A method for correcting distortion of Brillouin signal spectrum distribution based on spatial adaptive deconvolution

Abstract

The application relates to the technical field of optical fiber sensing, and discloses a Brillouin signal spectrum distribution distortion correction method based on spatial adaptive deconvolution, which comprises the following steps: when a Brillouin optical time domain reflection system is in a calibration mode, an initial response function is acquired, and a spatially-varying point spread function of a sensing optical fiber containing self-phase modulation is constructed in combination with the parameters of the sensing optical fiber in the Brillouin optical time domain reflection system; singular value decomposition is used to perform low-rank decomposition on the constructed spatially-varying point spread function of the sensing optical fiber, so that a forward convolution operator is determined; when the Brillouin time domain reflection system is switched to a sensing mode, a Brillouin gain spectrum distribution image of the sensing optical fiber affected by self-phase modulation is acquired, is input into an original-dual gradient algorithm, and is iteratively solved in combination with the determined forward convolution operator, so that a Brillouin gain spectrum distribution image subjected to distortion correction is obtained. The application realizes efficient and high-precision reconstruction of a severely-distorted BGS distribution and restores the sensing capability of the end of the sensing optical fiber.

Description

Technical Field

[0001] This invention relates to the field of fiber optic sensing technology, and more specifically, to a method for correcting Brillouin signal spectral distribution distortion based on spatial adaptive deconvolution. Background Technology

[0002] With the expansion of global infrastructure construction, the demand for real-time monitoring of large-scale equipment is increasing. Distributed fiber optic sensing technology, capable of simultaneously transmitting information and sensing the environment through a single optical fiber, has been widely applied in numerous industries. Among them, Brillouin Optical Time-Domain Reflectometry (BOTDR) technology utilizes the spontaneous Brillouin scattering signal generated by single-ended optical signal injection to demodulate the linear relationship between the Brillouin Frequency Shift (BFS) and the measured quantity, thereby achieving distributed physical quantity measurement.

[0003] However, because BOTDR systems rely on spontaneous Brillouin scattering signals, the scattered signal intensity is weak, resulting in a low signal-to-noise ratio, which limits the sensing distance and accuracy of BOTDR. The most direct way to improve the signal-to-noise ratio is to increase the power of the probe pulse, but this introduces nonlinear effects, leading to pulse spectral broadening and energy redistribution, which severely degrades signal quality or even makes demodulation impossible.

[0004] To address these issues, existing technologies typically employ hardware improvements such as distributed fiber amplification, pulse coding, or wavelength diversity, but these increase system complexity and cost. In recent years, software post-processing schemes based on image denoising and machine learning have attracted significant attention due to their lack of the need for additional hardware. For example, some studies treat the measured Brillouin Gain Spectrum (BGS) distribution as an image blurred by the point spread function (PSF), reconstructing the BGS through image deconvolution. However, existing deconvolution methods generally assume that the PSF is spatially invariant throughout the fiber. Practical research shows that when BOTDR systems use high-power pulses with wide rise or fall edges, self-phase modulation (SPM) introduces distance-dependent cumulative nonlinear phase shifts, causing dynamic spatial variations in the PSF. In this case, traditional deconvolution methods based on spatially invariant PSFs fail, unable to repair severe BGS distribution distortions caused by SPM, such as bimodal splitting, leading to a significant decrease in measurement accuracy.

[0005] Therefore, there is an urgent need for a BGS distortion correction method that can effectively adapt to the spatial variation characteristics of the point spread function and accurately correct the strong SPM effect-induced distortion without changing the existing system architecture. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention proposes a Brillouin signal spectrum distribution distortion correction method based on spatial adaptive deconvolution. It utilizes singular value decomposition to perform a low-rank approximation of the spatially varying point spread function, and combines it with the hybrid regularized primal-dual gradient algorithm. Without increasing the hardware architecture complexity of the BOTDR system, it achieves efficient and high-precision reconstruction of severely distorted BGS distributions, restoring the sensing capability of the sensing fiber end.

