A method for extracting spectral characteristics of infrared radiation of engine exhaust plume

Abstract

The present application relates to the technical field of infrared radiation characteristics, and particularly relates to an engine exhaust plume infrared radiation spectrum characteristic extraction method, which comprises the following steps: acquiring an infrared radiation thermal image of an engine exhaust plume; eliminating steady-state background interference by carrying out time mean elimination and normalization processing on the infrared thermal image sequence; converting flow field data in a Cartesian coordinate system into a cylindrical coordinate system through interpolation, so as to facilitate azimuth angle mode analysis; applying a Welch method to block the preprocessed infrared thermal image data, and applying a Hanning window to suppress spectrum leakage; carrying out discrete Fourier transform on the blocks, regrouping the data according to frequency, and constructing a cross-spectral density matrix under each frequency; carrying out eigenvalue decomposition on the cross-spectral density matrix, obtaining SPOD modes and eigenvalues under each frequency, and arranging the eigenvalues in descending order according to the mode energy, so as to extract a dominant mode as an infrared radiation spectrum core feature.

Description

Technical Field

[0001] This invention relates to the field of infrared radiation characteristics technology, specifically a method for extracting the infrared radiation spectrum characteristics of engine exhaust plumes. Background Technology

[0002] The rocket engine exhaust plume is a high-speed free jet formed by the high-temperature, multi-component combustion gases of the propellant being ejected into the atmosphere through a nozzle. Its polyatomic components, such as H2O, CO2, and CO, undergo vibrational-rotational transitions at high temperatures, generating infrared radiation in specific wavelengths, making it a core focus of infrared early warning and detection systems. The flow of the exhaust plume is essentially a complex, unsteady flow with strong coupling of multiple physical fields, accompanied by phenomena such as chemical reactions, turbulent pulsations, and acoustic resonances. This results in significant spatiotemporal unsteady characteristics in infrared radiation, specifically manifested as: drastic fluctuations in flow parameters over time, multi-scale evolution of vortex structures in space, and multi-frequency coupling in the spectrum.

[0003] Existing studies on the infrared radiation characteristics of engine exhaust plumes mostly rely on numerical simulations or experimental measurements, but these methods have significant limitations: numerical simulations often focus on macroscopic statistical parameters, making it difficult to decouple the dynamic relationship between multi-scale vortex structures and infrared radiation, resulting in low accuracy and poor stability in extracting infrared radiation spectrum features.

[0004] Therefore, there is an urgent need for a method that accurately correlates flow field with infrared radiation characteristics and efficiently extracts infrared radiation spectrum features to solve the problems of low spectrum extraction accuracy and unclear physical mechanism correlation in existing technologies. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for extracting the infrared radiation spectrum characteristics of engine exhaust flames. Through multi-step collaborative processing, a deep coupling analysis of exhaust flame flow field and infrared radiation is achieved, and the infrared radiation spectrum characteristics are accurately extracted.

[0006] This invention is achieved using the following technical solution:

[0007] A method for extracting the infrared radiation spectrum characteristics of engine exhaust plume includes the following steps:

[0008] Step S1: Acquire infrared thermal image data of the engine exhaust plume and perform preprocessing;

[0009] Step S2: Perform azimuth Fourier transform to reduce azimuth angle;

[0010] Step S3: Divide the preprocessed infrared thermal image data into blocks and apply a Hanning window to suppress spectral leakage; perform discrete Fourier transform on the blocks, regroup the data by frequency, and construct the cross spectral density matrix at each frequency;

[0011] Step S4: Perform eigenvalue decomposition on the cross-spectral density matrix to obtain the SPOD modes and eigenvalues ​​at each frequency. The magnitude of the eigenvalues ​​corresponds to the mode energy. After sorting them in descending order, extract the dominant mode as the core feature of the infrared radiation spectrum.

