Gas concentration multi-peak distribution high-precision reconstruction method for laser remote measurement

Abstract

The invention discloses a gas concentration multimodal distribution high-precision reconstruction method for laser remote measurement, which comprises the following steps: acquiring gas concentration information of each light beam at each angle, and reconstructing the gas concentration information by adopting an algebraic iteration reconstruction algorithm to obtain a projection matrix; a flat penalty term is introduced, and false peak errors are reduced; after the reconstruction is finished, introducing a local smoothing constraint of self-adaptive smoothing intensity, and carrying out smoothing constraint on the pixel point by adopting a smoothing intensity coefficient which changes along with the change of eight pixel values around the pixel point; and the gas concentration of a single pixel is calculated by using the Beer-Lambert law, so that a gas concentration two-dimensional distribution diagram is obtained. According to the method, a flat constraint penalty term is added in the iteration process, and the false peak error in the multi-peak gas distribution reconstruction process is reduced; and moreover, local smoothing constraint of self-adaptive smoothing intensity is introduced, the continuity and stability of the reconstructed gas distribution diagram are improved, noise, artifacts and false peak errors are reduced, and the convergence speed is increased.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of gas concentration distribution reconstruction, and particularly relates to a high-precision reconstruction method for multi-peak distribution of gas concentration for laser remote sensing, which is suitable for non-uniform gas distribution reconstruction in the fields of combustion diagnosis, environmental monitoring and the like. BACKGROUND

[0002] In the fields of combustion diagnosis, environmental monitoring, industrial process control and the like, non-invasive gas concentration distribution measurement is of great significance. The tunable diode laser absorption spectroscopy (TDLAS) has been widely used in the field of gas measurement, and a lot of achievements have been made in the related field at home and abroad. The TDLAS has the advantages of non-contact detection, fast speed, high sensitivity and the like, and becomes a new type of gas concentration measurement method for the to-be-measured region. However, the concentration detected by the TDLAS is generally the average concentration of the laser beam passing through the region, and the specific distribution information of the gas field cannot be obtained. Therefore, the TDLAS technology can be combined with the computer tomography technology, and the iterative algorithm is used to reconstruct the specific distribution information of the gas concentration.

[0003] At present, the reconstruction method basically adopts an algebraic iterative reconstruction algorithm. Although the algebraic iterative reconstruction algorithm is widely used in gas reconstruction, it still has the following inherent defects: false peak artifact: when reconstructing complex gas distribution such as double-peak or multi-peak distribution, false peaks are easily generated in the region between the peaks, which is caused by the error amplification due to the cross correlation of the ray path; noise sensitivity: the measurement noise can significantly reduce the reconstruction accuracy and the like. Therefore, it is an urgent problem to be solved to develop a gas distribution reconstruction algorithm which can improve the reconstruction image accuracy and reduce the high noise in the complex gas scene. SUMMARY

[0004] The present application aims to make up for the defects of the prior art, and provides a high-precision reconstruction method for multi-peak distribution of gas concentration for laser remote sensing.

[0005] The present application is realized by the following technical solutions:

[0006] A high-precision reconstruction method for multi-peak distribution of gas concentration for laser remote sensing comprises the following steps:

[0007] S1, measuring the multi-peak gas distribution at different angles to obtain the gas concentration information of each light beam at each angle, and reconstructing the gas concentration information by using an algebraic iterative reconstruction algorithm to obtain a projection matrix;

[0008] S2, introducing a flat penalty term to reduce the false peak error generated in the multi-peak distribution in the reconstruction process;

[0009] S3, after the end of reconstruction, introduce adaptive smoothing intensity of local smoothing constraint, using the smoothing intensity coefficient which changes with the change of 8 pixel values around the pixel point to constrain the pixel point, reduce noise and artifacts;

[0010] S4, based on the obtained projection matrix, using the Beer-Lambert law to calculate the gas concentration of a single pixel, thereby obtaining the two-dimensional distribution map of gas concentration.

