JP5327965B2 - Gas concentration measuring method and gas concentration measuring device - Google Patents

Description

  The present invention relates to a gas concentration measurement method and a gas concentration measurement device.

  When measuring the gas concentration, measurement using a gas detector tube is performed. The gas detection tube is filled with a reagent (detection agent) that causes a color change when exposed to a specific gas, and the gas concentration is detected by reading the length of the discolored layer. The gas detector tube is easy to handle and can detect the concentration of the target gas component simply and in a short time regardless of the level of skill of the operator.

  Conventionally, the color change length of the gas detection tube has been read visually, but the inventors of the present application have previously proposed a method for obtaining the color change layer by reading it with an optical scanner (see Patent Document 1). By using this method, data reproducibility is improved by automatic reading, and the amount of color change can be measured with good reproducibility even when the color change boundary is blurred. In addition, a slight change in color can be detected, and a slight change in color change length can also be measured. Furthermore, a lot of noise is mixed in the detector tube image, which can also be reduced by signal processing. In addition to these, if an accumulation effect by continuous suction is added, it is possible to achieve higher sensitivity by one digit or more than the visual reading device.

  In addition, an apparatus for obtaining the discoloration layer was developed using a one-dimensional CCD sensor, and an experiment for measuring the gas concentration distribution at the same time was performed using the apparatus. A one-dimensional CCD sensor can be manufactured in a smaller device than an optical scanner and can be controlled by a microcomputer, so that a personal computer is not required and can be directly connected to a wireless LAN. However, since only one type of gas detection tube can be measured at the same time, it is necessary to further reduce the size.

  Then, the apparatus which reads the image (the said discoloration layer) of a gas detection tube using the camera of the mobile telephone was proposed (refer patent document 2). In this apparatus, an image can be sent as an email attachment and multi-point gas concentration information can be easily collected. In addition, remote control can be performed by using a mobile phone. Furthermore, by using a camera, it is possible to read a color change that is difficult to determine visually, so that the sensitivity can be improved.

  However, as the measurement of the discoloration area (discoloration amount) of the discoloration layer by the camera proceeds, problems that did not appear visually have become apparent.

  That is, the white portion of the gas detector tube (the portion that is not discolored by a specific gas, the undiscolored portion) undergoes a color change due to humidity. This color change cannot be ignored when calculating the color change area from an image taken by a camera. Further, it was found that the discoloration portion discolored by aeration of gas varies depending on the humidity even at the same concentration. Specifically, when a gas having the same concentration is vented, the color of the discolored portion of the gas detection tube may become darker as the humidity of the gas increases. This humidity dependency cannot be ignored for gas concentration measurement.

Japanese Patent No. 4009725 JP 2007-218878 A

  As described above, when measuring the gas concentration using the gas detection tube, the influence of the humidity of the gas cannot be ignored. For this reason, there is a possibility that the gas concentration cannot be obtained correctly in a normal atmosphere where the humidity is not controlled.

  Therefore, an object of the present invention is to provide a gas concentration measuring method and a gas concentration measuring device capable of accurately (correctly) calculating a gas concentration using a gas detector tube.

  The gas concentration measurement method according to the present invention is a gas concentration measurement method using a gas detection tube, wherein a gas is passed through the gas detection tube, and the gas detection tube through which the gas has passed is photographed with a camera. An imaging step, analyzing the image data of the gas detector tube imaged by the camera, calculating a color change rate due to the gas of the gas detector tube, and a humidity sensor to determine the humidity of the gas. And a gas concentration calculation step of calculating the concentration of the gas by correcting the color change due to the humidity of the gas from the color change speed and the humidity.

  In this gas concentration measurement method, by performing a series of steps from the ventilation step to the gas concentration calculation step, the gas concentration is adjusted regardless of the humidity of the gas (correcting the humidity dependency of the color change of the gas detector tube). Can be calculated correctly.

In the gas concentration measurement method, the gas detection tube includes a plurality of gas detection tubes, and the discoloration speed calculation step extracts image data of each gas detection tube from the image data of the gas detection tube. It is characterized by analyzing.
In this gas concentration measurement method, the gas detector tube is composed of a plurality of gas detector tubes, and in the color change rate calculation step, the image data of each gas detector tube is cut out and analyzed from the image data of the gas detector tube. The gas concentration of a plurality of gases can be calculated with the gas detector tube.

