Wide-spectral-range double-frequency comb dynamic spectroscopy measurement method and wide-spectral-range double-frequency comb dynamic spectroscopy measurement system

Abstract

The invention discloses a wide-spectral-range double-frequency comb dynamic spectroscopy measurement method and system. Performing double-frequency comb spectroscopy measurement to obtain an interference signal frequency spectrum which contains actual spectrum information in each period and is folded to be within a frequency range of 0-0.5 * fs, extracting frequency spectrum characteristics and corresponding frequency values from the interference signal frequency spectrum, and recording the frequency spectrum characteristics and the corresponding frequency values as frequency measurement values fm; and when fm is close to 0 or 0.5 * fs, a signal change trend prediction algorithm is used for judging whether folding occurs, and when folding occurs, the frequency position of the actual signal spectral information is recovered according to a frequency correction value, so that the dynamic double-frequency comb measurement method and system for the wide-spectral-range spectral information are realized. According to the invention, dynamic measurement of spectrum information in a wide spectrum range can be realized.

Description

Technical Field

[0001] The present invention relates to the field of spectroscopy measurement technology, and in particular to a wide spectrum range dual-frequency comb dynamic spectroscopy measurement method and system. Background Art

[0002] Dynamic spectroscopy is a widely used method in scientific research and production, finding increasing application in a wide range of fields, including economic development and industrial production. The measurement range of dynamic spectroscopy determines the measurement range and application effectiveness of related application systems. Dynamic, wide-range spectroscopy measurements have important and extensive practical value.

[0003] There are many ways to interrogate the system to be measured in a dynamic spectral measurement system, mainly including wavelength division multiplexing (WDM), optical frequency domain reflectometry (OFDR) and optical time domain reflectometry (OTDR). Traditional WDM has a very limited number of multiplexed units to be measured because different units to be measured occupy different wavelength ranges. The OFDR method uses a swept laser as the light source, and the laser linewidth degrades under rapid tuning, resulting in limited spectral resolution under high-speed data acquisition. The OTDR method has a wide optical pulse width (ns level), so the spatial resolution of the system to be measured is only dm level. The dual-frequency comb technology based on femtosecond pulse technology has a fixed and accurate frequency interval and a narrow pulse at the femtosecond level, which can achieve high spectral resolution, high refresh rate and high spatial resolution measurement of the system to be measured.

[0004] There are many factors that limit the dynamic spectrum measurement range, including the light source spectral range, system noise, dynamic spectrum aliasing, etc. For systems using dual-frequency comb pulse light sources, the main reason affecting the dynamic spectrum measurement range is the spectral aliasing problem. When the dynamic range of the interference signal exceeds the spectral aliasing range f r *f s / (2*|f r -f s |), the excess interference signal spectrum will be folded to the range of 0 to 0.5*f s The spectral aliasing problem causes the dynamic spectral measurement range of the dual-frequency comb pulse source system to be limited by the repetition frequency of the light source.

[0005] Expanding the dynamic spectroscopy measurement range of the system under test can be achieved by addressing both system noise and spectral folding. Given the inherent limitations of dynamic spectroscopy measurement range, reducing noise can only provide a relative improvement. To truly address dynamic spectroscopy measurement issues, spectral aliasing must be addressed. Currently, there is no clear approach or method to expand the dynamic spectroscopy measurement range from the perspective of spectral aliasing.

[0006] The present invention proposes to obtain the actual spectral information in each period folded to the range of 0 to 0.5*f by using the dual frequency comb technology. s The interference signal spectrum within the frequency range is extracted, and the spectrum characteristics and their corresponding frequency values ​​are extracted from the interference signal spectrum and recorded as frequency measurement values. The frequency correction value is given according to the calculation method of the correction offset and the frequency correction value. The frequency change of the actual spectral information is obtained according to the frequency correction value, and the frequency position of the actual signal spectrum spectral information is restored to realize the dynamic measurement of spectral information in a wide spectrum range. Summary of the Invention

[0007] The purpose of the present invention is to provide a wide spectrum range dual-frequency comb dynamic spectroscopy measurement method and system.

