Phase unwinding method and device applied to ship collision bridge body vibration monitoring

Abstract

The invention provides a phase unwinding method and device applied to vibration monitoring of a ship colliding with a bridge body, and relates to the technical field of distributed optical fiber sound wave sensing. The method comprises the following steps: based on a self-attention mechanism, carrying out adaptive noise suppression on a window phase signal sequence to obtain a de-noised phase signal sequence; according to the de-noised phase signal sequence, performing multi-scale time-frequency feature fusion through a convolutional neural network to obtain a space-time matrix and time-frequency features; according to the space-time matrix, spatial correlation feature extraction is carried out through a two-dimensional convolutional network, and spatial features are obtained; based on a long-short-term memory network, performing weighted fusion according to the time-frequency features and the spatial features to obtain space-time fusion features; and according to the space-time fusion features, unwrapping phase mapping is carried out through a full connection layer, and a real phase signal sequence is obtained. The phase unwinding method is high in accuracy and good in robustness for distributed optical fiber sound wave vibration in a ship collision scene.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed optical fiber acoustic wave sensing, and in particular to a phase unwinding method and device for monitoring vibrations of a ship-collision bridge. Background Art

[0002] Ship collisions are a major safety threat to bridge structures, especially in waters with frequent shipping. Bridges are at risk of collisions caused by loss of control, overloading, or extreme weather. Distributed fiber-optic acoustic vibration sensing systems, as a high-precision, long-distance structural health monitoring technology, can capture vibration signals from bridge structures in real time through optical fiber sensing, providing critical data support for early warning and damage assessment of ship collision events. However, distributed fiber-optic acoustic vibration sensing systems suffer from significant phase wrapping issues in high-frequency vibration scenarios, such as shock waves generated by ship collisions. Due to the periodicity limitation of phase measurement (typically 2π), when the vibration frequency or amplitude exceeds the system's dynamic range, the phase signal undergoes modulo 2π truncation, resulting in severe distortion of amplitude, frequency, and phase information. This issue directly weakens the distributed fiber-optic acoustic vibration sensing system's ability to identify ship collision events, for example, making it impossible to accurately determine the impact energy, locate the impact location, or distinguish between the impact and other environmental noise (such as wind vibration and vehicle loads), significantly reducing the reliability of the early warning system.

[0003] Traditional phase unconvolution methods (such as time-domain integration and Doppler unconvolution) face multiple challenges in complex engineering scenarios. The instantaneous high-frequency impact signal generated by a ship collision exceeds the dynamic range of traditional algorithms, resulting in increased unconvolution errors. These methods rely on manually designed feature extraction rules and are difficult to adapt to the diversity of different bridge structures and ship collision scenarios. Traditional algorithms have high computational complexity and cannot meet the real-time requirements of the early warning system (response time often exceeds seconds), thus missing the best warning opportunity.

[0004] Distributed fiber-optic acoustic vibration sensing systems, leveraging the distributed sensing properties of optical fibers, can monitor vibration signals along long fiber optic lines in real time. They are widely used in bridge safety monitoring, geological disaster warning, and other fields. However, distributed fiber-optic acoustic vibration sensing systems often face the problem of phase wrapping during signal demodulation. This is due to the interferometric measurement principle, which causes the phase information to be periodically truncated within a range of 2π. This restricts the phase value of the original vibration signal to a range between -π and π, making it unable to directly reflect the true vibration amplitude and frequency information. To address this issue, existing technologies primarily rely on traditional mathematical algorithms and model-driven approaches, but these methods have significant limitations in practical applications.

[0005] The core of traditional phase unwrapping techniques is to expand truncated phase values into a continuous domain through mathematical operations. For example, the least squares method (LSM) estimates the unwrapped phase by fitting a polynomial or curve. However, this method is sensitive to noise, particularly in low signal-to-noise ratio environments, where it can easily introduce errors. It also cannot effectively handle the rapid phase jumps found in high-frequency vibration signals. While high-order polynomial fitting can improve fitting accuracy, it requires extensive prior knowledge to set the polynomial order and initial conditions, resulting in high computational complexity and difficulty in real-time processing. This is particularly true in long-haul fiber monitoring, where the exponential growth of data volume further reduces algorithm efficiency. Furthermore, model-based iterative unwrapping methods (such as phase unwrapping algorithms) rely on assumptions about the physical model of the vibration signal. When the actual vibration signal contains complex noise or nonlinear characteristics, model mismatches can lead to cumulative errors, resulting in systematic deviations in the unwrapping results. Some technologies have also attempted to incorporate hardware compensation, such as increasing the sampling frequency or introducing auxiliary reference signals, to alleviate the wrapping problem. However, hardware upgrades increase system costs, and completely eliminating wrapping is still difficult in dynamic environments or long-haul deployments.