In the gas concentration measuring method, the gas concentration calculating step calculates the concentration of the gas by using a non-linear mapping means.
In this gas concentration measurement method, the gas concentration can be calculated from the color change rate and the humidity by using a nonlinear mapping means in the gas concentration calculation step.

In the gas concentration measurement method, the nonlinear mapping means is a neural network.
In this gas concentration measurement method, in the gas concentration calculation step, the gas concentration can be accurately calculated by inputting the color change rate and the humidity into a neural network (after learning is completed) which is a nonlinear mapping means.

Further, in the gas concentration measuring method, the color change rate calculating step calculates the color change rate from a relationship between a color change area obtained using a division quadrature method and time.
In this gas concentration measurement method, in the discoloration rate calculation step, by analyzing the image data of the gas detector tube photographed by the camera, specifically, the discoloration area obtained using the sectional quadrature method, ) Calculate the color change rate (reaction rate) from the relationship with time. By using this sectional quadrature method, noise in the image data of the gas detector tube can be reduced, and an appropriate color change area (color change rate) can be calculated.

Further, in the gas concentration measurement method, the color change rate calculating step calculates the color change area by correcting the color change due to the humidity of the gas in an uncolored portion that is not changed by the gas in the gas detection tube. It is characterized by doing.
In this gas concentration measurement method, in the color change rate calculation step, the color change area is calculated by correcting the color change due to the humidity of the gas in the uncolored portion that has not been changed by the (specific) gas in the gas detector tube. As a result, the humidity dependency of the color change of the unchanged color portion can be corrected, and an appropriate color change area can be calculated.

Further, in the gas concentration measurement method, the discoloration rate calculating step includes calculating a difference between the image data of the gas detector tube photographed in advance before starting the measurement and the image data of the gas detector tube photographed in the photographing step. For the image data, correction of the discoloration due to the humidity of the gas in the uncolored portion is performed by a threshold method in which the pixel value equal to or less than a predetermined threshold is 0, and the discoloration area is calculated using the corrected image data. It is characterized by that.
In this gas concentration measurement method, in the color change rate calculating step, the humidity change of the color change of the unchanged color portion is corrected appropriately by performing the correction of the color change due to the humidity of the gas of the unchanged color portion by the threshold method. Can do.

Further, in the gas concentration measurement method, before starting measurement, the gas detection tube is preliminarily ventilated with air to saturate the humidity response of the gas detection tube. discoloration amount of the correction and said row Ukoto.
In this gas concentration measurement method, in the color change rate calculation step, the humidity change of the color change of the unchanged color portion is appropriately corrected by correcting the color change due to the humidity of the gas of the unchanged color portion by the preliminary ventilation method. be able to.

  A gas concentration measuring device (system) according to the present invention includes a gas detection tube, a pump for venting gas through the gas detection tube, a camera for photographing the gas detection tube through which the gas has passed, and the camera. Analyzing the imaged image data of the gas detector tube, calculating a color change rate by the gas of the gas detector tube, a humidity sensor for detecting the humidity of the gas, the color change rate and the humidity And a gas concentration calculating means for calculating the concentration of the gas by correcting the color change due to the humidity of the gas.

  In this gas concentration measuring device, the gas concentration can be correctly calculated regardless of the humidity of the gas by providing the pump to the gas concentration calculating means.

Further, in the gas concentration measuring apparatus, the gas detection tube is composed of a plurality of gas detection tubes, and the color change rate calculating means cuts out image data of each gas detection tube from the image data of the gas detection tube. It is characterized by analyzing.
In this gas concentration measuring apparatus, the gas detector tube is composed of a plurality of gas detector tubes, and the discoloration rate calculating means cuts out and analyzes the image data of each gas detector tube from the image data of the gas detector tube. The gas concentration of a plurality of gases can be calculated with this gas detector tube.