[0011] The specific content of step S1 is as follows:

[0012] S1.1, build a TDLAS detection system, and discretize the known gas distribution into an n*n grid;

[0013] S1.2, measure the multi-peak gas distribution at different angles to obtain the gas integral absorbance information of each light beam at each angle;

[0014] S1.3, the obtained integral absorbance information and the length of each light beam passing through the grid are brought into the algebraic iterative algorithm for reconstruction.

[0015] First, discretize the image, that is, discretize the entire two-dimensional gas concentration distribution image into N=n*n grids, and the gas parameters inside the grid are considered to be uniform. It is assumed that there are M rays passing through the image from different directions, and the rays are numbered in order, l i represents the i-th ray, and the length of the i-th ray in the j-th pixel is denoted as l ij , the projection of the i-th ray in the j-th pixel is defined as: D ij =l ij *x ij , then the total projection of the i-th ray is represented as

[0016]

[0017] where N is the total number of all pixels in the grid, x j is the integral absorbance in the jth grid;

[0018] The projection equation set of M rays is obtained from equation (1):

[0019]

[0020] D M represents the total projection value of the Mth ray;

[0021] The projection equation set is expressed in matrix form as LX=D, where L represents the length coefficient matrix, X represents the image matrix, and D represents the projection data matrix;

[0022] Solve using algebraic iteration method, the solution formula is:

[0023]

[0024] where λ is a relaxation factor, 0 < λ < 2; j represents the index of the N-dimensional image vector, 1 ≤ j ≤ N, is the integral absorbance of the jth pixel; K is the kth iteration, when or the iteration number is exceeded, the iteration is stopped.

[0025] The specific content of step S2 is as follows:

[0026] In view of the false peak error generated by the algebraic iterative reconstruction algorithm under the multi-peak distribution reconstruction, a flat penalty term is introduced The solving formula is:

[0027]

[0028] where is a flat penalty term, and its formula is:

[0029]

[0030] where N(j) is all points adjacent to pixel j, and α, β, and γ are fixed penalty parameters, is the integral absorbance of the jth pixel, is the integral absorbance of the pth pixel.

[0031] The specific content of step S3 is as follows:

[0032] After the reconstruction is completed, a local smoothing constraint with an adaptive smoothing strength is introduced, and a smoothing strength coefficient that changes with the change of the values of the eight pixels around the pixel point is used to perform a smoothing constraint on the pixel point, and the solving formula is:

[0033]

[0034] where N(j) is all points adjacent to pixel j, and θ is an adaptive smoothing strength coefficient, and its formula is:

[0035]

[0036] where ε is an initial smoothing strength coefficient, is the current grid pixel value.

[0037] The specific content of step S4 is as follows:

[0038] According to the Beer-Lambert law, when a certain laser beam i passes through a target area, the integral absorbance of the absorption line with a central frequency v is discretized as:

[0039]

[0040] wherein X i is the integral absorbance of the i-th light beam; S(T) is the line intensity at temperature T, P is the gas pressure, and X is the gas concentration, according to the Beer-Lambert law, x j = PXS(T); x j is the integral absorbance in the j-th grid, so as to obtain the concentration C j is:

[0041]

[0042] P is the pressure, L j is the optical path of the j-th pixel, and S(T) is the line intensity at temperature T.

[0043] A device for high-precision reconstruction of multi-peak distribution of gas concentration for laser remote sensing, comprising:

[0044] A data acquisition and reconstruction module: measuring the multi-peak gas distribution at different angles, acquiring the gas concentration information of each light beam at each angle, and reconstructing the gas concentration information using an algebraic iterative reconstruction algorithm to obtain a projection matrix;

[0045] A flat penalty term introduction module: introducing a flat penalty term to reduce the false peak error generated by the multi-peak distribution in the reconstruction process;

[0046] A smoothing constraint module: after the reconstruction is completed, a local smoothing constraint with adaptive smoothing strength is introduced, and a smoothing strength coefficient that changes with the change of the values of the 8 surrounding pixels is used to perform smoothing constraint on the pixel points;

[0047] A gas concentration calculation module: based on the obtained projection matrix, the gas concentration of a single pixel is calculated using the Beer-Lambert law to obtain a two-dimensional distribution map of the gas concentration.