[0006] The shortcomings of existing phase unwrapping technologies for distributed fiber-optic acoustic vibration sensing systems are primarily reflected in the following aspects: First, traditional phase unwrapping algorithms lack robustness to noise. This is particularly true in complex environments such as bridges, where factors such as wind vibration, vehicle loads, and ambient temperature fluctuations can introduce multiple noise sources, leading to distortion in the unwrapped phase signal. Second, these methods perform poorly when processing non-stationary vibration signals. The vibration frequency and amplitude of bridge structures vary significantly under different operating conditions, and traditional algorithms struggle to adaptively adjust model parameters, prone to local optimal solutions or unwrapping errors. Furthermore, the conflict between computational efficiency and real-time performance limits the engineering applications of distributed fiber-optic acoustic vibration sensing systems. For example, in long-distance monitoring, traditional algorithms require point-by-point or segmented phase data processing, which cannot meet the real-time alarm requirements of bridge safety monitoring. Furthermore, existing phase unwrapping methods are highly dependent on the initial phase estimate. If the initial value is biased, the subsequent unwrapping process will fail due to error accumulation. This problem is particularly prominent when the fiber routing is complex or there is interference from multiple nodes. Finally, traditional phase unwrapping techniques struggle to simultaneously achieve the desired unwrapping accuracy for different frequency components. Rapid phase changes in high-frequency vibration signals can exceed the algorithm's processing capabilities, while minute phase fluctuations in low-frequency signals are easily masked by noise, limiting overall monitoring performance. These shortcomings make it difficult for existing phase unwrapping techniques to meet the demands for high precision, high real-time performance, and robust environmental adaptability in practical applications. This is particularly true in scenarios like bridges, where safety and reliability are paramount. A new unwrapping solution that overcomes the limitations of traditional methods is urgently needed.

[0007] In the prior art, there is a lack of a phase unwrapping method for distributed optical fiber acoustic vibration with high accuracy and good robustness for ship impact scenarios. Summary of the Invention

[0008] To address the existing technical issues of insufficient vibration signal demodulation accuracy, poor adaptability to complex dynamic scenarios, and low real-time processing efficiency caused by phase wrapping when distributed fiber-optic acoustic vibration sensing systems are used for ship-collision bridge monitoring, the present invention provides a phase unwrapping method and device for ship-collision bridge vibration monitoring. The technical solution is as follows:

[0009] On the one hand, a phase unwinding method for monitoring vibration of a bridge body caused by a ship collision is provided. The method is implemented by a phase unwinding device and includes:

[0010] The distributed optical fiber vibration sensing system is used to collect monitoring information and obtain a phase signal sequence of bridge vibration. The phase signal sequence is segmented based on a sliding window of a preset length to obtain a window phase signal sequence.

[0011] Based on the self-attention mechanism, adaptive noise suppression is performed on the window phase signal sequence to obtain the denoised phase signal sequence;

[0012] According to the denoised phase signal sequence, multi-scale time-frequency feature fusion is performed through convolutional neural network to obtain the space-time matrix and time-frequency features;

[0013] According to the spatiotemporal matrix, spatial correlation features are extracted through a two-dimensional convolutional network to obtain spatial features;

[0014] Based on the long short-term memory network, the time-frequency features and spatial features are weightedly fused to obtain the spatiotemporal fusion features;

[0015] According to the spatiotemporal fusion features, the unwrapped phase mapping is performed through the fully connected layer to obtain the real phase signal sequence.

[0016] On the other hand, a phase unwinding device for monitoring vibrations of a ship-collision bridge body is provided. The device is applied to a phase unwinding method for monitoring vibrations of a ship-collision bridge body. The device comprises:

[0017] The signal acquisition module is used to collect monitoring information through a distributed optical fiber vibration sensing system to obtain a phase signal sequence of bridge vibration; based on a sliding window of a preset length, the phase signal sequence is segmented to obtain a window phase signal sequence;

[0018] The signal denoising module is used to perform adaptive noise suppression on the window phase signal sequence based on the self-attention mechanism to obtain a denoised phase signal sequence;

[0019] The time-frequency feature extraction module is used to perform multi-scale time-frequency feature fusion through a convolutional neural network based on the denoised phase signal sequence to obtain the space-time matrix and time-frequency features;

[0020] The spatial feature extraction module is used to extract spatial correlation features through a two-dimensional convolutional network based on the spatiotemporal matrix to obtain spatial features;

[0021] The feature fusion module is used to perform weighted fusion based on the long short-term memory network according to the time-frequency features and spatial features to obtain the spatiotemporal fusion features;

[0022] The signal deconvolution module is used to perform deconvolution phase mapping through a fully connected layer based on the spatiotemporal fusion features to obtain the real phase signal sequence.