[0012] Further, step S1: acquiring infrared thermal image data of the engine exhaust plume and performing preprocessing; includes the following processes:

[0013] The infrared thermal image data is subjected to time-de-meaning and normalization processing to eliminate steady-state background interference; and the data interpolation in the Cartesian coordinate system (x, y, z) is converted to the cylindrical coordinate system (x, r, θ) to facilitate azimuth mode analysis.

[0014] Construct the snapshot matrix:

[0015] ;

[0016] For multidimensional flow field data, q i It is converted into a vector of length n, where n equals the number of variables multiplied by the number of grid points;

[0017] The instantaneous energy at each instant or snapshot is represented by the spatial inner product:

[0018] ;

[0019] Where W is the positive definite Hermitian matrix that describes the component weights, and Ω represents the flow field spatial domain.

[0020] Further, step S2: performing an azimuth Fourier transform to achieve azimuth reduction; includes the following process:

[0021] In cylindrical coordinates (x, r, θ), the data in the azimuth direction of the exhaust plume is periodic. By performing a Fourier transform on the snapshot data in this direction, the azimuth mode can be obtained:

[0022] ;

[0023] Where, q im Let i be the data for the m-th azimuth angle of variable i. For the nth azimuth angle, The number of azimuth angles is given. The azimuth angle modes follow a conjugate symmetric distribution. The time-domain signals of the circumferential modes are all real numbers. The original information is obtained by performing an inverse Fourier transform on the azimuth angle modes.

[0024] ;

[0025] The transformed tail jet flow field data for each variable i can be stored in a complex matrix. middle:

[0026] .

[0027] Further, step S3 involves: dividing the preprocessed infrared thermal image data into blocks and applying a Hanning window to suppress spectral leakage; performing a discrete Fourier transform on the blocks, regrouping the data by frequency, and constructing a cross-spectral density matrix for each frequency; including the following processes:

[0028] The Welch method is used to process matrix X. m By uniformly dividing the data into blocks, we can obtain a block matrix with overlapping snapshots. The l-th block matrix is... It can be represented as:

[0029] ;

[0030] Wherein, the k-th snapshot of the l-th block matrix is N f N represents the number of snapshots contained in each block. o N represents the number of snapshots where two adjacent block matrices overlap. blk Indicates the number of block matrices:

[0031] ;

[0032] Set the number of blocks N blk =5, each block contains N f =512 snapshots, with 50% overlap between adjacent blocks. To obtain convergent spectral results, a Hanning window is applied to each block to reduce spectral leakage. ;

[0033] Because the use of window functions can lead to information loss at the boundaries, window functions are only applied in overlapping blocks; subsequently, a weighted discrete Fourier transform is performed:

[0034] ;

[0035] in, For weights;

[0036] The generated Fourier matrix Wherein, the k-th discrete frequency component in the l-th block is ;

[0037] Generate Fourier matrix Regrouped by frequency, at a given frequency ω k Data matrix under azimuth wavenumber m middle:

[0038] ;

[0039] Constructing frequency ω k The cross spectral density matrix below:

[0040] ;

[0041] Matrix W is the numerical integration coefficient matrix of the discrete grid.

[0042] Further, step S4 involves performing eigenvalue decomposition on the cross-spectral density matrix to obtain the SPOD modes and eigenvalues ​​at each frequency. The magnitude of the eigenvalues ​​corresponds to the mode energy. After sorting them in descending order, the dominant mode is extracted as the core feature of the infrared radiation spectrum. This includes the following process:

[0043] SPOD modes are obtained by solving for the eigenvalues ​​of the cross-spectral density matrix at each frequency:

[0044] ;

[0045] ;

[0046] in, Represents the SPOD mode. Represents the eigenvector matrix, The eigenvalue matrix represents the eigenvalues, with the eigenvalues ​​corresponding to the modal energies, and they are arranged in descending order of eigenvalue size. The SPOD modes obtained by decomposition are spatially orthogonal at each frequency, and each order of SPOD mode at different frequencies is orthogonal in the original time domain.

