A combustion flow field tomographic reconstruction method based on temperature zone division and spectral line selection

Abstract

The present disclosure relates to the technical field of optical tomography, and provides a combustion flow field tomographic reconstruction method based on temperature region division and spectral line selection, comprising: selecting a plurality of absorption spectral lines for a target molecule in a to-be-measured combustion flow field within an absorption spectral line range of a set waveband; reconstructing an absorption coefficient distribution of each absorption spectral line and preliminarily reconstructing a temperature distribution based on a linear inversion algorithm; constructing a candidate spectral line combination set based on the selected plurality of absorption spectral lines; evaluating a temperature measurement uncertainty of different spectral line combinations in the candidate spectral line combination set in different temperature ranges; dividing a temperature interval based on the temperature distribution, selecting an optimal spectral line combination for a grid corresponding to each temperature interval based on the temperature measurement uncertainty, and reconstructing a temperature and concentration distribution of the combustion flow field based on an absorption coefficient distribution of the optimal spectral line combination. The optimal temperature measurement spectral line combination with anti-noise and anti-interference capability is selected for the combustion flow field in different temperature regions, and global accurate and stable tomographic calculation is realized.

Description

Technical Field

[0001] This disclosure belongs to the field of optical tomography technology, and particularly relates to a combustion flow field tomography reconstruction method based on temperature zone division and spectral line selection. Background Technology

[0002] Laser absorption spectroscopy (LAS) is widely used for measuring temperature and component concentrations in complex combustion flow fields due to its advantages such as in-situ, quantitative, non-invasive, and high-frequency response. Combined with computational tomography, LAS can achieve spatiotemporally resolved measurements of thermophysical parameters in non-uniform combustion flow fields. Among these methods, the double-line thermometry method is currently the primary method for tomographic reconstruction of combustion flow fields due to its simple optical system and high data processing efficiency. When the temperature distribution is reconstructed using the double-line method, the concentration distribution can be quantitatively calculated using the absorption coefficient distribution of one of the absorption lines and Beer-Lambert's law. However, the spectral pairs selected in the traditional double-line thermometry method only have sufficient temperature sensitivity within a limited temperature range, making it difficult to cover combustion flow fields with high temperature gradients or large temperature spans. Therefore, the double-line thermometry method is sensitive to experimental noise and tomographic inversion errors, easily producing non-physical artifacts and local calculation distortions in the reconstruction results. This problem is more pronounced under finite or sparse projection conditions.

[0003] In recent years, multi-spectral absorption spectroscopy has gained increasing attention due to its ability to improve temperature inversion accuracy by utilizing multi-spectral information, particularly demonstrating strong applicability in nonlinear tomographic reconstruction methods. However, multi-spectral nonlinear tomographic methods generally suffer from high computational complexity and low reconstruction efficiency, making them unsuitable for high-resolution or large-scale reconstruction scenarios. Linear multispectral tomography is another computational method that integrates multi-spectral information, offering significant advantages in computational efficiency, but it requires high accuracy in reconstructing the absorption coefficient of each absorption line. When using multi-spectral tomography for temperature measurement, the inversion accuracy of each spectral line also significantly impacts the final temperature measurement result. Under conditions of large computational errors, multi-spectral information can even amplify the temperature measurement uncertainty. Therefore, there is an urgent need for an efficient and high-precision multi-spectral combustion flow field tomographic temperature measurement method to achieve wide-temperature-range, high-fidelity multi-parameter tomographic reconstruction of combustion flow fields, meeting the diagnostic needs of complex combustion environments. Summary of the Invention

[0004] To address the aforementioned issues, this disclosure provides a combustion flow field tomographic reconstruction method based on temperature zone division and spectral line selection. The optimal combination of thermometric spectral lines with strong noise and interference resistance is selected for each temperature zone of the combustion flow field to achieve globally accurate and stable tomographic calculations.

[0005] This disclosure provides a combustion flow field tomography reconstruction method based on temperature zone division and spectral line selection, including:

[0006] Within the absorption spectral range of the set band, multiple absorption spectral lines are selected for the target molecules in the combustion flow field to be tested. Among them, the selected multiple absorption spectral lines have different temperature sensitivities and cover a temperature range of greater than or equal to 296K and less than or equal to 2000K.

[0007] Based on the linear inversion algorithm, the absorption coefficient distribution of each absorption line is reconstructed, and the temperature distribution is initially reconstructed.