[0023] On the other hand, a phase unwinding device is provided, which includes: a processor; a memory, wherein computer-readable instructions are stored in the memory, and when the computer-readable instructions are executed by the processor, any one of the phase unwinding methods mentioned above for ship-collision bridge vibration monitoring is implemented.

[0024] On the other hand, a computer-readable storage medium is provided, in which at least one instruction is stored. The at least one instruction is loaded and executed by a processor to implement any one of the above-mentioned phase unwrapping methods for monitoring vibration of a ship-collision bridge body.

[0025] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0026] This paper proposes a phase unwrapping method for monitoring the vibration of ship-impact bridge beams. By using a deep learning model to learn the inherent physical laws and noise distribution characteristics of phase signals using large-scale training data, the method demonstrates enhanced robustness against complex interferences such as random noise, ambient temperature fluctuations, and the superposition of multiple vibration sources. It maintains high-precision unwrapping capabilities, especially in low signal-to-noise ratio environments, addressing the phase jumps and cumulative errors caused by noise sensitivity in traditional algorithms. The method enables real-time processing of massive amounts of phase data along long-distance optical fibers, overcoming the computational efficiency bottleneck caused by point-by-point expansion in traditional iterative algorithms and meeting the real-time alarm and rapid response requirements of bridge safety monitoring. The deep learning model, through an adaptive parameter optimization mechanism, dynamically adapts to the vibration signal characteristics of different frequency components, accurately capturing the rapid phase changes of high-frequency vibrations while effectively identifying phase fluctuations of low-frequency, minute vibrations. This significantly improves the system's compatibility with multi-band vibration signals and overall monitoring performance. The feature fusion capability of deep learning further enhances the ability to identify complex vibration patterns, providing highly reliable, high-resolution vibration sensing data for bridge structural health monitoring. This significantly reduces the system's reliance on hardware compensation methods and reduces engineering deployment costs.

[0027] This paper reconstructs the phase unwrapping algorithm framework through deep learning technology, fundamentally overcoming the limitations of traditional methods in terms of noise robustness, computational efficiency, dynamic adaptability, and global accuracy. This lays a technical foundation for the large-scale application of distributed fiber-optic acoustic vibration sensing systems in high-reliability scenarios such as bridge safety monitoring. This invention provides a highly accurate and robust phase unwrapping method for distributed fiber-optic acoustic vibration sensing in ship impact scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0029] Figure 1 This is a flow chart of a phase unwrapping method for vibration monitoring of a bridge body caused by a ship collision, provided by an embodiment of the present invention;

[0030] Figure 2 This is a block diagram of a phase unwinding device for monitoring vibrations of a bridge body caused by a ship collision, provided by an embodiment of the present invention;

[0031] Figure 3 It is a structural schematic diagram of a phase unwrapping device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0033] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as an "exemplary" in the present invention should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of the word "exemplary" is intended to present concepts in a concrete manner. Furthermore, in the embodiments of the present invention, "and / or" can mean both or either of the two.

[0034] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same. The terms "of," "corresponding," and "corresponding" may sometimes be used interchangeably. It should be noted that, when the distinction is not emphasized, the meanings they convey are the same.

[0035] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0036] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0037] The embodiment of the present invention provides a phase unwinding method for monitoring vibration of a ship-collision bridge. The method can be implemented by a phase unwinding device, which can be a terminal or a server. Figure 1 The flowchart of the phase unwrapping method applied to the vibration monitoring of a ship-collision bridge is shown. The processing flow of the method may include the following steps:

[0038] S1. Monitoring information is collected through a distributed optical fiber vibration sensing system to obtain a phase signal sequence of bridge vibration; the phase signal sequence is segmented based on a sliding window of a preset length to obtain a window phase signal sequence.

[0039] In a feasible embodiment, the received phase signal sequence of the distributed optical fiber vibration sensing system is To process, is the time index. Use a fixed-length sliding window to calculate the phase signal sequence Segmentation is performed to ensure the continuity of event features.

[0040] S2. Based on the self-attention mechanism, adaptive noise suppression is performed on the window phase signal sequence to obtain a denoised phase signal sequence.

