flying from a specific direction, that is, particulate matter flying from a specific source.

In the above analysis system, the data acquisition unit may acquire information on the particle size of the particulate matter as related data. In this case, the feature extraction unit may extract, from the related data, related data in which the particulate matter has a predetermined particle size range. This makes it possible to extract features related to the source of particulate matter, such as whether the source of particulate matter is near or far.

In the above analysis system, the data acquisition unit may acquire data related to the gas at the location where the particulate matter was sampled as related data. In this case, the feature extraction unit may extract relevant data indicating that the gas is a predetermined type of gas from the relevant data. As a result, it is possible to extract features related to the state of the atmosphere at the location where the data acquisition unit is installed, such as whether or not corrosive gas is present at the location where the data acquisition unit is installed.

In the above analysis system, the data acquisition unit may acquire data related to the wind speed at the location where the particulate matter was collected as related data. In this case, the feature extraction unit may extract relevant data in which the wind speed exceeds a predetermined threshold value or is equal to or less than a predetermined threshold value. This makes it possible to extract features related to the source of particulate matter, such as whether the source of particulate matter is near or far.

In the analysis system described above, the output unit may generate an alert when the feature extraction unit extracts related data that matches a predetermined index for a predetermined feature from among the related data. Thereby, it is possible to visually and/or audibly confirm that the relevant data matching the predetermined index has been acquired.

A server according to another aspect of the present invention is a server that acquires and analyzes relevant data regarding particulate matter. The server includes a feature extraction unit and an output unit. The feature extraction unit extracts features included in the related data by executing a predetermined feature extraction process with the related data as input. The output unit outputs information related to the features extracted by the feature extraction unit.

In the server described above, the feature extraction unit automatically executes a predetermined feature extraction process with input of related data related to particulate matter, and automatically extracts features included in the related data. In this manner, the features included in the related data are automatically extracted by the feature extraction unit, so that the features included in the related data can be efficiently and accurately extracted. Since the features included in the relevant data characterize the particulate matter to be analyzed, efficient and accurate extraction of the features included in the associated data enables efficient analysis of the particulate matter. Further, the output unit outputs information related to the features extracted by the feature extraction unit, so that the user can be shown what features have been extracted from the related data.

An analytical method according to still another aspect of the present invention is an analytical method for particulate matter. The analytical method comprises the following steps.
A step of obtaining relevant data on particulate matter.
A step of extracting features included in the related data by executing a predetermined feature extraction process with the related data as input.
A step of outputting information related to the extracted features.

In the above analysis method, a predetermined feature extraction process is automatically executed with the relevant data related to particulate matter as input, and the features included in the relevant data are automatically extracted. By automatically extracting the features included in the related data in this way, the features included in the related data can be efficiently and accurately extracted. Since the features included in the relevant data characterize the particulate matter to be analyzed, efficient and accurate extraction of the features included in the associated data enables efficient analysis of the particulate matter. Also, by outputting information related to the extracted features, it is possible to show the user what features have been extracted from the related data.

A program according to still another aspect of the present invention is a program for causing a computer to execute the above analysis method.

Features included in data on particulate matter can be extracted efficiently and accurately.

The figure which shows the structure of an analyzer. The figure which shows the structural example of a 1st analyzer. FIG. 4 is a diagram showing a functional block configuration of an analysis server; 4 is a flowchart showing peak search operation; The figure which shows an example when the related data whose peak value exceeded the 1st threshold value is displayed as a list. The figure which shows an example of the graph which shows the time-dependent change of the related data which have the peak value more than a 1st threshold value. 4 is a flowchart showing an average value search operation; FIG. 10 is a diagram showing an example of a list display of related data including peak values equal to or greater than a first threshold and related data including average values equal to or greater than a second threshold; The figure which shows an example of the graph which shows the time-dependent change of the related data containing the average value more than a 2nd threshold value. 4 is a flow chart showing an automatic correlation extraction operation; FIG. 10 is a diagram showing an example of a scatter diagram when multiple correlations are found between two pieces of related data; The figure which shows an example when only a scatter diagram with a large correlation is displayed. The figure which shows an example when the scatter diagram with a large correlation is highlighted. The figure which shows an example of the state which graphically displayed the change with time of correlation. The figure which shows an example of the state which graphically displayed the time-dependent change of the value of two related data with high correlation. The figure which shows the modification of a 1st analyzer.

1. First Embodiment (1) Analysis System An analysis system 100 will be described below. The analysis system 100 is a system for acquiring data on particulate matter (referred to as relevant data RD) and analyzing particulate matter through extraction of features contained in the acquired relevant data RD.

Particulate matter to be analyzed by the analysis system 100 includes, for example, combustion processes in factories, etc., brakes of various transportation devices (automobiles, ships, etc.), tires, internal combustion engines, steam engines, exhaust gas purifiers and motors, and volcanoes. It is micrometer-order particulate matter generated by natural disasters such as volcanic eruptions and mining development.

The configuration of the analysis system 100 will be described using FIG. FIG. 1 is a diagram showing the configuration of an analysis system. The analysis system 100 mainly includes a data acquisition section 1 and an analysis server 3 .

The data acquisition unit 1 is located at or near a source of particulate matter, and acquires various data related to particulate matter generated from the source as related data RD. The data acquisition unit 1 is arranged, for example, in or near a factory that may generate particulate matter, or along or near a road with heavy traffic (main road, expressway, etc.). In addition, the data acquisition unit 1 may be mounted on a moving body (for example, an automobile) so as to be movable.

The analysis server 3 is a computer system composed of a CPU, storage devices (RAM, ROM, SSD, HDD, etc.), various interfaces, and the like. The analysis server 3 is connected to the data acquisition unit 1 and collects and stores the related data RD acquired by the data acquisition unit 1 . The analysis server 3 also extracts features included in the related data RD by executing a predetermined feature extraction process with the related data RD collected from the data acquisition unit 1 as input.

The analysis server 3 is connected to the client terminal 5. The client terminal 5 is an information terminal such as a personal computer, tablet terminal, or smart phone used by the user. The user can use the client terminal 5 to access the analysis server 3 and browse the related data RD stored in the analysis server 3 and the analysis results of the related data RD output from the analysis server 3 .

FIG. 1 shows an example of the analysis system 100 in which one data acquisition unit 1, one analysis server 3, and one client terminal 5 are provided. , and the number of client terminals 5 are arbitrary.

(2) Data Acquisition Unit (2-1) Overall Configuration A specific configuration of the data acquisition unit 1 provided in the analysis system 100 will be described below with reference to FIG. The data acquisition unit 1 has a first analysis device 11 , a second analysis device 13 , a third analysis device 15 and a data collection device 17 .

The first analyzer 11 collects particulate matter present at the location of the data acquisition unit 1 at predetermined time intervals (for example, every hour), and the mass concentration of the collected particulate matter and the particulate matter and information about the elements contained in are acquired as related data RD. Here, "information about an element contained in particulate matter" means an element contained in particulate matter and the content of the element. This information may also include the composition ratio (element ratio) of the elements contained in the particulate matter. By having the first analysis device 11 in the data acquisition unit 1, in the analysis system 100, particles analysis can be performed on

Particulate matter may vary depending on the source, etc., but for example, chromium (Cr), copper (Cu), iron (Fe), aluminum (Al), silicon (Si), lead (Pb), zinc (Zn) , Mercury (Hg), Vanadium (V), Calcium (Ca), Potassium (K), Arsenic (As), Selenium (Se), Sulfur (S), elements that cause flame reactions (e.g., Strontium (Sr)) and so on. The first analysis device 11 can acquire at least information on these elements and other elements (elements included, content of the elements).

