An industrial waste gas detection method and system based on gas chromatography

Abstract

This invention discloses a method and system for detecting industrial waste gas based on gas chromatography. The method includes the following steps: S1, pre-constructing an environmental temperature difference chromatogram calibration model, which calibrates the chromatogram obtained by the gas chromatograph based on the temperature difference between the actual test environment and the standard test environment; S2, detecting the industrial waste gas to be detected using a gas chromatograph, obtaining the original chromatogram, and then calibrating it using the environmental temperature difference chromatogram calibration model to obtain the calibrated chromatogram; S3, performing peak finding processing on the calibrated chromatogram using a rapid peak finding model to obtain all chromatographic peaks; S4, analyzing the chromatographic peaks obtained in step S3 to obtain the components in the industrial waste gas to be detected and the content of each component. This invention can achieve rapid and accurate detection of industrial waste gas and can significantly reduce the influence of environmental temperature on the test results, showing great application potential.

Description

Technical Field

[0001] This invention relates to the field of gas detection, and in particular to a method and system for detecting industrial waste gas based on gas chromatography. Background Technology

[0002] The current method for detecting industrial waste gas is to take samples of the waste gas at regular intervals or to directly pass it into a detection device for testing, in order to determine whether the pollutant content of the waste gas meets the emission standards, and then determine whether it can be discharged based on the emission standards.

[0003] Gas chromatography is widely used in gas detection due to its high accuracy. Its measurement principle is that if a certain component is present in the gas entering the instrument, it will produce a peak at a corresponding time point. The concentration of the component can be calculated by the peak area, thus achieving qualitative and quantitative analysis. Examples include patents CN112798703B (an industrial waste gas detection device with remote control function), CN113640439A (a gas detection device and method based on gas chromatography technology), and CN119246711A (a gas detection method for public places based on gas chromatography).

[0004] The chromatographic column is a key component in gas chromatography and is significantly affected by changes in ambient temperature. Different ambient temperatures lead to variations in the retention times of substances. These retention time deviations are directly reflected in the obtained chromatogram, thus affecting the results of qualitative and quantitative analyses using the chromatogram. Furthermore, in quantitative analysis, a standard curve characterizing the relationship between peak area and component concentration is typically constructed at a standard ambient temperature. Using this standard curve for quantitative calculations in non-standard environments can further amplify the influence of ambient temperature on the quantitative analysis results. Adjusting the ambient temperature to a standard temperature before measurement can avoid these effects, but in many cases, this presents challenges such as high implementation difficulty or cost (e.g., using temperature control equipment) or inconvenience (e.g., during on-site testing). For example, patent CN115950988A discloses a chromatographic analysis method based on peak-finding technology, which provides a solution to reduce the impact of ambient temperature changes on testing; however, this method is relatively complex and primarily focuses on peak location. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide an industrial waste gas detection method and system based on gas chromatography, which addresses the shortcomings of the prior art.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: In its first aspect, the present invention provides a method for detecting industrial waste gas based on gas chromatography, comprising the following steps:

[0007] S1. A pre-constructed environmental temperature difference chromatogram correction model is used to correct the chromatograms obtained by the gas chromatograph based on the temperature difference between the actual test environment and the standard test environment.

[0008] S2. The industrial waste gas to be tested is detected by gas chromatograph. After obtaining the original chromatogram, it is corrected by the environmental temperature difference chromatogram correction model to obtain the corrected chromatogram.

[0009] S3. Use a rapid peak-finding model to perform peak-finding processing on the calibrated chromatogram to obtain all chromatographic peaks;

[0010] S4. Based on the chromatographic peaks obtained in step S3, the components in the industrial waste gas to be detected and the content of each component are obtained.

[0011] Preferably, the environmental temperature difference chromatogram correction model is constructed using the following method:

[0012] S1-1, Pre-obtain the maximum amount of each waste gas component contained in the industrial waste gas to be detected in the target scenario;

[0013] S1-2, Constructing the training dataset:

[0014] A single-component standard sample containing only one type of exhaust gas component was subjected to gas chromatography detection at standard ambient temperature. Then, gas chromatography detection was performed again after only changing the ambient temperature. A single-component training dataset S1 was constructed using all the obtained chromatograms.

[0015] A multi-component standard sample containing at least two waste gas components was subjected to gas chromatography detection at standard ambient temperature. Then, gas chromatography detection was performed again after only changing the ambient temperature. A multi-component training dataset S2 was constructed using all the obtained chromatograms.

[0016] Gas chromatography was performed on a full-component standard sample containing all exhaust gas components at a standard ambient temperature. Then, gas chromatography was performed again after only changing the ambient temperature. All the obtained chromatograms were used to construct a full-component training dataset S3.

[0017] S1-3, Model Training:

[0018] The CNN convolutional neural network was trained sequentially using a single-component training dataset S1, a multi-component training dataset S2, and a full-component training dataset S3. After training, the environmental temperature difference chromatogram correction model was obtained.

[0019] Preferably, step S1-2 specifically includes:

[0020] S1-2-1, Constructing a single-component training dataset S1:

[0021] At standard ambient temperature, a single-component standard sample containing only one type of waste gas component is obtained using a gas chromatograph for analysis, and the resulting chromatogram is recorded as the standard single-component chromatogram. Each waste gas component is tested individually, and the standard single-component chromatogram of each single-component standard sample is obtained separately.

[0022] For any single-component standard sample, changing only the ambient temperature yields chromatograms at different actual ambient temperatures, denoted as the measured chromatogram PC. BΔT Let ΔT = actual ambient temperature - standard ambient temperature; for any single-component standard sample C B ΔT, ΔT below C B Corresponding measured chromatogram PC BΔT And C B Standard single-component chromatogram PC B Combine them into a single-component training data s1, and combine all the obtained single-component training data s1 to construct a single-component training dataset S1.

[0023] S1-2-2, Constructing a multi-component training dataset S2:

[0024] At standard ambient temperature, a multi-component standard sample containing at least two waste gas components was obtained using a gas chromatograph for analysis, and the resulting chromatogram was recorded as a standard multi-component chromatogram. Then, the types and quantities of waste gas components in the standard sample were changed, and the samples were tested separately to obtain a standard multi-component chromatogram for each type of multi-component standard sample.

