High-precision FSI ranging method based on CEEMD algorithm
Through the combination of CEEMD algorithm and Hilbert transform, the problem of nonlinear impact of optical frequency scanning in the FSI range measurement system is solved, the ranging accuracy and stability are improved, the system structure is simplified and the cost is reduced.
Patent Information
- Application Number
- CN202211027810.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-08-25
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2042-08-25
AI Technical Summary
The FSI ranging system is affected by the nonlinearity of the laser optical frequency scanning, resulting in poor distance measurement accuracy and stability, and the existing technology is difficult to effectively solve this problem.
The CEEMD algorithm based on the supplementary population average empirical mode decomposition is used to decompose and reconstruct the interfering signal, and the phase difference is extracted in combination with the Hilbert transform, which improves the signal-to-noise ratio and reduces the nonlinear impact of optical frequency scanning.
It significantly improves the ranging accuracy and system stability, simplifies the ranging system structure, reduces the construction cost, improves the component modal aliasing problem, and improves the reconstruction quality of the interference signal.
Smart Images

Figure CN115435676B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of optical precision measurement, and further relates to an optical frequency scanning interferometry ranging method, specifically a high-precision optical frequency scanning interferometry FSI (Frequency Scanning Interferometry) ranging method based on a complementary ensemble average empirical mode decomposition (CEEMD) algorithm, which can be used to improve the signal-to-noise ratio of an optical interference signal, reduce the influence of noise and optical frequency scanning nonlinearity on the phase extraction of the interference signal, further improve the phase extraction accuracy of the interference signal, and improve the ranging accuracy of the FSI system. Background Art
[0002] As engineering machinery equipment develops towards precision, high-precision, large-range, non-contact large-size measurement technology has become one of the key technologies in the equipment manufacturing industry. As a typical large-size measurement technology, FSI ranging technology has been widely studied and applied in equipment manufacturing, space exploration, etc., but the optical frequency nonlinearity of the ranging system has always been an important factor limiting the measurement accuracy of this method. As the light source of the FSI system, the tunable external cavity semiconductor laser ECDL (External Cavity Diode Laser) adjusts the structure and size of the internal optical path of the laser by changing the piezoelectric ceramic PZT (Piezoelectric Ceramic Transducer) or the input voltage at the end of the motor to change the length of the resonant cavity, thereby achieving optical frequency tuning output. However, affected by factors such as PZT hysteresis characteristics and motor mechanical jitter, the output optical frequency of ECDL will change nonlinearly. Along with the nonlinear change of optical frequency, the measured interference signal has a non-stationary characteristic, which seriously affects the extraction of the interference signal phase and the ranging accuracy of the FSI system.
[0003] In the FSI ranging system, after the laser receives an external driving signal, the external driving signal drives the piezoelectric ceramic or motor to adjust the length of the laser's internal resonant cavity to achieve the light source tuning output. However, due to the hysteresis and creep characteristics of the piezoelectric ceramic or the jitter of the motor end, the optical frequency output cannot be linearly tuned. This phenomenon is called laser optical frequency scanning nonlinearity. Under this influence, the signal generated by interference becomes a non-stationary signal. Therefore, the laser optical frequency scanning nonlinearity directly affects the ranging accuracy and ranging stability of the FSI ranging system.
[0004] In order to eliminate the influence of the nonlinearity of the laser optical frequency scanning on the FSI ranging system, the following common technologies are currently available: 1) Front-end suppression of the optical frequency scanning nonlinearity caused by the laser, using feedback control methods, introducing devices such as phase-locked loops, and reducing the nonlinear changes of the laser frequency through feedback iteration to achieve stable output of the laser frequency; although this method has the ability to suppress the nonlinearity of optical frequency scanning, its scope of application is limited and it is not suitable for high-frequency optical frequency scanning, and additional devices will increase the complexity of the system structure and the system cost. 2) Under the continuous optical frequency scanning mechanism, the FSI system will generate an interference signal for distance measurement. If the signal processing method is used to perform back-end processing on the interference signal generated under the influence of the optical frequency nonlinear scanning, the influence of the optical frequency scanning nonlinearity on the ranging accuracy can be reduced or even eliminated. However, this technology requires the introduction of a set of reference interferometers with known distances. Under the same frequency-sweeping light source, the measured interference signal is resampled at the intermediate frequency domain interval points in the auxiliary interference signal, or the interference signal is subjected to FFT transformation after resampling. Although it can effectively improve the spatial resolution of the interference signal of the measuring optical path and reduce the influence of the optical frequency scanning nonlinearity on the interference signal, it requires an external auxiliary optical path, which has certain limitations on the measurement range and increases the complexity of the system, and is not applicable to practical applications. 3) Under the influence of optical frequency nonlinear scanning, the interference signal generated by the system becomes a non-stationary signal. Empirical mode decomposition (EMD) uses the distribution characteristics of the signal in the time domain to decompose and process the signal, so it has many advantages in processing non-stationary signals. Using empirical mode decomposition (EMD) and HT to directly extract the phase of the stationary interference signal can effectively reduce the influence of optical frequency scanning nonlinearity and improve the extraction accuracy of the interference signal phase under the influence of frequency scanning nonlinearity. However, EMD has problems such as aliasing interference between components when performing signal decomposition. This defect limits the accuracy and stability of EMD in extracting the phase of non-stationary interference signals. Summary of the invention
[0005] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art and to propose a high-precision FSI ranging method based on the CEEMD algorithm, which is used to solve the problem that the FSI ranging system is affected by the nonlinearity of the laser optical frequency scanning, resulting in poor accuracy and stability of the ranging system.
