Method and system for extracting sensing parameters of brillouin fiber with spatial resolution enhancement

Abstract

The application belongs to the field of optical fiber parameter processing, and provides a Brillouin optical fiber sensing parameter extraction method and system with enhanced spatial resolution, which acquires a low spatial resolution Brillouin gain spectrum image, processes the image to obtain an initial high spatial resolution Brillouin gain spectrum estimation, extracts and refines shallow features in the acquired Brillouin gain spectrum image to obtain a forward projection, divides the original image by the forward projection to perform embedding and calculate a backward projection, multiplies the initial high spatial resolution Brillouin gain spectrum estimation and the backward projection element by element to obtain an updated high spatial resolution Brillouin gain spectrum estimation, and further extracts and fuses deep features to obtain a high spatial resolution distributed optical fiber sensing parameter. The application effectively suppresses the noise amplification and ringing artifact problems of traditional deconvolution, further improves the performance and reduces the calculation cost.

Description

Technical Field

[0001] This invention belongs to the field of optical fiber parameter processing, specifically relating to a spatially resolution-enhanced Brillouin fiber optic sensing parameter extraction method and system. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Distributed fiber optic sensing technology, with its advantages of resistance to electromagnetic interference, corrosion resistance, distributed measurement capability, and ease of long-distance deployment, has been widely used in industrial fields such as structural health monitoring of large infrastructure, underground compressed energy storage, oil and gas pipeline monitoring, and power cable monitoring. Based on stimulated Brillouin scattering (BOTDA), distributed fiber optic sensing technology can achieve distributed temperature and strain measurement along the entire fiber optic link.

[0004] Spatial resolution is determined by the pump pulse width, which is typically limited by the phonon lifetime in silica optical fiber. While narrower pump pulses can achieve better spatial resolution, the energy of narrow pump pulses decays rapidly during transmission as the sensing distance increases, causing the Brillouin gain spectrum to broaden rapidly and reducing sensing accuracy. Simply increasing the peak power of the pulse can also induce strong nonlinear effects in the fiber, such as pump evacuation and self-phase modulation, which will similarly lead to inaccurate measured Brillouin frequency shifts and even lower spatial resolution.

[0005] To improve spatial resolution, existing technologies have proposed a serial differential pulse pair BOTDA (DPP-BOTDA) hardware scheme. This scheme employs a serial time-division multiplexing measurement mode, generating two wide pulses sequentially and acquiring two sets of Brillouin signals. Subsequent differential operations are performed by a computer. While this approach improves spatial resolution to some extent, the serial pulse generation mode doubles the modulation time, reduces the system sampling rate, and significantly decreases measurement efficiency. Furthermore, the differential operations rely on subsequent data processing, increasing system complexity and easily introducing data processing errors that affect measurement accuracy. Additionally, some deconvolution algorithms are highly sensitive to noise and can cause signal-to-noise ratio degradation, leading to ringing effects. These factors can cause significant errors in the deconvolution results, further impacting the accuracy of the sensing results. Summary of the Invention

[0006] To address the aforementioned problems, this invention proposes a spatially resolution-enhanced Brillouin fiber optic sensing parameter extraction method and system.

[0007] According to some embodiments, the present invention adopts the following technical solution: A spatially resolution-enhanced Brillouin fiber optic sensing parameter extraction method includes the following steps: Obtain low spatial resolution Brillouin gain spectrum images; Features are extracted from the acquired Brillouin gain spectrum image, dimensionality is reduced, and feature representation is enhanced to obtain a feature map. The feature map is restored to the same spatial size as the input image, and feature extraction and channel adjustment are performed. The processed image is added to the original input image to obtain the initial high spatial resolution Brillouin gain spectrum estimate. The shallow features in the acquired Brillouin gain spectrum image are extracted and refined, and then fused and stitched with the features after the feature representation ability is enhanced to obtain the forward projection. The original image is divided by the forward projection for embedding. Calculate the back projection based on the embedded image; The initial high spatial resolution Brillouin gain spectrum estimate is multiplied element-wise by the back projection to obtain the updated high spatial resolution Brillouin gain spectrum estimate. Feature extraction and dimensionality reduction are performed on the updated high spatial resolution Brillouin gain spectrum estimate. Deep feature extraction and fusion are then performed on the dimensionality-reduced features to obtain high spatial resolution distributed fiber optic sensing parameters.

