A method for measuring gas velocity in absorption spectra by constructing and matching absorption peak image features.

Abstract

This invention discloses a method for measuring gas velocity in absorption spectra by constructing and matching absorption peak image features. The method includes: performing multiple phase-shift downsampling on a detected wavelength-modulated laser signal; reconstructing the multiple phase-shift sequences to obtain a two-dimensional grayscale image matrix containing the direct absorption spectrum profile; sharpening the two-dimensional grayscale image matrix extracted from the laser signal where the flow field velocity is zero to obtain an image template characterizing the absorption peak position; matching the position of the image template from the two-dimensional grayscale images obtained from laser detection signals in two directions within the flow field under test, based on the absorption peak position image template, to determine the offset of the absorption peak positions in the two directions; and calculating the airflow velocity using the absorption peak offsets in the two detection directions. This invention improves the accuracy of solving the absorption spectrum frequency shift at different optical depths, thereby improving the velocity measurement accuracy in complex flow fields with varying operating conditions.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of gas parameter measurement, and particularly relates to an absorption spectrum gas velocity measurement method based on absorption peak image feature construction and matching. BACKGROUND

[0002] Rocket and aero-engine rely on continuous design and validation cycles to achieve continuous updates. Accurate measurement of gas velocity and temperature is crucial to accelerate this iterative process. Currently, the measurement of these parameters mainly adopts contact methods, such as using thermocouples to measure temperature and using pitot tubes to measure gas velocity. The contact method requires the probe to contact the flow field and sense the parameters, which will affect the original flow field and cause slow response. In recent years, many non-contact combustion monitoring methods have been developed for measuring temperature and gas velocity in high-temperature and high-speed flow fields. Among numerous non-contact methods, TDLAS (tunable diode laser absorption spectroscopy) technology has unique advantages in fast resolution, quantitative measurement and low cost. TDLAS achieves simultaneous measurement of multiple gas parameters, including gas concentration, temperature, velocity and pressure, by measuring the absorption spectrum of specific molecules in the flow field.

[0003] In terms of velocity measurement, TDLAS technology measures gas velocity by obtaining the Doppler frequency shift of the absorption spectrum of molecules in the flow field. Compared with other non-contact measurement methods such as PIV (particle image velocimetry) and KTV (krypton tagging velocimetry), TDLAS technology uses existing gas molecules in the flow field as tracer molecules without the need for additional tracer molecules or particles. Unlike LDV (laser Doppler velocimetry)

[18] , which measures velocity from scattered laser signals, TDLAS-based velocity measurement is extracted from transmitted signals, with higher signal-to-noise ratio. Therefore, TDLAS is considered as a promising non-invasive and high-precision flow rate measurement technology.

[0004] TDLAS measurement methods include DAS (direct absorption spectroscopy) and WMS (wavelength modulation spectroscopy). The DAS method uses a linear wavelength scanning laser to directly obtain the complete absorption spectrum including the absorption center and its two wings by comparing the transmitted laser intensity and the reference laser intensity, and derives the parameters by fitting the complete absorption spectrum. In addition, by comparing with other absorption spectra in static or laser paths with different laser directions, the Doppler frequency shift can be directly extracted. However, the accuracy of the absorption spectrum is sensitive to the reference laser intensity, which is usually fitted from the transmitted laser signal. In the actual measurement environment of high-temperature and high-speed airflow, non-absorption loss caused by beam steering effect or particle scattering will cause fluctuations in the transmitted signal, which affects the fitting accuracy of the reference laser intensity and limits the accuracy of the absorption spectrum measurement.

[0005] Unlike the DAS method, the WMS method modulates the laser at a high frequency and quadrature demodulates the transmitted WMS signal to obtain a high-precision harmonic signal of the absorption spectrum. The normalized 2f / 1f harmonic based on the WMS method can effectively avoid background noise and fluctuations in the intensity of the laser signal and is widely used in gas concentration and temperature detection. However, the speed measurement based on the Doppler effect depends on the measurement of the frequency shift of the laser, and the harmonic extracted from the transmitted WMS signal using the time-based lock-in amplification technology lacks clear frequency coordinate information. Therefore, it is not possible to directly use the extracted Doppler frequency shift information to obtain the speed from the harmonic signal. In some studies, the peak of the harmonic is usually used to correspond to the center wavelength of the absorption spectrum, but this assumption usually needs to meet complex restrictions, such as a phase difference of π between the wavelength modulation and the intensity modulation. Therefore, although the WMS method can obtain high-precision harmonic signals, the extraction of the Doppler frequency shift of the laser from the harmonic signal still requires a complex calibration process. SUMMARY

[0006] The technical problem to be solved by the present application is to provide an absorption spectrum gas speed measurement method for constructing and matching absorption peak image features.