[0047] Furthermore, based on the SPOD energy spectrum distribution, the dominant frequency and its corresponding spatial mode structure are identified; low-energy modes are truncated based on energy thresholds, while the dominant SPOD modes are retained, thus completing the spectral feature extraction.

[0048] This method can accurately extract the spatiotemporal-frequency-spectral characteristics of infrared radiation from the exhaust plumes of rocket engines and other power devices, providing key technical support for the detection and identification of infrared radiation targets. Attached Figure Description

[0049] Figure 1 This is the overall flowchart of the present invention.

[0050] Figure 2 This is a schematic diagram of the computational domain and grid distribution of the engine exhaust flow field in this invention.

[0051] Figure 3 This is a diagram showing the modal time coefficients and spectral distribution in this invention.

[0052] Figure 4 This is a diagram showing the modal energy percentage under different nozzle configurations in this invention. Detailed Implementation

[0053] The present invention will now be described in further detail with reference to the accompanying drawings and specific preferred embodiments.

[0054] In the description of this invention, it should be understood that the terms "left side," "right side," "inner side," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. The specific dimensions used in this embodiment are only for illustrating the technical solution and do not limit the scope of protection of this invention.

[0055] A method for extracting the infrared radiation spectrum characteristics of engine exhaust plumes, as shown in the attached figure. Figure 1 As shown, it includes the following steps:

[0056] Step S1: Acquire infrared thermal image data of the engine exhaust plume and perform preprocessing;

[0057] A computational model of the rocket engine exhaust flow field was established: Taking the rocket engine exhaust flame as the research object, the unsteady exhaust flame flow field ejected through the rocket engine nozzle was simulated using the large eddy simulation (LES) method based on computational fluid dynamics (CFD) and considering the combustion chemical reaction process. The obtained flow field data (including flow field temperature, pressure, density, and composition) was used as the underlying data for infrared radiation and spectral characteristic extraction. The apparent ray method was used to divide the three-dimensional flow field into several uniform isothermal layers along the detector observation direction. The particle scattering effect was ignored, and the absorption and emission of each medium were calculated. The radiative transfer equation was solved to obtain the infrared radiation intensity data of 2.7 μm (H2O characteristic band) and 4.3 μm (CO2 characteristic band), forming an infrared thermal image sequence.

[0058] A small solid rocket motor with a nozzle diameter of 25 mm was used as the subject, with a combustion chamber pressure of 4.3 MPa and a nozzle expansion ratio of 2.8. Based on computational fluid dynamics (CFD), the large eddy simulation (LES) method, which considers the combustion chemical reaction process, was used to simulate the unsteady tail jet flow field ejected through the rocket motor nozzle.

[0059] As attached Figure 2As shown, the computational domain has an axial length of 2.5 m and a radial radius of 0.5 m. A structured mesh is used, with local refinement near the nozzle and in the exhaust region, resulting in a mesh count of 4.3 million. Gas inlet parameters (temperature 1963 K, pressure 288000 Pa, Mach number 2.35, component mole fractions including H2O 0.400, CO2 0.136, etc.) are applied at the nozzle. Ambient back pressure and inlet parameters are applied at the inlet and outlet, respectively. The far-field boundary is set as the pressure far-field. Flow field snapshot data at 1024 time steps are obtained through numerical simulation, including parameters such as temperature, pressure, axial velocity, and mass fractions of H2O, CO, and CO2 components.

[0060] Data from 1024 time steps were collected. Based on the NASA-SP-3080 spectral database, a statistical narrow band model combined with a single-line group model was used to calculate the spectral absorption coefficients and transmittance of components such as H2O, CO2, and CO in the exhaust flame within the temperature range of 300~3000K.

[0061] Using the apparent ray method, the three-dimensional exhaust flow field is divided into several uniform isothermal layers along the detector observation direction. Ignoring the particle scattering effect, the absorption and emission of each medium are calculated recursively, and the radiative transfer equation is solved to obtain infrared radiation intensity data at 2.7 μm (H2O characteristic band) and 4.3 μm (CO2 characteristic band), forming an infrared thermal image sequence.