[0008] As an alternative implementation, the process of extracting features from the acquired Brillouin gain spectrum image, performing dimensionality reduction processing, and increasing feature expressiveness to obtain a feature map includes: extracting shallow features from the low spatial resolution Brillouin gain spectrum image through a convolutional layer, then performing feature extraction and dimensionality reduction through residual blocks, and passing the dimensionality-reduced feature map through a bottleneck layer to increase feature expressiveness.

[0009] As an alternative implementation, the process of restoring the feature map to the same spatial size as the input image and performing feature extraction and channel adjustment includes: restoring the feature map to the same spatial size as the input image through an upsampling layer; further extracting features from the processed feature map through a first convolutional layer, a residual block, and a second convolutional layer; adjusting the number of channels to a set value through a third convolutional layer; and adding the result to the original input image to obtain an initial high spatial resolution Brillouin gain spectrum estimate.

[0010] As an alternative implementation, the process of extracting and refining shallow features from the acquired Brillouin gain spectrum image, fusing and stitching them with features enhanced with increased feature expressiveness to obtain the forward projection includes: replacing the point spread function convolution operation with a learnable convolutional layer, calculating the forward projection, extracting shallow features from the acquired Brillouin gain spectrum image by the convolutional layer, refining them by the residual block, fusing and stitching them with the bottleneck layer features, and then inputting them into the convolutional layer and the residual block to obtain the forward projection.

[0011] As an alternative implementation, the process of calculating back projection based on the embedded image includes: replacing the point spread function transpose convolution operation with a second set of learnable convolutional layers, and passing the embedded image through the second set of learnable convolutional layers to calculate the back projection.

[0012] As an alternative implementation, the high spatial resolution distributed fiber optic sensing parameters include the distributed Brillouin gain coefficient, linewidth, and Brillouin frequency shift.

[0013] As an alternative implementation, feature extraction and dimensionality reduction are performed on the updated high spatial resolution Brillouin gain spectrum estimate. The process of deep feature extraction and fusion of the dimensionality-reduced features includes: feature extraction through convolutional layers, dimensionality reduction through max pooling layers, and deep feature extraction of the dimensionality-reduced features through multiple residual blocks.

[0014] As a further defined implementation, the residual block is divided into four stages, each stage including multiple residual blocks. Each residual block contains two convolutional layers, a batch normalization layer and a ReLU activation function, as well as a shortcut connection between the input and output, and then the features are mapped to the output matrix.

[0015] As a further defined implementation, the output of the i-th residual block is represented as: ,in The residual mapping function is represented by the formula: , Indicates batch normalization, This indicates a convolutional layer with a kernel size of 3×3. This indicates a modified linear activation function.

[0016] A spatially resolution-enhanced Brillouin fiber optic sensing parameter extraction system includes: an image acquisition module configured to acquire a low spatial resolution Brillouin gain spectrum image; The estimation module is configured to extract features from the acquired Brillouin gain spectrum image, perform dimensionality reduction processing, and increase the feature representation capability to obtain a feature map. The feature map is restored to the same spatial size as the input image, and feature extraction and channel adjustment are performed. The processed image is added to the original input image to obtain the initial high spatial resolution Brillouin gain spectrum estimate. The update module is configured to extract and refine the shallow features in the acquired Brillouin gain spectrum image, fuse and stitch them with the features after the feature representation is enhanced to obtain the forward projection, divide the original image by the forward projection for embedding; calculate the back projection based on the embedded image; and multiply the initial high spatial resolution Brillouin gain spectrum estimate by the back projection element by element to obtain the updated high spatial resolution Brillouin gain spectrum estimate. The distributed fiber optic parameter extraction module is configured to extract features and reduce the dimensionality of the updated high spatial resolution Brillouin gain spectrum estimate, and then perform deep feature extraction and fusion on the reduced features to obtain high spatial resolution distributed fiber optic sensing parameters.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention innovatively embeds the iterative process of the classic Richardson-Lucy (RL) deconvolution algorithm into the deep feature fusion obtained by the neural network. It not only inherits the interpretability of RL deconvolution, but also integrates the deconvolution iterative process into the feature-driven neural network model, effectively suppressing the amplification of noise and ringing artifacts caused by traditional deconvolution, while further improving performance and reducing computational cost.