[0007] To achieve the above technical objectives, the present invention adopts the following technical solution:

[0008] A Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution includes the following steps: Step S1: When the Brillouin optical time-domain reflectometry system is in calibration mode, obtain the initial response function of the Brillouin optical time-domain reflectometry system; Step S2: Based on the obtained initial response function and the parameters of the sensing fiber in the Brillouin optical time-domain reflectometry system, construct the spatial variation point spread function of the sensing fiber, which includes self-phase modulation-induced variation. Step S3: Use singular value decomposition to perform low-rank decomposition on the spatial variation point spread function of the constructed sensing fiber to determine the forward convolution operator. Step S4: When the Brillouin time-domain reflectometry system is switched to sensing mode, acquire the Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation. Step S5: Input the obtained Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation into the original-dual gradient algorithm, and iteratively solve the distortion-corrected Brillouin gain spectrum distribution image by combining the determined forward convolution operator.

[0009] Further, step S2 includes the following sub-steps: Step S2.1: Based on the obtained initial response function and the parameters of the sensing fiber in the Brillouin optical time-domain reflectometry system, combined with the nonlinear phase shift induced by self-phase modulation during the transmission of the probe pulse in the sensing fiber, calculate the pulse envelope function that evolves with the length of the sensing fiber. Step S2.2: Combine the pulse envelope function of the sensing fiber with the window function used in the short-time Fourier transform to determine the spatial variation point spread function of the sensing fiber.

[0010] Furthermore, the calculation process of the pulse envelope function of the sensing optical fiber is as follows:

[0011] in, Denotes the initial response function of the Brillouin optical time-domain reflectometry system. Indicates the sensing fiber in the time domain The pulse envelope function under the following conditions This indicates the attenuation value of the sensing fiber. Indicates the length of the sensing fiber. express The additional nonlinear phase induced by self-phase modulation during propagation It represents the imaginary unit.

[0012] Furthermore, the process for determining the spatial variation point spread function of the sensing optical fiber is as follows:

[0013] in, Indicates the sensing fiber in the time domain With frequency domain The two-dimensional spatial variation point spread function, This represents the integration variables used in the integration process. Represented in the time domain It is the complex conjugate of the window function after center translation.

[0014] Further, the specific process of step S3 is as follows: the spatial variation point spread function of the constructed sensing fiber is decomposed into a weighted result of the mean kernel and the feature kernel through singular value decomposition. The singular values ​​of the feature kernel are arranged in descending order, and the feature kernels with the first few singular values ​​exceeding the preset threshold are extracted as forward convolution operators.

[0015] Furthermore, step S5 includes the following sub-steps: Step S5.1: Use the Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation as the initial input of the primal-dual gradient algorithm, initialize the set of dual variable matrices, and construct the hybrid regularization objective function. Step S5.2: Predict the Brillouin gain spectrum distribution image using the primal-dual gradient algorithm, and deconvolve the predicted Brillouin gain spectrum distribution image using the forward convolution operator; Step S5.3: Calculate the value of the hybrid regularization objective function and update the dual variable matrix set based on the predicted Brillouin gain spectrum distribution image, the deconvolutioned Brillouin gain spectrum distribution image, and the obtained Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation. Step S5.4: Calculate the update direction of the Brillouin gain spectrum distribution image based on the updated dual variable matrix set, and update the predicted Brillouin gain spectrum distribution image; Step S5.5: Repeat steps S5.2-S5.4 with the updated Brillouin gain spectrum distribution image until the mixing regularization objective function converges; Step S5.6: Use the predicted Brillouin gain spectrum distribution image that meets the convergence condition as the distortion-corrected Brillouin gain spectrum distribution image.

[0016] Furthermore, the construction process of the hybrid regularization objective function is as follows:

[0017] in, Indicates the first The Brillouin gain spectrum distribution image predicted in the next iteration. This represents the forward convolution operator. This represents the Brillouin gain spectrum distribution image obtained from the sensing fiber under the influence of self-phase modulation. This represents the total variation term in the time dimension for suppressing noise along the length of the sensing fiber. express The regularization parameter, The total variation term representing the frequency dimension of the smoothed Brillouin gain spectral distribution image. express The regularization parameter, This represents the frequency dimension Tikhonov regularization term. express The regularization parameter.

