Submarine cable fault detection system

Abstract

The invention relates to the technical field of submarine cable fault detection, in particular to a submarine cable fault detection system. The method comprises the steps that an adjustable low-frequency pulse injection unit injects a low-frequency pulse excitation signal and collects a response signal generated by a submarine cable fault; the multi-source signal fusion acquisition unit acquires a reflected voltage signal, an acoustic signal and a leakage magnetic field signal, and constructs a multi-modal fusion response characteristic matrix; the intelligent broadband noise suppression unit is used for performing noise suppression processing on each modal sub-matrix to obtain a de-noised multi-modal fusion response characteristic matrix; and the intelligent fault positioning interpretation unit is used for constructing a three-dimensional signal fingerprint spectrogram, performing classification analysis on fault events based on a fuzzy clustering algorithm and a depth feature matching model to obtain a fault type, and performing calculation through a path residual fitting method to obtain a fault position. The invention realizes a detection system which does not depend on a special cable structure, can actively excite and fuse multi-source signals, and can position submarine cable faults at high precision.

Description

Technical Field

[0001] The invention belongs to the technical field of submarine cable fault detection, and in particular relates to a submarine cable fault detection system. Background Art

[0002] Submarine cables are critical infrastructure for marine power transmission and communications systems. Their operational status is directly related to the stability of long-distance power supply and the security of transoceanic communications. Because cables are constantly exposed to high voltage, high humidity, high voltage differentials, and corrosive seawater, they are prone to failures such as insulation aging, mechanical damage, partial discharge, and core breakage. Rapidly and accurately detecting cable anomalies and precisely locating the fault point in the early stages of a fault is a crucial technical issue for ensuring the stable operation of submarine cable systems. Currently, fault detection technologies primarily rely on methods such as optical time-domain reflectometry, distributed fiber optic sensing, low-frequency injection, electrical parameter monitoring, or manually towed ultrasonic / electromagnetic scanning. However, these methods still face challenges in detecting early-stage faults over large areas, long distances, and without breakage. Existing submarine cable fault detection technologies still have some difficult-to-solve problems: most methods rely on pre-buried optical fibers or special structures in the cable itself, which limits their adaptability to old cables or communication / power hybrid cables; passive detection methods have difficulty capturing short-term dynamic fault signals in a timely manner, and cannot achieve active response induction and high-sensitivity detection; single-modal signal acquisition methods (such as only electrical signals or only acoustic signals) have difficulty coping with multi-path propagation, nonlinear responses and background noise interference in complex submarine environments; the problems in the above three aspects will lead to the inability to perceive the initial state of the fault in a timely manner during submarine cable fault detection, which is prone to misjudgment or missed detection, and it is difficult to effectively identify and accurately locate the fault type in a complex submarine environment. Summary of the Invention

[0003] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a submarine cable fault detection system to solve the problems raised in the prior art, such as the inability to timely perceive the initial state of the fault during submarine cable fault detection, the susceptibility to misjudgment or missed detection, and the difficulty in effectively identifying and accurately locating the fault type in a complex submarine environment.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions: A submarine cable fault detection system comprising: An adjustable low-frequency pulse injection unit is used to inject a frequency-adjustable low-frequency pulse excitation signal into the submarine cable to stimulate the electrical response, acoustic response, and magnetic response signals generated by the submarine cable fault; The multi-source signal fusion acquisition unit is used to collect the reflected voltage signal in the electrical response, the acoustic signal in the acoustic response, and the leakage magnetic field signal in the magnetic response signal; and synchronize the collected signals, align the timestamps, and perform modal fusion to obtain a multi-modal fusion response feature matrix; An intelligent broadband noise suppression unit is used to perform noise suppression processing on each modal sub-matrix in the multimodal fusion response feature matrix to obtain a denoised multimodal fusion response feature matrix; The intelligent fault location and interpretation unit is used to extract the fault-related modal submatrix from the denoised multimodal fusion response feature matrix and construct a three-dimensional signal fingerprint spectrum; based on the three-dimensional signal fingerprint spectrum, the fault event is classified and analyzed to obtain the fault type and fault location.