[0062] The infrared thermal image sequence is subjected to time-de-meaning and normalization processing to eliminate steady-state background interference; and the data interpolation in the Cartesian coordinate system (x, y, z) is converted to the cylindrical coordinate system (x, r, θ) to facilitate azimuth mode analysis.

[0063] Construct the snapshot matrix:

[0064] ;

[0065] For multidimensional flow field data, q i It is converted into a vector of length n, where n equals the number of variables multiplied by the number of grid points;

[0066] The instantaneous energy at each instant or snapshot is represented by the spatial inner product:

[0067] ;

[0068] Where W is the positive definite Hermitian matrix describing the component weights, and Ω represents the flow field spatial domain. For statistically stationary data, a two-point cross-spectral density matrix is ​​constructed and its eigenmodes are solved. A convergent estimate of the spectral density needs to be obtained by spectral averaging across multiple independent processes of the flow; therefore, the Welch method is used to divide the preprocessed infrared thermal image data into blocks with a 50% overlap between adjacent blocks, and a periodic Hanning window is applied to suppress spectral leakage.

[0069] Step S2: Perform azimuth Fourier transform to reduce azimuth angle;

[0070] In cylindrical coordinates (x, r, θ), the data in the azimuth direction of the exhaust plume is periodic. By performing a Fourier transform on the snapshot data in this direction, the azimuth mode can be obtained:

[0071] ;

[0072] Where, q im Let i be the data for the m-th azimuth angle of variable i. For the nth azimuth angle, The number of azimuth angles is given. The azimuth angle modes follow a conjugate symmetric distribution. The time-domain signals of the circumferential modes are all real numbers. The original information is obtained by performing an inverse Fourier transform on the azimuth angle modes.

[0073] ;

[0074] The transformed tail jet flow field data for each variable i can be stored in a complex matrix. middle:

[0075] .

[0076] Step S3: Divide the preprocessed infrared thermal image data into blocks and apply a Hanning window to suppress spectral leakage; perform discrete Fourier transform on the blocks, regroup the data by frequency, and construct the cross spectral density matrix at each frequency;

[0077] The Welch method is used to process matrix X. m By uniformly dividing the data into blocks, we can obtain a block matrix with overlapping snapshots. The l-th block matrix is... It can be represented as:

[0078] ;

[0079] Wherein, the k-th snapshot of the l-th block matrix is N f N represents the number of snapshots contained in each block. o N represents the number of snapshots where two adjacent block matrices overlap. blk Indicates the number of block matrices:

[0080] ;

[0081] Set the number of blocks N blk =5, each block contains N f=512 snapshots, with 50% overlap between adjacent blocks. To obtain convergent spectral results, a Hanning window is applied to each block to reduce spectral leakage. ;

[0082] Because the use of window functions can lead to information loss at the boundaries, window functions are only applied in overlapping blocks; subsequently, a weighted discrete Fourier transform is performed:

[0083] ;

[0084] in, For weights;

[0085] The generated Fourier matrix Wherein, the k-th discrete frequency component in the l-th block is ;

[0086] Generate Fourier matrix Regrouped by frequency, at a given frequency ω k Data matrix under azimuth wavenumber m middle:

[0087] ;

[0088] Constructing frequency ω k The cross spectral density matrix below:

[0089] ;

[0090] Matrix W is the numerical integration coefficient matrix of the discrete grid.

[0091] Step S4: Perform eigenvalue decomposition on the cross-spectral density matrix to obtain the SPOD modes and eigenvalues ​​at each frequency. The magnitude of the eigenvalues ​​corresponds to the mode energy. After sorting them in descending order, extract the dominant mode as the core feature of the infrared radiation spectrum.

[0092] SPOD modes (i.e., spectral eigenorthogonal decomposition modes) are obtained by solving the eigenvalues ​​of the cross-spectral density matrix at each frequency.

