Dual optical comb multi-wavelength interference dynamic ranging signal processing method

Abstract

The present application belongs to the field of laser interferometric distance measurement, and provides a kind of double optical comb multi-wavelength interference dynamic ranging signal processing method, and the present application constructs the dual control mechanism of initial frequency shift capture and dynamic frequency shift prediction.The method first uses spectrum envelope extraction technology to carry out global scanning on double optical comb multi-wavelength interference signal, quickly estimates the initial Doppler frequency shift of target and presets the frequency of orthogonal reference signal for phase demodulation of phase-locked amplification;Then switch to dynamic frequency shift prediction mode.Using phase-locked amplification phase demodulation to extract real-time phase, and using adaptive frequency shift prediction window to process real-time phase for dynamic frequency shift prediction, according to the speed of target adaptive adjustment frequency shift prediction window length, and the frequency shift prediction value is fed back to phase-locked amplification phase demodulation, dynamic correction for the frequency of orthogonal reference signal used for mixing.The present application effectively solves the problem of invalidation of traditional phase-locked amplification phase demodulation method and limited start under high dynamic condition, and realizes high-precision double optical comb heterodyne interference dynamic ranging.

Description

Technical Field

[0001] This invention belongs to the field of laser interferometric ranging, specifically relating to a dual-comb multi-wavelength interferometric dynamic ranging signal processing method. Background Technology

[0002] Laser interferometric ranging is a large-scale, high-precision, non-contact absolute distance measurement method. The dual-comb heterodyne interferometric ranging method, based on the principle of multi-wavelength interferometric ranging, utilizes the characteristics of an optical frequency comb to synchronously generate a composite wavelength from multiple single wavelengths, possessing great potential for large-scale, high-precision, and rapid dynamic absolute measurement. However, existing dual-comb heterodyne interferometric ranging methods face a fundamental challenge in dynamic ranging: when using the traditional lock-in amplification (LIA) method to calculate the phase, an orthogonal reference signal with the same frequency as the measured signal is required. When measuring a moving target, the Doppler frequency shift caused by the relative velocity will mix into the measurement signal. To ensure the signal-to-noise ratio, the system is usually equipped with a narrowband low-pass filter. Once the Doppler frequency shift causes the signal frequency to deviate, the effective signal after mixing will not be able to pass through the system's low-pass filter, leading to phase extraction failure and thus making dynamic ranging impossible.

[0003] To address the phase extraction failure problem of traditional phase-locked amplification methods in dynamic ranging, two main solutions have been proposed both domestically and internationally: The first category is the iterative tracking method based on digital phase-locked loops (PLLs). This method essentially employs a frequency counting and error feedback mechanism. However, it has significant limitations in high-dynamic measurement capabilities at large accelerations: on the one hand, the frequency counter relies on statistical gate time, limiting the loop's dynamic response bandwidth. Some existing technologies can only achieve a maximum frequency change rate of 6 MHz / s under phase-locked conditions (corresponding to a measured acceleration of approximately 4.65 m / s² at a wavelength of 1550 nm). On the other hand, its core PI control is a second-order feedback model. According to control theory, its dynamic steady-state error (phase lag) is positively correlated with the magnitude of the target acceleration. This means that as the target acceleration increases, the phase measurement error will increase significantly; once the acceleration exceeds the aforementioned threshold, the excessive phase error will directly lead to loop lockout, thus causing it to fail.

[0004] The second category is based on the digital frequency comb frequency domain slicing method. This method generates a series of dense digital local oscillator signals (i.e., "digital combs") inside the FPGA, using this dense "digital ruler" to cover the frequency shift range that may occur in the input signal, and solves the measurement dead zone problem by switching between odd and even combs. However, although this method can measure large accelerations, it is only applicable to single-carrier signals. In dual-comb dynamic ranging, the input signal itself contains multiple discrete frequency components, and the output will simultaneously contain multiple low-frequency signal aliasing. The system backend cannot establish a mapping relationship between the digital comb teeth and specific wavelengths, and therefore cannot separate the independent phase of each wavelength, making it unsuitable for dual-comb dynamic ranging scenarios.