[0005] A further improvement of the present invention is: Preferably, the adjustable low pulse injection unit includes a signal generation module, a parameter control module and an excitation output interface module; The signal generation module is used to generate a periodic voltage signal within a set frequency range; the parameter control module is used to adjust the frequency, amplitude and duty cycle of the low-frequency pulse excitation signal according to a preset excitation configuration table; and the excitation output interface module is used to output the low-frequency pulse excitation signal.

[0006] Preferably, the parameter control module calibrates the output frequency in real time through a closed-loop feedback method based on the feedback from the multi-source signal fusion acquisition unit.

[0007] Preferably, the low-frequency pulse excitation signal is a continuous periodic sinusoidal signal mode, a step voltage signal mode or a single pulse signal mode.

[0008] Preferably, the multi-source signal fusion acquisition unit includes a voltage signal acquisition module, an acoustic signal acquisition module, a magnetic field signal acquisition module and a feature fusion processing module; The voltage signal acquisition module is used to acquire the reflected voltage signal in the electrical response; the acoustic signal acquisition module is used to acquire the acoustic signal in the acoustic response; the magnetic field signal acquisition module is used to acquire the leakage magnetic field signal in the magnetic response; The feature fusion processing module is used to uniformly calibrate the clock and align the timestamps of the reflected voltage signal, acoustic signal and leakage magnetic field signal, and after normalization and standardization, encode the standardized signal into a feature vector form, and summarize the feature vector according to the time series dimension, signal type dimension and spatial channel dimension to obtain a multimodal fusion response feature matrix.

[0009] Preferably, the intelligent broadband noise suppression unit includes a multimodal decomposition processing module, a time-frequency noise reduction module and a frequency calibration module; The multimodal decomposition processing module is used to perform dimensional decomposition processing on each modal submatrix in the multimodal fusion response feature matrix according to the signal type dimension to obtain the separated modal submatrices; the time-frequency noise reduction module is used to perform layered filtering on the noise components of different frequency bands in each modal submatrix; the frequency calibration module is used to detect the frequency deviation between the adjustable low-frequency pulse injection signal and each response signal, and dynamically adjust the feature matrix based on the frequency deviation to obtain the denoised multimodal fusion response feature matrix.

[0010] Preferably, the time-frequency noise reduction module uses an algorithm combining wavelet multi-scale decomposition and empirical mode decomposition to perform hierarchical filtering; The frequency calibration module dynamically adjusts the characteristic matrix through a frequency drift compensation algorithm.

[0011] Preferably, the intelligent fault location and interpretation unit includes a modal extraction module, a feature fingerprint construction module and a fault identification and location module; The modal extraction module is used to extract the abnormal response modal submatrix from the denoised multimodal fusion response feature matrix; the feature fingerprint construction module is used to construct a three-dimensional signal fingerprint spectrum of the fusion fault event based on the abnormal response modal submatrix; the fault identification and positioning module is used to classify and analyze the fault event to obtain the fault type and fault location.

[0012] Preferably, the three-dimensional signal fingerprint spectrum is constructed by using the time axis, signal type axis and signal amplitude axis as three orthogonal dimensions; The fault identification and positioning module uses a fuzzy clustering algorithm and a deep feature matching model to classify and analyze fault events to obtain the fault type, and calculates the fault location through a path residual fitting method.

