A flame temperature imaging measurement method fusing electrical imaging and single-path absorption spectroscopy

Abstract

The application provides a flame temperature imaging measurement method fusing electrical imaging and single-path absorption spectrum, mainly including three steps of parameter initialization, histogram mapping relationship construction and temperature distribution reconstruction. First, in the parameter initialization, a temperature histogram is calculated by using single-path laser absorption spectrum, and a dielectric constant histogram is established by using electrical imaging to reconstruct the dielectric constant on the laser path. Then, in the histogram mapping relationship construction, a mapping model of the dielectric constant histogram to the temperature histogram is constructed by using a neural network, the dielectric constant histogram corresponding to multiple straight lines passing through the imaging area is obtained, and the temperature histogram corresponding to the dielectric constant histogram is obtained by substituting the dielectric constant histogram into the mapping model. Finally, in the temperature distribution reconstruction, the temperature distribution in the whole area is solved according to the temperature histograms on all straight lines. The application fuses single-path laser absorption spectrum and electrical imaging, obtains the temperature distribution of the whole area, and has important application value for temperature measurement in extreme environments.

Description

TECHNICAL FIELD

[0001] The present application relates to a flame temperature imaging measurement method fusing electrical imaging and single-path absorption spectrum, and belongs to the field of electrical imaging. Two-dimensional dielectric constant reconstruction information of electrical imaging technology and single-angle laser absorption spectrum information are fully fused to construct a mapping model on the same path, so as to obtain temperature information corresponding to multiple straight lines of laser absorption spectrum passing through the imaging area of the entire region, and realize two-dimensional temperature field reconstruction. BACKGROUND

[0002] Combustion is the main form of energy conversion, and the safety, efficiency and cleanliness of the combustion process are crucial to ensure the normal operation of major fields such as power, energy, transportation, aerospace, etc. Laser absorption spectroscopy (LAS) has the advantages of no need for pretreatment, fast response, accurate data, simultaneous detection of multiple parameters, strong environmental adaptability, etc., but for combustion temperature field measurement, a certain number of light path layouts are required, and in the case of limited projection layout, combustion temperature field measurement cannot be realized, and only temperature measurement in the corresponding light path can be realized. Electrical tomography (ET) technology, as a visualization monitoring means of electrical parameter distribution of complex flow system, is widely used in flow pattern identification, petroleum pipeline working condition detection, flame electrical parameter distribution visualization monitoring and many other aspects, and has the characteristics of no radiation, non-invasive, fast response, low cost, no movable parts, reliable work, no need for optical window, etc.

[0003] Electrical tomography technology is a non-contact measurement method based on electrostatic field model. When the electrical parameters of the measured field change, the admittance between the two electrodes of the sensor will change, and by measuring the boundary admittance value and combining image reconstruction algorithm, the electrical parameter distribution of the medium in the field can be obtained. The electrical parameters of the medium include dielectric constant and conductivity. When focusing on the dielectric constant characteristics of the medium, electrical tomography is called electrical capacitance tomography (ECT); when focusing on the conductivity characteristics of the medium, electrical tomography is called electrical resistance tomography (ERT).

[0004] In the combustion field, a large number of neutral particles, free radicals and electrons are generated under the action of chemical ionization and thermal ionization, so that the electrical parameters of the flame change significantly compared with air, and therefore, the electrical imaging technology can be used to visually monitor the flame distribution. In addition, the electrical parameters of the flame are related to the temperature, particle species and particle concentration of the combustion field. Therefore, electrical imaging can monitor the dielectric constant and conductivity distribution in the flame to realize the imaging of electrical parameters; according to the theoretical model or simplified model of the electrical parameters and temperature, the direct reconstruction of ET for temperature can be realized; however, too many unknown parameters in the temperature measurement theoretical model are not mature in actual application, and it is necessary to be combined with other temperature measurement methods to construct the model of dielectric constant and temperature.