[0005] In summary, the current field of dynamic dual-comb heterodyne interferometric ranging technology lacks a signal processing method that can specifically target the characteristics of optical frequency comb signals, solve the failure problem of traditional lock-in amplification methods, and balance measurement speed and accuracy. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a dual-comb multi-wavelength interferometric dynamic ranging signal processing method to solve the problems in the prior art. The technical solution adopted by this invention is as follows: A dual-comb multi-wavelength interferometric dynamic ranging signal processing method includes: Step 1: Construct a dual-comb heterodyne interferometric ranging system model, acquire the measurement path signal and the reference path signal, calculate the synthesized wavelength, and use the phase of the synthesized wavelength to calculate the distance to be measured; Step 2: Extract the spectral envelope of the measurement path and reference path signals acquired in Step 1 to complete the initial frequency shift capture, and preset the frequency of the orthogonal reference signal in the phase-locked loop amplification phase solution; Step 3: Directly perform phase-locked amplification and phase calculation on the acquired signal, and use an adaptive frequency shift prediction window to set the frequency tracking of the orthogonal reference signal; Step 4: Adaptively adjust the window length of the adaptive frequency shift prediction window to determine the range of values ​​for the window length; Step 5: During measurement, the measurement system cycles through steps 1-4. When the measurement system is reset or restarted after a power failure, it triggers initial frequency shift acquisition and maintains dynamic frequency shift prediction in a stable state.

[0007] Furthermore, in step 1, the synthesized wavelength is calculated. The formula is: ; in, and Different wavelengths of comb teeth for optical combs; utilizing synthesized wavelengths phase Calculate the distance to be measured L= 。

[0008] Furthermore, in step 2, the initial frequency shift acquisition includes: extracting the spectral envelope of the input signal through initial FFT calculation to identify the initial position of the Doppler frequency shift; and calculating the initial frequency shift amount. The signal is converted into a frequency control value and written into the phase-locked loop amplifier for quadrature reference signal frequency setting, so that the effective signal falls within the passband range of the low-pass filter, thus completing the initial signal locking.

[0009] Furthermore, step 3 includes: Step 3.1: Establish a mathematical model for the frequency shift value and establish the synthetic wavelength. Corresponding phase change rate Dynamic mapping relationship; measuring three continuous phase data as: current point , the previous point And the first two points Set the phase update rate to ; The adaptive window length is Then based on the phase difference between the beginning and end of the window The fine frequency shift prediction model is expressed as: ; in, The wavelength of a single comb tooth; This provides real-time, fine-grained frequency shift predictions. Step 3.2: Set the initial length of the frequency shift prediction window, setting it to the maximum window length of the frequency shift prediction window. The range of values ​​for the maximum window length satisfies the following inequality: ; in, The maximum acceleration of the target to be measured; Step 3.3: Directly perform phase-locked amplification and phase calculation on the acquired signal, and use the inequality in Step 3.2, based on the fine frequency shift prediction model in Step 3.1, to track and set the frequency of the orthogonal reference signal in real time, so as to realize high-precision phase-locked amplification and phase calculation on the acquired signal.

[0010] Furthermore, step 4 includes: setting the standard deviation of the phase measurement results as... The upper cutoff frequency of the system filter is f t Window length The range of values ​​for satisfies the following inequality: .

[0011] Furthermore, step 5 includes: during measurement, the measurement system cycles through steps 1-4, triggering initial frequency shift acquisition when the measurement system is reset or restarted after a power failure, and maintaining dynamic frequency shift prediction in a stable state.