[0041] Optionally, based on a self-attention mechanism, adaptive noise suppression is performed on the window phase signal sequence to obtain a denoised phase signal sequence, including:

[0042] Long-distance dependency capture is performed according to the window phase signal sequence to obtain a weighted value vector of the intermediate quantity of the window phase signal sequence;

[0043] Calculate the attention output of the window phase signal sequence based on the weighted value vector and the preset output weight matrix;

[0044] Based on the attention output, dynamic residual analysis is performed according to the window phase signal sequence to obtain the adaptive threshold setting of statistical characteristics;

[0045] According to the adaptively set threshold and attention output, the window phase signal sequence is subjected to binary mask denoising to obtain a denoised phase signal sequence.

[0046] In one feasible implementation, a self-attention mechanism is used to analyze the window phase signal sequence. Processing, intermediate No. The weighted value vector of each position is as follows (1):

[0047] (1)

[0048] in, , , is a learnable matrix used to map the original features to the attention space; yes The i-th row vector of is the j-th column vector of K, yes The j-th row vector of TN is the window phase signal sequence. The number of elements.

[0049] The attention output h is as follows (2):

[0050] (2)

[0051] in, is the output weight matrix.

[0052] The noise and signal areas are distinguished by residual analysis and dynamic thresholding, and a mask is generated for noise suppression. The absolute difference between the original input and the attention output can be expressed as a residual vector, and the threshold is adaptively set according to the statistical characteristics of the current residual. As shown in formula (3):

[0053] (3);

[0054] in, is the coefficient.

[0055] The set of locations where the residual exceeds the threshold is defined as the noise region. The output X after denoising The value X of the position i As shown in formula (4):

[0056] (4);

[0057] in, is a binary mask. When the residual is less than or equal to the threshold, (keep the original signal), when the residual is greater than the threshold (Replaced with attention output).

[0058] S3. Based on the denoised phase signal sequence, multi-scale time-frequency feature fusion is performed through a convolutional neural network to obtain the space-time matrix and time-frequency features.

[0059] Optionally, based on the denoised phase signal sequence, multi-scale time-frequency feature fusion is performed through a convolutional neural network to obtain a space-time matrix and time-frequency features, including:

[0060] According to the preset signal conversion parameters, the denoised phase signal sequence is converted into a time-space matrix;

[0061] According to the spatiotemporal matrix, multi-scale feature extraction is performed through a convolutional neural network to obtain multi-scale time-frequency local features;

[0062] Based on the preset fusion weights, channel fusion is performed according to the multi-scale time-frequency local features to obtain time-frequency features.

[0063] In a feasible implementation, according to the preset signal conversion parameters including repetition frequency, data acquisition card sampling speed, etc., the noise-reduced window phase signal sequence X is converted into a time-space matrix . Indicates time Space-time position The spatial dimension of the space-time matrix corresponds to the sampling points along the sensing cable of the distributed optical fiber vibration sensing system, the temporal dimension corresponds to the sampling time points during continuous monitoring of the system, and the element values of the matrix correspond to the amplitude.

[0064] The spatiotemporal matrix is rectified by N parallel convolution kernels Extract local features at different time scales. The convolution kernel is in time The local features extracted at are as follows (5):

[0065] (5);

[0066] in, It is The convolution kernel is offset in time The weight parameter at It is The size of the convolution kernel, is a bias term used to adjust the linear combination of features

[0067] Multi-scale features After splicing, pass 1 1 convolution is used to perform channel fusion, and the fused time-frequency features are as follows (6):

[0068] (6);

[0069] in, It is The fusion weight of each scale feature is used to adjust the contribution of each scale feature to the final fusion result. is the bias term of the fusion layer.

[0070] Will The feature sequences of different scales are spliced according to the channel dimension to form a high-dimensional feature vector, which retains the information of all time scales. The fused feature sequence It integrates the time-frequency information of high-frequency, medium-frequency and low-frequency vibrations.

[0071] Among them, the convolution kernel of the convolutional neural network is in parallel relationship; the number of convolution kernels is 3; the size of the convolution kernel is 3 3, 5 5 and 7 7.

[0072] In one possible implementation, different convolution kernel sizes correspond to different time windows. Small kernels capture local details of high-frequency vibrations, while large kernels extract long-term trends of low-frequency vibrations. By processing in parallel, the network can simultaneously focus on features at different time scales.

[0073] S4. According to the spatiotemporal matrix, spatial correlation features are extracted through a two-dimensional convolutional network to obtain spatial features.

[0074] Among them, the two-dimensional convolutional network captures the spatial correlation of sensor information in the spatiotemporal matrix by establishing a local correlation model in the spatial dimension.

