A high-bandwidth flow rate diagnostic method based on molecular spectroscopy

Abstract

The application discloses a high-bandwidth flow rate diagnostic method based on molecular spectrum, applies a frequency locking technology to a gas flow rate measuring technology based on molecular spectrum, realizes high-bandwidth gas flow rate measurement, and improves data acquisition quality and analysis accuracy. The scheme is specifically as follows: a laser is controlled to output a single-wavelength laser signal of a target gas absorption line center wavelength to be measured. The signal is converted into a modulated optical signal by an electro-optical modulator, and a time-domain electrical signal is obtained by a gas flow field and a photoelectric detector. A high-speed frequency domain transformation model is built for frequency domain transformation, a main band frequency shift database and a sideband wavelength database are established according to the frequency domain electrical signal. According to the intensity difference between the left and right sidebands in the sideband wavelength database, frequency locking feedback control is performed on the driving current signal of the laser controller, so that the laser controller can stably track the laser center wavelength to the gas absorption peak; and the gas flow field speed is continuously calculated according to the center wavelength frequency shift amount in the main band frequency shift database.

Description

Technical Field

[0001] This invention relates to the field of optical diagnostic technology, and more specifically to a high-bandwidth flow rate diagnostic method based on molecular spectroscopy. Background Technology

[0002] Researching gas flow field velocity measurement technology is of great significance in addressing the challenges of measuring ultra-high-speed flow velocities in aerospace applications. This technology contributes accurate and crucial data for research in areas such as engine optimization and aircraft system improvement.

[0003] Existing gas velocity measurement technologies can be divided into contact and non-contact methods. Contact measurement is a traditional method, including methods such as pneumatic probes and hot-wire anemometers. Compared with contact methods, non-contact methods have advantages such as not interfering with the measured field, higher spatiotemporal resolution, and a larger velocity measurement range, and are therefore widely used in scientific research and engineering applications.

[0004] Gas flow rate measurement technology based on molecular spectroscopy is a non-contact measurement method that utilizes the absorption characteristics of gas molecules to measure gas flow rate. This technology is based on the absorption of light by gas molecules at specific wavelengths during light propagation, generating an absorption spectrum. The flow rate is then diagnosed by analyzing the frequency shift of the absorption spectrum. Currently, this technology uses laser scanning over a specific wavelength range, reflecting the flow rate information in the frequency shift of the molecular spectral lines within that specific wavelength range. This results in obtaining flow rate information only once per cycle. However, the use of laser scanning (obtaining flow rate information only once per cycle) severely limits the number of flow rate measurements per unit time, resulting in low temporal resolution. Therefore, it cannot meet the requirements for measuring faster and more non-uniform velocities in real-world operating conditions.

[0005] Frequency locking techniques are frequently used to improve measurement resolution when measuring other physical quantities of gases. While widely applied in experiments measuring physical quantities in the time domain, frequency locking has not yet been applied to gas flow velocity measurement techniques based on molecular spectroscopy (frequency domain).

[0006] In summary, while there has been much research on scanning-based molecular spectroscopy-based gas velocity measurement techniques, a non-scanning molecular spectroscopy-based gas velocity measurement technique has not yet been developed using frequency locking technology to achieve higher bandwidth velocity measurement. Summary of the Invention

[0007] In view of this, the present invention provides a high-bandwidth flow velocity diagnostic method based on molecular spectroscopy, which can apply frequency locking technology in molecular spectroscopy to measure high-bandwidth gas flow field velocities, improve data acquisition quality and analysis accuracy, and achieve higher bandwidth flow velocity measurement.

[0008] To achieve the above objectives, the technical solution of the present invention includes the following steps:

[0009] S1: Set the laser controller to generate a control signal to control the laser to output a single-wavelength laser signal corresponding to the center wavelength of the absorption line of the target gas.

[0010] S2: A modulation signal is generated by a signal generator and input into an electro-optic modulator. The electro-optic modulator externally modulates the single-wavelength laser signal at a modulation frequency, so that the optical signal generates symmetrical sideband wavelengths on both sides of the center wavelength, and obtains the modulated optical signal.