In the gas concentration measuring apparatus, the gas concentration calculating means is a non-linear mapping means.
In this gas concentration measuring apparatus, since the gas concentration calculating means is a non-linear mapping means, the gas concentration can be calculated from the color change rate and the humidity.

In the gas concentration measuring apparatus, the nonlinear mapping means is a neural network.
In this gas concentration measuring apparatus, the non-linear mapping means is a neural network, so that the gas concentration can be accurately calculated.

Further, in the gas concentration measuring apparatus, the color change rate calculating means calculates the color change rate from a relationship between a color change area obtained using a division quadrature method and time.
In this gas concentration measuring device, the discoloration rate calculating means analyzes the image data of the gas detector tube photographed by the camera, specifically, the discoloration area obtained using the sectional quadrature method, the time, The discoloration speed is calculated from the relationship. By using this sectional quadrature method, noise in the image data of the gas detector tube can be reduced, and an appropriate discoloration area can be calculated.

Further, in the gas concentration measuring apparatus, the discoloration rate calculating means calculates the discoloration area by correcting the discoloration due to the humidity of the gas in an undiscolored portion not discolored by the gas of the gas detection tube. It is characterized by doing.
In this gas concentration measuring apparatus, the color change rate calculating means corrects the color change due to the humidity of the gas in the uncolored portion that has not been changed by the gas in the gas detection tube, and calculates the color change area. As a result, the humidity dependency of the color change of the unchanged color portion can be corrected, and an appropriate color change area can be calculated.

Further, in the gas concentration measuring device, the discoloration speed calculating means may calculate a difference between the image data of the gas detector tube photographed in advance before starting the measurement and the image data of the gas detector tube photographed in the photographing step. For the image data, correction of the discoloration due to the humidity of the gas in the uncolored portion is performed by a threshold method in which the pixel value equal to or less than a predetermined threshold is 0, and the discoloration area is calculated using the corrected image data. It is characterized by that.
In this gas concentration measuring device, the color change rate calculating means appropriately corrects the humidity dependency of the color change of the unchanging color portion by performing the correction of the color change due to the humidity of the gas of the unchanging color portion by the threshold method. Can do.

Further, in the gas concentration measuring apparatus, the color change rate calculating means may correct the color change due to the humidity of the gas in the uncolored portion by previously ventilating the atmosphere to the gas detection tube before starting the measurement. The pre-venting method is used to saturate the humidity response of the gas detector tube .
In this gas concentration measurement device, the color change rate calculating means appropriately corrects the humidity dependence of the color change of the unchanged color portion by correcting the color change due to the humidity of the gas of the unchanged color portion by the pre-ventilation method. be able to.

  According to the present invention, it is possible to provide a gas concentration measuring method and a gas concentration measuring device capable of accurately (correctly) calculating a gas concentration using a gas detection tube.

It is a flowchart which shows the gas concentration measuring method which concerns on this embodiment. It is a figure which shows the structure of the gas concentration measuring apparatus which concerns on this embodiment. It is a figure which shows the calculation method of a discoloration area. It is a figure which shows the relationship between time and a discoloration area. It is a figure which shows a discoloration part and an unchanging color part. It is a figure which shows the humidity response example (relationship of X coordinate and f (x)) in an unchanging color part. It is a figure which shows the humidity response example (relationship of time and a discoloration area) in an unchanging color part. It is a figure which shows the effect (The relationship between X coordinate and f (x)) of the threshold method. It is a figure which shows the effect (relationship of time and a discoloration area) of a threshold method. It is a figure which shows the effect (relationship of an X coordinate and f (x)) of the prior ventilation method. It is a figure which shows the effect (relationship of time and a discoloration area) of the prior ventilation method. It is a figure which shows the preparation methods of the same concentration and different humidity gas. It is a figure which shows the measurement result (the relationship between humidity and reaction rate) by the same concentration and different humidity gas. It is a figure which shows the measurement result (relationship of gas concentration and reaction rate) by the same concentration and different humidity gas. It is a figure which shows the relationship between a true gas concentration and the estimated gas concentration by a neural network.

  An embodiment of the present invention will be described with reference to FIGS.

