A method and system for submarine cable state recognition based on dynamic fingerprint matching

Abstract

The present application relates to a kind of submarine cable state identification method and system based on dynamics fingerprint matching, it is related to distributed optical fiber vibration sensing and submarine cable health monitoring technical field.The method obtains vibration time series data distributed along submarine cable, forms space-time two-dimensional data matrix, submarine cable includes submarine cable and submarine optical fiber composite cable;Target mileage section is processed by multi-scale space sliding window and time window, and the measured dynamics feature related to submarine cable-sea bed support state is extracted;Call preset submarine cable state dynamics fingerprint library, match the measured feature with the dynamics fingerprint under different buried, exposed, scour support weakening or suspected suspension state, output state category, abnormal section, risk score and confidence.Fingerprint library can be constructed by numerical simulation, model test, historical inspection data or sea measurement data.Compared with prior art, the present application can realize the continuous identification and inspection auxiliary decision of submarine cable buried state degradation and scouring exposure risk.

Description

Technical Field

[0001] This invention relates to the fields of submarine cable health monitoring, distributed optical fiber vibration sensing, marine engineering dynamics, submarine cable-seabed coupling response identification, and seabed scour exposure risk assessment, and in particular to a submarine cable condition identification method and system based on dynamic fingerprint matching. Background Technology

[0002] After being laid, submarine cables are subject to long-term influences from tides, waves, seabed migration, localized scouring, anchor disturbances, and changes in geological conditions. While a cable may initially be in a stable buried state, scouring and sediment redistribution can gradually lead to shallow burial, semi-buried, partially exposed, or weakened seabed support, and even, in certain sections, apparent suspension. These conditions alter the contact stiffness, damping, friction, and boundary conditions between the cable and the seabed, thereby affecting the cable's dynamic response under marine environmental stimuli.

[0003] Traditional inspections primarily rely on side-scan sonar, multibeam sonar, ROV, shallow seismic profiling, magnetic positioning, or manual marine survey reports. While these methods can provide spatial evidence of exposure or erosion, they typically suffer from long inspection cycles, high costs, difficulty in continuous monitoring, and inability to reflect changes in dynamic response during operation. Distributed fiber optic vibration sensing enables long-distance, dense observation of submarine cables, providing conditions for online identification of cable burial degradation.

[0004] Existing anomaly detection methods based on distributed optical fibers often use local energy enhancement or spectral peaks as alarm criteria, making it difficult to distinguish between ship far-field disturbances, enhanced local coupling, shallow burial or exposure caused by seabed erosion, and true free-suspension modal responses. Forcibly applying all anomalies to a fully suspended tension beam model can easily lead to over-interpretation of physics, especially in field conditions where most submarine cables are still in contact with the seabed, where the tension beam model is not always applicable.

[0005] Therefore, there is a need for an identification scheme for the actual degradation status of in-service submarine cables. This scheme should leverage the advantages of continuous monitoring using distributed optical fibers to link measured vibration characteristics with the coupling dynamics mechanism of the submarine cable and seabed. It should also classify and assess the degradation status of the cables by using a pre-set dynamic fingerprint database, rather than relying solely on a single abnormal energy threshold or a single suspended beam inversion model. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a method and system for identifying the state of submarine cables based on dynamic fingerprint matching. This addresses the problem that existing technologies struggle to distinguish between different degradation states of submarine cables, such as shallow burial, semi-buried, partially exposed, weakened support, and suspected suspension, and reduces the risk of misinterpreting incomplete suspension anomalies as responses of suspended tension beams. Figure 3As shown, submarine cables can exhibit various states under the influence of scouring and seabed migration, such as complete burial, shallow burial, semi-burial or partial exposure, weakened support due to scouring, and suspected suspension. This invention identifies and assesses the risks associated with these continuous degradation states.

[0007] The objective of this invention can be achieved through the following technical solutions: A method for identifying the state of submarine cables based on dynamic fingerprint matching, the method comprising: Vibration time-series data from multiple spatial sampling points along the submarine cable are acquired using distributed optical fiber vibration sensing links laid along or accompanying the submarine cable; the submarine cable includes submarine optical cable and submarine optical-electric composite cable. The vibration time series data is organized into a two-dimensional space-time matrix based on spatial and temporal sampling points, and the data segment to be tested is constructed based on the target inspection mileage. The data segment to be tested is divided into multi-scale spatial sliding windows and / or time windows to obtain several spatial-time windows. For each spatial-time window, the measured dynamic features related to the cable-seabed support state are extracted. The measured dynamic features include at least one of the following: frequency-space response features, frequency band energy features, spatial coherence features, and modal energy distribution features. A preset submarine cable state dynamic fingerprint database is invoked, and the measured dynamic features are matched, classified, and regressed with the dynamic feature templates in the submarine cable state dynamic fingerprint database to obtain the state category, anomaly score, exposure probability, and scour risk score corresponding to each space-time window; the dynamic fingerprint database includes dynamic feature templates for different burial states, exposure states, scour states, support weakening states, and suspected suspended states. The spatial-temporal windows that are adjacent and whose status categories meet the preset status category consistency condition or whose anomaly scores, exposure probabilities or scour risk scores meet the preset threshold conditions are merged into submarine cable status anomaly segments and their dynamic anomaly lengths are calculated. Within the same mileage segment, multiple sets of measured dynamic features are extracted according to the preset time window length and time step, and the repetition ratio of state categories, stability of risk scores, repetition rate of spectral peaks, shift of anomaly center, and fluctuation of anomaly length are statistically analyzed to correct the identification confidence of each segment. The degradation risk sections identified by distributed optical fibers are spatially aligned with exposed, scoured, or shallowly buried sections in side-scan sonar, multibeam sonar, ROV, shallow seismic profiling, magnetic positioning, or artificial marine survey reports. Spatial overlap, center deviation, length error, and risk level consistency are calculated, and the scour risk score is corrected or inspection verification results are generated. Output the start and end mileage, dynamic abnormal length, status category, risk level, and confidence level of the abnormal section of the submarine cable to complete the submarine cable status identification; the status category includes at least one of the following: fully buried, shallowly buried, partially buried, partially exposed, scour pit developed, seabed support weakened, and suspected of being suspended.