[0011] S3: By constructing an optical path, the modulated optical signal is incident into the gas flow field at a certain angle. The modulated optical signal interacts with the target gas and undergoes energy absorption. The laser transmission signal exiting the gas flow field is received by the photodetector, which converts the optical signal into a time-domain electrical signal.

[0012] S4: Build a high-speed frequency domain transformation model, which is used to convert time-domain electrical signals into frequency-domain electrical signals.

[0013] S5: Establish a main band frequency shift database and a side band wavelength database.

[0014] The sideband wavelengths are extracted from the frequency domain electrical signal and stored in the sideband wavelength database. Based on the intensity difference between the left and right sidebands of the main band in the sideband wavelength database, the error signal required for tracking and locking the laser center wavelength to the absorption peak of the gas to be measured is calculated. Based on the error signal, a frequency-locked feedback controller is used to control the drive current signal in real time and return to S1, so that the laser controller can stably track the laser center wavelength to the gas absorption peak under the action of the drive current signal.

[0015] The main band center wavelength is extracted by the frequency domain electrical signal, and then the difference between it and the center wavelength of the absorption line of the target gas is calculated. The absolute value of the difference is used as the main band frequency shift and stored in the main band frequency shift database. The gas flow field velocity is continuously calculated based on the center wavelength frequency shift in the main band frequency shift database.

[0016] Preferably, in S2, a modulation signal is generated by a signal generator, wherein the frequency of the frequency modulation signal generated by the signal generator is above GHz.

[0017] Preferably, in S4, a high-speed frequency domain transformation model is built, with a transformation speed of nanoseconds or higher.

[0018] Preferably, in S4, a high-speed frequency domain transformation model is constructed, specifically a Convolutional Neural Network (CNN) model. This model includes an input layer, an output layer, a combination layer of two convolutional layers and a pooling layer, and a fully connected layer. The output of each layer incorporates a ReLU activation function to introduce nonlinearity and fit the nonlinear mapping relationship between the input and output. The input layer receives a time-domain electrical signal, and the number of features in the input layer is determined by the number of data points in the time-domain electrical signal. The output layer outputs a frequency-domain electrical signal, and the number of features in the output layer is determined by the number of data points in the frequency-domain electrical signal. The CNN model uses a combination of the ADAM algorithm and the backpropagation algorithm to update the weights of the neural network. By calculating the gradient direction of the loss function, the weights are adjusted along the gradient direction to reduce the absolute value of the difference between the model's prediction and the actual result; the MSE function is selected as the loss function.

[0019] Preferably, based on the intensity difference between the left and right sidebands of the main band in the sideband wavelength database, the error signal required for laser center wavelength tracking and locking to the absorption peak of the gas to be measured is calculated. Specifically, the peak value difference between the left and right sidebands of the main band in the sideband wavelength database is used as the error signal, and the error signal formula is expressed as follows:

[0020] e(ω)=J1(ω)-J -1 (ω)

[0021] J n (ω) represents the nth-order Bessel function with angular frequency ω, where n = 1 or -1; when the error signal value e(ω) is 0, it means that the output wavelength of the laser is located at the center wavelength of the absorption line, indicating that the sideband has been leveled.

[0022] Preferably, based on the intensity difference between the left and right sidebands of the main band in the sideband wavelength database, the error signal required for laser center wavelength tracking and locking to the absorption peak of the gas to be measured is calculated. Specifically, the peak ratio of the left and right sidebands of the main band in the sideband wavelength database is used as the error signal, and the error signal formula is expressed as follows:

[0023] e(ω)=J1(ω) / J -1 (ω)

[0024] J n (ω) represents the nth-order Bessel function with angular frequency ω, where n = 1 or -1; when the error signal value e(ω) is 1, it means that the output wavelength of the laser is located at the center wavelength of the absorption line, indicating that the sideband has been leveled.

[0025] Preferably, the frequency-locked feedback controller is a PID controller, an active disturbance rejection controller, a sliding mode controller, or an adaptive controller, etc.

[0026] Furthermore, based on the error signal, a frequency-locked feedback controller is used to control the drive current signal in real time, enabling the laser controller to stably track the laser center wavelength to the gas absorption peak under the action of the drive current signal. Specifically, the control signal output by the PID controller is:

[0027]

[0028] Where K P K I K D These are the proportional coefficient, integral coefficient, and derivative coefficient, respectively; t is the time variable; and e(t) is the difference between the input error signal and the actual output.