[0006] The basic idea of implementing the present invention is as follows: first, an optical frequency scanning interferometric ranging system is constructed to generate the interference signal required for distance solution; a laser is used to continuously tune the optical frequency and output it, and a detector is used to detect the interference signal after passing through the interference optical path to generate a sinusoidal signal that changes continuously in the time domain; at the same time, the system monitors the synchronous changes of the laser frequency and the interference signal, and uses the CEEMD algorithm to decompose and reconstruct the sinusoidal interference signal, thereby improving the signal-to-noise ratio of the sinusoidal signal while reducing the influence of the optical frequency scanning nonlinearity on the sinusoidal interference signal; then, the Hilbert Transform (HT) is used to extract the phase of the processed interference signal, and the distance measurement result is obtained by solving the phase difference of two sinusoidal interference signals that change continuously at different times within a certain optical frequency change range.
[0007] The present invention achieves the above-mentioned purpose by the following specific steps:
[0008] (1) Using a tunable external cavity semiconductor laser ECDL, a collimator, a beam splitter, a Fabry-Perot etalon, a photodetector, and a Michelson interferometer optical path to build an FSI ranging system, and equip the system with a data acquisition device;
[0009] (2) The tunable external cavity semiconductor laser ECDL of the FSI ranging system generates a tunable light source S whose frequency changes with time t and outputs it to the collimator;
[0010] (3) After the collimator collimates the tuned light source S, it is divided into two light beams S1 and S2 using a beam splitter prism, and S1 enters the Fabry-Perot etalon, and S2 enters the Michelson interferometer optical path;
[0011] (4) The Fabry-Perot FP etalon obtains the transmission signal of S1, and uses a photodetector to detect the transmission signal to obtain the FP signal; the photodetector in the Michelson interference optical path detects S2 that has entered the optical path and has undergone splitting, reflection, and convergence interference to obtain an interference signal;
[0012] (5) Synchronously sample the FP signal and the interference signal obtained in step (4) using a data acquisition device, and preprocess them:
[0013] (5.1) The tuning range Δv of the laser ECDL output tuning light source is calculated based on the number of resonance peaks of the FP signal:
[0014] Δv=(r-1)·FSR,
[0015] Where r represents the number of resonance peaks of the FP signal, and FSR represents the optical frequency difference between two adjacent resonance peaks;
[0016] (5.2) The interference signal is intercepted by using the time when the resonance peaks of the first and last two FP signals in the sampling interval are located;
[0017] (5.3) dividing the interference signal intercepted in step (5.2) according to the period, into an integer period interference signal and a fractional period interference signal, and further dividing the fractional period interference signal to obtain a starting fractional period interference signal and an ending fractional period interference signal;
[0018] (6) The fractional period interference signal is processed using the CEEMD algorithm, that is, the starting fractional period interference signal and the ending fractional period interference signal are decomposed, and the decomposed components are arranged in order from high to low frequency, and the high-frequency noise component is removed according to the arrangement result, and the real stationary component reflecting the characteristics of the interference signal is retained, and then the retained real stationary component is superimposed and reconstructed, and finally a reconstructed fractional period interference signal with a high signal-to-noise ratio is obtained;
[0019] (7) Using Hilbert transform to solve the phase difference of the reconstructed fractional period interference signal The implementation steps are as follows:
[0020] (7.1) Perform Hilbert transform on the starting fractional period interference signal and the ending fractional period interference signal obtained in step (5.3), take the Hilbert transformed signal as the imaginary part and the original signal as the real part, and construct the analytical function of the fractional period interference signal:
[0021]
[0022] Where i represents an imaginary number; b(t) represents the intensity amplitude function of the coherent term of the fractional period interference signal; Represents the instantaneous phase function of the fractional periodic interference signal;
[0023] (7.2) Calculate the instantaneous phase value of the fractional period interference signal:
[0024]
[0025] (7.3) Obtain multiple instantaneous phase values corresponding to different times, and perform phase unwrapping on them to obtain a continuous instantaneous phase function that changes with time Then, by calculating the instantaneous phase values corresponding to the first and last moments t1 and t2 and Make a difference and get the phase difference of the fractional period interference signal to be calculated
[0026] (8) The total phase difference Δφ is obtained by the following steps:
[0027] (8.1) Using the peak number of the integer-period interference signal, the phase difference Δφ1 of the integer-period interference signal is obtained according to the following formula:
[0028] Δφ1=(δ-1)·2π
[0029] Among them, δ represents the number of peaks;
[0030] (8.2) Use the solution method of the fractional period interference signal obtained in step (7.3) to solve the phase difference corresponding to the starting fractional period interference signal and the ending fractional period interference signal, and obtain the starting fractional phase difference Δφ s The phase difference Δφ with the ending decimal e .
[0031] (8.3) to obtain the total phase difference Δφ:
[0032] Δφ=Δφ1+Δφ s +Δφ e ;
[0033] (9) Calculate the distance L to be measured according to the following formula to complete the distance measurement:
[0034]
[0035] Where n is the refractive index of air and c is the speed of light.