[0013] A fault detection method of the above-mentioned submarine cable fault detection system comprises the following steps: S1, injects a frequency-adjustable low-frequency pulse excitation signal into the submarine cable to stimulate the electrical response, acoustic response and magnetic response signals generated by the submarine cable fault; S2, collects the reflected voltage signal in the electrical response, the acoustic signal in the acoustic response, and the leakage magnetic field signal in the magnetic response signal; and performs synchronization marking, time stamp alignment, and modal fusion on the collected signals to obtain a multimodal fusion response feature matrix; S3, performing noise suppression processing on each modal sub-matrix in the multimodal fusion response feature matrix to obtain a denoised multimodal fusion response feature matrix; S4, extract the fault-related modal submatrix from the denoised multimodal fusion response feature matrix and construct a three-dimensional signal fingerprint spectrum; based on the three-dimensional signal fingerprint spectrum, classify and analyze the fault events to obtain the fault type and fault location.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present invention discloses a submarine cable fault detection system. The system first injects a low-frequency pulse excitation signal into the cable through an adjustable low-frequency pulse injection unit, stimulating the electrical, acoustic, and magnetic multimodal responses generated by the fault. A multi-source signal fusion acquisition unit simultaneously collects reflected voltage, acoustic signals, and leakage magnetic field signals, and generates a multimodal fusion response feature matrix through time alignment and modal fusion. An intelligent broadband noise suppression unit performs noise suppression on each modal submatrix and outputs a denoised feature matrix. Finally, an intelligent fault location and interpretation unit extracts the fault-associated modal submatrix from the matrix, constructs a three-dimensional signal fingerprint spectrum based on time, signal type, and amplitude, and uses classification analysis and path calculation based on the spectrum to identify the fault type and accurately locate the fault. The system forms a closed-loop fault detection chain through multimodal collaborative perception, dynamic noise reduction, and intelligent graph diagnosis. In the system of the present invention, based on actively injecting a frequency-adjustable low-frequency pulse excitation signal into the submarine cable, it is possible to actively induce fault response and synchronously collect multi-source signals without modifying the cable structure, effectively detecting various types of submarine cable faults including leakage, breakage, insulation degradation, etc., and improving the comprehensiveness and sensitivity of fault identification; by constructing a multimodal fusion response feature matrix and combining broadband noise suppression and path residual fitting positioning algorithms, it is possible to achieve robust de-interference, accurate identification and three-dimensional positioning of fault signals in complex submarine environments, thereby improving the accuracy of fault interpretation and shortening the response time of cable fault processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a principle block diagram of a submarine cable fault detection system proposed by the present invention; Figure 2 This is a flow chart of a submarine cable fault detection method proposed by the present invention; Figure numerals: 1. Adjustable low-frequency pulse injection unit; 11. Signal generation module; 12. Parameter control module; 13. Excitation output interface module; 2. Multi-source signal fusion acquisition unit; 21. Voltage signal acquisition module; 22. Acoustic signal acquisition module; 23. Magnetic field signal acquisition module; 24. Feature fusion processing module; 3. Intelligent broadband noise suppression unit; 31. Multimodal decomposition processing module; 32. Time-frequency noise reduction module; 32. Frequency calibration module; 4. Intelligent fault location and interpretation unit; 41. Modal extraction module; 42. Feature fingerprint construction module; 42. Fault identification and location module. DETAILED DESCRIPTION

[0016] Hereinafter, the terms "first," "second," "third," and "fourth" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the technical features indicated. Thus, a feature identified as "first," "second," "third," or "fourth" may explicitly or implicitly include one or more of such features.

[0017] The co-shooting method provided in the embodiments of the present application can be applied to terminal devices such as mobile phones, tablet computers, wearable devices, vehicle-mounted devices, augmented reality (AR) / virtual reality (VR) devices, laptop computers, ultra-mobile personal computers (UMPCs), netbooks, and personal digital assistants (PDAs). The embodiments of the present application do not impose any restrictions on the specific types of terminal devices.

[0018] It should be noted that the terms "first," "second," and the like in the description and drawings of the present invention are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatus.

[0019] A first aspect of the present invention provides a submarine cable fault detection system, comprising: An adjustable low-frequency pulse injection unit 1 is used to inject a low-frequency pulse excitation signal with adjustable frequency into the submarine cable to stimulate the electrical response, acoustic response and magnetic response signals generated by the submarine cable fault; The adjustable low-frequency pulse injection unit 1 includes a signal generation module 11, a parameter control module 12 and an excitation output interface module 13; Among them, the signal generation module 11 is used to generate a periodic voltage signal within a set frequency range; the parameter control module 12 is used to dynamically adjust the frequency, amplitude and duty cycle of the low-frequency pulse excitation signal according to a preset excitation configuration table; the excitation output interface module 13 is used to apply the generated excitation signal to the conductor port of the submarine cable.

[0020] In this embodiment, the parameter control module 12 is constructed based on an adjustable oscillator and a digital controller, and uses closed-loop feedback to calibrate the output frequency in real time, compensating for frequency offsets due to changes in ambient temperature and cable impedance. Specifically, the parameter control module 12 calibrates the output frequency in real time based on the reflected voltage signal, acoustic signal, and leakage magnetic field signal collected by the multi-source signal fusion acquisition unit 2.

