WO2015063821A1

WO2015063821A1 - Analysis system

Info

Publication number: WO2015063821A1
Application number: PCT/JP2013/079070
Authority: WO
Inventors: 洋平川口; 永野　久志; 真人戸上
Original assignee: 株式会社日立製作所
Priority date: 2013-10-28
Filing date: 2013-10-28
Publication date: 2015-05-07
Also published as: JPWO2015063821A1; JP6121557B2

Abstract

In order to allow qualitative or quantitative analysis even if the number of positions at which measurement is to be performed is more than the number of sensors, this analysis system is provided with the following: a plurality of collection units; a collection control unit that makes the amounts of samples collected from the respective collection units vary over time; a sensor that observes a mixture of samples collected simultaneously from the plurality of collection units; and a separation processing unit that estimates the sample concentration at each collection unit on the basis of a time-series signal outputted by the sensor and a collection-amount signal that specifies the sample collection amount for each collection unit at each point in time.

Description

Analysis system

The present invention relates to an analysis system.

Japanese Patent Laid-Open No. 2008-015862 (Patent Document 1) is one of the documents related to a technique for detecting the occurrence of an abnormal situation using a plurality of sensors. The summary part of this patent document states that “input means 14 for receiving signal values from a plurality of sensors 1 to 6 for detecting an abnormal situation, and changing one or more of a plurality of signal values received by the input means 14. The weighting means 15 for weighting the plurality of signal values is weighted so that the degree of change of the changed signal value is different from the degree of change of the other changed or not changed signal value. Based on a plurality of signal values, the calculation means 16a for calculating a comprehensive abnormality index value and comparing the abnormality index value with a predetermined reference value to determine whether the abnormality index value indicates abnormality, And a determination means 17a that outputs a signal indicating that when it is determined that an abnormality detection device is indicated. "

JP 2008-015862 A

Currently, sensors such as mass spectrometer (MS), ion mobility spectrometer (IMS), gas chromatography (GC), liquid chromatography (LC), gas sensor, ion sensor, biosensor, microbial sensor, optical smoke detector are used. There are a wide variety of systems that observe the state of substances, microorganisms, and fine particles contained in the target space and analyze the observed signals. The substance or fine particles to be observed generally exist in the state of vapor, mist liquid, fine particles, or the like, but in this specification, these states are not distinguished and are also referred to as “samples”.

As described above, Patent Document 1 describes a method of determining whether or not an abnormal situation has occurred based on signals from a plurality of sensors. However, the analysis is performed when the number of measurement target positions is greater than the number of sensors. The method is not disclosed. In fact, in the analysis system shown in Patent Document 1, when the number of measurement target positions is larger than the number of sensors, it is impossible to determine whether or not an abnormal situation has occurred at a position where no sensor is installed.

The present invention has been made to solve such technical problems, and an analysis system that enables qualitative analysis and quantitative analysis at each measurement target position even when the number of measurement target positions is larger than the number of sensors. provide.

In order to solve the above problems, for example, the configuration described in the claims is adopted. The present specification includes a plurality of means for solving the above problems. To give an example, a plurality of collection units, a collection control unit that changes the amount of sample collected from each collection unit over time, and a plurality of units are provided. Based on a sensor that observes the sample collected simultaneously from the collection unit in a mixed state, a time-series signal output from the sensor, and a collection amount signal that specifies the collection amount of the sample at each time point to each collection unit And a separation processing unit that estimates the concentration of the sample in the collection unit.

According to the present invention, even when the number of measurement target positions (that is, collection units) is larger than the number of sensors, the concentration of the sample at each measurement target position (that is, the collection unit) is estimated, and qualitative analysis and / or quantitative analysis is performed. it can. Problems, configurations, and effects other than those described above will be clarified in the following description of embodiments.

1 is a diagram illustrating a hardware configuration of an analysis system according to Embodiment 1. FIG. 1 is a diagram illustrating a functional block configuration of an analysis system according to Embodiment 1. FIG. 5 is a flowchart for explaining a processing operation in the analysis system according to the first embodiment. The flowchart explaining the starting process of an analysis system. The flowchart explaining the standby process of an analysis system. The flowchart explaining the collection control processing (the 1) of an analysis system. The figure explaining the relationship between the collection amount signal a (t) and the switching control voltage value of the multi-channel DA converter 111 when the processing method shown in FIG. 6 is adopted. The flowchart explaining the collection control processing (the 2) of an analysis system. FIG. 9 is a diagram for explaining the relationship between the collection amount signal a (t) and the switching control voltage value of the multichannel DA converter 111 when the processing method shown in FIG. 8 is adopted. The flowchart explaining the separation process of an analysis system. The flowchart explaining the measurement stop process of an analysis system. The figure which shows the example of a reception screen of a noise level. The figure which shows the example of a result screen which a result output part outputs. 10 is a flowchart for explaining separation processing in the analysis system according to the second embodiment. FIG. 10 is a diagram illustrating a functional block configuration of an analysis system according to a third embodiment. 9 is a flowchart for explaining separation processing (part 1) in the analysis system according to the third embodiment. The figure which shows the example of a result screen which a result output part outputs. FIG. 10 is a diagram illustrating a functional block configuration of an analysis system according to a fourth embodiment. The flowchart explaining the collection control processing (the 3) of an analysis system. The flowchart explaining the collection control processing (the 4) of an analysis system. FIG. 10 is a diagram illustrating a hardware configuration of an analysis system according to a fifth embodiment. FIG. 10 is a diagram illustrating a functional block configuration of an analysis system according to a fifth embodiment. 10 is a flowchart for explaining separation processing in the analysis system according to the fifth embodiment. FIG. 10 is a diagram illustrating a hardware configuration of an analysis system according to a sixth embodiment. The figure which shows the example of a setting which changes a noise level sharply in a time-axis direction. The figure which shows the example of a setting which changes a noise level gently in a time-axis direction. 9 is a flowchart for explaining separation processing (part 2) in the analysis system according to the third embodiment. The flowchart explaining the collection control processing (the 5) of an analysis system.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. The embodiment of the present invention is not limited to the examples described later, and various modifications are possible within the scope of the technical idea.

[Example 1]
[Hardware configuration of analysis system]
In this embodiment, an analysis system capable of qualitative analysis and quantitative analysis at each measurement target position (collection unit) even when the number of measurement target positions (collection unit) is larger than the number of sensors will be described. One application example of this embodiment is an atmospheric analysis system that uses, for example, a mass analyzer as a sensor.

FIG. 1 shows a hardware configuration of the analysis system 10 according to the embodiment. The analysis system 10 according to the present embodiment includes a sensor 100, a pipe 101 that connects the sensor 100 and the mixing unit 116, a central processing unit 104 that controls the sensor 100 and processes the detection signal, a user interface unit 105, and a storage medium 109. , A volatile memory 110, a multi-channel DA (Digital-to-Analog) converter 111 that distributes collection control signals generated in the central processing unit 104 to the collection units 115_1 to 115_N, and collection units 115_1 to 115_N arranged at measurement target positions. The mixing unit 116 mixes samples acquired from the storage units 115_1 to 115_N through a pipe (not shown).

In the case of the present embodiment, the sensor 100 uses a mass analyzer composed of an ionization unit 102, a high-frequency power source 103, a detector 106, an ion transport unit 107, an ion trap 108, and vacuum pumps 112 to 114. The vacuum pumps 112 to 114 are used to maintain the pressure of each connected chamber at an appropriate value.

The collection units 115_1 to 115_N are provided with a steam inlet or a particulate collection mechanism. When the collecting units 115_1 to 115_N are steam inlets, each collecting unit opens and closes an electromagnetic valve based on a voltage value (control signal) given from the multi-channel DA converter 111, and the inhaled steam or mist droplets Control the flow rate. However, the control based on the voltage value is not limited to the two values of the “open” state and the “closed” state, and may be multi-value control that controls the tilt angle of the electromagnetic valve to an arbitrary angle.

When the collecting units 115_1 to 115_N are fine particle collecting mechanisms, each collecting unit determines the amount of fine particles to be collected depending on whether or not to cause a cyclone phenomenon based on the voltage value (control signal) given from the multi-channel DA converter 111. Control. However, the control based on the voltage value is not limited to the binary value of on / off of the cyclone phenomenon, but may be multi-value control for controlling the intensity of the cyclone phenomenon to an arbitrary intensity.

Vapor or mist droplets or fine particles collected by the collecting units 115_1 to 115_N are mixed through the mixing unit 116 and the pipe 101 and sent to the sensor 100. The sensor 100 introduces the vapor or mist droplets or fine particles introduced from the pipe 101 into the ionization unit 102 having an ion source and ionizes them. As the ionization method, for example, an electrospray ionization method, a sonic spray ionization method, or the like is used.