[0008] According to one aspect of the present invention, a method for measuring dual-frequency comb dynamic spectroscopy over a wide frequency range is provided, comprising:

[0009] Step 1: Use a repetition frequency f r The wide spectrum excitation pulse is input into the system to be tested, generating a signal pulse with continuously changing spectral information within the spectrum range of the excitation pulse; using a pulse with a repetition frequency f s The wide spectrum sampling pulse samples the signal pulse to generate a periodic interference signal; the spectrum width of the wide spectrum excitation pulse and the wide spectrum sampling pulse are both greater than the spectrum aliasing range f r *f s / (2*|f r -f s |);

[0010] Step 2: Perform time-frequency transformation on the interference signal of the first period containing the actual spectral information in the period to obtain the signal folded to the range of 0 to 0.5*f s Interference signal spectrum within the frequency range; extract spectrum features from the interference signal spectrum, and record the frequency value corresponding to the spectrum feature as the frequency measurement value f m (1), f m (1) The value is between 0 and 0.5*f s Within the frequency range; let the correction offset cor be 0. If the frequency value of the spectrum characteristic of the actual spectral information is greater than 0, the frequency correction value f c The calculation method is f c (n) = cor + f m (n), n is the number of cycles; if the frequency value of the spectrum characteristic of the actual spectral information is less than 0, the frequency correction value f c The calculation method is f c (n) = cor-f m (n); according to the frequency correction value f c Calculation method, calculate f c (1);

[0011] Step 3: Starting from the second period, perform time-frequency transformation on the interference signal of the nth period containing the actual spectral information in the period, and obtain the signal folded to the range of 0 to 0.5*f s Interference signal spectrum within the frequency range; extract spectrum features from the interference signal spectrum, and record the frequency value corresponding to the spectrum feature as the frequency measurement value f m (n), f m The value of (n) is between 0 and 0.5*f s Within the frequency range;

[0012] When f m (n) is close to 0 or 0.5*f s When the signal change trend prediction algorithm is used, the signal change trend prediction algorithm is based on the frequency correction value (f c (n-1),f c (n-2)…) gives the frequency prediction value f of the current cycle p (n), according to the previous f c 、f p (n) and f m (n) to determine whether folding occurs:

[0013] 1) If folding occurs, then a) when f m When (n) is close to 0, if f p (n) is greater than f c (n-

[0014] 1), then cor remains unchanged, and the frequency correction value f c (n) is calculated as f c (n) is equal to cor + f m (n); if f p (n) is less than f c (n-1), then cor remains unchanged, and the frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor-f m (n); b) when f m (n) is close to 0.5*f s When f p (n) is greater than f c (n-1), then cor increases by f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor-f m (n); if f p (n) is less than f c(n-1), then cor decreases f s , frequency correction value f c (n) is calculated as follows: the frequency correction value fc(n) is equal to cor+f m (n);

[0015] 2) If it is determined that no folding occurs, the cor value remains unchanged, and the frequency correction value calculation method remains unchanged;

[0016] Calculate f c (n);

[0017] When f m (n) is not close to 0 or 0.5*f s When , the cor value remains unchanged, and the calculation method of the frequency correction value remains unchanged;

[0018] Step 4: Calculate f based on the value of cor and the frequency correction value calculation method. c (n);

[0019] Obtain the frequency variation of the actual spectral information according to the frequency correction value fc(n) and recover the frequency position of the actual signal spectral information;

[0020] Step 5. Repeat steps 3 to 4 until the measurement is completed.

[0021] In one example, in steps 2 and 3, the spectral characteristics of the interference signal spectrum are the intensity peak of the interference signal spectrum, the center of mass of the intensity spectrum, the specific phase value position of the phase spectrum, or the similarity or correlation between multiple points within a certain frequency range of the interference amplitude or phase spectrum or all points of the interference amplitude spectrum relative to different times.

[0022] In one example, in step 3, the signal change trend prediction algorithm includes but is not limited to one or more methods such as multi-point interpolation, spline interpolation, linear interpolation, Hermite interpolation, Kalman filtering, synovial observation, Lemberg observation, moving horizon prediction, implicit dynamic feedback, filter deviation update, least squares curve fitting, and autoregressive prediction.