[0048] A computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the high-precision reconstruction method for multi-peak distribution of gas concentration for laser remote sensing when executing the computer program.

[0049] A computer-readable storage medium having a computer program stored thereon, the computer program implementing the steps of the high-precision reconstruction method for multi-peak distribution of gas concentration for laser remote sensing when executed by a processor.

[0050] The present application has the advantages that: the present application provides a gas concentration multi-peak distribution high-precision reconstruction method for laser remote sensing, a flat constraint penalty term is added in the iteration process, and false peak error in the multi-peak gas distribution reconstruction process is reduced; and a local smoothing constraint with adaptive smoothing strength is further introduced, the continuity and stability of the reconstructed gas distribution map are improved, noise, artifacts and false peak error are reduced, and the convergence speed is accelerated;

[0051] The present application introduces a flat penalty term to suppress noise, when the current pixel value is higher than the neighbor value, the penalty term will lower the pixel value; when the current pixel value is lower than the neighbor value, the penalty term will appropriately raise the pixel value; because the false peak center pixel is high relative to the neighbor pixel value, and the real peak has continuity, the center pixel has little difference relative to the neighbor pixel value, so the penalty term can suppress the false peak error;

[0052] The present application introduces a local smoothing constraint with adaptive smoothing strength, improves the continuity and stability of the reconstructed gas distribution map, reduces noise sensitivity and artifact problems; the local smoothing constraint can force the value of each grid point to approach the surrounding grid, when a sharp fluctuation or unreasonable peak value of the edge grid is generated, for example, a false peak, the smoothing constraint effectively reduces such abnormalities by forcing the adjacent grid values to be similar;

[0053] The present application further changes the fixed smoothing strength coefficient to an adaptive smoothing strength coefficient, which dynamically responds to the value of the corresponding pixel value, if the current value is significantly higher than the average value of the neighbors, the smoothing strength coefficient decreases, and vice versa, the smoothing strength coefficient increases, ensuring that the most appropriate smoothing strength coefficient is used when smoothing the pixel value, and avoiding excessive smoothing or local optimal solution. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 is a TDLAS-based gas multi-peak distribution concentration high-precision reconstruction method flowchart of the present application;

[0055] Figure 2 is a gas discretization area in the embodiment of the present application;

[0056] Figure 3a is a top view of the original reconstructed double-peak gas distribution map in the embodiment of the present application, Figure 3b is a perspective view of the original reconstructed double-peak gas distribution map;

[0057] Figure 4a is a top view of the double-peak gas distribution reconstructed by using the ART algorithm in the embodiment of the present application, Figure 4b is a perspective view of the double-peak gas distribution reconstructed by using the ART algorithm;

[0058] Figure 5aIt is a top view of a double-peak gas distribution reconstructed by a gas multi-peak distribution concentration high-precision algorithm used in an embodiment of the present application, Figure 5b It is a perspective view of a double-peak gas distribution reconstructed by a gas multi-peak distribution concentration high-precision algorithm,

[0059] Figure 6a It is an error graph reconstructed by an ART algorithm used in an embodiment of the present application, Figure 6b It is an error graph reconstructed by a gas multi-peak distribution concentration high-precision algorithm. DETAILED DESCRIPTION

[0060] In order to better understand the purpose, structure and function of the present application, a gas concentration multi-peak distribution high-precision reconstruction method for laser remote sensing of the present application is further described in detail below in combination with the drawings.

[0061] As shown in the drawings, Figure 1 A gas concentration multi-peak distribution high-precision reconstruction method for laser remote sensing of the present application comprises the following steps: S1, measuring a multi-peak gas distribution at different angles to obtain gas concentration information of each light beam at each angle, reconstructing the gas concentration information by using an algebraic iterative reconstruction algorithm to obtain a projection matrix;

[0062] S1.1, building a TDLAS detection system to discretize a known gas distribution into an n*n grid;

[0063] S1.2, measuring a multi-peak gas distribution at different angles to obtain gas integral absorbance information of each light beam at each angle;

[0064] S1.3, bringing the obtained integral absorbance information and the length of each light beam passing through the grid into the algebraic iterative algorithm for reconstruction.