[0005] Laser absorption spectroscopy is a line-of-sight measurement technique that uses the absorption of specific wavelength laser by gas to obtain the absorption rate of gas based on Beer-Lambert law. The absorption rate of light intensity in the light path is strongly coupled with the temperature, component concentration and other parameters of the absorbing molecules. By using double-spectrum colorimetry and multi-spectrum measurement, each parameter can be decoupled to realize the quantitative measurement of multiple parameters in the combustion field. The basic principle is that when a laser passes through the gas flow field to be measured, the light intensity of the laser will be attenuated due to the absorption of molecules. The attenuation value is related to the temperature, component concentration, pressure and path length of the absorbing molecules. By scanning the wavelength of the incident laser within a certain range, the absorption spectrum of the measured molecules can be obtained, and then the parameters in the laser path can be solved. According to the absorption spectrum data of a single path, the average value of the combustion field parameters along the laser path can be solved, but the spatial distribution of the parameters cannot be reflected. Therefore, by combining laser absorption spectroscopy with tomographic imaging technology, the absorption spectrum data of multiple angles and multiple light paths can be obtained, and the distribution of the combustion field parameters can be solved by tomographic imaging algorithm. In 2014, Lan Lijuan et al. published an article titled "Theoretical and experimental study on the measurement of gas temperature in vacuum environment by tunable diode laser absorption spectroscopy" in the Physical Review, Vol. 63, No. 8, pp. 101-109. The authors used tunable diode laser absorption spectroscopy (TDLAS) to measure the vibration-rotation temperature of gas molecules in a vacuum environment. When the absorption rate of the gas to the laser is greater than 2%, TDLAS can be used to measure the vibration-rotation temperature of gas molecules in a vacuum environment, with high response speed and high consistency with the gas temperature measured by a standard platinum resistance and the temperature of a constant temperature bath. The error between the two is not more than ±0.2℃ when the temperature of the constant temperature bath is stable. In 2015, Liu C et al. published an article titled "Development of a fan-beam TDLAS-based tomographic sensor for rapid imaging of temperature and gas concentration" in Optics Express, Vol. 23, No. 17, pp. 22494-22511. The authors designed a five-angle fixed fan-beam TDLAS sensor, which reduced the distance between two adjacent detectors and improved the spatial resolution to 0.78 centimeters. The authors collected two water molecule absorption spectra at a frame rate of 83.3 frames / s on 60 measurement light paths, and monitored the temperature distribution of McKenna planar flame and dynamic flame with different shapes. The experimental results show that the reconstructed temperature and H2O concentration distribution are consistent with the expected distribution.In the current single-path laser absorption spectrum system (such as patent: gas temperature probability density distribution fitting reconstruction method based on single-path multi-spectrum, application number: CN201710469033.4), using multiple absorption spectra can reconstruct the gas temperature probability density distribution. Compared with the traditional LAS which can only obtain the temperature mean value information, this method can obtain more information, but cannot obtain the position information of the temperature value on the path, and more projection angle temperature histograms are needed to reconstruct the temperature corresponding to each grid in the entire region.

[0006] However, when laser absorption spectroscopy technology is used for two-dimensional temperature field imaging, a complete optical window needs to be opened. Single-path LAS can measure the average temperature on the path by opening two holes only in the opposite position of the combustion chamber, but cannot obtain the temperature information of each grid on the path, and cannot reconstruct the two-dimensional temperature distribution of the entire field.

[0007] Electrical imaging technology is based on the principle that the change of dielectric constant in the measured region will cause the change of impedance between the electrodes on the sensor. The two-dimensional distribution of dielectric constant and conductivity in the measured region is reconstructed by using reconstruction algorithm. Because it belongs to non-invasive measurement, it has the advantages of fast response, low imaging cost, no need to open window, etc. It has been applied in the measurement of electrical parameters of combustion field. In 2000, R.C. Waterfall published the paper "Imaging Combustion Using Electrical Capacitance Tomography" in IEE Seminar Advanced Sensors and Instrumentation Systems for Combustion Processes. It was used to observe the position of the jet flame in the combustion tank. It could be installed in the actual combustor and had a time resolution of several milliseconds, which was enough to monitor and control the jet engine. In 2008, Lis S et al. published the paper "Preliminary study on ECT imaging of flames in porous media" in Measurement Science and Technology, Vol. 19, No. 9. They analyzed the influence of flame ionization effect on dielectric constant and used ECT technology to study the flame distribution in porous media. The flame was imaged, which could track the movement of the flame and also detect multiple flames. The color distribution of the flame image corresponded well to the combustion intensity, which could reflect the shape, position and gas flow change of the flame. In 2010, Gut Z. and Wolanski P. published the paper "Flame Imaging Using 3D Electrical Capacitance Tomography" in Combustion Science and Technology, Vol. 182, No. 11-12, pp. 1580-1585. They made a three-plane ECT sensor with 18 electrodes (6 electrodes in each plane and 3 rows of electrodes), which could obtain more realistic flame reconstruction images.In 2013, Jia Y et al. published "Theoretical Analysis of the Dielectric Characteristics of Plasma Flame and Imaging plasma flame Using Electrical Capacitance Tomography" in the 2013 IEEE International Conference on Imaging Systems and Techniques. Based on the analysis of plasma flame ionization phenomena, polarization modes, bound electrons and free electric displacement polarization mechanism, a theoretical model of the dielectric coefficient of plasma flame was proposed. Combined with the mathematical model of ECT sensor and multi-point calibration measurement data, the distribution of plasma flame dielectric constant in the measurement area was obtained, providing a new method and valuable data for ECT research on plasma combustion. In 2021, Hu D et al. published "Revised Calderon Method of Annular ECT for Imaging Flashback Flame of a Bluff-Body Burner" in IEEE Transactions on Instrumentation and Measurement, Vol. 70. Eight-electrode circular sensors were used to visualize and monitor the flashback process of different combustion conditions in a bluff-body burner. The reconstructed gray value was used to quantitatively characterize the flashback intensity and flashback time, providing a new means for the study of combustion stability mechanism. In 2024, Jin S et al. published "Estimation of Temperature for Premixed Flame by Relative Permittivity and Conductivity" in IEEE Transactions on Instrumentation and Measurement, Vol. 73. A method for estimating the temperature of premixed flame based on relative permittivity and conductivity was proposed. A mechanism model was established between the relative permittivity, conductivity and temperature of premixed flame, and the temperature estimation of premixed flame was realized.