A specific configuration example of the first analysis device 11 capable of acquiring the mass concentration of particulate matter and information on the elements contained in the particulate matter will be described later.

The second analysis device 13 is an anemoscope that acquires the wind direction and wind speed at the location of the data acquisition unit 1 (where the particulate matter is sampled) as related data RD at predetermined time intervals (for example, hourly intervals). be. Particulate matter tends to be carried on the wind from its source. Therefore, by having the second analysis device 13 in the data acquisition unit 1, it is possible to specify from which direction the particulate matter collected by the first analysis device 11 flew.

The third analysis device 15 is a device that analyzes the gas contained in the atmosphere around the location of the data acquisition unit 1 (where the particulate matter is sampled) at predetermined time intervals (for example, hourly intervals). Specifically, the third analysis device 15 identifies the gas contained in the atmosphere around the position where the data acquisition unit 1 is arranged and/or acquires the concentration of the gas as the related data RD. Gases that can be analyzed by the third analyzer 15 include, for example, hydrocarbons, carbon monoxide (CO), carbon dioxide (CO ₂ ), nitrogen oxides (NOx), ozone (O ₃ ), sulfur oxides (SOx). , gases such as hydrogen sulfide (H ₂ S), and/or Volatile Organic Compounds (VOCs) such as acetone, ethanol, toluene, benzene, freon.

The data collection device 17 is a data logger that acquires the related data RD acquired by the first analysis device 11 to the third analysis device 15 and transmits it to the analysis server 3. The data collection device 17 considers the time lag between each analysis device and the data collection device 17 to determine the timing of acquiring the related data RD from each analysis device. As a result, the data collection device 17 does not miss the related data RD obtained by each analysis device. In addition, the data collection device 17 associates the relevant data RD acquired from each analysis device with the time (time stamp) at which the relevant relevant data RD was acquired. When transmitting the related data RD to the analysis server 3 , the data collection device 17 also transmits the time stamp associated with the related data RD to the analysis server 3 .

By having the above configuration, the data acquisition unit 1 can obtain the mass concentration of particulate matter, information about elements contained in the particulate matter, wind direction, wind speed, and information about gas contained in the surrounding atmosphere. It can be acquired as relevant data RD related to the substance and provided to the analysis server 3 . In addition, since the related data RD acquired by each analysis device is acquired at predetermined time intervals, the related data RD is data whose results change over time. That is, the related data RD is data in which the results (values, etc.) obtained at each time are arranged in chronological order.

Note that the data acquisition unit 1 may have other measurement devices in addition to the first analysis device 11 to the third analysis device 15 described above. For example, it may have a positioning device such as GPS that acquires the arrangement position of the data acquisition unit 1 .

(2-2) Configuration of First Analysis Apparatus A specific configuration example of the first analysis apparatus 11 will be described with reference to FIG. FIG. 2 is a diagram showing a configuration example of the first analysis device. The first analysis device 11 has a collection filter 111 , a collection section 113 , a first analysis section 115 , a second analysis section 117 and a control section 119 .

The collection filter 111 is formed, for example, of a porous fluororesin-based material having pores capable of collecting particulate matter on a reinforcing layer formed of nonwoven fabric of a polymeric material (such as polyethylene). It is a tape-shaped member formed by stacking collection layers (sometimes called collection regions). As the collection filter 111, for example, other filters such as a one-layer glass filter and a one-layer fluororesin material filter can be used.

In this embodiment, the collection filter 111 is wound in the length direction (the direction indicated by the thick arrow in FIG. 2) by rotating the take-up reel 111b. can move.

The collection part 113 is provided so as to correspond to the first position P1 in the length direction of the collection filter 111 . For example, the collection unit 113 blows the air sucked by the suction force of the suction port 135 connected to the suction pump 131 from the discharge port 133 to the collection region existing at the first position P1 of the collection filter 111. , causing the collection area to collect particulate matter contained in the atmosphere.

The first analysis unit 115 measures the amount of particulate matter collected by the collection filter 111 . Specifically, the first analysis unit 115 has a β-ray source 51 and a β-ray detector 53 . The β-ray source 51 is provided at the outlet 133 of the collection unit 113 and emits β-rays to the collection region of the collection filter 111 arranged at the first position P1. The β-ray source 51 is, for example, a β-ray source using carbon-14 (14C).

The β-ray detector 53 is provided to face the β-ray source 51 at the suction port 135 of the collection unit 113, and measures the intensity of β-rays transmitted through the particulate matter collected in the collection region at the first position P1. to measure. The β-ray detector 53 is, for example, a photomultiplier tube with a scintillator. The trapped amount (mass concentration) of particulate matter is calculated based on the intensity of β rays measured by the β ray detector 53 .

The second analysis unit 117 is provided to correspond to a second position P2 in the length direction of the collection filter 111, and measures data regarding fluorescent X-rays generated from particulate matter present at the second position P2. Specifically, the second analysis unit 117 has an X-ray source 71 and a detector 73 .

The X-ray source 71 irradiates the particulate matter present at the second position P2 with X-rays. The X-ray source 71 is, for example, a device that irradiates a metal such as palladium with an electron beam to generate X-rays. The detector 73 detects fluorescent X-rays generated from particulate matter. The detector 73 is, for example, a silicon semiconductor detector or a silicon drift detector.

The control unit 119 acquires data for calculating the mass concentration of particulate matter using the first analysis unit 115 provided at the first position P1. Further, the control unit 119 controls the take-up reel 111b to move the collection filter 111 in order to acquire the elemental analysis result using the second analysis unit 117 provided at the second position P2. Specifically, every time the collection of particulate matter by collection unit 113 is completed and the measurement of the amount of area) is moved from the first position P1 where the first analysis unit 115 is provided toward the second position P2 where the second analysis unit 117 is provided.

After the collection region reaches the second position P2, the control unit 119 irradiates X-rays from the X-ray source 71 toward the second position P2. The fluorescent X-ray obtained is acquired as data for elemental analysis.

The control unit 119 calculates the mass concentration of particulate matter as the related data RD based on the intensity of the β rays measured by the first analysis unit 115 . In addition, the control unit 119 acquires the fluorescent X-ray data (for example, the fluorescent X-ray spectrum) obtained by the second analysis unit 117, and based on the fluorescent X-ray data, the elements contained in the particulate matter and the The content is calculated as related data RD. The calculated related data RD is transmitted to the data collection device 17 .

(3) Functional Block Configuration of Analysis Server The functional block configuration of the analysis server 3 will be described below with reference to FIG. FIG. 3 is a diagram showing the functional block configuration of the analysis server. The functional blocks of the analysis server 3 described below may be stored in the storage device of the analysis server 3 and implemented by programs executable by the analysis server 3 . Also, some of the functional blocks may be implemented by hardware that constitutes the analysis server 3 . The analysis server 3 has a storage unit 31, a data reception unit 33, a feature extraction unit 35, and an output unit 37 as functional blocks.

The storage unit 31 stores various data, programs, setting values, etc. used by the analysis server 3 . Specifically, the storage unit 31 stores the related data RD acquired by the data acquisition unit 1 . By having the storage unit 31, the analysis server 3 functions as a database of the related data RD.

The data receiving section 33 receives the related data RD from the data collecting device 17 of the data acquiring section 1 and stores it in the storage section 31 . The data receiving unit 33 receives the related data RD accumulated in the data collection device 17 at a predetermined timing. Note that the data receiving unit 33 may receive the related data RD via the data collection device 17 immediately after each analysis device outputs the related data RD (that is, the data collection device 17 accumulates the related data RD). not), the related data RD may be received at the timing when the related data RD is accumulated in the data collection device 17 to some extent.