[0075] In one feasible implementation, a two-dimensional convolutional network is used to process the phase data along the optical fiber at adjacent time points, and the spatial features As shown in formula (7):

[0076] (7);

[0077] The two-dimensional convolution kernel simultaneously considers the phase values of the current spatial position and its adjacent spatial positions at time t, and captures the spatial correlation between sensors by modeling local correlation in the spatial dimension.

[0078] S5. Based on the long short-term memory network, weighted fusion is performed according to the time-frequency features and spatial features to obtain the spatiotemporal fusion features.

[0079] Optionally, based on the long short-term memory network, weighted fusion is performed according to the time-frequency features and the spatial features to obtain spatiotemporal fusion features, including:

[0080] Based on the gating mechanism and time-frequency characteristics, the long short-term memory network is used to capture dynamic long-term dependencies and obtain hidden states.

[0081] Based on the preset attention mechanism weights, the spatial attention weights are calculated according to the spatial features and hidden states;

[0082] Based on the spatial attention weight, the time-frequency features and spatial features are weightedly fused to obtain the spatiotemporal fusion features.

[0083] In one feasible implementation, a long short-term memory network (LSTM) is used to perform the fusion of feature sequences. For time series modeling, LSTM dynamically updates the hidden state through the gating mechanism (forget gate, input gate, output gate) , can memorize long-term time dependencies, that is, the periodic or trend changes of vibration signals, and capture long-term dependencies. The hidden state As shown in formula (8):

[0084] (8);

[0085] in, It is a hidden state and is dynamically updated through a gating mechanism (forget gate, input gate, output gate).

[0086] The spatial features are combined with the temporal features and weighted by the attention mechanism. The process is as follows (9):

[0087] (9);

[0088] in, is the spatial attention weight, which is calculated by the spatial attention module. The process is as follows (10):

[0089] (10);

[0090] in, It is a linear combination of the time-frequency and spatial features based on the weight of the attention mechanism.

[0091] S6. Based on the spatiotemporal fusion features, unwrapped phase mapping is performed through the fully connected layer to obtain the true phase signal sequence.

[0092] In one feasible implementation, spatiotemporal features are established through a fully connected layer With the true phase Nonlinear mapping:

[0093] (11);

[0094] in, is the weight matrix of the fully connected layer, which is used for linear mapping of features to frequencies; is the bias term of the fully connected layer.

[0095] This invention uses a deep learning approach to unwind the phase wrapping problem in a distributed fiber-optic acoustic vibration sensing system, significantly improving the system's monitoring accuracy, real-time performance, and environmental adaptability in complex environments. Compared to traditional mathematical algorithms, the deep learning model automatically extracts multi-scale vibration features from phase signals through an end-to-end feature learning mechanism, effectively overcoming the effects of noise interference and nonlinear distortion.

[0096] In addition, the present invention does not rely on manual setting of the polynomial order, initial phase value or physical model parameters, avoiding the systematic deviation caused by insufficient prior knowledge or model mismatch in traditional methods. Especially in bridge scenarios with complex fiber optic layout paths and multi-node interference, it can effectively suppress the error accumulation effect and ensure the global consistency of the unwinding results.

[0097] This paper proposes a phase unwrapping method for monitoring the vibration of ship-impact bridge beams. By using a deep learning model to learn the inherent physical laws and noise distribution characteristics of phase signals using large-scale training data, the method demonstrates enhanced robustness against complex interferences such as random noise, ambient temperature fluctuations, and the superposition of multiple vibration sources. It maintains high-precision unwrapping capabilities, especially in low signal-to-noise ratio environments, addressing the phase jumps and cumulative errors caused by noise sensitivity in traditional algorithms. The method enables real-time processing of massive amounts of phase data along long-distance optical fibers, overcoming the computational efficiency bottleneck caused by point-by-point expansion in traditional iterative algorithms and meeting the real-time alarm and rapid response requirements of bridge safety monitoring. The deep learning model, through an adaptive parameter optimization mechanism, dynamically adapts to the vibration signal characteristics of different frequency components, accurately capturing the rapid phase changes of high-frequency vibrations while effectively identifying phase fluctuations of low-frequency, minute vibrations. This significantly improves the system's compatibility with multi-band vibration signals and overall monitoring performance. The feature fusion capability of deep learning further enhances the ability to identify complex vibration patterns, providing highly reliable, high-resolution vibration sensing data for bridge structural health monitoring. This significantly reduces the system's reliance on hardware compensation methods and reduces engineering deployment costs.