The generated ions are sent from the ionization unit 102 to the ion trap 108 via the ion transport unit 107. As the ion trap 108, for example, a quadrupole ion trap, a linear trap, or the like is used. For mass analysis, linear ion trap mass spectrometer, quadrupole ion trap mass spectrometer, quadrupole filter mass spectrometer, triple quadrupole mass spectrometer, time-of-flight mass spectrometer, magnetic field mass spectrometer Alternatively, a method such as ion mobility may be used.

The high frequency power supply 103 supplies a high frequency voltage to the ion trap 108 and traps ions inside the ion trap 108. The high frequency voltage applied to the ion trap 108 is temporally changed by the central processing unit 104. Due to the time change of the high-frequency voltage, the ions trapped in the ion trap 108 are sent to the detector 106 at different times according to the mass-to-charge ratio (m / z). The detector 106 converts the amount of ions that have reached into an electrical signal that is expressed as a voltage value, and sends the electrical signal to the central processing unit 104.

The central processing unit 104 converts the time to the ion mass-to-charge ratio (m / z) with respect to the time-series voltage signal, thereby representing the intensity representing the amount of ions at each mass-to-charge ratio (m / z). And is stored in the volatile memory 110. The mass spectrum at time t is stored as x (t) = (x_t1,..., X_tM) expressed in an array format with M elements. Further, the central processing unit 104 stores the observation signal X, which is a time series signal of the mass spectrum x (t), in the volatile memory 110.

The central processing unit 104 executes separation processing based on the observation signal X stored in the volatile memory 110. This separation process is executed based on a program stored in the storage medium 109. The user interface unit 105 presents the spectrum of each measurement target position output by the separation process. The user interface unit 105 may be, for example, a monitor having a touch panel, or may be a monitor, a mouse, a keyboard, or the like connected via another PC connected via a network. An output device may be used.

The central processing unit 104 executes collection control processing based on a program stored in the storage medium 109. The multi-channel DA converter 111 converts the voltage time series corresponding to each collection unit output by the collection control process into an actual voltage value and sends it to the collection units 115_1 to 115_N. Based on these voltage values, the flow rates of the vapor and the mist droplets by the collection units 115_1 to 115_N are controlled.

[Functional structure of analysis system]
In FIG. 2, the functional block diagram of the analysis system 10 of a present Example is shown. 2 corresponding to those in FIG. 1 are denoted by the same reference numerals. The operation input unit 204 is used for receiving a start operation, a measurement start operation, a measurement stop operation, and the like by a user, and outputs an operation signal corresponding to each operation to the collection control unit 201 and the sensor 100. Here, the start operation, the measurement start operation, and the measurement stop operation are executed, for example, by pressing a button included in the user interface unit 105.

The operation input unit 204 is also used for receiving an input of a noise level by a user. The noise level input to the operation input unit 204 is output to the separation processing unit 202. The collection control unit 201 determines each flow rate a (t) = (a_t1,..., A_tN) of the collection units 115_1 to 115_N according to the collection control process described later, and sends the determined values to the collection units 115_1 to 115_N. Output. Hereinafter, each flow rate a (t) is referred to as a “collected amount signal”. The collected amount signal is also output to the sensor 100. Here, the function of the collection control unit 201 is provided through a program executed on the central processing unit 104.

The sensor 100 performs mass analysis of the mixed sample introduced from the pipe 101 based on a prescribed measurement sequence. The measurement sequence includes processes such as an accumulation process, an exhaust waiting process, a mass scanning process, and an exclusion process, and a time-series control signal for controlling voltage applied to a plurality of electrodes, opening / closing of an electromagnetic valve, and on / off of the detector 106. Executed through. In the case of the present embodiment, the sensor 100 outputs the time series of the mass spectrum x (t) to the separation processing unit 202 as the observation signal X. The function of the separation processing unit 202 is provided through a program executed on the central processing unit 104.

The separation processing unit 202 receives the observation signal X, the collection amount signal, and the noise level as input, and performs separation processing of the observation signal X, and N separation signals (separation signals at each position) corresponding to the collection units 115_1 to 115_N, respectively. ) Is output.

The result output unit 203 presents the input separation signal to the user. The separation signal presentation method includes, for example, presentation of image information by the user interface unit 105, presentation of information by a braille display, presentation of information by voice, printing of image information through a printer, and the like. The function of the result output unit 203 is also provided through a program executed on the central processing unit 104.

[Processing of analysis system]
FIG. 3 shows processing operations executed in the analysis system 10 according to the present embodiment. The following processing operations are provided through programs executed on the central processing unit 104.
(Step S301)
The central processing unit 104 that functions as a control device of the analysis system 10 starts a processing operation of the analysis system 10 when receiving an activation operation through the operation input unit 204. At the same time, the central processing unit 104 executes the activation process of the sensor 100 and shifts to a standby state for accepting a measurement start operation.

(Step S302)
In this step, the central processing unit 104 functioning as the operation input unit 204 controls the sensor 100 and the collection control unit 201 to a standby state. In the standby process, the central processing unit 104 executes a monitoring process for a user operation input and a monitoring process for the apparatus state.

(Step S303)
In this step, the central processing unit 104 determines whether or not a measurement start operation has been performed by the user (whether or not the measurement start operation has been accepted). While the measurement start operation is not performed, the central processing unit 104 returns to step S302. On the other hand, when the measurement start operation is performed, the central processing unit 104 proceeds to step S304.

(Step S304)
In this step, the central processing unit 104 functioning as the collection control unit 201 (hereinafter referred to as “collection control unit 201”) stores “1” at the time number t.
(Step S305)
In this step, the collection control unit 201 determines whether or not the time number t is less than the threshold value (TH). If the time number t is less than the threshold value, the process proceeds to step S306.

(Step S306)
In this step, the collection control unit 201 determines whether a measurement stop operation has been performed. If the measurement stop operation has not been performed, the process proceeds to step S307. If the measurement stop operation has been performed, the process proceeds to step S312. .

(Step S307)
In this step, the collection control unit 201 determines a collection amount signal corresponding to the flow rate a (t) of each of the collection units 115_1 to 115_N. The determined collection amount signal is provided from the collection control unit 201 to each of the collection units 115_1 to 115_N.

(Step S308)
In this step, the central processing unit 104 functioning as a control device for the sensor 100 controls the operation of the sensor 100 and executes mass spectrometry. The sensor 100 obtains a mass spectrum x by mass-analyzing the mixed sample. Here, the mass spectra x corresponding to the time numbers t = 1 to T are combined in time series and then output from the sensor 100 as the observation signal X.

(Step S309)
In this step, the central processing unit 104 functioning as the signal separation unit 202 performs signal separation on the observation signal X based on the collection amount signal, and separates signals corresponding to each of the collection units 115_1 to 115_N, and results of qualitative analysis. The result of quantitative analysis is calculated.

(Step S310)
In this step, the central processing unit 104 functioning as the result output unit 203 performs a result output process. In the result output process, the result output unit 203 presents the separation signal, the result of the qualitative analysis, and the result of the quantitative analysis to the user. In addition, each scheduled information is accumulated in the storage medium 109.

(Step S311)
In this step, the collection control unit 201 adds “1” to the time number t and returns to step S305. Thereafter, the central processing unit 104 that controls the operation of the sensor 100 determines again whether or not the measurement stop operation has been performed in step S306. If the measurement stop operation has been performed, the measurement stop processing step S312 is performed. Proceed to

(Step S312)
In this step, the central processing unit 104 that functions as a control device for the sensor 100 executes measurement stop processing for the sensor 100. The measurement stop process is a process for stopping the apparatus normally. When the measurement stop process is completed, the analysis system 10 stops.

[Detailed operation of each process]
[Start process]
FIG. 4 shows the detailed operation of the activation process S301.
(Step S401)
In this step, the central processing unit 104 executes a vacuum degree initialization process. In the vacuum degree initialization process, the vacuum pumps 112 to 114 execute an evacuation operation, and the chambers connected to the pumps are reduced to appropriate pressures and maintained at those pressures.

(Step S402)
In this step, the central processing unit 104 executes a cleaning process. In the cleaning process, the central processing unit 104 requests the user to introduce a sample such as ammonia into the sensor 100, and waits for the sample to be introduced into the sensor 100 before starting the measurement. By executing the cleaning process, the substance (carryover substance) adsorbed inside the sensor 100 at the previous measurement is cleaned. Note that a sample introduction request to the user is presented through the user interface unit 105.

