JPH0820392B2 - Multiple gas component concentration determination method - Google Patents

Description

TECHNICAL FIELD The present invention relates to a method capable of simultaneously and continuously quantifying the concentrations of a plurality of gas components in a gas body composed of a plurality of gas components.

The present invention is used for analysis of components in exhaust gas or gas components in a closed system in various manufacturing industries.

[Prior Art] In various manufacturing industries, analysis of components in exhaust gas is important in terms of manufacturing process control or environmental problems. on the other hand,
Sample gas sampling is required for gas analysis using general analytical equipment, and a large amount of sample gas cannot be introduced into the detector, such as with gas chromatographs and ICP emission spectroscopy analyzers that are mainly used as gas analyzers. There are problems such as long time. Furthermore, in order to perform “in-situ” analysis, simple, small size, and maintenance-free are required. Therefore, using these analyzers, multiple gas components can be quickly and easily in a gas distribution path such as an exhaust gas path. It is difficult to quantify continuously and simultaneously.

The semiconductor gas sensor is a sensor that uses a metal oxide sintered body such as tin oxide as a material, and detects a gas body by utilizing a change in electric conductivity due to adsorption and desorption of gas generated on the surface thereof. Its applications have been limited to level monitors such as detectors for gas alarms, but the reason for this is that it is extremely difficult to provide component selectivity and that the gas concentration-sensor response characteristics have a non-linear relationship. It is a thing.
Therefore, it is rarely used for quantitative analysis, especially for quantitative analysis of multiple components, and few technologies have been put to practical use. As a study on the use for quantitative analysis, an example of simultaneous determination of multiple gas components using multiple semiconductor gas sensors (7th Chemical Sensor Research Conference 177,1988 or Sensor Technology vol.7, No. 5,1987). In all of these, the sensor response (sensor resistance) to multiple gas components is considered to be represented by the parallel sum or parallel + series sum of the sensor resistances for a single gas component, and this is formulated,
Solving as a system of equations quantifies multiple gas components. However, since the semiconductor gas sensor has a non-linear relationship between the gas concentration-sensor response (sensor resistance) characteristics, the derived relational expression becomes complicated. Solving this as a simultaneous equation requires a long time even if a computer is used. It costs. Furthermore, in the computer-based solution, the value when the calculation error is less than or equal to a certain set value is adopted as the quantitative value, so in order to improve the quantitative accuracy, it is necessary to give fairly strict setting conditions. Also causes a long calculation time. Furthermore, in these methods, it is difficult to construct a sensor response model in consideration of changes in the sensor response itself by detecting the interaction between gas components or a plurality of components, and these actions are remarkable. In this case, the quantification accuracy may be deteriorated.

[Problems to be solved by the invention] Rapid analysis of components in exhaust gas in various manufacturing industries,
Simultaneous quantification of multiple components is essential along with continuity. However, it is difficult for the conventional gas analyzer to introduce the gas directly into the detector from the gas distribution path, and sampling is required. Furthermore, the analyzer cannot be installed in a bad environment with a lot of dust, vibration, etc., which is not suitable for “in-situ” analysis. Therefore, the conventional gas analyzer has problems in terms of quickness and continuity in terms of simultaneous quantification of a plurality of gas components in the gas distribution path. One law,
The semiconductor gas sensor can perform rapid and continuous analysis with a single gas component, but it has been difficult to simultaneously quantify a plurality of gas components.

Therefore, the present invention provides a method capable of rapidly and continuously quantifying a plurality of gas components in a gas body composed of a plurality of gas components.

[Means for Solving the Problems] In order to solve the above problems, the present inventors have found that the following method is effective. That is, in the distribution path of the gas body composed of a plurality of gas components,
A gas composed of m kinds of components to be quantified in advance is provided with an output signal obtained by arranging different kinds of semiconductor gas sensors whose number is equal to or larger than the number of gas components to be quantified and each sensor detecting a plurality of gas components in the gas body. For the body, N samples with different component mixture ratios were measured by n kinds of sensors satisfying the condition of m <n, and output signals (R _i1 , R _i2 , ..., R _in ) of each sensor (i = 1,2, ..., N) (However, C ₁ , ..., C _k , ..., C _m (k = 1,2, ..., m) are the concentrations of the 1, ..., k, ..., m-th gas components in the gas body, R _{1 respectively.} ,
R ₂ , ..., R _n are the values (C _i1 ′,…, C _ik ′,) obtained by substituting the output signals of the 1, 2,…, n-th sensor respectively.
…, C _im ′) (i = 1,2,…, N) and its sample value (C _i1 ,…,
The partial regression coefficient b _kj (k = 1,2, ..., m; j = 0,1,) by the multiple regression analysis method so that the sum of squares of the differences with C _ik , ..., C _im ) is minimized.
It is a method of simultaneously quantifying a plurality of gas component concentrations by applying the values of 2, ..., N) to the above formula.