[0036] Compared with the prior art, the present invention has the following advantages:
[0037] First, since the interference signal obtained by the data acquisition system is affected by noise and the nonlinearity of the laser frequency sweep, which affects the accuracy of the distance solution, the present invention achieves the purpose of eliminating adverse effects and improving the system ranging accuracy by only processing the key signals involved, thereby eliminating the need to add other auxiliary measurement equipment and measurement devices. Under the premise of ensuring the system ranging accuracy, the complexity of the ranging system is simplified and the construction cost of the measurement system is reduced.
[0038] Second, since the present invention uses the CEEMD algorithm to directly decompose and reconstruct the sinusoidal interference signal to obtain a high-quality interference signal, and then uses the Hilbert transform HT to extract the phase of the reconstructed sinusoidal interference signal, it effectively improves the accuracy of the Hilbert transform in solving the phase difference of the interference signal, thereby significantly improving the ranging accuracy.
[0039] Third, the method of the present invention utilizes the characteristic of uniform distribution of white noise in time-frequency space, and automatically maps component signals of different time-frequency scales to appropriate time-frequency scales related thereto through the white noise background, thereby ensuring the continuity of each component in the time domain, thereby reducing modal aliasing; in addition, the time-domain zero-mean characteristic of white noise is also utilized, and the CEEMD algorithm is used to add white noise pairs multiple times, and multiple groups of IMF components of the same order are averaged to eliminate the influence of the added white noise on the original signal; thereby, while taking into account the performance of EMD adaptive decomposition, the component modal aliasing problem caused by EMD decomposition can be effectively improved, thereby improving the reconstruction quality of the sinusoidal interference signal and effectively improving the ranging accuracy under the influence of the nonlinearity of laser optical frequency scanning. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] Figure 1 It is a flow chart of the implementation of the present invention;
[0041] Figure 2 It is a structural schematic diagram of the FSI system in the present invention;
[0042] Figure 3 is the Hilbert transform integration area of the product signal;
[0043] Figure 4 It is a phase decomposition diagram of the simulated interference signal;
[0044] Figure 5 It is a schematic diagram of the decomposition and reconstruction of non-stationary interference signals by EMD;
[0045] Figure 6 This is a schematic diagram of the decomposition and reconstruction of the interference signal by CEEMD. DETAILED DESCRIPTION
[0046] The present invention will be further described below in conjunction with the accompanying drawings.
[0047] Embodiment 1:
[0048] See attached Figure 1 The present invention proposes a high-precision optical frequency scanning interferometry (FSI) ranging method based on a supplementary population average empirical mode decomposition (CEEMD) algorithm, which specifically includes the following steps:
[0049] Step 1. Build an FSI ranging system using a tunable external cavity semiconductor laser ECDL, a collimator, a beam splitter, a Fabry-Perot FP standard, a photodetector and a Michelson interference optical path, and equip the system with a data acquisition device; the beam splitter is provided with two, namely a first beam splitter prism BS1 and a second beam splitter prism BS2; wherein the first beam splitter prism BS1 is used to split the tuned light source after collimation by the collimator, and the second beam splitter prism BS2 is located in the Michelson interference optical path, and is used to split the light beam entering the optical path again.
[0050] Step 2. Generate a tuned light source S whose frequency changes with time t through the tunable external cavity semiconductor laser ECDL of the FSI ranging system and output it to the collimator;
[0051] Step 3. After the collimator collimates the tuned light source S, it is divided into two light beams S1 and S2 using a beam splitter prism, and S1 enters the Fabry-Perot FP etalon and S2 enters the Michelson interferometer optical path;
[0052] Step 4. The Fabry-Perot FP etalon obtains the transmission signal of S1, and uses a photodetector to detect the transmission signal to obtain an FP signal; the photodetector in the Michelson interference optical path detects S2 that has entered the optical path and has undergone splitting, reflection, and convergence interference to obtain an interference signal;
[0053] The light splitting, reflection and convergence interference are as follows: after S2 enters the Michelson interference optical path, it is split into two new light beams S2 by the second light splitting prism BS2. 21 and S 22 , where S 21 As the reference light of the ranging optical path, S 22 The two are respectively reflected by the corner cone reflection prisms arranged at different positions, and then converge and interfere at the second beam splitter prism BS2.
[0054] The interference signal, according to the light wave interference theorem, is expressed as follows:
[0055]
[0056] Among them, I r (t) represents the intensity of the reference light, I m (t-τ) represents the intensity of the measured light, τ is the delay of the measured light relative to the reference light in the time domain; v0 represents the initial optical frequency output by the laser, and β(t) represents the rate of change of the optical frequency; the above expression is simplified as follows:
[0057] I(t)=a(t)+b(t)cosθ(t),
[0058] Where, a(t) = I r (t)+I m (t-τ) represents the light intensity amplitude function of the low-frequency DC term of the interference signal; A function of the intensity amplitude representing the coherence term of the interference signal.
[0059] Step 5. Use a data acquisition device to synchronously sample the FP signal and the interference signal obtained in step 4, and pre-process them:
[0060] (5.1) The tuning range Δv of the laser ECDL output tuning light source is calculated based on the number of resonance peaks of the FP signal:
[0061] Δv=(r-1)·FSR,
[0062] Wherein, r represents the number of resonance peaks of the FP signal, and FSR represents the optical frequency difference corresponding to two adjacent resonance peaks; in this embodiment, the optical frequency difference FSR corresponding to two adjacent resonance peaks is 1.5 GHz.