[0008] Based on the selected multiple absorption lines, a set of candidate spectral line combinations is constructed.

[0009] Under a given noise level, evaluate the temperature measurement uncertainty of different spectral line combinations in the candidate spectral line combination set within different temperature ranges;

[0010] Temperature ranges are divided based on the preliminary reconstructed temperature distribution. The optimal spectral line combination is selected for each temperature range based on the temperature measurement uncertainty. The temperature and concentration distribution of the combustion flow field are reconstructed based on the absorption coefficient distribution of each absorption line of the selected optimal spectral line combination.

[0011] Furthermore, based on the initially reconstructed temperature distribution, temperature intervals are divided. Based on the temperature measurement uncertainty, the optimal spectral line combination is selected for each temperature interval's corresponding grid. Based on the absorption coefficient distribution of each absorption line in the selected optimal spectral line combination, the temperature and concentration distribution of the combustion flow field are reconstructed, including:

[0012] Step 1: Based on the temperature measurement uncertainty, select the combination of temperature measurement spectral lines with the highest reliability over a wide temperature range. Using the selected combination of spectral lines and the absorption coefficient distribution of the absorption spectral lines, the temperature and concentration distribution of the combustion flow field are initially reconstructed.

[0013] Step 2: Divide the temperature range of the combustion flow field reconstructed in the previous step; based on the temperature measurement uncertainty, select the corresponding spectral line combination for the grid corresponding to each temperature range; and use the selected spectral line combination and the absorption coefficient distribution of the absorption spectral line to reconstruct the temperature and concentration distribution of the combustion flow field in each temperature range.

[0014] Step 3: Go back to step 2 until you get the best result.

[0015] Furthermore, after the first step and before the second step, the method further includes:

[0016] The temperature and concentration distribution of the initially reconstructed combustion flow field are input into the trained neural network model, which outputs the temperature distribution of the combustion flow field without outliers. The trained neural network model is used to identify reconstructed outliers and provides reference values ​​for outliers based on the physical information of the smoothness of the combustion flow field distribution.

[0017] After the second step and before the third step, the method further includes:

[0018] The temperature distribution of the combustion flow field reconstructed in different temperature ranges is input into the trained neural network model, which outputs the temperature distribution of the combustion flow field reconstructed in different temperature ranges without outliers.

[0019] Furthermore, the neural network model consists of four convolutional layers and four fully connected layers, wherein the input and output of the convolutional layers introduce skip connections.

[0020] Furthermore, within the defined absorption spectral range, multiple absorption lines are selected for the target molecules in the combustion flow field to be measured, including:

[0021] Within the set absorption spectral range, multiple absorption lines are selected for water molecules in the combustion flow field to be tested.

[0022] Furthermore, within the defined absorption spectral range, multiple absorption lines are selected for water molecules in the combustion flow field under test, including:

[0023] Four absorption lines were selected from the near-infrared absorption spectrum of water molecules in the combustion flow field to be tested.

[0024] Furthermore, the center frequencies of the four absorption lines are 7161.4 cm⁻¹. -1 7185.597 cm -1 7444.352cm -1 and 6807.82 cm -1 .

[0025] Furthermore, based on the initially reconstructed temperature distribution, temperature intervals are divided. Based on the temperature measurement uncertainty, the optimal spectral line combination is selected for each temperature interval's corresponding grid. Based on the absorption coefficient distribution of each absorption line in the selected optimal spectral line combination, the temperature and concentration distribution of the combustion flow field are reconstructed, including:

[0026] Step 1: For the aforementioned temperature range, 7185.597 cm was selected. -1 + 6807.82 cm -1 The spectral line combination, based on 7185.597 cm⁻¹ -1 The absorption coefficient distribution and 6807.82 cm -1 The absorption coefficient distribution was used to preliminarily reconstruct the temperature and concentration distribution of the combustion flow field;

[0027] Step 2: Divide the temperature distribution calculation results from Step 1 into two temperature ranges: T ≥ 1000 K and T < 1000 K. For the range above 1000 K, a grid size of 7444.36 cm is selected. -1+ 6807.82 cm -1 The spectral line combination, based on 7444.36cm -1 The absorption coefficient distribution and 6807.82 cm -1 The temperature and concentration distribution of the combustion flow field were reconstructed using the absorption coefficient distribution. A 7161.4 cm² grid was used for meshes below 1000 K. -1 + 7185.597 cm -1 + 7444.36 cm -1 The spectral line combination is based on 77161.4 cm. -1 The absorption coefficient distribution, 7185.597 cm⁻¹ -1 The absorption coefficient distribution and 6807.82 cm -1 The absorption coefficient distribution reconstructs the temperature and concentration distribution of the combustion flow field;