[Operation] Hereinafter, the present invention will be specifically described.

In the present invention, before application to the gas body to be analyzed, the correlation between the gas concentration and the sensor response is created by the multiple regression analysis method for the gas body composed of a plurality of gas components to be quantified in advance. According to the multiple regression analysis method, the concentration of each component in the gas body consisting of m components is C _k (k = 1,2, ..., m)
And the sensor response of n types of sensors when measuring this gas body is y _j (j = 1,2, ..., n), the sensor response y _j
On the other hand, if the component concentration C _k is additive as in equation (1), the component concentration C _k in the gas body can be determined as in equation (2) by solving this simultaneous equation. You can

That is, if the relational expression (2) can be created for a gas body composed of a plurality of gas components to be quantified in advance before being applied to the gas body to be analyzed, the plurality of gas components in the gas body to be analyzed Sensor response obtained by detecting
By applying y _j to the equation (2), the component concentration C _k can be quickly and simultaneously quantified.

Here, the partial regression coefficient b _kj in the multiple regression equation (2) can be all determined by the following method. Now, regarding a gas body consisting of m kinds of components, the mixture ratio of the components is different N
Let n _i , y _i2 , ..., Y _in (i = 1, 2, ..., N) be sensor responses of n types of sensors when measuring one sample. For the k-th gas component, y _i1 , y _i2 , ..., y _in are substituted into the equation (2), the value obtained is C _ik ′, and the sampled value is C _ik ′.
_{If ik} , the partial regression coefficient b _kj can be determined by minimizing the sum of squares of the residuals e _ik (= C _ik −C _ki ′) of both. That is, E (b _k0 , b _k1 , ..., Of equation (3)
For b _kn ), a simultaneous equation (normal equation) (4) obtained by finding ∂E / ∂b _kj = 0 (j = 0,1,2, ..., n)
By solving for b _kj, b _kj is determined all.

On the other hand, the correlation between the gas concentration C of the semiconductor gas sensor and the sensor response (sensor resistance) is generally expressed by the following equation (5).

^{R = a'C -b 'log R =} a + b logC ... (5) (5) formula is a model for a single gas component, a gas mixture consisting of m kinds of components as an extension of the model (concentration C _1, Consider the following equation (6) as a sensor response model of the semiconductor gas sensor for C ₂ , ..., C _m ).

log R = a ₀ + a ₁ logC ₁ + a ₂ logC ₂ + ... + am _m logC _m (6) That is, the relational expression of formula (6) is created for all semiconductor gas sensors of which the number is different from the number of gas components to be quantified. And the partial regression coefficient (a ₀ , a ₁ , a ₂ ,
, _Am ) and solving them as simultaneous equations, it is possible to obtain the number of relational expressions of the equation (7) corresponding to the equation (2) corresponding to the number of components to be quantified.

When the relational expression of the equation (7) is obtained, the respective sensor responses (R ₁ , R ₂ , ..., R _n ) obtained by detecting the gas body to be analyzed are applied to the equation (7). By doing so, it is possible to quickly and simultaneously quantify a plurality of gas components in the gas body.

Here, regarding the semiconductor gas sensor, the relation of the formula (6) is established very well, so the relational formula of the formula (7) can be derived. However, in order to eliminate the error in the calculation process, the gas concentration C-sensor It is better to directly obtain the equation (7) by the multiple regression analysis method from the measurement result of the response (sensor resistance R) characteristic. However, this is based on the premise that Equation (6) is extremely well established.

Although the formulas (6) and (7) use the sensor resistance value R of the semiconductor gas sensor as the sensor response, even if the relational formula is created by using the voltage applied to the load resistance arranged in series with the sensor instead of the sensor resistance. Good. This is because there is a one-to-one correspondence between the voltage applied to the sensor resistance and the voltage applied to the load resistance, and the conversion between the two can be easily performed.

In the method of the present invention, since the gas concentration-sensor response characteristics for a plurality of gas components are directly obtained by using a gas body composed of a plurality of gas components, it is different from the conventional expansion from a single gas component. Differently, the relational expression is created in consideration of the interaction between the gas components or the change in the sensor response itself due to the detection of a plurality of components. Therefore, even if these effects are remarkable, the effects can be ignored in the method of the present invention.

