Design method of multi-band combined concentration retrieval model

Abstract

The present application relates to the technical field of spectral detection, and more particularly to a design method of a multi-waveband combined concentration inversion model, comprising: S1: building an experimental platform for collecting the transmittance of a to-be-detected gas; S2: taking all stable wavebands with a waveband width greater than a spectral width threshold as an effective waveband set; S3: establishing a multi-waveband combined concentration inversion model; S4: calculating residual weight coefficients, dominant proportions and significant variable coefficients of each effective waveband by using a weight comparison method; S5: obtaining a weight proportion of the residual weight coefficients and the significant variable coefficients based on an AHP (Analytic Hierarchy Process) method; S6: obtaining a final multi-waveband combined concentration inversion model according to the weight proportion of the residual weight coefficients and the significant variable coefficients; S7: repeating steps S2-S4 to obtain h groups of data sets; and S8: regressing the h groups of data sets by using a multiple linear regression algorithm to obtain system errors of each effective waveband. The present application realizes high-precision gas concentration inversion.

Description

Technical Field

[0001] This invention relates to the field of spectral detection technology, and in particular to a design method for a multi-band joint concentration inversion model. Background Technology

[0002] Gas concentration detection technology plays a crucial role in fields such as chemical engineering, metallurgy, mining, medicine, and aerospace. Current optical methods for gas concentration detection mainly include optical interferometry, spectral analysis, and laser cavity ring-down methods. Among these, spectral analysis, with its sensitive and stable non-contact detection capabilities, has become a research focus in gas concentration detection. The spectral concentration inversion algorithm is a critical factor affecting the accuracy of gas concentration detection. While existing spectral concentration inversion algorithms can perform accurate spectral concentration inversion—analyzing gases within a known concentration range in a single absorption band based on Lambert-Beer's law—a limited number of sample points can affect the fitting effect during the gas concentration inversion process, thus impacting the inversion accuracy. Research on multi-band joint concentration inversion models is relatively limited, and existing multi-band joint concentration inversion models suffer from transmittance deviations from the stable range due to gas saturation absorption and weak absorption. Summary of the Invention

[0003] This invention addresses the problem of transmittance deviating from the stable range due to gas saturation absorption and weak absorption in existing multi-band joint concentration inversion models. It provides a design method for a multi-band joint concentration inversion model, which solves the problem of transmittance deviating from the stable range due to gas saturation absorption and weak absorption, and achieves high-precision gas concentration inversion.

[0004] The design method for the multi-band joint concentration inversion model proposed in this invention specifically includes the following steps: S1: Construct an experimental platform for collecting the transmittance of the gas to be tested. The experimental platform includes a Fourier transform spectrometer, a gas cell, a blackbody, a gas cylinder, and a gas mixer. The gas cell is placed between the Fourier transform spectrometer and the blackbody, and the gas cell is connected to the outlet of the gas mixer. The gas cylinder is connected to the inlet of the gas mixer. The transmittance of the gas to be tested stored in the gas cell is collected by the Fourier transform spectrometer. S2: Select stable bands for transmittance based on the stable transmittance range, and take all stable bands with a bandwidth greater than the spectral width threshold as the effective band set: in, , , and These represent the first wavenumber positions of the first effective band, the second effective band, the third effective band, and the i-th effective band, respectively. , , and These represent the second wavenumber positions of the first effective band, the second effective band, the third effective band, and the i-th effective band, respectively. The transmittance of the gas to be measured is... LL < τ gas <UL This represents the stable transmittance range. S3: Establish a multi-band joint concentration inversion model using the following formula: (2); in, The concentration value of the gas to be measured. C i ( i =1,2,…, n () represents the concentration value of the i-th effective band. e i Let be the systematic error of the i-th effective band. G i The residual weighting coefficient for the i-th effective band is... S i Let be the significant variable coefficient of the i-th effective band. A G The weighting ratio of the residual weighting coefficient. A S The weighting of the coefficients of significant variables. ( i =1,2,…, n T is the sum of squared residuals of the i-th effective band. Cal The standard transmittance spectrum is given, and cal is the calculated value. S4: Calculate the residual weight coefficient, dominance ratio, and significant variable coefficient of each effective band in the effective band set using the weighted comparison method; S5: Use the advantage ratio as an element of the decision matrix, and process the decision matrix based on the AHP analytic hierarchy process to obtain the weight ratio of the residual weight coefficients and the weight ratio of the coefficients of significant variables. S6: Substitute the weight ratios of the residual weight coefficients and the weight ratios of the significant variable coefficients into equation (2) to obtain the final multi-band joint concentration inversion model: (4); in, The dominant proportion of the i-th effective band; S7: Reacquire the transmittance of the gas to be tested using a Fourier transform spectrometer, repeat steps S2-S4, and obtain h sets of data including the concentration values, residual weight coefficients, dominance ratios, and significant variable coefficients of each effective band. S8: Use the multiple linear regression algorithm to regress the h-group dataset to obtain the systematic error of each effective band.