[0028] Step 3: Divide the calculation results from Step 2 into three temperature ranges: T < 600 K, 600 K ≤ T < 1000 K, and T > 1000 K. For the range T < 600 K, a grid of 7161.4 cm is used. -1 + 7185.597 cm -1 The spectral line combination, based on 7161.4 cm -1 The absorption coefficient distribution and 7185.597 cm -1 The temperature and concentration distribution of the combustion flow field are reconstructed from the absorption coefficient distribution. The grids for 600 K ≤ T < 1000 K and T ≥ 1000 K retain the reconstruction results from the second step.

[0029] Furthermore, the noise level is set to a range of greater than or equal to 2% and less than or equal to 8%.

[0030] Furthermore, the linear inversion algorithm is the Tikhonov Regularization algorithm.

[0031] Compared with the prior art, this disclosure has the following advantages:

[0032] 1. Multiple absorption lines, such as four, can be selected to have different temperature sensitivities, covering the range from room temperature to high temperature, thus solving the problem of insufficient sensitivity in traditional dual-line temperature measurement.

[0033] 2. A three-step temperature zone division and spectral line selection strategy is proposed. Without prior information on flow field temperature, the optimal spectral line combination can be applied to a specific temperature zone, reducing the problem of temperature measurement noise sensitivity caused by the direct superposition of multiple spectral lines.

[0034] 3. By identifying abnormal reconstructed values ​​through neural networks, the problem of local distortion in tomographic reconstruction can be effectively solved.

[0035] Other features and advantages of this disclosure will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the disclosure. The objects and other advantages of this disclosure may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 A flowchart of a combustion flow field tomography reconstruction method based on temperature zone division and spectral line selection according to an embodiment of the present disclosure is shown;

[0038] Figure 2 This invention illustrates a neural network model framework for identifying outliers in tomographic reconstruction, as shown in an embodiment of the present disclosure.

[0039] Figure 3 The simulation results of the temperature noise resistance uncertainty of eight candidate spectral line combinations in the embodiments of this disclosure are shown.

[0040] Figure 4 A flowchart illustrating the three-step temperature range division and spectral line selection strategy proposed in this disclosure is shown.

[0041] Figure 5 The simulation results of tomographic reconstruction of the combustion flow field model with a double Gaussian peak configuration using different spectral line combination schemes are shown in the embodiments of this disclosure.

[0042] Figure 6 The results of temperature and water molecule concentration reconstruction for experimental testing of the McKenna combustion flame according to embodiments of this disclosure are shown.

[0043] Figure 7 A radial comparison is shown between thermocouple measurement data from embodiments of the present disclosure and experimental tomography results with different spectral line combinations. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0045] Figure 1 This diagram illustrates a flow chart of a combustion flow field tomography reconstruction method based on temperature zone division and spectral line selection according to an embodiment of the present disclosure. The method, based on temperature zone division and spectral line selection, specifically includes the following steps:

[0046] Step 101: Within the absorption spectral range of the set band, select multiple absorption spectral lines for the target molecules in the combustion flow field to be tested.

[0047] In step 101, there are three or more absorption lines selected. These lines have different temperature sensitivities and cover a temperature range of ≥296K and ≤2000K, covering room temperature (approximately 296K) to high temperature (approximately 2000K). The purpose is to provide multispectral information for the tomographic reconstruction of combustion flow fields over a wide temperature range.

[0048] Specifically, water molecules, the main combustion product of hydrocarbon fuels, can be selected as the target molecules. The wavelength range can be set to the near-infrared band. In this case, step 101 involves selecting multiple absorption lines for water molecules in the combustion flow field to be measured within the absorption spectral range of the near-infrared band.

[0049] Step 102: Based on the linear inversion algorithm, reconstruct the absorption coefficient distribution of each absorption spectral line and initially reconstruct the temperature distribution.

[0050] This step 102 specifically includes:

[0051] Step 102-1: Use a linear function to fit the data and obtain the integrated absorbance of each absorption line at different projection positions.

[0052] Specifically, the linear function can be Voigt.