[0018] Furthermore, the set of dual variable matrices includes: data fidelity term dual variables, time domain total variation dual variables, and frequency domain total variation dual variables.

[0019] Furthermore, the update process of the dual variable matrix set is as follows:

[0020]

[0021]

[0022] in, Indicates the first The data fidelity term dual variable in the next iteration Indicates the step size. Indicates the weight of data fidelity items. Indicates the first The time-domain total variational dual variable under the next iteration Indicates along the time domain The first-order difference operator, This indicates an element-wise projection operation. Show the first Frequency domain total variation dual variable under the next iteration This represents the first-order difference operator along the frequency domain.

[0023] Furthermore, the calculation process for the update direction of the Brillouin gain spectrum distribution image is as follows:

[0024] in, Indicates the first The update direction of the Brillouin gain spectrum distribution image in the next iteration. express The adjoint operator, express The adjoint operator, express The adjoint operator.

[0025] Compared with the prior art, the present invention has the following beneficial effects: (1) The Brillouin signal spectrum distribution distortion correction method based on spatial adaptive deconvolution of the present invention uses singular value decomposition technology to decompose the spatially varying point spread function that continuously evolves with the length of the sensing fiber into a linear combination of a very small number of feature kernels and smooth weight fields, thereby realizing a compact parameterized expression of spatial variation characteristics. While ensuring the accuracy of the point spread function, it avoids the explicit construction of large-scale convolution operators and significantly improves the post-processing efficiency of long-distance distributed sensing data. (2) The Brillouin signal spectrum distribution distortion correction method based on spatial adaptive deconvolution of this invention adopts a framework of hybrid regularization objective function and primal-dual gradient algorithm for iterative solution. The primal-dual gradient algorithm can reconstruct the Brillouin gain spectrum distribution image under the harsh condition of SNR as low as 10-15dB by monitoring the gap of dual variables without matrix inversion, thus ensuring the reliability of BOTDR system in extreme environments. At the same time, random noise is suppressed and spatial abrupt change edges are maintained by the total variation term of time axis in the hybrid regularization objective function, and high-frequency artifacts common in deconvolution are eliminated by combining the frequency dimension Tikhonov term, which effectively prevents the staircase effect and overfitting phenomenon.

[0026] In summary, this invention combines singular value decomposition of the spatially varying point spread function with a hybrid regularized primal-dual gradient algorithm, which can restore a severely distorted bimodal Brillouin gain spectrum distribution image to an interpretable unimodal Brillouin gain spectrum distribution image, thus achieving distortion correction. Attached Figure Description

[0027] Figure 1 This is an overall flowchart of the Brillouin signal spectrum distribution distortion correction method based on spatial adaptive deconvolution of the present invention; Figure 2 This is a schematic diagram of the BOTDR system structure used in this invention; Figure 3This is a simulation verification diagram of BGS distribution distortion correction in an embodiment of the present invention, wherein, Figure 3 In the diagram, (a) represents the simulated intrinsic BGS distribution. Figure 3 (b) in the figure represents the simulated BGS distribution unaffected by SPM. Figure 3 In the diagram, (c) represents the simulated BGS distribution affected by SPM. Figure 3 In the figure, (d) represents the simulated BGS distribution affected by SPM, corrected by the present invention. Figure 4 This is a diagram illustrating the experimental verification effect of BGS distribution distortion correction in an embodiment of the present invention. Figure 4 In the diagram, (a) represents the BGS distribution unaffected by SPM. Figure 4 (b) in the figure represents the BGS distribution affected by SPM. Figure 4 (c) in the figure represents the BGS distribution after correction by the present invention, which is affected by SPM; Figure 5 This is a measurement error analysis diagram of the Brillouin frequency shift in an embodiment of the present invention, wherein, Figure 5 In the figure, (a) represents the BFS curves with and without SPM, after correction, and the ideal BFS curve. Figure 5 In the figure, (b) represents the absolute error curves relative to the ideal BFS with and without SPM and after correction. Detailed Implementation

[0028] The technical solution of the present invention will be further explained and described below with reference to the accompanying drawings.