[0025] For any multi-component standard sample, changing only the ambient temperature yields chromatograms at different actual ambient temperatures, denoted as the measured chromatogram PCd. BΔT Let ΔT = actual ambient temperature - standard ambient temperature; for any multi-component standard sample Cd B ΔT, ΔT under Cd B Measured chromatogram of PCd BΔT and Cd B Standard multicomponent chromatogram PCd B Combine them into a single multi-component training data s2, and combine all the obtained multi-component training data s2 to construct a multi-component training dataset S2.

[0026] S1-2-3, Construct the full-component training dataset S3:

[0027] At standard ambient temperature, a full-component standard sample containing all waste gas components was obtained using a gas chromatograph and measured to obtain a chromatogram, which was recorded as the standard full-component chromatogram. The concentration ratio of the waste gas components in the full-component standard sample was changed only to obtain the standard full-component chromatogram of the full-component standard sample with each different concentration ratio.

[0028] For any full-component standard sample, changing only the ambient temperature yields chromatograms at different actual ambient temperatures, denoted as the measured chromatogram PCa. BΔT Let ΔT = actual ambient temperature - standard ambient temperature; for any full-component standard sample Ca B ΔT, ΔT under Ca B Measured chromatogram of PCa BΔT and Ca B Standard full-component chromatogram PCa B Combine them into a single full-component training data s3, and combine all the obtained full-component training data s3 to construct the full-component training dataset S3.

[0029] The standard ambient temperature is 20–25℃.

[0030] Preferably, steps S1-3 specifically include:

[0031] S1-3-1, Using a single-component training dataset S1, with ΔT and PC BΔT For input, PC B To output the CNN convolutional neural network, a first-order correction model is obtained after training.

[0032] S1-3-2, Using a multi-component training dataset S2, with ΔT and PCd BΔT For input, PCd B To output the first-order correction model, a second-order correction model is obtained after training.

[0033] S1-3-3, using the full-component training dataset S3, with ΔT, PCa BΔT For input, PCa B The second-order calibration model is trained to output the final environmental temperature difference chromatogram calibration model.

[0034] Preferably, the method for peak finding processing using the fast peak finding model includes the following steps:

[0035] S3-1. Obtain all peak vertices in the corrected chromatogram. The set of all peak vertices is denoted as Q. p ;

[0036] S3-2, Obtain Q p The starting point of the peak corresponding to each peak vertex;

[0037] S3-3, Obtain Q p The peak endpoint corresponding to each peak vertex in the middle;

[0038] S3-4. Based on the obtained peak apex, the peak start point and peak end point corresponding to each peak apex, obtain all chromatographic peaks and index them on the chromatographic curve of the corrected chromatogram, thereby obtaining the processed chromatogram with indexed chromatographic peaks.

[0039] Preferably, the method for peak finding processing using the fast peak finding model includes the following steps:

[0040] S3-1, Obtaining the peak vertex:

[0041] S3-1-1. On the chromatographic curve of the calibrated chromatogram, starting from the origin and moving from left to right, take J consecutive data points as a data unit. For the last data unit, take the actual remaining data points as a data unit. Let the number of data points in the last data unit be J' and the total number of data units be m. Then J'≤J, and the total number of data points N=m*(J-1)+J'.

[0042] S3-1-2, Simultaneously, the following steps are used to obtain the peak vertices for all data units:

[0043] For a given data cell, take any data point P in that data cell excluding the left and right endpoints. h If the following conditions a and b are met, then the data point P is... h As the peak:

[0044] a. Data point P h The signal strength is simultaneously greater than the signal strength of the first data point to its left and the first data point to its right;

[0045] b. From data point P h There are at least n consecutive n directions to the left T The signal strength of each data point decreases sequentially, starting from data point P. h There are at least n consecutive n directions to the right T The signal strength of each data point decreases sequentially.

[0046] Traverse all data points in the data unit except for the left and right endpoints, and construct a set Q of all the resulting peak vertices. P1 ;

[0047] S3-1-3. On the chromatogram of the corrected chromatogram, for all data points at the left and right endpoints of all data units, if conditions a and b above are satisfied simultaneously, the current data point is taken as the peak vertex, and all the obtained peak vertices are constructed into a peak vertex set Q. P2 ;

[0048] S3-1-4, Q P1 and Q P2 By combining the peaks, we obtain the set Q of all peak vertices in the corrected chromatogram.p Q p =Q P1 +Q P2 ;

[0049] S3-2, Obtain the set of peak vertices Q p The starting point of each peak corresponding to the apex of the middle peak:

[0050] For Q p Peak P in n From P n Calculate the slope of the line connecting any two adjacent data points to the left, and denote it as: P n and P n-1 The slope between them is k n P n-1 and P n-2 The slope between them is k n-1 ,...,P n-(m-1) and P n-m The slope between them is k n-(m-1) P n-m and P n-(m+1) The slope between them is k n-m ;

[0051] When the first data point that satisfies either condition c or d appears, that data point P is selected. n-m As the peak P n The corresponding peak starting point:

[0052] c, k n-(m-1) >0, and k n-m ≤0

[0053] d, k n-(m-1) >0, k n-m >0, and k n-(m-1) -k n-m ≤Δk T1 ;Δk T1 The preset slope change threshold is a constant that is not less than 0;

[0054] S3-3, Obtain the set of peak vertices Q p The peak endpoint of each peak corresponding to the apex of the middle peak:

[0055] For Q p Peak P in n From P n Calculate the slope of the line connecting any two adjacent data points to the right, and denote it as: P n and P n+1 The slope between them is k n+1 P n+1 and P n+2 The slope between them is k n+2,...,P n+(m'-1) and P n+m' The slope between them is k n+m' P n+m' and P n+(m'+1) The slope between them is k n+(m'+1) ;

[0056] When the first data point that satisfies either condition e or f appears, then that data point P is selected. n+m' As the peak P n The corresponding peak endpoint:

[0057] e, k n+m' <0, and k n+(m'+1) ≥0;

[0058] f、k n+m' <0,k n+(m'+1) <0, and |k n+m' |-|k n+(m'+1) |≤Δk T2 Δk T2 The preset slope change threshold is a constant that is not less than 0;

[0059] S3-4. Based on the obtained peak apex, the peak start point and peak end point corresponding to each peak apex, obtain all chromatographic peaks and index them on the chromatographic curve of the corrected chromatogram, thereby obtaining the processed chromatogram containing chromatographic peaks.