[0008] Furthermore, the distributed optical fiber vibration sensing link is a distributed acoustic sensing, distributed vibration sensing, phase-sensitive optical time-domain reflection, or an equivalent optical fiber vibration measurement link along the line. The vibration time-series data includes strain along the line, strain rate, phase change, vibration intensity, and its transformation quantity. The space-time two-dimensional matrix includes sampling frequency, spatial sampling interval, mileage offset, and data dimension direction; the target inspection mileage is determined by the mapping relationship between KP value, channel number, or geographical coordinates and submarine cable route.

[0009] Furthermore, the preset submarine cable state dynamic fingerprint database is constructed from numerical simulation, physical model test, historical distributed optical fiber inspection data, side-scan sonar data, multibeam bathymetry data, ROV image data, shallow seismic profile data, magnetic positioning data, artificial marine survey reports, or a combination thereof. The state fingerprint in the dynamic fingerprint database includes one or more of the following: frequency-space response map, low-frequency energy ratio, spatial coherence length, spectral peak frequency, spectral peak significance, spectral entropy, modal energy distribution, dominant mode width, environmental noise cross-correlation response, time-frequency energy evolution, cross-channel coherence matrix, and multi-time window stability index; wherein, the low-frequency energy ratio is determined by the ratio of the target low-frequency band energy to the total analysis band energy, the spatial coherence length is determined by the maximum continuous spatial distance of the cross-channel coherence coefficient exceeding a preset threshold, the spectral entropy is determined by the normalized spectral energy distribution, and the modal energy distribution is determined by the proportion of eigenvalues ​​or singular values ​​of the spatial covariance matrix, cross-correlation response matrix, or frequency-space characteristic matrix.

[0010] Furthermore, the dynamic fingerprint database is generated based on a finite element method, multibody dynamics, multiphysics coupling, or an equivalent cable-seabed coupling dynamic model. The model is constructed by changing one or more of the following: burial depth, exposed ratio, scour length, scour depth, soil stiffness, soil damping, seabed contact stiffness, friction coefficient, flow velocity, wave conditions, cable tension, bending stiffness, mass per unit length, and boundary constraints, to obtain the equivalent distributed optical fiber dynamic response under different burial degradation states. The equivalent distributed fiber dynamic response is obtained by mapping the submarine cable structure response to the fiber sampling interval, instrument sampling frequency and spatial resolution, and adding instrument noise, ocean background noise or channel coupling differences.

[0011] Furthermore, the extraction process of the measured dynamic features includes: The measured data segments in each space-time window are subjected to detrending, mean removal, standardization, bandpass filtering, short-time Fourier transform, wavelet transform, power spectrum estimation, cross-channel coherent estimation, or frequency band energy integration to obtain frequency domain, time-frequency domain, and spatial domain characteristics related to scouring exposure or buried support weakening.

[0012] Furthermore, the extraction of the measured dynamic features also includes: selecting reference spatial sampling points, performing environmental noise cross-correlation on the vibration time series of each spatial sampling point within the target window and the reference spatial sampling point, and forming a cross-correlation response matrix; The cross-correlation response matrix is ​​subjected to orthogonal mode decomposition, singular value decomposition, or principal component decomposition to obtain the dominant spatial mode, corresponding time coordinates, and mode energy proportion; and the spatial range of support weakening or exposure anomaly is determined based on the weighted energy profile of the dominant spatial mode.

[0013] Furthermore, the similarity matching between the measured dynamic features and the dynamic feature templates in the dynamic fingerprint database includes: The measured dynamic characteristics are constructed into feature vectors, feature matrices, or frequency-space feature maps; Cosine similarity, correlation coefficient, Euclidean distance, dynamic time regularization distance, Mahalanobis distance, probability classifier, support vector machine, random forest, partial least squares, neural network or combination thereof are used to calculate or classify the similarity with the state templates in the dynamic fingerprint database, and the embedding state degradation identification result is determined according to the state template with the highest matching score or the state category with the highest classification probability. The exposure probability and scour risk score are generated by weighting the similarity between measured dynamic features and multiple dynamic feature templates, the continuous length of the abnormal window, the degree of low-frequency energy enhancement, the change in spatial coherence length, the degree of modal energy concentration, the proportion of repeated occurrences in multiple time windows, and the overlap of marine survey data.

[0014] Furthermore, the dynamic abnormal length is obtained through a multi-scale spatial sliding window along the mileage direction of the coastal cable; when the state category or risk score of multiple adjacent windows meets the preset merging conditions, the multiple adjacent windows are merged into the same scour exposure or burial state degradation risk section.

[0015] Furthermore, the method also includes: for segments whose output state category is suspected to be suspended or whose modal energy is highly concentrated, identifying the family of natural frequencies that meet the harmonic consistency condition, and inverting the suspension length and equivalent axial tension based on the tension beam model; By minimizing the frequency residual between the observed natural frequency and the model-predicted natural frequency, the suspected suspended length and equivalent axial tension are solved within the geometric and tension boundaries. The reliability of the inversion results is judged based on the frequency residual, geometric boundary, tension boundary, Nyquist frequency, and the number of observation periods at the lowest frequency. The expression for the tension beam model is: in, Let be the nth natural frequency, L be the suspension length, T be the equivalent axial tension, μ be the mass per unit length, and EI be the bending stiffness.

[0016] A submarine cable status identification system based on dynamic fingerprint matching, the system comprising: A distributed optical fiber vibration sensing unit is used to acquire vibration time-series data from multiple spatial sampling points along the coastal cable. The data organization and preprocessing unit is used to form a two-dimensional spatial-temporal matrix and determine the target inspection mileage; The feature extraction unit is used to perform spatial sliding window and / or temporal window processing on the target inspection mileage and extract the measured dynamic features; The status fingerprint database unit is used to store the dynamic response characteristics under different burial, scouring, or exposed conditions. The fingerprint matching and risk assessment unit is used to match the measured dynamic features with the state fingerprint database and output the state category, exposure probability, scour risk score, dynamic anomaly length and confidence level. The inspection output unit is used to generate inspection results of the degradation status of submarine cable installation. The state fingerprint database unit includes a simulation database construction module, a measured sample database entry module, and a marine survey data annotation module; the simulation database construction module is configured to generate equivalent distributed optical fiber dynamic responses under different burial degradation states based on finite element, multibody dynamics, multiphysics coupling, or equivalent submarine cable-seabed coupling dynamic models. The inspection output unit is configured to output the risk map along the mileage, the distribution map of state categories, the frequency-spatial characteristic map, the list of abnormal sections, the overlap of marine survey data, the center deviation, the length error, the confidence level, the recommended review sections, and the optional tension beam inversion results of suspected completely suspended sections.