[0063] (5.2) The interference signal is intercepted by using the time when the resonance peaks of the first and last two FP signals in the sampling interval are located;
[0064] (5.3) dividing the interference signal intercepted in step (5.2) according to the period, into an integer period interference signal and a fractional period interference signal, and further dividing the fractional period interference signal to obtain a starting fractional period interference signal and an ending fractional period interference signal;
[0065] Step 6. Use the CEEMD algorithm to process the fractional period interference signal, that is, decompose the starting fractional period interference signal and the ending fractional period interference signal, and arrange the decomposed components in descending order according to the frequency, remove the high-frequency noise component according to the arrangement result, retain the real stationary component reflecting the characteristics of the interference signal, and then superimpose and reconstruct the retained real stationary component, and finally obtain a reconstructed fractional period interference signal with a high signal-to-noise ratio;
[0066] The supplementary population average empirical mode decomposition (CEEMD) algorithm decomposes the intercepted interference signal, specifically: a positive and a negative white noise pair is added to the original signal in sequence, and the component signals of different time-frequency scales are automatically mapped to the time-frequency scale related thereto through the white noise background by utilizing the characteristic that the white noise is evenly distributed in the time-frequency space, and then the empirical mode decomposition is performed respectively to obtain multiple intrinsic mode function (IMF) components; and the time-domain zero mean characteristic of the white noise is utilized to add the white noise pair multiple times and perform the empirical mode decomposition, and the multiple groups of IMF components of the same order are averaged to obtain the final component.
[0067] Step 7. Use Hilbert transform to solve the phase difference of the reconstructed fractional period interference signal The implementation steps are as follows:
[0068] The Hilbert transform, for any signal x(t), is expressed as follows:
[0069]
[0070] in, and denote Fourier transform and inverse Fourier transform respectively, and H[ω] is the transfer function of the Hilbert change.
[0071] (7.1) Perform Hilbert transform on the starting fractional period interference signal and the ending fractional period interference signal obtained in step (5.3), take the Hilbert transformed signal as the imaginary part and the original signal as the real part, and construct the analytical function of the fractional period interference signal:
[0072]
[0073] Where i represents an imaginary number; b(t) represents the intensity amplitude function of the coherent term of the fractional period interference signal; Represents the instantaneous phase function of the fractional periodic interference signal;
[0074] (7.2) Calculate the instantaneous phase value of the fractional period interference signal:
[0075]
[0076] (7.3) Obtain multiple instantaneous phase values corresponding to different times, and perform phase unwrapping on them to obtain a continuous instantaneous phase function that changes with time Then, by calculating the instantaneous phase values corresponding to the first and last moments t1 and t2 and Make a difference and get the phase difference of the fractional period interference signal to be calculated
[0077] Step 8. Obtain the total phase difference Δφ as follows:
[0078] (8.1) Using the peak number of the integer period interference signal, the integer phase difference Δφ1 is obtained according to the following formula:
[0079] Δφ1=(δ-1)·2π
[0080] Among them, δ represents the number of peaks;
[0081] (8.2) Use the solution method of the fractional period interference signal obtained in step (7.3) to solve the phase difference corresponding to the starting fractional period interference signal and the ending fractional period interference signal, and obtain the actual fractional phase difference Δφ s The phase difference Δφ with the ending decimal e .
[0082] (8.3) to obtain the total phase difference Δφ:
[0083] Δφ=Δφ1+Δφ s +Δφ e ;
[0084] Step 9. Calculate the distance L to be measured according to the following formula to complete the distance measurement:
[0085]
[0086] Where n is the refractive index of air and c is the speed of light.
[0087] Example 2: Reference Figure 2 The implementation steps of the method of the present invention are the same as those of the first embodiment. Now, the specific structure of the FSI system and its ranging principle in the present invention are further described:
[0088] 1. Structure of FSI ranging system:
[0089] The optical frequency scanning interference FSI ranging system includes: a tunable external cavity semiconductor laser ECDL, a collimator, a first beam splitter prism BS1, a Fabry-Perot FP etalon, a first photodetector PD1, a Michelson interference optical path and a data acquisition device;
[0090] The tunable external cavity semiconductor laser ECDL is used to generate a tunable light source with a continuously changing frequency. After being collimated by a collimator, the tunable light source is split into two light beams S1 and S2 at a first beam splitter prism BS1. One light beam S1 directly enters a Fabry-Perot FP etalon, and a first photodetector PD1 detects a transmission signal generated by the Fabry-Perot FP etalon to obtain an FP signal; the other light beam S2 enters a Michelson interference optical path;
[0091] The Michelson interference optical path includes a second beam splitter prism BS2, a first corner reflector prism RR1, a second corner reflector prism RR2 and a second photodetector PD2; S2 is again split into two new beams S by the second beam splitter prism BS2. 21 and S 22 , where S 21 As the reference light of the ranging optical path, S 22The two are respectively reflected by the first corner pyramid prism RR1 and the second corner pyramid prism RR2, and then converge and interfere at the second beam splitter prism BS2, and are detected by the second photodetector PD2 to obtain an interference signal;
[0092] The data acquisition device is used to achieve synchronous sampling of the FP signal and the interference signal;
[0093] By processing the interference signal and FP signal, the optical frequency variation range Δv and the interference signal phase difference are realized. The solution is obtained and the target distance measurement is finally completed.
[0094] 2. The ranging principle of FSI ranging system:
[0095] In the FSI ranging system, the output frequency of the laser can be expressed as:
[0096] v(t)=v0+β·t <1>
[0097] Among them, v0 is the initial optical frequency output by the laser, and β is the rate of change of the optical frequency.