[0023] In one example, in step 3, the above-mentioned c 、f p (n) and f m The method to determine whether folding occurs is to compare the cor value of the n-1 cycle with f m (n) The value calculated according to the frequency correction value calculation method of the n-1th cycle is the same as f p (n) values ​​are compared. If the two values ​​are located at the folding boundary (f s Integer multiples of f sIf the two sides are on different sides of (integer + 0.5) times), it is judged that folding has occurred; otherwise, it is judged that folding has not occurred.

[0024] In one example, in step 3, the f m (n) is close to 0 or 0.5*f s The judgment method is: f m (n) and 0 or 0.5*f s The distance is less than the frequency measurement value f m 3 times the mean square error.

[0025] The present invention also provides a wide spectrum range dual-frequency comb dynamic spectroscopy measurement system, comprising:

[0026] It includes a wide spectrum dual-frequency comb source, a system to be measured, a detection device and a data acquisition and processing unit; wherein the dual-frequency comb source generates a repetition frequency f r and f s The wide spectrum excitation pulse and sampling pulse are input to the system to be measured, and the system to be measured outputs the signal pulse generated by the excitation pulse; the signal pulse and the sampling pulse generate a periodic interference signal through the detection device; the data acquisition and processing unit collects and transforms the interference signal into a time-frequency signal, and obtains the actual spectral information in each period, which is folded to the position between 0 and 0.5*f s The interference signal spectrum within the frequency range is extracted, and the spectrum characteristics and their corresponding frequency values ​​are extracted from the interference signal spectrum and recorded as the frequency measurement value f m ; Assume the correction offset cor, its initial value is 0, according to cor and f m Calculate the frequency correction value f c When f m Close to 0 or 0.5*f s When the signal change trend prediction algorithm is used based on the frequency correction value of no more than K cycles before, the frequency prediction value f of the current cycle is given. p , according to the previous f c 、The current f m With f p The relationship between the two determines whether folding occurs. When folding occurs, the value of cor and the calculation method of the frequency correction value are updated to give the current period f c ; According to the frequency correction value f c The frequency variation of the actual spectral information is obtained and the frequency position of the actual signal spectral information is restored, thereby realizing a dynamic measurement system for spectral information in a wide spectrum range.

[0027] The wide-spectrum dual-frequency comb source can be composed of one or two lasers and a resonant cavity. The wavelengths of the excitation pulse, sampling pulse, and signal pulse can be in the ultraviolet, visible, infrared, X-ray, terahertz, and electromagnetic wave frequency bands.

[0028] The system under test consists of a series of units under test that are not completely distributed in the same space and can transmit, reflect and scatter the excitation pulse.

[0029] The detection device can be a spectrum analyzer or a photodetector. When the signal acquisition unit is a spectrum analyzer, the interference signal spectrum of different sampling periods can be directly obtained. When the signal acquisition unit is a photodetector, the time domain electrical signal is obtained, and it is necessary to perform time-frequency transformation on it to obtain the interference signal spectrum of each sampling period.

[0030] The data acquisition and processing unit collects and transforms the interference signal into time-frequency signal, and obtains the signal containing the actual spectrum information in each period, which is folded to the position between 0 and 0.5*f s The interference signal spectrum within the frequency range is extracted, and the spectrum characteristics and their corresponding frequency values ​​are extracted from the interference signal spectrum and recorded as frequency measurement values. The frequency correction value is given according to the calculation method of the correction offset and the frequency correction value. The frequency change of the actual spectrum information is obtained according to the frequency correction value, and the frequency position of the actual signal spectrum information is restored to realize a dynamic measurement system for spectral information in a wide spectrum range.

[0031] The wide-spectrum dual-frequency comb source can be composed of one or two lasers and a resonant cavity. The wavelengths of the excitation pulse, sampling pulse and signal pulse can be in the ultraviolet, visible, infrared, X-ray, terahertz and electromagnetic wave frequency bands.

[0032] The spectrum width of the wide spectrum excitation pulse and the wide spectrum sampling pulse is greater than the spectrum aliasing range f r *f s / (2*|f r -f s |).

[0033] The spectrum of the signal pulse changes within the spectrum of the wide spectrum excitation pulse. The width of the spectrum of the interference signal after sampling the signal pulse is less than 0.5*f s . BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The present invention will be further described in detail below with reference to the accompanying drawings, in which:

[0035] Figure 1 This is a flow chart of a dual-frequency comb dynamic spectroscopy measurement method over a wide spectrum range;

[0036] Figure 2 This is a system diagram of a wide spectrum range dual-frequency comb dynamic spectroscopy measurement system;

[0037] Figure 3 This is the system structure diagram of Example 1;

[0038] Figure 4The frequency measurement value curve, correction offset curve and frequency correction value curve in Example 1 are shown in FIG.