[0018] This invention effectively solves the pain point of the Brillouin optical time domain analysis technology, which is difficult to balance spatial resolution, measurement accuracy and measurement efficiency. It can directly recover key sensing parameters such as Brillouin frequency shift, linewidth and gain with high spatial resolution from the low spatial resolution Brillouin gain spectrum acquired by a single long pump pulse.

[0019] This invention requires no hardware modification, significantly reduces data volume and measurement time, and improves system real-time performance and spatial resolution, providing new ideas and technical support for high-precision and low-cost applications of distributed fiber optic sensing in complex engineering scenarios.

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0021] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0022] Figure 1 This is a schematic diagram of the system structure of one embodiment; Figure 2 This is a schematic diagram of the structure of an RL deconvolution module according to one embodiment; Figure 3 This is a schematic diagram of a division operation in one embodiment; Figure 4 This is a schematic diagram of a multiplication operation in one embodiment; Figure 5 This is a schematic diagram of a distributed optical fiber parameter extraction module according to one embodiment; Figure 6The results of a 60ns single-pulse sensing simulation experiment are shown in one embodiment, where (a) is a low spatial resolution Brillouin gain spectrum image and (b) is a high spatial resolution image. Figure 7 The results of a 25km fiber optic sensing experiment are shown in one embodiment. (a) is a 25km low spatial resolution Brillouin gain spectrum image; (b) is an image of a standard test piece at the end of a 25km fiber optic cable; and (c) is an extracted high spatial resolution sensing image. Figure 8 The following are experimental results of 45km fiber optic sensing in one embodiment: (a) is a low spatial resolution Brillouin gain spectrum image of 45km fiber optic cable; (b) is an image of a standard test piece at the end of a 45km fiber optic cable; and (c) is an extracted high spatial resolution sensing image. Detailed Implementation

[0023] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0024] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0025] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0026] Where there is no conflict, the embodiments and features described in this application may be combined with each other.

[0027] Example 1 A spatially resolution-enhanced Brillouin fiber optic sensing parameter extraction method includes the following steps: Obtain low spatial resolution Brillouin gain spectrum images; Features are extracted from the acquired Brillouin gain spectrum image, dimensionality is reduced, and feature representation is enhanced to obtain a feature map. The feature map is restored to the same spatial size as the input image, and feature extraction and channel adjustment are performed. The processed image is added to the original input image to obtain the initial high spatial resolution Brillouin gain spectrum estimate. The shallow features in the acquired Brillouin gain spectrum image are extracted and refined, and then fused and stitched with the features after the feature representation ability is enhanced to obtain the forward projection. The original image is divided by the forward projection for embedding. Calculate the back projection based on the embedded image; The initial high spatial resolution Brillouin gain spectrum estimate is multiplied element-wise by the back projection to obtain the updated high spatial resolution Brillouin gain spectrum estimate. Feature extraction and dimensionality reduction are performed on the updated high spatial resolution Brillouin gain spectrum estimate. Deep feature extraction and fusion are then performed on the dimensionality-reduced features to obtain high spatial resolution distributed fiber optic sensing parameters.

[0028] The steps described above in this embodiment are implemented using an embedded Richardson-Lucy deconvolutional neural network, the core of which includes an RL deconvolution module and a distributed fiber parameter extraction module, such as... Figure 1 As shown below, a detailed introduction will be provided.