The feature extraction unit 35 extracts features included in the related data RD by executing a predetermined feature extraction process with the related data RD stored in the storage unit 31 as input. Specifically, the feature extraction unit 35 statistically analyzes the related data RD as predetermined feature extraction processing, and extracts features included in the related data RD. More specifically, each value included in the related data RD whose values change over time is screened to extract features included in the related data RD. This feature extraction processing is called a data screening function.

Specifically, as a data screening function, the feature extraction unit 35 extracts the feature that the peak value (instantaneous value) included in the related data RD exceeds the first threshold, and the peak value exceeds the first threshold. The output unit 37 is notified of the time when the exceeding peak value was measured and information (for example, data name) identifying the related data RD whose peak value exceeds the first threshold. This feature extraction processing is called a peak search function.

In addition, as another data screening function, the feature extraction unit 35 extracts the feature that the average value of the values included in the related data RD exceeds the second threshold, and the average value exceeds the second threshold. The output unit 37 is notified of the time and information identifying the related data RD whose average value exceeds the second threshold. This feature extraction processing is called an average value search function. Note that the above average value is, for example, an average value of a plurality of values included within a predetermined time range before and after a specific time included in the related data RD. In this case, "the time when the average value exceeds the second threshold" is the reference time when the average value is calculated. Note that in this data screening function, the median value of the values included in the related data RD may be used instead of the average value of the values included in the related data RD.

The above threshold used in the data screening function may be a predetermined fixed value, or may be a value that can be changed according to the type of related data RD. If the threshold can be changed, for example, the standard deviation of the related data RD to be subjected to data screening can be calculated, and an integer multiple (eg, 1 or 2 times) of the standard deviation can be set as the threshold.

Furthermore, the feature extraction unit 35 automatically calculates the correlation of the plurality of related data RD stored in the storage unit 31, and information related to the plurality of related data RD whose calculated correlation is higher than or equal to the third threshold value. is extracted as a feature included in the related data RD and notified to the output unit 37 . This feature extraction processing is called an automatic correlation extraction function.

The feature extraction unit 35 may execute the data screening function and automatic correlation extraction function described above in response to a command from the client terminal 5, or may automatically execute them at a predetermined timing. Further, when executing the automatic correlation extraction function, a plurality of related data RD to be calculated for correlation may be selectable by the user using the client terminal 5, or the feature extraction unit 35 may A plurality of related data RD for which correlation is to be calculated may be automatically extracted.

The output unit 37 outputs information related to the features extracted by the feature extraction unit 35 to the client terminal 5 . When the data screening function is executed in the feature extraction unit 35, the output unit 37 outputs the time at which the peak value equal to or greater than the first threshold occurs and/or the time at which the average value or median value becomes equal to or greater than the second threshold. are displayed on the client terminal 5 as a list. Each listed time is provided with a link related to related data RD including the above peak value and/or average value (median value). When this link is selected, the output unit 37 displays the data value of the corresponding related data RD and/or the graph of the related data RD on the client terminal 5 .

On the other hand, when the feature extraction unit 35 executes the automatic correlation extraction function, the output unit 37 outputs to the client terminal 5 information related to the plurality of related data RD for which the calculated correlation is higher than the third threshold. do. For example, the output unit 37 displays on the client terminal 5 a scatter diagram of a plurality of related data RD whose correlation is equal to or greater than the third threshold.

Alternatively, the output unit 37 outputs a scatter diagram for a plurality of combinations of the related data RD to the client terminal 5, and among the displayed scatter diagrams, for the combinations of the related data RD whose correlation is equal to or greater than the third threshold, Scatterplots may be highlighted. Specifically, for example, the related data RD whose correlation is smaller than the third threshold is surrounded by a frame of a predetermined color (for example, red) for the combination of the related data RD whose correlation is equal to or greater than the third threshold. It can be highlighted by a method such as displaying the scatterplot for the combination of

In addition, the output unit 37 can display a scatter diagram for a plurality of related data RD specified by the user using the client terminal 5. For example, the output unit 37 can generate and display a scatter diagram of the content of a combination of two elements among the plurality of elements selected by the user using the client terminal 5 . For example, when three elements are selected by the user, for each combination of two elements (three combinations) further selected from the three elements, the client displays a scatter diagram of the contents of the two elements. It can be displayed on the terminal 5. Thereby, the user can visually confirm the combination of related data RD with high correlation based on the displayed scatter diagrams.

(4) Analysis Operation of Related Data in Analysis System (4-1) Peak Search Operation The analysis operation of the related data RD in the analysis system 100 will be described below. First, the operation of the peak search function (peak search operation) as one of the data screening functions will be described with reference to FIG. FIG. 4 is a flow chart showing the peak search operation. The peak search operation shown in the flow chart of FIG. 4 is executed by the analysis server 3 .

Note that the peak search operation may be performed on all related data RD (including values that change over time) stored in the storage unit 31, or may be performed on a plurality of related data RD specified in advance. may be performed for For example, other related data RD acquired during a period in which a specific wind direction is indicated in the related data RD related to the wind direction acquired by the second analysis device 13 (for example, related data RD related to the content of a specific element ). A peak search of the relevant data RD obtained at a particular wind direction can, for example, monitor the state of particulate matter from a particular source.

For example, when the relevant data RD representing the content of elements contained in particulate matter is targeted for peak search, when a specific wind direction is observed, the peak of the content of elements contained in particulate matter Relevant data RD whose value exceeds the first threshold can be extracted. As a result, it is possible to extract related data indicating anomalies regarding particulate matter flying from a specific direction, that is, particulate matter flying from a specific source.

First, in step S11, the feature extraction unit 35 searches for a peak value included in the related data RD that is the target of the peak search. For example, if the related data RD is data on the content of elements contained in the particulate matter, the peak value of the content is searched. For example, the feature extraction unit 35 scans the values included in the related data RD one by one, and if the currently scanned value is larger than the values before and after it, the currently scanned value is determined as the sub-peak value. Then, the maximum value among all the sub-peak values included in the related data RD is determined as the peak value.

Next, in step S12, the feature extraction unit 35 determines whether or not the peak value found by executing step S11 is greater than or equal to the first threshold. For example, it is determined whether or not the peak value of the content of the element is equal to or greater than the first threshold. The first threshold can be appropriately determined depending on the type of element and the like. If the peak value is not equal to or greater than the first threshold ("No" in step S12), the feature extraction unit 35 determines that the relevant data RD currently being searched for peaks does not include a peak value equal to or greater than the first threshold. to end the peak search operation. If there is another related data RD to be peak searched, the peak search operation is executed for the related data RD.

On the other hand, if the peak value found by executing step S11 is greater than or equal to the first threshold value ("Yes" in step S12), the feature extraction unit 35 determines the time at which the peak value occurred in the related data RD in step S13. and information identifying the related data RD including this peak value are notified to the output unit 37 . For example, when the peak value of the content of the element is equal to or greater than the first threshold value, the time when the peak value of the content of the element occurred and the identification information of the related data RD including this peak value are sent to the output unit 37. Output.

Upon receiving the time and the identification information of the related data RD, the output unit 37 causes the client terminal 5 to display the time when the peak value equal to or greater than the first threshold occurs in step S14. When the client terminal 5 accesses the analysis server 3 using a web browser or the like, the output unit 37 generates an HTML file for displaying a table listing the times as shown in FIG. 5, for example. In the list display shown in FIG. 5, the time at which the peak value occurred is displayed in the "Date" column, and a display (Peak) indicating that the peak value is greater than or equal to the first threshold value is displayed in the "Type" column. be. FIG. 5 is a diagram showing an example of displaying a list of related data whose peak value exceeds the first threshold.