[0012] The present invention has the following beneficial effects: (1) This invention avoids the low spectral resolution defect of the frequency counting mechanism under high dynamic measurement and transforms the dynamic lag of PI feedback that deteriorates with acceleration into a precise and controllable settling time of the adaptive frequency shift prediction window. Compared with the existing digital phase-locked loop (DPLL) method that relies on zero-crossing frequency counting and PI control process with low information utilization, this invention uses the frequency shift prediction window mechanism to replace the asymptotic convergence process of PI control, obtains high-precision phase information of the entire waveform, and directly calculates the frequency shift based on the phase change, effectively avoiding the influence of loop bandwidth limitation and dynamic tracking error, at an acceleration of 24 m / s². 2 Even under these conditions, it can maintain sub-nanometer level measurement accuracy. This ensures the system has the capability for high-precision measurement of highly dynamic, high-acceleration objects.

[0013] (2) This invention avoids the output signal spectral aliasing problem of digital frequency comb spectrum segmentation methods in dual-comb dynamic ranging applications and greatly reduces hardware resource consumption. Addressing the many-to-many aliasing and mapping failure problems of existing digital frequency comb spectrum segmentation methods when processing discrete physical wavelength components, this invention employs a dynamic frequency shift compensation method based on spectral envelope extraction and adaptive frequency shift prediction control. This invention does not rely on dense digital frequency combs to passively match the input signal, but instead actively follows the dynamic frequency shift of the measurement path signal for overall control by real-time correction of the orthogonal reference signal frequency. This mechanism ensures that each tooth of the physical optical comb maintains a relatively fixed spectral structure during frequency shift, guaranteeing that the system backend can accurately separate and calculate the independent phase of each wavelength. Furthermore, the frequency shift prediction window of this method only requires logic resources of the maximum initial window length, greatly reducing FPGA logic resource consumption and making it easy to implement on lightweight hardware embedded platforms. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the process of the present invention; Figure 2 A schematic diagram of the phase calculation results obtained by applying this invention to a target that starts from a stationary state and undergoes uniform acceleration with variable acceleration. Figure 3 The figure shows the results of distance measurement accuracy verification and error analysis of this invention in dynamic measurement. Detailed Implementation

[0015] The following will be described in conjunction with embodiments of the present invention. Figures 1-3The technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Unless otherwise specified, the technical means used in the embodiments are conventional means well known to those skilled in the art.

[0016] This invention addresses the problem of phase extraction failure in traditional phase-locked loop (PLL) methods caused by Doppler frequency shift due to dynamic target motion by constructing a dual control mechanism of "initial frequency shift acquisition - dynamic frequency shift prediction". This method first uses spectral envelope extraction technology to perform a global scan of the signal, quickly estimating the initial Doppler frequency shift of the target and presetting the frequency of the orthogonal reference signal; then it switches to dynamic frequency shift prediction mode. The real-time phase is extracted from the orthogonal I / Q signal output by the PLL amplification phase calculation method using the CORDIC phase calculation method, and dynamic frequency shift prediction is performed using an adaptive frequency shift prediction window implemented with a FIFO queue. The predicted value is fed back to the PLL amplification phase calculation to dynamically correct the frequency of the orthogonal reference signal. This invention effectively solves the failure and start-up limitations of traditional PLL methods under high dynamic conditions, achieving high-precision dual-comb heterodyne interferometry dynamic ranging.

[0017] like Figure 1 This invention proposes a dual-comb multi-wavelength interferometric dynamic ranging signal processing method, comprising the following steps: Step 1: Construct a model of the dual-comb heterodyne interferometric ranging system, acquire signals from the measurement path and the reference path, and calculate the synthesized wavelength. Utilizing the synthesized wavelength phase Calculate the distance to be measured L= ; Step 2: Extract the spectral envelope of the measurement path and reference path signals acquired in Step 1 during the startup phase to complete the initial frequency shift capture, and preset the frequency of the orthogonal reference signal in the phase-locked loop amplification phase solution.

