Welding arc emission coefficient reconstruction parameter tuning method based on PSO-AwASD-POCS algorithm

Through the PSO-AwASD-POCS algorithm, the reconstruction parameters of welding arc emission coefficient reconstruction are automatically tuned, which solves the problem of lack of theoretical selection of parameter settings of AwASD-POCS algorithm and improves the accuracy of welding arc spectrum diagnosis. level of sex and automation.

CN114299003BActive Publication Date: 2024-11-26INNER MONGOLIA UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111603186.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-24
Publication Date
2024-11-26
Estimated Expiration
2041-12-24

AI Technical Summary

Technical Problem

The AwASD-POCS algorithm needs to manually set 8 parameters during welding arc reconstruction. There is no theoretical selection criteria, which limits its promotion and application.

Method used

The parameter tuning method of welding arc emission coefficient reconstruction based on PSO-AwASD-POCS algorithm is adopted. The emission coefficient distribution model is constructed by acquiring arc feature spectrum lines images, multi-angle projection and Gaussian white noise superposition, initializing the particle swarm position and velocity, and iterating Calculate to automatically tune the reconstruction parameters.

Benefits of technology

It realizes adaptive automatic selection of appropriate reconstruction parameters, improves the accuracy and automation level of welding arc spectroscopy diagnosis, reduces artificial intervention, and provides a reliable basis for welding arc spectroscopy diagnosis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114299003B_ABST
    Figure CN114299003B_ABST
Patent Text Reader

Abstract

The present invention discloses a welding arc emission coefficient reconstruction parameter tuning method based on the PSO-AwASD-POCS algorithm, and the steps include: obtaining an arc characteristic spectral line image, and constructing an emission coefficient distribution mathematical model according to the arc characteristic spectral line image; projecting the emission coefficient distribution mathematical model from different angles to obtain multi-angle characteristic spectral line intensity, and superimposing Gaussian white noise on the multi-angle characteristic spectral line intensity; setting algorithm parameters, and setting the reconstruction parameter value range; initializing the particle group position and speed, and combining the multi-angle characteristic spectral line intensity, algorithm parameters and reconstruction parameter value range, calculating the fitness of the particle group after initialization; updating the particle group position and speed and performing iterative calculation, according to the iterative result and the fitness of the particle group after initialization, obtaining the tuning reconstruction parameters. The present invention automatically selects appropriate reconstruction parameters, effectively improving the accuracy and automation level of spectral diagnosis.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of welding arc spectrum diagnosis, and in particular to a welding arc emission coefficient reconstruction parameter tuning method based on a PSO-AwASD-POCS algorithm. Background Art

[0002] Arc spectrum diagnosis is a non-contact detection method with the advantages of high sensitivity, good selectivity, non-intrusiveness, and high temporal and spatial resolution. Using the welding arc spectrum information to quantitatively characterize the temperature distribution and electron density inside the arc helps to understand the arc heat generation mechanism, discover welding defects, and provide a theoretical basis for improving welding technology and improving weld quality. As an implementation algorithm of compressed sensing theory, AwASD-POCS can well solve the problem of ill-posedness in solving the objective function. It can obtain satisfactory welding arc reconstruction results without the need for spectral data acquisition at each angle. It can not only significantly reduce the cost of acquisition equipment, but also reduce the probability of introducing noise signals due to cumbersome acquisition.

[0003] However, since the AwASD-POCS algorithm requires eight parameters to be properly set during the reconstruction process, and these parameters have no theoretical selection criteria and can only rely on the experience of researchers, this seriously restricts the promotion and application of the algorithm. Summary of the invention

[0004] In view of the defects in the prior art, the present invention provides a welding arc emission coefficient reconstruction parameter tuning method based on the PSO-AwASD-POCS algorithm, the steps comprising:

[0005] Acquire an arc characteristic spectral line image, and construct an emission coefficient distribution mathematical model according to the arc characteristic spectral line image;