[0039] Figure 5 It is the spectrum information directly recovered according to the frequency measurement value in Example 1 and the actual signal spectrum information recovered according to the frequency correction value. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present invention are described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0041] Example 1

[0042] The system structure diagram of this example is as follows Figure 3 As shown. The dual-frequency comb source used in this example is a dual-wavelength pulse light source, the system to be measured is a fiber Bragg grating, the detection device is a balanced photodetector, and the data acquisition and processing unit is an analog-to-digital conversion acquisition card, a computer, and data processing software; wherein the dual-wavelength pulse light source generates a repetition frequency f r and f s The wide-spectrum excitation light pulse and the sampling light pulse are inputted into the fiber Bragg grating through the optical circulator, and the fiber Bragg grating outputs the signal light pulse generated by the excitation light pulse; the signal light pulse is outputted by the circulator and coupled with the sampling light pulse through the optical coupler, and after coupling, the sampling light pulse samples the signal light pulse through the balanced photodetector to generate a periodic interference signal; the spectral width of the wide-spectrum excitation light pulse and the wide-spectrum sampling light pulse and the width of the spectral overlap of the wide-spectrum excitation light pulse and the wide-spectrum sampling light pulse are both greater than 20nm, which is greater than the spectrum aliasing range f r *f s / (2*|f r -f s |)=4.11nm; the data acquisition and processing unit collects and transforms the interference signal into a time-frequency signal, and obtains the actual spectral information in each period, which is folded to the position between 0 and 0.5*f s The interference signal spectrum within the frequency range is extracted from the interference signal spectrum, and the intensity spectrum centroid characteristics and their corresponding frequency values ​​are recorded as frequency measurement values ​​f m ; Assume the correction offset cor, its initial value is 0, according to cor and f m Calculate the frequency correction value f c When f m Close to 0 or 0.5*f s When the Kalman filter signal change trend prediction algorithm is used based on the frequency correction value of no more than K = 100 cycles before, the frequency prediction value f of the current cycle is given. p , according to the previous f c 、The current fm With f p The relationship between the two determines whether folding occurs. When folding occurs, the value of cor and the calculation method of the frequency correction value are updated to give the current period f c ; According to the frequency correction value f c The frequency variation of the actual spectral information is obtained and the frequency position of the actual signal spectral information is restored, thereby realizing a dynamic measurement system for fiber Bragg grating spectral information in a wide spectrum range greater than the spectrum aliasing range of 4.11nm.

[0043] In this example, the dual-frequency comb source used is a dual-wavelength pulsed laser with a single resonant cavity. The laser consists of a pump laser tube, a wavelength division multiplexer, an optical isolator, an erbium-doped fiber, a carbon nanotube saturated absorber, an in-line polarizer with a polarization-maintaining pigtail, an optical coupler, and a polarization controller. The order in which these components are connected can be varied. A dual-wavelength laser can simultaneously generate optical pulses of two wavelengths. Because the two pulses have different spectral center wavelengths, their repetition frequencies differ.

[0044] In this example, a dual-wavelength pulse light source is used as a dual-frequency comb source. The dual-frequency comb source can also be composed of one or two lasers, or generated by one or two resonant cavities. The wavelengths of the excitation pulses, sampling pulses, and signal pulses can be in the ultraviolet, visible, infrared, X-ray, terahertz, and electromagnetic wave frequency bands.

[0045] In this example, the dual-wavelength pulse light source generates pulses with repetition frequencies f r =50.602155MHz and f s =50.600145MHz excitation pulse and sampling pulse; after power amplification and broadening, the spectrum width of the wide spectrum excitation pulse is 30nm, and the spectrum width of the wide spectrum sampling pulse is 40nm. The spectrum widths of the wide spectrum excitation pulse and the wide spectrum sampling pulse are both greater than the spectrum aliasing range f r *f s / (2*|f r -f s |)=4.11nm, and the interference spectral range width of the excitation pulse spectrum and the sampling pulse spectrum is 14nm.