(Step S403)
In this step, the central processing unit 104 executes a mass-to-charge ratio calibration process. In the mass-to-charge ratio calibration process, the central processing unit 104 requests the user to introduce a standard material sample having a peak at a known mass-to-charge ratio (m / z) into the sensor 100, and introduces the sample. Wait and start measurement. The central processing unit 104 creates a correspondence table b (m) between each element number on the mass spectrum array and the mass-to-charge ratio (m / z) based on the measured peak position of the spectrum.

(Step S404)
In this step, the central processing unit 104 executes normal / abnormal determination processing (blank check). In the normality / abnormality determination process (blank check), the central processing unit 104 requests the user to introduce a known sample that does not contain the measurement target component into the sensor 100, and waits for the introduction of the sample before starting the measurement.

In this step, the central processing unit 104 determines whether or not the measured spectrum satisfies a preset condition. If a positive result is obtained, the central processing unit 104 determines that the measured spectrum is normal, and ends the activation process. On the other hand, if a negative result is obtained, the central processing unit 104 determines that the measured spectrum is abnormal, and returns to the cleaning process (step S402).

One of the conditions for determining that the measured spectrum is normal is, for example, that a large peak does not exist in the measured spectrum. One of the other conditions is that, when the measured spectrum is regarded as an M-dimensional vector, the cosine similarity with a reference spectrum measured in the past is higher than a threshold value. A known appropriate method may be used to determine whether or not the measured spectrum is normal.

[Standby processing]
FIG. 5 shows the detailed operation of the standby process S302.
(Step S501)
In this step, the central processing unit 104 functioning as the operation input unit 204 (hereinafter referred to as “operation input unit 204”) determines whether or not an operation signal is input. When the operation signal is input, the central processing unit 104 proceeds to step S502. When the operation signal is not input, the central processing unit 104 ends the standby process as it is.

(Step S502)
In this step, the operation input unit 204 outputs a control signal corresponding to the operation input to the sensor 100 and the collection control unit 201.

[Collection control processing (part 1)]
FIG. 6 shows a detailed operation example of the collection control process S307.
(Step S601)
In this step, the collection control unit 201 determines an N-dimensional vector collection amount signal a (t) that specifies the collection amount of each of the N collection units at each time number t.

At this time, the central processing unit 104 determines the individual collection amount signals a (t) so that the similarities between the collection amount signals a (t) at the respective time numbers t are reduced. In this embodiment, a method of storing independent random numbers in each element of the N collection units at each time number t is adopted. For example, when G collection unit numbers are randomly selected from elements of a set of numbers S = {1, 2, ..., N} representing each collection unit, the kth selected element number is represented by f (k ), The central processing unit 104 stores a set S1 (t) = {f (1), f (2), ..., f (k),. , F (G)}.

(Step S602)
In this step, the central processing unit 104 represents an N-dimensional vector in which the elements of the collection unit numbers included in the set S1 (t) are represented by “1” and the other elements are represented by “0 (zero)”. Store in a (t). Here, each N-dimensional element of the collection amount signal a (t) represents the flow rate of each of the N collection units.

(Step S603)
In this step, the central processing unit 104 converts the collection amount signal a (t) into a multi-channel open / close control voltage value through the multi-channel DA converter 111 and outputs it to the corresponding collection unit. FIG. 7 shows an example of the open / close control voltage value output from the multi-channel DA converter 111. FIG. 7 shows a case where N = 4 and G = 3.

At time number t, the collection unit to which the element of the collection signal a (t) is given with a dimension of “1” opens the electromagnetic valve or generates a cyclone and collects the sample. On the other hand, the collection unit to which the element of the collection amount signal a (t) is given with a dimension of “0” closes the electromagnetic valve or does not generate a cyclone and does not collect the sample.

[Collection control processing (2)]
FIG. 8 shows another detailed operation example of the collection control process S307.
(Step S801)
In this step, the central processing unit 104 functioning as the collection control unit 201 (hereinafter referred to as “collection control unit 201”) has N pieces of N given by real numbers “0” to “1” that follow an independent uniform distribution. An N-dimensional vector whose elements are random numbers is stored in the collected signal a (t).

(Step S802)
In this step, the collection controller 201 converts the multichannel DA converter 111 into a multichannel open / close control voltage value, and outputs it to the corresponding collector. Here, when the collection amount can be specified only in two ways, “0” or “1”, the collection amount is controlled by a binary signal as shown in FIG. In the case of this example that can be designated, the continuous value random number is converted as it is into a multi-channel switching control voltage value as in the example shown in FIG. FIG. 9 also shows a case where N = 4. When the continuous value random number is used, it is possible to select a (t) having a small similarity between the collection amount signals a (t) at each time number t, and the accuracy of the subsequent separation process is increased.

9 assumes that the flow rate of the collected sample is proportional to the opening / closing time of the electromagnetic valve, and encodes each element of the collection amount signal a (t), which is a continuous value, with the opening / closing time of each collecting unit. Yes. In particular, a device such as an electromagnetic valve cannot effectively control the instantaneous flow rate, and therefore coding based on opening and closing times is effective. In the case of an apparatus capable of continuously controlling the instantaneous flow rate, as described above, the collected amount signal a (t) may be used as it is as a multi-channel switching control voltage value.

[Separation process]
FIG. 10 shows a detailed operation example of the separation process S309. In the case of the present embodiment, the separation process is executed using a model of X = AP + E. In the model, the observation signal X is a t × M matrix. Further, A = (a (1), a (2),..., A (t)) is a t × N matrix representing the collection amount signal. The N × M matrix P is an unknown mass spectrum that will be observed when collecting fine particles separately from each collection unit. E is noise.

Note that the M mass points m of the unknown mass spectrum may be handled independently. In the following description, only the mass point m is considered and expressed as y = Ap + e.
However, y = (x_1m, ..., x_tm) ^ T and p = (p_1, ..., p_N) ^ T.
Here, by estimating p at each mass point m and arranging them for all the mass points m, the mass spectrum P at each collection unit can be obtained. This problem is to find N (t <N) unknowns from t equations.

In the separation process S309 in this embodiment, during the time number t that is not sufficiently smaller than N, the position (source of suspicion, luggage, etc.) where the explosive or chemical agent substance is generated and the amount of the substance do not change greatly. Assume a case.

(Step S1001)
In this step, the central processing unit 104 functioning as the signal separation unit 202 (hereinafter referred to as “signal separation unit 202”) normalizes each element of the collection amount signal A for each row τ based on Equation 1. .

This normalization process is simultaneously small when controlling the flow rate of vapor or mist droplets to be sucked by opening and closing the electromagnetic valve, or when controlling the amount of particulates collected depending on whether or not a cyclone phenomenon occurs. The actual collection amount is less than the value of the collection signal a_nτ (where nτ is a subscript) at the time when a larger number of collection units collect the sample at the same time than when the number of collection units collects the sample. Executed to solve the problem. By normalizing the collection amount signal, it is possible to reduce a deviation between the collection amount signal a_nτ (nτ is a subscript) corresponding to each collection unit n at each time τ and the actual collection amount.

(Step S1002)
In this step, the signal separation unit 202 multiplies the collection amount signal a_nτ (nτ is a subscript) corresponding to each time and each collection unit by p_n (n is a subscript) that is a separation result of each collection unit n. Weighted post-multiplication signals (Equation 3 shown below) corresponding to each time obtained by adding the obtained post-multiplication signals (Equation 2 shown below) over all collection units n (n = 1 to N), and The separation signal (signal of each collection unit) of each collection unit is estimated so that the degree of similarity with the corresponding observation signal y_τ increases.

At this time, the signal separation unit 202 calculates a separation signal p that minimizes the square of the error e | e | ^ 2. Such p is obtained by the following equation (4). Note that ε = (ε_t−T + 1,..., Ε_t) is a noise level input by the user. T is the time length of the observation signal used for estimation.

The longer the time length T of the observation signal used for estimation, the lower the follow-up to the temporal change in the amount of the substance to be detected and the amount of fine particles, but the estimation accuracy improves. On the other hand, the shorter the time length T, the better the follow-up to the temporal change in the amount of the substance to be detected and the amount of fine particles.

Further, the signal separation unit 202 may calculate a separation signal p that minimizes the square of the error e | e | ^ 2 under the constraint that each element of p is 0 or more. Since the value of the mass spectrum is a value proportional to the number of ions having each mass-to-charge ratio, all values are in principle 0 or more. By utilizing this property, it can be expected that the accuracy of separation is improved. Such a p can be obtained by repeating the following Expression 6 until the number of τ in which d_τ in Expression 5 below is less than a predetermined noise level ε_τ is equal to or less than a certain number. . Note that ε = (ε_1,..., Ε_τ) (τ is a subscript) is a noise level input by the user.