[0006] Projecting the emission coefficient distribution mathematical model from different angles to obtain multi-angle characteristic spectral line intensities, and superimposing Gaussian white noise on the multi-angle characteristic spectral line intensities;

[0007] Set algorithm parameters and set the value range of reconstruction parameters;

[0008] Initializing the position and speed of the particle swarm, and calculating the fitness of the particle swarm after the initialization based on the multi-angle characteristic spectral line intensity, algorithm parameters, and reconstruction parameter value range;

[0009] The particle swarm position and velocity are updated and iterative calculation is performed, and the tuning reconstruction parameters are obtained according to the iterative results and the fitness of the initialized particle swarm.

[0010] Furthermore, the arc characteristic spectral line image is obtained, and a mathematical model of emission coefficient distribution is constructed according to the arc characteristic spectral line image, specifically:

[0011] Call the imread function to obtain the arc characteristic spectrum line image according to the specified path;

[0012] Calling the cftool toolbox, and constructing a mathematical model of emission coefficient distribution according to the data distribution characteristics of each row of the arc characteristic spectrum line image;

[0013] The emission coefficient distribution mathematical model is normalized.

[0014] Furthermore, the emission coefficient distribution mathematical model is projected from different angles to obtain multi-angle characteristic spectral line intensities, and Gaussian white noise is superimposed on the multi-angle characteristic spectral line intensities, specifically:

[0015] Selecting a plurality of projection angles within a range of 0° to 179°, wherein the interval angles between the plurality of projection angles are the same;

[0016] Determine the number of rays, project the emission coefficient distribution mathematical model from the multiple projection angles, and obtain multi-angle characteristic spectral line intensities;

[0017] The standard deviation of the arc characteristic spectral line image is obtained, and Gaussian white noise is calculated according to the standard deviation of the arc characteristic spectral line image, and the Gaussian white noise is superimposed on the multi-angle characteristic spectral line intensity.

[0018] Furthermore, the algorithm parameters include AwASD-POCS algorithm parameters and PSO algorithm parameters.

[0019] The AwASD-POCS algorithm parameters include but are not limited to arc-detector distance, number and size of voxels, number and size of detectors, and ray casting method;

[0020] The PSO algorithm parameters include but are not limited to inertia factor, learning factor, maximum number of iterations, number of particles and particle dimension.

[0021] Furthermore, the reconstruction parameters are 8 reconstruction parameters of the AwASD-POCS algorithm, and the setting range of reconstruction parameter values ​​is specifically:

[0022] Set the lambda parameter to a value range of 0.1 to 2.5, the lambda_red parameter to a value range of 0.01 to 1, the tviter parameter to a value range of 1 to 300, and the alpha parameter to a value range of 10. -8~50, set the value range of alpha_red parameter to 0.01~1, set the value range of rmax parameter to 0~10, set the value range of maxl2err parameter to 10 -4 ~1000, set the delta parameter value range to 10 -5 ~5.

[0023] Furthermore, the initialization of the particle swarm position and velocity, and combining the multi-angle characteristic spectral line intensity, algorithm parameters and reconstruction parameter value range, calculates the fitness of the particle swarm after initialization, specifically including performing the following steps on each particle of the particle swarm in turn:

[0024] Initialize the particle's position and velocity;

[0025] Calculating the fitness of the particle according to the position and speed of the particle after the initialization;

[0026] Update the global optimal fitness and the global optimal position according to the fitness;

[0027] Repeat the above steps to initialize the next particle until the fitness calculation of the last particle is completed.

[0028] Furthermore, the fitness of the particle is calculated according to the position and speed of the particle after the initialization, specifically:

[0029] According to the position of the initialized particle, combined with the multi-angle characteristic spectral line intensity, algorithm parameters and reconstruction parameter value range, the AwASD-POCS algorithm is used to calculate the reconstruction result;

[0030] Performing median filtering and normalization processing on the reconstruction result;

[0031] Calculating the fitness of the initialized particles according to the processed reconstruction results;

[0032] Record the historical optimal position and historical optimal fitness of the particle.