[0008] The mapping model of permittivity, conductivity and temperature needs to be established for the temperature measurement of combustion field by electrical imaging technology. However, the model requires the parameters such as the type, concentration and collision frequency of particles in the combustion field, which are difficult to obtain. Simplifying the model by simulation is not necessarily reasonable. Therefore, the mapping model of permittivity, conductivity and temperature needs to be established by integrating ECT with other temperature measurement methods. In 2016, Chen Q et al. published an article titled 'Direct Measurements of Permittivity of Plasma-Assisted Combustion Using Electrical Capacitance Tomography' in IEEE Transactions on Plasma Science, Vol. 44, No. 12, pp. 3009-3016. In this article, the temperature of single-point flame was measured by thermocouple, and the electron density and electron temperature were measured by Langmuir probe. The theoretical value of the dielectric coefficient of the flame was obtained and compared with the ECT measurement results. This method uses invasive measurement to obtain the temperature and electrical parameters of the combustion field. Although it disturbs the flow field and has low spatial resolution of measurement results, it still provides a new idea for the temperature parameter monitoring of combustion field by electrical imaging technology. In 2018, Liu J et al. published an article titled 'Internal structure visualization of flow and flame by process tomography and PLIF data fusion' in Journal of Thermal Science, Vol. 27, No. 1, pp. 64-73. In this article, electrical imaging and laser-induced fluorescence technology were used to monitor the combustion process of the flame simultaneously. The high permittivity part in the ECT reconstructed image can represent the interface between the inner cone and the outer shell of the flame. The reaction of this interface is the strongest and the permittivity is the highest. The OH-PLIF captured image shows the distribution of OH radicals in the flame on the same sheet. The two images show good consistency in shape, showing a strong correlation between the permittivity and OH radicals in the flame, which proves the feasibility of using ECT to measure the flame.In 2020, Hu D et al. published an article titled "Estimation of Combustion Temperature Field from the Electrical Admittivity Distribution Obtained by Electrical Tomography" in IEEE Transactions on Instrumentation and Measurement, Vol. 69, No. 9, pp. 6271-6280. To address the complexity of the permittivity and conductivity to temperature mapping model, ion probes and thermocouples were used to measure the electrical parameters and temperature at different positions of an alcohol lamp flame. Polynomial, exponential, and extreme gradient boosting (XGBoost) regression modeling methods were proposed to fit the mapping model and used in combustion field temperature estimation. The conductivity and permittivity distributions at different heights of a Bunsen burner flame were reconstructed. However, the thermocouples and ion probes would disrupt the flow field and cause severe interference with the permittivity and conductivity distribution in the sensitive field. Therefore, the above method can only be used for offline calibration of electrical tomography and cannot monitor the same flame under the same combustion state. In 2024, Tian Y et al. published an article titled "Dynamic Cross-Sectional Temperature Imaging From LAS Labeled Electrical Tomography" in IEEE Transactions on Instrumentation and Measurement, Vol. 73. The permittivity and conductivity distributions of a Bunsen burner flame under multiple fuel-rich combustion conditions were reconstructed. LAS was used to reconstruct the temperature distribution in the same cross-section. By changing the equivalence ratio of fuel and air, the combustion state of the flame was changed. A random forest regression model was used to construct the mapping of electrical parameters to temperature under different combustion conditions as a temperature measurement model, achieving temperature distribution monitoring of the combustion field under different combustion conditions using electrical tomography technology. However, this method requires a two-dimensional LAS temperature measurement system, which requires a complete optical window and many optical paths.

[0009] Combining the advantages of LAS and ECT technology in the combustion field measurement, a single path LAS signal is taken as a temperature reference to establish a mapping relationship between the dielectric constant and the temperature, and then the temperature of the combustion field is reconstructed; especially in some conditions that are not conducive to the use of optical measurement methods, such as the blocking of solid particles to the light signal in a harsh environment, the optical window is not easy to open, and the two-dimensional temperature distribution of the combustion field can be obtained, which has important engineering application prospects.