[0060] Preferably, where J = 5-50, n T =2-5.

[0061] Preferably, where 0 ≤ Δk T1 ≤0.15, 0≤Δk T2 ≤0.15.

[0062] Preferably, step S4 specifically includes:

[0063] S4-1. Compare the processed chromatogram with the standard full-component chromatogram to obtain the component corresponding to each chromatographic peak in the processed chromatogram:

[0064] For any chromatographic peak R in the processed chromatogram i Record the chromatographic peak R i The peak is P i , obtain P i x-coordinate t i In the standard full-component chromatogram, the horizontal axis range is (t i -0.5d min )~(t i +0.5d min Within the range, obtain the x-coordinate t of the peak.Bi With t i The peak P corresponding to the minimum absolute value of the difference Bi , will P Bi The component C corresponding to the peak i As chromatographic peak R i Corresponding ingredients;

[0065] When the standard full-component chromatogram is in the range of (t) on the horizontal axis... i -0.5d min )~(t i +0.5d min No chromatographic peak was found within the specified range; the chromatographic peak R was manually confirmed. i The corresponding components; where d min It is the minimum distance between the peak vertices of two adjacent chromatographic peaks on the horizontal axis in a standard full-component chromatogram;

[0066] S4-2. Based on the area of ​​the chromatographic peaks in the processed chromatogram, and combined with the standard curve of the relationship between the chromatographic peak area and component concentration in the pre-constructed characterization standard full component chromatogram, the concentration of the component corresponding to each chromatographic peak is calculated.

[0067] A second aspect of the present invention provides an industrial waste gas detection system based on gas chromatography, which performs industrial waste gas detection using the method described above. The system includes:

[0068] A gas chromatograph is used to detect industrial waste gas and obtain raw chromatograms.

[0069] The environmental temperature difference chromatogram correction model obtains the corrected chromatogram using the method in step S2.

[0070] The rapid peak-finding model uses the method in step S3 to perform peak-finding processing on the calibrated chromatogram to obtain all chromatographic peaks.

[0071] And a chromatographic peak analysis module, which uses the method in step S4 to analyze the obtained chromatographic peaks to obtain the components in the industrial waste gas to be detected and the content of each component.

[0072] The beneficial effects of this invention are:

[0073] This invention provides a gas chromatography-based method for detecting industrial waste gas, which can achieve rapid and accurate detection of industrial waste gas and significantly reduce the influence of ambient temperature on test results, and has good application prospects.

[0074] This invention utilizes an environmental temperature difference chromatogram correction model built based on machine learning algorithms to correct chromatograms collected under non-standard environmental temperatures to those collected under standard temperatures. This reduces analytical errors caused by retention time shifts due to temperature changes, and also reduces errors from using standard curves built under standard temperatures for quantitative analysis under non-standard temperatures. This improves the accuracy of subsequent qualitative and quantitative analyses based on chromatograms.

[0075] The rapid peak-finding model of this invention can quickly obtain the peak apex, peak start point, and peak end point, achieving rapid and accurate peak finding. After peak finding, the processed chromatogram containing the chromatographic peaks is compared with the standard full-component chromatogram obtained during the construction of the environmental temperature difference chromatogram correction model. This allows for the rapid acquisition of the components and concentrations corresponding to each chromatographic peak in the processed chromatogram, ultimately achieving rapid qualitative and quantitative analysis of industrial waste gas. Attached Figure Description

[0076] Figure 1 This is a flowchart of the industrial waste gas detection method based on gas chromatography of the present invention;

[0077] Figure 2 This is a flowchart illustrating the method for constructing the environmental temperature difference chromatogram correction model of the present invention;

[0078] Figure 3 This is a detailed flowchart of steps S1-2 of the present invention;

[0079] Figure 4 This is a detailed flowchart of steps S1-3 of the present invention;

[0080] Figure 5 A flowchart of a method for performing peak finding processing on the fast peak finding model of the present invention;

[0081] Figure 6 This is a flowchart illustrating step S4 of the present invention.

[0082] Figure 7 This is the standard chromatogram P1 obtained in the test example of the present invention at an ambient temperature of 25°C;

[0083] Figure 8 This is the measured chromatogram P2 obtained in the test example of the present invention at an ambient temperature of 5°C;

[0084] Figure 9 The corrected chromatogram P3 obtained in the test example of this invention;

[0085] Figure 10 S is a test example of the present invention. P1 A schematic diagram of the area represented;

[0086] Figure 11 This is a schematic diagram of the region represented by ΔS in the test example of the present invention;

[0087] Figure 12 This is a schematic diagram of the industrial waste gas detection system based on gas chromatography in Embodiment 2 of the present invention. Detailed Implementation

[0088] The present invention will be further described in detail below with reference to embodiments, so that those skilled in the art can implement it based on the description.

[0089] It should be understood that terms such as “having,” “comprising,” and “including” as used herein do not exclude the presence or addition of one or more other elements or combinations thereof.

[0090] Example 1

[0091] A method for detecting industrial waste gas based on gas chromatography includes the following steps:

[0092] S1. A pre-constructed environmental temperature difference chromatogram correction model is used to correct the chromatograms obtained by the gas chromatograph based on the temperature difference between the actual test environment and the standard test environment.

[0093] S2. The industrial waste gas to be tested is detected by gas chromatograph. After obtaining the original chromatogram, it is corrected by the environmental temperature difference chromatogram correction model to obtain the corrected chromatogram.

[0094] S3. Use a rapid peak-finding model to perform peak-finding processing on the calibrated chromatogram to obtain all chromatographic peaks;

[0095] S4. Based on the chromatographic peaks obtained in step S3, the components in the industrial waste gas to be detected and the content of each component are obtained.

[0096] In this embodiment, the environmental temperature difference chromatogram correction model was constructed using the following method:

[0097] S1-1. Pre-obtain the maximum concentration of each waste gas component in the industrial waste gas to be detected under the target scenario; that is, for the main application target scenarios, pre-obtain all possible components contained in the industrial waste gas generated under these target scenarios manually, and these components can be denoted as C1, C2, ..., C... L Any one of the waste gas components can be denoted as C. i Therefore, the data used to construct the environmental temperature difference chromatogram calibration model must be determined based on the actual detection target.