[0033] Furthermore, the updating of the particle swarm position and velocity and performing iterative calculation, and obtaining the tuning reconstruction parameters according to the iterative result and the fitness of the initialized particle swarm, specifically includes sequentially performing the following steps on each particle of the particle swarm:

[0034] Update the particle's velocity and position;

[0035] Perform iterative calculation according to the updated particle speed and position to obtain fitness;

[0036] Update the global optimal fitness and the global optimal position according to the fitness;

[0037] Draw the particle position distribution diagram and the fitness curve changing with the number of iterations;

[0038] Repeat the above steps to perform the update iterative calculation of the next particle until the update iterative calculation of the last particle is completed;

[0039] According to the global optimal fitness and the global optimal position at the end of the iteration, the tuning reconstruction parameters are obtained.

[0040] Further, the updating of the global optimal fitness and the global optimal position according to the fitness is specifically:

[0041] When performing the first iterative calculation, the fitness obtained by the first iterative calculation is compared with the fitness of the initialized particle swarm, and the smaller fitness is taken as the global optimal fitness, and the position corresponding to the smaller fitness is taken as the global optimal position;

[0042] When performing the second and subsequent iterative calculations, the fitness obtained by the iterative calculation is compared with the fitness obtained by the previous iterative calculation, and the smaller fitness is taken as the global optimal fitness, and the position corresponding to the smaller fitness is taken as the global optimal position.

[0043] Furthermore, it also includes performing the AwASD-POCS algorithm according to the tuned reconstruction parameters to obtain the welding arc reconstruction emission coefficient distribution, specifically:

[0044] The global optimal position is used as a tuned reconstruction parameter to perform the AwASD-POCS algorithm to obtain a welding arc reconstruction emission coefficient with the minimum iteration error, and a distribution diagram of the reconstruction emission coefficient is drawn;

[0045] The error between the reconstructed emission coefficient of the welding arc and the original emission coefficient is analyzed.

[0046] The beneficial effects of the present invention are reflected in: it can automatically select appropriate reconstruction parameters in an adaptive manner according to the characteristics of welding arc spectrum acquisition, and can obtain satisfactory reconstruction results by using a small amount of projection characteristic spectral line intensity information without human intervention, thereby effectively improving the accuracy and automation level of spectral diagnosis, and laying a reliable foundation for subsequent arc spectrum diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0048] Figure 1 A flowchart of a welding arc emission coefficient reconstruction parameter tuning method based on a PSO-AwASD-POCS algorithm provided in an embodiment of the present invention;

[0049] Figure 2 An emission coefficient distribution diagram of a certain row of an arc characteristic spectrum line image provided by an embodiment of the present invention;

[0050] Figure 3 The projection diagram of the emission coefficient distribution with an angle interval of 15° provided by the embodiment of the present invention;

[0051] Figure 4 The initialization particle swarm position distribution diagram and its fitness provided by the embodiment of the present invention;

[0052] Figure 5 A particle swarm position distribution diagram and its fitness at the time of iteration termination provided by an embodiment of the present invention;

[0053] Figure 6 A curve diagram showing the change of fitness with the number of iterations provided in an embodiment of the present invention;

[0054] Figure 7 A distribution diagram of welding arc emission coefficient after reconstruction provided by an embodiment of the present invention;

[0055] Figure 8 This is an error diagram between the original transmission coefficients and the reconstruction results provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The following embodiments of the technical solution of the present invention are described in detail in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only used as examples, and cannot be used to limit the protection scope of the present invention.

[0057] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by those skilled in the art to which the present invention belongs.