[0010] Based on the above background, the present application proposes a flame temperature imaging measurement method combining electrical imaging and single-path absorption spectrum, which uses single-path LAS to obtain the light intensity change on the path, uses SART algorithm to obtain temperature histogram, uses electrical imaging to obtain one-dimensional dielectric constant distribution on the same path of LAS, and obtains dielectric constant histogram according to the selected maximum and minimum values of dielectric constant; a neural network is used to construct a mapping from the dielectric constant histogram to the temperature histogram, a plurality of straight lines are set, the straight line equation of each straight line is determined, the dielectric constant distribution of each straight line is obtained according to the straight line equation, and the corresponding dielectric constant histogram is obtained, which is substituted into the mapping relationship to obtain the temperature histogram corresponding to all straight lines; finally, the temperature distribution in the whole region is obtained by using SART algorithm. The advantage of this method is to fully utilize the dielectric constant information, fuse the histogram information of single-path LAS, and expand to several straight lines, so that the temperature field information of the measured flame can be calculated, and only two optical holes are needed, the structure of the system is simple, and the cost is low. SUMMARY

[0011] (1) Technical problems to be solved

[0012] The purpose of the present application is to propose a flame temperature imaging measurement method combining electrical imaging and single-path absorption spectrum, which fully utilizes the two-dimensional dielectric constant reconstruction information of electrical imaging technology and the single-angle laser absorption spectrum information to fuse and construct the mapping relationship on the same straight line path, so as to obtain the temperature information corresponding to the laser absorption spectrum straight line of the whole region, and realize two-dimensional temperature distribution imaging.

[0013] (2) Technical solutions

[0014] The present application is a flame temperature imaging measurement method combining electrical imaging and single-path absorption spectrum, mainly including the following steps:

[0015] Step 1: Obtain the temperature histogram on the path by a single-path laser absorption spectrum system; the system includes a laser generator and a controller, a measured flame gas and a photodetector, a specific waveband laser passes through the center of the flame, the target gas will absorb the laser, so that the light intensity is attenuated, the photodetector receives the attenuation of the light intensity signal, and the laser beam v(t) after high-frequency modulation can be expressed as:

[0016]

[0017] wherein, denotes the center wave number before modulation, a d modulation depth, f is the modulation frequency of the high-frequency sinusoidal signal. The laser intensity I0(t) after high-frequency modulation can be expressed as:

[0018]

[0019] wherein, is the average light intensity of low-frequency scanning, b is the laser intensity modulation coefficient, is the harmonic component of each modulation frequency of the laser. According to the Beer-Lambert law, the laser with the center wave number at v0[cm -1 ] passes through the measured gas, and the absorption rate a(v) of the light intensity at the wave number v[cm -1 ] is:

[0020]

[0021] Substituting formulas (1) and (2) into formula (3), the light intensity of the laser beam after passing through the target gas is:

[0022]

[0023] wherein, τ[v(t)] is the transmission coefficient of the laser.

[0024] For the case of small absorption of the target gas, the transmission coefficient can be approximated as:

[0025]

[0026] Expanding formula (5) into Fourier series form:

[0027]

[0028] Expand the Fourier coefficients:

[0029]

[0030] For a single-path measurement system, M temperature values T m , m = 1, 2, …, M, corresponding to the occupied absorption distance L m , m = 1, 2, …, M; K absorption spectra per unit path length β k (v), k = 1, 2, …, K; corresponding to the Fourier coefficients satisfy:

[0031]

[0032] The first and second harmonic components are extracted from the transmitted light intensity signal by using a lock-in amplifier and a low-pass filter, and the normalized second harmonic signal S 2f / 1f is obtained.

[0033]

[0034] where R 2f and R 1f are the amplitudes of the second and first harmonic signals, respectively, and X 1f and Y 1f are the X and Y components of the first harmonic, and X 2f and Y 2f are the X and Y components of the second harmonic.

[0035] For a single-path LAS measurement system, M discrete gas temperature values are selected, and the normalized second harmonic signal R 2f / 1f,i at unit path length for the i-th pair of gas temperature values is simulated. According to equation (9), the single-path, normalized second harmonic signal and the normalized second harmonic signal per unit absorption of the M discrete gas temperature values satisfy:

[0036]

[0037] According to equation (10), the path length corresponding to the selected M discrete gas temperature values can be obtained, and a temperature histogram is established.

[0038] Step 2, calculate the dielectric constant value at the same path of the single-path LAS, and establish a dielectric constant histogram; solve the dielectric constant distribution through the obtained capacitance value, and the sensitive field of electrical imaging is:

[0039]

[0040] where z = x + iy is a complex number representing the coordinates (x, y), and γ(z) = σ(z) + iωε(z) and are the complex conductivity and the corresponding electric potential at point z, respectively.

[0041] Let any analytic function in the region be v(z), then:

[0042]

[0043] Since:

[0044]

[0045] By combining equation (13) and equation (12), we have:

[0046]

[0047] By divergence theorem, the integral of the changing dielectric constant in space can be linked to the boundary voltage and current measurement, i.e. according to divergence theorem

[0048]

[0049] The voltage to current density mapping on the measured field domain boundary, i.e. Dirichlet-to-Neumann mapping is:

[0050]

[0051] where, represents the voltage to current density mapping when there is a dielectric constant distribution γ0(z) in the sensitive field Ω, represents the voltage to current density mapping when there is a dielectric constant distribution (γ0+Δγ)(z) in the sensitive field Ω, then we have:

[0052]

[0053] Subtracting equation (18) from equation (17), and assuming that the potential distribution is approximately invariant when the dielectric constant variation Δγ(z) is small according to Calderon method, i.e. then we have:

[0054]

[0055] Since satisfies Laplace equation, i.e. Let be a particular solution, and let k=k1+ik2, z=x+iy. So we have:

[0056]

[0057] where t(k) is the scattering transform, which satisfies the form of Fourier transform.