[0029] The structure of the RL deconvolution module is as follows: Figure 2 As shown in the estimation module, the low spatial resolution Brillouin gain spectrum image X is processed through convolutional layers with kernel sizes of 3*3 and 1*1, followed by residual blocks for feature extraction and dimensionality reduction. The dimensionality-reduced feature map is then processed through a bottleneck structure using 5*5 and 1*1 convolutional layers to enhance feature representation. Subsequently, an upsampling layer restores the feature map to the same spatial size as the input image. The processed feature map is then further processed through 3*3, 1*1 convolutional layers, a residual block, and another 3*3 convolutional layer for feature extraction. Finally, a 1*1 convolutional layer is used to adjust the number of channels to 1, and the result is added to the original input image to obtain the initial high spatial resolution Brillouin gain spectrum estimate E. k .

[0030] The update module is broken down into four steps: ; ; ; ; On the other hand, this embodiment does not use the fixed analytical form of the physical point spread function (such as Gaussian or Lorentz type) in traditional RL deconvolution. Instead, it replaces the point spread function convolution operation in the Richardson-Lucy algorithm with learnable convolutional layers and residual blocks. In forward projection, this embodiment uses a set of learnable convolutional layers. To replace the point spread function convolution operation, the forward projection FP is calculated. The input image X is processed by 3x3 and 1x1 convolutional layers to extract shallow features, which are then refined by a Residual Block. These features are then fused and concatenated with the bottleneck layer features. Further input of 3x3 and 1x1 convolutional layers and a Residual Block yields the forward projection FP. The FP is then input into the Richardson-Lucy Division module for division, embedding the division of the original image X by FP to obtain the DV, as shown in the diagram. Figure 3 As shown. In back projection, this embodiment replaces the point spread function transpose convolution operation with another set of learnable convolutional layers b. The DV is passed through b to calculate the back projection BP. Finally, the initial high spatial resolution Brillouin gain spectrum E is estimated. k Element-wise multiplication with the back projection BP, i.e., the multiplication operation, yields the updated high spatial resolution BGS estimate E. k+1 ,like Figure 4 As shown. This approach improves the flexibility and adaptability of the model while adhering to the mathematical principles of traditional RL algorithms.

[0031] Distributed fiber optic parameter extraction module, such as Figure 5 As shown. This module aims to estimate E from the reconstructed high spatial resolution Brillouin Gain Spectrum (BGS). k+1 High spatial resolution distributed fiber optic sensing parameters, including Brillouin Frequency Shift (BFS), are extracted from the features. Brillouin line width and Brillouin gain coefficient .

[0032] In this embodiment, the distributed fiber parameter extraction module consists of an improved ResNet18 network. The high spatial resolution BGS feature layer E, initially output by the RL deconvolution module, is then processed. k+1 It first extracts features through a 7x7 convolutional layer, and then reduces the spatial dimension through a max-pooling layer. Here, B is the number of base channels.

[0033] Then, the features are further extracted using 16 residual blocks for deep feature extraction. The residual blocks are divided into four stages, with the number of residual blocks in each stage being [3, 4, 6, 3]. The output of the i-th residual block can be expressed as: ,in The residual mapping function is represented by the formula: , Indicates batch normalization, This indicates a convolutional layer with a kernel size of 3×3. This represents a modified linear activation function. Each residual block contains two 3x3 convolutional layers, a BN layer, a ReLU activation function, and a shortcut connection between the input and output. Finally, the network maps the intermediate features to a 3xW output matrix through 3x3 convolutional layers, where W is the number of sampling points along the fiber length. This allows for the extraction of high spatial resolution distributed fiber sensing information, including Brillouin Frequency Shift (BFS). Brillouin line width and Brillouin gain coefficient .

[0034] Figure 6 To simulate a low spatial resolution BGS, a pump pulse with a width of 60 ns was used, with a sampling rate of 1 Gs / s and a sensing fiber length of 55 meters. This fiber, with a frequency range of 10.65 GHz–11.05 GHz, consisted of multiple fiber segments of varying lengths (6 m, 3 m, 1 m, 0.6 m, 0.5 m, 0.4 m, 0.3 m, 0.2 m, and 0.1 m) with a BFS of 10.95 GHz. The low spatial resolution BGS is shown below. Figure 6 As shown in (a), since the pump pulse width is 60 ns and the spatial resolution is 6 m, the BGS of fiber segments shorter than 6 m will be distorted due to SR limitations. The method proposed in this invention clearly distinguishes fiber segments of all lengths, demonstrating a significant advantage in high spatial resolution distributed fiber optic sensing information extraction.