Further, based on the information identifying the related data RD received from the feature extraction unit 35 in step S13, the output unit 37 causes the client terminal 5 to display a graph showing changes over time in the values in the related data RD. Generate a link for In the time list display, the output unit 37 attaches the generated link to the display portion of the time at which the peak value included in the related data RD occurs. In the list display shown in FIG. 5, an underline attached to each time indicates that the above link is attached.

　On the client terminal 5, by selecting (for example, clicking) the time with the above link, for example, a graph as shown in FIG. 6 is displayed on the client terminal 5. The graph shown in FIG. 6 is a graph showing changes in the content of the specific element A over time. FIG. 6 is a diagram showing an example of a graph showing temporal changes in related data having peak values equal to or greater than the first threshold.

By the above peak search operation, the analysis server 3 can automatically, efficiently and accurately extract the feature that the related data RD includes a large peak value. For example, when a peak value that exceeds the first threshold indicates the occurrence of an abnormality, the analysis server 3 retrieves the related data RD that has a peak value that exceeds the first threshold and includes an abnormality by the above-described peak search operation. It can be automatically extracted and presented to the user in which related data RD anomaly occurred. Further, by graphically displaying the related data RD including a large peak value on the client terminal 5, the user can visually confirm temporal changes in the related data RD including anomalies, for example.

(4-2) Average Search Operation Next, the operation of the average search function (average search operation) as another data screening function will be described with reference to FIG. FIG. 7 is a flow chart showing the average value search operation. The average value search operation shown in the flowchart of FIG. 7 is executed by the analysis server 3 .

Note that the average value search operation may also be executed for all related data RD stored in the storage unit 31, similarly to the peak search operation, or may be executed for a plurality of previously specified related data RD. may be

For example, when the related data RD representing the content of the element contained in the particulate matter is targeted for the average value search operation, when a specific wind direction is observed, the content of the element contained in the particulate matter Relevant data RD whose average value or median value of exceeds the second threshold can be extracted. As a result, it is possible to extract related data indicating anomalies regarding particulate matter flying from a specific direction, that is, particulate matter flying from a specific source.

First, in step S21, the feature extraction unit 35 calculates the average value of the related data RD that is the target of average value search. For example, when the related data RD is data on the content of elements contained in the particulate matter, the average value of the content is calculated. For example, the feature extraction unit 35 uses one value included in the related data RD as a reference, and calculates the average value of the reference value and a predetermined number of values before and after the reference value. The feature extraction unit 35 calculates the average value while changing the reference value from the leading value to the trailing value of the related data RD. As a result, for example, when the related data RD contains N values that change over time, when calculating the average value of the three values around the reference value, N−2 average values is calculated. That is, by executing step S21, a plurality of average values are calculated for one piece of related data RD.

Note that the above average value can also be calculated by a method of calculating a moving average of the related data RD. Also, a median value may be calculated instead of the average value.

Next, in step S22, the feature extraction unit 35 determines whether or not any of the multiple average values calculated by executing step S21 is equal to or greater than the second threshold. For example, it is determined whether or not the average value of the element contents is equal to or greater than the second threshold. The second threshold can be appropriately determined depending on the type of element and the like. If none of the plurality of average values is greater than or equal to the second threshold ("No" in step S22), the feature extraction unit 35 determines that the related data RD does not include an average value greater than or equal to the second threshold, End the mean value search operation. If there is another related data RD to be searched for the mean value, the mean value search operation is executed for the related data RD.

On the other hand, if any of the plurality of average values calculated in step S21 is equal to or greater than the second threshold ("Yes" in step S22), the feature extraction unit 35 determines in step S23 that the average value is equal to or greater than the second threshold. The output unit 37 is notified of the time and information identifying the related data RD for which the average value was calculated. For example, when the average value of the element content is equal to or greater than the second threshold, the time when the average value of the element content becomes equal to or greater than the second threshold, the identification information of the related data RD including the average value, is output to the output unit 37 .

Upon receiving the time and the identification information of the related data RD, the output unit 37 causes the client terminal 5 to display the time when the average value becomes equal to or greater than the second threshold in step S24. As shown in FIG. 8, this time is displayed together with the time of occurrence of the peak value in a table that lists the times of occurrence of peak values equal to or greater than the first threshold. In the row (the third row in FIG. 8) corresponding to the time when the average value in the "Type" column is equal to or greater than the second threshold, an indication (Average ) is done. FIG. 8 is a diagram showing an example of a list display of related data including peak values equal to or greater than the first threshold and related data including average values equal to or greater than the second threshold.

Further, based on the information identifying the related data RD received from the feature extraction unit 35 in step S23, the output unit 37 causes the client terminal 5 to display a graph showing changes over time in the values in the related data RD. A link is generated for this purpose, and the link is attached to the display of the time when the average value becomes equal to or greater than the second threshold.

By selecting (for example, clicking) the linked time on the client terminal 5, for example, a graph as shown in FIG. 9 is displayed on the client terminal 5. The graph shown in FIG. 9 is a graph showing changes over time in the content of the specific element B. FIG. FIG. 9 is a diagram showing an example of a graph showing temporal changes in related data including average values equal to or greater than the second threshold. Also, by selecting the above link, the values included in the corresponding related data RD may be displayed.

By means of the above average value search operation, the analysis server 3 can automatically and efficiently and accurately extract the feature that the related data RD includes a large average value or median value. For example, if the average value or median value exceeding the second threshold indicates the occurrence of an abnormality, the analysis server 3 determines whether the average value or median value exceeds the second threshold by the above-described average value search operation. It is possible to automatically extract the related data RD containing anomalies and present to the user in which related data RD the anomaly occurred. In addition, by displaying the related data RD including a large average value or median value in a graph on the client terminal 5, the user can visually confirm temporal changes in the related data RD including anomalies, for example.

(4-3) Auto-correlation Extraction Operation Further, the operation of the auto-correlation extraction function (auto-correlation extraction operation) will be described with reference to FIG. FIG. 10 is a flow chart showing the automatic correlation extraction operation. The automatic correlation extraction operation shown in the flowchart of FIG. 10 is executed by the analysis server 3 .

First, in step S31, the feature extraction unit 35 selects a plurality of related data RD (two related data RD) for which correlation is to be calculated. For example, before step S31 is executed, the user uses the client terminal 5 to select a plurality of related data RD (a plurality of data items for which correlation is to be investigated) for which the correlation is to be investigated. After that, in step S31, the feature extraction unit 35, which has received the selection result from the user, selects two related data RD for which the correlation is to be calculated from the plurality of related data RD selected by the user.

Alternatively, the feature extraction unit 35 may automatically select two pieces of related data RD whose correlation is to be investigated from all the related data RD stored in the storage unit 31 . Furthermore, a plurality of related data RD when a specific wind direction is observed in the related data RD related to wind direction may be used as targets for calculating the correlation.

Next, in step S32, the feature extraction unit 35 calculates the correlation between the two pieces of related data RD selected by executing step S31. Specifically, the feature extraction unit 35 uses the value contained in one of the two relational data RD as the first element (x) and the value contained in the other as the second element (y) to A coefficient of determination (square of correlation coefficient) representing the degree of correlation of data RD is calculated.

Since the data acquisition unit 1 acquires the value of the related data RD at predetermined time intervals, the related data RD contains a plurality of values that change over time. When the relevant data RD contains many values, that is, when the relevant data RD is obtained as a result of long-term measurements, as shown in FIG. A correlation may be seen. In the example shown in FIG. 11, two large correlations are observed in the elliptical area. Note that the coefficient of determination calculated for the value of the related data RD containing multiple large correlations tends to be small. In other words, the coefficient of determination calculated from the long-term related data RD may not show that there are multiple large correlations. FIG. 11 is a diagram showing an example of a scatter diagram when multiple correlations are found between two pieces of related data.