[Measurement stop processing]
FIG. 11 shows a detailed operation example of the measurement stop process S312.
(Step S1101)
In this step, the central processing unit 104 executes the same process as the cleaning process S402.
(Step 1102)
In this step, the central processing unit 104 executes a high frequency power supply stop process and stops the high frequency power supply 103.
(Step 1103)
After completing the stop of the high-frequency power supply 103, the central processing unit 104 executes a vacuum pump stop process and stops all the vacuum pumps 112 to 114.

[Noise level reception screen example]
FIG. 12 shows an example of a screen used when receiving an input of the noise level ε used by the separation processing unit 202 through the operation input unit 204 (user interface unit 105). As the noise level ε is larger, a deviation from the model, that is, noise is allowed, and therefore, it is less likely to be affected by noise included in the input signal. In this case, however, it is difficult to estimate a small value derived from a low concentration substance.

On the other hand, as the noise level ε is reduced, estimation with an emphasis on deviation from the model is performed, so that a slight value derived from a low-concentration substance can be estimated. However, in that case, it is easily affected by noise of the input signal.

Therefore, when there are a lot of contaminants in the environment or when the accuracy of the sensor is low, the user can obtain a high-precision separation result by setting the noise level ε high. Conversely, when there are few contaminants in the environment or when the accuracy of the sensor is high, the user can obtain a highly accurate separation result by setting the noise level ε low. In this way, separation processing that meets the needs of the application area can be performed.

Furthermore, since the noise level can be set for each time, the noise level at the old time is increased, and the noise level at the new time is decreased, thereby making an estimation that places more importance on the observation signal at the new time than at the old time. It can be carried out. When the amount of the detection target substance or fine particles changes every moment, the detection target substance is set by changing the noise level at the old time and the noise level at the new time sharply as shown in FIG. And follow-up to changes in the amount of fine particles can be made faster. However, in that case, the estimation accuracy deteriorates.

On the other hand, by setting the noise level at the old time and the noise level at the new time to change gradually as shown in FIG. 26, the estimation accuracy can be increased. However, in this case, the follow-up to the change in the amount of the substance to be detected and the fine particles is delayed.

When the change in the amount of the substance or fine particles to be detected is fast, or when it is required to shorten the measurement time required for detection, the noise level is set to change sharply as shown in FIG. The tracking accuracy can be increased. Conversely, when the change in the amount of the detection target substance or fine particles is slow, or when the measurement time required for detection may be long, the noise level is set to change gently as shown in FIG. Thus, the estimation accuracy can be increased.

[Result screen example]
FIG. 13 shows an example of a result screen output by the result output unit 203 (user interface unit 105). As shown in FIG. 13, mass spectra separated for each collection unit at different positions (the horizontal axis represents the mass-to-charge ratio and the vertical axis represents the detection intensity) are associated with the individual collection units 115_1 to N. Displayed with status. However, this display form is an example, and the qualitative analysis result and / or quantitative analysis result corresponding to the collection unit may be displayed on the screen together with the mass spectrum or independently.

[Summary]
As described above, if the analysis system according to the present embodiment is used, even when the number of measurement target positions (collection units) is larger than the number of sensors 100, qualitative analysis and quantitative analysis at each measurement target position can be realized. it can.

In addition, qualitative analysis and quantitative analysis at each measurement target position are dynamically variably controlled from the measurement target position (collection unit) and are collected from each measurement target position (collection unit) at each time point. It is not limited to the method of mixing the fine particles to be introduced into the sensor 100. For example, you may use a method of switching the collection unit that sequentially collects particles one by one (sequential point collection method), or a method of collecting particles from all the collection units simultaneously (all point mixed collection method) May be.

However, in the case of the sequential point collection method, the time required to measure fine particles from all the collection units is inevitably long. For this reason, the sequential point collection method cannot be overlooked when the location where the explosive or chemical agent material is generated (such as the suspect or baggage) moves at high speed or the amount of the substance changes at high speed. It is easy to occur, and in this application, high estimation accuracy cannot be expected for qualitative analysis and quantitative analysis.

Further, in the case of the sequential point collection method, fine particles can be collected only once for each measurement target position (collection unit). For this reason, when the error at each time of the observation signal X of the sensor or the error at the time of the collection amount of the collection unit is large, the error may greatly affect the results of qualitative analysis and quantitative analysis. The estimation accuracy is lowered.

On the other hand, in the case of the all-point mixed collection method, the measurement time can be shortened, but the chemical noise tends to increase as the number of collection units increases, and the estimation accuracy of qualitative analysis and quantitative analysis tends to decrease. Furthermore, qualitative analysis and quantitative analysis for each measurement target position (collection unit) are impossible.

In contrast to these methods, in the case of the present embodiment, since the collection of particulates from a plurality of measurement target positions (collecting units) is introduced into the sensor 100 in a mixed state, all the collecting units 115_1 to 115_N are measured. The time required to complete the process is short, and there are fewer oversights than the sequential point collection method.

In the case of the present embodiment, since the fine particles are collected a plurality of times for each measurement target position (collection unit), the error or collection of the observation signal X of the sensor 100 for each time as compared with each point collection method sequentially. It is difficult to be affected by the error of the collection amount of each part at each time. In addition, since the number of collection units mixed at each time point is smaller than the total number of measurement target positions (collection units), chemical noise can be reduced compared to the case where all points are collected simultaneously. The estimation accuracy of quantitative analysis is increased. Furthermore, in the case of the present embodiment, qualitative analysis and quantitative analysis for each measurement target position (collection unit), which is impossible in the case of the all-point mixed collection method, can be performed.

[Example 2]
In this example, when the assumption that only a small number of positions where substances having the same mass-to-charge ratio (m / z) are generated at the same time is established, a high-speed qualitative analysis at each measurement target position (collection unit) An analysis system capable of high-speed quantitative analysis will be described. The basic configuration of the analysis system according to the present embodiment is the same as that shown in FIG. 1, and the basic measurement procedure is the same as that shown in FIG. Below, only the processing procedure peculiar to a present Example is demonstrated.

[Outline of separation processing]
When the number T of equations is greater than or equal to the number N of unknowns, p can be estimated by the least square method under the assumption that the noise e is normally distributed. However, if the number of equations T is smaller than the number of unknowns N, the number of unknowns is relatively “underdetermined” and there are multiple candidate solutions that contain incorrect solutions other than the original solution. Can do. Therefore, in order to narrow down solution candidates, the present embodiment introduces an assumption that there are few “non-zero (zero)” elements included in p.

When this example is an atmospheric analysis system or a water quality analysis system, this assumption is that there are a small number of emission sources (factories, facilities, etc.) where substances with the same mass-to-charge ratio (m / z) are simultaneously emitted. It corresponds to the property of being limited to. If this example is an explosive trace detection system or chemical agent detection system for security applications, this assumption is based on the origin of the same mass-to-charge ratio (m / z) explosive or chemical agent material. Corresponds to the property that only a small number of locations (such as suspects and luggage) are available.

In Example 1, the separation signal p is output so that the square of the error e | y | ^ 2 at y = Ap + e is minimized. On the other hand, in the present embodiment, even if | e | ^ 2 is the same, the separated signal p is output so that there are fewer “non-zero” elements. As an example of a separation method for calculating a separation signal with fewer “non-zero” elements, a method based on compressed sensing will be described. An example of a method based on compressed sensing is Orthogonal Matching Pursuit (OMP) shown below.

[Details of separation processing]
FIG. 14 shows a detailed operation example of the separation process S309 used in the analysis system according to the present embodiment.
(Step S1401)
In this step, the central processing unit 104 functioning as the signal separation unit 202 (hereinafter referred to as “signal separation unit 202”) sets the collection amount signal A to Equation 1 as in step S1001 (FIG. 10) of the first embodiment. Based on this, normalization is performed for each row τ.

(Step S1402)
In this step, the signal separation unit 202 initializes the residual vector d by substituting y = (x_ (t−T + 1) m,..., X_tm) ^ T. However, a_nτ (nτ is a subscript) is the flow rate of the collection unit n at time τ. The central processing unit 104 performs steps S1403 to S1405 described later at most N times until the number of τs in which each element d_τ (τ is a subscript) of the residual vector d is less than a predetermined noise level ε_τ is equal to or less than a certain number. Iterate.

(Step S1403)
In this step, the signal separation unit 202 calculates a_j = (a_ (t−T + 1) j ... a_tj) ^ T having the maximum projection length b = d ^ T a_n with respect to the residual vector d. Find from the column vector of A. Then, the obtained j is stored in the array B.

(Step S1404)
In this step, the signal separation unit 202 obtains {q_u} that minimizes the distance of the following equation (7), and further updates the residual signal d by the following equation (8) based on the q_u.