[0098] S1-2, Constructing the training dataset:

[0099] S1-2-1, Constructing a single-component training dataset S1:

[0100] At standard ambient temperature, a single-component standard sample containing only one type of waste gas component is obtained using a gas chromatograph for analysis, and the resulting chromatogram is recorded as the standard single-component chromatogram. Each waste gas component is tested individually, and the standard single-component chromatogram of each single-component standard sample is obtained separately.

[0101] For any single-component standard sample, changing only the ambient temperature yields chromatograms at different actual ambient temperatures, denoted as the measured chromatogram PC. BΔT Let ΔT = actual ambient temperature - standard ambient temperature; for any single-component standard sample C B ΔT, ΔT below C B Corresponding measured chromatogram PC BΔT And C B Standard single-component chromatogram PC B Combine them into a single-component training data s1, and combine all the obtained single-component training data s1 to construct a single-component training dataset S1.

[0102] In this embodiment, in order to ensure the coverage of training data and improve model accuracy, the single-component standard samples corresponding to each exhaust gas component need to be measured and their chromatograms collected at multiple different temperatures, and then used to construct single-component training data.

[0103] S1-2-2, Constructing a multi-component training dataset S2:

[0104] At standard ambient temperature, a multi-component standard sample containing at least two waste gas components was obtained using a gas chromatograph for analysis, and the resulting chromatogram was recorded as a standard multi-component chromatogram. Then, the types and quantities of waste gas components in the standard sample were changed, and the samples were tested separately to obtain a standard multi-component chromatogram for each type of multi-component standard sample.

[0105] For any multi-component standard sample, changing only the ambient temperature yields chromatograms at different actual ambient temperatures, denoted as the measured chromatogram PCd. BΔT Let ΔT = actual ambient temperature - standard ambient temperature; for any multi-component standard sample Cd B ΔT, ΔT under Cd B Measured chromatogram of PCd BΔT and Cd B Standard multicomponent chromatogram PCd B Combine them into a single multi-component training data s2, and combine all the obtained multi-component training data s2 to construct a multi-component training dataset S2.

[0106] In this embodiment, to ensure the coverage of training data and improve model accuracy, for each waste gas component, additional components are added to form a multi-component standard sample. Chromatographic data are collected at multiple different temperatures to serve as multi-component training data. A more diverse combination of multi-component standard samples results in richer training data. For example, this could include several multi-component standard samples with 2, 3, 4, ..., (L-1) components. In this embodiment, the number of multi-component standard samples used to construct the training data is at least 200*L.

[0107] S1-2-3, Construct the full-component training dataset S3:

[0108] At standard ambient temperature, a full-component standard sample containing all waste gas components was obtained using a gas chromatograph and measured to obtain a chromatogram, which was recorded as the standard full-component chromatogram. The concentration ratio of the waste gas components in the full-component standard sample was changed only to obtain the standard full-component chromatogram of the full-component standard sample with each different concentration ratio.

[0109] For any full-component standard sample, changing only the ambient temperature yields chromatograms at different actual ambient temperatures, denoted as the measured chromatogram PCa. BΔT Let ΔT = actual ambient temperature - standard ambient temperature; for any full-component standard sample Ca B ΔT, ΔT under Ca B Measured chromatogram of PCa BΔT and Ca B Standard full-component chromatogram PCa B Combine them into a single full-component training data s3, and combine all the obtained full-component training data s3 to construct the full-component training dataset S3.

[0110] The standard ambient temperature is 25℃. The actual ambient temperature range used when constructing the training dataset is 0 to 50℃, specifically including 0℃, 1℃, 2℃, 3℃, 4℃, ..., 48℃, 49℃, and 50℃. Data is collected in temperature intervals of 1℃ to provide training data with an appropriate temperature range and rich temperature data, thereby ensuring the accuracy of the model.

[0111] S1-3, Model Training:

[0112] S1-3-1, Using a single-component training dataset S1, with ΔT and PC BΔT For input, PC B To output the CNN convolutional neural network, a first-order correction model is obtained after training.

[0113] S1-3-2, Using a multi-component training dataset S2, with ΔT and PCd BΔT For input, PCd B To output the first-order correction model, a second-order correction model is obtained after training.

[0114] S1-3-3, using the full-component training dataset S3, with ΔT, PCa BΔT For input, PCa B The second-order calibration model is trained to output the final environmental temperature difference chromatogram calibration model.

[0115] Changes in ambient temperature can affect the retention time of substances in a chromatograph. Shifts in retention time are directly reflected in the obtained chromatogram, thus impacting the results of qualitative and quantitative analyses. Furthermore, in quantitative analysis, a standard curve characterizing the relationship between peak area and component concentration is typically constructed at a standard ambient temperature. Using this standard curve for quantitative calculations in non-standard environments can further amplify the influence of ambient temperature on the results. Adjusting the ambient temperature to a standard temperature before measurement can avoid these effects; however, this approach is often difficult or costly to implement (e.g., using temperature control equipment) or inconvenient (e.g., during on-site testing).

[0116] This invention employs a different approach: by using a model built based on machine learning algorithms, chromatograms collected under non-standard ambient temperatures are corrected to those at standard temperatures. This reduces analytical errors caused by temperature variations leading to shifts in substance retention times, and also reduces errors when using standard curves built at standard temperatures for quantitative analysis at non-standard temperatures. This improves the accuracy of subsequent qualitative and quantitative analyses based on chromatograms.

[0117] Furthermore, in the model training process of this invention, a single-component training dataset S1 is first used to train the CNN convolutional neural network, enabling the CNN to learn the transformation relationship of chromatograms under a single exhaust gas component to the standard temperature under different temperatures. Then, a multi-component training dataset S2 is used for training, enabling the initially trained first-order correction model to further learn the transformation relationship of chromatograms under multiple exhaust gas components to the standard temperature under different temperatures. Finally, a full-component training dataset S3 is used for training, enabling the further trained second-order correction model to further learn the transformation relationship of chromatograms containing all exhaust gas components but with different concentrations to the standard temperature under different temperatures. By using training datasets that gradually increase in complexity to train the model in stages, the correction capability and accuracy of the obtained environmental temperature difference chromatogram correction model can be better guaranteed.