[0098] According to the light wave interference theorem, the interference signal can be expressed as:
[0099]
[0100] Among them, I r is the intensity of the reference light, I m is the intensity of the measured light, τ is the delay of the measured light relative to the reference light in the time domain, and thus the instantaneous phase of the interference signal I(t) is:
[0101]
[0102] The phase change of the interference signal I(t) from time t1 to time t2 It can be expressed as:
[0103]
[0104] Combined <1> The phase difference of the interference signal from time t1 to time t2 is:
[0105]
[0106] Among them, Δv is the range of optical frequency change from time t1 to time t2.
[0107] When the distance to be measured is L, the delay τ can be expressed as:
[0108]
[0109] Where n is the refractive index of air and c is the speed of light.
[0110] Combined <5> , <6> We can get:
[0111]
[0112] Among them, the speed of light c and the refractive index of air n are constant in a specific measurement environment, so the ranging accuracy of the system depends on the phase difference of the interference signal The extraction of and calculation of the optical frequency variation range Δv. Figure 1 In the FSI measurement system shown, the optical frequency variation range is measured by the FP standard tool. The optical frequency variation range Δv and the number of resonance peaks r measured by FP have the following relationship:
[0113] Δv=(r-1)·FSR <8>
[0114] Among them, FSR is the optical frequency difference corresponding to two adjacent resonance peaks.
[0115] The present invention uses Hilbert transform to process the interference signal to obtain the phase change of the interference signal. Substitute the obtained optical frequency change and phase change into the formula <7> , complete the acquisition of target distance.
[0116] Example 3: Reference Figure 3 and Figure 4 The implementation steps of the method of the present invention are the same as those of the first embodiment. The phase solution principle based on Hilbert transform in the present invention is further described as follows:
[0117] It is known that Hilbert transform HT can realize the definition and calculation of instantaneous parameters of complex signals, and is a powerful tool for analyzing phase changes of complex signals, and HT has been successfully applied in FSI ranging. However, there is still a lack of rigorous theoretical derivation and detailed theoretical analysis for the engineering application of HT under specific conditions. The present invention will make a theoretical derivation and analysis on the rationality and feasibility of the application of HT in FSI ranging technology, and use HT to realize the solution of the phase difference of the interference signal in the FSI ranging system. The theoretical derivation is as follows.
[0118] For any signal x(t), its Hilbert transform can be expressed as:
[0119]
[0120] in, and is the Fourier transform and inverse Fourier transform, H[ω] is the transfer function of the Hilbert change, which can be specifically expressed as:
[0121]
[0122] However, in the FSI ranging system, the interference signal expression is a complex polynomial, so it is necessary to analyze and derive the Hilbert transform of the interference signal. <2> , the interference signal I(t) can be expressed as:
[0123]
[0124] Among them, a(t)=I r (t)+I m (t-τ) represents the light intensity amplitude function of the low-frequency DC term of the interference signal; represents the intensity amplitude function of the coherent term of the interference signal; the phase information of the interference signal is implicit in In the above example, let ω a ,ω b ,ω c They are a(t), b(t), The frequency, ω a With ω b is the frequency of the low-frequency DC signal, ω c is the frequency of the beat signal, and the three satisfy ω c >>ω b ≈ω a > 0, so there must exist ω such that <18> Established:
[0125]
[0126] From the convolution theorem of Fourier transform, we can get:
[0127]
[0128] According to the above derivation, there exists ω1∈(ω a ,ω c -2ω b ) makes:
[0129]
[0130] Therefore, a high-pass filter can be used to filter out the low-frequency DC term a(t), leaving the product term containing the phase information of the interference signal. After high-pass filtering, the interference signal I(t) can be expressed as:
[0131]
[0132] Among them, W(ω) is the window function of the high-pass filter, which can be expressed as follows:
[0133]
[0134] At this time, you need to <15> The product signal The Hilbert transform is analyzed and deduced to obtain the phase information of the interference signal. In fact, some people have made a detailed discussion on the Hilbert transform in the form of a product, providing theoretical experience for our analysis of the application of the Hilbert transform in FSI. Let the signals f(t) and g(t) be functions that vary with time t, and their Fourier transforms be F(ω f ) and G(ω g ), respectively. Therefore, f(t)·g(t) can be expressed in the form of its inverse Fourier transform:
[0135]
[0136] Among them, the common Hilbert transform formula is deduced and summarized, and the Hilbert transform formula of the exponential function is given, as shown in Equation (18).
[0137] H[e ibt =isgn(b)e ibt <18>
[0138] Therefore, combining Equation <9>, the Hilbert transform of f(t)·g(t) can be expressed as:
[0139]
[0140] Next, the integral domain that satisfies Equation <19> is analyzed as follows. As Figure 3 shown, when F(ω f ) and G(ω g ) satisfy |ω f |<ω and |ω g |>ω, F(ω f )·G(ω g )≠0, and the Hilbert transform of the signal product form exists; Figure 3 shown, when ω f >0 and ω g >0, F(ω f )·G(ω g )≠0, and the Hilbert transform of the signal product form exists; Combining the above two conditions, in practical engineering applications, when the frequency satisfies ω g >ω>ω f >0, the Hilbert transform of the signal product form exists. Under this condition, we can obtain:
[0141] sgn(ω f +ω g )=sgn(ω g ) <20>
[0142] Therefore, Equation <19> can be written as:
[0143]
[0144] because, So we can get:
[0145] H[f(t)·g(t)]=f(t)·H[g(t)] <22>
[0146] In summary, when the Fourier transform of f(t) and g(t) satisfies That is, when f(t) is a low-pass signal and g(t) is a high-pass signal, the formula is <22> It holds true, so we can conclude that the Hilbert transform of the product of the low-pass signal and the high-pass signal is equivalent to the product of the Hilbert transform of the low-pass signal and the high-pass signal.