[0029] like Figure 1 This invention provides a method for correcting Brillouin signal spectral distribution distortion based on spatial adaptive deconvolution, comprising the following steps: Step S1: When the optical switch in the BOTDR system is switched to the calibration branch, the probe light returns to the circulator via the reflector. The adjustable attenuator is adjusted to ensure that the reflected light does not exceed the range of the photodetector. The photodetector converts the optical signal into an electrical signal and sends it to the data acquisition card for demodulation to obtain the initial response function of the BOTDR system, which includes the effects of light source pulses, device responses, etc.

[0030] Step S2: Based on the obtained initial response function and the parameters of the sensing fiber in the Brillouin optical time-domain reflectometry system, construct the spatial variation point spread function of the sensing fiber, including self-phase modulation-induced diffusion; this includes the following sub-steps: Step S2.1: Based on the obtained initial response function and the parameters of the sensing fiber in the Brillouin optical time-domain reflectometry system, and combined with the nonlinear phase shift induced by self-phase modulation during the transmission of the probe pulse in the sensing fiber, calculate the pulse envelope function that evolves with the length of the sensing fiber:

[0031] in, Denotes the initial response function of the Brillouin optical time-domain reflectometry system. Indicates the sensing fiber in the time domain The pulse envelope function under the following conditions This indicates the attenuation value of the sensing fiber. Indicates the length of the sensing fiber. express The additional nonlinear phase induced by self-phase modulation during propagation depends on the nonlinear coefficient of the light, the peak power of the pulse, and the effective core area of ​​the sensing fiber. The SPM effect causes instantaneous frequency offsets that vary with time on the rising and falling edges of the pulse. It represents the imaginary unit.

[0032] Step S2.2: Combine the pulse envelope function of the sensing fiber with the window function used in the short-time Fourier transform to determine the spatial variation point spread function of the sensing fiber:

[0033] in, Indicates the sensing fiber in the time domain With frequency domain The two-dimensional spatial variation point spread function, This represents the integration variables used in the integration process. Represented in the time domain It is the complex conjugate of the window function after center translation.

[0034] Step S3: To simplify computation, singular value decomposition (SVD) is used to perform low-rank decomposition on the spatial variation point spread function of the constructed sensing fiber, determining the forward convolution operator. SVD transforms the complex spatial variation convolution of the continuously evolving spatial variation point spread function with the length of the sensing fiber into a weighted superposition of a small number of spatially invariant convolutions, achieving a compact parameterized expression of spatial variation characteristics. While ensuring the accuracy of the point spread function, this avoids explicitly constructing a large-scale convolution operator, significantly improving the post-processing efficiency of long-distance distributed sensing data and reducing costs. Specifically: The spatial variation point spread function of the constructed sensing fiber is expressed as a weighted result of the mean kernel and the characteristic kernel through singular value decomposition. ,in, Indicates the mean kernel. Indicates the number of feature kernels, Indicates the first Each feature kernel, Indicates the first The weight coefficients of each feature kernel; The singular values ​​of the feature kernels are arranged in descending order, and the feature kernels with the highest proportion of singular values ​​exceeding a preset threshold are extracted and used as forward convolution operators.

[0035] Step S4: Switch the optical switch in the BOTDR system to the sensing branch of the sensing fiber under test, and obtain the Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation. At the end of the sensing fiber, the BGS distribution is broadened and deformed due to the strong SPM effect, and the original distribution acquired shows severe bimodal distortion.

[0036] Step S5: Input the Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation into the original-dual gradient algorithm, and iteratively solve the distortion-corrected Brillouin gain spectrum distribution image by combining the determined forward convolution operator. This invention adopts a framework of hybrid regularization objective function and original-dual gradient algorithm for iterative solution. The original-dual gradient algorithm can reconstruct the Brillouin gain spectrum distribution image under the harsh condition of SNR as low as 10-15dB by monitoring the gap of dual variables without matrix inversion, thus ensuring the reliability of the BOTDR system in extreme environments.