[0118] In this embodiment, the method for peak finding processing using the fast peak finding model includes the following steps:

[0119] S3-1, Obtaining the peak vertex:

[0120] S3-1-1. On the chromatographic curve of the calibrated chromatogram, starting from the origin and moving from left to right, take J consecutive data points as a data unit. For the last data unit, take the actual remaining data points as a data unit. Let the number of data points in the last data unit be J' and the total number of data units be m. Then J'≤J, and the total number of data points N=m*(J-1)+J'.

[0121] J can be selected according to the actual situation and needs. When the value of J is appropriate, the accuracy is high without increasing the amount of calculation too much. For example, J = 5-50. In this embodiment, J = 10 is selected.

[0122] S3-1-2, Simultaneously, the following steps are used to obtain the peak vertices for all data units:

[0123] For a given data cell, take any data point P in that data cell excluding the left and right endpoints. h If the following conditions a and b are met, then the data point P is... h As the peak:

[0124] a. Data point P h The signal strength is simultaneously greater than the signal strength of the first data point to its left and the first data point to its right;

[0125] b. From data point P h There are at least n consecutive n directions to the left T The signal strength of each data point decreases sequentially, starting from data point P. h There are at least n consecutive n directions to the right T The signal strength of each data point decreases sequentially.

[0126] If condition a is satisfied, it means that P h The signal strength is higher than both the left and right sides, satisfying b, indicating that P h The left curve shows an upward trend within a certain range, and the right curve shows a downward trend within a certain range. Therefore, if conditions a and b are satisfied simultaneously, P can be considered... h The point is located at the peak; where n T The value n is determined based on the actual situation and requirements. T The larger the value, the higher the recognition accuracy; however, if it is too large, there is a certain risk of missed recognition. Its typical value can be n. T =2-5, in this embodiment, n is selected. T =3.

[0127] Traverse all data points in the data unit except for the left and right endpoints, and construct a set Q of all the resulting peak vertices. P1 ;

[0128] S3-1-3. On the chromatogram of the corrected chromatogram, for all data points at the left and right endpoints of all data units, if conditions a and b above are satisfied simultaneously, the current data point is taken as the peak vertex, and all the obtained peak vertices are constructed into a peak vertex set Q. P2 ;

[0129] S3-1-4, Q P1 and Q P2 By combining the peaks, we obtain the set Q of all peak vertices in the corrected chromatogram. p Q p =Q P1 +Q P2 ;

[0130] S3-2, Obtain the set of peak vertices Q p The starting point of each peak corresponding to the apex of the middle peak:

[0131] For Q p Peak P in n From P n Calculate the slope of the line connecting any two adjacent data points to the left, and denote it as: P n and P n-1 The slope between them is k n P n-1 and P n-2 The slope between them is k n-1 ,...,P n-(m-1) and P n-m The slope between them is k n-(m-1) P n-m and P n-(m+1) The slope between them is k n-m ;

[0132] When the first data point that satisfies either condition c or d appears, that data point P is selected. n-m As the peak P n The corresponding peak starting point:

[0133] c, k n-(m-1) >0, and k n-m ≤0;

[0134] d, k n-(m-1) >0, k n-m >0, and k n-(m-1) -k n-m ≤Δk T1 ;Δk T1The pre-set threshold for slope change is a constant that is not less than 0; for example, 0 ≤ Δk. T1 ≤0.15, in this embodiment, Δk T1 =0.12;

[0135] When condition c is satisfied, it means that from the peak P n To the left, the slope of the curve gradually decreases from a positive value, and from P... n-(m-1) To P n-(m+1) When time changes from positive to negative, it indicates that P n-(m-1) and P n-(m+1) Point P between n-m This is the starting point where the slope increases, that is, the starting point of the peak on the left side of the peak;

[0136] When condition d is satisfied, it means that from the peak P n To the left, the slope of the curve gradually decreases from a positive value, and from P... n-(m-1) To P n-(m+1) At that point, the degree of decrease in the slope can be considered close to zero, so P n-(m-1) and P n-(m+1) Point P between n-m This can be considered the starting point where the slope becomes significantly larger, that is, the peak starting point on the left side of the peak; the above method can quickly and accurately identify the peak starting point.

[0137] S3-3, Obtain the set of peak vertices Q p The peak endpoint of each peak corresponding to the apex of the middle peak:

[0138] For Q p Peak P in n From P n Calculate the slope of the line connecting any two adjacent data points to the right, and denote it as: P n and P n+1 The slope between them is k n+1 P n+1 and P n+2 The slope between them is k n+2 ,...,P n+(m'-1) and P n+m' The slope between them is k n+m' P n+m' and P n+(m'+1) The slope between them is k n+(m'+1) ;

[0139] When the first data point that satisfies either condition e or f appears, then that data point P is selected. n+m' As the peak P n The corresponding peak endpoint:

[0140] e, k n+m' <0, and k n+(m'+1) ≥0;

[0141] f、k n+m' <0,k n+(m'+1) <0, and |k n+m' |-|k n+(m'+1) |≤Δk T2 Δk T2 The pre-set threshold for slope change is a constant that is not less than 0; for example, 0 ≤ Δk. T2 ≤0.15, in this embodiment, Δk T2 =0.12;

[0142] When condition e is satisfied, it indicates that from the peak P n To the right, the slope of the curve gradually increases from negative values, and from P... n+(m'-1) To P n+(m'+1) When the time changes from negative to positive, it indicates that P n+(m'-1) and P n+(m'+1) Point P between n+m' This is the point where the slope increases, which is also the end point of the peak on the right side of the peak.

[0143] When condition d is satisfied, it means that from the peak P n To the right, the slope of the curve gradually increases from negative values, and from P... n+(m'-1) To P n+(m'+1) At that point, the increase in slope can be considered close to zero, so P n+(m'-1) and P n+(m'+1) Point P between n+m' It can be considered as the point where the slope increases, that is, the peak endpoint on the right side of the peak; the above method can quickly and accurately identify the peak endpoint.