The fact that multiple large correlations are seen between multiple related data RDs indicates, for example, that the characteristics of particulate matter generated by a specific source changed during the acquisition period of related data RDs. In analyzing particulate matter, it is important to know when the characteristics of the particulate matter changed during the acquisition period of the related data RD.

Therefore, the feature extraction unit 35 not only calculates the coefficient of determination using the values of the entire period for which the related data RD is acquired, but also divides the period into a plurality of sub-periods and uses the values within the sub-periods to determine the coefficient of determination. Calculate the coefficient. For example, if the relevant data RD for which the correlation is to be examined is data acquired over a year, the coefficient of determination is calculated using the values for the first month included in the relevant data RD, and the coefficient of determination for the next month is Calculating the coefficient of determination using the values is repeatedly executed, and a total of 12 coefficients of determination can be calculated for the related data RD acquired over one year.

As described above, for the related data RD obtained over a long period of time, by calculating the coefficient of determination for each of a plurality of small periods included in the long period of time, information on events occurring in the relevant small period can be obtained. be done. For example, when the coefficient of determination in a specific short period is small (correlation between related data RD is small), while the coefficient of determination in other short periods is large (correlation between related data RD is large), these Between the two sub-periods, it can be assumed that special events occurred with respect to particulate matter generation.

The above short period may be variable. For example, the coefficient of determination can be calculated using the values for the first month included in the related data RD, and then the coefficient of determination can be calculated using the values for the first two months included in the related data RD. In this way, by allowing the above period to be set flexibly, it is possible to specify in which period the correlation of the related data changed. For example, the related data RD for the first month did not show a large correlation (the coefficient of determination was small), while the related data RD for the first two months showed a relatively large correlation (the coefficient of determination was relatively large), it can be specified that there was a change in the correlation of the relevant data RD during at least the latter half of the two months. That is, it can be identified that the properties of the particulate matter have changed at least during the latter half of the month.

Alternatively, the above small period may be divided into smaller sub-intervals and the coefficient of determination calculated for each sub-interval. For example, the related data RD obtained over a period of one year can be divided into sub-periods of one month, and the sub-periods of one month can be divided into sub-sections of one week. In this case, it is possible to obtain more detailed information about the event that occurred in the small section. Furthermore, the sub-intervals described above may be variable as well as the sub-periods.

As described above, when calculating the correlation (coefficient of determination) between two pieces of related data RD, by dividing the related data RD into short periods and calculating the coefficient of determination, the timing at which the characteristic change of the particulate matter occurs can be analyzed in more detail, and the timing of occurrence of specific events at the source of particulate matter can be analyzed in detail.

After calculating the correlation between the two pieces of related data RD, the feature extraction unit 35 determines in step S33 whether or not the correlation (coefficient of determination) calculated in step S32 is greater than or equal to the third threshold. Since a plurality of coefficients of determination are calculated in step S32, the feature extraction unit 35 determines whether or not there is a coefficient of determination greater than or equal to the third threshold among the plurality of coefficients of determination. Note that the third threshold for evaluating the magnitude of correlation can be appropriately determined depending on how large the coefficient of determination is to determine that there is correlation.

If none of the calculated determination coefficients exceeds the third threshold ("No" in step S33), the feature extraction unit 35 determines that the correlation between the currently selected two pieces of related data RD is small, and proceeds to step S35. move on.

On the other hand, if any of the calculated coefficients of determination is greater than or equal to the third threshold ("Yes" in step S33), the feature extraction unit 35 determines that the correlation between the currently selected two pieces of related data RD is large. In S34, the output unit 37 is notified of information identifying the currently selected two related data RD. At this time, the feature extraction unit 35 notifies the output unit 37 of information for identifying the two pieces of related data RD and information on the period during which the coefficient of determination is equal to or greater than the third threshold.

After executing the above steps S32 to S34 using the currently selected combination of the two related data RD, the feature extraction unit 35 extracts the data specified by the user or stored in the storage unit 31 in step S35. It is determined whether correlations (coefficients of determination) have been calculated for all combinations of two pieces of related data RD included in all related data RD. If correlations have not been calculated for all combinations ("No" in step S35), the feature extraction unit 35 returns to step S31, selects another combination of the two pieces of related data RD, and Steps S32 to S34 are executed for .

On the other hand, if correlations are calculated for all combinations ("Yes" in step S35), the output unit 37 identifies, in step S35, the two related data RD notified by the feature extraction unit 35 in step S34. and the period during which the correlation (coefficient of determination) of the two pieces of related data RD was equal to or greater than the third threshold, a predetermined graph is displayed for a plurality of pieces of related data RD with high correlation.

The output unit 37 generates a scatter diagram of the values of the two highly correlated related data RD within the notified period and outputs it to the client terminal 5 .

When outputting a scatter diagram of two pieces of related data RD having a correlation (coefficient of determination) greater than or equal to the third threshold, as shown in FIG. may be displayed on the client terminal 5, or, as shown in FIG. Scatter plots with three or more thresholds may be highlighted by surrounding them with a frame. FIG. 12 is a diagram showing an example when only scatter diagrams with high correlation are displayed. FIG. 13 is a diagram showing an example when a scatter diagram with a large correlation is highlighted.

Further, the output unit 37 outputs the information for identifying the two related data RD notified from the feature extraction unit 35 in step S34 and the period during which the correlation between the two related data RD is equal to or greater than the third threshold value. , the temporal change of a plurality of highly correlated related data RD can be graphically displayed. Whether to display a scatterplot, a graph of changes over time, or both can be optionally determined.

For example, as a result of executing the automatic correlation extraction function, between time T1 and time T2, there is a high correlation between the content of element C and the content of element D (the coefficient of determination is the third threshold (TH3) or more ), when there is a high correlation between the content of element E and the content of element F, and the content of element G and the content of element H, when there is a high correlation between these three elements A graph such as that shown in FIG. 14 can be displayed for each of the combinations. In FIG. 14, the time change graph of the correlation between the content of the element C and the content of the element D is represented by a solid line, and the time change graph of the correlation between the content of the element E and the content of the element F is represented by a solid line. is represented by a dashed line, and a time change graph of the correlation between the content of the element G and the content of the element H is represented by a dashed-dotted line. FIG. 14 is a diagram showing an example of a state in which changes in correlation over time are displayed graphically.

By the automatic correlation extraction operation described above, the analysis server 3 can automatically efficiently and accurately extract the feature that a plurality of related data RD has a large correlation. For example, when there is a large correlation between the contents of multiple elements, it is assumed that particulate matter with the same contents (element ratio) of the multiple elements is measured during the period when the correlation is observed. I can guess.

Also, by displaying a scatter diagram of two pieces of related data RD having a large correlation, it is possible to visually confirm the magnitude of the correlation between the two pieces of related data RD. In addition, by displaying a plurality of scatter diagrams regardless of the magnitude of the correlation and highlighting the scatter diagram with a large correlation among the displayed plurality of scatter diagrams, It is possible to visually confirm which data RD is.

Furthermore, by graphically displaying the temporal change in the correlation of the plurality of related data RD with high correlation, it is possible to visually confirm how the correlation of the plurality of related data RD changes over time.

For example, if the related data RD is data representing the content of elements contained in particulate matter, it is possible to visually confirm how the correlation of the content of a plurality of elements changes over time. . A high correlation between the contents of a plurality of elements means that the particulate matter contains a plurality of highly correlated elements at a constant composition ratio. The types and composition ratios of the elements contained in the particulate matter greatly depend on the properties of the particulate matter, such as the generation source and/or the generation conditions of the particulate matter.