[0029] After processing by the PID control module, the control signal is output to the laser controller, which then controls the laser's output wavelength to remain stable at the center of the absorption line in real time according to the entire PID feedback control loop.

[0030] Furthermore, based on the center wavelength frequency shift in the main band frequency shift database, the gas flow field velocity is continuously calculated. Specifically, the main band frequency shift Δν in the main band frequency shift database is used... shift The final gas flow field velocity is calculated using the following formula:

[0031]

[0032] Where u is the gas velocity, ν is the value to be determined; θ is the angle at which the modulated light signal enters the gas flow field; and ν0 is the center frequency of the absorption line of the target gas.

[0033] Beneficial effects:

[0034] This invention provides a high-bandwidth flow velocity diagnostic method based on molecular spectroscopy. The method combines real-time frequency locking technology, time-frequency domain conversion algorithms, and laser electro-optic external modulation methods in molecular spectroscopy to achieve continuous measurement of gas flow field velocity. This overcomes the time resolution limitations of scanning (discrete) gas flow velocity measurement, broadens the applicable measurement range, and changes the current situation where frequency locking technology has not been applied to gas flow velocity measurement based on molecular spectroscopy. It also achieves higher measurement accuracy and solves the technical challenge of lacking a high-bandwidth (continuous, non-scanning) high-precision gas flow velocity measurement method based on molecular spectroscopy. Attached Figure Description

[0035] Figure 1 Diagram of a high-bandwidth gas velocity measurement system;

[0036] Figure 2 A flowchart of a high-bandwidth gas flow field velocity measurement technique provided for a preferred embodiment of the frequency locking method;

[0037] Figure 3 These are time-domain and frequency-domain signal diagrams before and after absorption in an embodiment of the present invention;

[0038] Figure 4 Preferred embodiment of the high-speed frequency domain conversion model: Convolutional neural network model diagram;

[0039] Figure 5 This is a control block diagram of one embodiment of a frequency-locked feedback controller. Detailed Implementation

[0040] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0041] This invention provides a high-bandwidth gas flow rate measurement method based on molecular spectroscopy. The measurement system is shown in the attached figure. Figure 1 As shown, the system includes a laser controller 1, a laser 2, a high-frequency modulation signal source 3, an electro-optic modulator 4, a gas flow field 5, a photodetector 6, a high-speed frequency domain conversion model 7, a main band frequency shift database 8, a side band frequency domain database 9, an error signal generator 10, a frequency-locked feedback controller 11, and a smooth speed database 12.

[0042] Laser controller 1 controls the output optical signal of laser 2 through drive current. High-frequency modulation signal source 3 provides a high-frequency signal to electro-optic modulator 4. Electro-optic modulator 4 performs external phase modulation on the optical signal output by laser 2, changing the waveform of the optical signal. The modulated optical signal passes through gas flow field 5 and is transmitted into photodetector 6. Photodetector 6 converts the optical signal into a time-domain electrical signal. High-speed frequency domain conversion model 7 converts the time-domain electrical signal generated by photodetector 6 into a frequency-domain signal, which is then classified and stored in the main band frequency shift database 8 and the side band frequency domain database 9. Error signal generator 10 generates an error signal based on the information in the side band frequency domain database 9. Frequency locking feedback controller 11 outputs drive current to control laser controller 1 based on the error signal, causing laser 2 to lock onto the optical signal at the desired output frequency. Under the premise of real-time frequency locking control, the final smooth speed database 12 information is calculated using the information in the main band frequency shift database 8 and the side band frequency domain database 9.

[0043] The procedure for measuring high-bandwidth gas flow velocity is attached. Figure 2 As shown. The specific steps are as follows:

[0044] S1: Set the laser controller to generate a control signal to control the laser to output a single-wavelength laser signal corresponding to the center wavelength of the absorption line of the target gas.

[0045] S2: A modulation signal is generated by a signal generator and input into an electro-optic modulator. The electro-optic modulator is used to externally modulate the single-wavelength laser signal at a modulation frequency, so that the optical signal generates symmetrical sideband wavelengths on both sides of the center wavelength, and the modulated optical signal is obtained. In this embodiment of the invention, the frequency of the frequency modulation signal generated by the signal generator is above GHz. Figure 3 This demonstrates the sideband frequency ω0-ω caused by external phase modulation. m and ω0+ω m , where ω m ω is the modulation frequency, and ω0 is the center frequency.