[0018] Step 3: After startup, the acquired signal is directly subjected to phase-locked amplification and phase solution, and the orthogonal reference signal frequency tracking setting is performed using an adaptive frequency shift prediction window.

[0019] Step 4: Adaptively adjust the window length of the adaptive frequency shift prediction window.

[0020] Step 5: During measurement, the measurement system cycles through steps 1-4. When the measurement system is reset or restarted after a power failure, it triggers initial frequency shift acquisition and maintains dynamic frequency shift prediction in a stable state.

[0021] Specifically, the details of each step are as follows: Step 1: Construct a dual-comb heterodyne interferometric ranging system model. Dual-comb interferometric ranging technology is based on classical multi-wavelength interferometric ranging technology. It replaces the multiple laser sources in multi-wavelength interferometric ranging with two optical frequency combs whose center frequencies are fixed and whose tooth frequencies are slightly different. These combs are formed by frequency shifting of the same laser source via an acousto-optic frequency shifter and modulation by an electro-optic modulator. A suitable composite wavelength (e.g., ...) is selected from the interference signal. and ) Calculate the resulting composite wavelength : ; in, and The wavelengths of the different comb teeth.

[0022] Using the phase corresponding to this synthesized wavelength Calculate the distance to be measured L= This mechanism extends the range of unambiguous measurements to the synthetic wavelength scale. Under this mechanism, the Doppler frequency shift caused by the target motion... This can cause a shift in the frequency of the measured signal. When using traditional phase-locked array (LIA) calculations that rely on narrowband filtering to ensure the signal-to-noise ratio, the Doppler frequency shift introduced by high-speed motion will cause the frequency of the mixed signal to deviate from the baseband. Once this deviation exceeds the passband range of the low-pass filter, the effective signal will be filtered out, directly leading to phase extraction failure and a blind zone in dynamic ranging.

[0023] Step 2: Extract the spectral envelope of the input signal through initial FFT calculation to identify the initial position of the Doppler frequency shift; calculate the initial frequency shift amount. The signal is converted into a frequency control word and written into the phase-locked loop amplifier for phase calculation to set the frequency of the quadrature reference signal, so that the effective signal falls within the passband range of the system's low-pass filter, thus completing the initial signal locking.

[0024] Step 3: Use a signal generator to simulate the comb signal of the interference. Set the system front-end sampling rate to 125MHz, the center frequency of the interference signal to 6MHz, the interference signal spectrum to contain 0, ±1, ±2, ±3, and ±4 order comb teeth, the comb tooth spacing to 1.1MHz, and the phase update rate to... =50kHz.

[0025] Step 3.1: Establish a mathematical model for the frequency shift value. Establish the synthesized wavelength. Corresponding phase change rate The dynamic mapping relationship is described. To implement this model in a digital system, a frequency shift prediction window is introduced for discretization. Let the three continuous phase data measured by this method be: the current point... , the previous point And the first two points Set the phase update rate to 0.5. The adaptive window length is Then based on the phase difference between the beginning and end of the window The real-time fine frequency shift prediction model can be expressed as follows: ; in, The wavelength of a single comb tooth; This provides real-time, fine-grained frequency shift predictions. Step 3.2: Set the maximum window length for hardware initialization .

[0026] The maximum window length for frequency shift prediction is determined by two main factors. First, a window that is too short will prevent the length from being reduced according to the model given in step 3.1, causing the adaptive strategy to fail. Second, a window that is too long will lead to resource consumption issues; a large window will consume excessive logical resources and computational complexity. Therefore, given the target maximum acceleration... and the standard deviation of the assumed system phase measurement results Given a value of 0.01°, what is the upper cutoff frequency of the system filter? f t Maximum window length The range of values ​​for should satisfy the following inequality: ; Where a is the maximum acceleration of the target to be measured.

[0027] Based on the inequality constraints derived above, determine the maximum window length. =128.