[0065] S2, introducing a flat penalty term to reduce the false peak error generated by the multi-peak distribution in the reconstruction process;

[0066] S3, after the reconstruction is completed, introducing a local smoothing constraint with adaptive smoothing strength, and using a smoothing strength coefficient that changes with the change of 8 pixel values around a pixel point to perform smoothing constraint on the pixel point;

[0067] S4, based on the obtained projection matrix, using the Beer-Lambert law to calculate the gas concentration of a single pixel to obtain a two-dimensional gas concentration distribution graph.

[0068] As shown in the drawings, Figure 2 and Figure 3a, b shows, first of all, the image is discretized, that is, the entire gas concentration two-dimensional distribution image is discretized into N = n * n grid, the gas parameters inside the grid are considered to be uniform, the known gas distribution is discretized into n * n grid, first of all, it is assumed that there are M rays passing through the image from different directions, the rays are numbered in order, l i represents the i-th ray. The ray passes through the image, and the length of the i-th ray in the j-th pixel is denoted as l ij , the projection of the i-th ray in the j-th pixel is defined as: D ij = l ij *x ij , the total projection of the i-th ray can be represented as:

[0069]

[0070] Thus, the projection equation set of M rays is obtained as:

[0071]

[0072] The projection equation set can be represented in matrix form as LX = D, L represents the length coefficient matrix, X represents the image matrix, and D represents the projection data matrix;

[0073] Solve by algebraic iteration method, the general solution formula is:

[0074]

[0075] Wherein, λ (0 < λ < 2) is the relaxation factor; j (1 ≤ j ≤ N) represents the index of N-dimensional image vector, is the integral absorbance of the j-th pixel; when or the number of iterations is exceeded, the iteration is stopped;

[0076] According to Beer-Lambert (Beer-Lambert) law, when a certain laser i passes through the target area, the integral absorbance of the absorption line with central frequency v can be discretized as:

[0077]

[0078] Wherein, X i is the integral absorbance of the i-th light; S (T) is the line intensity when the temperature is T; x j is the integral absorbance in the j-th grid, so the concentration C j of the j-th pixel is:

[0079]

[0080] P is the pressure, L j is the optical path of the j-th pixel, and S (T) is the line intensity when the temperature is T.

[0081] As Figure 4a is the top view of the reconstructed bi-modal gas distribution using the ART algebraic iterative algorithm, Figure 4b is the perspective view of the reconstructed bi-modal gas distribution using the ART algebraic iterative algorithm;

[0082] On the basis of the ART algebraic iterative algorithm, the flat constraint penalty term and the local smoothing constraint with adaptive strength are introduced. In order to solve the false peak error generated by the reconstruction algorithm in the bi-modal or even multi-modal distribution reconstruction, the flat penalty term is introduced The solving formula is:

[0083]

[0084] Wherein is the flat penalty term, and its formula is:

[0085]

[0086] Wherein N(j) is all the points adjacent to the pixel j, and a, b, g are penalty parameters;

[0087] After the reconstruction is completed, the local smoothing constraint with adaptive smoothing strength is introduced, and the smoothing strength coefficient which changes with the change of the value of the 8 pixels around the pixel is used to constrain the pixel, and the solving formula is:

[0088]

[0089] Wherein N(j) is all the points adjacent to the pixel j, and a, b, g are penalty parameters;

[0090]

[0091] Wherein ε is the initial smoothing strength coefficient, is the current grid pixel value

[0092] As Figure 5a is the top view of the reconstructed bi-modal gas distribution using the ART algebraic iterative algorithm, Figure 5b is the perspective view of the reconstructed bi-modal gas distribution using the ART algebraic iterative algorithm;

[0093] As Figure 6a shown, the image is the error distribution diagram of the image reconstructed by the ART algebraic iterative reconstruction algorithm and the image before reconstruction, wherein the highest error point is concentrated in the area between the two peak values, and the maximum value is 0.046664923. The total relative error R of the whole image is calculated by the formula:

[0094]

[0095] wherein X i is the image after reconstruction, Y i is the image before reconstruction;