[0037] Step S5 includes the following sub-steps: Step S5.1: Use the Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation as the initial input of the original-dual gradient algorithm, initialize the set of dual variable matrices, and construct the hybrid regularization objective function. The set of dual variable matrices includes: data fidelity term dual variable, time domain total variation dual variable, and frequency domain total variation dual variable, all of which are initialized to zero matrices of the same dimension as the initial input. Step S5.2: Predict the Brillouin gain spectrum distribution image using the primal-dual gradient algorithm, and deconvolve the predicted Brillouin gain spectrum distribution image using the forward convolution operator; Step S5.3: Calculate the value of the hybrid regularization objective function and update the dual variable matrix set based on the predicted Brillouin gain spectrum distribution image, the deconvolutioned Brillouin gain spectrum distribution image, and the obtained Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation. Step S5.4: Calculate the update direction of the Brillouin gain spectrum distribution image based on the updated dual variable matrix set, and update the predicted Brillouin gain spectrum distribution image; Step S5.5: Repeat steps S5.2-S5.4 with the updated Brillouin gain spectrum distribution image until the mixing regularization objective function converges; Step S5.6: Use the predicted Brillouin gain spectrum distribution image that meets the convergence condition as the distortion-corrected Brillouin gain spectrum distribution image.

[0038] This invention combines singular value decomposition of the spatial variation point spread function with a hybrid regularized primal-dual gradient algorithm, which can restore a severely distorted bimodal Brillouin gain spectrum distribution image to an interpretable unimodal Brillouin gain spectrum distribution image, thus achieving distortion correction.

[0039] In one technical solution of this invention, the hybrid regularization objective function introduces a time-dimensional total variation term to suppress noise along the length of the sensing fiber, a frequency-dimensional total variation term to smooth the Brillouin gain spectrum distribution image, and a frequency-dimensional Tikhonov regularization term. The time-dimensional total variation term to suppress noise along the length of the sensing fiber is used to suppress random noise and preserve spatial abrupt change edges. The frequency-dimensional Tikhonov regularization term is used to alleviate the staircase effect caused by the total variation and enhance the spectral physical line shape, eliminating high-frequency artifacts commonly encountered during deconvolution, effectively preventing staircase effects and overfitting. The construction process of the hybrid regularization objective function in this invention is as follows:

[0040] in, Indicates the first The Brillouin gain spectrum distribution image predicted in the next iteration. This represents the forward convolution operator. This represents the Brillouin gain spectrum distribution image obtained from the sensing fiber under the influence of self-phase modulation. This represents the total variation term in the time dimension for suppressing noise along the length of the sensing fiber. express The regularization parameter, The total variation term representing the frequency dimension of the smoothed Brillouin gain spectral distribution image. express The regularization parameter, This represents the frequency dimension Tikhonov regularization term. express The regularization parameter.

[0041] In one technical solution of the present invention, the update process of the dual variable matrix set is as follows:

[0042]

[0043]

[0044] in, Indicates the first The data fidelity term dual variable in the next iteration Indicates the step size. Indicates the weight of data fidelity items. Indicates the first The time-domain total variational dual variable under the next iteration Indicates along the time domain The first-order difference operator, This represents an element-wise projection operation, used to restrict the dual variable to a range determined by the regularization parameter. Show the first Frequency domain total variation dual variable under the next iteration This represents the first-order difference operator along the frequency domain.

[0045] In one technical solution of the present invention, the calculation process of the update direction of the Brillouin gain spectrum distribution image is as follows:

[0046] in, Indicates the first The update direction of the Brillouin gain spectrum distribution image in the next iteration. express The adjoint operator, express The adjoint operator, express The adjoint operator.