[0046] In this example, the system under test is a fiber Bragg grating (FBG) system. The stress on the FBG changes continuously, causing the reflected signal pulse spectrum to change within the range of the wide-spectrum excitation pulse, with a variation range of 14nm. The spectral width of the signal pulse reflected by the FBG is 0.08nm, and the spectral width of the interference signal after sampling is 0.5MHz, which is less than 0.5*f s .

[0047] In this example, the data acquisition and processing unit includes an analog-to-digital conversion acquisition card, a computer, and data processing software. The analog-to-digital conversion acquisition card performs analog-to-digital conversion on the interference signal to obtain a digital signal. The digital signal is input into the data processing software in the computer. The data processing software uses the Fourier transform method to perform time-frequency conversion on the digital signal to obtain the interference signal spectrum containing the actual spectral information in each period, and extracts the intensity spectrum centroid characteristics and the corresponding frequency value from the interference signal spectrum and records them as the frequency measurement value f m .

[0048] The analog-to-digital conversion acquisition card acquires the interference signal of the first cycle to obtain a digital signal, which is input into the data processing software in the computer. The data processing software uses the Fourier transform method to perform time-frequency transformation on the digital signal to obtain the interference signal spectrum containing the actual spectrum information in the first cycle, and extracts the intensity spectrum centroid characteristics and the corresponding frequency value from the interference signal spectrum and records them as the frequency measurement value f m (1), f m (1) The value is between 0 and 0.5*f s Within the frequency range; assuming the correction offset cor is 0, according to the spectrum shape, the frequency value of the spectrum characteristic of the actual spectrum information is greater than 0, and the frequency correction value f c The calculation method is f c (n) = cor + f m (n), n is the number of cycles; calculate f c (1) = 0 + f m (1);

[0049] Assume that the correction offset cor is 0. If the frequency value of the spectrum characteristic of the actual spectrum information is greater than 0, the frequency correction value f c The calculation method is f c (n) = cor + f m (n), n is the number of cycles; if the frequency value of the spectrum characteristic of the actual spectrum information is less than 0, the frequency correction value f c The calculation method is f c (n) = cor-f m (n); according to the frequency correction value f c Calculation method, calculate f c (1);

[0050] Starting from the second period, the interference signal of the nth period containing the actual spectral information in the period is Fourier transformed to obtain the interference signal folded to the range of 0 to 0.5*f s Interference signal spectrum within the frequency range; extract the intensity spectrum centroid feature of the spectrum from the interference signal spectrum, and record the frequency value corresponding to the spectrum feature as the frequency measurement value f m (n), fm The value of (n) is between 0 and 0.5*f s Within the frequency range;

[0051] In this example, the f m (n) is close to 0 or 0.5*f s The judgment method is f m (n) and 0 or 0.5*f s The distance is less than the frequency measurement value f m The frequency measurement value f from the 2nd to the 20th period is 3 times the mean square error of m with 0 or 0.5*f s The distance is greater than the frequency measurement value f m 3 times the mean square error of f m Not close to 0 or 0.5*f s , so no folding occurs, so the cor value remains unchanged, and the frequency correction value calculation method remains unchanged. At this time, cor=0, and the frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor + f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c (n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0052] In this example, the frequency measurement value f of the 21st cycle is m (21) and 0.5*f s The distance is less than the frequency measurement value f m 3 times the mean square error of f m (21) Close to 0.5*f s , using the Kalman filter algorithm, based on the frequency correction value (f c (n-1),f c (n-2)…) gives the frequency prediction value f of the current cycle p (21), the formula is as follows:

[0053] f p (n)=Φ[cor+f m (n-1)]+BU z-1

[0054] Where Φ is the discrete state transfer matrix, BU z-1 Indicates the control quantity, the cor value of the 20th cycle and f m (21) The value calculated by the frequency correction value calculation method of the 20th cycle is the same as f p(21) are located at 0.5 times f s On different sides of , it is judged that folding occurs, because f p (21) greater than f c (20), so cor increases f s , at this time cor=f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor-f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c (n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0055] Frequency measurement values ​​f from the 22nd to the 208th period m with 0 or 0.5*f s The distance is greater than the frequency measurement value f m 3 times the mean square error of , it is judged that no folding occurs, so the cor value remains unchanged, and the calculation method of the frequency correction value remains unchanged. At this time, cor=f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor-f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c (n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0056] Frequency measurement value f of the 209th cycle m (21) is close to 0, and the Kalman filter algorithm is used to give the frequency prediction value f of the current cycle p (209), the cor value and f in the 208th cycle m (209) The value calculated by the frequency correction value calculation method of the 208th cycle is the same as f p (209) are located at 1 times f s On different sides of , it is judged that folding occurs, because f p (209) is less than f c (208), so cor remains unchanged, and cor=f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor + f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c(n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0057] Frequency measurement values ​​f from the 210th to the 399th cycles m with 0 or 0.5*f s The distance is greater than the frequency measurement value f m 3 times the mean square error of , it is judged that no folding occurs, so the cor value remains unchanged, and the calculation method of the frequency correction value remains unchanged. At this time, cor=f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor + f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c (n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0058] Frequency measurement value f of the 400th cycle m (400) is close to 0.5*f s , using the Kalman filter algorithm, gives the frequency prediction value f of the current cycle p (400), the cor value and f in the 399th cycle m (400) The value calculated by the frequency correction value calculation method of the 399th cycle is the same as f p (400) are located at 1.5 times f s On different sides of , it is judged that folding occurs, because f p (400) is greater than f c (399), so cor increases f s , at this time cor=2*f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor-f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c (n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0059] Frequency measurement value f from the 401st to 600th cycles m with 0 or 0.5*f s The distance is greater than the frequency measurement value f m3 times the mean square error of , it is judged that no folding occurs, so the cor value remains unchanged, and the calculation method of the frequency correction value remains unchanged. At this time, cor=2*f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor-f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c (n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0060] Frequency measurement value f of the 601st cycle m (601) is close to 0.5*f s , using the Kalman filter algorithm, gives the frequency prediction value f of the current cycle p (601), the cor value and f of the 600th period m (601) The value calculated according to the frequency correction value calculation method of the 600th cycle is the same as f p (601) are located at 1.5 times f s On different sides of , it is judged that folding occurs, because f p (601) is less than f c (600), so cor decreases f s , at this time cor=f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor + f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c (n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0061] Frequency measurement values ​​f from the 601st to the 837th cycles m with 0 or 0.5*f s The distance is greater than the frequency measurement value f m 3 times the mean square error of , it is judged that no folding occurs, so the cor value remains unchanged, and the calculation method of the frequency correction value remains unchanged. At this time, cor=f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor + f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c(n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0062] Frequency measurement value f of the 838th cycle m (838) is close to 0, and the Kalman filter algorithm is used to give the frequency prediction value f of the current cycle. p (838), the cor value and f in the 837th cycle m (838) The value calculated by the frequency correction value calculation method of the 837th cycle is the same as f p (838) is located at 1 times f s On the right side, it is judged that there is no folding, so the cor value remains unchanged, and the calculation method of the frequency correction value remains unchanged. At this time, cor=f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor + f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c (n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0063] Frequency measurement value f from the 839th to the 997th period m with 0 or 0.5*f s The distance is greater than the frequency measurement value f m 3 times the mean square error of , it is judged that no folding occurs, so the cor value remains unchanged, and the calculation method of the frequency correction value remains unchanged. At this time, cor=f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor + f m (n), according to the value of cor and the frequency correction value calculation method, f is calculated c (n); obtaining the frequency change of the actual spectral information according to the frequency correction value fc(n) and recovering the frequency position of the actual signal spectral information;

[0064] The measurement ends after the 997th cycle.

[0065] In this example, the spectral feature of the interference signal spectrum used is the centroid of the intensity spectrum. A spectral feature can also be the intensity peak of the interference signal spectrum, the position of a specific phase value in the phase spectrum, or the similarity or correlation between multiple points within a certain frequency range of the interference amplitude or phase spectrum, or between all points of the interference amplitude spectrum at different times.

[0066] In this example, the signal change trend prediction algorithm used is the Kalman filter method. The signal change trend prediction algorithm may also include, but is not limited to, one or more methods such as multi-point interpolation, spline interpolation, linear interpolation, Hermite interpolation, synovial observation, Lemberg observation, moving horizon prediction, implicit dynamic feedback, filter deviation update, least squares curve fitting, and autoregressive prediction.