[0017] Compared with the prior art, the beneficial effects of the present invention include: 1. The submarine cable status identification of this invention is not limited to the identification of a single completely suspended state. Instead, it expands the identification object to the continuous degradation process of the submarine cable burial status. It can identify fully buried, shallow buried, semi-buried, partially exposed, scour pit developed, seabed support weakened, and suspected suspended states. It covers the entire evolution chain of submarine cables from stable burial to support failure, which is more in line with the actual situation in the field where partial exposure, shallow burial, semi-buried and support weakened coexist. It can provide accurate basis for differentiated operation and maintenance and risk classification management.

[0018] 2. This invention constructs a standardized dynamic fingerprint database based on the coupling dynamic mechanism of submarine cables and seabed. State discrimination is achieved by matching measured features with mechanistic templates, replacing the traditional approach that relies on threshold alarm logic based on single vibration energy and spectral peak values. This invention can effectively distinguish between far-field ship disturbances, fiber optic channel coupling differences, ocean background noise fluctuations, and actual seabed erosion support degradation, reducing the probability of false alarms. Furthermore, the fingerprint database can be constructed offline using methods such as simulation, without relying on a large number of field anomaly annotation samples.

[0019] 3. The fingerprint database of this invention can be constructed by combining numerical simulation, physical experiment, historical inspection data and marine survey data. It does not depend on specific simulation software or a single model, and is adaptable to different sea areas and different types of submarine cables. The solution has strong versatility and practicality.

[0020] 4. This invention retains the advantages of DAS inspection, such as spatial sliding window, spectrum analysis, cross-correlation, modal energy, and multi-time window stability, and can output risk maps along the mileage, anomaly length, and confidence level, which is convenient for engineering review.

[0021] 5. In this invention, complex physical quantities such as burial depth, soil parameters, sea state parameters, and submarine cable mechanical parameters are only used for constructing and updating the fingerprint database in the offline stage. In the online identification stage, only distributed optical fiber vibration data needs to be input. There is no need to collect the above complex parameters in real time to complete the status classification, risk scoring, and section location. This architecture reduces the computing power requirements for online monitoring and the cost of on-site data collection. It can be directly connected to the existing distributed optical fiber sensing system of in-service submarine cables without the need to install a large number of additional sensing devices. It is suitable for continuous online monitoring of the entire route of long-distance submarine cables.

[0022] 6. The present invention limits the tension beam inversion to a subordinate embodiment of a suspected fully suspended or strong modal response segment, thereby avoiding inappropriate suspended model interpretations of exposed or shallowly buried states that are still in contact with the seabed.

[0023] 7. This invention can generate traceable inspection and verification results by analyzing spatial overlap, center deviation, and length error with side-scan sonar, multibeam sonar, ROV, shallow profiling, magnetic survey, or marine survey reports. Attached Figure Description

[0024] Figure 1This is a flowchart of the method of the present invention.

[0025] Figure 2 This is a block diagram of the system structure of the present invention.

[0026] Figure 3 A schematic diagram of the construction of the submarine cable-seabed coupling dynamic fingerprint database.

[0027] Figure 4 This is a schematic diagram of the actual DAS frequency-space feature map and state fingerprint template matching.

[0028] Figure 5 This is a schematic diagram illustrating the dynamic abnormal length formed by merging multi-scale sliding windows along the mileage.

[0029] Figure 6 This is a schematic diagram illustrating the weak prior fusion and verification evaluation of marine survey data.

[0030] Figure 7 Diagram of marine survey verification and inspection output. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0032] Example 1 This embodiment discloses a submarine cable state identification method based on dynamic fingerprint matching, the method as follows: Figure 1 As shown, it includes steps S1-S9, as follows: Figure 3 As shown, submarine cables, under the influence of scouring, seabed migration, or changes in local support conditions, can exhibit various degradation states, including complete burial, shallow burial, semi-burial or partial exposure, weakened support due to scouring, and suspected suspension. This method identifies and assesses the risks associated with these continuous degradation states.

[0033] The specific steps are described below: Step S1, Data Acquisition and Spatiotemporal Matrix Construction: Vibration time-series data from multiple spatial sampling points along the submarine cable are obtained by using distributed optical fiber vibration sensing links laid along or accompanying the submarine cable; submarine cables include submarine optical cables and submarine optical-electric composite cables. The vibration time series data is organized into a two-dimensional space-time matrix based on spatial and temporal sampling points, and the data segment to be tested is constructed based on the target inspection mileage.

[0034] The distributed fiber optic vibration sensing link is a distributed acoustic sensing, distributed vibration sensing, phase-sensitive optical time-domain reflection or equivalent fiber optic vibration measurement link along the line. The vibration time series data includes strain, strain rate, phase change, vibration intensity and its transformation quantity along the line.

[0035] The space-time two-dimensional matrix includes sampling frequency, spatial sampling interval, mileage offset, and data dimension direction; the target inspection mileage is determined by the mapping relationship between KP value, channel number, or geographic coordinates and submarine cable route.

[0036] Step S2, Multi-scale sliding window and data preprocessing: The data segment to be tested is divided into multi-scale spatial sliding windows and / or time windows to obtain several spatial-time windows. Detrending, mean removal, standardization and / or bandpass filtering preprocessing are performed on the data in each window in sequence. An environmental noise cross-correlation matrix is ​​constructed by selecting reference spatial sampling points. The dominant spatial mode is obtained by singular value decomposition or orthogonal mode decomposition to locate the spatial range of support weakening and exposed anomalies.

[0037] Step S3, Extraction of measured dynamic features: For each space-time window, the measured dynamic features related to the cable-seabed support state are extracted. The measured dynamic features include at least one of the following: frequency-space response features, frequency band energy features, spatial coherence features, and modal energy distribution features.