Therefore, by graphically displaying the changes in correlation over time for the contents of a plurality of elements with high correlations, for example, the user can visually confirm the change in correlation over time and change the properties of the particulate matter, i.e. , at what timing and from which source the particulate matter is flying, and/or at what timing and how the generation conditions of the particulate matter have changed.

When a scatter diagram of two pieces of related data RD having a large correlation is displayed and this scatter diagram is designated, the output unit 37 outputs the two pieces of related data RD that generated the scatter diagram, as shown in FIG. Changes in each value over time may be displayed graphically. FIG. 15 is a diagram showing an example of a state in which temporal changes in values of two related data having a high correlation are displayed graphically. In FIG. 15, by specifying a scatter diagram of the content of element C and the content of element D, which have a large correlation as shown in FIG. It is an example in which the temporal change of the content is displayed side by side in a graph.

With only a scatter diagram like FIG. 12, it is unclear in which period the correlation between the two related data RD increases. It can be estimated whether the correlation is large. In the example shown in FIG. 14, between time T1 and time T2, the tendency of increase/decrease in the content of element C and the tendency of increase/decrease in the content of element D are the same. It can be estimated that the correlation between the content and the content of element D is increasing.

When the output unit 37 generates a scatter diagram of two pieces of related data RD as shown in FIG. As a result, even if two pieces of related data RD include a plurality of periods with high correlation, it is possible to identify a plurality of correlations between two pieces of highly correlated data RD by the color of the dots in the scatter diagram. For example, when two related data RD are data related to particulate matter generated from the same source, it is possible to visualize that the properties of the particulate matter (for example, the composition ratio of elements, etc.) change with time. can be confirmed.

(5) Modification of First Analysis Apparatus The first analysis apparatus 11 that acquires related data RD on particulate matter is not limited to the configuration shown in FIG. Specifically, as shown in FIG. 16, the first analyzer 11 includes a light source 51′ and a scattered light detector instead of the β-ray source 51 and the β-ray detector 53 for measuring the trapped amount of particulate matter. A portion 53' may be provided. FIG. 16 is a diagram showing a modification of the first analysis device.

The light source 51 ′ emits a laser beam L toward the inside of the discharge port 133 . The scattered light detection unit 53 ′ detects scattered light generated by the laser light L being scattered by particulate matter while passing through the outlet 133 . The scattered light detector 53' is, for example, a photodetector such as a photodiode. Accordingly, the control unit 119 can acquire information about particulate matter contained in the atmosphere at the position where the first analysis device 11 is arranged, based on the intensity of the scattered light detected by the scattered light detection unit 53'. Specifically, based on the intensity of the scattered light, data on the particle size of particulate matter contained in the atmosphere (for example, particle size distribution of particulate matter) can be acquired as the related data RD. Further, based on the intensity of the scattered light, it is possible to acquire data on the content of particulate matter, such as the number of particulate matter contained in the atmosphere, as related data RD.

(6) Application Examples of Analysis System An application example of the analysis system 100 will be described below. The analysis system 100 described above can be applied, for example, to environmental management of storage locations for products and the like. Specifically, for example, in order to analyze whether corrosive gas, corrosive particulate matter, etc. that affects the product, etc. exists in the storage location of the product, etc., the above analysis System 100 can be used.

More specifically, the first analysis device 11 to the third analysis device 15 are installed in a storage place for products, etc., and related data RD related to elements contained in particulate matter existing in the space of the storage place, Relevant data RD about the particle size of the particulate matter, relevant data RD about the gas existing in the space of the storage location, related data RD about the wind direction of the storage location, and related data RD about the wind speed of the storage location are obtained. and based on these related data RD environmental management of the storage location can be carried out.

As a method of acquiring the related data RD related to the particle size of the particulate matter, the first analyzer 11 as shown in FIG. A method of obtaining related data RD can be used.

In addition, it is also possible to obtain related data RD on particle size using a plurality of first analysis devices 11 as shown in FIG. Specifically, each of the plurality of first analysis devices 11 is provided with a collection unit 113 that collects particulate matter with different particle size ranges, and the related data RD ( By classifying the related data RD related to the element contained in the particulate matter and the related data RD related to the content (mass concentration) of the particulate matter according to which first analysis device 11 the related data RD is acquired , the relevant data RD sorted by the particle size of the particulate matter (ie the relevant data RD containing data relating to the particle size of the particulate matter) can be obtained.

Further, in addition to the first analysis device 11 to the third analysis device 15, another analysis device capable of measuring the particle size of particulate matter is connected to the data collection device 17, and this analysis device Relevant data RD regarding the diameter may be measured.

In this case, the feature extraction unit 35 of the analysis server 3 executes the feature extraction process, and out of the related data RD stored in the storage unit 31, the related data indicating that the particulate matter has a predetermined particle size range RD can be extracted. For example, when the feature extraction unit 35 extracts the related data RD indicating that the particle size range is specified, the output unit 37 of the analysis server 3 stores particles having the specified particle size range in the space of the storage location. It can output information about the presence of substances such as

For example, when the feature extraction unit 35 extracts the related data RD indicating that the particulate matter has a large particle size equal to or larger than a predetermined threshold value, the output unit 37 stores the product in the storage space. It can output information that it contains particulate matter generated in a nearby location. On the other hand, when the feature extraction unit 35 extracts the related data RD indicating that the particulate matter has a small particle size equal to or smaller than the predetermined threshold value, the output unit 37 outputs can output information that it contains particulate matter generated in a relatively distant place.

The feature extracting unit 35 extracts information about the particle size of the particulate matter, information about the direction of the wind, information about the wind speed, information about the element contained in the particulate matter, and the like, and extracts the information about the specific element from the specific element. and can extract relevant data RD for particulate matter having a particular particle size range. This provides more detailed information about the source of particulate matter. When the feature extraction unit 35 extracts the related data RD related to the particulate matter generated from the specific source, the output unit 37 outputs information indicating that the particulate matter generated from the specific source exists in the storage space. can be output.

As described above, by extracting the relevant data RD based on the information on the particle size of the particulate matter, the information on the wind direction, the information on the wind speed, and the information on the elements contained in the particulate matter, the relevant data RD is It is possible to determine whether the related data RD is related to particulate matter artificially generated by the operation of a plant or the like, or related data RD is related to naturally occurring particulate matter such as soil.

For example, when the feature extraction unit 35 extracts related data RD related to particulate matter that comes from a specific direction, contains a specific element that is likely to be generated from a predetermined factory, and has a large particle size, the output The unit 37 can output information indicating that the storage location contains particulate matter originating from a factory located near a specific direction. Further, for example, when the feature extracting unit 35 extracts related data RD related to particulate matter that has a small particle size and contains a specific element contained in soil and that flies from another direction, the output unit 37 can output information that particulate matter originating from soil located far away in the other direction is present in the storage location.

Further, when the feature extraction unit 35 extracts the related data RD indicating that the wind speed exceeds the predetermined threshold from the related data RD stored in the storage unit 31, the output unit 37 extracts the particulate matter It is possible to output information suggesting the possibility that is flying from a relatively distant position. On the other hand, when the relevant data RD indicating that the wind speed is equal to or less than the predetermined threshold is extracted, the output unit 37 outputs information suggesting the possibility that the particulate matter is coming from a relatively close position. can.

Furthermore, when the feature extraction unit 35 extracts the related data RD indicating that the particle size of the particulate matter is small and the wind speed exceeds the predetermined threshold value, the output unit 37 outputs the It can output information that suggests the possibility that it is flying from a distant position. On the other hand, when the relevant data RD indicating that the particle size of the particulate matter is large and the wind speed is equal to or lower than the predetermined threshold is extracted, the output unit 37 outputs It can output information that suggests the possibility that it is flying.