[0096] The total relative error of the ART algebraic iterative reconstruction algorithm is 0.17408;

[0097] As Figure 6b shown, the error distribution diagram of the image after reconstruction using the gas multi-peak distribution concentration high-precision algorithm and the image before reconstruction is shown, wherein the highest error point is concentrated in the region between the two peaks, the maximum value is 0.024822018, and the total relative error is 0.1252;

[0098] Table 1

[0099]

[0100] Compared with the ART algebraic iterative reconstruction algorithm, the gas multi-peak distribution concentration high-precision algorithm reduces the maximum error by 35.4%, reduces the relative error by 28.08%, reduces the number of iterations by 19.6%, and greatly reduces the false peak error between peaks;

[0101] A gas concentration multi-peak distribution high-precision reconstruction device for laser remote sensing comprises:

[0102] A data acquisition and reconstruction module: measuring the multi-peak gas distribution at different angles, acquiring the gas concentration information of each light beam at each angle, and reconstructing the gas concentration information using an algebraic iterative reconstruction algorithm to obtain a projection matrix;

[0103] A flat penalty term introduction module: introducing a flat penalty term to reduce the false peak error generated by the multi-peak distribution in the reconstruction process;

[0104] A smoothing constraint module: after the reconstruction is completed, a local smoothing constraint with an adaptive smoothing intensity is introduced, and a smoothing intensity coefficient that changes with the change of the values of the 8 surrounding pixels is used to perform smoothing constraint on the pixel points;

[0105] A gas concentration calculation module: based on the obtained projection matrix, the gas concentration of a single pixel is calculated using the Beer-Lambert law to obtain a two-dimensional gas concentration distribution map.

[0106] A computer device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the gas concentration multi-peak distribution high-precision reconstruction method for laser remote sensing when executing the computer program.

[0107] A computer readable storage medium, having stored thereon a computer program, the computer program being executed by a processor to implement steps of a gas concentration multi-peak distribution high-precision reconstruction method for laser telemetry.

[0108] The application is described in detail in conjunction with the drawings and several embodiments, and those skilled in the art can make various changes or equivalent replacements to the features and embodiments without departing from the essential spirit and scope of the application. In addition, under the guidance of the application, the skilled person can modify these features and embodiments according to specific conditions without departing from the core idea of the application. Therefore, the protection scope of the application is not limited to the specific embodiments, and all embodiments meeting the scope of the claims of the application belong to the protection scope of the application. All the technical contents claimed are described in detail in the claims. Any modification and improvement of the technical scheme of the application within the scope of the technical concept of the application should be regarded as within the protection scope of the application.

Claims

1. A method for high-precision reconstruction of multi-peak gas concentration distribution for laser telemetry, characterized in that, Includes the following steps: S1. Measure the multi-peak gas distribution at different angles to obtain the gas concentration information of each beam at each angle. Reconstruct the gas concentration information using an algebraic iterative reconstruction algorithm to obtain the projection matrix. S2. To reduce the false peak error caused by the multi-peak distribution during the reconstruction process, a flattening penalty term is introduced; S3. After reconstruction, local smoothing constraints with adaptive smoothing intensity are introduced. The smoothing intensity coefficient that changes with the value of the 8 surrounding pixels is used to smooth the pixels. S4. Based on the obtained projection matrix, the gas concentration of a single pixel is calculated using Beer-Lambert's law, thereby obtaining a two-dimensional distribution map of gas concentration.

2. The method for high-precision reconstruction of multi-peak gas concentration distribution for laser telemetry according to claim 1, characterized in that, The specific details of step S1 are as follows: S1.1 Build a TDLAS detection system and discretize the known gas distribution into an n*n grid; S1.2 Measure the multi-peak gas distribution at different angles and obtain the gas integrated absorbance information of each beam at each angle; S1.

3. The obtained integrated absorbance information and the length of each beam passing through the grid are fed into the algebraic iterative algorithm for reconstruction.