[0005] Preferably, step S4 specifically includes the following steps: S41: Calculate the residual weighting coefficient of the i-th effective band using the following formula. G i : (5); (6); in, Let be the sum of squared residuals for the i-th effective band. Let w be the residual of the w-th wavenumber position of the i-th effective band; S42: Calculate the dominance ratio and significance coefficient of the i-th effective band using the following formula: (7); (8); Where M is the total number of channels in the effective band, and crad is the number of channels to find. The number of items whose absolute value is less than 0.1.

[0006] Preferably, in step S5, when the residual weight coefficient and the significant variable coefficient are the same, the weight ratio of the residual weight coefficient and the weight ratio of the significant variable coefficient are both 1; when the residual weight coefficient and the significant variable coefficient are different, the weight ratio of the significant variable coefficient relative to the weight ratio of the residual weight coefficient is 1:(1- Z i ).

[0007] Preferably, the spectral width threshold is eight times the spectral resolution of the Fourier transform spectrometer.

[0008] Preferably, the effective band set contains no less than two effective bands.

[0009] Preferably, the gas cylinder is used to deliver gas to the gas pool, the gas mixer is used to change the gas concentration of the gas to be tested stored in the gas pool, the gas pool is used to store the gas to be tested, and the surface source blackbody is used to provide stable parallel light background radiation.

[0010] Compared with the prior art, the present invention can achieve the following beneficial effects: This invention can solve the problem of transmittance deviating from the stable range due to gas saturation absorption and weak absorption in the concentration inversion process of existing models, and improves the utilization rate and inversion accuracy of multi-band joint concentration inversion information.

[0011] This invention combines the transmittance stability range and spectral width threshold to achieve adaptive selection of effective bands for gas concentration inversion. At the same time, it uses the concentration information of each effective band of the gas to be measured to establish a multi-band joint concentration inversion model, and uses the AHP hierarchical analysis method and multiple linear regression method to determine the parameters in the multi-band joint concentration inversion model, thereby achieving high-precision gas concentration inversion. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the structure of an experimental platform for collecting the transmittance of a gas to be tested, according to an embodiment of the present invention. Figure 2 This is a flowchart illustrating the design method of a multi-band joint concentration inversion model provided in an embodiment of the present invention; Figure 3 This is a fitting curve diagram of adaptive band selection provided according to an embodiment of the present invention; Figure 4 This is a graph showing the transmittance of the gas under test in the first effective band and the residual distribution curve of the nonlinear least squares fitting, according to an embodiment of the present invention. Figure 5 This is a graph showing the transmittance of the gas under test in the second effective band and the residual distribution curve of the nonlinear least squares fitting, according to an embodiment of the present invention.

[0013] Figure labels: 1. Fourier transform spectrometer, 2. Gas cell, 3. Surface source blackbody, 4. Gas cylinder, and 5. Gas mixing instrument. Detailed Implementation

[0014] In the following description, embodiments of the invention will be described with reference to the accompanying drawings. In the description below, the same modules are denoted by the same reference numerals. Where the same reference numerals are used, their names and functions are also the same. Therefore, their detailed description will not be repeated.