[0067] The frequency measurement value curve, correction offset curve and frequency correction value curve from the 1st to the 997th cycle are as follows Figure 4 shown.

[0068] The spectral information before correction directly recovered from the frequency measurement value and the spectral information of the actual signal after correction recovered from the frequency correction value are as follows: Figure 5 As shown in FIG, the spectrum range of the signal spectrum information before correction is only 4.11 nm, and the spectrum range of the actual signal spectrum information after correction is restored to 14 nm.

[0069] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art may make appropriate changes or modifications within the technical scope disclosed in the present invention, and such changes or modifications shall be included in the scope of protection of the present invention.

Claims

1. A wide-spectrum dual-frequency comb dynamic spectroscopy measurement method, characterized in that: The specific steps are as follows: Step 1: Use a repetition frequency f r The wide spectrum excitation pulse is input into the system to be tested, generating a signal pulse with continuously changing spectral information within the spectrum range of the excitation pulse; using a pulse with a repetition frequency f s The wide spectrum sampling pulse samples the signal pulse to generate a periodic interference signal; the spectrum width of the wide spectrum excitation pulse and the wide spectrum sampling pulse are both greater than the spectrum aliasing range f r *f s / (2*|f r -f s |); Step 2: Perform time-frequency transformation on the interference signal of the first period containing the actual spectral information in the period to obtain the signal folded to the range of 0 to 0.5*f s Interference signal spectrum within the frequency range; extract spectrum features from the interference signal spectrum, and record the frequency value corresponding to the spectrum feature as the frequency measurement value f m (1), f m (1) The value is between 0 and 0.5*f s Within the frequency range; let the correction offset cor be 0. If the frequency value of the spectrum characteristic of the actual spectral information is greater than 0, the frequency correction value f c The calculation method is f c (n) = cor + f m (n), n is the number of cycles; if the frequency value of the spectrum characteristic of the actual spectral information is less than 0, the frequency correction value f c The calculation method is f c (n) = cor-f m (n); according to the frequency correction value f c Calculation method, calculate f c (1); Step 3: Starting from the second period, perform time-frequency transformation on the interference signal of the nth period containing the actual spectral information in the period, and obtain the signal folded to the range of 0 to 0.5*f s Interference signal spectrum within the frequency range; extract spectrum features from the interference signal spectrum, and record the frequency value corresponding to the spectrum feature as the frequency measurement value f m (n), f m The value of (n) is between 0 and 0.5*f s Within the frequency range; When f m (n) is close to 0 or 0.5*f s When the signal change trend prediction algorithm is used, the signal change trend prediction algorithm is based on the frequency correction value (f c (n-1),f c (n-2)…) gives the frequency prediction value f of the current cycle p (n), according to the previous f c 、f p (n) and f m (n) to determine whether folding occurs: 1) If folding occurs, then a) when f m When (n) is close to 0, if f p (n) is greater than f c (n-1), then cor remains unchanged, and the frequency correction value f c (n) is calculated as f c (n) is equal to cor + f m (n); If f p (n) is less than f c (n-1), then cor remains unchanged, and the frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor-f m (n); b) when f m (n) is close to 0.5*f s When f p (n) is greater than f c (n-1), then cor increases by f s , frequency correction value f c (n) is calculated as the frequency correction value f c (n) is equal to cor-f m (n); if f p (n) is less than f c (n-1), then cor decreases f s , frequency correction value f c (n) is calculated as follows: the frequency correction value fc(n) is equal to cor+f m (n); 2) If it is determined that folding has not occurred, the cor value remains unchanged, and the frequency correction value calculation method remains unchanged; the calculated f c (n); When f m (n) is not close to 0 or 0.5*f s When , the cor value remains unchanged, and the calculation method of the frequency correction value remains unchanged; Step 4: Calculate f based on the value of cor and the frequency correction value calculation method. c (n); Obtain the frequency variation of the actual spectral information according to the frequency correction value fc(n) and recover the frequency position of the actual signal spectral information; Step 5. Repeat steps 3 to 4 until the measurement is completed.