[0058] Calculate the scattering transform:

[0059]

[0060] where, P is the number of measurement electrodes, θ i represents the center of the plate of the i-th electrode, and A is the area of the plate.

[0061] Performing two-dimensional inverse Fourier transform on t(k) can obtain the change value of the dielectric constant Δε(z):

[0062] ​

[0063] The dielectric constant distribution along the path is calculated using the Calderon method, and formula (22) is expressed as:

[0064]

[0065] where R is the radius of the numerical integration region.

[0066] The dielectric constant distribution along the path is calculated using formula (23), and the maximum dielectric constant ε max and the minimum dielectric constant ε min are found. max and ε min are divided into N values, i.e. ε n , n = 1, 2, …, N; the grid corresponding to the single-path LAS is found along the path, and the dielectric constant value of each grid is put into ε n , n = 1, 2, …, N, to establish the dielectric constant histogram along the path.

[0067] Step three, the mapping relationship between the dielectric constant histogram and the temperature histogram along the same path is established using a neural network, and the temperature histogram matrix of Q straight lines passing through the imaging region is obtained; the fully connected neural network is a multi-layer perceptron structure, which includes three layers of input layer, hidden layer and output layer, and the number of hidden layers in this method is set to one. The input of the neural network is the N values of the dielectric constant histogram, and the output is the M values of the temperature histogram.

[0068] The number of neurons in the hidden layer is selected according to the empirical formula:

[0069]

[0070] where n Neurons is the number of neurons, N is the number of indices of the input layer, i.e. the number of values of the dielectric constant histogram, M is the number of indices of the output layer, i.e. the number of values of the temperature histogram, and c is an empirical constant, usually a number in the range of [1, 10].

[0071] The output of the jth node in the hidden layer is:

[0072]

[0073] where h j is the output value, ω ij is the connection weight between the ith node of the input layer and the jth node of the hidden layer, x i is the input variable of the ith node of the input layer, β j is the threshold value of the jth node of the hidden layer, and f(·) is the activation function.

[0074] The output of the mth node of the output layer is:

[0075]

[0076] where y out is the output value, η jm is the connection weight value between the jth node of the hidden layer and the mth node of the output layer, σ m is the threshold value of the mth node of the output layer.

[0077] The imaging area is discretized into D grids, and a rectangular coordinate system is established with the center of the imaging area as the origin. The equations of Q straight lines passing through the imaging area are:

[0078] xcosθ Q×1 +ysinθ Q×1 =ρ Q×1 (27)

[0079] where θ Q×1 is the set of angles between the normal of each of the selected Q straight lines and the horizontal axis, and ρ Q×1 is the set of distances from the origin to each of the selected Q straight lines.

[0080] The grid passed by the straight line equation is determined using formula (27), and the dielectric constant value of each straight line passing through the grid is calculated using formula (23) and placed in the dielectric constant value ε n , n = 1, 2, …, N, so that each straight line establishes a dielectric constant histogram on the corresponding path, i.e. Q dielectric constant histograms. The Q dielectric constant histograms are substituted into the full connection neural network mapping model of formulas (25) and (26) to obtain Q temperature histograms corresponding to each straight line.

[0081] Step four, two-dimensional temperature distribution imaging; a sensitivity matrix W Q×D is established according to the layout of the selected Q straight lines, where the element w i,j in the ith row and jth column is the length of the ith path passing through the jth grid. A 0-1 binary matrix Y D×M is used to represent the temperature distribution of the target gas field, and each row of the matrix Y D×M represents a grid, and there is an element 1 in the row and the rest of the elements are 0. According to the column corresponding to the element 1, the temperature value at that position is obtained, i.e.

[0082]

[0083] where T represents the temperature values of the D grids, and the superscript R represents the reconstructed distribution.

[0084] The histogram matrix L Q×MThe path length of M kinds of gas states respectively occupying Q straight lines is represented, and the Q straight lines satisfying formula (10) are written in matrix form as:

[0085] S Q×K =L Q×M ·R M×K (29)

[0086] wherein the M kinds of gas parameters in formula (10) are normalized second harmonic R 2f / 1f,i The harmonic base matrix R M×K is arranged row by row, K is the data point number of the normalized second harmonic signal, and the matrix of the normalized second harmonic signal arranged row by row is S Q×K .

[0087] The definition formula of the sensitivity matrix is:

[0088] L Q×M =W Q×D ·Y D×M (30)

[0089] Since the histogram of multiple straight lines passing through the imaging area satisfies that the sum of lengths is equal to the total length of the straight line, the constraint is satisfied:

[0090] L Q×M ·e M×1 =W Q×D ·e D×1 (31)

[0091] wherein the column vectors e M×1 and e D×1 are unit vectors.