[0015] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not constitute a limitation thereof.

[0016] This invention is based on the transmittance stability range LL < τ gas <UL (0.1<τ) gas<0.9) Selecting bands for transmittance, obtaining stable bands within the stable transmittance range, and then truncating the range of stable bands according to the spectral width threshold, finally obtaining the effective bands. For a gas to be measured with multiple effective bands, this invention first performs gas concentration inversion on each effective band simultaneously, then calculates the concentration inversion results and corresponding residual distributions by combining the gas concentration information of each effective band, and calculates the residual weight coefficients, significance coefficients, and dominance ratios of each effective band using the weight comparison method. Then, the dominance ratio is used as the judgment matrix, and the matrix is ​​normalized using the AHP analytic hierarchy process to obtain the weight ratios of the residual weight coefficients and significance coefficients. Finally, through repeated experiments, h sets of datasets are obtained, consisting of the concentration values, residual weight coefficients, dominance ratios, and significance coefficients of each band. The h sets of datasets are calculated using a multiple linear regression algorithm to obtain the systematic errors of each band. The multi-band joint concentration inversion model proposed in this invention realizes the adaptive adjustment of the effective band for gas concentration inversion, solves the problem of transmittance deviating from the stable range due to gas saturated absorption and weak absorption, and achieves high-precision gas concentration inversion.

[0017] Figure 1 The structure of an experimental platform for collecting the transmittance of a gas to be tested, according to an embodiment of the present invention, is shown.

[0018] The experimental platform for collecting the transmittance of the gas to be tested includes a Fourier transform spectrometer 1, a gas cell 2, a blackbody 3, a gas cylinder 4, and a gas mixer 5. The gas cell 2 is placed between the Fourier transform spectrometer 1 and the blackbody 3, and the gas cell 2 is connected to the outlet of the gas mixer 5. The gas cylinder 4 is connected to the inlet of the gas mixer 5. The gas cylinder 4 supplies gas to the gas cell 2. The gas mixer 5 changes the gas concentration of the gas to be tested stored in the gas cell 2. The gas cell 2 stores the gas to be tested, and the transmittance of the gas to be tested stored in the gas cell 2 is collected by the Fourier transform spectrometer 1.

[0019] Figure 2 The flowchart illustrates the design method of the multi-band joint concentration inversion model provided by an embodiment of the present invention. like Figure 2 As shown, the design method of the multi-band joint concentration inversion model proposed in this invention specifically includes the following steps: S1: Build an experimental platform for collecting the transmittance of the gas to be tested, and collect the transmittance of the gas to be tested stored in the gas cell 2 using a Fourier transform spectrometer 1.

[0020] S2: Select stable bands for transmittance based on the stable transmittance range, and take all stable bands with a bandwidth greater than the spectral width threshold as the effective band set: in, , , and These represent the first wavenumber positions of the first effective band, the second effective band, the third effective band, and the i-th effective band, respectively. , , and These represent the second wavenumber positions of the first effective band, the second effective band, the third effective band, and the i-th effective band, respectively. The transmittance of the gas to be measured is... LL < τ gas <UL This represents the stable transmittance range.

[0021] The spectral width threshold is eight times the spectral resolution of Fourier transform spectrometer 1. The effective band set contains no fewer than two effective bands.

[0022] S3: Establish a multi-band joint concentration inversion model using the following formula: (2); in, The concentration value of the gas to be measured. C i ( i =1,2,…, n () represents the concentration value of the i-th effective band. e i Let be the systematic error of the i-th effective band. G i The residual weighting coefficient for the i-th effective band is... S i Let be the significant variable coefficient of the i-th effective band. A G The weighting ratio of the residual weighting coefficient. A S The weighting of the coefficients of significant variables. ( i =1,2,…, n ) represents the sum of squared residuals for the i-th effective band.

[0023] Step S4 specifically includes the following steps: S41: Calculate the residual weighting coefficient of the i-th effective band using the following formula. G i : (4); (5); in, Let be the sum of squared residuals for the i-th effective band. Let w be the residual of the w-th wavenumber position of the i-th effective band; S42: Calculate the dominance ratio and significance coefficient of the i-th effective band using the following formula: (6); (7); Where M is the total number of channels in the effective band, and crad is the number of channels to find. The number of items whose absolute value is less than 0.1.