[0007] To achieve the above-mentioned purpose, the technical scheme adopted by the present application is as follows:

[0008] A laser absorption spectrum gas speed measurement method, comprising:

[0009] Step S1, multiple phase shift down-sampling is performed on the detected wavelength modulated laser signal, and a two-dimensional gray scale image matrix containing a direct absorption spectrum profile is obtained by reconstructing the multiple phase shift sequence;

[0010] Step S2, an image template representing the absorption peak position is obtained according to the two-dimensional gray scale image matrix;

[0011] Step S3, the offset of the absorption peak position in two directions is determined according to the image template of the absorption peak position;

[0012] Step S4, the flow speed is calculated using the absorption peak offset in two detection directions.

[0013] Preferably, in step S2, the two-dimensional gray scale image matrix extracted from the laser signal with a flow field speed of zero is sharpened to obtain the image template representing the absorption peak position.

[0014] Preferably, in step S3, the positions of the matched image templates in the two-dimensional gray scale images obtained from the laser detection signals in two directions of the flow field to be measured are determined according to the image template of the absorption peak position, and the offset of the absorption peak position in two directions is determined.

[0015] The present application reconstructs a two-dimensional image matrix containing a direct absorption spectrum profile from a phase shift sequence obtained by multiple phase shifting and down sampling of a WMS signal, extracts a Doppler frequency shift amount of the absorption spectrum from a sharpening feature of the direct absorption spectrum profile, and realizes airflow velocity measurement by using a template image for image feature matching. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only constitute a part of the embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on the provided drawings.

[0017] Figure 1 The flow chart of the absorption spectrum gas velocity measurement method for constructing and matching the absorption peak image features of the embodiments of the present application;

[0018] Figure 2 The laser absorption spectrum optical path diagram;

[0019] Figure 3 The cross optical path test schematic diagram;

[0020] Figure 4 The discretized wavelength modulation signal schematic diagram;

[0021] Figure 5 The synthesized down-sampling and rearranged matrix plan view;

[0022] Figure 6 The synthesized down-sampling and rearranged matrix perspective view;

[0023] Figure 7 The image schematic diagram after using filter sharpening;

[0024] Figure 8 The velocity schematic diagram obtained by using image feature matching. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0026] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0027] Embodiment 1

[0028] As shown in the drawings, the embodiment of the present application provides an absorption spectrum gas velocity measurement method for constructing and matching absorption peak image features, comprising: Figure 1

[0029] Step S1, multiple phase shift down-sampling is performed on the detected wavelength modulated laser signal, and the multiple phase shift sequence is reconstructed to obtain a two-dimensional gray scale image matrix containing a direct absorption spectrum profile;

[0030] Step S2, sharpening processing is performed on the two-dimensional gray scale image matrix extracted from the laser signal with a flow field velocity of zero to obtain an image template representing the absorption peak position;

[0031] Step S3, according to the image template of the absorption peak position, the positions of the matched image template in the two-dimensional gray scale images obtained from the laser detection signals in two directions of the flow field to be measured are determined, and the offset amounts of the absorption peak positions in the two directions are determined;

[0032] Step S4, the flow velocity is calculated using the absorption peak offset amounts in the two detection directions.

[0033] As an embodiment of the present application, in step S1, the WMS is a typical TDLAS method with excellent noise resistance. The driving current of the emitted WMS laser is composed of two components, the low-frequency component comes from the low-frequency sawtooth wave, and the high-frequency component comes from the sinusoidal wave scanning. The emitted laser wave number can be expressed as:

[0034]

[0035] Where v0 is the center frequency of the absorption spectrum. v s (t) is the function of the slow scanning of the laser frequency as a sawtooth profile, v m and α m are the modulation depth and phase shift of the modulated laser at mf H . Where the emitted laser intensity I0 is expressed as:

[0036]

[0037] Where, I s (t) is the laser intensity driven by the low-frequency sawtooth wave, b n and β n ​The intensity amplitude and phase shift of the laser modulated in the high frequency sinusoidal wave. When the emitted laser passes through the target gas, the emitted WMS laser signal is then detected by a photodiode detector. The intensity of the transmitted WMS laser signal is then recorded by a DAQ (Data Acquisition) with a sampling rate of f s .