[0053] When the frequency is v [cm -1 When the laser beam passes through the target reconstruction region, its intensity along the optical path changes due to the absorption effect of water molecules. L [cm] attenuation. This process can be quantitatively described by Beer-Lambert's law as integral absorbance, which can be expressed as the following formula (1):

[0054] (1)

[0055] in: [cm -1 ] is the first i The integrated absorbance of the beam path; For the first i The intensity of incident light in the beam path; For the first i The intensity of transmitted light in the beam path; [cm-2 [Frequency] v [cm -1 The laser beam passed through the first j The absorption coefficient of each grid cell. [cm -2 ] = J represents the total number of grid cells obtained after discretizing the target reconstruction region; [atm] is the number j Local pressure in each grid; To absorb the gas in the first j Mole fraction of each grid cell; [cm -2 atm -1 ] is the first j The intensity of the spectral lines in each grid depends on temperature. For the first i The beam of light passes through the first j Optical path of each grid.

[0056] Step 102-2: Determine the projection weight matrix based on the optical path arrangement.

[0057] Specifically, the element values ​​in the projection weight matrix are the path lengths of the laser beam through a specific grid. Once the optical path arrangement is determined, the projection weight matrix can be calculated using algebraic geometric relationships.

[0058] Step 102-3: Based on the obtained integral absorbance and the determined projection weight matrix, the absorption coefficient distribution of each absorption line is obtained using a linear inversion algorithm.

[0059] Specifically, the linear inversion algorithm can be the Tikhonov Regularization algorithm.

[0060] The absorption coefficients of different absorption lines were solved using Tikhonov Regularization. α ν Distribution, i.e., calculating the least squares solution to the following equation:

[0061]

[0062] in: It is a regularization term. μ It is a weighting factor for the degree of regularization, which can be dynamically determined using the L-curve method; W It is the projection weight matrix. The frequency is v The integral absorbance vector, The frequency is v The absorption coefficient vector.

[0063] Step 103: Based on the selected multiple absorption lines, construct a set of candidate spectral line combinations;

[0064] Step 104: Under the given noise level, evaluate the temperature measurement uncertainty of different spectral line combinations in the candidate spectral line combination set within different temperature ranges;

[0065] Specifically, the noise level is set to a range of greater than or equal to 2% and less than or equal to 8%. In this disclosure, the noise level used to evaluate the noise resistance of different spectral line combination methods is set to 5%, which can be used to simulate the local absorption coefficient error caused by a combination of factors such as electronic noise, spectral fitting error, and absorption coefficient tomographic inversion.

[0066] Temperature measurement uncertainty reflects the noise resistance of spectral line combinations within a corresponding temperature range; the smaller the value, the stronger the noise resistance.

[0067] Step 105: Divide the temperature range based on the preliminary reconstructed temperature distribution, select the optimal spectral line combination for each temperature range based on the temperature measurement uncertainty, and reconstruct the temperature and concentration distribution of the combustion flow field based on the absorption coefficient distribution of each absorption spectral line of the selected optimal spectral line combination.

[0068] After obtaining the absorption coefficients of different absorption lines within each grid through inversion, the temperature and concentration values ​​of each grid are calculated using the absorption coefficients of the absorption lines included in the optimal spectral combination selected for the grid, through the Boltzmann plot method. The calculation formula is as follows:

[0069]

[0070] In equation (3), A The integral absorption coefficient, S ( T 0) [cm -2 atm -1 [For temperature dependent] T Spectral line intensity of 0 T 0 represents the reference temperature of 296 K. h is Planck's constant. c At the speed of light, k Here, E' is the Boltzmann constant, and E' represents the lower energy level of the spectral line. The center frequency of the beam. For the partition function, P For pressure, L For optical path, X For concentration, T The temperature of the grid to be calculated.

[0071] By fitting a first-order polynomial of ln[A / S(T0)] to the low-state energy levels of different absorption spectra, the temperature and concentration values ​​can be inferred from the fitting slope and intercept, respectively.

[0072] Specifically, step 105 includes the following three steps:

[0073] Step 1: Based on the temperature measurement uncertainty, select the combination of temperature measurement spectral lines with the highest reliability over a wide temperature range. Using the selected combination of spectral lines and the absorption coefficient distribution of the absorption spectral lines, the temperature and concentration distribution of the combustion flow field are initially reconstructed.