[0024] S5: Use the advantage ratio as an element of the decision matrix, and process the decision matrix based on the AHP (Analytic Hierarchy Process) to obtain the weight ratio of the residual weight coefficients and the weight ratio of the coefficients of significant variables.

[0025] In step S5, when the residual weight coefficient and the significance coefficient are the same, the weight ratio of both the residual weight coefficient and the significance coefficient is 1; when the residual weight coefficient and the significance coefficient are different, the weight ratio of the significance coefficient relative to the residual weight coefficient is 1: (1- Z i ).

[0026] S6: Substitute the weight ratios of the residual weight coefficients and the weight ratios of the significant variable coefficients into equation (2) to obtain the final multi-band joint concentration inversion model: (8); in, The dominant proportion of the i-th effective band; S7: Use Fourier transform spectrometer 1 to reacquire the transmittance of the gas to be measured, repeat steps S2-S4, and obtain h sets of data including the concentration values, residual weight coefficients, dominance ratios and significant variable coefficients of each effective band. S8: Use the multiple linear regression algorithm to regress the h-group dataset to obtain the systematic error of each effective band.

[0027] Systematic error e i It is a non-random error, which originates from fixed factors such as instrument processing and adjustment errors, and does not change with the gas concentration. Therefore, the multivariate linear regression algorithm can be used to solve for the systematic error.

[0028] To understand the specific process of the multiple linear regression algorithm regressing the h sets of data, the following detailed explanation uses 6% N2O gas as an example. Eight experiments were conducted on N2O gas using an experimental platform for collecting the transmittance of the gas to be tested, and the experimental data shown in Table 1 were obtained.

[0029] Table 1

[0030] Where C1 and C2 are the concentration inversion values ​​in the 2130~2170cm⁻¹ and 2537~2584cm⁻¹ bands, respectively, Creal is the concentration of the experimental gas N₂O (6%), and η = ΣC real -C i [G i (1-Z i ) / (2-Z i )+S i / (2-Z i )], where η is the dependent variable.

[0031] To solve for constant coefficients e i By constructing a multiple linear regression model and further rearranging equation (8), we can obtain: (9); Based on equation (9), multiple linear regression was performed on the data in Table 1 to obtain the regression coefficients for each experimental band, i.e., the constant coefficients. e i Ultimately, the height is 2130 ~ 2170cm. -1 Constant coefficients of the band e 1 represents +1.1559, 2537 ~ 2584cm -1 Constant coefficients of the band e 2 is -1.1697.

[0032] Figure 3 The fitted curve for adaptive band selection provided according to an embodiment of the present invention is shown.

[0033] like Figure 3 As shown, a~b represent the first effective band of the gas to be measured, and c~d represent the second effective band. The specific selection method is as follows: draw lines within the range of 0.1 and 0.9 to select many bands, and then perform a secondary selection on the bands among these bands that have a number greater than 8 times the resolution. The bands that ultimately meet the target are the first and second effective bands of the gas to be measured.

[0034] Figure 4 The diagram shows the transmittance of the gas under test in the first effective band and the residual distribution curve of the nonlinear least squares fitting, according to an embodiment of the present invention. Figure 5 The diagram shows the transmittance of the gas under test in the second effective band and the residual distribution curve of the nonlinear least squares fitting, according to an embodiment of the present invention.

[0035] like Figure 4As shown, the gas under test is in the wavelength range of 2144~2176 cm. -1 The concentration inversion fitting effect is good, and the residual values ​​are all distributed around 0. For example... Figure 5 As shown, the gas under test is in the wavelength range of 2519~2552 cm⁻¹ -1 The concentration inversion fitting effect is good, with residual values ​​all distributed around 0, but in the band 2552~2600 cm⁻¹ -1 The fitting effect is poor.