[0046] S3: By constructing an optical path, the modulated optical signal is incident into the gas flow field at a certain angle. The modulated optical signal interacts with the target gas, resulting in energy absorption. The laser transmission signal exiting the gas flow field is received by a photodetector, which converts the optical signal into a time-domain electrical signal. The offset varies depending on the angle at which the modulated optical signal is incident; θ is the angle between the laser beam and the flow velocity, and a value closer to 0° is preferable.

[0047] S4: Construct a high-speed frequency domain transformation model. This model is used to convert time-domain electrical signals into frequency-domain electrical signals. In this embodiment of the invention, the conversion speed is at the nanosecond level or higher. In practice, conversion speeds at the second, microsecond, millisecond, and nanosecond levels can be achieved. A preferred embodiment of the ultra-high-speed frequency domain transformation model of this invention is detailed in the appendix. Figure 4 As shown in this embodiment, 100,000 sets of high-frequency time-domain signals with a simulated frequency range of 0.01GHz-100GHz are set and transformed into corresponding frequency-domain signal data using the Fast Fourier Transform algorithm, and a training dataset is constructed based on this. The training dataset is divided into training set:validation set:test set = 8:1:1. The training set and validation set are used for model training, while the test set is only used for evaluating the model's generalization performance. A convolutional neural network (CNN) model is constructed, consisting of an input layer, an output layer, a combination layer of two convolutional layers and a pooling layer, and a fully connected layer. The output of each layer incorporates a ReLU activation function to introduce nonlinearity and fit the nonlinear mapping between the input and output. The input layer receives a time-domain electrical signal, and the number of features in the input layer is determined by the number of data points in the time-domain electrical signal. The output layer outputs a frequency-domain electrical signal, and the number of features in the output layer is also determined by the number of data points in the frequency-domain electrical signal. The CNN model uses a combination of the ADAM algorithm and backpropagation to update the neural network weights. By calculating the gradient direction of the loss function, the weights are adjusted along the gradient direction to reduce the absolute value of the difference between the model's prediction and the actual result. The MSE function is selected as the loss function.

[0048] The number of convolutional layers, pooling layers, and fully connected layers, as well as hyperparameters such as batch size, learning rate, and number of iterations, all need to be continuously optimized during training to determine the optimal and most concise combination of parameters and architecture.

[0049] The ultra-high-speed frequency domain transform model, obtained through the above training method, is first used to predict simulated time-domain data on a test set to verify its generalization performance. Finally, the noisy time-domain signal obtained in step S3 is input into the ultra-high-speed frequency domain transform model for analysis to obtain the predicted frequency-domain electrical signal.

[0050] S5: Collect center wavelength data and sideband wavelength data of frequency domain electrical signals, and establish a main band frequency shift database and a sideband wavelength database; the specific method for constructing the main band frequency shift database and sideband wavelength data in this embodiment of the invention is as follows:

[0051] The center wavelength of the main band is extracted by the frequency domain electrical signal, and then the difference between the center wavelength of the absorption line of the target gas is calculated. The absolute value of the difference is used as the main band frequency shift and stored in the main band frequency shift database.

[0052] Sideband wavelengths are extracted from frequency domain electrical signals and stored in a sideband wavelength database.

[0053] After obtaining the main band frequency shift database and the sideband wavelength database as described above, the data in the sideband wavelength database is used for sideband leveling. The specific leveling method is as follows: Based on the intensity difference between the left and right sidebands of the main band in the sideband wavelength database, the error signal required for tracking and locking the laser center wavelength to the absorption peak of the gas under test is calculated; based on the error signal, a frequency-locked feedback controller is used to control the driving current signal in real time, so that the laser controller stably tracks the laser center wavelength to the gas absorption peak under the action of the driving current signal, and then returns to step S1. In this embodiment of the invention, the frequency-locked feedback controller can be a PID controller, an active disturbance rejection controller, a sliding mode controller, or an adaptive controller, etc.