3. The method for high-precision reconstruction of multi-peak gas concentration distribution for laser telemetry according to claim 2, characterized in that, First, the image is discretized, that is, the entire two-dimensional gas concentration distribution image is discretized into an N = n*n grid. The gas parameters within the grid are considered uniform. Assume that M rays pass through the two-dimensional gas concentration distribution image from different directions, and the rays are numbered sequentially. i Let l represent the i-th ray, which passes through a two-dimensional gas concentration distribution image. Let l be the length of the i-th ray within the j-th pixel. ij Let the projection of the i-th ray into the j-th pixel be: D ij =l ij *x ij The total projection of the i-th ray is expressed as: Where N is the total number of pixels in the grid, x j The integral absorbance within the j-th grid; From equation (1), we obtain the projection equations for the M rays: D M This represents the total projection value of the Mth ray; The projection equations are represented in matrix form as LX = D, where L represents the length coefficient matrix, X represents the image matrix, and D represents the projection data matrix. The solution is obtained using the algebraic iteration method, and the formula is: Where λ is the relaxation factor, 0 < λ < 2; j represents the index of the N-dimensional image vector, 1 ≤ j ≤ N. Let K be the integral absorbance of the j-th pixel; K is the integral absorbance of the k-th iteration, when... The iteration may stop if the number of iterations is exceeded.

4. The method for high-precision reconstruction of multi-peak gas concentration distribution for laser telemetry according to claim 3, characterized in that, The specific details of step S2 are as follows: To address the spurious peak error generated by the algebraic iterative reconstruction algorithm under multi-peak distribution reconstruction, a flattening penalty term is introduced. The solution formula is: in The flatness penalty term is formulated as follows: Where N(j) represents all points adjacent to pixel j, and α, β, γ are fixed penalty parameters. Let J be the integral absorbance of the j-th pixel. Let be the integral absorbance of the p-th pixel.

5. The method for high-precision reconstruction of multi-peak gas concentration distribution for laser telemetry according to claim 4, characterized in that, The specific details of step S3 are as follows: After reconstruction, a local smoothing constraint with adaptive smoothing intensity is introduced. The smoothing intensity coefficient, which changes with the values ​​of the eight surrounding pixels, is used to constrain the smoothing of the pixel. The solution formula is as follows: Where N(j) represents all points adjacent to pixel j, and θ is the adaptive smoothing intensity coefficient, whose formula is: Where ε is the initial smoothing intensity coefficient. This represents the current grid pixel value.

6. The method for high-precision reconstruction of multi-peak gas concentration distribution for laser telemetry according to claim 5, characterized in that, The specific details of step S4 are as follows: According to Beer-Lambert's law, when a laser beam i passes through a target region, the integrated absorbance of the absorption line with center frequency v is discrete as follows: Among them, X i Let S(T) be the integral absorbance of the i-th ray; S(T) be the linear intensity at temperature T; P be the gas pressure; X be the gas concentration; and according to Beer-Lambert's law, x... j =PXS(T); x j Let C be the integral absorbance within the j-th grid, thus obtaining the concentration C of the j-th pixel. j for: P is pressure, L j Let S be the optical path length of the j-th pixel, and S(T) be the linear intensity at temperature T.

7. A high-precision reconstruction device for multi-peak gas concentration distribution in laser telemetry, characterized in that, Including: Data acquisition and reconstruction module: Measure the multi-peak gas distribution at different angles, acquire the gas concentration information of each beam at each angle, and reconstruct the gas concentration information using an algebraic iterative reconstruction algorithm to obtain the projection matrix; A flattening penalty term module is introduced: a flattening penalty term is introduced to reduce the false peak error caused by the multi-peak distribution during the reconstruction process; Smoothing constraint module: After reconstruction, local smoothing constraint with adaptive smoothing intensity is introduced. The smoothing intensity coefficient changes with the value of the 8 surrounding pixels to perform smoothing constraint on the pixel. Gas concentration calculation module: Based on the obtained projection matrix, the gas concentration of a single pixel is calculated using Beer-Lambert's law, thereby obtaining a two-dimensional gas concentration distribution map.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the high-precision reconstruction method for multi-peak gas concentration distribution for laser telemetry as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the high-precision reconstruction method for multi-peak gas concentration distribution for laser telemetry as described in any one of claims 1-6.

Images