  As shown in FIG. 1, the gas concentration measurement method according to the present embodiment is a gas concentration measurement method using a gas detection tube, and includes a ventilation step S1 for passing gas through the gas detection tube, and a gas through which the gas has passed. A photographing step S2 for photographing the detection tube with a camera, an image data of the gas detection tube photographed with the camera, a color change rate calculating step S3 for calculating a color change rate due to gas in the gas detection tube, and a humidity sensor Detection step S4 for detecting the humidity of the gas, and a gas concentration calculation step S5 for calculating the gas concentration by correcting the color change due to the gas humidity from the color change speed and humidity.

  A gas detection tube 2 (see FIG. 2) that constitutes the gas concentration measurement device 1 used in the gas concentration measurement method includes alumina, a glass tube with a predetermined scale having a constant inner diameter of about several millimeters to several tens of millimeters, A detection agent in which particles having a particle diameter of about several tens of μm made of silica gel or the like are coated with a reagent that reacts with a gas to be detected and changes color is filled. By passing the sample gas (atmosphere) from one end of the glass tube, a discoloration layer is formed from the gas introduction end (upstream end) to the exhaust end (downstream end) according to the amount of gas component to be detected. It grows. The length of the discoloration layer is longer as the gas concentration to be detected is higher and shorter as the gas flow rate (volume) is constant.

  In the gas detection tube 2, the detection agent differs depending on the type of detection gas, and the color to be changed also differs. The gas detection tube 2 is commercially available, for example, No. manufactured by Gastec Corporation. 4LL (for hydrogen sulfide), No. 4 122P (for toluene), No. 71 (for methyl mercaptan) can be used, and gas detector tubes corresponding to various gases can be used.

  A plurality of (six) gas detection tubes 2 used for measurement are supported by a support member, and a detection tube holder (housing) 3 that shields the gas detection tubes 2 from external light (see FIG. 2). , 21 cm × 16.5 cm × 14.5 cm). The configuration of the support member and other light shielding members in the detection tube holder 3 is disclosed in detail in the above-mentioned Patent Document 2.

  The ventilation step S <b> 1 is a step of venting gas through the gas detection tube 2 by the pump 4.

The pump 4 (see FIG. 2) is connected to the downstream end of the gas detection pipe 2 through a pipe, and is sucked to the pump 4 side by driving the pump 4 to vent the gas (atmosphere) into the gas detection pipe 2. The power line of the pump 4 is connected to a 5V DC power supply via a power MOSFET, and ON (drive) / OFF (stop) is controlled by the controller.
The pump 4 connects one pump of the same type to each of the six gas detection tubes 2 and sucks each gas detection tube 2 with an equal suction force.
In addition, as long as the pump 4 has a suction force capable of sucking a plurality of gas detection tubes 2 equally, six downstream pipes may be assembled and sucked by a single pump.
Further, the drive power source of the pump 4 is not limited to the DC power source, and may be an AC power source. Furthermore, you may comprise so that a some pump may be combined.

  The photographing step S2 is a step of photographing with the camera the gas detection tube 2 in which gas has been ventilated in the aeration step S1. Specifically, this is a step of photographing a transmitted light image transmitted through the gas detection tube 2 illuminated by light from the illumination means (light source) 5 (see FIG. 2) with the camera module 6 (see FIG. 2). This camera is composed of an illumination means 5 and a camera module 6.

The illuminating means 5 constitutes a surface emitting light source using a cold cathode tube as a light source. In order to obtain uniform surface-emitting illumination light, the cold-cathode tube has a bent shape within the surface (housing). Further, the light emitted from the cold cathode tube is diffused by a diffuser plate made of a white acrylic plate, and surface-emitting illumination light with excellent uniformity of luminance distribution in the light emitting surface can be obtained.
The electrodes at both ends of the cold-cathode tube are connected to an inverter for DC-AC conversion, and emit light by supplying high-voltage, high-frequency AC. The power supply cable of the illumination means 5 is connected to a power supply via a control circuit using a power MOSFET so that ON (lighting) / OFF (lighting off) can be controlled by the controller.