(Step S1405)
In this step, the signal separation unit 202 excludes the column vector a_j from A. As a result of the above iteration, the observation signal y is decomposed into N values {q_u} corresponding to each collection unit. However, in the above algorithm, q_u for as many u as possible is calculated to be “0”.

In addition to the Orthogonal-Matching-Pursuit algorithm, there are Least-Angle-Regression-Stagewise (LARS) algorithm, Thresholding algorithm, Iterative-Reweighed-Least-Squares (IRLS) algorithm, etc. Also good.

Also in the present embodiment, setting the noise level has an effect of enhancing robustness against noise.
Estimation with greater emphasis on the assumption that the higher the noise level ε, the fewer non-zero elements included in p than the deviation from the model (the assumption that there are more zero elements included in p) Therefore, it is difficult to be affected by noise included in the input signal. In this case, however, it is difficult to estimate a small value derived from a low concentration substance.

On the other hand, the smaller the noise level ε, the greater the deviation from the model than the assumption that there are fewer non-zero elements in p (the assumption that there are more zero elements in p). Therefore, a slight value derived from a low-concentration substance can be estimated. However, in that case, it is easily affected by noise of the input signal.

[Summary]
The analysis system according to the present embodiment enables high-speed qualitative analysis and high-speed analysis at each measurement target position (collection unit) when the number of positions where substances having the same mass-to-charge ratio (m / z) are generated simultaneously is limited to a small number. Quantitative analysis is possible.

On the other hand, when the number of measurement target positions is larger than the number of sensors, in addition to the method described above, for example, a method of sequentially switching the collection unit for collecting particles one by one (sequential point collection method) is used. Alternatively, a method of collecting fine particles simultaneously from all the collecting units (all-point mixed collecting method) may be used.

However, in the case of the sequential point collection method, the time required to measure fine particles from all the collection units is inevitably long. For this reason, the sequential point collection method cannot be overlooked when the location where the explosive or chemical agent material is generated (such as the suspect or baggage) moves at high speed or the amount of the substance changes at high speed. It is easy to occur, and in this application, high estimation accuracy cannot be expected for qualitative analysis and quantitative analysis. In addition, when the error at each time of the sensor observation signal X and the error at the time of the collection amount of the collection unit are large, the error may greatly affect the results of the qualitative analysis and the quantitative analysis. Accuracy is lowered.

On the other hand, in the case of the all-point mixed collection method, the measurement time can be shortened, but the chemical noise tends to increase as the number of collection units increases, and the estimation accuracy of qualitative analysis and quantitative analysis tends to decrease.

In contrast to these methods, in the case of the present embodiment, the collection of particles from each measurement target position (collection unit) is mixed while dynamically controlling the collection of the particles from a plurality of measurement target positions (collection unit). By introducing the sensor 100 into the sensor 100 in such a state, it is possible to perform qualitative analysis and quantitative analysis at a higher speed and higher accuracy than other methods.

Since it is as above, since the analysis system concerning a present example is introduced into sensor 100 in the state where particles from a plurality of measurement object positions (collection part) were mixed like Example 1, all the collection parts The time required to measure 115_1 to 115_N is shortened, and overlooking is less than the sequential point collection method.

In the case of the present embodiment, since the fine particles are collected a plurality of times for each measurement target position (collection unit), the error or collection of the observation signal X of the sensor 100 at each time as compared with the respective point collection method. It is difficult to be affected by the error of the collection amount of each part at each time. In addition, since the number of collection units mixed at each time point is smaller than the total number of measurement target positions (collection units), chemical noise is smaller than when all points are collected simultaneously. The estimation accuracy of quantitative analysis is increased. Furthermore, it is possible to perform qualitative analysis and quantitative analysis for each measurement target position (collection unit), which is impossible in the case of the all-point mixed collection method.

Furthermore, when the analysis system according to this embodiment is used in an atmospheric analysis system, a water quality analysis system, an explosive trace detection system for security use, or a chemical agent detection system, substances having the same mass-to-charge ratio (m / z) are simultaneously used. By preferentially outputting separation results that occur in a small number of positions, even if the measurement time is short, that is, when the number of measurements is small, multiple candidate separation results can be appropriately narrowed down. it can. Therefore, high-speed qualitative analysis and high-speed quantitative analysis at each measurement target position are possible.

[Example 3]
In this embodiment, even when the number of measurement target positions (collection units) is larger than the number of sensors and many substances having values at the same point on the spectrum can be generated at the same time, qualitative analysis and quantification can be performed with high accuracy. An analysis system capable of performing analysis will be described.

The difference between the present embodiment and the above-described embodiment 2 is that a target substance database is prepared for the spectrum separated by the separation process, and the qualitative process and the quantitative process are executed based on the database. FIG. 15 illustrates a functional block configuration of the analysis system according to the third embodiment. In FIG. 15, parts corresponding to those in FIG. Hereinafter, a configuration unique to the present embodiment will be described.

The separation processing unit 1501 receives an observation signal X, which is a time-series signal of the mass spectrum x, a collected signal, a noise level, a spectrum template of the target substance database 1502, and a calibration curve, based on these information. The observation signal X is separated, and the separation signal of each measurement target position (collection unit) and the result of qualitative analysis and the result of quantitative analysis at each measurement target position (collection unit) are output.

The result output unit 1503 presents the input separation signal, the result of qualitative analysis of each measurement target position (collection unit), and the result of quantitative analysis to the user. The separation signal presentation method includes, for example, presentation of image information by the user interface unit 105, presentation of information by a braille display, presentation of information by voice, printing of image information through a printer, and the like. The function of the result output unit 1503 is also provided through a program executed on the central processing unit 104.

[Outline of separation processing]
The separation processing unit 1501 according to the present exemplary embodiment performs separation processing based on a model represented by X = AQR + E. Here, the observation signal X is a T × M matrix. A = (a (t−T + 1), a (2),..., A (t)) is a T × N matrix representing the collection amount signal. T is the time length of the observation signal used for estimation. Q, which is an N × J matrix, represents an index value corresponding to the concentration of the i-th substance in the collection unit n, with the element q_ni in the n-th row and i-th column (ni is a subscript). The matrix R which is J × M is a mass spectrum whose row vector of the i-th row is measured when the i-th substance is present in a unit concentration. E represents noise. R is a sequence of spectral templates R_i = (r_i1,..., R_iM) stored in the target substance database 1502, and is known.

The model can be converted into the model shown in the following equation (9).

Both sides of Equation 7 are a t × J matrix. Each of J substances i may be handled independently. Hereinafter, only the i-th column vector corresponding to the i-th substance is considered and expressed as y = Aq + e. Here, y represents a vector in the i-th column of XR ^ +. e represents the i-th column vector of ER ^ +. q = (q_1i, ..., q_Ni) ^ T. If the qs of all the substances are arranged after estimating the q of each substance i, the index value Q corresponding to the concentration of each substance in each collecting unit can be obtained.

This problem is to find N unknowns from T equations. As in the case of the first embodiment, when T is greater than or equal to N, q can be estimated by the method of least squares under the assumption that e is a normal distribution. However, when T is smaller than N, the number of unknowns is relatively “underdetermined” and there may be a plurality of solution candidates including erroneous solutions other than the original solution. Therefore, in order to narrow down the solution candidates, the assumption that there are few “non-zero” elements included in q is introduced.

When this embodiment is an atmospheric analysis system or a water quality analysis system, this assumption corresponds to the property that the positions of emission sources (factories, facilities, etc.) from which substances of the same component are simultaneously discharged are limited to a small number. When this embodiment is an explosive trace detection system or chemical agent detection system for security applications, this assumption is based on the location of the source of the same component explosive or chemical agent substance (suspected person, luggage, etc.) Corresponds to the property of being limited to a small number.

For reference, the assumption used in Example 2 was that there were a small number of positions where substances having the same mass-to-charge ratio (m / z) were generated simultaneously, but the same mass-to-charge ratio (m / z). This assumption does not hold when multiple types of substances having the same) are likely to be generated at the same time. The assumption of this embodiment is valid even in such a case. Therefore, although there are many kinds of substances having the same mass-to-charge ratio (m / z), this embodiment is effective in a region where the same substance is not frequently generated at the same time.

[Details of separation processing]
FIG. 16 illustrates a detailed operation example of the separation process S309 used in the analysis system according to the present embodiment.
(Step S1601)
In this step, the central processing unit 104 (hereinafter referred to as “signal separation unit 1501”) functioning as the signal separation unit 1501 sets the collection amount signal A to Equation 1 as in step S1001 (FIG. 10) of the first embodiment. Based on this, normalization is performed for each row τ.