[0058] like Figure 1 As shown, the welding arc emission coefficient reconstruction parameter tuning method based on the PSO-AwASD-POCS algorithm includes the following steps:

[0059] S1: Acquire an arc characteristic spectral line image, and construct an emission coefficient distribution mathematical model according to the arc characteristic spectral line image;

[0060] S2: Projecting the emission coefficient distribution mathematical model from different angles to obtain multi-angle characteristic spectral line intensities, and superimposing Gaussian white noise on the multi-angle characteristic spectral line intensities;

[0061] S3: Set algorithm parameters and set the value range of reconstruction parameters;

[0062] S4: Initializing the position and speed of the particle swarm, and calculating the fitness of the particle swarm after the initialization based on the multi-angle characteristic spectral line intensity, algorithm parameters, and reconstruction parameter value range;

[0063] S5: updating the particle swarm position and velocity and performing iterative calculations, and obtaining tuning and reconstruction parameters according to the iterative results and the fitness of the initialized particle swarm.

[0064] Furthermore, in step S1, an arc characteristic spectral line image is obtained, and a mathematical model of emission coefficient distribution is constructed according to the arc characteristic spectral line image, which specifically includes:

[0065] S11: Call the imread function to obtain the arc characteristic spectrum line image according to the specified path;

[0066] S12: calling the cftool toolbox, and constructing a mathematical model of emission coefficient distribution according to the data distribution characteristics of each row of the arc characteristic spectrum line image;

[0067] S13: normalizing the emission coefficient distribution mathematical model.

[0068] Specifically, the arc characteristic line image is collected by a high-speed camera and stored in a specified path. The imread function is called to read the welding arc characteristic line image in the specified path. Since the arc characteristic line image exported by the high-speed camera is in index mode, it is necessary to call the ind2gray function to convert the index mode image into grayscale mode and force the data to be converted into double type for subsequent spectral data processing. After the conversion is completed, the cftool toolbox is called, and the grayscale value of a row of the arc characteristic line image is used as the ordinate. Formula (1) is used to calculate the fitting parameters and obtain the mathematical model of the emission coefficient distribution:

[0069]

[0070] In the formula, ε is the emission coefficient, A1 and A2 are the amplitudes of the emission coefficient distribution curve, b1 and b2 are the centers of the curve, σ1 and σ2 are the standard deviations, and x and y are the horizontal and vertical coordinates, respectively. The max function and min function are called to calculate the maximum and minimum values ​​of the emission coefficient distribution mathematical model, and the emission coefficient distribution mathematical model is normalized using formula (2):

[0071]

[0072] In the formula, ε normalized is the normalized emission coefficient, εmin is the minimum value of the emission coefficient, ε max is the maximum value of the emission coefficient. The emission coefficient distribution of a row of the arc characteristic line image is as follows: Figure 2 shown.

[0073] Further, in step S2, the emission coefficient distribution mathematical model is projected from different angles to obtain multi-angle characteristic spectral line intensities, and Gaussian white noise is superimposed on the multi-angle characteristic spectral line intensities, specifically including:

[0074] S21: selecting a plurality of projection angles within a range of 0° to 179°, wherein the interval angles between the plurality of projection angles are the same;

[0075] S22: Determine the number of rays, project the emission coefficient distribution mathematical model from the multiple projection angles, and obtain multi-angle characteristic spectral line intensities;

[0076] S23: Obtaining the standard deviation of the arc characteristic spectrum line image, and calculating Gaussian white noise according to the standard deviation of the arc characteristic spectrum line image, and superimposing the Gaussian white noise on the multi-angle characteristic spectrum line intensity.

[0077] Specifically, multiple projection angles are selected within the range of 0° to 179°, and the interval angles between the multiple projection angles are the same. In this embodiment, the projection angle interval is set to 15° as an example, and 0°, 15°, 30°, 45°, 60°, 75°, 90°, 105°, 120°, 135°, 150°, and 165° are selected as projection angles. The number of rays is determined according to formula (3):

[0078]

[0079] In the formula, n is the number of rays, W is the number of pixels in the width direction of the arc characteristic spectrum line image, and round indicates rounding operation. The radon function is called to project the emission coefficient distribution mathematical model from different projection angles, such as Figure 3 As shown. According to formula (4), Gaussian white noise is obtained:

[0080] e=SD×randn(n) (4)

[0081] Where SD is the standard deviation of the arc characteristic spectral line image obtained by calling the Noisele function, and randn represents the standard normal distribution pseudo-random number with a generated sequence number of n. The calculated Gaussian white noise is linearly superimposed on the characteristic spectral line intensity at different projection angles to meet the situation of noise interference during the actual spectral line intensity acquisition process.