As such an illuminating means 5, for example, a surface emitting cold cathode tube backlight set M-416 manufactured by Akizuki Denshi Co., Ltd. can be used.
The illumination means 5 is not limited to the one using a cold cathode tube as a light source, but an LED (light emitting diode), an organic EL (electroluminescence), or the like may be used as long as the light quantity necessary for photographing can be obtained and surface emission is possible. A light source may be used.

  The camera module 6 uses a digital camera equipped with a two-dimensional image sensor in order to photograph a plurality of gas detection tubes 2 simultaneously. The image pickup unit included in the camera module 6 includes a camera lens and a two-dimensional image pickup device, and converts an image formed on the two-dimensional image pickup device by the camera lens into an electric signal (image data). The converted image data is appropriately subjected to AD (analog-digital) conversion and image compression processing in the camera body.

  The camera module 6 includes digital output terminal means, and sends the image data to an FPGA (Field Programmable Gate Array, Cyclone, Altera) 7 (see FIG. 2) which will be described later. Since the clock of the camera module 6 is as fast as 27 MHz, this image data is sent to the FPGA 7, and then stored in the SRAM (512k × 8bit, IDT71V424S12PH, IDT) 8 (see FIG. 2).

As such a camera module 6, for example, MTV54KOD manufactured by Akizuki Denshi Co., Ltd. can be used.
The camera module 6 is not particularly limited as long as it can perform digital output and obtain a color image. The processing result of the FPGA 7 is transmitted to the host computer 9 via the RS232C, but the transmission means may be other transmission means such as a wireless LAN, a wired LAN, a USB, a mobile phone, and a PHS. The host computer 9 may be in a remote place.

  In addition, photographing (measurement) conditions such as a photographing interval of the gas detector tube 2 by the camera and selection of the gas detector tube 2 to be used are input to a dedicated user interface (in the host computer 9, see FIG. 2). Yes. Then, the input photographing condition is transmitted to the FPGA 7 (under that condition), photographing is started. The specific shooting interval is set to transmit image data to the FPGA 7 about once every several tens of seconds (a fixed time interval).

  The color change rate calculation step S3 is a step of analyzing the image data of the gas detection tube 2 taken by the camera and calculating the color change rate due to the gas in the gas detection tube 2. Specifically, this is a step of calculating the color change rate from the relationship between the color change area obtained using the sectional quadrature method and the (aeration) time.

The image data of the gas detector tube 2 photographed by the camera in the photographing step S2 is sent to the FPGA 7 (color change speed calculating means). Then, the FPGA 7 (including the CPU) that has received the image data calculates the color change area and the color change speed by analyzing (processing) the image data. The image data used for this analysis is data of a reference image and a photographing (measurement) image. Note that the capacity of the SRAM 8 for storing image data is set to a capacity capable of storing signals for two pieces of image data (for reference and for measurement).
When the gas detector tube 2 is composed of a plurality of gas detector tubes, the FPGA 7 cuts out and analyzes the image data of each gas detector tube from the image data of the gas detector tube 2.

First, as a method of calculating the discoloration area, reference image data that is manually captured before starting measurement and measurement image data that is captured at regular time intervals are used. And the difference from each measurement image data (difference processing is performed), and the discoloration area is calculated from the difference image.
The method for calculating the discoloration area includes known processes such as an averaging process and a moving average process in addition to the difference process.

In this difference image, as shown in FIG. 3, the position (discoloration distance) in the gas traveling direction (discoloration direction) of the gas detector tube 2 (scale not shown) is x, and the luminance change is f (x). It is a graph which shows the relationship between both. The gas detector tube 2 changes color from the left end toward the right side, and the hatched portion of this graph indicates a discolored region. FIG. 3 schematically shows the gas detection tube 2 in comparison.
In FIG. 3, f _ya is a value of a luminance change when the color is completely changed. On the other hand, in the unchanging color area, the luminance change value is 0 by the difference processing. The area where the luminance change gradually changes in the middle of the graph is an area where discoloration is in progress. In this embodiment, the discoloration area is used instead of the discoloration length so that the discoloration amount can be appropriately detected even when the boundary between the discoloration layer (discoloration part) and the non-discoloration layer (undiscoloration part) is ambiguous. .