[0092] By combining formula (29), (30) and (31), the reconstructed temperature field model is obtained:

[0093] W Q×D ·Y D×M ·(R M×K e M×1 )=(S Q×K W Q×D ·e D×1 ) (32)

[0094] The matrix Y D×M is solved by iteration using formula (32), and the two-dimensional temperature distribution image is reconstructed according to formula (28).

[0095] (Three) beneficial effects

[0096] The beneficial effect of this invention lies in proposing a flame temperature imaging measurement method that integrates electrical imaging and single-path absorption spectroscopy. Electrical imaging monitors flames without requiring an optical window, but it can only obtain the distribution of the dielectric constant, not temperature information; while single-path laser absorption spectroscopy cannot reconstruct a two-dimensional temperature field, only obtaining the average temperature along the path. This invention integrates electrical imaging and single-path laser absorption spectroscopy, utilizing an electrical imaging system and a single laser optical path to obtain a dielectric constant histogram and a single-path laser absorption spectrum temperature histogram. A mapping model is constructed on the histograms along the same path, and temperature histograms are obtained along multiple straight lines passing through the imaging region, ultimately achieving two-dimensional temperature field reconstruction. Attached Figure Description

[0097] Appendix Figure 1 : Detailed implementation diagram of this method

[0098] Appendix Figure 2 The reconstruction system structure diagram consists of the following parts: laser control and generation module (101), capacitor array sensor (102), imaging area (103), photodetector (104), laser absorption spectrum data acquisition module (105), electrical tomography acquisition module (106), and host computer (107).

[0099] Appendix Figure 3 The area between the two dashed lines represents the flame distribution in the experiment, which is used for temperature field reconstruction. The region between the two dashed lines represents the location for electrical imaging measurements, and the area between the two dashed lines represents the location for laser absorption spectroscopy measurements.

[0100] Appendix Figure 4 Temperature distribution reconstructed using a 45×45 grid. Detailed Implementation

[0101] Reference Appendix Figure 1 The following is a detailed implementation diagram of this method; Appendix 2 is a schematic diagram of the imaging system. Figure 3 For the flame distribution in the experiment, see attached. Figure 4 To reconstruct the temperature field distribution, the specific steps are as follows, using an example:

[0102] Step 1: Obtain the temperature histogram along the path using a single-path LAS; (See attached image) Figure 2 The single-channel LAS system shown includes a laser generator and controller, the measured flame area and photodetector, as well as a data acquisition module and a host computer, using a 7185.6cm... -1 and 7444.4cm -1Two laser lines pass through the center of the flame, the wave number model is based on 1 kHz sawtooth wave superimposed 200 kHz sinusoidal modulation, the modulation depth is 0.1, the laser of a specific wave band passes through the center of the flame and is absorbed by the target gas, so that the light intensity decays, the detector receives the decay of the light intensity signal, and the data acquisition module transmits it to the upper computer.

[0103] Six temperature values T1, T2, T3, T4, T5 and T6 are discretely selected, and the corresponding occupied absorption distances are L1, L2, L3, L4, L5 and L6. The absorption spectra β1(v) and β2(v) of the two kinds of uniform gas parameters per unit path length are calculated. For a single-path LAS measurement system, six discrete gas temperature values are selected, and the normalized second harmonic signal R 2f / 1f,i Simulation is carried out. The normalized second harmonic signal per unit absorption of the six discrete gas temperature values satisfies:

[0104]

[0105] According to formula (1), the path lengths corresponding to the selected six gas temperature values are obtained, and a temperature histogram is established.

[0106] Step two, obtain the dielectric constant value on the same path as the single-path LAS, and establish a dielectric constant histogram; as shown in the figure Figure 2 In the electrical imaging system shown in the figure, when reconstructing the dielectric constant distribution, the two-dimensional field to be measured is evenly divided into a 45x45 grid, the sensitive field is circular, and a total of 1517 grids are included. The number of grids passing through the path is 45. The dielectric constant distribution is calculated using Calderon:

[0107]

[0108] The dielectric constant distribution on the path is obtained by using formula (2), and the maximum dielectric constant ε max and the minimum dielectric constant ε min are found. Divide ε max and ε min into six values, i.e. ε1, ε2, ε3, ε4, ε5 and ε6. Find the 45 grids on the path corresponding to the single-path LAS, put the dielectric constant value of each grid into ε1, ε2, ε3, ε4, ε5 and ε6, and establish a dielectric constant histogram on the path.

[0109] Step three, using neural network to establish the mapping relationship between permittivity histogram and temperature histogram on the same path, and obtain the temperature histogram matrix of 189 straight lines passing through the imaging area; the full connection neural network is a multi-layer perceptron structure, which includes three layers of input layer, hidden layer and output layer, the number of hidden layers in the method is set to 1, and the number of neurons is 9. The full connection neural network is used to realize the forward transmission of permittivity histogram to temperature histogram.