[0098] This paper reconstructs the phase unwrapping algorithm framework through deep learning technology, fundamentally overcoming the limitations of traditional methods in terms of noise robustness, computational efficiency, dynamic adaptability, and global accuracy. This lays a technical foundation for the large-scale application of distributed fiber-optic acoustic vibration sensing systems in high-reliability scenarios such as bridge safety monitoring. This invention provides a highly accurate and robust phase unwrapping method for distributed fiber-optic acoustic vibration sensing in ship impact scenarios.

[0099] Figure 2 This is a block diagram of a phase unwinding device for monitoring vibrations of a ship-collision bridge according to an exemplary embodiment. The device is used to implement a phase unwinding method for monitoring vibrations of a ship-collision bridge. Figure 2 The device includes a signal acquisition module 210, a signal denoising module 220, a time-frequency feature extraction module 230, a spatial feature extraction module 240, a feature fusion module 250, and a signal deconvolution module 260.

[0100] The signal acquisition module 210 is used to collect monitoring information through a distributed optical fiber vibration sensing system to obtain a phase signal sequence of bridge vibration; based on a sliding window of a preset length, the phase signal sequence is segmented to obtain a window phase signal sequence;

[0101] A signal denoising module 220 is configured to perform adaptive noise suppression on the window phase signal sequence based on a self-attention mechanism to obtain a denoised phase signal sequence;

[0102] The time-frequency feature extraction module 230 is used to perform multi-scale time-frequency feature fusion through a convolutional neural network based on the denoised phase signal sequence to obtain a space-time matrix and time-frequency features;

[0103] A spatial feature extraction module 240 is configured to extract spatial correlation features using a two-dimensional convolutional network based on the spatiotemporal matrix to obtain spatial features;

[0104] The feature fusion module 250 is used to perform weighted fusion based on the long short-term memory network according to the time-frequency features and the spatial features to obtain the spatiotemporal fusion features;

[0105] The signal deconvolution module 260 is configured to perform deconvolution phase mapping through a fully connected layer according to the spatiotemporal fusion features to obtain a true phase signal sequence.

[0106] Optionally, the signal denoising module 220 is further configured to:

[0107] Long-distance dependency capture is performed according to the window phase signal sequence to obtain a weighted value vector of the intermediate quantity of the window phase signal sequence;

[0108] Calculate the attention output of the window phase signal sequence based on the weighted value vector and the preset output weight matrix;

[0109] Based on the attention output, dynamic residual analysis is performed according to the window phase signal sequence to obtain the adaptive threshold setting of statistical characteristics;

[0110] According to the adaptively set threshold and attention output, the window phase signal sequence is subjected to binary mask denoising to obtain a denoised phase signal sequence.

[0111] Optionally, the time-frequency feature extraction module 230 is further configured to:

[0112] According to the preset signal conversion parameters, the denoised phase signal sequence is converted into a time-space matrix;

[0113] According to the spatiotemporal matrix, multi-scale feature extraction is performed through a convolutional neural network to obtain multi-scale time-frequency local features;

[0114] Based on the preset fusion weights, channel fusion is performed according to the multi-scale time-frequency local features to obtain time-frequency features.

[0115] Among them, the convolution kernel of the convolutional neural network is in parallel relationship; the number of convolution kernels is 3; the size of the convolution kernel is 3 3, 5 5 and 7 7.

[0116] Among them, the two-dimensional convolutional network captures the spatial correlation of sensor information in the spatiotemporal matrix by establishing a local correlation model in the spatial dimension.

[0117] Optionally, the feature fusion module 250 is further configured to:

[0118] Based on the gating mechanism and time-frequency characteristics, the long short-term memory network is used to capture dynamic long-term dependencies and obtain hidden states.

[0119] Based on the preset attention mechanism weights, the spatial attention weights are calculated according to the spatial features and hidden states;

[0120] Based on the spatial attention weight, the time-frequency features and spatial features are weightedly fused to obtain the spatiotemporal fusion features.

[0121] This paper proposes a phase unwrapping method for monitoring the vibration of ship-impact bridge beams. By using a deep learning model to learn the inherent physical laws and noise distribution characteristics of phase signals using large-scale training data, the method demonstrates enhanced robustness against complex interferences such as random noise, ambient temperature fluctuations, and the superposition of multiple vibration sources. It maintains high-precision unwrapping capabilities, especially in low signal-to-noise ratio environments, addressing the phase jumps and cumulative errors caused by noise sensitivity in traditional algorithms. The method enables real-time processing of massive amounts of phase data along long-distance optical fibers, overcoming the computational efficiency bottleneck caused by point-by-point expansion in traditional iterative algorithms and meeting the real-time alarm and rapid response requirements of bridge safety monitoring. The deep learning model, through an adaptive parameter optimization mechanism, dynamically adapts to the vibration signal characteristics of different frequency components, accurately capturing the rapid phase changes of high-frequency vibrations while effectively identifying phase fluctuations of low-frequency, minute vibrations. This significantly improves the system's compatibility with multi-band vibration signals and overall monitoring performance. The feature fusion capability of deep learning further enhances the ability to identify complex vibration patterns, providing highly reliable, high-resolution vibration sensing data for bridge structural health monitoring. This significantly reduces the system's reliance on hardware compensation methods and reduces engineering deployment costs.