This discoloration area is calculated by the sectional quadrature method. Specifically, since the sum of the luminance change in the entire range of the layer filled with the detection agent (the area shown by hatching in FIG. 3) is the discoloration area S, this sum is expressed by S = Σf (x). The color change area S is calculated.
By using the sectional quadrature method in this way, noise in the image data of the gas detector tube 2 can be reduced, and an appropriate color change area can be calculated.

  Next, as a method of calculating the color change rate, the color change area is calculated in each measurement image, and the relationship between the color change area and the ventilation time is plotted (see FIG. 4). The discoloration area increases with time, and since the gradient at this time is the discoloration rate (reaction rate), the discoloration rate is calculated by calculating the gradient.

When calculating the above-mentioned color change area, the color change area is calculated by correcting the color change due to the humidity of the gas in the uncolored portion (see FIG. 5).
The humidity response (humidity dependency of the color change) of the unchanging color portion will be described.

  The humidity response of the non-discolored portion is a minute discoloration other than the discoloration caused by a specific gas, and is a change that can be barely confirmed by comparing with the gas detector tube 2 before opening with the naked eye.

From FIG. 6 showing the humidity response example (relationship between the X coordinate of the gas detection tube 2 and f (x)) of the uncolored portion, the response due to humidity after the color change region due to the target gas (hydrogen sulfide in the present embodiment). It can be seen that this response saturates after a few minutes.
When the above-described discoloration area and discoloration speed are calculated in a state in which the response due to the humidity of the undiscolored portion remains included, as shown in FIG. 7, the discoloration speed (slope) immediately after the start of measurement increases and depends on time. It turns out that it is not constant. This is because the color change due to the humidity of the gas in the uncolored portion that is not changed by the target gas is added to the color change area.

  As a method for correcting the color change due to the humidity of the gas in the uncolored portion, there are a threshold method and a preliminary ventilation method.

The threshold method is a method of calculating a discoloration area by regarding a change in the pixel as 0 when the change value of the luminance change f (x) at each position (pixel) is equal to or less than the threshold in the difference image.
When this method is used, the FPGA 7 (discoloration speed calculation means) is configured to have the function.

By using this method, the humidity response of the unchanging color portion can be removed (see FIG. 8).
Further, from FIG. 9, it is possible to obtain a constant result regardless of time without increasing the color changing speed immediately after the start of measurement due to the response of the unchanged color part. Note that the color change area S is 0 immediately after the start of measurement because the color change caused by gas is small.

The pre-venting method is a method in which measurement is performed after allowing the atmosphere to pass through the gas detection tube 2 for a sufficient time (empirically about 15 minutes) to saturate the humidity response of the uncolored portion.
Note that when this method is used, the color change speed calculation means includes an FPGA 7 and a pump 4.

By using this method, it is possible to remove the humidity response at the unchanging color portion as in the threshold method (see FIG. 10).
In addition, as in the threshold method, it is possible to obtain a result that is constant regardless of time without increasing the discoloration speed immediately after the start of measurement, as in the threshold method, from FIG.
In addition, unlike the threshold method, it can be seen that an appropriate discoloration area and discoloration speed can be calculated immediately after the start of measurement.

  As described above, the correction of the color change due to the humidity of the gas in the unchanging color portion is performed by the threshold method or the pre-venting method, whereby the humidity dependency of the color change in the unchanging color portion can be appropriately corrected. From this, the discoloration speed can be calculated for an appropriate (correct) discoloration area.

  As described above, when calculating the discoloration area, the discoloration due to the humidity of the gas in the undiscolored portion was corrected by using a threshold method or a pre-venting method.

The detection step S4 is a step in which the humidity sensor 10 detects the humidity of the gas.
Note that the detection step S4 may be before the photographing step S2 or before the discoloration speed photographing step S3 as long as it is after the ventilation step S1.

The humidity sensor 10 (see FIG. 2) is a sensor that detects the humidity of the gas, and can detect the humidity of the gas before the gas is vented into the gas detection tube 2 (when the gas is vented). It is installed in front of the gas detector tube 2.
The humidity sensor 10 is capable of digital output and is connected to the FPGA 7. The detected humidity data is output to the FPGA 7 under measurement conditions. The FPGA 7 that has received the humidity data transmits the humidity data and the color change rate data to the host computer 9.