[0038] When the laser signal transmits through the gas to be measured, such as Figure 2 The wavelength-dependent gas absorption rate change due to the absorption effect can be expressed as the ratio of the detected transmitted laser light intensity I t to the emitted laser intensity I0, which can also be expressed as:

[0039]

[0040] where l[m] is the laser absorption path length, P[atm] is the pressure, T(x)[K] and X abs (x) are the gas temperature and concentration along the gas absorption path, respectively, Δv u is the laser Doppler effect caused by the gas flow velocity , which can be expressed as,

[0041]

[0042] where c represents the speed of light, θ represents the propagation direction of the detected laser and the flow field velocity direction, as shown in Figure 3 , the flow field velocity value can be expressed as u:

[0043]

[0044] where θ1 and θ2 are the angles between the two laser propagation directions and the gas flow velocity direction and represent the laser Doppler effects of the two directions, respectively. Therefore, for the digitized representation of the detected transmitted laser light intensity I t in a single scanning period, it can be expressed as

[0045]

[0046] where N is the number of sampling points in a single scanning period. Let the low frequency value be f L , then N can be expressed as:

[0047] N = f S / f L (7)

[0048] Down-sampling is a signal processing technique that involves selecting a certain number of values from a continuous sequence of samples to form a new sequence with a reduced sampling frequency. As shown in Fig. 1 for the original sequence of a WMS signal for digital transmission, let Q be the total number of periods that a high-frequency sinusoidal wave goes through during a low-frequency sawtooth wave scanning process, and let P be the total number of sampling points in a single sinusoidal period. Since the time interval between two adjacent sampling points is equal, the phase difference between two adjacent sampling points is defined as Figure 4

[0049]

[0050] Down-sampling still needs to follow the sampling theorem, otherwise it can lead to signal distortion and aliasing. Let M be defined as the down-sampling interval:

[0051]

[0052] where f s / M means that the sampling frequency of the new down-sampled sequence is reduced to f i / M. At the same time, let each sampling point (numbered P) in the first sinusoidal period be the independent starting point of each down-sampled sequence.

[0053] Name these down-sampled sequences as polyphase sequences. Let I i denote the set of the i-th polyphase sequence, denoted as:

[0054] I i = [I t (k)|k = jM + i, 0 < j < Q, j ∈ Z] (10)

[0055] Substitute equation (7) into equation (6) to get:

[0056]

[0057] where, and are the invariants in the i-th polyphase sequence set because i is an invariant. According to this, the waveform of I i is the waveform of the sawtooth wave scanning as shown in Fig. 2, and can also be regarded as the transmitted DAS profile signal. Figure 5

[0058] Further, rearrange these polyphase sequences to create a Q x P matrix, named as D-R matrix (Down-sampling and Rearrangement matrix), denoted as:

[0059]

[0060] In different rows of I, these are different polyphase sequences, and when the gas velocity is 0, the position of the absorption peak will change with ​And the change, the function can be expressed as:

[0061]

[0062] As an embodiment of the present application, in step S2, the intensity change of the down-sampled signal rearranged in the D-R matrix represents the position of the absorption peak caused by the gas molecule absorption effect, and the position of the absorption peak can be determined by the feature matching method. However, as shown in Figure 6 , the absorption peak feature is covered by the intensity change caused by the sawtooth wave laser scanning, and sharpening this feature helps to accurately extract the position of the absorption spectrum in a noisy environment. Removing the intensity change caused by laser scanning helps to sharpen this feature and further improves the ability to accurately extract the position of the absorption spectrum in a high noise environment and small absorption condition.

[0063] Let I G represent the sharpened D-R matrix, and the sharpening kernel is G. According to the down-sampling rule, for any random point in the i-th row and j-th column of the D-R matrix, its adjacent point in the original digital transmission WMS signal is rearranged in the (i+1)-th row and (j-1)-th column and the (i-1)-th row and (j+1)-th column. Therefore, the sharpening operation is represented as:

[0064]

[0065] Where "*" represents convolution operation. I N represents I G normalized in the range of -1 to 1. The above operation makes the feature of the absorption spectrum significantly enhanced, and there is a transition from the maximum value to the minimum value near the absorption peak.

[0066] As shown in Figure 7 , the contrast of the absorption peak feature after the sharpening process is more obvious, and then by calculating the pixel position change of the absorption spectrum feature in the image, the frequency shift of the absorption spectrum caused by the Doppler effect can be calculated, so as to confirm the gas flow velocity.