[0038] The process of extracting measured dynamic characteristics includes: Short-time Fourier transform, wavelet transform, power spectrum estimation, cross-channel coherent estimation, or frequency band energy integration are performed on the data segments to be measured in each space-time window after preprocessing to obtain frequency domain, time-frequency domain, and spatial domain characteristics related to scouring of exposed surfaces or weakening by buried supports.

[0039] The extraction of measured dynamic characteristics specifically includes: selecting reference spatial sampling points, performing environmental noise cross-correlation on the vibration time series of each spatial sampling point within the target window and the reference spatial sampling point, and forming a cross-correlation response matrix; Orthogonal mode decomposition, singular value decomposition, or principal component decomposition are performed on the cross-correlation response matrix to obtain the dominant spatial mode, corresponding time coordinates, and mode energy proportions; and the spatial range of support weakening or exposure anomalies is determined based on the weighted energy profile of the dominant spatial mode.

[0040] Step S4, dynamic fingerprint database matching and discrimination: The preset submarine cable state dynamic fingerprint database is invoked, and the measured dynamic features are matched, classified, and regressed with the dynamic feature templates in the submarine cable state dynamic fingerprint database to obtain the state category, anomaly score, exposure probability, and scour risk score corresponding to each space-time window. The dynamic fingerprint database includes dynamic feature templates for fully buried, shallowly buried, partially buried, exposed, scoured, weakened support, and suspected suspended states.

[0041] The preset submarine cable state dynamic fingerprint database is constructed from numerical simulation, physical model test, historical distributed fiber optic inspection data, side-scan sonar data, multibeam bathymetry data, ROV image data, shallow seismic profile data, magnetic positioning data, manual marine survey reports, or a combination thereof.

[0042] The state fingerprints in the dynamic fingerprint database include one or more of the following: frequency-space response map, low-frequency energy ratio, spatial coherence length, spectral peak frequency, spectral peak significance, spectral entropy, modal energy distribution, dominant mode width, environmental noise cross-correlation response, time-frequency energy evolution, cross-channel coherence matrix, and multi-time window stability index.

[0043] Among them, the low-frequency energy ratio is determined by the ratio of the target low-frequency band energy to the total analysis band energy; the spatial coherence length is determined by the maximum continuous spatial distance of the cross-channel coherence coefficient exceeding the preset threshold; the spectral entropy is determined by the normalized spectral energy distribution; and the modal energy distribution is determined by the proportion of eigenvalues ​​or singular values ​​of the spatial covariance matrix, cross-correlation response matrix, or frequency-space characteristic matrix.

[0044] Specifically, the fingerprint database is not limited to any particular simulation software and can be generated from finite element, multibody dynamics, multiphysics coupling, or equivalent cable-seabed coupling dynamic models. It can also be supplemented by physical model experiments, historical DAS inspection data, side-scan sonar, multibeam sonar, ROV, shallow seismic profiling, magnetic positioning, or data annotated in manual marine survey reports. The fingerprint database includes response characteristics under conditions such as fully buried, shallowly buried, partially buried, partially exposed, scour-prone, weakened support, and suspected suspension.

[0045] In one numerical simulation-based database construction method, the dynamic fingerprint database is generated by using a dynamic model based on finite element method, multibody dynamics, multiphysics coupling, or equivalent submarine cable-seabed coupling. The model obtains the equivalent distributed optical fiber dynamic response under different burial degradation states by changing one or more of the following: burial depth, exposed ratio, scour length, scour depth, soil stiffness, soil damping, seabed contact stiffness, friction coefficient, flow velocity, wave conditions, submarine cable tension, bending stiffness, mass per unit length, and boundary constraints, thereby constructing the dynamic fingerprint database.

[0046] It should be noted that parameters such as burial depth, exposed ratio, scour length, scour depth, soil stiffness, damping, flow velocity, wave conditions, cable tension, bending stiffness, and mass per unit length are mainly used for offline construction, expansion, or correction of the dynamic fingerprint database. The online identification stage of this invention does not require real-time measurement of all the above parameters. Instead, it uses the already constructed dynamic fingerprint database to match the measured dynamic characteristics of the distributed optical fiber, thereby obtaining the cable condition category and risk assessment results.

[0047] The equivalent distributed fiber dynamic response is obtained by mapping the submarine cable structure response to the fiber sampling interval, instrument sampling frequency, and spatial resolution, and adding instrument noise, ocean background noise, or channel coupling differences.

[0048] Similarity matching between measured dynamic features and dynamic feature templates in the dynamic fingerprint database includes: The measured dynamic characteristics are constructed into eigenvectors, eigenmatrices, or frequency-space feature maps; Cosine similarity, correlation coefficient, Euclidean distance, dynamic time warping distance, Mahalanobis distance, probability classifier, support vector machine, random forest, partial least squares, neural network or combination thereof are used to calculate or classify the similarity with the state templates in the dynamic fingerprint database, and the buried state degradation identification result is determined according to the state template with the highest matching score or the state category with the highest classification probability. Exposure probability and scour risk scores are generated by weighting the similarity between measured dynamic features and multiple dynamic feature templates, the continuous length of anomaly windows, the degree of low-frequency energy enhancement, the change in spatial coherence length, the degree of modal energy concentration, the proportion of repeated occurrences in multiple time windows, and the overlap of marine survey data. in It is a normalized indicator. It's the weight.

[0049] In a preferred embodiment, the vibration data along the line is represented as a two-dimensional space-time matrix. ,in The number of spatial sampling points. The number of time sampling points, matrix elements Indicates the first The spatial sampling point at the th Vibration response, strain, strain rate, or phase change at each time sampling point. For any space-time window within the target inspection mileage. Extracting feature vectors: in, For low-frequency energy ratio, The spatial coherence length, For spectral entropy, The main spectral peak frequency. For the significance of the spectral peak, For the first The proportion of energy in each mode. These are multi-time-window stability indicators. The above characteristics are used to characterize the differences in dynamic response of submarine cables as they degrade from fully buried to shallow buried, semi-buried, partially exposed, weakened by scour support, or suspected suspended state.

[0050] Step S5, Merging and merging abnormal segments: Adjacent spatial-temporal windows that meet the preset consistency condition for state category or whose anomaly score, exposure probability, or scour risk score meets the preset threshold condition are merged into submarine cable state anomaly segments, and their dynamic anomaly length is calculated.