As such a humidity sensor 10, for example, SHT75 manufactured by Sension can be used.
The humidity sensor 10 is not particularly limited as long as it can perform digital output.

  The gas concentration calculation step S5 is a step of calculating the gas concentration by correcting the color change due to the humidity of the gas from the color change speed and the humidity. Specifically, by inputting the discoloration speed and humidity data (discoloration speed and humidity values) transmitted from the FPGA 7 to the host computer 9 into a neural network (in the host computer 9, gas concentration calculation means), This is a step of calculating the gas concentration.

As described above, the gas concentration is calculated by using the color change rate and the humidity to correct the humidity response (humidity dependency of the color change) of the color change part (see FIG. 5) (correct the color change due to the gas humidity). ) Is based on. The humidity response of the color changing portion means that the color change state varies depending on the humidity of the gas even if the gas has the same concentration.
In the present embodiment, the color change due to the humidity of the gas is corrected using the color change speed and the humidity. However, the correction can be performed using the difference image and the humidity of the image data (whole) of the gas detection tube 2. If it is, it will not be limited to the discoloration speed in particular.

  The humidity response of the discolored portion was specifically examined using the same concentration / different humidity gas produced by the same concentration / different humidity gas production method shown in FIG. The results are shown in FIG. 13 and FIG. From FIG. 13 and FIG. 14, the humidity dependence of the color change rate changes depending on the gas concentration, and there is a complicated relationship among the three (humidity, color change rate, gas concentration), and a simple linear format (regression). (Equation) etc., it was found that it cannot be corrected.

The method for producing the same concentration / different humidity gas shown in FIG. 12 is a method in which the sample solution is completely volatilized in the sampling bag.
As a specific procedure, a sample solution (hydrogen sulfide dissolved in water) is injected into a sampling bag using a microsyringe. And using a flow regulator, humidity is added to dry air, the air is put into a sampling bag, and a sample solution is vaporized. As shown in FIG. 12, the humidity is adjusted by changing the flow rate ratio with the two flow regulators. As the flow regulator, a mass flow controller that can set the flow rate without depending on the load pressure is desirable.
The fine adjustment of the density is performed using a PID (Photo Ionization Detector) detector.

  As described above, the relationship between the three (humidity, color change rate, gas concentration) is complicated without being linearly superimposed, and it is difficult to simply correct it. Therefore, the humidity response of the discoloration part is corrected using a neural network (nonlinear mapping means), and the gas concentration is calculated (estimated).

The neural network used (MLP: Multi-Layer Perception) is standardized by changing the color change rate and gas concentration by 1/1000 times and humidity by 1/100 times on MATLAB as a programming language, and then using the Back Propagation method. It was learned, and the gas concentration was calculated by the Leave-one-out method.
The neural network portion may be another neural network capable of nonlinear mapping, for example, SOM (Self-Organized Map), and uses a method other than the neural network capable of nonlinear mapping, such as LWR (Locally Weighted Regression) method. (Non-linear mapping means).

The input signal of this neural network is humidity and the color change rate (the number of input layers is 2), and the output signal is the gas concentration (the number of output layers is 1).
When the number of intermediate layers, the number of learning times, and the learning constant are changed as parameters, the number of intermediate layers is 8, the number of learning times is 20000, and the learning constant is 0.15. The sum of squares of errors between the estimated gas concentration and the true gas concentration at 14 points (see FIGS. 13 and 14) was the minimum value (see FIG. 15).
From FIG. 15, it can be seen that the estimated gas concentration substantially matches the true gas concentration.

  By learning the neural network in this way and then inputting the color change rate and the humidity, the gas concentration can be automatically (correctly) accurately calculated.

  As described above, when calculating the gas concentration, the humidity dependency of the color change of the color change portion was corrected by using the color change speed and the humidity.

  As described above, by performing the ventilation step S1 to the gas concentration calculation step S5, an accurate gas concentration can be calculated without being influenced by the humidity of the gas (without depending on the humidity).