[0028] Step 3.3: The acquired signal is directly subjected to phase-locked amplification (PLA) phase calculation. The quadrature I / Q data stream containing phase information output from the PLA phase calculation enters the CORDIC phase calculation pipeline to calculate the actual phase value. The phase value output by CORDIC is pushed into an adaptive frequency shift prediction window, which is implemented by a FIFO queue with an initial maximum depth of 128, completing the sliding storage of data. Real-time reading of data from three nodes in the FIFO queue: the current point... , the previous point And the first two points The second-order difference value of the phase is calculated to characterize the instantaneous acceleration of the target. An adaptive frequency shift prediction mechanism (i.e., the inequality in step 3.2) is introduced using the prediction window designed above. Based on the fine frequency shift prediction model given in step 3.1, the frequency of the orthogonal reference signal is tracked and set in real time, thereby realizing high-precision phase-locked amplification and phase solution of the acquired signal.

[0029] Step 3.1 employs a linear difference method based on the phases at both ends of the window. Alternatively, a least-squares fitting method based on all data points within the window can be used as an alternative. This approach is suitable for scenarios with higher noise suppression requirements and ample computational resources. Specifically: The adaptive frequency shift prediction model established in step 3.1 and the initial maximum window length determined in step 3.2 Based on this, read all the data stored in the FIFO queue. Point phase data Using the least squares method to analyze this... Linear regression was performed on the data points to calculate the rate of change of phase over time. The optimal estimate minimizes the sum of squared residuals between the fitted result and all phase measurement points within the window. Let i be the sampling point number, and the rate of change can be expressed as: ; After obtaining the fitted rate of change k, it replaces the value in the main scheme. Substituting into the frequency shift prediction model, we get: ; This alternative approach utilizes all data within the window for fitting, which, compared to the difference method using only the first and last two points, further reduces the impact of sudden noise at individual sampling points on the prediction results, resulting in smoother and more stable frequency shift predictions and ensuring the robustness of the system in low signal-to-noise ratio environments.

[0030] Step 4: Establish adaptive window length The mathematical model. The window length of the frequency shift prediction window is mainly affected by two factors. On the one hand, it is limited by the dynamic response capability, if... If the value is too large, it will exacerbate the hysteresis, causing the predicted value to fail to keep up with the actual frequency shift, resulting in phase extraction failure; on the other hand, it is limited by noise suppression capabilities, if If the value is too small, it cannot effectively average out the random phase noise within the window, causing the absolute value of the noise derivative to exceed the system tolerance, thus leading to phase extraction failure. The standard deviation of the system phase measurement error measured under given static conditions. Given that the angle is 0.01°, the upper cutoff frequency of the system filter is: f t =25kHz.

[0031] Assume the standard deviation of the phase measurement results is The upper cutoff frequency of the system filter is f t Window length The range of values ​​for should satisfy the following inequality: ; This model provides the theoretical foundation for the implementation of adaptive frequency shift prediction windows. Based on this step, the... Constraints are used to dynamically determine the optimal window length at the current moment. Historical data at the corresponding depth address is indexed from the FIFO. This enables adaptive adjustment.

[0032] Step 5: For actual measurements, ±3rd order comb frequency signals are used. Data is continuously processed through the acquired measurement path signal and reference path signal, and phase-locked loop (PLL) amplification phase calculation is performed. The system cyclically executes steps 1 to 4, processing the data at each moment: frequency shift capture is triggered when the system is reset or restarted after a power failure, and dynamic frequency shift prediction and tracking are maintained in a stable state. The phase result output by the PLL amplification phase calculation is finally stored in the buffer pool and transmitted to the host computer.