[0110] The imaging area is discretized into 1517 grids, and a rectangular coordinate system is established with the center of the imaging area as the origin. The equation of the 189 straight lines passing through the imaging area is:

[0111] xcosθ 189×1 +ysinθ 189×1 =ρ 189×1 (3)

[0112] Where, θ Q×1 is the set of angles between the normal of each of the 189 selected straight lines and the horizontal axis, and ρ 189×1 is the set of distances from the origin to each of the 189 selected straight lines.

[0113] The grid passed by the straight line equation is found using formula (3), the permittivity value of each straight line passing through the grid is calculated using formula (2), and is put into permittivity values ε1, ε2, ε3, ε4, ε5, ε6, thereby establishing the permittivity histogram on the corresponding path, i.e. 189 permittivity histograms. The 189 permittivity histograms are substituted into the full connection neural network mapping model to obtain 189 temperature histograms corresponding to each straight line.

[0114] Step four, two-dimensional temperature distribution imaging; according to the layout of the 189 selected straight lines, a sensitivity matrix W 189×1517 is established, where the element w i,j in the ith row and jth column is the length of the ith path passing through the jth grid. A 0-1 binary matrix Y D×M is used to represent the temperature distribution of the target gas field, and each row of the matrix Y 1517×6 represents a grid, and there is an element 1 in the row and the rest of the elements are 0. According to the column corresponding to the element 1, the temperature value at that position is obtained:

[0115]

[0116] Where, represents the temperature values of the 1517 grids, and the superscript R represents the reconstructed distribution.

[0117] Histogram matrix L 189×6The 189 straight lines represent the path length of 6 temperature states, respectively, and the 189 straight lines satisfy formula (1) in matrix form:

[0118] S 189×1180 =L 189×6 ·R 6×1180 (5)

[0119] Wherein, two spectral lines are selected near the peak of 520 and 660 sampling points, and two spectral lines constitute the measurement data matrix S of 189 light paths 189×1180 The 6 gas temperature parameters in formula (1) are normalized second harmonic R 2f / 1f,i Arranged row by row to form the harmonic base matrix R 6×1180 .

[0120] From the definition of the sensitivity matrix:

[0121] L 189×6 =W 189×1517 ·Y 1517×6 (6)

[0122] Since the histogram of the 189 straight lines passing through the imaging area satisfies the sum of the lengths equal to the total length of the straight line, it satisfies the constraint:

[0123] L 189×6 ·e 6×1 =W 189×1517 ·e 1517×1 (7)

[0124] Wherein, column vectors e 6×1 And e 1517×1 Are unit vectors.

[0125] Together with formula (5), (6) and (7), the reconstructed temperature field model can be obtained:

[0126] W 189×1517 ·Y 1517×6 ·(R 6×1180 e 6×1 )=(S 189×1180 W 189×1517 ·e 1517×1 ) (8)

[0127] Using formula (8) to solve the matrix Y 1517×6 , the two-dimensional temperature distribution image is reconstructed according to formula (4).

Claims

1. A flame temperature imaging measurement method that fuses electrical imaging and single-pass absorption spectroscopy, characterized by, The imaging system comprises a circular cross-section electrical imaging sensor, a single-channel tunable laser generation module, a photodetector, and a data acquisition and processing module, wherein the circular cross-section electrical imaging sensor has P electrodes with an area of A, one electrode applies an alternating voltage, and the other electrodes are grounded or kept at the same potential, the capacitance value change in the sensor is obtained, and the dielectric constant distribution is solved; the laser passes through the center region of the circular cross-section electrical imaging sensor, and the photodetector obtains the laser absorption spectrum; the flow field and normalized second harmonic signal simulation are performed to estimate the temperature range of the flame, M temperature values are selected, and the temperature histogram is calculated; the imaging area is discretized into D grids, all dielectric constant values of the grids on the same path as the laser absorption spectrum are obtained, N dielectric constant values are selected according to the distribution interval of the dielectric constant values, and the dielectric constant histogram is obtained; a mapping model of the dielectric constant histogram to the temperature histogram is established by using a neural network, Q straight lines passing through the imaging area are selected, the dielectric constant distribution corresponding to each straight line is reconstructed by using electrical imaging, the dielectric constant histogram of each straight line passing through the grid is obtained, the temperature histogram of the grid on each straight line is obtained through the mapping model, and a sensitivity matrix is established according to the Q straight lines passing through the imaging area, so that the two-dimensional temperature distribution image in the whole region is calculated by using the temperature histogram and the sensitivity matrix.