As described above, the feature extraction unit 35 extracts features that are a combination of features related to wind speed and features related to particle size of particulate matter, so that the source of particulate matter can be identified more accurately.

By extracting the features related to particulate matter as described above, it is possible to determine how to deal with the products stored in the storage location based on the properties of the particulate matter present in the space of the storage location. For example, when the feature extraction unit 35 extracts related data RD indicating that particulate matter that affects the product such as corrosion is present in the space of the storage location, the product may be transferred to another location. measures to prevent particulate matter from adhering to the product, such as placing a cover over the product, and washing the product, etc., by sprinkling water.

In addition, the feature extraction unit 35 can perform feature extraction processing to extract, from among the related data RD stored in the storage unit 31, related data RD indicating that a predetermined type of gas is included. For example, when the feature extraction unit 35 extracts the related data RD indicating that a predetermined type of gas is contained, the output unit 37 detects that the predetermined type of gas exists in the space of the storage location. You can output information about

For example, when the feature extraction unit 35 extracts the relevant data RD indicating the presence of corrosive gases such as sulfur oxides (SOx) and hydrogen sulfide (H ₂ S), the output unit 37 outputs the data of the storage location. Information indicating the presence of corrosive gas in the space can be output. As a result, when information indicating that corrosive gas exists in the space of the storage location is output, countermeasures such as moving the product to another location can be taken.

The feature extraction unit 35 determines whether the particle size of the particulate matter exceeds a predetermined threshold, the wind direction indicates a specific direction, the wind speed exceeds a predetermined threshold, and the particulate matter contains a predetermined element. The output unit 37 can also generate an alert when it extracts relevant data RD that match predetermined indicators for a predetermined characteristic, such as the presence of a predetermined type of gas, which is contained in the air. The output unit 37 can generate an alert by, for example, emitting a sound or displaying a predetermined display on the display unit of the analysis server 3 .

Generating an alert when relevant data RD matching a predetermined index is extracted indicates, for example, if there is particulate matter and/or gas in the storage space that could affect the product. The user can be made to visually and/or audibly recognize that there is. As a result, when an alert occurs, countermeasures such as moving products can be quickly executed.

2. Features of Embodiments The embodiments described above can also be expressed as follows.
(1) The analysis system is a system that analyzes particulate matter. The analysis system includes a data acquisition section, a feature extraction section, and an output section. The data acquisition unit acquires relevant data regarding particulate matter. The feature extraction unit extracts features included in the related data by executing a predetermined feature extraction process with the related data as input. The output unit outputs information related to the features extracted by the feature extraction unit.

(2) In the analysis system of (1) above, the feature extraction unit may extract related data whose instantaneous value exceeds the first threshold. In this case, the output unit may display a list of information related to related data whose instantaneous value exceeds the first threshold. As a result, it is possible to automatically extract relevant data whose instantaneous value exceeds the first threshold value and include anomaly, and to present to the user in which relevant data an anomaly has occurred.

(3) In the analysis system (1) or (2) above, the feature extraction unit may extract related data whose average value or median value exceeds the second threshold. In this case, the output unit may display a list of information related to related data whose average value or median value exceeds the second threshold. Thereby, it is possible to automatically extract related data whose average value or median value exceeds the second threshold value and include anomaly, and to present to the user in which related data an anomaly has occurred.

(4) In the analysis system of (2) or (3) above, the output unit may graphically display related data corresponding to the specified information among the information displayed in the list above. This makes it possible to visually confirm changes in related data including anomalies.

(5) In the analysis system of (1) to (4) above, the feature extraction unit calculates the correlation of the plurality of related data, and extracts information related to the related data for which the calculated correlation is equal to or greater than the third threshold. may be extracted. This eliminates the need for the user to calculate and analyze correlations for various combinations of multiple pieces of related data, so that related data with high correlation can be efficiently and accurately extracted.

(6) In the analysis system of (5) above, the output unit may display a scatter diagram of a plurality of related data whose correlation is equal to or greater than the third threshold. This makes it possible to visually confirm the magnitude of the correlation of the extracted related data.

(7) In the analysis system of (5) above, the output unit may display a plurality of scatter diagrams of the plurality of related data, and highlight the scatter diagrams of the plurality of related data whose correlation is equal to or greater than the third threshold. good. Thereby, it is possible to visually recognize which of all related data has a large correlation.

(8) In the analysis systems of (5) to (7) above, the related data may be data that changes over time. In this case, the feature extraction unit may calculate the correlation of multiple pieces of related data in a predetermined time interval. This allows calculation of the correlation of multiple pieces of related data in a specific time interval. Correlation of relevant data in a particular time interval provides information about events that occurred during that time interval.

(9) In the analysis system of (8) above, the predetermined time interval may be variable. This makes it possible to flexibly set the target time interval for calculating the correlation of a plurality of pieces of related data.

(10) In the analysis system of (8) or (9) above, the feature extraction unit may calculate the correlation of the plurality of related data in each of the plurality of small intervals included in the predetermined time interval. This makes it possible to obtain more detailed information about an event that occurred in a specific period by correlating related data in a specific small interval.

(11) In the analysis system of (8) to (10) above, the output unit may graphically display chronological changes in the plurality of related data whose correlation is equal to or greater than the third threshold. Thereby, it is possible to visually confirm changes over time in a plurality of related data.

(12) In the analysis system of (1) to (11) above, the data acquisition unit acquires the mass concentration of the particulate matter and information related to the elements contained in the particulate matter as related data. good. This allows analysis of the particulate matter based on features extracted from information about the mass concentration of the particulate matter and/or the elements contained in the particulate matter.

(13) In the analysis system of (12) above, the data acquisition unit may acquire the direction of the wind at the location where the particulate matter was collected as related data. In this case, the feature extraction unit determines whether the instantaneous value of the content of the element contained in the particulate matter in the related data in the specific wind direction exceeds the first threshold, or the content of the element included in the particulate matter Relevant data in which the mean or median of exceeds a second threshold may be extracted. As a result, it is possible to extract related data indicating anomalies regarding particulate matter flying from a specific direction, that is, particulate matter flying from a specific source.

(14) In the analysis system of (1) to (13) above, the data acquisition unit may acquire information on the particle size of the particulate matter as related data. In this case, the feature extraction unit may extract, from the related data, related data in which the particulate matter has a predetermined particle size range. This makes it possible to extract features related to the source of particulate matter, such as whether the source of particulate matter is near or far.

(15) In the analysis system of (1) to (14) above, the data acquisition unit may acquire, as related data, data relating to the gas at the location where the particulate matter was sampled. In this case, the feature extraction unit may extract relevant data indicating that the gas is a predetermined type of gas from the relevant data. As a result, it is possible to extract features related to the state of the atmosphere at the location where the data acquisition unit is installed, such as whether or not corrosive gas is present at the location where the data acquisition unit is installed.

(16) In the analysis system of (1) to (15) above, the data acquisition unit may acquire, as related data, data relating to the wind speed at the location where the particulate matter was sampled. In this case, the feature extraction unit may extract relevant data in which the wind speed exceeds a predetermined threshold value or is equal to or less than a predetermined threshold value. This makes it possible to extract features related to the source of particulate matter, such as whether the source of particulate matter is near or far.

(17) In the analysis system of (1) to (16) above, when the feature extraction unit extracts related data that matches a predetermined index for a predetermined feature from the related data, the output unit issues an alert. may occur. Thereby, it is possible to visually and/or audibly confirm that the relevant data matching the predetermined index has been acquired.

(18) The server is a server that acquires and analyzes relevant data regarding particulate matter. The server includes a feature extraction unit and an output unit. The feature extraction unit extracts features included in the related data by executing a predetermined feature extraction process with the related data as input. The output unit outputs information related to the features extracted by the feature extraction unit.