[0144] S3-4. Based on the obtained peak apex, the peak start point and peak end point corresponding to each peak apex, obtain all chromatographic peaks and index them on the chromatographic curve of the corrected chromatogram, thereby obtaining the processed chromatogram containing chromatographic peaks.

[0145] In this embodiment, step S4 specifically includes:

[0146] S4-1. Compare the processed chromatogram with the standard full-component chromatogram to obtain the component corresponding to each chromatographic peak in the processed chromatogram:

[0147] For any chromatographic peak R in the processed chromatogram i Record the chromatographic peak R i The peak is P i , obtain P i x-coordinate t i In the standard full-component chromatogram, the horizontal axis range is (t i -0.5d min )~(t i+0.5d min Within the range, obtain the x-coordinate t of the peak. Bi With t i The peak P corresponding to the minimum absolute value of the difference Bi , will P Bi The component C corresponding to the peak i As chromatographic peak R i Corresponding ingredients;

[0148] The elution time of any component in the measured chromatogram (i.e., the horizontal axis of the chromatogram) will be consistent with or close to the elution time of the same component in the standard full-component chromatogram (especially after correction using the environmental temperature difference chromatogram correction model in this invention, it will be even closer). Therefore, for any chromatographic peak to be determined, by comparing the measured chromatogram with the standard full-component chromatogram, and searching within a certain range in the standard full-component chromatogram for the peak whose horizontal axis is closest to that of the peak to be determined, the component of the chromatographic peak corresponding to that peak is the same as that of the chromatographic peak to be determined. Thus, the component corresponding to each chromatographic peak in the processed chromatogram can be obtained quickly.

[0149] When the standard full-component chromatogram is in the range of (t) on the horizontal axis... i -0.5d min )~(t i +0.5d min No chromatographic peak was found within the specified range; the chromatographic peak R was manually confirmed. i The corresponding ingredients.

[0150] Where, d min This represents the minimum distance between the peak vertices of two adjacent chromatographic peaks on the horizontal axis in a standard full-component chromatogram; through d min To determine the search range, locate the horizontal search range (t). i -0.5d min )~(t i +0.5d min This ensures that no two chromatographic peaks will exist simultaneously within this range.

[0151] S4-2. Based on the peak areas of the processed chromatograms, and combined with a pre-constructed standard curve showing the relationship between peak area and component concentration in the full-component chromatograms of the characterizing standard, the concentration of the component corresponding to each peak is calculated. The peak area is positively correlated with the component concentration. Pre-constructing a standard curve and then using peak area analysis to obtain the component concentration is a common method in the field, and this invention does not limit its application; specific methods are not described in this embodiment.

[0152] Test case

[0153] A mixed gas containing benzene, toluene, and styrene was used as the test sample to simulate industrial waste gas, and chromatographic detection was performed using a gas chromatograph (Agilent 7890A).

[0154] 1. Test of the calibration effect of the environmental temperature difference chromatogram calibration model:

[0155] First, detect the standard chromatogram P1 at an ambient temperature of 25℃, such as... Figure 7 As shown in the figure, the chromatographic peaks of benzene, toluene, and styrene are marked ①, ②, and ③ in sequence.

[0156] Then, the measured chromatogram P2 at an ambient temperature of 5℃ was detected, as shown below. Figure 8 As shown; the measured chromatogram P2 was corrected using the environmental temperature difference chromatogram correction model to obtain the corrected chromatogram P3, as shown. Figure 9 As shown.

[0157] Compare the coefficients of difference ε between P1 and P2. Δs And the difference coefficient ε between P1 and P3 Δs :

[0158] Define the difference coefficient ε Δs The calculation formula is:

[0159]

[0160] Where, ε Δs The smaller the value, the smaller the difference; S P1 This represents the area of ​​the enclosed region between the chromatographic curve in the standard chromatogram and the horizontal axis and the vertical lines connecting the left and right endpoints of the chromatographic curve. (Refer to...) Figure 10 ;

[0161] ΔS represents the area of ​​intersection. For the difference coefficients between P1 and P2, when P1 and P2 are plotted on the same chromatogram, ΔS is the area of ​​the closed region formed by the intersection of the chromatographic curves in P1 and P2. (Refer to...) Figure 11 For the difference coefficient between P1 and P3, P1 and P3 are plotted on the same chromatogram, and ΔS is the area of ​​the closed region between the chromatographic curves of P1 and P3.

[0162] For further comparison, the measured chromatogram P2 was corrected using the first-order correction model and the second-order correction model, respectively, to obtain P4 and P5, and the difference coefficients between the corrected chromatograms P4 and P1, and P5 and P1 were calculated.

[0163] The test results are shown in Table 1 below:

[0164] Table 1

[0165] <![CDATA[P1 and P2]]> <![CDATA[P1 and P3]]> <![CDATA[P1 and P4]]> <![CDATA[P1 and P5]]> <![CDATA[Coefficient of difference ε Δs > 1.56％ 0.05％ 0.67％ 0.24％

[0166] The test results show that there is a significant difference between the measured chromatogram P2 and the standard chromatogram P1 after temperature change. When the first-order correction model and the second-order correction model are used for correction, the difference decreases in that order. However, the difference is minimized when the environmental temperature difference chromatogram correction model is used.

[0167] 2. Quantitative analysis results of exhaust gas:

[0168] The test samples were subjected to chromatographic detection at ambient temperatures of 5°C and 15°C, respectively, using the method described in Example 1. As a control, the method in the control example was used as in the example, but without using the ambient temperature difference chromatogram correction model. Instead, the original chromatogram was directly processed using the method in step S3, followed by peak finding, and then analyzed using the method in S4 to calculate the test error coefficient η. C :

[0169]

[0170] Among them, C T C represents the actual concentration measured at different ambient temperatures (5℃, 15℃, 25℃). 标 η represents the actual concentration of the sample in the test sample. C The smaller the value, the higher the accuracy.

[0171] In this case, when the ambient temperature is 25℃, the ambient temperature difference chromatogram correction model is not used for correction. Instead, the original chromatogram is directly processed by the method in step S3, and then the analysis is performed by the method in S4.

[0172] The test results are shown in Table 2 below:

[0173] Table 2

[0174]

[0175] As can be seen from the test results in Table 2, the test accuracy of Example 1 is very high. However, without the use of an environmental temperature difference chromatogram correction model, the test results are greatly affected by temperature, and the test accuracy is significantly reduced.