[0047] In one technical solution of the present invention, since the point spread function changes slowly along the length of the sensing fiber, the spatial axis of the sensing fiber can be divided into several local processing units of equal length. Within each local processing unit, the point spread function is assumed to be spatially invariant, and the weight coefficients of the low-rank decomposition are shared to accelerate the solution. Within each local processing unit, the weight coefficients of the low-rank decomposition can be approximated as the average value or center value within that unit. This processing method allows multiple sampling rows within the same local unit to share the same set of low-rank point spread function weight coefficients, avoiding the repeated construction of forward convolution operators for each sampling position, thereby improving the iterative solution speed.

[0048] Example To further verify the effectiveness of the Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution in the BOTDR system, this embodiment constructs a BOTDR system, as follows: Figure 2A continuous-wave laser with a center wavelength of 1550 nm is used as the light source. The continuous-wave laser is split into upper and lower paths via a 50:50 fiber coupler. The upper path is the probe light branch, and the lower path is the local oscillator light branch. In the probe light branch, the continuous-wave light is first pulse-modulated by a semiconductor optical amplifier and a first electro-optic modulator to form a probe pulse for injection into the sensing fiber. The time-domain waveform of the probe pulse is precisely controlled by an arbitrary waveform generator. Subsequently, the probe pulse is amplified by an erbium-doped fiber amplifier and filtered by a first optical bandpass filter to remove spontaneous emission noise introduced during amplification. The filtered probe pulse is injected into the sensing fiber through an optical circulator, generating a backscattered Brillouin signal within the sensing fiber. In the local oscillator light branch, the continuous-wave light is modulated by a second electro-optic modulator and filtered by a second optical bandpass filter to obtain a local oscillator light of approximately 10.3 GHz. The local oscillator light is used for coherent beat frequency detection with the Brillouin scattered light returned from the sensing fiber. To reduce the impact of polarization state changes on the detection results, a polarization scrambler is installed in the BOTDR system to reduce the influence of polarization fading on the Brillouin gain spectrum measurement.

[0049] To rigorously compare the Brillouin gain spectrum distribution under conditions of self-phase modulation and those without significant self-phase modulation, this embodiment maintains the optical path structure, receiving link, and data acquisition method unchanged, only altering the time-domain waveform of the probe pulse using an arbitrary waveform generator to obtain two sets of experimental data. First, the arbitrary waveform generator outputs a rectangular pulse with a pulse width of 25 ns, causing the probe optical branch to generate a 25 ns rectangular probe pulse. When using this pulse for BOTDR measurement, a Brillouin gain spectrum distribution with a single peak is obtained at the end of the sensing fiber. The measured Brillouin gain spectrum does not show significant broadening, shoulder peaks, or peak distortion in the distance-frequency two-dimensional distribution. Therefore, this measurement result is used as the reference Brillouin gain spectrum data without significant self-phase modulation. Subsequently, while maintaining the same probe pulse peak power, the output waveform of the arbitrary waveform generator is switched to a Gaussian pulse, causing the probe optical branch to generate a Gaussian probe pulse with a full width at half maximum (FWHM) of 25 ns. Because Gaussian pulses have a continuously varying power envelope in the time domain, they introduce more significant nonlinear phase modulation when propagating in sensing optical fibers, causing the point spread function of the returned Brillouin scattering signal to change with the fiber distance.

[0050] Figure 3 This represents the simulated BGS distribution under the same real-world experimental environment, where... Figure 3 (a) shows the simulated intrinsic BGS distribution. Figure 3 (b) shows the simulated BGS distribution without considering the SPM effect. Figure 3 (c) shows that the simulated BGS distribution is significantly distorted when the SPM effect is considered. Figure 3Figure (d) shows that, considering the SPM effect, the simulated BGS distribution, after being corrected by the Brillouin signal spectrum distribution distortion correction method based on spatial adaptive deconvolution of this invention, has been restored from the original bimodal structure to a single peak, and its shape is similar to that of the BGS distribution unaffected by SPM.