[0122] This paper reconstructs the phase unwrapping algorithm framework through deep learning technology, fundamentally overcoming the limitations of traditional methods in terms of noise robustness, computational efficiency, dynamic adaptability, and global accuracy. This lays a technical foundation for the large-scale application of distributed fiber-optic acoustic vibration sensing systems in high-reliability scenarios such as bridge safety monitoring. This invention provides a highly accurate and robust phase unwrapping method for distributed fiber-optic acoustic vibration sensing in ship impact scenarios.

[0123] Figure 3 : is a schematic structural diagram of a phase unwrapping device provided by an embodiment of the present invention, such as Figure 3 As shown, the phase unwrapping device may include the above Figure 2 The phase unwrapping device shown is applied to the vibration monitoring of a ship-collision bridge. Optionally, the phase unwrapping device 310 may include a first processor 2001 .

[0124] Optionally, the phase unwrapping device 310 may further include a memory 2002 and a transceiver 2003 .

[0125] The first processor 2001, the memory 2002 and the transceiver 2003 may be connected via a communication bus.

[0126] The following combination Figure 3 The components of the phase unwrapping device 310 are described in detail:

[0127] The first processor 2001 is the control center of the phase unwrapping device 310 and can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs) or one or more field programmable gate arrays (FPGAs).

[0128] Optionally, the first processor 2001 may execute various functions of the phase unwrapping device 310 by running or executing a software program stored in the memory 2002 and calling data stored in the memory 2002 .

[0129] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 3 CPU0 and CPU1 are shown in FIG.

[0130] In a specific implementation, as an embodiment, the phase unwrapping device 310 may also include multiple processors, such as Figure 3 1 and 2. The first processor 2001 and the second processor 2004 are shown in FIG. Each of these processors can be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). A processor herein can refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0131] The memory 2002 is used to store the software program for executing the solution of the present invention, and is controlled by the first processor 2001 for execution. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0132] Alternatively, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, an optical disc storage (including a compact disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or exist independently and accessed through the interface circuit ( Figure 3 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0133] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0134] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 3 The receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0135] Optionally, the transceiver 2003 may be integrated with the first processor 2001, or may exist independently and communicate with the first processor 2001 through the interface circuit ( Figure 3 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0136] It should be noted that Figure 3 The structure of the phase unwrapping device 310 shown in the figure does not constitute a limitation on the router. The actual knowledge structure recognition device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0137] In addition, the technical effects of the phase unwinding device 310 can refer to the technical effects of the phase unwinding method applied to the vibration monitoring of a ship-collision bridge body described in the above method embodiment, and will not be repeated here.

[0138] It should be understood that the first processor 2001 in the embodiment of the present invention may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor, or the processor may be any conventional processor, etc.

[0139] It should also be understood that the memory in the embodiments of the present invention may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0140] The above embodiments can be implemented in whole or in part via software, hardware (e.g., circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0141] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0142] In this disclosure, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of a, b, or c" can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or plural.

[0143] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0144] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0145] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0146] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of the device or unit, which can be electrical, mechanical or other forms.

[0147] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0148] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0149] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical disks.

[0150] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A phase unwrapping method for monitoring vibration of a bridge body caused by a ship collision, characterized in that: The method comprises: The distributed optical fiber vibration sensing system is used to collect monitoring information and obtain a phase signal sequence of bridge vibration. The phase signal sequence is segmented based on a sliding window of a preset length to obtain a window phase signal sequence. Based on the self-attention mechanism, adaptive noise suppression is performed on the window phase signal sequence to obtain the denoised phase signal sequence; According to the denoised phase signal sequence, multi-scale time-frequency feature fusion is performed through convolutional neural network to obtain the space-time matrix and time-frequency features; According to the spatiotemporal matrix, spatial correlation features are extracted through a two-dimensional convolutional network to obtain spatial features; Based on the long short-term memory network, the time-frequency features and spatial features are weightedly fused to obtain the spatiotemporal fusion features; According to the spatiotemporal fusion features, the unwrapped phase mapping is performed through the fully connected layer to obtain the real phase signal sequence.