[0176] Example 2

[0177] An industrial waste gas detection system based on gas chromatography is provided, which uses the method of Example 1 to detect industrial waste gas, referring to... Figure 12 The system includes:

[0178] A filter, used to filter industrial waste gas to be tested;

[0179] A gas chromatograph is used to detect filtered industrial waste gas and obtain raw chromatograms.

[0180] The rapid peak-finding model uses the method in step S3 to perform peak-finding processing on the calibrated chromatogram to obtain all chromatographic peaks.

[0181] And a chromatographic peak analysis module, which uses the method in step S4 to analyze the obtained chromatographic peaks to obtain the components in the industrial waste gas to be detected and the content of each component.

[0182] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0183] Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details.

Claims

1. A method for detecting industrial waste gas based on gas chromatography, characterized in that, Includes the following steps: S1. A pre-constructed environmental temperature difference chromatogram correction model is used to correct the chromatograms obtained by the gas chromatograph based on the temperature difference between the actual test environment and the standard test environment. S2. The industrial waste gas to be tested is detected by gas chromatograph. After obtaining the original chromatogram, it is corrected by the environmental temperature difference chromatogram correction model to obtain the corrected chromatogram. S3. Use a rapid peak-finding model to perform peak-finding processing on the calibrated chromatogram to obtain all chromatographic peaks; S4. Based on the chromatographic peaks obtained in step S3, the components in the industrial waste gas to be detected and the content of each component are obtained. The environmental temperature difference chromatogram correction model was constructed using the following method: S1-1, Pre-obtain the maximum amount of each waste gas component contained in the industrial waste gas to be detected in the target scenario; S1-2, Constructing the training dataset: A single-component standard sample containing only one type of exhaust gas component was subjected to gas chromatography detection at standard ambient temperature. Then, gas chromatography detection was performed again after only changing the ambient temperature. A single-component training dataset S1 was constructed using all the obtained chromatograms. A multi-component standard sample containing at least two waste gas components was subjected to gas chromatography detection at standard ambient temperature. Then, gas chromatography detection was performed again after only changing the ambient temperature. A multi-component training dataset S2 was constructed using all the obtained chromatograms. Gas chromatography was performed on a full-component standard sample containing all exhaust gas components at a standard ambient temperature. Then, gas chromatography was performed again after only changing the ambient temperature. All the obtained chromatograms were used to construct a full-component training dataset S3. S1-3, Model Training: The CNN convolutional neural network was trained sequentially using a single-component training dataset S1, a multi-component training dataset S2, and a full-component training dataset S3. After training, the environmental temperature difference chromatogram correction model was obtained. The method for peak finding processing using the fast peak finding model includes the following steps: S3-1. Obtain all peak vertices in the corrected chromatogram. The set of all peak vertices is denoted as Q. p ; S3-2, Obtain Q p The starting point of the peak corresponding to each peak vertex; S3-3, Obtain Q p The peak endpoint corresponding to each peak vertex in the middle; S3-4. Based on the obtained peak apex, the peak start point and peak end point corresponding to each peak apex, obtain all chromatographic peaks and index them on the chromatographic curve of the corrected chromatogram, thereby obtaining the processed chromatogram with indexed chromatographic peaks.

2. The industrial waste gas detection method based on gas chromatography according to claim 1, characterized in that, Step S1-2 specifically includes: S1-2-1, Constructing a single-component training dataset S1: At standard ambient temperature, a single-component standard sample containing only one type of waste gas component is obtained using a gas chromatograph for analysis, and the resulting chromatogram is recorded as the standard single-component chromatogram. Each waste gas component is tested individually, and the standard single-component chromatogram of each single-component standard sample is obtained separately. For any single-component standard sample, changing only the ambient temperature yields chromatograms at different actual ambient temperatures, denoted as the measured chromatogram PC. BΔT Let ΔT = actual ambient temperature - standard ambient temperature; for any single-component standard sample C B ΔT, ΔT below C B Corresponding measured chromatogram PC BΔT And C B Standard single-component chromatogram PC B Combine them into a single-component training data s1, and combine all the obtained single-component training data s1 to construct a single-component training dataset S1. S1-2-2, Constructing a multi-component training dataset S2: At standard ambient temperature, a multi-component standard sample containing at least two waste gas components was obtained using a gas chromatograph for analysis, and the resulting chromatogram was recorded as a standard multi-component chromatogram. Then, the types and quantities of waste gas components in the standard sample were changed, and the samples were tested separately to obtain a standard multi-component chromatogram for each type of multi-component standard sample. For any multi-component standard sample, changing only the ambient temperature yields chromatograms at different actual ambient temperatures, denoted as the measured chromatogram PCd. BΔT Let ΔT = actual ambient temperature - standard ambient temperature; for any multi-component standard sample Cd B ΔT, ΔT under Cd B Measured chromatogram of PCd BΔT and Cd B Standard multicomponent chromatogram PCd B Combine them into a single multi-component training data s2, and combine all the obtained multi-component training data s2 to construct a multi-component training dataset S2. S1-2-3, Construct the full-component training dataset S3: At standard ambient temperature, a full-component standard sample containing all waste gas components was obtained using a gas chromatograph and measured to obtain a chromatogram, which was recorded as the standard full-component chromatogram. The concentration ratio of the waste gas components in the full-component standard sample was changed only to obtain the standard full-component chromatogram of the full-component standard sample with each different concentration ratio. For any full-component standard sample, changing only the ambient temperature yields chromatograms at different actual ambient temperatures, denoted as the measured chromatogram PCa. BΔT Let ΔT = actual ambient temperature - standard ambient temperature; for any full-component standard sample Ca B ΔT, ΔT under Ca B Measured chromatogram of PCa BΔT and Ca B Standard full-component chromatogram PCa B Combine them into a single full-component training data s3, and combine all the obtained full-component training data s3 to construct the full-component training dataset S3. The standard ambient temperature is 20–25℃.