[0051] Step S6, Confidence of multi-time-window stability correction: Within the same mileage segment, multiple sets of measured dynamic features are extracted according to the preset time window length and time step, and the repetition rate of state categories, stability of risk scores, repetition rate of spectral peaks, shift of anomaly center, and fluctuation of anomaly length are statistically analyzed to correct the identification confidence of each segment.

[0052] Step S7, Risk value of marine survey data fusion and correction: The degradation risk sections identified by distributed optical fibers are spatially aligned with exposed, scoured, or shallowly buried sections in side-scan sonar, multibeam sonar, ROV, shallow seismic profiling, magnetic positioning, or artificial marine survey reports. Spatial overlap, center deviation, length error, and risk level consistency are calculated, and scour risk scores are corrected or inspection verification results are generated.

[0053] Step S8, quantitative inversion of suspected suspended sections: For segments whose output state category is suspected to be suspended or where modal energy is highly concentrated, the family of natural frequencies that meet the harmonic consistency condition is identified, and the suspension length and equivalent axial tension are inverted based on the tension beam model.

[0054] By minimizing the frequency residual between the observed natural frequency and the model-predicted natural frequency, the suspected suspended length and equivalent axial tension are solved within the geometric and tension boundaries. The reliability of the inversion results is then judged based on the frequency residual, geometric boundary, tension boundary, Nyquist frequency, and the number of observation periods at the lowest frequency.

[0055] The expression for the tension beam model is: in, Let be the nth natural frequency, L be the suspension length, T be the equivalent axial tension, μ be the mass per unit length, and EI be the bending stiffness.

[0056] Step S9, Inspection results output: Output the start and end mileage, dynamic anomaly length, status category, risk level, marine survey matching error, and on-site verification suggestions for abnormal submarine cable sections. Optional output of inversion parameters for suspended sections completes submarine cable status identification. Status categories include at least one of the following: fully buried, shallowly buried, partially buried, partially exposed, scour pit developed, seabed support weakened, and suspected suspended.

[0057] The dynamic abnormal length is obtained through a multi-scale spatial sliding window along the mileage of the coastal cable; when the state category or risk score of multiple adjacent windows meets the preset merging conditions, the multiple adjacent windows are merged into the same scour exposed or buried state degradation risk section.

[0058] Example 2 This embodiment discloses a submarine cable status identification system based on dynamic fingerprint matching, based on Embodiment 1 above. The system is as follows: Figure 2 The following are included: A distributed optical fiber vibration sensing unit is used to acquire vibration time-series data from multiple spatial sampling points along the coastal cable. The data organization and preprocessing unit is used to form a two-dimensional spatial-temporal matrix and determine the target inspection mileage; The feature extraction unit is used to perform spatial sliding window and / or temporal window processing on the target inspection mileage and extract the measured dynamic features; The status fingerprint database unit is used to store the dynamic response characteristics under different burial, scouring, or exposed conditions. The fingerprint matching and risk assessment unit is used to match the measured dynamic features with the state fingerprint database and output the state category, exposure probability, scour risk score, dynamic anomaly length and confidence level. The inspection output unit is used to generate inspection results of the degradation status of submarine cable burial.

[0059] The state fingerprint database unit includes a simulation database construction module, a measured sample database entry module, and a marine survey data annotation module. The simulation database construction module is configured to generate equivalent distributed optical fiber dynamic responses under different burial degradation states based on finite element, multibody dynamics, multiphysics coupling, or equivalent submarine cable-seabed coupling dynamic models.

[0060] The inspection output unit is configured to output the risk map along the mileage, the distribution map of the state category, the frequency-spatial characteristic map, the list of abnormal sections, the overlap of marine survey data, the center deviation, the length error, the confidence level, the recommended review sections, and the optional tension beam inversion results of suspected completely suspended sections.

[0061] For details regarding the above modules, please refer to the relevant descriptions and effects in Example 1 for further understanding.

[0062] Example 3 This embodiment, based on embodiments 1 and 2 above, discloses an embodiment for identifying the state degradation of in-service DAS data.

[0063] In this embodiment, a distributed fiber optic vibration sensing link continuously collects vibration data along the coastal cable. The data organization and preprocessing unit reads the sampling frequency, spatial sampling interval, and mileage offset, converting the data into a channel-time or time-channel two-dimensional matrix. Inspection personnel can select the section to be tested according to KP mileage, channel number, or geographical coordinates; the system can also automatically perform a sliding window scan of the entire route. Figure 5 As shown, the system divides DAS space-time vibration data into multiple spatial and temporal windows, and extracts frequency-space features, frequency band energy features, spatial coherence features, and modal energy distribution features at different scales.

[0064] The feature extraction unit performs detrending, mean removal, standardization, and necessary frequency band processing on each spatial window, and then extracts the low-frequency energy ratio, frequency band energy distribution, spectral peak frequency, spectral peak significance, spectral entropy, cross-channel coherence length, and frequency-spatial energy map. For windows with coherent responses, an environmental noise cross-correlation matrix can also be constructed, and the dominant spatial modes and mode energy distribution can be obtained through orthogonal mode decomposition.

[0065] The fingerprint matching and risk assessment unit matches the measured features with state templates in the dynamic fingerprint database. For example... Figure 6 As shown, the system inputs the measured dynamic features into the state fingerprint matching engine and calls the preset dynamic fingerprint database for similarity matching, classification, or regression to obtain the state category, matching score, confidence level, and risk score. If the measured features are closer to the fully embedded template, a low risk is output; if the low-frequency energy is enhanced, the spatial coherence length is increased, and the feature map is similar to the shallowly embedded or semi-embedded template, a shallowly embedded or semi-embedded risk is output; if abnormal energy appears continuously along a certain length and is similar to the locally exposed or scour pit template, a locally exposed or scour risk is output; if there is a concentration of strong modal energy and a family of harmonic peaks, a suspected suspended or strong modal response prompt is output.

[0066] In one embodiment, the measured dynamic characteristics can be calculated as follows.

[0067] First, for the target window The vibration data within the area are analyzed by spectrum or time-frequency analysis to obtain the power spectral density. Let the target low-frequency band be... The total analysis bandwidth is The low-frequency energy ratio is defined as: This index is used to characterize the degree of enhancement in low-frequency vibration response after submarine cable support is weakened or exposed.