[0051] The spatial adaptive deconvolutional Brillouin spectrum correction performed using the method of this invention yields the following experimental BGS distribution results: Figure 4 As shown: Figure 4 In (b), the distribution affected by SPM distortion clearly shows a bimodal distribution in the frequency domain, while it exhibits a skewed distribution in the time domain. Figure 4 (c) shows the BGS distribution reconstructed using the distortion correction method of this invention and Figure 4 The BGS distributions in (a) that are not affected by SPM are very similar: the bimodal distribution of BGS has converged to a unimodal distribution, and the tilting phenomenon has been significantly suppressed.

[0052] Figure 5 The diagram shows the measurement error analysis of the Brillouin frequency shift, by... Figure 5 As can be seen in (a), the original Brillouin gain spectrum, which is severely affected by self-phase modulation, cannot be reliably demodulated. The extracted Brillouin frequency shift distribution exhibits significant fluctuations and deviates considerably from the ideal Brillouin frequency shift distribution. To quantitatively evaluate the demodulation accuracy, this embodiment calculates the standard deviation of the Brillouin frequency shift and the mean absolute error for three different cases. The calculation results are as follows: Figure 5 In (b), the standard deviation of the BFS curve reconstructed by the spatial adaptive deconvolution-based Brillouin signal spectrum distortion correction method of this invention is 0.748 MHz, and the mean absolute error is 1.088 MHz. In contrast, the standard deviation of the rectangular pulse reference without significant self-phase modulation is 0.977 MHz, and the mean absolute error is 1.129 MHz; while the standard deviation of the Brillouin frequency shift in the original Gaussian pulse measurement results affected by self-phase modulation is 5.172 MHz, and the mean absolute error is 5.304 MHz, accompanied by significant random fluctuations, making it difficult to use as a reliable sensing result. These results demonstrate that after adopting the distortion correction method of this invention, the demodulation accuracy of the Brillouin gain spectrum affected by self-phase modulation is significantly improved, and its Brillouin frequency shift extraction accuracy is close to that of the reference case without significant self-phase modulation. Therefore, this invention can effectively restore the Brillouin gain spectrum shape and the temperature or strain sensing capability of the BOTDR system when there is spectral distortion caused by self-phase modulation.

[0053] In one technical solution of the present invention, a computer-readable storage medium is also provided, storing a computer program that enables a computer to execute the Brillouin signal spectrum distribution distortion correction method based on spatial adaptive deconvolution of the present invention.

[0054] In one technical solution of the present invention, an electronic device is also provided, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the Brillouin signal spectrum distribution distortion correction method based on spatial adaptive deconvolution of the present invention.

[0055] In the embodiments disclosed in this application, a computer storage medium may be a tangible medium that may contain or store programs for use by or in conjunction with an instruction execution system, apparatus, or device. The computer storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of computer storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, and portable compact disc read-only memory (CD). ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0056] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0057] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A method for correcting Brillouin signal spectral distribution distortion based on spatial adaptive deconvolution, characterized in that, Includes the following steps: Step S1: When the Brillouin optical time-domain reflectometry system is in calibration mode, obtain the initial response function of the Brillouin optical time-domain reflectometry system; Step S2: Based on the obtained initial response function and the parameters of the sensing fiber in the Brillouin optical time-domain reflectometry system, construct the spatial variation point spread function of the sensing fiber, which includes self-phase modulation-induced variation. Step S3: Use singular value decomposition to perform low-rank decomposition on the spatial variation point spread function of the constructed sensing fiber to determine the forward convolution operator. Step S4: When the Brillouin time-domain reflectometry system is switched to sensing mode, acquire the Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation. Step S5: Input the obtained Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation into the original-dual gradient algorithm, and iteratively solve the distortion-corrected Brillouin gain spectrum distribution image by combining the determined forward convolution operator.

2. The Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution according to claim 1, characterized in that, Step S2 includes the following sub-steps: Step S2.1: Based on the obtained initial response function and the parameters of the sensing fiber in the Brillouin optical time-domain reflectometry system, combined with the nonlinear phase shift induced by self-phase modulation during the transmission of the probe pulse in the sensing fiber, calculate the pulse envelope function that evolves with the length of the sensing fiber. Step S2.2: Combine the pulse envelope function of the sensing fiber with the window function used in the short-time Fourier transform to determine the spatial variation point spread function of the sensing fiber.