[0082] Further, in step S3, the algorithm parameters include AwASD-POCS algorithm parameters and PSO algorithm parameters, wherein the AwASD-POCS algorithm parameters include but are not limited to two-dimensional linear set structure parameters such as arc-detector distance, number and size of voxels, number and size of detectors, and ray projection method, and the PSO algorithm parameters include but are not limited to parameters such as inertia factor, learning factor, maximum number of iterations, number of particles, and particle dimension. For the setting of PSO algorithm parameters, the inertia factor is adjusted in a linear decreasing manner to take into account the global and local search capabilities of the PSO algorithm. In this embodiment, the initial inertia factor is set to 0.95, and the termination inertia factor is set to 0.4. At the same time, in order to ensure the calculation accuracy and efficiency of the algorithm, the learning factor is set to 2, the maximum number of iterations is set to 10, and the number of particles is set to 20.

[0083] Preferably, the reconstruction parameters are the 8 reconstruction parameters of the AwASD-POCS algorithm, so the particle dimension of the PSO algorithm is set to 8. According to the reconstruction parameter characteristics of the AwASD-POCS algorithm, the value range of the reconstruction parameters is set: the value range of the lambda parameter is set to 0.1 to 2.5, the value range of the lambda_red parameter is set to 0.01 to 1, the value range of the tviter parameter is set to 1 to 300, and the value range of the alpha parameter is set to 10 -8 ~50, set the value range of alpha_red parameter to 0.01~1, set the value range of rmax parameter to 0~10, set the value range of maxl2err parameter to 10 -4 ~ 1000, set the delta parameter value range to 10 -5 The upper and lower limits of particle velocity are determined by 30% of the difference between the upper and lower limits of the corresponding particle position.

[0084] Further, in step S4, the position and velocity of the particle swarm are initialized, and the fitness of the particle swarm after initialization is calculated in combination with the multi-angle characteristic spectral line intensity, algorithm parameters and reconstruction parameter value range, which specifically includes performing the following steps on each particle of the particle swarm in turn:

[0085] S41: Initialize the position and velocity of the particle;

[0086] S42: Calculating the fitness of the particle according to the position and speed of the particle after the initialization;

[0087] S43: updating the global optimal fitness and the global optimal position according to the fitness;

[0088] S44: Repeat the above steps to initialize the next particle until the fitness calculation of the last particle is completed.

[0089] Specifically, for each particle, its initial position and velocity are calculated by formula (5):

[0090] P(i)=V(i)=P imin +rand×(P imax -P imin ) (5)

[0091] Where P is the particle position, V is the particle velocity, i is the particle dimension, and P imax is the upper bound of the corresponding reconstruction parameter, P imin is the lower bound of the corresponding reconstruction parameter, and rand represents the generation of uniformly distributed pseudo-random numbers between 0 and 1.

[0092] Furthermore, the step S42 of calculating the fitness of the particle according to the position and speed of the initialized particle is specifically as follows:

[0093] S421: According to the position of the initialized particle, combined with the multi-angle characteristic spectral line intensity, algorithm parameters and reconstruction parameter value range, the AwASD-POCS algorithm is used to calculate the reconstruction result;

[0094] S422: performing median filtering and normalization processing on the reconstruction result;

[0095] S423: Calculating the fitness of the initialized particles according to the processed reconstruction results;

[0096] S424: Record the historical optimal position and historical optimal fitness of the particle.