[0068] Secondly, the correlation coefficient or coherence coefficient is calculated for the vibration response between spatial sampling points. Let... Indicates spatial interval as The average coherence coefficient between channels, If a preset coherence threshold is set, then the spatial coherence length is defined as: in, Used to characterize the continuous range of abnormal responses along the mileage direction of coastal cables.

[0069] Next, the normalized spectral energy within the target window is expressed as: The spectral entropy is defined as: in, It is used to characterize the degree of concentration of spectral energy. The lower the spectral entropy, the more concentrated the energy is in a few frequency components; the higher the spectral entropy, the closer the response is to a broadband random perturbation.

[0070] For windows with spatially coherent responses, a spatial covariance matrix or cross-correlation response matrix can be constructed, followed by eigenvalue decomposition or singular value decomposition. Let the... The eigenvalues ​​corresponding to the first mode are Then the first The energy proportion of the first mode is: in, Used to characterize whether the dynamic response of the target section exhibits concentrated modal characteristics.

[0071] Example 4 This embodiment, based on Embodiments 1 and 2 above, discloses a numerical simulation construction of a dynamic fingerprint library.

[0072] In this embodiment, the system constructs a submarine cable state dynamic fingerprint database offline. For example... Figure 4 As shown, this embodiment constructs a submarine cable state dynamic fingerprint database offline. The fingerprint database is used to provide dynamic templates for different burial states, exposed states, scour support weakening states, and suspected suspended states for online DAS measured feature matching. It does not require real-time measurement of all seabed, sea conditions, and submarine cable mechanical parameters during the online identification stage.

[0073] Specifically, a coupled dynamic model of the submarine cable and seabed is established. This model can employ finite element methods, multibody dynamics, multiphysics coupling models, or equivalent simplified dynamic models, without limitation to specific commercial software or solver types. In the model, the submarine cable can be represented by beam elements, cable elements, cable-pipe composite elements, or equivalent continuum; the seabed support can be represented by distributed spring-damped elements, contact elements, friction boundaries, equivalent soil stiffness, or combinations thereof; and environmental excitation can be represented by ocean currents, waves, random background excitation, or equivalent broadband excitation.

[0074] In the modeling process, parameters can be divided into three categories: basic parameters, state parameters, and environmental and boundary parameters. Basic parameters include one or more of the following: cable outer diameter, mass per unit length, bending stiffness, axial stiffness, damping ratio, fiber optic spatial sampling interval, instrument sampling frequency, and measurement length. State parameters include one or more of the following: burial depth, exposed proportion, scour length, scour depth, support weakening length, seabed contact range, and suspected suspended span. Environmental and boundary parameters include one or more of the following: soil stiffness, soil damping, contact stiffness, friction coefficient, flow velocity, wave height, wave period, initial cable tension, and boundary constraint type.

[0075] It should be noted that the above parameters are used for offline construction, expansion, or correction of the dynamic fingerprint database, and are not input parameters that must be collected in real time during the online identification phase. When the target sea area lacks measured parameters such as soil stiffness, soil damping, current velocity, wave, or in-situ tension of submarine cables, multiple sets of state fingerprint templates can be generated using engineering design values, historical inspection data, marine survey reports, typical seabed categories, empirical value ranges, or parameter hierarchical combinations.

[0076] In one implementation, one or more sets of parameter combinations are set according to state categories such as fully buried, shallowly buried, partially buried, partially exposed, scour-prone, weakened seabed support, and suspected suspension. For each set of parameter combinations, the displacement, velocity, acceleration, strain, strain rate, curvature, or their equivalent response of the submarine cable under environmental excitation are calculated. Subsequently, the structural dynamic response is equivalently mapped according to the spatial sampling interval, measurement length, and sampling frequency of the distributed optical fiber to generate an equivalent DAS response consistent with the measured distributed optical fiber data scale. In one embodiment, each state fingerprint record in the dynamic fingerprint database includes: state label, simulation or experimental parameters, equivalent DAS response, feature vector, applicable conditions, and confidence weight. Each state fingerprint record can be represented as... ,in For state categories, For model parameter combinations, As a dynamic characteristic template, This refers to template weights or applicable confidence levels.

[0077] Example 5 This embodiment, based on Embodiments 1 and 2 above, discloses a method for marine survey data verification and weak prior fusion.

[0078] In this embodiment, the system accesses one or more types of marine survey data, including side-scan sonar, multibeam sonar, ROV imagery, shallow seismic profiling, magnetic positioning, manual marine survey reports, or historical inspection records. For example... Figure 7 As shown, by using the submarine cable route coordinates, KP mileage, or channel-mileage mapping relationship, the system can spatially align the abnormal sections identified by DAS with the exposed, shallowly buried, scoured, or suspected suspended marked sections in the marine survey data, and output the spatial overlap, center deviation, length error, corrected confidence, and inspection and verification suggestions.

[0079] After spatial alignment, the system calculates the spatial overlap, center deviation, length error, and state consistency between the DAS-identified segment and the marine survey-marked segment. Among them, the spatial overlap can be determined by the ratio of the intersection length to the union length of the two types of segments, the center deviation can be determined by the difference in the center mileage of the two types of segments, and the length error can be determined by the difference between the DAS dynamic anomaly length and the marine survey mark length.

[0080] Marine survey data can be used as weak priors in the identification process, and also as posterior verification to correct the confidence level of the DAS identification results. When high-risk DAS sections and abnormal marine survey sections highly overlap and are in the same state, the confidence level of that section is increased; when there is a significant deviation between the two, a suggested review section is output to indicate possible data time differences, changes in seabed state, positioning errors, or insufficient environmental stimulation. Finally, the system outputs the DAS identified section, the marine survey marked section, the corrected confidence level, and inspection and review suggestions.

[0081] Example 6 This embodiment, based on Embodiments 1 and 2 above, discloses a tension beam inversion method for a suspected completely suspended section.

[0082] In a few sections, if the measured DAS characteristics show significant spatial modal energy concentration, and at least two intrinsic frequency peaks satisfying harmonic consistency appear in the dominant time coordinate spectrum, then this section can be marked as a suspected completely suspended or strong modal response section. For such sections, the system can further use a tension beam model to invert the suspension length. and equivalent axial tension .

[0083] In one implementation, the first The first natural frequency can be expressed as: in, For bending stiffness, The mass per unit length. Under geometric and tension boundary constraints, the system is solved by minimizing the residual between the observed frequencies and the model predicted frequencies. and .