[0074] Step 2: Divide the temperature range of the combustion flow field reconstructed in the previous step; based on the temperature measurement uncertainty, select the corresponding spectral line combination for the grid corresponding to each temperature range; and use the selected spectral line combination and the absorption coefficient distribution of the absorption spectral line to reconstruct the temperature and concentration distribution of the combustion flow field in each temperature range.

[0075] Step 3: Go back to step 2 until you get the best result.

[0076] By using the above three-step temperature zone division and the spectral selection strategy based on the temperature measurement noise resistance capability within different temperature ranges, under the prior condition of unknown flow field temperature distribution, multiple rounds of temperature zone division can be autonomously implemented. For flow field grids in different temperature zones, the above-mentioned spectral line combination scheme with the best noise resistance capability within a specific temperature range can be selected to calculate the temperature and component concentration values ​​of the current grid.

[0077] Because the Boltzmann chart method for temperature measurement, which integrates all candidate spectral lines, is heavily influenced by experimental noise and chromatographic errors, especially in the low and high temperature ranges, the calculation results are significantly affected by experimental noise and chromatographic errors. Therefore, this invention selects the optimal combination of absorption spectral lines for each grid to achieve high-precision temperature measurement calculations over a wide temperature range.

[0078] Furthermore, during testing of the proposed three-step temperature zone division and spectral line selection strategy, it was found that under complex flow field distributions or finite projections, the reconstruction error of the local absorption coefficient exceeded 5%, even reaching 10%, causing some local anomalies in the first-step reconstruction. These local anomalies led to errors in the second-step temperature zone division, resulting in the selection of inappropriate spectral line combinations for calculation. Therefore, these local anomalies were gradually retained in the final reconstruction result. To address this problem, this invention constructs a neural network model to identify reconstruction anomalies and, based on the physical information of smooth flow field distribution, provides a reference value for the anomalies to classify the grid temperature zones.

[0079] The neural network model framework for anomaly recognition constructed in this invention is as follows: Figure 2 As shown, the neural network model consists of four convolutional layers and four fully connected layers. Skip connections are introduced into the input and output of the convolutional layers to retain more original information.

[0080] In this disclosed scheme, the training set of the neural network model contains 5000 images simulating the temperature distribution of combustion flow fields, each image being a 55×55 pixel two-dimensional grid. The original images are normal images. To construct a supervised learning task for anomaly point identification, up to 100 anomaly points are randomly added to each image. The location and number of anomaly points are randomized within each image, exhibiting a certain degree of distribution diversity. The values ​​of the anomaly values ​​differ significantly from the background of the original image, simulating local mutations or anomalous perturbations in the image, helping the model learn pixel-level anomaly detection capabilities. In this neural network model, the anomaly values ​​can take any value within the ranges of -2000 K to -300 K and 300 K to 2000 K.

[0081] After the trained neural network model is inserted into the first and second steps of the three-step temperature range division and spectral line selection strategy for reconstruction, the problem of local distortion in tomographic reconstruction can be effectively solved.

[0082] Specifically, after the first step mentioned above, the temperature distribution of the initially reconstructed combustion flow field is input into the trained neural network model, and the temperature distribution of the combustion flow field without outliers is output. The trained neural network model is used to identify reconstructed outliers and to provide reference values ​​for outliers based on the physical information of the smoothness of the combustion flow field distribution.

[0083] After the second step mentioned above, the temperature distribution of the combustion flow field reconstructed by temperature interval is input into the trained neural network model, and the temperature distribution of the combustion flow field without outliers reconstructed by temperature interval is output.

[0084] By using the above method, the temperature reconstruction results are input into the model, and the output results without local anomalies are obtained, thereby achieving accurate temperature zone classification for each grid.

[0085] In the scheme of this embodiment, no prior information on flow field temperature is required. The optimal spectral line combination can be applied to a specific temperature range, making full and effective use of multispectral information to achieve global high-precision tomographic reconstruction of combustion flow field with a wide temperature range.

[0086] The following example illustrates the above method using a water molecule as the target molecule, the infrared band as the wavelength, and a selection of 4 absorption spectral lines.

[0087] 1) Within the near-infrared absorption spectrum, the four absorption lines described are the absorption lines of water molecules in the flow field under test. Water molecules, the main combustion product of hydrocarbon fuels, are selected as the target, and the thermometric absorption line is 7161.4 cm⁻¹. -1 7185.597cm -1 7444.352 cm -1 and 6807.82 cm -1It exhibits sufficient temperature sensitivity within the temperature range of hydrocarbon fuel combustion flow field. The spectral parameters of each absorption line are shown in Table 1.