2. The method of claim 1, wherein the method is a flame temperature imaging measurement method that combines electrical imaging and single-pass absorption spectroscopy. The mapping of the dielectric constant histogram to the temperature histogram is obtained by using one optical path, the temperature histogram of Q straight lines passing through the imaging area is obtained by using electrical imaging technology, the two-dimensional temperature distribution image in the whole region is calculated by combining the grid division corresponding to all straight lines and the sensitivity matrix constructed by the straight lines, and the specific steps include the following steps: Step one: obtaining the temperature histogram on the path by a single-channel laser absorption spectrum system; the laser of a specific waveband passes through the center of the flame and is absorbed by the target gas, resulting in a decrease in light intensity, and the light intensity after the high-frequency modulated laser beam passes through the target gas is: where I0(t) is the incident light intensity of the laser, τ[v(t)] is the transmission coefficient of the laser, denotes the center wave number before modulation, a d modulation depth, f is the modulation frequency of the high-frequency sinusoidal signal, and a(v) is the absorption rate of the light intensity at the wave number v [cm -1 ] The flow field and normalized second harmonic signal simulation is performed to estimate the temperature range of the flame, and M temperature values {T1, T2, …, T m ,…,T M} are discretely selected, where 1≤m≤M, and the corresponding absorption lengths are {L1, L2, …, L m ,…,L M}; K absorption spectra β k (v) of unit path length are calculated, k=1, 2, …, K; the first harmonic and second harmonic components are extracted from the transmitted light intensity signal, and the normalized second harmonic signal is obtained; the normalized second harmonic signal of unit path length at the selected mth gas temperature value is simulated; and the single-path normalized second harmonic signal and the M discrete gas temperature values of unit absorption normalized second harmonic signal satisfy: S 1×K = L 1×M · R M×K (2) wherein R M×K is a harmonic base matrix composed of M kinds of gas parameter normalized second harmonic wave arranged row by row, K is the number of data points of the normalized second harmonic wave signal, S 1×K is a matrix of the normalized second harmonic wave signal arranged row by row; The path length L corresponding to the selected M discrete gas temperature values is obtained according to formula (2) 1×M and thereby a temperature histogram is established; Step two: obtaining the dielectric constant distribution on the same path as the single-channel laser absorption spectrum to establish a dielectric constant histogram; the dielectric constant distribution on the path is calculated by using the Calderon method to inverse the dielectric constant distribution on the path: where R is the radius of the numerical integration region, ΔC is the capacitance change value, (x, y) is the coordinate value on the path, r is the polar radius, θ is the polar angle, P is the number of electrodes, and A is the area of the electrode plate. θ k represents the center of the electrode plate of the kth electrode, θ k = 2πk / P; The dielectric constant distribution on the path is obtained from equation (3), and the maximum value ε max and the minimum value ε min of the dielectric constant in the path are found max . min The maximum value ε min and the minimum value ε n are divided into N values, i.e., {ε max 1, ε2,…, ε min N}, where 1≤n≤N; the imaging area is discretized into D grids, the grid on the path corresponding to the single-path laser absorption spectrum is found, and the dielectric constant value of each grid is put into {ε n 1, ε2,…, ε max N} to establish the dielectric constant histogram on the path. Step three: establishing the mapping relationship of the dielectric constant histogram to the temperature histogram on the same path by using a neural network to obtain the temperature histogram matrix of Q straight lines passing through the imaging area; the fully connected neural network is a multi-layer perceptron structure, which comprises three layers of input layer, hidden layer and output layer; the input of the neural network is N values of the dielectric constant histogram, and the output is M values of the temperature histogram; A rectangular coordinate system is established with the center of the imaging area as the origin, and the equations of the Q straight lines passing through the imaging area satisfy: xcosθ Q×1 +ysinθ Q×1 = p Q×1 (4) where θ Q×1 is the set of angles between the normal of each of the selected Q straight lines and the horizontal axis, and p Q×1 is the set of distances from the origin to each of the selected Q straight lines. The grid through each straight line is found by using formula (4), the dielectric constant value of each straight line through the grid is calculated by using formula (3), and is placed into the dielectric constant value {ε min ,ε2,…,ε n ,…,ε max} to establish a dielectric constant histogram on the path corresponding to each straight line, that is, Q dielectric constant histograms; the Q dielectric constant histograms are substituted into the full connection neural network mapping model to obtain Q corresponding straight line temperature histogram matrices; Step four, two-dimensional temperature distribution imaging; sensitivity matrix W is established according to the layout of selected Q straight lines Q×D wherein the element w i,j is the length of the ith path passing through the jth grid; matrix Y D×M is the temperature distribution of the target gas, Y D×M Each row of Y represents a grid, there is an element 1 in the row and the rest of the elements are 0, and the temperature value at the column corresponding to the element 1 is obtained. wherein, represents the temperature values of D grids, the superscript R represents the reconstructed distribution; The histograms on the multiple straight lines passing through the imaging area satisfy that the sum of the lengths is equal to the total length of the straight line, and the reconstructed temperature field model is: W Q×D ·Y D×M ·(R M×K e M×1 )=(L Q×M ·R M×K W Q×D ·e D×1 ) (6) where e M×1 and e D×1 are unit vectors, and W Q×D is a sensitivity matrix. The matrix Y of equation (6) D×M is solved iteratively and the two-dimensional temperature distribution image is reconstructed according to equation (5).

Images