[0093] ;

[0094] ;

[0095] in, Represents the SPOD mode. Represents the eigenvector matrix, The eigenvalue matrix represents the eigenvalues, with the eigenvalues ​​corresponding to the modal energies, and they are arranged in descending order of eigenvalue size. The SPOD modes obtained by decomposition are spatially orthogonal at each frequency, and each order of SPOD mode at different frequencies is orthogonal in the original time domain.

[0096] Based on the SPOD energy spectrum distribution, the dominant frequency and its corresponding spatial mode structure are identified; low-energy modes are truncated based on energy thresholds, while the dominant SPOD modes are retained, thus completing the spectral feature extraction.

[0097] Appendix Figure 3 This is a time coefficient and power spectral density plot of the 1st, 10th, 30th, and 100th modes of a temperature field provided in an embodiment of the present invention. From Figure 3 As can be seen, the temporal fluctuation amplitude of the first-order mode is significantly higher than that of other higher-order modes, indicating that this mode plays a dominant role in the overall flow field. With increasing mode order, the amplitude of the mode coefficients gradually decreases, and the fluctuations become more intense. This indicates that the higher-order modes represent stronger flow variability, and their contribution to the flow field is relatively smaller. It can be observed that the spectrum of the first-order mode is mainly concentrated in the low-frequency range, with a dominant frequency of approximately 292 Hz and a frequency range roughly between 100-1000 Hz, indicating that this mode mainly exhibits low-frequency oscillations and occupies the majority of energy in the flow. With increasing mode order, the mode frequency gradually increases, while the peak frequency tends to be lower, reflecting that the flow characteristics corresponding to higher-order modes have higher time-varying frequencies and exhibit faster and more fragmented evolutionary characteristics. This phenomenon reveals the temporal evolution characteristics of different order modes: lower-order modes dominate large-scale, low-frequency flow patterns and contain more energy; while higher-order modes are more complex, have less energy, and exhibit rapid changes over time, representing local details and high-frequency fluctuations in the flow field.

[0098] Figure 4 This is a comparison of the infrared radiation spectra of a circular nozzle and a square nozzle in the 2.7 μm and 4.3 μm bands, according to an embodiment of the present invention. As can be seen from the figure, the SPOD spectra of the infrared thermal images all exhibit a free jet spectral distribution, meaning that the main energy distribution in the thermal image is in the low-frequency range, and the modal energy decreases rapidly with increasing frequency. With increasing SPOD mode order, the modal energy also decays rapidly, indicating that low-order SPOD modes occupy the vast majority of the thermal image energy at the corresponding frequencies.

[0099] In the description of this invention, it should be understood that the indicated orientation or positional relationship is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing this invention and simplifying the description, and is not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0100] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for extracting the infrared radiation spectrum characteristics of engine exhaust plumes, characterized in that: Includes the following steps: Step S1: Acquire the infrared thermal image sequence of the engine exhaust plume and perform preprocessing; Step S2: Perform azimuth Fourier transform to reduce azimuth angle; Step S3: Divide the preprocessed infrared thermal image data into blocks and apply a Hanning window to suppress spectral leakage; perform discrete Fourier transform on the blocks, regroup the data by frequency, and construct the cross spectral density matrix at each frequency; Step S4: Perform eigenvalue decomposition on the cross-spectral density matrix to obtain the SPOD modes and eigenvalues ​​at each frequency. The magnitude of the eigenvalues ​​corresponds to the mode energy. After sorting them in descending order, extract the dominant mode as the core feature of the infrared radiation spectrum. Step S3: The preprocessed infrared thermal image data is divided into blocks, and a Hanning window is applied to suppress spectral leakage; the blocks are subjected to discrete Fourier transform, and the data is regrouped by frequency to construct the cross-spectral density matrix at each frequency; including the following processes: The Welch method is used to process matrix X. m By uniformly dividing the data into blocks, we can obtain a block matrix with overlapping snapshots. The l-th block matrix is... It can be represented as: ； Wherein, the k-th snapshot of the l-th block matrix is N f N represents the number of snapshots contained in each block. o N represents the number of snapshots where two adjacent block matrices overlap. blk Indicates the number of block matrices: ； Set the number of blocks N blk =5, each block contains N f =512 snapshots, with 50% overlap between adjacent blocks. To obtain convergent spectral results, a Hanning window is applied to each block to reduce spectral leakage. ; Because the use of window functions can lead to information loss at the boundaries, window functions are only applied in overlapping blocks; subsequently, a weighted discrete Fourier transform is performed: ； in, For weights; The generated Fourier matrix Wherein, the k-th discrete frequency component in the l-th block is ; Generate Fourier matrix Regrouped by frequency, at a given frequency ω k Data matrix under azimuth wavenumber m middle: ； Constructing frequency ω k The cross spectral density matrix below: ； Matrix W is the numerical integration coefficient matrix of the discrete grid.