[0036] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0037] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A design method for a multi-band joint concentration inversion model, characterized in that, Specifically, the steps include the following: S1: Construct an experimental platform for collecting the transmittance of the gas to be tested. The experimental platform includes a Fourier transform spectrometer, a gas cell, a blackbody, a gas cylinder, and a gas mixer. The gas cell is placed between the Fourier transform spectrometer and the blackbody, and the gas cell is connected to the outlet of the gas mixer. The gas cylinder is connected to the inlet of the gas mixer. The transmittance of the gas to be tested stored in the gas cell is collected by the Fourier transform spectrometer. S2: Select stable bands for transmittance based on the stable transmittance range, and take all stable bands with a band width greater than the spectral width threshold as the effective band set: in, , , and These represent the first wavenumber positions of the first effective band, the second effective band, the third effective band, and the i-th effective band, respectively. , , and These represent the second wavenumber positions of the first effective band, the second effective band, the third effective band, and the i-th effective band, respectively. The transmittance of the gas to be measured. LL < τ gas <UL This refers to the stable transmittance range; S3: Establish a multi-band joint concentration inversion model using the following formula: （2）； in, The concentration value of the gas to be measured. C i Let i be the concentration value of the i-th effective band. i =1,2,…, n , e i Let be the systematic error of the i-th effective band. G i The residual weighting coefficient for the i-th effective band is... S i Let be the significant variable coefficient of the i-th effective band. A G The weighting ratio of the residual weighting coefficient. A S The weighting of the coefficients of significant variables. T is the sum of squared residuals for the i-th effective band. Cal The standard transmittance spectrum is given, and cal is the calculated value. S4: Calculate the residual weight coefficient, dominance ratio, and significant variable coefficient of each effective band in the effective band set using the weighted comparison method; S5: Use the advantage ratio as an element of the decision matrix, and process the decision matrix based on the AHP analytic hierarchy process to obtain the weight ratio of the residual weight coefficient and the weight ratio of the significant variable coefficient. S6: Substitute the weight ratios of the residual weight coefficients and the weight ratios of the significant variable coefficients into equation (2) to obtain the final multi-band joint concentration inversion model: （4）； in, The dominant proportion of the i-th effective band; S7: Reacquire the transmittance of the gas to be tested using the Fourier transform spectrometer, repeat steps S2-S4, and obtain h sets of datasets including the concentration values, residual weight coefficients, dominance ratios, and significant variable coefficients of each effective band. S8: Use the multiple linear regression algorithm to regress the h-group dataset to obtain the systematic error of each effective band.

2. The design method of the multi-band joint concentration inversion model according to claim 1, characterized in that, Step S4 specifically includes the following steps: S41: Calculate the residual weighting coefficient of the i-th effective band using the following formula. G i : （5）； （6）； in, Let be the sum of squared residuals for the i-th effective band. Let w be the residual of the w-th wavenumber position of the i-th effective band; S42: Calculate the dominance ratio and significance coefficient of the i-th effective band using the following formula: （7）； （8）； Where M is the total number of channels in the effective band, and crad is the number of channels to find. The number of items whose absolute value is less than 0.

1.

3. The design method of the multi-band joint concentration inversion model according to claim 2, characterized in that, In step S5, when the residual weight coefficient and the significant variable coefficient are the same, the weight ratio of the residual weight coefficient and the weight ratio of the significant variable coefficient are both 1; when the residual weight coefficient and the significant variable coefficient are different, the weight ratio of the significant variable coefficient relative to the weight ratio of the residual weight coefficient is 1: (1- Z i ).

4. The design method of the multi-band joint concentration inversion model according to claim 1, characterized in that, The spectral width threshold is eight times the spectral resolution of the Fourier transform spectrometer.

5. The design method of the multi-band joint concentration inversion model according to claim 1, characterized in that, The effective band set contains no less than two effective bands.

6. The design method of the multi-band joint concentration inversion model according to claim 1, characterized in that, The gas cylinder supplies gas to the gas pool, the gas mixer changes the gas concentration of the gas to be tested stored in the gas pool, the gas pool stores the gas to be tested, and the surface source blackbody provides stable parallel light background radiation.

Images