(19) The analysis method is for particulate matter. The analytical method comprises the following steps.
A step of obtaining relevant data on particulate matter.
A step of extracting features included in the related data by executing a predetermined feature extraction process with the related data as input.
A step of outputting information related to the extracted features.

(20) A program according to still another aspect of the present invention is a program for causing a computer to execute the above analysis method.

3. 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 required.
(A) The order and/or processing contents of each step shown in the flowcharts of FIGS. 4, 7, and 10 may be changed without departing from the gist of the invention.

(B) The above peak search function and average value search function may be provided in the data collection device 17 of the data acquisition unit 1 . The data collection device 17 having a peak search function and an average value search function determines whether the related data RD accumulated in the data collection device 17 includes a peak value equal to or greater than the first threshold and/or is equal to or greater than the second threshold. If the average value of

More specifically, for example, among the related data RD accumulated in the data collection device 17, a state in which the concentration of a specific component detected in a wind direction of a specific azimuth exceeds a specific value has passed for a specific period of time. When the related data RD exists, the data collection device 17 can issue a warning mail.

(C) In the automatic correlation extraction operation, when the correlation is calculated using the element ratio of the elements contained in the particulate matter, the correlation (coefficient of determination) is calculated by removing the data in which the element ratio is extremely small or large. Calculations may be performed.

(D) In the above first embodiment, the analysis system 100 is a so-called "client-server system" having the analysis server 3, but it is not limited to this. For example, the data acquisition unit 1 (each analysis device thereof, or the data collection device 17) is provided with the function of the feature extraction unit 35 and/or the output unit 37 to form an analysis system as so-called "edge computing". can also In this case, the data acquisition unit 1 may transmit, in addition to the related data RD, features extracted from the related data RD to the analysis server.

(E) The feature extraction unit 35 may implement feature extraction processing by, for example, a machine learning algorithm such as a neural network. Specifically, for example, a trained model of the neural network is generated using the related data RD having the feature to be extracted and the feature of the related data RD as teacher data, and this can be used as the feature extraction unit 35 . By inputting the relevant data RD to be analyzed to the feature extraction unit 35, which is a learning model, the features included in the relevant relevant data RD can be extracted.

(F) The feature extraction unit 35 extracts features included in the related data RD by other statistical analysis methods such as data mining for the related data RD, in addition to the statistical analysis method described in the first embodiment. can.

(G) The feature extraction unit 35 sets a predetermined threshold value and executes a number of algorithms for determining whether or not the data value included in the related data RD exceeds the threshold value, thereby extracting the features included in the related data RD. may be extracted.

(H) The analysis server 3 detects particles in the analyzers (first analyzer 11 to third analyzer 15) based on information on various processes in factories (for example, temperature, humidity, pressure, gas flow rate, etc.) It may be possible to predict the analysis result (related data RD) regarding the substance. Specifically, for example, models of various processes for predicting analysis results can be created, and the analysis results can be predicted from the calculation results of the model.

The present invention can be widely applied to analysis systems for particulate matter analysis.

100: Analysis system 1 Data acquisition unit 11 First analysis device 111 Collection filter 111a Delivery reel 111b Take-up reel P1 First position P2 Second position 113 Collection unit 115 First analysis unit 51 β-ray source 53 β-ray detector 51' light source 53' scattered light detection unit 117 second analysis unit 71 X-ray source 73 detector 119 control unit 131 suction pump 133 discharge port 135 suction port 13 second analysis device 15 third analysis device 17 data collection device 3 analysis server 31 storage unit 33 data reception unit 35 feature extraction unit 37 output unit 5 client terminal RD related data

Claims

An analysis system for analyzing particulate matter,
a data acquisition unit that acquires relevant data about the particulate matter;
a feature extraction unit that extracts features included in the related data by executing a predetermined feature extraction process with the related data as input;
an output unit that outputs information related to the feature extracted by the feature extraction unit;
An analysis system comprising:
The feature extraction unit extracts the related data whose instantaneous value exceeds a first threshold,
The output unit displays a list of information related to the related data whose instantaneous value exceeds a first threshold.
The analysis system according to claim 1.
The feature extraction unit extracts the relevant data whose average value or median value exceeds a second threshold,
The output unit displays a list of information related to the related data whose average value or median value exceeds a second threshold,
The analysis system according to claim 1 or 2.
4. The analysis system according to claim 2 or 3, wherein said output unit graphically displays said related data corresponding to specified information among said list-displayed information.
The analysis system according to any one of claims 1 to 4, wherein the feature extraction unit calculates correlations of a plurality of related data, and extracts information about related data for which the calculated correlations are equal to or greater than a third threshold.
The analysis system according to claim 5, wherein the output unit displays a scatter diagram of a plurality of related data whose correlation is equal to or greater than the third threshold.
The analysis system according to claim 5, wherein the output unit displays a plurality of scatter diagrams of a plurality of related data, and highlights a scatter diagram of a plurality of related data whose correlation is equal to or greater than the third threshold.
the relevant data is data that changes over time;
8. The analysis system according to any one of claims 5 to 7, wherein said feature extraction unit calculates correlations of a plurality of related data in a predetermined time interval.
The analysis system according to claim 8, wherein the predetermined time interval is variable.
10. The analysis system according to claim 8 or 9, wherein said feature extraction unit calculates correlations of a plurality of related data in each of a plurality of sub-intervals included in said predetermined time interval.
The analysis system according to any one of claims 8 to 10, wherein said output unit graphically displays changes over time of a plurality of related data whose correlation is equal to or greater than said third threshold.
The analysis system according to any one of claims 1 to 11, wherein the data acquisition unit acquires the mass concentration of the particulate matter and information related to elements contained in the particulate matter as the related data. .
The data acquisition unit acquires, as the relevant data, a wind direction at a location where the particulate matter is collected,
The feature extraction unit determines whether the instantaneous value of the content of the element contained in the particulate matter in the related data in a specific wind direction exceeds a first threshold, or the content of the element included in the particulate matter 13. The analysis system according to claim 12, wherein the relevant data whose mean or median amount exceeds a second threshold is extracted.
The data acquisition unit acquires information about the particle size of the particulate matter as the related data,
14. The analysis system according to any one of claims 1 to 13, wherein said feature extraction unit extracts, from said related data, said related data in which said particulate matter has a predetermined particle size range.
The data acquisition unit acquires, as the relevant data, data relating to the gas at the location where the particulate matter was sampled,
15. The analysis system according to any one of claims 1 to 14, wherein said feature extractor extracts, from said related data, said related data that said gas is a predetermined type of gas.
The data acquisition unit acquires, as the relevant data, data relating to the wind speed at the location where the particulate matter was sampled,
16. The analysis system according to any one of claims 1 to 15, wherein said feature extraction unit extracts, from said related data, said related data in which said wind speed exceeds a predetermined threshold or is equal to or less than a predetermined threshold.
17. The output unit generates an alert when the feature extraction unit extracts the related data that matches a predetermined index with respect to a predetermined feature from the related data. analysis system.
A server for obtaining and analyzing relevant data regarding particulate matter, comprising:
a feature extraction unit that extracts features included in the related data by executing a predetermined feature extraction process with the related data as input;
an output unit that outputs information related to the feature extracted by the feature extraction unit;
A server.
An analysis method for particulate matter,
obtaining relevant data about the particulate matter;
a step of extracting features included in the relevant data by executing a predetermined feature extraction process using the relevant data as an input;
outputting information related to the extracted features;
A method of analysis comprising:
obtaining relevant data about particulate matter;
a step of extracting features included in the relevant data by executing a predetermined feature extraction process using the relevant data as an input;
outputting information related to the extracted features;
A program for causing a computer to execute an analysis method including