[0054] In a preferred embodiment of the error signal generation method, the peak difference between two sidebands in the sideband wavelength database is used as the error signal, and the error signal formula can be expressed as follows:

[0055] e(ω)=J1(ω)-J -1 (ω)

[0056] J n (ω) represents the nth-order Bessel function with angular frequency ω. When the error signal value e(ω) is 0, it means that the output wavelength of the laser is located at the center wavelength of the absorption line, indicating that the sideband has been leveled.

[0057] Another preferred embodiment of the error signal generation method uses the ratio of the peak values ​​of the two sidebands in the sideband wavelength database as the error signal, and the error signal formula is expressed as:

[0058] e(ω)=J1(ω) / J -1 (ω)

[0059] J n (ω) represents the nth-order Bessel function with angular frequency ω, where n = 1 or -1; when the error signal value e(ω) is 1, it means that the output wavelength of the laser is located at the center wavelength of the absorption line, indicating that the sideband has been leveled.

[0060] The data in the main band frequency shift database is used to calculate the velocity of continuous gas flow fields. The calculation method is as follows: using the main band frequency shift Δν from the main band frequency shift database... shift The final gas flow field velocity is calculated using the following formula:

[0061]

[0062] Where u is the gas velocity, ν is the value to be determined; θ is the angle at which the modulated light signal enters the gas flow field; and ν0 is the center frequency of the absorption line of the target gas.

[0063] Figure 2 The flowchart outlines an embodiment of a frequency-locking feedback controller, employing PID control as the frequency-locking feedback controller. This embodiment is described in detail below, as shown in the attached diagram. Figure 5 As shown, the control signal output by the PID controller can be expressed as:

[0064]

[0065] Where K P K I K D These are the proportional coefficient, integral coefficient, and derivative coefficient, respectively; t is the time variable; and e(t) is the difference between the input error signal and the actual output. After processing by the PID control module, the control signal is output to the laser controller, which can then control the laser's output wavelength to be stabilized at the center of the absorption line in real time according to the entire PID feedback control loop.

[0066] The calculation of the gas flow field velocity is completed in a very short time. The sideband leveling operation is performed simultaneously with the calculation of the gas flow field velocity, and the sideband leveling operation is basically completed instantaneously.

[0067] In summary, the above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A high-bandwidth flow rate diagnostic method based on molecular spectroscopy, characterized in that, Includes the following steps: S1: Set the laser controller to generate a control signal to control the laser to output a single-wavelength laser signal corresponding to the center wavelength of the absorption line of the target gas. S2: A modulation signal is generated by a signal generator and input into an electro-optic modulator. The electro-optic modulator is used to externally modulate the single-wavelength laser signal at a modulation frequency, so that the optical signal generates symmetrical sideband wavelengths on both sides of the center wavelength, and the modulated optical signal is obtained. S3: By constructing an optical path, the modulated optical signal is incident into the gas flow field at a certain angle. The modulated optical signal interacts with the target gas and undergoes energy absorption. The laser transmission signal exiting the gas flow field is received by the photodetector, which converts the optical signal into a time-domain electrical signal. S4: Construct a high-speed frequency domain transformation model, which is used to transform the time-domain electrical signal into a frequency-domain electrical signal; S5: Establish a main band frequency shift database and a side band wavelength database; The sideband wavelengths are extracted from the frequency domain electrical signal and stored in the sideband wavelength database. Based on the intensity difference between the left and right sidebands of the main band in the sideband wavelength database, the error signal required for tracking and locking the laser center wavelength to the absorption peak of the gas to be measured is calculated. Based on the error signal, a frequency-locked feedback controller is used to control the driving current signal in real time and return to S1, so that the laser controller can stably track the laser center wavelength to the gas absorption peak under the action of the driving current signal. The main band center wavelength is extracted by the frequency domain electrical signal, and then the difference between it and the center wavelength of the absorption line of the target gas is calculated. The absolute value of the difference is used as the main band frequency shift and stored in the main band frequency shift database. The gas flow field velocity is continuously calculated based on the center wavelength frequency shift in the main band frequency shift database.

2. The high-bandwidth flow rate diagnostic method based on molecular spectroscopy as described in claim 1, characterized in that, In step S2, a modulation signal is generated by a signal generator, wherein the frequency of the modulation signal generated by the signal generator is above GHz.