1 Gas concentration measuring device 2 Gas detector tube 3 Detector tube holder (housing)
4 Pump 5 Illumination means 6 Camera module 7 FPGA
8 SRAM
9 Host computer 10 Humidity sensor S1 Ventilation process S2 Photography process S3 Discoloration speed calculation process S4 Detection process S5 Gas concentration calculation process

Claims (16)

A gas concentration measurement method using a gas detector tube,
A venting step for venting gas through the gas detection tube;
A photographing step of photographing the gas detection tube through which the gas has passed with a camera;
Analyzing the image data of the gas detector tube photographed by the camera, and calculating a color change rate by the gas of the gas detector tube,
A detection step of detecting the humidity of the gas by a humidity sensor;
A gas concentration calculation step of calculating the concentration of the gas by correcting the color change due to the humidity of the gas from the color change speed and the humidity;
A gas concentration measurement method comprising: The gas detector tube comprises a plurality of gas detector tubes,
2. The gas concentration measuring method according to claim 1, wherein the discoloration rate calculating step extracts and analyzes image data of each gas detector tube from the image data of the gas detector tube.   The gas concentration measuring method according to claim 1 or 2, wherein the gas concentration calculating step calculates the concentration of the gas by using a non-linear mapping means.   4. The gas concentration measuring method according to claim 3, wherein the nonlinear mapping means is a neural network.   5. The color change rate calculating step calculates the color change rate from a relationship between a color change area obtained using a division quadrature method and time. 6. The gas concentration measuring method described.   6. The discoloration speed calculating step calculates the discoloration area by correcting a discoloration due to humidity of the gas in an undiscolored portion that has not discolored by the gas of the gas detection tube. The gas concentration measuring method as described in 2. The discoloration speed calculating step includes a difference between the image data of the gas detector tube imaged in advance before starting the measurement and the image data of the gas detector tube imaged in the imaging step of a predetermined threshold value or less. The color change area is calculated using the corrected image data by correcting the color change due to the humidity of the gas in the non-change color part by setting the pixel value to 0. 7. The color change area according to claim 6, wherein the color change area is calculated using the corrected image data. Gas concentration measurement method. 7. The gas concentration measuring method according to claim 6, wherein before the measurement is started, the atmosphere is preliminarily ventilated to saturate the humidity response of the gas detector tube . A gas detector tube;
A pump for venting gas through the gas detector tube;
A camera for photographing the gas detection tube through which the gas has passed;
Analyzing the image data of the gas detector tube photographed by the camera, a color change rate calculating means for calculating the color change rate of the gas detector tube due to the gas,
A humidity sensor for detecting the humidity of the gas;
Gas concentration calculating means for correcting the color change due to the humidity of the gas from the color change speed and the humidity and calculating the concentration of the gas;
A gas concentration measuring device comprising: The gas detector tube comprises a plurality of gas detector tubes,
10. The gas concentration measuring apparatus according to claim 9, wherein the discoloration speed calculating means cuts out and analyzes the image data of each gas detector tube from the image data of the gas detector tube.   The gas concentration measuring device according to claim 9 or 10, wherein the gas concentration calculating means is a nonlinear mapping means.   12. The gas concentration measuring apparatus according to claim 11, wherein the non-linear mapping means is a neural network.   The color change rate calculating means calculates the color change rate from a relationship between a color change area obtained by using a division quadrature method and time, according to any one of claims 9 to 12. The gas concentration measuring device described.   14. The discoloration speed calculating means calculates the discoloration area by correcting discoloration due to humidity of the gas in an undiscolored portion that has not discolored by the gas of the gas detection tube. The gas concentration measuring device according to 1. The discoloration speed calculating means is configured to obtain a difference between the image data of the gas detector tube imaged in advance before starting the measurement and the image data of the gas detector tube imaged in the imaging process, below a predetermined threshold value. 15. The color change area is calculated using the corrected image data by correcting the color change due to the humidity of the gas in the non-change color portion by setting the pixel value to 0. Gas concentration measuring device. 15. The gas concentration measuring apparatus according to claim 14, wherein the discoloration rate calculating means saturates the humidity response of the gas detection tube by ventilating the gas detection tube in advance before starting the measurement. .

Images