[0021] In this embodiment, the low-frequency pulse excitation signal is injected into the submarine cable using three signal modes, including a continuous periodic sinusoidal signal mode, a step voltage signal mode, and a single pulse signal mode; Among them, the continuous periodic sinusoidal signal mode is used to stimulate low-frequency resonance response; the step voltage signal mode is used to stimulate cable impedance jump reflection; and the single pulse signal mode is used to simulate the broken arc impact and induce acoustic response.

[0022] Multi-source signal fusion acquisition unit 2, which is used to collect reflected voltage signals, acoustic signals and leakage magnetic field signals, and perform synchronization marking, timestamp alignment and modal fusion to construct a multi-modal fusion response feature matrix; The multi-source signal fusion acquisition unit 2 includes a voltage signal acquisition module 21, an acoustic signal acquisition module 22, a magnetic field signal acquisition module 23 and a feature fusion processing module 24; Among them, the voltage signal acquisition module 21 is used to acquire the reflected voltage signal in the electrical response; the acoustic signal acquisition module 22 is used to acquire the acoustic signal in the acoustic response; and the magnetic field signal acquisition module 23 is used to acquire the leakage magnetic field signal in the magnetic response.

[0023] In this embodiment, the feature fusion processing module 24 is used to perform unified clock calibration and timestamp alignment on the reflected voltage signal, acoustic signal and leakage magnetic field signal, and after normalization and standardization processing, encode the standardized signal into a feature vector form to construct a multimodal fusion response feature matrix organized according to the time series dimension, signal type dimension and spatial channel dimension.

[0024] Specifically, the reflected voltage signal, acoustic signal, and leakage magnetic field signal are first uniformly calibrated through hardware, such as through a GPS disciplined clock or IEEE 1588 PTP (Precision Time Protocol). Furthermore, timestamp alignment is achieved through software, with a four-element timestamp attached to each sampling point. Based on the moment of excitation signal injection, all sampling points are time-shifted to achieve timestamp alignment.

[0025] Specifically, the normalization is to perform modality-level normalization on the calibrated and aligned signals, and the standardization processing is to perform channel-level normalization on the calibrated and aligned signals.

[0026] Specifically, encoding the red-marked signal into a feature vector format involves extracting features using a sliding window, generating an N-dimensional feature vector for each window, and obtaining a feature vector. The feature vectors are aggregated from the time series dimension, signal type dimension, and spatial channel dimension to obtain a modal fusion response feature matrix.

[0027] An intelligent broadband noise suppression unit 3 is used to perform noise suppression processing on each modal sub-matrix in the multimodal fusion response feature matrix to obtain a denoised multimodal fusion response feature matrix; In this embodiment, the intelligent broadband noise suppression unit 3 includes a multimodal decomposition processing module 31, a time-frequency noise reduction module 32 and a frequency calibration module 33; Among them, the multimodal decomposition processing module 31 is used to decompose each modal submatrix in the multimodal fusion response feature matrix according to the signal type dimension to obtain the separated modal submatrices; the time-frequency noise reduction module 32 adopts an algorithm that combines wavelet multi-scale decomposition and empirical mode decomposition to perform layered filtering on the noise components of different frequency bands in each modal submatrix; the frequency calibration module 33 is used to detect the frequency deviation between the adjustable low-frequency pulse injection signal and each response signal, and dynamically adjust the feature matrix through the frequency drift compensation algorithm to obtain the denoised multimodal fusion response feature matrix.

[0028] Specifically, the time-frequency denoising module 32 first performs wavelet multi-scale decomposition on the noise in different frequency bands in each modal sub-matrix, then performs empirical mode decomposition on the decomposed sub-band signals, and performs layered filtering on the noise components in different frequency bands after decomposition using a threshold method.

[0029] Intelligent fault location and interpretation unit 4 is used to extract the fault-related modal submatrix from the denoised multimodal fusion response feature matrix, construct a three-dimensional signal fingerprint spectrum, and classify and analyze the fault events based on the fuzzy clustering algorithm and the deep feature matching model to obtain the fault type, and calculate the fault location through the path residual fitting method; In this embodiment, the intelligent fault location and interpretation unit 4 includes a mode extraction module 41, a feature fingerprint construction module 42 and a fault identification and location module 43; Among them, the modal extraction module 41 is used to extract the abnormal response modal submatrix from the denoised multimodal fusion response feature matrix; the feature fingerprint construction module 42 constructs a three-dimensional signal fingerprint spectrum of the fusion fault event based on the abnormal response modal submatrix; the fault identification and positioning module 43 uses the fuzzy clustering algorithm and the deep feature matching model to classify and analyze the fault event to obtain the fault type, and calculates the fault location through the path residual fitting method.