2. The method for extracting the infrared radiation spectrum characteristics of engine exhaust plume according to claim 1, characterized in that: Step S1: Acquire infrared thermal image data of the engine exhaust plume and perform preprocessing; The process includes the following: Time-de-averaging and normalization are performed on the infrared thermal image sequence to eliminate steady-state background interference; The data interpolation in the Cartesian coordinate system (x, y, z) is converted to the cylindrical coordinate system (x, r, θ) to facilitate azimuth mode analysis. Construct the snapshot matrix: ； For multidimensional flow field data, q i It is converted into a vector of length n, where n equals the number of variables multiplied by the number of grid points; The instantaneous energy at each instant or snapshot is represented by the spatial inner product: ； Where W is the positive definite Hermitian matrix that describes the component weights, and Ω represents the flow field spatial domain.

3. The method for extracting the infrared radiation spectrum characteristics of engine exhaust plume according to claim 2, characterized in that: Step S2: Perform azimuth Fourier transform to reduce azimuth angle; The process includes the following: In cylindrical coordinates (x, r, θ), the data in the azimuth direction of the exhaust plume is periodic. By performing a Fourier transform on the snapshot data in this direction, the azimuth mode can be obtained: ； Where, q im Let i be the data for the m-th azimuth angle of variable i. For the nth azimuth angle, The number of azimuth angles is given. The azimuth angle modes follow a conjugate symmetric distribution. The time-domain signals of the circumferential modes are all real numbers. The original information is obtained by performing an inverse Fourier transform on the azimuth angle modes. ； The transformed tail jet flow field data for each variable i can be stored in a complex matrix. middle: 。 4. The method for extracting the infrared radiation spectrum characteristics of engine exhaust plume according to claim 3, characterized in that: Step S4: Perform eigenvalue decomposition on the cross-spectral density matrix to obtain the SPOD modes and eigenvalues ​​at each frequency. The magnitude of the eigenvalues ​​corresponds to the mode energy. After sorting them in descending order, the dominant mode is extracted as the core feature of the infrared radiation spectrum. include The following process: SPOD modes are obtained by solving for the eigenvalues ​​of the cross-spectral density matrix at each frequency: ； ； in, Represents the SPOD mode. Represents the eigenvector matrix, This represents the eigenvalue matrix, where the magnitude of each eigenvalue corresponds to the magnitude of the modal energy, and the eigenvalues ​​are sorted in descending order. The decomposed SPOD modes are spatially orthogonal at each frequency, and each order of SPOD mode at different frequencies is orthogonal in the original time domain.

5. The method for extracting the infrared radiation spectrum characteristics of engine exhaust plume according to claim 4, characterized in that: Based on the SPOD energy spectrum distribution, the dominant frequency and its corresponding spatial mode structure are identified; low-energy modes are truncated based on energy thresholds, while the dominant SPOD modes are retained, thus completing the spectral feature extraction.

Images