2. The phase unwrapping method for monitoring vibration of a ship-collision bridge according to claim 1 is characterized in that: The method of adaptively suppressing noise on the window phase signal sequence based on the self-attention mechanism to obtain a denoised phase signal sequence includes: Long-distance dependency capture is performed according to the window phase signal sequence to obtain a weighted value vector of the intermediate quantity of the window phase signal sequence; Calculate the attention output of the window phase signal sequence based on the weighted value vector and the preset output weight matrix; Based on the attention output, dynamic residual analysis is performed according to the window phase signal sequence to obtain the adaptive threshold setting of statistical characteristics; According to the adaptively set threshold and attention output, the window phase signal sequence is subjected to binary mask denoising to obtain a denoised phase signal sequence.

3. The phase unwrapping method for monitoring vibration of a ship-collision bridge according to claim 1 is characterized in that: The method of performing multi-scale time-frequency feature fusion by a convolutional neural network based on the denoised phase signal sequence to obtain a space-time matrix and time-frequency features includes: According to the preset signal conversion parameters, the denoised phase signal sequence is converted into a time-space matrix; According to the spatiotemporal matrix, multi-scale feature extraction is performed through a convolutional neural network to obtain multi-scale time-frequency local features; Based on the preset fusion weights, channel fusion is performed according to the multi-scale time-frequency local features to obtain time-frequency features.

4. The phase unwrapping method for monitoring vibration of a ship-collision bridge according to claim 3 is characterized in that: The convolution kernels of the convolutional neural network are in parallel relationship; the number of the convolution kernels is 3; the size of the convolution kernel is 3 3, 5 5 and 7 7.

5. The phase unwrapping method for monitoring vibration of a ship-collision bridge according to claim 1 is characterized in that: The two-dimensional convolutional network captures the spatial correlation of sensor information in the spatiotemporal matrix by establishing a local correlation model in the spatial dimension.

6. The phase unwrapping method for monitoring vibration of a ship-collision bridge according to claim 1 is characterized in that: The long short-term memory network is based on the weighted fusion of time-frequency features and spatial features to obtain spatiotemporal fusion features, including: Based on the gating mechanism and time-frequency characteristics, the long short-term memory network is used to capture dynamic long-term dependencies and obtain hidden states. Based on the preset attention mechanism weights, the spatial attention weights are calculated according to the spatial features and hidden states; Based on the spatial attention weight, the time-frequency features and spatial features are weightedly fused to obtain the spatiotemporal fusion features.

7. A phase unwinding device for monitoring vibrations of a ship-collision bridge, wherein the phase unwinding device is used to implement the phase unwinding method for monitoring vibrations of a ship-collision bridge as described in any one of claims 1 to 6, and is characterized in that: The device comprises: The signal acquisition module is used to collect monitoring information through a distributed optical fiber vibration sensing system to obtain a phase signal sequence of bridge vibration; based on a sliding window of a preset length, the phase signal sequence is segmented to obtain a window phase signal sequence; The signal denoising module is used to perform adaptive noise suppression on the window phase signal sequence based on the self-attention mechanism to obtain a denoised phase signal sequence; The time-frequency feature extraction module is used to perform multi-scale time-frequency feature fusion through a convolutional neural network based on the denoised phase signal sequence to obtain the space-time matrix and time-frequency features; The spatial feature extraction module is used to extract spatial correlation features through a two-dimensional convolutional network based on the spatiotemporal matrix to obtain spatial features; The feature fusion module is used to perform weighted fusion based on the long short-term memory network according to the time-frequency features and spatial features to obtain the spatiotemporal fusion features; The signal deconvolution module is used to perform deconvolution phase mapping through a fully connected layer based on the spatiotemporal fusion features to obtain the real phase signal sequence.

8. The phase unwinding device for monitoring vibration of a ship-collision bridge according to claim 7 is characterized in that: The signal denoising module is further configured to: Long-distance dependency capture is performed according to the window phase signal sequence to obtain a weighted value vector of the intermediate quantity of the window phase signal sequence; Calculate the attention output of the window phase signal sequence based on the weighted value vector and the preset output weight matrix; Based on the attention output, dynamic residual analysis is performed according to the window phase signal sequence to obtain the adaptive threshold setting of statistical characteristics; According to the adaptively set threshold and attention output, the window phase signal sequence is subjected to binary mask denoising to obtain a denoised phase signal sequence.

9. A phase unwinding device, characterized in that: The phase unwinding device comprises: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program code, which can be called by a processor to execute the method according to any one of claims 1 to 6.

Images