[0084] The inversion process includes unit consistency verification, geometric boundary setting, tension boundary setting, frequency residual evaluation, boundary fit judgment, Nyquist frequency judgment, and minimum frequency period count judgment. When the minimum natural frequency observation period count is insufficient, the inversion length is close to the upper limit of the spatial window, the frequency residual exceeds the threshold, the tension solution is close to the boundary, the model predicted frequency exceeds the Nyquist frequency, or the number of harmonic peaks is insufficient, the system reduces the level of the suspended inversion result and uses it only as evidence of candidate modes, rather than as the sole basis for identifying exposure or erosion degradation.

[0085] The inversion process includes unit consistency verification, geometric boundary setting, tension boundary setting, frequency residual evaluation, boundary fit judgment, Nyquist frequency judgment, and minimum frequency cycle number judgment. If the inversion result does not meet the reliability gating, it is retained only as candidate modal evidence and not used as the sole basis for identifying exposure or erosion degradation. When any of the following conditions are met, the system lowers the level of the suspended inversion result and uses it only as candidate modal evidence, not as the sole basis for identifying exposure or erosion degradation: 1) The number of observation periods within the analysis time window for the lowest natural frequency is less than the preset threshold; 2) The inversion length is close to the upper limit of the spatial window; 3) The frequency residual is greater than the preset threshold; 4) The tension solution is close to the tension boundary; 5) The model's prediction frequency exceeds the Nyquist frequency; 6) Insufficient number of harmonic peaks.

[0086] Example 7 This embodiment, based on Embodiments 1 and 2 above, discloses a simplified state recognition under parameter missing conditions.

[0087] In this embodiment, when the target sea area lacks measured parameters such as soil stiffness, soil damping, ocean currents, waves, or in-situ tension of submarine cables, the system does not require real-time measurement of all of the above parameters. Instead, it constructs a dynamic fingerprint template using engineering design values, historical inspection data, marine survey reports, typical seabed categories, or preset parameter ranges.

[0088] In the online identification phase, the system forms a data segment to be tested based on distributed optical fiber vibration data, sampling frequency, spatial sampling interval, and channel-mileage mapping relationship, and extracts measured dynamic features such as low-frequency energy ratio, spatial coherence length, spectral entropy, frequency-space characteristics, and modal energy distribution. Subsequently, the measured dynamic features are matched with a preset dynamic fingerprint database to output the state category, abnormal segment, dynamic anomaly length, risk level, and confidence level.

[0089] Therefore, this invention can still be implemented even in the absence of complete seabed, sea state, and submarine cable mechanical parameters. The complex physical parameters are mainly used to improve the accuracy and confidence level of the dynamic fingerprint database, and are not necessary inputs for online status identification.

[0090] 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 person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for identifying the state of submarine cables based on dynamic fingerprint matching, characterized in that, The method includes: Vibration time-series data from multiple spatial sampling points along the submarine cable are acquired using distributed optical fiber vibration sensing links laid along or accompanying the submarine cable; the submarine cable includes submarine optical cable and submarine optical-electric composite cable. The vibration time series data is organized into a two-dimensional space-time matrix based on spatial and temporal sampling points, and the data segment to be tested is constructed based on the target inspection mileage. The data segment to be tested is divided into multi-scale spatial sliding windows and / or time windows to obtain several spatial-time windows. For each spatial-time window, the measured dynamic features related to the cable-seabed support state are extracted. The measured dynamic features include at least one of the following: frequency-space response features, frequency band energy features, spatial coherence features, and modal energy distribution features. A preset submarine cable state dynamic fingerprint database is invoked, and the measured dynamic features are matched, classified, and regressed with the dynamic feature templates in the submarine cable state dynamic fingerprint database to obtain the state category, anomaly score, exposure probability, and scour risk score corresponding to each space-time window; the dynamic fingerprint database includes dynamic feature templates for different burial states, exposure states, scour states, support weakening states, and suspected suspended states. The spatial-temporal windows that are adjacent and whose status categories meet the preset status category consistency condition or whose anomaly scores, exposure probabilities or scour risk scores meet the preset threshold conditions are merged into submarine cable status anomaly segments and their dynamic anomaly lengths are calculated. Within the same mileage segment, multiple sets of measured dynamic features are extracted according to the preset time window length and time step, and the repetition ratio of state categories, stability of risk scores, repetition rate of spectral peaks, shift of anomaly center, and fluctuation of anomaly length are statistically analyzed to correct the identification confidence of each segment. The degradation risk sections identified by distributed optical fibers are spatially aligned with exposed, scoured, or shallowly buried sections in side-scan sonar, multibeam sonar, ROV, shallow seismic profiling, magnetic positioning, or artificial marine survey reports. Spatial overlap, center deviation, length error, and risk level consistency are calculated, and the scour risk score is corrected or inspection verification results are generated. Output the start and end mileage, dynamic abnormal length, status category, risk level, and confidence level of the abnormal section of the submarine cable to complete the submarine cable status identification; the status category includes at least one of the following: fully buried, shallowly buried, partially buried, partially exposed, scour pit developed, seabed support weakened, and suspected of being suspended.

2. The method for identifying the state of a submarine cable based on dynamic fingerprint matching according to claim 1, characterized in that, The distributed optical fiber vibration sensing link is a distributed acoustic sensing, distributed vibration sensing, phase-sensitive optical time-domain reflection or equivalent optical fiber vibration measurement link along the line. The vibration time series data includes strain along the line, strain rate, phase change, vibration intensity and its transformation quantity. The space-time two-dimensional matrix includes sampling frequency, spatial sampling interval, mileage offset, and data dimension direction; the target inspection mileage is determined by the mapping relationship between KP value, channel number, or geographical coordinates and submarine cable route.