[0147] According to the Hilbert transform condition of the signal product form, there must exist ω2 such that <23> Established:
[0148]
[0149] In the FSI ranging system, due to ω c >>ω b , so there must exist ω2∈(ω b ,ω c ) makes the formula <23> Established. Combined <14> The condition for this to be true is when ω2∈(2ω b ,ω c ), the formula <14> Japanese style <23> At the same time, the Hilbert transform of the interference signal can be expressed as:
[0150]
[0151] In order to facilitate the analysis and calculation of interference signals, based on the above analysis, the Hilbert transform is used to construct the analytical function of the interference signal: It can be expressed as:
[0152]
[0153] Combined <9> , <21> According to the linear properties of Fourier transform, we can get:
[0154]
[0155] Combined <24> , we can get the analytical function of the interference signal
[0156]
[0157] By <27> The phase function of the interference signal can be obtained
[0158]
[0159] According to the properties of inverse trigonometric functions, the phase function obtained by Hilbert transform is The value range of changes repeatedly in the interval [-0.5π, 0.5π]. Therefore, it is necessary to Perform phase unwrapping to obtain a continuously transformed phase function At this time, the phase change of the interference signal I(t) between any two moments t1 and t2 is It can be expressed as:
[0160]
[0161] According to the light wave interference theorem, the phase difference between adjacent peaks of the interference signal is 2π. Therefore, the phase of the interference signal can be decomposed as follows: Figure 4 The integer phase shown Starting fractional phase and the cutoff fractional phase Three parts.
[0162] Integer Phase It can be expressed as:
[0163]
[0164] The integer phase is determined by the peak number δ of the interference signal, and the fractional phase Using HT we can get:
[0165]
[0166] In the above formula, t p1 is the moment when the first peak of the interference signal occurs, t p2 is the moment when the last peak of the interference signal occurs. Therefore, the phase of the interference signal can be expressed as:
[0167]
[0168] The optical frequency variation range Δv and the phase variation of the interference signal Substitute the distance solution formula <7> The measured distance L can be calculated.
[0169] Example 4: Reference Figure 5 and Figure 6 The implementation steps of the method of the present invention are the same as those of Example 1. Now, the nonlinear suppression principle based on CEEMD in the present invention is further described in detail:
[0170] In fact, due to the hysteresis characteristics of PZT in ECDL and the mechanical jitter of the motor end, the interference signal becomes a non-stationary signal with time-varying frequency. Under this influence, it is difficult for HT to directly extract the phase difference of the non-stationary interference signal. Relevant research shows that as the signal quality improves, the ability of HT to analyze the phase change of the signal will also improve. Therefore, it is necessary to process the non-stationary interference signal first.
[0171] As a powerful tool for analyzing non-stationary signals, EMD can adaptively decompose interference signals and reconstruct high-quality interference signals, thereby improving the accuracy of interference signal phase extraction. In terms of signal decomposition, EMD does not rely on basis functions, but adaptively decomposes based on the distribution of extreme points of the signal itself, decomposing complex signals containing multiple frequency components step by step into the form of the sum of multiple single-frequency components. It is suitable for the analysis and processing of complex signals with non-stationary characteristics. The specific process of EMD decomposition is described as follows:
[0172] a) First extract the upper and lower extreme points of the original signal s(t), and use cubic spline to fit the upper and lower extreme points to obtain the upper envelope signal e u (t) and the lower envelope signal e d (t).
[0173] b) Obtain the mean envelope signal m(t) of the upper and lower envelopes, remove m(t) from the initial signal s(t) and obtain the signal c(t).
[0174] c) Consider c(t) as a new s(t) signal and repeat steps ab until the latest decomposed c(t) meets the decomposition termination condition. At this time, c(t) is regarded as the first-order component IMF1.
[0175] d) Remove IMF1 from s(t), and record the remaining signal as r(t). Treat r(t) as the new s(t) signal, and repeat steps ac to obtain the IMF of each order component in turn. i .
[0176] The decomposition form of the non-stationary interference signal can be expressed as:
[0177]
[0178] Where, IMF1, IMF2, …, IMF n It is the various order components obtained by EMD decomposition of non-stationary signals, also known as intrinsic mode function IMF (Intrinsic Modal Function), such as Figure 5 The figure shows the schematic diagram of EMD decomposition and reconstruction of non-stationary interference signals.
[0179] Even though EMD has many advantages in analyzing non-stationary signals, the mutual aliasing between component modes greatly limits the application of EMD in engineering. Ideally, each IMF contains only one characteristic frequency. In fact, in the process of EMD decomposing the signal, different characteristic frequencies are mixed in the same IMF component, or the same characteristic frequency exists in different IMFs. This phenomenon is defined as modal aliasing. Due to the existence of modal aliasing, the two adjacent IMF components obtained by EMD decomposition interfere with each other, and the characteristic frequencies are mixed and cannot be distinguished. Therefore, in the process of decomposing and reconstructing non-stationary signals, it is inevitable that the real components of the signal itself will be eliminated or false components will be introduced into the reconstructed signal, which seriously affects the reconstruction quality of the signal. For example Figure 5 As shown in the IMF4 component, this component is generally regarded as the most important component reflecting the characteristics of the interference signal. Obviously, due to the influence of modal aliasing, other frequency components are mixed into the IMF4 component, causing it to have certain fluctuations in the time domain, which directly affects the reconstructed interference signal I r The quality of (t) affects the extraction accuracy of the interference signal phase.