3. The industrial waste gas detection method based on gas chromatography according to claim 2, characterized in that, Steps S1-3 specifically include: S1-3-1, Using a single-component training dataset S1, with ΔT and PC BΔT For input, PC B To output the CNN convolutional neural network, a first-order correction model is obtained after training. S1-3-2, Using a multi-component training dataset S2, with ΔT and PCd BΔT For input, PCd B To output the first-order correction model, a second-order correction model is obtained after training. S1-3-3, using the full-component training dataset S3, with ΔT, PCa BΔT For input, PCa B The second-order calibration model is trained to output the final environmental temperature difference chromatogram calibration model.

4. The industrial waste gas detection method based on gas chromatography according to claim 1, characterized in that, The method for peak finding processing using the fast peak finding model includes the following steps: S3-1, Obtain the peak vertex: S3-1-1. On the chromatographic curve of the calibrated chromatogram, starting from the origin and moving from left to right, take J consecutive data points as a data unit. For the last data unit, take the actual remaining data points as a data unit. Let the number of data points in the last data unit be J' and the total number of data units be m. Then J'≤J, and the total number of data points N=m*(J-1)+J'. S3-1-2, Simultaneously, the following steps are used to obtain the peak vertices for all data units: For a data cell, any data point P in that data cell excluding the left and right endpoints... h If the following conditions a and b are met, then the data point P is... h As the peak: a. Data point P h The signal strength is simultaneously greater than the signal strength of the first data point to its left and the first data point to its right; b. From data point P h There are at least n consecutive n directions to the left T The signal strength of each data point decreases sequentially, starting from data point P. h There are at least n consecutive n directions to the right T The signal strength of each data point decreases sequentially. Traverse all data points in the data unit except for the left and right endpoints, and construct a set Q of all the resulting peak vertices. P1 ; S3-1-3. On the chromatogram of the corrected chromatogram, for all data points at the left and right endpoints of all data units, if conditions a and b above are satisfied simultaneously, the current data point is taken as the peak vertex, and all the obtained peak vertices are constructed into a peak vertex set Q. P2 ; S3-1-4, Q P1 and Q P2 By combining the peaks, we obtain the set Q of all peak vertices in the corrected chromatogram. p Q p =Q P1 +Q P2 ; S3-2, Obtain the set of peak vertices Q p The starting point of each peak corresponding to the apex of the middle peak: For Q p Peak P in n From P n Calculate the slope of the line connecting any two adjacent data points to the left, and denote it as: P n and P n-1 The slope between them is k n P n-1 and P n-2 The slope between them is k n-1 ,...,P n-(m-1) and P n-m The slope between them is k n-(m-1) P n-m and P n-(m+1) The slope between them is k n-m ; When the first data point that satisfies either condition c or d appears, that data point P is selected. n-m As the peak P n The corresponding peak starting point: c, k n-(m-1) >0, and k n-m ≤0; d, k n-(m-1) >0, k n-m >0, and k n-(m-1) -k n-m ≤Δk T1 ;Δk T1 The preset slope change threshold is a constant that is not less than 0; S3-3, Obtain the set of peak vertices Q p The peak endpoint of each peak corresponding to the apex of the middle peak: For Q p Peak P in n From P n Calculate the slope of the line connecting any two adjacent data points to the right, and denote it as: P n and P n+1 The slope between them is k n+1 P n+1 and P n+2 The slope between them is k n+2 ,...,P n+(m'-1) and P n+m' The slope between them is k n+m' P n+m' and P n+(m'+1) The slope between them is k n+(m'+1) ; When the first data point that satisfies either condition e or f appears, then that data point P is selected. n+m' As the peak P n The corresponding peak endpoint: e, k n+m' <0, and k n+(m'+1) ≥0; f、k n+m' <0,k n+(m'+1) <0, and |k n+m' |-|k n+(m'+1) |≤Δk T2 Δk T2 The preset slope change threshold is a constant that is not less than 0; S3-4. Based on the obtained peak apex, the peak start point and peak end point corresponding to each peak apex, obtain all chromatographic peaks and index them on the chromatographic curve of the corrected chromatogram, thereby obtaining the processed chromatogram containing chromatographic peaks.

5. The industrial waste gas detection method based on gas chromatography according to claim 4, characterized in that, in, J=5~50,n T <2~5.

6. The industrial waste gas detection method based on gas chromatography according to claim 4, characterized in that, in, 0≤Δk T1 ≤0.15,0≤Δk T2 ≤0.15。 7. The industrial waste gas detection method based on gas chromatography according to claim 4, characterized in that, Step S4 is as follows: S4-1. Compare the processed chromatogram with the standard full-component chromatogram to obtain the component corresponding to each chromatographic peak in the processed chromatogram: For any chromatographic peak R in the processed chromatogram i Record the chromatographic peak R i The peak is P i , obtain P i x-coordinate t i In the standard full-component chromatogram, the horizontal axis range is (t i -0.5d min )~(t i +0.5d min Within the range, obtain the x-coordinate t of the peak. Bi With t i The peak P corresponding to the minimum absolute value of the difference Bi , will P Bi The component C corresponding to the peak i As chromatographic peak R i Corresponding ingredients; When the standard full-component chromatogram is in the range of (t) on the horizontal axis... i -0.5d min )~(t i +0.5d min No chromatographic peak was found within the specified range; the chromatographic peak R was manually confirmed. i The corresponding components; where d min It is the minimum distance between the peak vertices of two adjacent chromatographic peaks on the horizontal axis in a standard full-component chromatogram; S4-2. Based on the area of ​​the chromatographic peaks in the processed chromatogram, and combined with the standard curve of the relationship between the chromatographic peak area and component concentration in the pre-constructed characterization standard full component chromatogram, the concentration of the component corresponding to each chromatographic peak is calculated.

8. An industrial waste gas detection system based on gas chromatography, characterized in that, It uses the method described in any one of claims 1-7 to detect industrial waste gas, and the system includes: A gas chromatograph is used to detect industrial waste gas and obtain raw chromatograms. The environmental temperature difference chromatogram correction model obtains the corrected chromatogram using the method in step S2. The rapid peak-finding model uses the method in step S3 to perform peak-finding processing on the calibrated chromatogram to obtain all chromatographic peaks. And a chromatographic peak analysis module, which uses the method in step S4 to analyze the obtained chromatographic peaks to obtain the components in the industrial waste gas to be detected and the content of each component.

Images