3. The high-bandwidth flow rate diagnostic method based on molecular spectroscopy as described in claim 1, characterized in that, In S4, a high-speed frequency domain transformation model is built, with a transformation speed on the order of nanoseconds.

4. The high-bandwidth flow rate diagnostic method based on molecular spectroscopy as described in claim 1, characterized in that, In S4, a high-speed frequency domain transformation model is built, specifically: a convolutional neural network model CNN, which includes an input layer, an output layer, a combination layer of two convolutional layers and a pooling layer, and a fully connected layer. The output of each layer is given a ReLU activation function to introduce nonlinear factors to fit the nonlinear mapping relationship between the input and the output. The input received by the input layer is the time-domain electrical signal, and the number of features of the input layer is determined by the number of data points of the time-domain electrical signal; The output of the output layer is the frequency domain electrical signal, and the number of features of the output layer is determined according to the number of data points of the frequency domain electrical signal. The Convolutional Neural Network (CNN) model uses a combination of the ADAM algorithm and the backpropagation algorithm to update the weights of the neural network. By calculating the gradient direction of the loss function, the weights are adjusted along the gradient direction, so that the absolute value of the difference between the model's prediction result and the actual result is reduced. The MSE function was chosen as the loss function.

5. The high-bandwidth flow rate diagnostic method based on molecular spectroscopy as described in claim 1, characterized in that, The step involves calculating the error signal required for laser center wavelength tracking and locking to the absorption peak of the gas under test, based on the intensity difference between the left and right sidebands of the main band in the sideband wavelength database. Specifically: Using the peak difference between the left and right sidebands of the main band in the sideband wavelength database as the error signal, the error signal formula is expressed as: in J n (ω) represents an angular frequency of ω. n The order Bessel function, n=1 or -1; when the error signal value e When 0(ω) is 0, it means that the output wavelength of the laser is located at the center wavelength of the absorption line, indicating that the sideband has been leveled.

6. The high-bandwidth flow rate diagnostic method based on molecular spectroscopy as described in claim 1, characterized in that, The step involves calculating the error signal required for laser center wavelength tracking and locking to the absorption peak of the gas under test, based on the intensity difference between the left and right sidebands of the main band in the sideband wavelength database. Specifically: The peak ratio of the left and right sidebands in the main band of the sideband wavelength database is used as the error signal. The error signal formula is as follows: in J n (ω) represents an angular frequency of ω. n The order Bessel function, n=1 or -1; when the error signal value e When (ω) is 1, it means that the output wavelength of the laser is located at the center wavelength of the absorption line, indicating that the sideband has been leveled.

7. A high-bandwidth flow rate diagnostic method based on molecular spectroscopy as described in any one of claims 1 to 6, characterized in that, The frequency-locked feedback controller is a PID controller, an active disturbance rejection controller, a sliding mode controller, or an adaptive controller.

8. A high-bandwidth flow rate diagnostic method based on molecular spectroscopy as described in any one of claims 1 to 6, characterized in that, Based on the error signal, a frequency-locked feedback controller is used to control the drive current signal in real time, so that the laser controller stably tracks the laser center wavelength to the gas absorption peak under the action of the drive current signal. Specifically: The frequency locking feedback controller is a PID controller; The control signal output by the PID controller is: in K P , K I , K D These are the proportional coefficient, integral coefficient, and differential coefficient, respectively, where t is the time variable. This is the difference between the input error signal and the actual output. After processing by the PID controller, the control signal is output to the laser controller, which then controls the laser's output wavelength to remain stable at the center of the absorption line in real time according to the entire PID feedback control loop.

9. A high-bandwidth flow rate diagnostic method based on molecular spectroscopy as described in any one of claims 1 to 6, characterized in that, Based on the center wavelength frequency shift in the main band frequency shift database, the gas flow field velocity is continuously calculated, specifically as follows: Using the main band frequency shift amount in the main band frequency shift database The final gas flow field velocity is calculated using the following formula: in u Let be the gas flow rate, which is the value to be determined. θ The angle at which the modulated optical signal enters the gas flow field. ν 0 represents the center frequency of the absorption line of the target gas.

Images