[0180] In actual engineering applications, most of the signals that need to be analyzed are multi-component signals similar to non-stationary interference signals. In fact, the components that make up the signal have a fixed position distribution in the time-frequency space, and some of the components show "non-stationary" time-frequency characteristics such as intermittent, jumping, and disorder in the time-frequency space. Due to the existence of these non-stationary components, there is a modal aliasing phenomenon when EMD decomposes the interference signal. In order to overcome the modal aliasing problem, the CEEMD algorithm was proposed based on EMD. The CEEMD algorithm first adds pairs of white noise (one positive and one negative) to the original signal, and then performs EMD decomposition separately. In this process, the CEEMD algorithm uses the uniform distribution of white noise in the time-frequency space to combine component signals of different time-frequency scales into IMF. i The white noise background is automatically mapped to the appropriate time-frequency scale related to it, so as to ensure the continuity of each component in the time domain and reduce modal aliasing. In addition, by using the zero mean characteristic of white noise in the time domain, the CEEMD algorithm adds white noise pairs multiple times and averages multiple groups of IMF components of the same order to eliminate the influence of the added white noise on the original signal. The following is the detailed process of decomposition and reconstruction of the CEEMD algorithm.
[0181] First, add paired white noise signals n(t) to the original signal s(t), that is:
[0182]
[0183] The synthesized signals after adding the white noise pair to the original signal s(t) for the jth time are s j1(t) and s j2 (t), respectively for s j1 (t) and s j2 (t) Perform EMD decomposition. Each noise addition decomposition will obtain two sets of n-layer IMF components, namely:
[0184]
[0185] Taking the average of the same-order components of each layer, we can get:
[0186]
[0187] When <10> ~ <12> When j = 1, 2, ..., m, we can get a component matrix with n rows and m columns, that is:
[0188]
[0189] By averaging the components of the same order in each layer of the component matrix, the component matrix of the CEEMD decomposition signal can be obtained, that is:
[0190]
[0191] like Figure 6 The schematic diagram of the decomposition and reconstruction of non-stationary interference signals by the CEEMD algorithm is shown in Figure 1. The components of each order are arranged in descending order according to the frequency. By removing the high-frequency false components and residual components, the remaining real components are reconstructed to obtain a reconstructed interference signal with a high signal-to-noise ratio. r (t).
[0192] Compared with the EMD algorithm decomposing the interference signal, the CEEMD algorithm decomposes more components, the characteristic frequencies of each order component are more clearly distinguished, and the time domain distribution of the main components IMF5 and IMF6 reflecting the characteristics of the interference signal is smoother. This shows that the CEEMD algorithm reduces the mutual interference between adjacent IMFs to a certain extent, the distinction between real components and false components is more obvious, and the CEEMD algorithm effectively suppresses modal aliasing. Figure 5 and Figure 6 Reconstruct the interference signal I r From the quality of (t), it can be seen that the reconstruction quality of the interference signal by the CEEMD algorithm is better than that of the EMD algorithm. Obviously, the CEEMD algorithm will help to improve the phase solution accuracy of the interference signal and improve the ranging accuracy of the FSI system.
[0193] The present invention directly processes the measured interference signal without adding any auxiliary equipment, improves the system ranging accuracy without increasing the system complexity, and is conducive to the promotion and application of FSI technology; the algorithm used in the scheme is an adaptive processing algorithm, which performs adaptive decomposition processing according to the characteristics of the signal itself, so the algorithm itself will not restrict the effect of signal processing and has wide applicability; at the same time, the present invention improves the problem of modal aliasing when decomposing multi-component interference signals, improves the reconstruction quality of non-stationary interference signals, and significantly improves the ranging accuracy of the system.
[0194] Parts of the present invention that are not described in detail belong to common knowledge among those skilled in the art.
[0195] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Obviously, for professionals in this field, after understanding the content and principles of the present invention, they may make various modifications and changes in form and details without departing from the principles and structures of the present invention. However, these modifications and changes based on the ideas of the present invention are still within the scope of protection of the claims of the present invention.