[0088] Table 1

[0089]

[0090] Where: S represents the temperature-dependent spectral line intensity, E" represents the lower energy level, and e represents a power of 10.

[0091] 2) Based on the linear inversion algorithm, reconstruct the absorption coefficient distribution of the above four absorption lines.

[0092] 3) Construct a set of candidate spectral line combinations from the four selected absorption lines, which includes eight candidate spectral line combination schemes, as shown in Table 2.

[0093] Table 2

[0094]

[0095] 4) The Boltzmann spectral thermometry method was used to evaluate the temperature measurement uncertainty of different spectral line combinations in the range of 296 K to 2000 K under certain noise conditions.

[0096] To evaluate the accuracy of temperature measurement calculations for various spectral line combinations under noise interference over a wide temperature range, simulations were performed. The absorption coefficients of each absorption line were calculated using the HITRAN database within a temperature range of 296 K to 2000 K and a water molecule concentration of 5%. Then, 5% random Gaussian noise was added to each absorption coefficient, and the temperature measurement uncertainty for each spectral line combination was calculated using the Boltzmann plot method. The simulation results are as follows: Figure 3 As shown, Figure 3 The horizontal axis represents the set temperature value, and the vertical axis represents the calculated temperature. The shaded area represents the uncertainty range of 1000 calculations. Simulation results show that, although the selected spectral lines cover room temperature to high temperatures, the four-wavelength Boltzmann chart thermometry exhibits extremely high uncertainty below 600 K and above 1200 K. Furthermore, the spectral line combination 7185.597 cm⁻¹, commonly used for combustion flow field tomography diagnostics, can be observed. -1 + 7444.36 cm -1 Above 1500 K, the uncertainty is significant, making it unsuitable for high-temperature conditions. Nevertheless, simulation results show that some spectral line combination schemes have high temperature measurement noise resistance within a specific temperature range.

[0097] 5) A three-step temperature zone division and spectral line selection strategy is adopted to apply the optimal spectral line combination to the calculation of a specific temperature range, and a method is proposed to make full and reasonable use of multi-spectral line information to realize combustion flow field tomography calculation over a wide temperature range.

[0098] The implementation process of the three-step temperature zone division and spectral line selection strategy is as follows: Figure 4 As shown.

[0099] The first step is to select 7185.597 cm. -1 + 6807.82 cm -1 The spectral lines are combined to perform global temperature calculation; that is, to perform preliminary reconstruction. Then the reconstruction results are input into the neural network, and the output of the neural network is used as the input for the second step.

[0100] The second step divides the temperature calculation results output from the first step into two temperature ranges: T ≥ 1000 K and T < 1000 K. For the range above 1000 K, a grid of 7444.36 cm is used. -1 + 6807.82 cm -1 The temperature was recalculated using spectral line combinations, with a grid size of 7161.4 cm⁻¹ used for grids below 1000 K. -1 + 7185.597 cm -1 + 7444.36 cm -1 The temperature is recalculated by combining the spectral lines; at this point, the temperature intervals are reconstructed, and the reconstruction results of the temperature intervals are input into the neural network. The output of the neural network is used as the input for the third step.

[0101] The third step divides the output of the second step into three temperature ranges: T < 600 K, 600 K ≤ T < 1000 K, and T ≥ 1000 K. The grid for T < 600 K is 7161.4 cm². -1 + 7185.597 cm -1 The spectral line combinations are recalculated, and the grids for 600 K ≤ T < 1000 K and T ≥ 1000 K retain the calculation results from the second step.

[0102] By performing multiple rounds of reconstruction in this way, and making full and effective use of multispectral information, a global high-precision tomographic reconstruction of the combustion flow field with a wide temperature range can be achieved.

[0103] Comparison Example

[0104] First, a combustion flow field tomography reconstruction method based on temperature zone division and spectral line selection proposed in this disclosure is simulated and tested, and compared with other spectral line combination schemes. The flow field model used for testing is a double Gaussian peak, with the same configuration for temperature and water molecule concentration. The temperature distribution ranges from 296 K to 2000 K, and the water molecule concentration ranges from 0.01 to 0.15. The target reconstruction region is discretized into 55 × 55 grid cells with a grid resolution of 1 cm. Four projection angles are used, with 55 beams evenly spaced at each angle. The specific simulation process includes the following steps:

[0105] Step 1: Simulate the absorbance of each grid in the simulated flow field model using the HITRAN database.