[0097] Specifically, after obtaining the position and velocity of the particle after initialization, the characteristic spectral line intensity at different angles, reconstruction parameters, and two-dimensional linear geometric structure parameters are introduced into the AwASD-POCS algorithm to calculate the reconstruction result. The medfilt2 function is called to perform median filtering on the reconstruction result, and normalization is performed in a manner similar to formula (2). The fitness of the particle is calculated by formula (6):

[0098]

[0099] Where P cost is the fitness of the particle, Z original is the mathematical model of the original emission coefficient distribution, Z reconstruct To reconstruct the result, norm represents the norm of the calculation matrix. If the calculated particle fitness is less than the global optimal fitness, the global optimal fitness and the global optimal position are updated. The position and fitness of the particle swarm after initialization are as follows: Figure 4 shown.

[0100] Furthermore, in step S5, the particle swarm position and velocity are updated and iterative calculation is performed, and the tuning reconstruction parameters are obtained according to the iterative result and the fitness of the initialized particle swarm, which specifically includes sequentially performing the following steps on each particle of the particle swarm:

[0101] S51: Update the velocity and position of the particle;

[0102] S52: performing iterative calculation according to the updated particle speed and position to obtain fitness;

[0103] S53: updating the global optimal fitness and the global optimal position according to the fitness;

[0104] S54: Draw the particle position distribution diagram and the fitness curve changing with the number of iterations;

[0105] S55: Repeat the above steps to perform update iterative calculation of the next particle until the update iterative calculation of the last particle is completed;

[0106] S56: Obtaining tuning reconstruction parameters according to the global optimal fitness and the global optimal position when the iteration is terminated.

[0107] Specifically, the particle velocity is updated using formula (7):

[0108]

[0109] Where V n is the speed of the nth particle, k is the number of iterations, and w max is the initial inertia factor, w min is the termination inertia factor, ITER is the maximum number of iterations, c1 and c2 are learning factors, and Pbest n is the best historical fitness of the nth particle, Gbest is the global best fitness, P n Indicates the position of the nth particle, rand represents the generation of uniformly distributed pseudo-random numbers between 0 and 1. If the calculated particle speed exceeds the upper and lower bounds of the corresponding particle speed, the particle speed is forced to be set to the upper and lower bounds.

[0110] Use formula (8) to update the particle position:

[0111] P n (k+1) =P n (k) +V n (k) (8)

[0112] Where P nrepresents the position of the nth particle, V n is the speed of the nth particle, and k is the number of iterations. If the calculated particle position exceeds the upper and lower bounds of the corresponding particle position, the particle position is forced to be set to the upper and lower bounds.

[0113] After updating the particle position and velocity, the algorithm parameters set in step S3 are brought into the AwASD-POCS algorithm to calculate the reconstruction result, and the fitness of the particle after iterative calculation is calculated using formula (6).

[0114] If the calculated particle fitness is less than the global optimal fitness, the global optimal fitness and the global optimal position are updated as follows:

[0115] When the first iterative calculation is performed, the fitness obtained by the first iterative calculation is compared with the fitness of the particle swarm after initialization, and the smaller fitness is taken as the global optimal fitness, and the position corresponding to the smaller fitness is taken as the global optimal position;

[0116] When performing the second and subsequent iterative calculations, the fitness obtained by the iterative calculation is compared with the fitness obtained by the previous iterative calculation, and the smaller fitness is taken as the global optimal fitness, and the position corresponding to the smaller fitness is taken as the global optimal position.

[0117] Then, according to the particle dimensions, the particle position distribution diagram is drawn in pairs, which is convenient for researchers to further optimize the range of reconstruction parameter values, and the fitness curve with the number of iterations is drawn, which is convenient for researchers to determine the iterative convergence, so as to further adjust the number of particles and the maximum number of iterations and improve the reconstruction efficiency of the algorithm. Among them, the particle position is represented by a solid dot, and the global optimal position is represented by a hollow circle.