[0035] Figure 7 (a) is the low spatial resolution distributed Brillouin gain spectrum obtained from a long-distance sensing experiment of 25km. The standard fiber test piece is connected to the tail of the fiber after 25km (the test section lengths are 10cm, 20cm, 30cm, 40cm, 50cm and 1m). Figure 7 (b) is an enlarged view of the tail fiber test standard section in the low spatial resolution BGS. Due to the pump pulse width of 60 ns and the spatial resolution of 6 m, the BGS of the fiber test standard will be indistinguishable due to SR limitations. Figure 7 (c) shows the result after processing with the present invention, which successfully recovered fiber segments of 30cm, 40cm, 50cm and 1m, demonstrating its significant advantages in high spatial resolution distributed optical fiber sensing information extraction.

[0036] Figure 8 For the 45km long-distance sensing experiment, this embodiment uses two 10km fiber optic segments and one 25km fiber optic segment connected in series, and connects the test standard to the end of the fiber optic cable after the 25km segment. Figure 8(b) is an enlarged view of the tail fiber test standard section in the low spatial resolution BGS. Due to the pump pulse width of 60 ns and the spatial resolution of 6 m, the BGS of the fiber test standard will be indistinguishable due to SR limitations. Figure 8 (c) shows the result after processing with the present invention, in which 50cm and 1m fiber segments were successfully recovered.

[0037] Example 2 A spatially resolution-enhanced Brillouin fiber optic sensing parameter extraction system includes: The image acquisition module is configured to acquire low spatial resolution Brillouin gain spectrum images; The estimation module is configured to extract features from the acquired Brillouin gain spectrum image, perform dimensionality reduction processing, and increase the feature representation capability to obtain a feature map. The feature map is restored to the same spatial size as the input image, and feature extraction and channel adjustment are performed. The processed image is added to the original input image to obtain the initial high spatial resolution Brillouin gain spectrum estimate. The update module is configured to extract and refine the shallow features in the acquired Brillouin gain spectrum image, fuse and stitch them with the features after the feature representation is enhanced to obtain the forward projection, divide the original image by the forward projection for embedding; calculate the back projection based on the embedded image; and multiply the initial high spatial resolution Brillouin gain spectrum estimate by the back projection element by element to obtain the updated high spatial resolution Brillouin gain spectrum estimate. The distributed fiber optic parameter extraction module is configured to extract features and reduce the dimensionality of the updated high spatial resolution Brillouin gain spectrum estimate, and then perform deep feature extraction and fusion on the reduced features to obtain high spatial resolution distributed fiber optic sensing parameters.

[0038] The estimation module and the update module are combined into an RL deconvolution module.

[0039] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can be implemented in one or more computer-usable storage media (including, but not limited to, disk storage, etc.) containing computer-usable program code. CD - ROM It takes the form of a computer program product implemented on (such as optical memory, etc.).