3. The Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution according to claim 2, characterized in that, The calculation process for the pulse envelope function of the sensing optical fiber is as follows: in, Denotes the initial response function of the Brillouin optical time-domain reflectometry system. Indicates the sensing fiber in the time domain The pulse envelope function under the following conditions This indicates the attenuation value of the sensing fiber. Indicates the length of the sensing fiber. express The additional nonlinear phase induced by self-phase modulation during propagation It represents the imaginary unit.

4. The Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution according to claim 3, characterized in that, The process for determining the spatial variation point spread function of the sensing optical fiber is as follows: in, Indicates the sensing fiber in the time domain With frequency domain The two-dimensional spatial variation point spread function, This represents the integration variables used in the integration process. Represented in the time domain It is the complex conjugate of the window function after center translation.

5. The Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution according to claim 1, characterized in that, The specific process of step S3 is as follows: the spatial variation point spread function of the constructed sensing fiber is decomposed into a weighted result of mean kernel and feature kernel through singular value decomposition. The singular values ​​of the feature kernel are arranged in descending order, and the feature kernels with the first few singular values ​​exceeding the preset threshold are extracted as forward convolution operators.

6. The Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution according to claim 1, characterized in that, Step S5 includes the following sub-steps: Step S5.1: Use the Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation as the initial input of the primal-dual gradient algorithm, initialize the set of dual variable matrices, and construct the hybrid regularization objective function. Step S5.2: Predict the Brillouin gain spectrum distribution image using the primal-dual gradient algorithm, and deconvolve the predicted Brillouin gain spectrum distribution image using the forward convolution operator; Step S5.3: Calculate the value of the hybrid regularization objective function and update the dual variable matrix set based on the predicted Brillouin gain spectrum distribution image, the deconvolutioned Brillouin gain spectrum distribution image, and the obtained Brillouin gain spectrum distribution image of the sensing fiber affected by self-phase modulation. Step S5.4: Calculate the update direction of the Brillouin gain spectrum distribution image based on the updated dual variable matrix set, and update the predicted Brillouin gain spectrum distribution image; Step S5.5: Repeat steps S5.2-S5.4 with the updated Brillouin gain spectrum distribution image until the mixing regularization objective function converges; Step S5.6: Use the predicted Brillouin gain spectrum distribution image that meets the convergence condition as the distortion-corrected Brillouin gain spectrum distribution image.

7. The Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution according to claim 6, characterized in that, The process of constructing the hybrid regularization objective function is as follows: in, Indicates the first The Brillouin gain spectrum distribution image predicted in the next iteration. This represents the forward convolution operator. This represents the Brillouin gain spectrum distribution image obtained from the sensing fiber under the influence of self-phase modulation. This represents the total variation term in the time dimension for suppressing noise along the length of the sensing fiber. express The regularization parameter, The total variation term representing the frequency dimension of the smoothed Brillouin gain spectral distribution image. express The regularization parameter, This represents the frequency dimension Tikhonov regularization term. express The regularization parameter.

8. The Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution according to claim 7, characterized in that, The set of dual variable matrices includes: data fidelity term dual variables, time domain total variation dual variables, and frequency domain total variation dual variables.

9. The Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution according to claim 8, characterized in that, The update process for the set of dual variable matrices is as follows: in, Indicates the first The data fidelity term dual variable in the next iteration Indicates the step size. Indicates the weight of data fidelity items. Indicates the first The time-domain total variational dual variable under the next iteration Indicates along the time domain The first-order difference operator, This indicates an element-wise projection operation. Show the first Frequency domain total variation dual variable under the next iteration This represents the first-order difference operator along the frequency domain.

10. The Brillouin signal spectral distribution distortion correction method based on spatial adaptive deconvolution according to claim 9, characterized in that, The calculation process for the update direction of the Brillouin gain spectrum distribution image is as follows: in, Indicates the first The update direction of the Brillouin gain spectrum distribution image in the next iteration. express The adjoint operator, express The adjoint operator, express The adjoint operator.

Images