[0118] Finally, the tuning reconstruction parameters are obtained according to the global optimal fitness and global optimal position at the end of the iteration. Taking the parameters provided in the above embodiment as an example, the position and fitness of the particle swarm at the end of the iteration are as follows: Figure 5 As shown in the figure, the curve of fitness changing with the number of iterations is as follows Figure 6 As shown in the figure, the final 8 reconstruction parameters are: lambda is 0.1, lambda_red is 1, tviter is 115, alpha is 10 -8 , alpha_red is 0.3719, rmax is 10, maxl2err is 129.7, and delta is 0.5106, at which time the global optimal fitness can be achieved.

[0119] Furthermore, the step also includes performing an AwASD-POCS algorithm according to the tuned reconstruction parameters to obtain a welding arc reconstruction emission coefficient distribution.

[0120] Specifically, the global optimal position is used as the tuned reconstruction parameter, and the characteristic spectral line intensity at different angles, the tuned reconstruction parameters and the two-dimensional linear geometric structure parameters are brought into the AwASD-POCS algorithm to calculate the welding arc reconstruction emission coefficient with the minimum iterative error, and draw the distribution diagram of the reconstruction emission coefficient, as shown in Figure 7 Randomly select two rows of reconstruction results and analyze the error between the original emission coefficients and the reconstruction results, as shown in Figure 8 shown.

[0121] The present invention can automatically select appropriate reconstruction parameters in an adaptive manner according to the characteristics of welding arc spectrum acquisition. Without human intervention, it can use a small amount of projection characteristic spectral line intensity information to obtain satisfactory reconstruction results, thereby effectively improving the accuracy and automation level of spectral diagnosis and laying a reliable foundation for subsequent arc spectrum diagnosis.

[0122] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A welding arc emission coefficient reconstruction parameter tuning method based on PSO-AwASD-POCS algorithm, characterized in that the steps include: Acquire an arc characteristic spectral line image, and construct an emission coefficient distribution mathematical model according to the arc characteristic spectral line image; Projecting the emission coefficient distribution mathematical model from different angles to obtain multi-angle characteristic spectral line intensities, and superimposing Gaussian white noise on the multi-angle characteristic spectral line intensities; Set algorithm parameters and set the value range of reconstruction parameters; Initializing the position and speed of the particle swarm, and calculating the fitness of the particle swarm after the initialization based on the multi-angle characteristic spectral line intensity, algorithm parameters, and reconstruction parameter value range; The particle swarm position and velocity are updated and iterative calculations are performed, and the tuning and reconstruction parameters are obtained according to the iterative results and the fitness of the initialized particle swarm, which specifically includes performing the following steps on each particle of the particle swarm in turn: Update the particle's velocity and position; Perform iterative calculation according to the updated particle speed and position to obtain fitness; Update the global optimal fitness and the global optimal position according to the fitness; Draw the particle position distribution diagram and the fitness curve changing with the number of iterations; Repeat the above steps to perform the update iterative calculation of the next particle until the update iterative calculation of the last particle is completed; According to the global optimal fitness and the global optimal position at the end of the iteration, an optimized reconstruction parameter is obtained; The global optimal fitness and the global optimal position are updated according to the fitness, specifically: When performing the first iterative calculation, the fitness obtained by the first iterative calculation is compared with the fitness of the initialized particle swarm, and the smaller fitness is taken as the global optimal fitness, and the position corresponding to the smaller fitness is taken as the global optimal position; When performing the second and subsequent iterative calculations, the fitness obtained by the iterative calculation is compared with the fitness obtained by the previous iterative calculation, and the smaller fitness is taken as the global optimal fitness, and the position corresponding to the smaller fitness is taken as the global optimal position.

2. The welding arc emission coefficient reconstruction parameter tuning method based on the PSO-AwASD-POCS algorithm according to claim 1 is characterized in that: The arc characteristic spectral line image is obtained, and a mathematical model of emission coefficient distribution is constructed according to the arc characteristic spectral line image, specifically: Call the imread function to obtain the arc characteristic spectrum line image according to the specified path; Calling the cftool toolbox, and constructing a mathematical model of emission coefficient distribution according to the data distribution characteristics of each row of the arc characteristic spectrum line image; The emission coefficient distribution mathematical model is normalized.