(Step S1602)
In this step, the signal separation unit 1501 calculates XR ^ + by multiplying the observation signal X as an input signal by a pseudo inverse matrix R ^ + of R from the right. That is, the observation signal X is converted to XR ^ +. Here, R is configured by reading the template R_i = (r_i1,..., R_iM) of the spectrum of each substance i from the target substance database 1502 and arranging R_i.

(Step S1603 to Step S1606)
In these steps, processing similar to that in steps S1402 to S1405 is executed. However, the vector in the i-th column of XR ^ + is y, and q is estimated by the same algorithm as in the second embodiment.
(Step S1607)
In this step, the signal separation unit 1501 estimates the concentration for each substance at each measurement target position (collection unit). In the concentration estimation, a calibration curve c_i (w) is read from the target substance database 1502, and q is estimated using the calibration curve. Specifically, c_i (q_ni) is calculated by substituting the index q_ni corresponding to the concentration of substance i at the measurement target position (collection unit) n into the calibration curve, and the concentration of substance i at the measurement target position (collection unit) n is calculated. Estimate c_ni.

(Step S1608)
In this step, the signal separation unit 1501 estimates the presence or absence of the substance i based on whether or not the concentration c_ni exceeds a prescribed concentration.

FIG. 27 shows a detailed operation example of the separation process S309 used in the analysis system according to the present embodiment. The central processing unit 104 performs steps S2701 to S2709, which will be described later, a maximum of W times until the number V of the collection units in which the target substance is estimated to be “present” in the previous loop does not change compared to that in the previous loop. Iterate. Where W is a specified positive constant.

(Steps S2701 to S2708)
In these steps, processing similar to that in steps S1601 to S1608 is executed.

(Step S2709)
In this step, the signal separation unit 1501 determines the time length T of the observation signal used for estimation based on the following Equation 10. However, C is a predetermined positive constant, and V is the number of collection parts in which the substance i is estimated to be “present”.

As described above, the longer the time length T of the observation signal used for estimation, the lower the follow-up to the temporal change in the amount of the substance to be detected and the amount of fine particles, but the estimation accuracy improves. On the other hand, the shorter the time length T, the better the follow-up to the temporal change in the amount of the substance to be detected and the amount of fine particles. In consideration of the fact that the larger V is, the longer the observation signal is required, the longer T is, and the smaller V is, the shorter the observation signal is, the better estimation is possible. By shortening, the time length necessary to obtain sufficient estimation accuracy is set to T. Thereby, in this step, the balance between the estimation accuracy and the followability to the time change can be automatically adjusted.

[Result screen example]
FIG. 17 shows an example of a result screen output by the result output unit 1503 (user interface unit 105). In the present embodiment, the spatial distribution of the concentration c_ni corresponding to a certain target substance is displayed as a two-dimensional image so as to be superimposed on the arrangement map of the individual collection units 115_1 to N. Thereby, the user can know the concentration distribution of the measurement target substance at each measurement target position.

[Summary]
If the analysis system according to the present embodiment is used, since the fine particles from each measurement target position (collection unit) are mixed as in the case of the second embodiment, the time until all the collection units are measured can be shortened. Less overlooked than each point collection method. In addition, since the analysis system according to the present embodiment collects the microparticles multiple times from each measurement target position (collection unit), the error and the collection unit of the observation signal of the sensor at each time are sequentially compared with each point collection method. It is hard to be influenced by the error of the collected amount for each time.

In addition, since the analysis system according to the present embodiment requires a smaller number of collecting units to be mixed as compared with the all-point mixed collecting method, compared with the case of collecting particles from all measurement target positions (collecting units) at the same time. Chemical noise is small, and estimation accuracy of qualitative analysis and quantitative analysis is high. Furthermore, the analysis system according to the present embodiment can perform qualitative analysis and quantitative analysis for each measurement target position, which is impossible with the all-point mixed collection method.

Furthermore, when the analysis system according to the present embodiment is used in an atmospheric analysis system, a water quality analysis system, an explosive trace detection system for security use, or a chemical agent detection system, various types having the same mass-to-charge ratio (m / z). Even if these substances are generated at the same time, there is a property that the same component is generated at a small number of positions simultaneously. The analysis system according to the present embodiment uses this property and outputs by giving priority to a separation result in which there are a small number of measurement target positions where the same component is generated at the same time. Even when the number is small, a plurality of separation result candidates can be appropriately narrowed down. Therefore, even when many types of substances having the same mass-to-charge ratio (m / z) are generated at the same time, high-speed qualitative analysis and high-speed quantitative analysis at each measurement target position are possible.

[Example 4]
In this embodiment, an analysis system capable of performing qualitative analysis and quantitative analysis with high accuracy even when the number of measurement target positions (collection units) is larger than the number of sensors and high-speed throughput is required will be described. . The difference between the analysis system according to the present embodiment and the analysis system according to the third embodiment described above is that the collection control process is performed based on the result of the qualitative analysis output from the separation processing unit 1501.

FIG. 18 shows a functional block configuration of the analysis system according to the fourth embodiment. In FIG. 18, the same reference numerals are assigned to the corresponding parts to those in FIG. Hereinafter, a configuration unique to the present embodiment will be described. As illustrated in FIG. 18, the collection control unit 1801 according to the present embodiment has a path for feeding back the qualitative analysis result output from the separation processing unit 1501. The collection control unit 1801 determines the flow rate a (t) = (a_t1,..., A_tN) to be assigned to the collection units 115_1 to 115_N based on the result of the qualitative analysis of each position output from the separation processing unit 1501, The value is output to the collection units 115_1 to 115_N.

[Collection control processing (part 1)]
FIG. 19 shows a detailed operation example of the collection control process S307 used in the analysis system according to the present embodiment. In the following processing, a set of all numbers assigned to the collection unit is set as a total collection unit number set S, and sets generated by arbitrarily dividing the set S into two equal parts are S1 (t) and S2 (t ). Further, it is assumed that the collection unit number set S1 (-1) at the start of measurement (t = 0) stores all collection unit number sets S. That is, it has been initialized. The collection unit number sets S1 (t) and S2 (t) are both held and used until the execution of the collection control process S307 after the next time measurement.

(Step S1901)
In this step, the central processing unit 104 functioning as the collection control unit 1801 (hereinafter referred to as “collection control unit 1801”) determines whether S1 (t−1) is the total collection unit number set S or not. If S1 (t-1) is the total collection unit number set S, the collection control unit 1801 proceeds to step S1902, and otherwise proceeds to S1905.

(Step S1902)
In this step, the collection control unit 1801 obtains the t-1th element of the observation signal y for the target substance i that is the vector in the i-th column of XR ^ + (that is, the signal value y_ (t at the previous time t-1). -1)) It is determined whether or not it exceeds a certain threshold. In the case of “greater than or equal to the threshold value”, the collection control unit 1801 determines that the substance i is “present”. When it is determined that any target substance i is “present”, the collection control unit 1801 proceeds to step S1903, and otherwise proceeds to step S1904.

(Step S1903)
In this step, the collection control unit 1801 randomly divides the entire collection unit number set S into two equal parts, and stores the set elements after bisection into the collection unit number sets S1 (t) and S2 (t).
(Step S1904)
In this step, the collection control unit 1801 stores all the collection unit number sets S in the collection unit number set S1 (t).

(Step S1905)
In this step, the collection control unit 1801 obtains the t-1th element of the observation signal y for the target substance i that is the vector in the i-th column of XR ^ + (that is, the signal value y_ (t at the previous time t-1). -1)) Determine if it is above a certain threshold. In the case of “greater than or equal to the threshold value”, the collection control unit 1801 determines that the substance i is “present”. If it is determined that any target substance i is “present”, the collection control unit 1801 proceeds to step S1906, and otherwise proceeds to step S1907.

(Step S1906)
In this step, the collection control unit 1801 randomly divides the collection unit number set S1 (t-1) into two equal parts, and sets the set elements after the bisection into the collection unit number sets S1 (t) and S2 (t). Store.
(Step S1907)
In this step, the collection control unit 1801 randomly divides the collection unit number set S2 (t-1) into two equal parts, and sets the set elements after the bisection into the collection unit number sets S1 (t) and S2 (t). Store.

(Step S1908)
In this step, the collection control unit 1801 sets the element of the collection unit number included in the collection unit number set S1 (t) to “1” and other elements (that is, the collection unit number set S ′ (= S−S1). The N-dimensional vector generated so that the element of the collection unit number included in (t)) is set to “0” is stored in the collection amount signal a (t). Each N-dimensional element of a (t) represents the flow rate of each of the N collection units.
(Step S603)
In this step, the collection control unit 1801 converts the collection amount signal a (t) into a multi-channel switching control voltage value and outputs it.