3. The method for identifying the state of a submarine cable based on dynamic fingerprint matching according to claim 1, characterized in that, The preset submarine cable state dynamic fingerprint database is constructed from numerical simulation, physical model test, historical distributed optical fiber inspection data, side-scan sonar data, multibeam bathymetry data, ROV image data, shallow seismic profile data, magnetic positioning data, manual marine survey reports, or a combination thereof. The state fingerprint in the dynamic fingerprint database includes one or more of the following: frequency-space response map, low-frequency energy ratio, spatial coherence length, spectral peak frequency, spectral peak significance, spectral entropy, modal energy distribution, dominant mode width, environmental noise cross-correlation response, time-frequency energy evolution, cross-channel coherence matrix, and multi-time window stability index; wherein, the low-frequency energy ratio is determined by the ratio of the target low-frequency band energy to the total analysis band energy, the spatial coherence length is determined by the maximum continuous spatial distance of the cross-channel coherence coefficient exceeding a preset threshold, the spectral entropy is determined by the normalized spectral energy distribution, and the modal energy distribution is determined by the proportion of eigenvalues ​​or singular values ​​of the spatial covariance matrix, cross-correlation response matrix, or frequency-space characteristic matrix.

4. The method for identifying the state of a submarine cable based on dynamic fingerprint matching according to claim 1, characterized in that, The dynamic fingerprint database is generated by using a dynamic model based on finite element method, multibody dynamics, multiphysics coupling, or equivalent cable-seabed coupling. The model is constructed by changing one or more of the following: burial depth, exposed ratio, scour length, scour depth, soil stiffness, soil damping, seabed contact stiffness, friction coefficient, flow velocity, wave conditions, cable tension, bending stiffness, mass per unit length, and boundary constraints, to obtain the equivalent distributed optical fiber dynamic response under different burial degradation states. The equivalent distributed fiber dynamic response is obtained by mapping the submarine cable structure response to the fiber sampling interval, instrument sampling frequency and spatial resolution, and adding instrument noise, ocean background noise or channel coupling differences.

5. The method for identifying the state of a submarine cable based on dynamic fingerprint matching according to claim 1, characterized in that, The extraction process of the measured dynamic features includes: The measured data segments in each space-time window are subjected to detrending, mean removal, standardization, bandpass filtering, short-time Fourier transform, wavelet transform, power spectrum estimation, cross-channel coherent estimation, or frequency band energy integration to obtain frequency domain, time-frequency domain, and spatial domain characteristics related to scouring exposure or buried support weakening.

6. The method for identifying the state of a submarine cable based on dynamic fingerprint matching according to claim 1, characterized in that, The extraction of the measured dynamic features also includes: selecting reference spatial sampling points, performing environmental noise cross-correlation on the vibration time series of each spatial sampling point in the target window and the reference spatial sampling point to form a cross-correlation response matrix; The cross-correlation response matrix is ​​subjected to orthogonal mode decomposition, singular value decomposition, or principal component decomposition to obtain the dominant spatial mode, corresponding time coordinates, and mode energy proportion; and the spatial range of support weakening or exposure anomaly is determined based on the weighted energy profile of the dominant spatial mode.

7. The method for identifying the state of a submarine cable based on dynamic fingerprint matching according to claim 1, characterized in that, The similarity matching between the measured dynamic features and the dynamic feature templates in the dynamic fingerprint database includes: The measured dynamic characteristics are constructed into feature vectors, feature matrices, or frequency-space feature maps; Cosine similarity, correlation coefficient, Euclidean distance, dynamic time regularization distance, Mahalanobis distance, probability classifier, support vector machine, random forest, partial least squares, neural network or combination thereof are used to calculate or classify the similarity with the state templates in the dynamic fingerprint database, and the embedding state degradation identification result is determined according to the state template with the highest matching score or the state category with the highest classification probability. The exposure probability and scour risk score are generated by weighting the similarity between measured dynamic features and multiple dynamic feature templates, the continuous length of the abnormal window, the degree of low-frequency energy enhancement, the change in spatial coherence length, the degree of modal energy concentration, the proportion of repeated occurrences in multiple time windows, and the overlap of marine survey data.

8. The method for identifying the state of a submarine cable based on dynamic fingerprint matching according to claim 1, characterized in that, The dynamic abnormal length is obtained through a multi-scale spatial sliding window along the mileage direction of the coastal cable; when the state category or risk score of multiple adjacent windows meets the preset merging conditions, the multiple adjacent windows are merged into the same scour exposure or burial state degradation risk section.

9. The method for identifying the state of a submarine cable based on dynamic fingerprint matching according to claim 1, characterized in that, The method further includes: for segments whose output state category is suspected to be suspended or whose modal energy is highly concentrated, identifying the family of natural frequencies that meet the harmonic consistency condition, and inverting the suspension length and equivalent axial tension based on the tension beam model; By minimizing the frequency residual between the observed natural frequency and the model-predicted natural frequency, the suspected suspended length and equivalent axial tension are solved within the geometric and tension boundaries. The reliability of the inversion results is judged based on the frequency residual, geometric boundary, tension boundary, Nyquist frequency, and the number of observation periods at the lowest frequency. The expression for the tension beam model is: in, Let be the nth natural frequency, L be the suspension length, T be the equivalent axial tension, μ be the mass per unit length, and EI be the bending stiffness.

10. A submarine cable status identification system based on dynamic fingerprint matching, characterized in that, The system includes: A distributed optical fiber vibration sensing unit is used to acquire vibration time-series data from multiple spatial sampling points along the coastal cable. The data organization and preprocessing unit is used to form a two-dimensional spatial-temporal matrix and determine the target inspection mileage; The feature extraction unit is used to perform spatial sliding window and / or temporal window processing on the target inspection mileage and extract the measured dynamic features; The status fingerprint database unit is used to store the dynamic response characteristics under different burial, scouring, or exposed conditions. The fingerprint matching and risk assessment unit is used to match the measured dynamic features with the state fingerprint database and output the state category, exposure probability, scour risk score, dynamic anomaly length and confidence level. The inspection output unit is used to generate inspection results of the degradation status of submarine cable burial. The state fingerprint database unit includes a simulation database construction module, a measured sample database entry module, and a marine survey data annotation module; the simulation database construction module is configured to generate equivalent distributed optical fiber dynamic responses under different burial degradation states based on finite element, multibody dynamics, multiphysics coupling, or equivalent submarine cable-seabed coupling dynamic models. The inspection output unit is configured to output the risk map along the mileage, the distribution map of state categories, the frequency-spatial characteristic map, the list of abnormal sections, the overlap of marine survey data, the center deviation, the length error, the confidence level, the recommended review sections, and the optional tension beam inversion results of suspected completely suspended sections.

Images