3. The welding arc emission coefficient reconstruction parameter tuning method based on the PSO-AwASD-POCS algorithm according to claim 1 is characterized in that: The projecting of the emission coefficient distribution mathematical model from different angles to obtain multi-angle characteristic spectral line intensities and superimposing Gaussian white noise on the multi-angle characteristic spectral line intensities are specifically as follows: Selecting a plurality of projection angles within a range of 0° to 179°, wherein the interval angles between the plurality of projection angles are the same; Determine the number of rays, project the emission coefficient distribution mathematical model from the multiple projection angles, and obtain multi-angle characteristic spectral line intensities; The standard deviation of the arc characteristic spectral line image is obtained, and Gaussian white noise is calculated according to the standard deviation of the arc characteristic spectral line image, and the Gaussian white noise is superimposed on the multi-angle characteristic spectral line intensity.

4. The welding arc emission coefficient reconstruction parameter tuning method based on the PSO-AwASD-POCS algorithm according to claim 1 is characterized in that: The algorithm parameters include AwASD-POCS algorithm parameters and PSO algorithm parameters. The AwASD-POCS algorithm parameters include but are not limited to arc-detector distance, number and size of voxels, number and size of detectors, and ray casting method; The PSO algorithm parameters include but are not limited to inertia factor, learning factor, maximum number of iterations, number of particles and particle dimension.

5. The welding arc emission coefficient reconstruction parameter tuning method based on the PSO-AwASD-POCS algorithm according to claim 4 is characterized in that: The reconstruction parameters are 8 reconstruction parameters of the AwASD-POCS algorithm, and the value range of the set reconstruction parameters is specifically: Set the lambda parameter to a value range of 0.1 to 2.5, the lambda_red parameter to a value range of 0.01 to 1, the tviter parameter to a value range of 1 to 300, and the alpha parameter to a value range of 10. -8 ~50, set the value range of alpha_red parameter to 0.01~1, set the value range of rmax parameter to 0~10, set the value range of maxl2err parameter to 10 -4 ~1000, set the delta parameter value range to 10 -5 ~5.

6. The welding arc emission coefficient reconstruction parameter tuning method based on the PSO-AwASD-POCS algorithm according to claim 1, characterized in that: The initialization of the particle swarm position and velocity, and the calculation of the fitness of the particle swarm after initialization in combination with the multi-angle characteristic spectral line intensity, algorithm parameters and reconstruction parameter value range, specifically includes sequentially performing the following steps on each particle of the particle swarm: Initialize the particle's position and velocity; Calculating the fitness of the particle according to the position and speed of the particle after the initialization; Update the global optimal fitness and the global optimal position according to the fitness; Repeat the above steps to initialize the next particle until the fitness calculation of the last particle is completed.

7. The welding arc emission coefficient reconstruction parameter tuning method based on the PSO-AwASD-POCS algorithm according to claim 6 is characterized in that: The fitness of the particle is calculated according to the position and speed of the particle after the initialization, specifically: According to the position of the initialized particle, combined with the multi-angle characteristic spectral line intensity, algorithm parameters and reconstruction parameter value range, the AwASD-POCS algorithm is used to calculate the reconstruction result; Performing median filtering and normalization processing on the reconstruction result; Calculating the fitness of the initialized particles according to the processed reconstruction results; Record the historical optimal position and historical optimal fitness of the particle.

8. The welding arc emission coefficient reconstruction parameter tuning method based on the PSO-AwASD-POCS algorithm according to claim 1, characterized in that: It also includes performing the AwASD-POCS algorithm according to the tuned reconstruction parameters to obtain the welding arc reconstruction emission coefficient distribution, specifically: The global optimal position is used as a tuned reconstruction parameter to perform the AwASD-POCS algorithm to obtain a welding arc reconstruction emission coefficient with the minimum iteration error, and a distribution diagram of the reconstruction emission coefficient is drawn; The error between the reconstructed emission coefficient of the welding arc and the original emission coefficient is analyzed.