Claims
1. A high-precision optical frequency scanning interferometry (FSI) ranging method based on the supplementary population average empirical mode decomposition (CEEMD) algorithm, characterized in that: The steps include: (1) Using a tunable external cavity semiconductor laser ECDL, a collimator, a beam splitter, a Fabry-Perot etalon, a photodetector, and a Michelson interferometer optical path to build an FSI ranging system, and equip the system with a data acquisition device; (2) The tunable external cavity semiconductor laser ECDL of the FSI ranging system generates a tunable light source S whose frequency changes with time t and outputs it to the collimator; (3) After the collimator collimates the tuned light source S, it is divided into two light beams S1 and S2 using a beam splitter prism, and S1 enters the Fabry-Perot etalon, and S2 enters the Michelson interferometer optical path; (4) The Fabry-Perot FP etalon obtains the transmission signal of S1, and uses a photodetector to detect the transmission signal to obtain the FP signal; the photodetector in the Michelson interference optical path detects S2 that has entered the optical path and has undergone splitting, reflection, and convergence interference to obtain an interference signal; (5) Synchronously sample the FP signal and the interference signal obtained in step (4) using a data acquisition device, and preprocess them: (5.1) The tuning range Δv of the laser ECDL output tuning light source is calculated based on the number of resonance peaks of the FP signal: Δv=(r-1)·FSR, Where r represents the number of resonance peaks of the FP signal, and FSR represents the optical frequency difference between two adjacent resonance peaks; (5.2) The interference signal is intercepted by using the time when the resonance peaks of the first and last two FP signals in the sampling interval are located; (5.3) dividing the interference signal intercepted in step (5.2) according to the period, into an integer period interference signal and a fractional period interference signal, and further dividing the fractional period interference signal to obtain a starting fractional period interference signal and an ending fractional period interference signal; (6) The fractional period interference signal is processed using the CEEMD algorithm, that is, the starting fractional period interference signal and the ending fractional period interference signal are decomposed, and the decomposed components are arranged in order from high to low frequency, and the high-frequency noise component is removed according to the arrangement result, and the real stationary component reflecting the characteristics of the interference signal is retained, and then the retained real stationary component is superimposed and reconstructed, and finally a reconstructed fractional period interference signal with a high signal-to-noise ratio is obtained; (7) Using Hilbert transform to solve the phase difference of the reconstructed fractional period interference signal The implementation steps are as follows: (7.1) Perform Hilbert transform on the starting fractional period interference signal and the ending fractional period interference signal obtained in step (5.3), take the Hilbert transformed signal as the imaginary part and the original signal as the real part, and construct the analytical function of the fractional period interference signal: Where i represents an imaginary number; b(t) represents the intensity amplitude function of the coherent term of the fractional period interference signal; Represents the instantaneous phase function of the fractional periodic interference signal; (7.2) Calculate the instantaneous phase value of the fractional period interference signal: (7.3) Obtain multiple instantaneous phase values corresponding to different times, and perform phase unwrapping on them to obtain a continuous instantaneous phase function that changes with time Then, by calculating the instantaneous phase values corresponding to the first and last moments t1 and t2 and Make a difference and get the phase difference of the fractional period interference signal to be calculated (8) The total phase difference Δφ is obtained by the following steps: (8.1) Using the peak number of the integer-period interference signal, the phase difference Δφ1 of the integer-period interference signal is obtained according to the following formula: Δφ1=(δ-1)·2π Among them, δ represents the number of peaks; (8.2) Use the solution method of the fractional period interference signal obtained in step (7.3) to solve the phase difference corresponding to the starting fractional period interference signal and the ending fractional period interference signal, and obtain the starting fractional phase difference Δφ s The phase difference Δφ with the ending decimal e; (8.3) to obtain the total phase difference Δφ: Δφ=Δφ1+Δφ s +Df e ; (9) Calculate the distance L to be measured according to the following formula to complete the distance measurement: Where n is the refractive index of air and c is the speed of light.
2. The method according to claim 1, characterized in that: In step (1), there are two beam splitter prisms, namely a first beam splitter prism BS1 and a second beam splitter prism BS2; wherein the first beam splitter prism BS1 is used to split the tuned light source after collimation by the collimator, and the second beam splitter prism BS2 is located in the Michelson interference light path and is used to split the light beam entering the Michelson interference light path again.
3. The method according to claim 1, characterized in that: The splitting, reflection and convergence interference in step (4) are as follows: after S2 enters the Michelson interference optical path, it is split into two new beams S2 by the second beam splitter prism BS2. 21 and S 22 , where S 21 As the reference light of the ranging optical path, S 22 The two are respectively reflected by the corner cone reflection prisms arranged at different positions, and then converge and interfere at the second beam splitter prism BS2.
4. The method according to claim 1, characterized in that: The interference signal in step (4) is expressed as follows according to the light wave interference theorem: Among them, I r (t) represents the intensity of the reference light, I m (t-τ) represents the intensity of the measured light, τ is the delay of the measured light relative to the reference light in the time domain; v0 represents the initial optical frequency output by the laser, and β(t) represents the rate of change of the optical frequency; the above expression is simplified as follows: I(t)=a(t)+b(t)cosθ(t), Among them, a(t)=I r (t)+I m (t-τ) represents the light intensity amplitude function of the low-frequency DC term of the interference signal; A function of the intensity amplitude representing the coherence term of the interference signal.
5. The method according to claim 1, characterized in that: The optical frequency difference FSR corresponding to the two adjacent resonance peaks described in step (5.1) is 1.5 GHz.
6. The method according to claim 1, characterized in that: The intercepted interference signal is decomposed by the supplementary population average empirical mode decomposition (CEEMD) algorithm described in step (6), specifically: a positive and a negative white noise pair is added to the original signal successively, and the component signals of different time-frequency scales are automatically mapped to the time-frequency scale related thereto through the white noise background by utilizing the characteristic that the white noise is evenly distributed in the time-frequency space, and then the empirical mode decomposition is performed respectively to obtain multiple intrinsic mode function (IMF) components; By using the zero mean characteristic of white noise in the time domain, white noise pairs are added multiple times and empirical mode decomposition is performed. The multiple groups of IMF components of the same order are averaged to obtain the final component.
7. The method according to claim 1, characterized in that: The Hilbert transform in step (7) is expressed as follows for any signal x(t): in, and denote Fourier transform and inverse Fourier transform respectively, and H[ω] is the transfer function of the Hilbert change.