[0106] Step 2, based on the absorption weight matrix determined by the optical path arrangement. W The element sizes in the matrix are used to calculate the absorption spectra of all beams passing through the target reconstruction region.

[0107] Step 3: Fit the simulated absorption spectrum using the Voigt spectral model to obtain the integrated absorbance of the target absorption line.

[0108] Step 4: Using the Tikhonov Regularization calculation method and the fitting method from Step 3, the integral absorbance distribution is used to invert the absorption coefficient distribution.

[0109] For each absorption line, steps 1 to 4 are performed until the absorption coefficient distribution of each absorption line is obtained.

[0110] Step 5: Input the calculated absorption coefficient distribution of each absorption line into the temperature calculation framework for adaptively dividing temperature zones and selecting absorption lines proposed in this invention to perform partitioned inversion of the flow field temperature distribution.

[0111] Following the simulation reconstruction steps described above, the temperature distribution of the double-Gaussian peak combustion flow field used for testing was reconstructed, and the results are as follows: Figure 5 As shown in the figure. Simulation results demonstrate that the combustion flow field tomography reconstruction method based on temperature zone division and spectral line selection proposed in this invention effectively suppresses reconstruction artifacts and improves reconstruction accuracy across the entire temperature range. Under the simulated optical path arrangement conditions described above, the reconstruction error of this invention is 1.5%, which is more than 50% lower than other spectral line combination schemes.

[0112] Verification Example

[0113] To verify that the combustion flow field tomography reconstruction method based on temperature zone division and spectral line selection proposed in this disclosure can also achieve high-precision tomography reconstruction over a wide temperature zone in practical applications, this embodiment provides further explanation through experimental measurement.

[0114] The experiment measured a planar flame produced by a standard McKenna burner, with the measurement height 5 mm above the burner. Four projection angles were used, each with 55 equally spaced parallel beams, achieving a resolution of 2 mm. The multi-angle and multi-beam configuration was achieved by controlling the combined beam using a rotating platform and a translation stage. The flame fuel was a premixed CH4 / air gas, with a methane flow rate of 1.733 L / min and an air flow rate of 20.63 L / min, resulting in an equivalence ratio of 0.8. To stabilize the flame, a nitrogen escort gas flow rate of 20 L / min was used. Four lasers with center wavelengths of 1396 nm, 1392 nm, 1343 nm, and 1469 nm were used, and water molecules at 7161.4 cm⁻¹ were measured using time-division multiplexing. -1 7185.597 cm -1 7444.352 cm -1 and 6807.82 cm -1 The four absorption lines at that location.

[0115] The experimental absorption spectrum was obtained by fitting the detected laser intensity to a baseline. Following steps 3 to 5 in the control example, the experimental data were processed to reconstruct the two-dimensional temperature distribution 5 mm above the McKenna burner. The experimental reconstruction results for different spectral line selection schemes are shown below. Figure 6 As shown.

[0116] Experimental results show that the combustion flow field tomography reconstruction method based on temperature zone division and spectral line selection proposed in this disclosure can effectively suppress tomography reconstruction artifacts and restore the temperature and concentration flatness characteristics of the central region of the McKenna combustion flame in actual combustion diagnostic applications.

[0117] To verify the experimental reconstruction effect of the combustion flow field tomography reconstruction method based on temperature zone division and spectral line selection proposed in this disclosure, thermocouple measurements were performed. The radial distance at the same height above the burner was measured. Y The temperature distribution at 0 mm was analyzed and compared with the experimental reconstruction results, such as... Figure 7 As shown in the figure. The results show that the present invention effectively reconstructs the temperature of the combustion flame center region, and the thermal gradient boundary from the high temperature center to the environment is closer to the trend of thermocouple measurement results, realizing high-precision two-dimensional tomographic reconstruction over a wide temperature range from room temperature to high temperature (~1800 K).

[0118] The above-mentioned tomographic reconstruction simulation and experimental measurements both show that the method disclosed herein can effectively suppress reconstruction artifacts and achieve high-precision combustion flow field tomographic reconstruction across the entire temperature range.

[0119] Although the present disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure.