[0033] Starting from a stationary state with variable acceleration (from 12 m / s²) 2 up to 24 m / s 2 Distance measurement is performed on a test object undergoing uniform acceleration. After processing using this method, the measurement results are as follows: Figure 2 As shown, the phase resolution curve processed using this method is smooth and continuous, effectively solving the problems of method failure and limited startup in traditional phase-locked loop methods during dynamic measurements. To evaluate the phase measurement accuracy of this method, residual calculations were performed on the phase results measured using this method and the theoretical values. The results are shown below. Figure 3 As shown, the ranging accuracy of dynamic ranging can reach 0.1nm, meeting the requirements of high-precision ranging.

[0034] The above embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Any modifications, alterations, alterations, or substitutions made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for processing dynamic ranging signals using dual-comb multi-wavelength interferometry, characterized in that, include: Step 1: Construct a dual-comb heterodyne interferometric ranging system model, acquire the measurement path signal and the reference path signal, calculate the synthesized wavelength, and use the phase of the synthesized wavelength to calculate the distance to be measured; Step 2: Extract the spectral envelope of the measurement path and reference path signals acquired in Step 1 to complete the initial frequency shift capture, and preset the frequency of the orthogonal reference signal in the phase-locked loop amplification phase solution; Step 3: Directly perform phase-locked amplification and phase calculation on the acquired signal, and use an adaptive frequency shift prediction window to set the frequency tracking of the orthogonal reference signal; Step 4: Adaptively adjust the window length of the adaptive frequency shift prediction window to determine the range of values ​​for the window length; Step 5: During measurement, the measurement system cycles through steps 1-4. When the measurement system is reset or restarted after a power failure, it triggers initial frequency shift acquisition and maintains dynamic frequency shift prediction in a stable state. In step 1, the synthesized wavelength is calculated. The formula is: in, and Different wavelengths of comb teeth for optical combs; utilizing synthesized wavelengths phase Calculate the distance to be measured L= ; Step 3 includes: Step 3.1: Establish a mathematical model for the frequency shift value and establish the synthetic wavelength. Corresponding phase change rate Dynamic mapping relationship; measuring three continuous phase data as: current point , the previous point And the first two points Set the phase update rate to ; The adaptive window length is Then based on the phase difference between the beginning and end of the window The fine frequency shift prediction model is expressed as: in, The wavelength of a single comb tooth; This provides real-time, fine-grained frequency shift predictions. Historical data; Step 3.2: Set the initial length of the frequency shift prediction window, setting it to the maximum window length of the frequency shift prediction window. The range of values ​​for the maximum window length satisfies the following inequality: in, The maximum acceleration of the target to be measured; f t This is the upper cutoff frequency of the system filter; Step 3.3: Directly perform phase-locked amplification and phase calculation on the acquired signal, and use the inequality in Step 3.2, based on the fine frequency shift prediction model in Step 3.1, to track and set the frequency of the orthogonal reference signal in real time, so as to realize high-precision phase-locked amplification and phase calculation on the acquired signal.

2. The dual-comb multi-wavelength interferometric dynamic ranging signal processing method according to claim 1, characterized in that, In step 2, the initial frequency shift acquisition includes: extracting the spectral envelope of the input signal through initial FFT calculation to identify the initial position of the Doppler frequency shift; and calculating the initial frequency shift amount. The signal is converted into a frequency control value and written into the phase-locked loop amplifier for quadrature reference signal frequency setting, so that the effective signal falls within the passband range of the low-pass filter, thus completing the initial signal locking.

3. The dual-comb multi-wavelength interferometric dynamic ranging signal processing method according to claim 1, characterized in that, Step 4 includes: setting the standard deviation of the phase measurement results as... The upper cutoff frequency of the system filter is f t Window length The range of values ​​for satisfies the following inequality: 。 4. The dual-comb multi-wavelength interferometric dynamic ranging signal processing method according to claim 1, characterized in that, Step 5 includes: During measurement, the measurement system cycles through steps 1-4. When the measurement system is reset or restarted after a power failure, initial frequency shift acquisition is triggered, and dynamic frequency shift prediction is maintained in a stable state.

Images