2. A wide spectrum range dual-frequency comb dynamic spectroscopy measurement method according to claim 1, characterized in that: The spectral characteristics of the interference signal spectrum described in steps 2 and 3 are the intensity peak of the interference signal spectrum, the center of mass of the intensity spectrum, the specific phase value position of the phase spectrum, or the similarity or correlation between multiple points within a certain frequency range of the interference amplitude or phase spectrum or all points of the interference amplitude spectrum relative to different times.

3. The wide spectrum range dual-frequency comb dynamic spectroscopy measurement method according to claim 1, characterized in that: The signal change trend prediction algorithm described in step 3 includes, but is not limited to, one or more methods selected from the group consisting of multi-point interpolation, spline interpolation, linear interpolation, Hermite interpolation, Kalman filtering, synovial observation, Lemberg observation, moving horizon prediction, implicit dynamic feedback, filter deviation update, least squares curve fitting, and autoregressive prediction.

4. The wide spectrum range dual-frequency comb dynamic spectroscopy measurement method according to claim 1, characterized in that: According to the previous f c 、f p (n) and f m The method to determine whether folding occurs is to compare the cor value of the n-1 cycle with f m (n) The value calculated according to the frequency correction value calculation method of the n-1th cycle is the same as f p (n) values ​​are compared. If the two values ​​are located at the folding boundary (f s Integer multiples of f s If the two sides are on different sides of (integer + 0.5) times), it is judged that folding has occurred; otherwise, it is judged that folding has not occurred.

5. The wide spectrum range dual-frequency comb dynamic spectroscopy measurement method according to claim 1, characterized in that: f described in step 3 m (n) is close to 0 or 0.5*f s The judgment method is: f m (n) and 0 or 0.5*f s The distance is less than the frequency measurement value f m 3 times the mean square error.

6. A wide spectrum range dual frequency comb dynamic spectroscopy measurement system, characterized in that: The wide spectrum range dual-frequency comb dynamic spectroscopy measurement system comprises a wide spectrum dual-frequency comb source, a system to be measured, a detection device and a data acquisition and processing unit; wherein the dual-frequency comb source generates a repetition frequency f r and f s The wide spectrum excitation pulse and sampling pulse are input to the system to be measured, and the system to be measured outputs the signal pulse generated by the excitation pulse; the signal pulse and the sampling pulse generate a periodic interference signal through the detection device; the data acquisition and processing unit collects and transforms the interference signal into a time-frequency signal, and obtains the actual spectral information in each period, which is folded to the position between 0 and 0.5*f s The interference signal spectrum within the frequency range is extracted, and the spectrum characteristics and their corresponding frequency values ​​are extracted from the interference signal spectrum and recorded as the frequency measurement value f m ; Assume the correction offset cor, its initial value is 0, according to cor and f m Calculate the frequency correction value f c When f m Close to 0 or 0.5*f s When the signal change trend prediction algorithm is used based on the frequency correction value of no more than K cycles before, the frequency prediction value f of the current cycle is given. p , according to the previous f c 、The current f m With f p The relationship between the two determines whether folding occurs. When folding occurs, the value of cor and the calculation method of the frequency correction value are updated to give the current period f c ; According to the frequency correction value f c The frequency variation of the actual spectral information is obtained and the frequency position of the actual signal spectral information is restored, thereby realizing a dynamic measurement system for spectral information in a wide spectrum range.

7. A wide spectrum range dual-frequency comb dynamic spectroscopy measurement system according to claim 6, characterized in that: The broadband dual-frequency comb source can be composed of one or two lasers and a resonant cavity. The wavelengths of the excitation pulse, sampling pulse and signal pulse can be in the ultraviolet, visible, infrared, X-ray, terahertz and electromagnetic wave frequency bands.

8. The wide spectrum range dual-frequency comb dynamic spectroscopy measurement system according to claim 6, characterized in that: The spectrum width of the wide spectrum excitation pulse and the wide spectrum sampling pulse are both larger than the spectrum aliasing range f r *f s / (2*|f r -f s |).

9. The wide spectrum range dual-frequency comb dynamic spectroscopy measurement system according to claim 6, characterized in that: The spectrum of the signal pulse varies within the spectrum of the broadband excitation pulse.

10. The wide spectrum range dual-frequency comb dynamic spectroscopy measurement system according to claim 6, characterized in that: The width of the spectrum of the sampled interference signal of the signal pulse is less than 0.5*f s .

Images