According to the processing operation shown in FIG. 19, the collection operation can be controlled so that fine particles are preferentially collected for the collection unit corresponding to the collection unit number set S1 (t) in which the target substance may exist. . As a result, the detection is completed at high speed as the analysis system.

[Collection control processing (2)]
FIG. 20 shows another detailed operation example of the collection control process S307. In FIG. 20, parts corresponding to those in FIG. The difference between the processing operation shown in FIG. 20 and the processing operation shown in FIG. 19 is that step S2001 exists between step 1908 and (step S1903, step S1904, step S1906, step S1907).

(Step S2001)
In this step, the collection control unit 1801 collects the elements of the collection unit numbers included in the elements other than the collection unit number set S1 (t) used for collection (that is, the collection unit number set S ′ (= S−S1 (t)). ) Are randomly selected from the list and added to the collection unit number set S1 (t). In the case of this processing as well, as in the processing operation shown in FIG. 19, the collection amount so as to preferentially collect fine particles for the collection unit that is an element of the collection unit number set S1 (t) that may contain the target substance. Can be controlled. For this reason, detection can be completed at high speed. Further, in the processing operation shown in FIG. 20, several collections are also made from collection units (collection unit number elements included in the collection unit number set S ′ (= S−S1 (t)) once excluded from collection targets. Since a part is selected and added to the collection target, detection omission due to the influence of the error of the observation signal X of the sensor 100 for each time and the error of the collection amount of the collection part for each time can be made difficult.

[Collection control processing (part 3)]
FIG. 28 shows a detailed operation example of the collection control process S307. In FIG. 28, parts corresponding to those in FIG. The difference between the processing operation shown in FIG. 28 and the processing operation shown in FIG. 6 is that step 2801 exists between the start and (step S601, step S602, step S603).

(Step S2801)
In this step, the collection control unit 1801 determines G according to Equation 11. However, C is a predetermined positive constant, and V is the number of collection parts in which the substance i is estimated to be “present”.

In this step, according to Equation 11, the collection control unit 1801 decreases G as V increases, and increases G as V decreases.

When the number of positions (collection units) where the measurement target substance or fine particles exist is limited to a small number, the present embodiment has a function of mixing and measuring substances or fine particles from a plurality of collection units, and an observation signal. The combination of functions for estimating the presence or absence of substances or fine particles in each collection unit has the effect of reducing measurement time and speeding up qualitative analysis and high-speed quantitative analysis at each measurement target position.

As the most extreme example, if there are no substances or fine particles to be measured in any collection unit, the "all-point mixed collection method" that mixes all the collection units at the same time is optimal if chemical noise is ignored. . This is because in such a case, it is not estimated that the substance is present even if the substances or fine particles of all the collecting parts are mixed. It is because it understands that it does not do.

On the other hand, when there are a large number of positions (collecting units) where the substances or fine particles to be measured are present, the effect of reducing the measurement time is small. In particular, when substances or fine particles to be measured are present in all the collection units, the collection is performed for each collection unit, that is, the sequential point collection method described above is optimal. This is because, in such a case, no matter what collection unit is combined and mixed, the substance or fine particles to be measured is “present”, so the amount of information obtained by mixing does not increase. If it is> T, it is an underdetermined condition, and conversely, the information amount is reduced.

In this operation example, G is controlled according to V, which is an estimate of the number of positions (collection units) where the measurement target substance or fine particles are present, so that the maximum amount of information can be obtained. It is possible to improve the accuracy of qualitative analysis and quantitative analysis as much as possible even with the same measurement time.

[Summary]
As described above, when the analysis system according to the present embodiment is used, the collected amount signal is preferentially collected from the collection unit where the target substance may exist using the information of the result of the qualitative analysis. Therefore, qualitative analysis and quantitative analysis can be performed at high speed and with high accuracy.

[Example 5]
In the present embodiment, a case where the path length from each collecting unit to the sensor 100 is different for each collecting unit will be considered. When there is a difference in path length from each collecting unit to the sensor 100, the amount actually collected even if the same collecting amount signal is given to the collecting unit in which the same amount of fine particles or the like exists (ie, sensitivity) May differ from one collection unit to another.

In addition, when it takes time for the fine particles or gas collected from the collection units 115_1 to 115_N to reach the sensor 100, high-speed control cannot be performed, and the collection time signal switching time and the actual observation signal X time A delay occurs between For example, even when the collected signal is open (ie, “1”), the amount actually collected is less than expected, or even when the collected signal is closed (ie, “0”). The amount of time that occurs is greater than expected. Therefore, in this embodiment, an example of an analysis system capable of performing qualitative analysis and quantitative analysis with high accuracy even when the sensitivity differs according to the collection unit and time or when high-speed control is difficult will be described.

[Hardware configuration of analysis system]
The difference between the analysis system according to the present embodiment and the analysis system according to Embodiment 1 is that a different standard agent is installed in the vicinity of each collection unit. FIG. 21 illustrates a hardware configuration of the analysis system 2101 according to the present embodiment. In FIG. 21, parts corresponding to those in FIG. As shown in FIG. 21, in the case of the present embodiment, standard agents 2102_1 to 2102_N are installed in the vicinity of the collecting units 115_1 to 115_N. Standard agents 2102_1 to 2102_N are substances having peaks at different mass points.

[Functional structure of analysis system]
FIG. 22 shows a functional block configuration of the analysis system 2101 of this embodiment. Note that, in FIG. 22, the same reference numerals are given to portions corresponding to FIG. 15. The characteristic parts of this embodiment are a separation processing unit 2202 and a standard agent database 2201. The standard agent database 2201 stores a spectrum template and a calibration curve for each standard agent. The separation processing unit 2202 reads the template and the calibration curve from the standard agent database 2201, and separates the wrinkle promotion sign X into separation signals corresponding to each measurement target position (collection unit).

[Details of separation processing]
FIG. 23 shows a detailed operation example of the separation processing S309 executed in the present embodiment.
(Step S2301)
In this step, the central processing unit 104 (hereinafter referred to as “separation processing unit 2202”) functioning as the separation processing unit 2202 obtains a spectrum template R_n = from each standard agent n (corresponding to each collection unit n) from the standard agent database 2201. (r_n1,..., r_nM) and a calibration curve c_n (w) are read out, and the collected amount signal is corrected using these. Specifically, the inner product value w of the mass spectrum x′_n and R_n obtained by weighted averaging of the flow rate a_tn of the collection unit n is calculated, and w is substituted into the calibration curve c_n (w), so that the collection unit n is opened. The sensitivity c_n at the time of the state (ie, “1”) is estimated. The collection amount after the sensitivity c_n is multiplied by the element in the n-th column of the collection amount signal matrix A = (a (1), a (2), ..., a (t)) and the sensitivity is corrected Get the signal matrix A.

(Step S2302)
In this step, the separation processing unit 2202 determines whether or not the sensitivity c_n of each collection unit n is within a specified range. The separation processing unit 2202 determines that the mixing is normal when each sensitivity c_n is within the specified range and proceeds to step S2304, and determines that the mixing is abnormal when each sensitivity c_n is outside the specified range. Then, the process proceeds to step S2303.
(Step S2303)
In this step, the separation processing unit 2202 presents information on the collection unit n determined that the sensitivity c_n is out of the specified range through the user interface unit 105.

(Steps S2304 to S2309)
These steps are executed when it is determined that the sensitivities c_n corresponding to the respective collection units n are all within the specified range. Since each step corresponds to the described step, detailed description is omitted. For example, step S2304 (input signal conversion processing) corresponds to step S1602, and step S2305 (residual vector initialization processing) corresponds to step S1402. Step S2306 (nearest neighbor mixed pattern search process) corresponds to step S1403, and step S2307 (residual signal update process) corresponds to step S1404. Step S2308 (nearest neighbor mixed pattern removal process) corresponds to step S1405. Step S2309 (quantitative processing) corresponds to step S1607, and step S2310 (threshold processing) corresponds to step S1608.

[Summary]
If the analysis system according to the present embodiment is used, the separation process is executed only when the sensitivity is the same even when the sensitivity varies depending on the collection unit and time. For this reason, high detection accuracy can be guaranteed. Furthermore, since it is possible to easily know the collection unit in which an abnormality is found in sensitivity, the maintenance work of the collection unit can be made efficient.

[Example 6]
In this embodiment, when the number of measurement target positions (collection units) is larger than the number of sensors 100, an example of an analysis system that can perform qualitative analysis and quantitative analysis with high accuracy even when the sensitivity differs for each collection unit. Will be explained.