[0030] In this embodiment, the abnormal response modal submatrix is ​​obtained by screening through a preset response threshold or identifying and extracting based on a baseline comparison algorithm.

[0031] In this embodiment, the three-dimensional signal fingerprint spectrum is constructed with the time axis, signal type axis and signal amplitude axis as three orthogonal dimensions, as follows: Time axis design: The moment the low-frequency pulse excitation signal is injected is set as the time reference zero (t=0). All collected electrical, acoustic, and magnetic response signals are time-aligned based on this reference point. The time axis represents the delay time of each signal response relative to the excitation moment. Signal type axis design: The electrical response, acoustic response, and magnetic response signals are defined as three modes, each corresponding to an independent sub-channel; Signal amplitude axis design: The original signal value at each sampling time point and each signal mode is normalized according to a preset benchmark to obtain the normalized amplitude as the longitudinal amplitude distribution. In this embodiment, in the fault identification and positioning module 43, a fuzzy clustering algorithm is used to cluster abnormal areas of the three-dimensional signal fingerprint spectrum to determine the duration and coverage mode of the abnormal fault response; the deep feature matching model performs multi-layer semantic matching on the input three-dimensional signal fingerprint spectrum based on the fault feature fingerprint spectrum library and outputs a fault type label.

[0032] The path residual fitting method is used to invert wave propagation path errors and calculate the signal source location. It applies a least-squares correction to the propagation path error, combined with cable structural parameters, to determine the fault location of submarine cables. Specifically, the path residual fitting method estimates the wave propagation path error by analyzing the residual error (the difference between the observed value and the theoretical value) of the signal arrival time (or phase) and calculates the fault signal source location. This method, combined with detailed cable structural parameters (such as length, segmented wave velocity, and connector location), utilizes a least-squares optimization algorithm to correct the propagation path error, ultimately determining the precise location of submarine cable faults.

[0033] A second aspect of the present invention discloses a method for detecting a submarine cable fault, comprising the following steps: S1, injecting a frequency-adjustable low-frequency pulse excitation signal into the submarine cable and collecting the electrical response, acoustic response, and magnetic response signals generated by the submarine cable fault; S2 collects the reflected voltage signal, acoustic signal, and leakage magnetic field signal corresponding to the feedback of the low-frequency pulse excitation signal, and performs synchronization marking, timestamp alignment, and modal fusion to construct a multimodal fusion response feature matrix; S3, performing noise suppression processing on each modal sub-matrix in the multimodal fusion response feature matrix to obtain a denoised multimodal fusion response feature matrix; S4, extract the fault-related modal submatrix from the denoised multimodal fusion response feature matrix and construct a three-dimensional signal fingerprint spectrum; based on the three-dimensional signal fingerprint spectrum, classify and analyze the fault events to obtain the fault type and calculate the fault location.

[0034] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0035] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0036] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0037] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A submarine cable fault detection system, characterized in that: include: An adjustable low-frequency pulse injection unit (1) is used to inject a frequency-adjustable low-frequency pulse excitation signal into the submarine cable to stimulate the electrical response, acoustic response and magnetic response signals generated by the submarine cable fault; A multi-source signal fusion acquisition unit (2) is used to acquire a reflected voltage signal in the electric response, an acoustic signal in the acoustic response, and a leakage magnetic field signal in the magnetic response signal; The collected signals are synchronized, timestamp aligned, and modally fused to obtain a multimodal fusion response feature matrix; An intelligent broadband noise suppression unit (3) is used to perform noise suppression processing on each modal sub-matrix in the multimodal fusion response feature matrix to obtain a denoised multimodal fusion response feature matrix; The intelligent fault location and interpretation unit (4) is used to extract the fault-related mode submatrix in the denoised multimodal fusion response feature matrix and construct a three-dimensional signal fingerprint spectrum; based on the three-dimensional signal fingerprint spectrum, the fault event is classified and analyzed to obtain the fault type and fault location.