[0067] Let I N|u＝0 represent the sharpened D-R matrix of the transmission laser signal when the gas flow velocity is 0. Then the matching template I K about the absorption feature is extracted from the normalized sharpened D-R matrix, represented as:

[0068]

[0069] Where "°" represents the Hadamard product of two matrices. In equation (12), I K is set to a (S2-S1) x Q matrix. The absorption feature exists in the S1-th row to the S2-th row. In addition, in order to simplify the solution, IK is set to have Q columns to avoid feature matching in the column direction.

[0070] As an embodiment of the present application, in step S3,

[0071] Let I N|u denote the sharpened D-R matrix transmitting the laser signal at flow rate u. According to equation (13), the position of the absorption peak can be expressed as:

[0072]

[0073] where j u and j0denote the position of the absorption peak when the gas velocity is u and 0, respectively. Generally, the wavelength scanning function of the sawtooth wave can be adjusted to a first-order linear function, so that further:

[0074]

[0075] where Δj u denotes the position shift of the absorption peak feature caused by the Doppler frequency shift when the flow rate is u. a is the first-order coefficient of v s .

[0076] Further, as Figure 8 shown, the position of the absorption spectrum feature is calculated using image feature matching with a template image, let p u denote the normalized cross-correlation coefficient between I K and I N|u , expressed as:

[0077]

[0078] where, denotes the average intensity of the local part in I N|u , the calculation formula is:

[0079]

[0080] Let p u be the position of the maximum value in p u (p):

[0081]

[0082] where p u denotes the position change of the absorption peak; when determining p u , a point close to p u is selected to fit the extreme point, and a non-integer p u can be obtained to improve the resolution of the measurement.

[0083] As an embodiment of the present application, in step S4, the gas velocity can be determined as:

[0084]

[0085] where, and are the values of p when u takes the values of u1 and u2. u

[0086] The present application improves the accuracy of gas velocity measurement based on TDLAS, especially for the velocity measurement in high temperature and high speed environment such as engine exhaust nozzle. The multi-phase sequence extracted from the transmitted WMS signal by down-sampling strategy is rearranged as a gray-scale image corresponding to the laser frequency coordinate, i.e. the 1-D WMS signal is converted to a 2-D gray-scale image with each row being a DAS profile with laser frequency coordinate. An image filtering method is used to sharpen the absorption spectral features in the two-dimensional gray-scale image instead of the reference laser intensity fitting in the traditional DAS method. It can avoid the non-absorption loss that affects the accuracy of reference laser intensity fitting and limits the accuracy of absorption spectrum measurement. An image feature matching method is used to calculate the Doppler shift to derive the velocity, and both the intensity and two-dimensional distribution of the absorption spectral features are used to improve the matching accuracy within a wide absorption range. The proposed image processing method plays an important role in promoting advanced TDLAS parameter solutions and provides the possibility for future online fast solution of TDLAS parameters using GPU devices that are good at parallel image calculation.

[0087] The above-described embodiments are merely intended to describe the preferred modes of the present application, and are not intended to limit the scope of the present application. Various modifications and improvements of the present application made by those skilled in the art without departing from the design spirit of the present application shall fall within the scope of the present application as defined in the claims.​

Claims

1. A method of laser absorption spectroscopy gas velocity measurement, characterized by, The method comprises the following steps: Step S1, multiple phase shift undersampling is performed on the detected wavelength modulated laser signal, a multiple phase shift sequence is reconstructed, and a two-dimensional gray image matrix containing a direct absorption spectrum profile is obtained; Step S2, an image template representing the absorption peak position is obtained according to the two-dimensional gray image matrix; Step S3, the offset of the absorption peak position in two directions is determined according to the image template of the absorption peak position; Step S4, the gas flow velocity is calculated by using the absorption peak offset in two detection directions; The gas flow velocity is represented as: where, is the total number of sampling points in a single sinusoidal wave period, is the first order coefficient of is the function of the slow scan of the laser frequency as a sawtooth wave profile; and is the value of when and represents the change of the position of the absorption peak; when determining , a point close to is selected to fit the extreme point, and a non-integer is obtained to improve the resolution of the measurement; and are the angles between the two laser propagation directions and the airflow velocity direction, respectively; is the center frequency of the absorption spectrum; c represents the speed of light;​​​ In step S2, the two-dimensional gray image matrix extracted from the laser signal with a zero flow field velocity is sharpened to obtain the image template representing the absorption peak position; In step S3, the offset of the absorption peak position in two directions is determined according to the position of the matched image template in the two-dimensional gray image obtained from the laser detection signal in two directions of the flow field to be measured.

Images