[Hardware configuration of analysis system]
The difference between the analysis system according to the present embodiment and the analysis system according to the first embodiment is that a plurality of sensors are used. FIG. 24 illustrates a hardware configuration of the analysis system 2401 according to the present embodiment. The analysis system 2401 of this embodiment includes a plurality of collection units 2402_1 to 2402_N, a plurality of mixing units 2403_1 to 2403_K, a plurality of pipes 2404_1 to 2404_K, and a plurality of sensors 2405_1 to 2405_K. However, N and K are natural numbers that satisfy N> K.

For example, when the collection units 2402_1 to 2402_N are steam inlets, the flow rate of the steam or mist droplets to be sucked can be controlled by opening and closing the electromagnetic valve based on the voltage value input from the multi-channel DA converter 111. Each of the collecting units 2402_1 to 2402_N has K electromagnetic valves arranged in parallel, and the opening and closing thereof can be controlled independently.

For example, when the collection units 2402_1 to 2402_N are fine particle collection mechanisms, the amount of fine particles to be collected can be controlled based on whether or not the cyclone phenomenon occurs based on the voltage value input from the multi-channel DA converter 111. In this case, each of the collection units 2402_1 to 2402_N has K cyclone mechanisms arranged in parallel, and can start and stop them independently.

Vapor-young or mist-like droplets or fine particles collected by the collecting units 2402_1 to 2402_N are collected by K mixing units 2403_1 to 2403_N arranged in parallel. K pipes 2404_1 to 2402_K are connected to the mixing units 2403_1 to 2403_K in a one-to-one relationship, and steam or mist-like droplets or fine particles collected from the collecting units 2402_1 to 2402_N in the pipes are connected. It is sent to the sensors 2405_1 to 2405_K. For example, in the case of mass analysis processing, mass analysis is performed in each of the sensors 2405_1 to 2405_K, and a mass spectrum given by an M-dimensional vector is obtained. The MK-dimensional vector x = (x_1 ... x_K) obtained by connecting these M-dimensional K vectors in series is used as the observation signal X, and the same processing as in the above-described embodiment is executed.

[Summary]
If a plurality of sensors are used as in the analysis system according to the present embodiment, the amount of information of the observation signal X increases, and the accuracy of qualitative analysis and quantitative analysis can be improved. In particular, when the sensor price per unit and the maintenance cost are relatively low, by increasing the number of sensors as in this embodiment, the accuracy of signal separation, even with a smaller number of time points, The accuracy of qualitative analysis and quantitative analysis can be improved.

[Other embodiments]
In addition, this invention is not limited to an above-described Example, Various modifications are included. For example, the above-described embodiments have been described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described. Further, a part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment. Further, it is possible to add, delete, and replace other configurations for a part of the configuration of each embodiment.

In addition, each of the above-described configurations, functions, processing units, processing means, and the like may be realized by hardware by designing a part or all of them with, for example, an integrated circuit. Each of the above-described configurations, functions, and the like may be realized by software by interpreting and executing a program that realizes each function by the processor. Information such as programs, tables, and files for realizing each function can be stored in a recording device such as a memory, a hard disk, or an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, or a DVD.

Also, the control lines and information lines indicate what is considered necessary for the explanation, and not all the control lines and information lines on the product are necessarily shown. Actually, it may be considered that almost all the components are connected to each other.

DESCRIPTION OF SYMBOLS 10 ... Analysis system, 100 ... Sensor, 101 ... Piping, 104 ... Central processing unit, 105 ... User interface part, 111 ... Multichannel DA converter, 115_1-115_N ... Collection part, 116 ... Mixing part, 201 ... Collection control part , 202 ... separation processing unit, 1501 ... separation processing unit, 1502 ... target substance database, 1503 ... result output unit, 1801 ... collection control unit, 2101 ... analysis system, 2102 ... standard agent, 2201 ... standard agent database, 2202 ... separation Processing unit, 2401 ... analysis system, 2401-2_1 to 2402_N ... collection unit, 2403_1 to 2403_K ... mixing unit, 2404_1 to 2404_K ... piping, 2405_1 to 2405_K ... sensor.

Claims

N (a natural number greater than or equal to 2) collection units collecting samples from the outside,
Mixing the samples individually collected from the N collecting units, and outputting a mixed sample; and
K sensors (a natural number less than or equal to 1 and less than N) that observe the type or amount of substances contained in the mixed sample and output an observation signal;
A collection control unit that outputs N collection amount signals corresponding to the N collection units on a one-to-one basis, and controls the amount of the sample collected from each of the collection units at each time point. A collection control unit that switches the N collected amount signals at each time so that the similarity between the collected amount signals is low,
Based on the relationship between the observed signal observed at each time and the N collected signal at each time, the sample is collected from only one of the N collecting units. An analysis system comprising: a separation processing unit that estimates a separation signal corresponding to an observation signal in the case of measurement for all of the N collection units.
In the analysis system according to claim 1,
The separation processing unit is
After multiplication by multiplying the separated signal p_n (n is 1 to N) corresponding to each of the collection units by the collection amount signal a_nτ (n is 1 to N) corresponding to each time τ and each of the collection units A process of adding a signal for all the collecting units to generate a weighted added signal corresponding to each time;
Performing the process of estimating the separated signals corresponding to the respective collection units so that the similarity between the generated post-weighted addition signal and the observed signal y_τ corresponding to each time τ is increased. An analysis system characterized by
The analysis system according to claim 2,
The collection control unit sets the collection amount signal a_nτ (n is 1 to N) corresponding to each time τ and the individual collection unit to a large positive value as the collection amount of the sample is larger and the collection unit. The collection amount signal a_nτ (n is 1 to N) corresponding to each time τ and the individual collection units is set to a value close to 0 (zero) as the time and the collection unit with a small collection amount of the sample are controlled. An analysis system characterized by control.
The analysis system according to claim 1,
The separation processing unit estimates the separation signal corresponding to each of the collection units so that the number of non-zero (zero) elements included in the N separation signals is smaller. system.
The analysis system according to claim 1, further comprising a target substance database that stores spectrum templates and calibration curves of a plurality of substances,
The separation processing unit estimates an index for the amount of the detection target substance or the detection target fine particles contained in the sample collected from each of the collection units based on the spectrum template and the calibration curve read from the target substance database. An analysis system characterized by
The analysis system according to claim 5,
The collection control unit changes the N collection amount signals over time so as to preferentially collect the sample from the collection unit that is highly likely to contain a detection target substance or detection target fine particles based on the index. An analysis system characterized by
The analysis system according to claim 4,
It further has an operation input unit that receives an input of noise tolerance,
The separation processing unit places more importance on the smaller number of non-zero elements compared to the degree of similarity between the weighted added signal and the observed signal as the input noise tolerance is larger An analysis system characterized in that the separated signal is estimated.
The analysis system according to claim 1,
N standard agents arranged in the vicinity of the N collection units, each having a peak at a different mass point;
A standard database storing spectral templates and calibration curves of the N standard drugs;
The separation processing unit corrects the collected amount signal used at the time of the estimation based on the spectrum template and the calibration curve read from the standard agent database.
The analysis system according to claim 8,
The separation processing unit detects an abnormality of each collection unit according to the corrected collection amount signal.
The analysis system according to claim 2,
It further has an operation input unit that accepts an input of noise tolerance at each time,
The separation processing unit is configured to perform the weighted addition at a time when the noise tolerance is small as compared to a degree of similarity between the input signal after the weighted addition at the time when the noise tolerance is large and the observation signal. An analysis system characterized in that a separated signal is estimated with an emphasis on the degree of similarity between a post signal and the observed signal.
The analysis system according to claim 2,
The separation processing unit further executes a process of estimating the separation signal corresponding to each collection unit under a constraint that the value of the separation signal p_n (n is 1 to N) is 0 or more. Analysis system.
The analysis system according to claim 5,
The collection control unit is configured to control the collection amount signal to a positive value as the number of the collection units that are highly likely to contain the detection target substance or the detection target fine particles is estimated based on the index. Reduce the number of
The number of collection units that control the collection amount signal to a positive value is increased as the number of the collection units that are highly likely to contain a detection target substance or detection target fine particles is estimated based on the index. An analysis system characterized by
The analysis system according to claim 5,
The amount of collection used for the separation process increases as the number of the collection units that are likely to contain the detection target substance or the detection target fine particles is high based on the index at each time t. Increase the time length T of the signal a_nτ (n is 1 to N, τ is tT-1 to t) and the observed signal y_τ (n is 1 to N, τ is tT-1 to t),
The collection amount signal a_nτ (n is 1 to N, τ is used for the separation process) as the number of the collection units having a high possibility that the detection target substance or the detection target fine particles are present based on the index is estimated to be small. tT-1 to t) and the observation signal y_τ (where n is 1 to N and τ is tT-1 to t) are shortened.