[Examples] Examples will be described below.

Eight semiconductor gas sensors of different types were used to quantify H ₂ and CO in a gas body containing an arbitrary concentration of H ₂ 30 to 300 ppm and CO 350 to 3500 ppm in Ar gas. The mixed gas (25 levels) in the above concentration range was supplied to the sensor, and the result of examining the relationship of the equation (6) as the correlation between the gas concentration and the sensor response is shown in Table 1. Table 1 shows the partial regression coefficient, multiple correlation coefficient and partial correlation coefficient obtained for the eight semiconductor gas sensors used.

The multiple correlation coefficient for each sensor is 0.97 or more, and between the gas concentrations C _{H 2} , C _CO and the sensor response R,
It was confirmed that the relationship of equation (6) was established very well. Based on these results, Table 2 shows the results of the multiple regression equation corresponding to equation (7) for H ₂ and CO. Table 2 shows the partial regression coefficient in the multiple regression equation obtained for H ₂ and CO, and its multiple correlation coefficient.

25 levels of H ₂ , performed using the relations obtained here,
The quantitative results of CO are shown in Table 3. H ₂ and CO quantification accuracy is maximum error 3.50% and 16.60%, average error 1.00
% And 5.91%. Table 3 also shows the error when quantifying H ₂ and CO of the same concentration by the conventional method, but it can be seen that the quantification accuracy is significantly improved by using the method of the present invention.

Similarly, using 8 kinds of semiconductor gas sensors, H ₂ 20 to 300 ppm, CO 230 to 3300 ppm, CH ₄ 50 to 450 ppm in Ar gas.
As a result of quantifying H ₂ , CO, and CH _{4 in} a gas body containing any concentration within the range, it was possible to quantify each with an average error of 10% or less.

Regarding the analysis time of the above example, it takes several hours to measure the gas concentration-sensor response characteristics for obtaining the multiple regression equation corresponding to the equation (7), but once the relational expression of the equation (7) was obtained. Then, the time required for the subsequent quantification of the multiple components in the gas body is several seconds.

That is, if the calculation formula (7) is created at the laboratory stage before installing the analyzer using the method of the present invention in the gas flow passage such as the exhaust gas passage, it is applied to the gas body to be analyzed. In this case, the analysis time for quantification is almost negligible.

In addition, as the amount of data for creating equation (7),
In order to improve the quantification accuracy, it is desirable to use data that is three times or more the number of sensors used.

[Advantages of the Invention] By the method of the present invention, it becomes possible to quickly and continuously simultaneously quantify the concentrations of a plurality of gas components in a gas body composed of a plurality of gas components. The effect of contributing to a manufacturing process that requires continuous monitoring of the concentrations of multiple gas components is great.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Shigehiko Kanaya 1-1 Hibarigaoka, Tenhaku-cho, Toyohashi-shi, Aichi Within Toyohashi University of Technology (56) Reference JP-A-57-66347 (JP, A) JP Sho 60-67850 (JP, A) Actually opened Sho 60-188368 (JP, U)

Claims (1)

[Claims] 1. A semiconductor gas sensor having different types of gas components to be quantified is arranged in a flow path of a gas body composed of a plurality of gas components, and each sensor detects a plurality of gas components in the gas body. The output signal obtained by detecting the gas signal composed of m kinds of components to be quantified in advance,
N samples having different component mixture ratios were measured by n kinds of sensors satisfying the condition of m <n, and output signals (R _i1 , R _i2 , ..., R _in ) of each sensor (i = 1,2 , ..., N) (However, C ₁ , ..., C _k , ..., C _m (k = 1,2, ..., m) are the concentrations of the 1, ..., k, ..., m-th gas components in the gas body, R _{1 respectively.} ,
R ₂ , ..., R _n are the values (C _i1 ′,…, C _ik ′,) obtained by substituting the output signals of the 1, 2,…, n-th sensor respectively.
…, C _im ′) (i = 1,2,…, N) and its sample value (C _i1 ,…,
Partial regression coefficient b _kj (k = 1,2, ..., m; j = 0,1,) by the multiple regression analysis method so that the sum of squares of the differences with C _ik , ..., C _im ) is minimized.
A method for quantitatively determining a concentration of a plurality of gas components, characterized in that the concentrations of a plurality of gas components are simultaneously quantified by applying the values of 2, ..., N) to the above formula.