Claims

1. A combustion flow field tomography reconstruction method based on temperature zone division and spectral line selection, characterized in that, The method comprises the following steps: In the range of absorption spectral lines in a set wave band, a plurality of absorption spectral lines are selected for a target molecule in a to-be-detected combustion flow field, wherein the selected plurality of absorption spectral lines have different temperature sensitivities and cover a temperature width range of greater than or equal to 296K and less than or equal to 2000K; Based on a linear inversion algorithm, an absorption coefficient distribution of each absorption spectral line is reconstructed, and a temperature distribution is preliminarily reconstructed; Based on the selected plurality of absorption spectral lines, a candidate spectral line combination set is constructed; In a set noise level, the temperature measurement uncertainty of different spectral line combinations in different temperature ranges in the candidate spectral line combination set is evaluated; Based on the preliminarily reconstructed temperature distribution, a temperature interval is divided, and the optimal spectral line combination is selected for a grid corresponding to each temperature interval based on the temperature measurement uncertainty; and based on the absorption coefficient distribution of each absorption spectral line of the selected optimal spectral line combination, the temperature and concentration distribution of the combustion flow field is reconstructed, including:

2. The method of claim 1, wherein, The range of the set noise level is greater than or equal to 2% and less than or equal to 8%.

3. The method of claim 1, wherein, The linear inversion algorithm is a Tikhonov Regularization algorithm. The neural network model is composed of four convolutional layers and four fully connected layers, wherein the input and output of the convolutional layers introduce a skip connection.

4. The method of claim 3, wherein, In the range of absorption spectral lines in a set wave band, a plurality of absorption spectral lines are selected for a target molecule in a to-be-detected combustion flow field, including: In the range of absorption spectral lines in a set wave band, a plurality of absorption spectral lines are selected for water molecules in a to-be-detected combustion flow field.

5. The method of claim 4, wherein, 4 The central frequencies of the absorption lines are 7161.4 cm -1 -1, 7185.597 cm -1 -1, 7444.352 cm -1 -1 and 6807.82 cm -1 -1, respectively.

6. The method of claim 5, wherein, In the range of absorption spectral lines in a set wave band, a plurality of absorption spectral lines are selected for water molecules in a to-be-detected combustion flow field, including: First step: Based on the absorption coefficient distribution of 7185.597 cm -1 and 6807.82 cm -1 , the temperature and concentration distribution of the combustion flow field is preliminarily reconstructed by selecting the spectral line combination of 7185.597 cm -1 and 6807.82 cm -1 ​ Second step: the temperature distribution calculation results of the first step are divided into two temperature intervals of T ≥ 1000 K and T < 1000 K, wherein the spectral line combination of 7444.36 cm -1 + 6807.82 cm -1 is selected for the grid above 1000 K, the temperature and concentration distribution of the combustion flow field is reconstructed based on the absorption coefficient distribution of 7444.36 cm -1 and the absorption coefficient distribution of 6807.82 cm -1 , the spectral line combination of 7161.4 cm -1 + 7185.597 cm -1 + 7444.36 cm -1 is selected for the grid below 1000 K, and the temperature and concentration distribution of the combustion flow field is reconstructed based on the absorption coefficient distribution of 77161.4 cm -1 , the absorption coefficient distribution of 7185.597 cm -1 and the absorption coefficient distribution of 6807.82 cm -1 . Third step: The results of the second step are divided into three temperature intervals, T < 600 K, 600 K ≤ T < 1000 K, T > 1000 K, where the grid of T < 600 K selects the spectral line combination of 7161.4 cm -1 + 7185.597 cm -1 , reconstructs the temperature and concentration distribution of the combustion flow field based on the absorption coefficient distribution of 7161.4 cm -1 and the absorption coefficient distribution of 7185.597 cm -1 , and the grids of 600 K ≤ T < 1000 K and T ≥ 1000 K retain the reconstruction results of the second step.

7. The method of claim 1, wherein, From the absorption spectral lines of water molecules in the near-infrared wave band of the to-be-detected combustion flow field, four absorption spectral lines are selected.

8. The method of claim 1, wherein, Based on the preliminarily reconstructed temperature distribution, a temperature interval is divided, and the optimal spectral line combination is selected for a grid corresponding to each temperature interval based on the temperature measurement uncertainty; and based on the absorption coefficient distribution of each absorption spectral line of the selected optimal spectral line combination, the temperature and concentration distribution of the combustion flow field is reconstructed, including: The range of the set noise level is greater than or equal to 2% and less than or equal to 8%. The linear inversion algorithm is a Tikhonov Regularization algorithm.

Images