[0040] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0041] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0042] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made by those skilled in the art without creative effort within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A spatially resolution-enhanced Brillouin fiber optic sensing parameter extraction method, characterized in that, Includes the following steps: Obtain low spatial resolution Brillouin gain spectrum images; Features are extracted from the acquired Brillouin gain spectrum image, dimensionality reduction is performed, and feature expressiveness is increased to obtain a feature map. Specifically, shallow features are extracted from the low spatial resolution Brillouin gain spectrum image through a convolutional layer, and then feature extraction and dimensionality reduction are performed through residual blocks. The dimensionality-reduced feature map is then passed through a bottleneck layer to increase feature expressiveness. The feature map is restored to the same spatial size as the input image, and feature extraction and channel adjustment are performed. The processed image is added to the original input image to obtain the initial high spatial resolution Brillouin gain spectrum estimate. Specifically, the feature map is restored to the same spatial size as the input image through an upsampling layer. The processed feature map is further extracted through a first convolutional layer, a residual block, and a second convolutional layer. The number of channels is adjusted to a set value through a third convolutional layer. The initial high spatial resolution Brillouin gain spectrum estimate is obtained by adding it to the original input image. Shallow features are extracted and refined from the acquired Brillouin gain spectrum image. These features are then fused and stitched together with the features enhanced with increased feature expressiveness to obtain the forward projection. Specifically, this involves replacing the point spread function convolution operation with a learnable convolutional layer, calculating the forward projection, extracting shallow features from the acquired Brillouin gain spectrum image using a convolutional layer, refining it with residual blocks, fusing and stitching it with bottleneck layer features, and then inputting it into the convolutional layer and residual blocks to obtain the forward projection. Divide the original image by the forward projection to embed it; Based on the embedded image, back projection is calculated, specifically by replacing the point spread function transpose convolution operation with a second set of learnable convolutional layers, and passing the embedded image through the second set of learnable convolutional layers to calculate the back projection. The initial high spatial resolution Brillouin gain spectrum estimate is multiplied element-wise by the back projection to obtain the updated high spatial resolution Brillouin gain spectrum estimate. Feature extraction and dimensionality reduction are performed on the updated high spatial resolution Brillouin gain spectrum estimate. Deep feature extraction and fusion are then performed on the dimensionality-reduced features to obtain high spatial resolution distributed fiber optic sensing parameters.

2. The spatial resolution-enhanced Brillouin fiber optic sensing parameter extraction method as described in claim 1, characterized in that, The high spatial resolution distributed fiber optic sensing parameters include distributed Brillouin gain coefficient, linewidth, and Brillouin frequency shift.

3. The spatial resolution-enhanced Brillouin fiber optic sensing parameter extraction method as described in claim 1, characterized in that, The process of feature extraction and dimensionality reduction of the updated high spatial resolution Brillouin gain spectrum estimate, and deep feature extraction and fusion of the dimensionality-reduced features includes: feature extraction through convolutional layers, dimensionality reduction through max pooling layers, and deep feature extraction of the dimensionality-reduced features through multiple residual blocks.

4. The spatial resolution-enhanced Brillouin fiber optic sensing parameter extraction method as described in claim 3, characterized in that, The residual block is divided into four stages, each stage includes multiple residual blocks, each residual block contains two convolutional layers, a batch normalization layer and a ReLU activation function, as well as a shortcut connection between the input and output, and then the features are mapped to the output matrix.

5. The spatial resolution-enhanced Brillouin fiber optic sensing parameter extraction method as described in claim 3, characterized in that, the first... The output of the i+1 residual blocks is represented as follows: ,in, The residual mapping function is represented by the formula: , Indicates batch normalization, This indicates a convolutional layer with a kernel size of 3×3. This indicates a modified linear activation function.

6. A spatially resolution-enhanced Brillouin fiber optic sensing parameter extraction system, employing the method described in claim 1, characterized in that, include: The image acquisition module is configured to acquire low spatial resolution Brillouin gain spectrum images; The estimation module is configured to extract features from the acquired Brillouin gain spectrum image, perform dimensionality reduction processing, and increase the feature representation capability to obtain a feature map. The feature map is restored to the same spatial size as the input image, and feature extraction and channel adjustment are performed. The processed image is added to the original input image to obtain the initial high spatial resolution Brillouin gain spectrum estimate. The update module is configured to extract and refine the shallow features in the acquired Brillouin gain spectrum image, fuse and stitch them with the features after the feature representation is enhanced to obtain the forward projection, divide the original image by the forward projection for embedding, and calculate the back projection based on the embedded image. The initial high spatial resolution Brillouin gain spectrum estimate is multiplied element-wise by the back projection to obtain the updated high spatial resolution Brillouin gain spectrum estimate. The distributed fiber optic parameter extraction module is configured to extract features and reduce the dimensionality of the updated high spatial resolution Brillouin gain spectrum estimate, and then perform deep feature extraction and fusion on the reduced features to obtain high spatial resolution distributed fiber optic sensing parameters.

Images