2. A submarine cable fault detection system according to claim 1, characterized in that: The adjustable low pulse injection unit (1) comprises a signal generation module (11), a parameter control module (12) and an excitation output interface module (13); The signal generation module (11) is used to generate a periodic voltage signal within a set frequency range; the parameter control module (12) is used to adjust the frequency, amplitude and duty cycle of the low-frequency pulse excitation signal according to a preset excitation configuration table; The excitation output interface module (13) is used to output the low-frequency pulse excitation signal.

3. A submarine cable fault detection system according to claim 2, characterized in that: The parameter control module (12) calibrates the output frequency in real time through a closed-loop feedback method based on the feedback of the multi-source signal fusion acquisition unit (2).

4. A submarine cable fault detection system according to claim 1, characterized in that: The low-frequency pulse excitation signal is in a continuous periodic sinusoidal signal mode, a step voltage signal mode or a single pulse signal mode.

5. The submarine cable fault detection system according to claim 1, characterized in that: The multi-source signal fusion acquisition unit (2) comprises a voltage signal acquisition module (21), an acoustic signal acquisition module (22), a magnetic field signal acquisition module (23) and a feature fusion processing module (24); The voltage signal acquisition module (21) is used to acquire the reflected voltage signal in the electrical response; the acoustic signal acquisition module (22) is used to acquire the acoustic signal in the acoustic response; and the magnetic field signal acquisition module (23) is used to acquire the leakage magnetic field signal in the magnetic response. The feature fusion processing module (24) is used to uniformly calibrate the clock and align the timestamps of the reflected voltage signal, the acoustic signal and the leakage magnetic field signal, and after normalization and standardization, encode the standardized signal into a feature vector form, and summarize the feature vector according to the time series dimension, the signal type dimension and the spatial channel dimension to obtain a multimodal fusion response feature matrix.

6. A submarine cable fault detection system according to claim 1, characterized in that: The intelligent broadband noise suppression unit (3) includes a multimodal decomposition processing module (31), a time-frequency noise reduction module (32) and a frequency calibration module (33); The multimodal decomposition processing module (31) is used to perform dimensional decomposition processing on each modal submatrix in the multimodal fusion response feature matrix according to the signal type dimension to obtain each separated modal submatrix; the time-frequency noise reduction module (32) is used to perform layered filtering on noise components of different frequency bands in each modal submatrix; The frequency calibration module (33) is used to detect the frequency deviation between the adjustable low-frequency pulse injection signal and each response signal, and dynamically adjust the characteristic matrix based on the frequency deviation to obtain a denoised multimodal fusion response characteristic matrix.

7. A submarine cable fault detection system according to claim 6, characterized in that: The time-frequency noise reduction module (32) uses an algorithm combining wavelet multi-scale decomposition and empirical mode decomposition to perform hierarchical filtering; The frequency calibration module (33) dynamically adjusts the characteristic matrix through a frequency drift compensation algorithm.

8. The submarine cable fault detection system according to claim 1, characterized in that: The intelligent fault location and interpretation unit (4) includes a modal extraction module (41), a feature fingerprint construction module (42) and a fault identification and location module (43); The modal extraction module (41) is used to extract an abnormal response modal submatrix from the denoised multimodal fusion response feature matrix; the feature fingerprint construction module (42) is used to construct a three-dimensional signal fingerprint spectrum of the fusion fault event based on the abnormal response modal submatrix; The fault identification and positioning module (43) is used to classify and analyze fault events to obtain the fault type and fault location.

9. A submarine cable fault detection system according to claim 8, characterized in that: The three-dimensional signal fingerprint spectrum is constructed by using the time axis, signal type axis and signal amplitude axis as three orthogonal dimensions; The fault identification and positioning module (43) uses a fuzzy clustering algorithm and a deep feature matching model to classify and analyze the fault event to obtain the fault type, and calculates the fault location through a path residual fitting method.

10. A fault detection method of the submarine cable fault detection system according to claim 1, characterized in that: The following steps are involved: S1, injects a frequency-adjustable low-frequency pulse excitation signal into the submarine cable to stimulate the electrical response, acoustic response and magnetic response signals generated by the submarine cable fault; S2, collecting the reflected voltage signal in the electrical response, the acoustic signal in the acoustic response, and the leakage magnetic field signal in the magnetic response signal; The collected signals are synchronized, timestamp aligned, and modally fused to obtain a multimodal fusion response feature matrix; S3, performing noise suppression processing on each modal sub-matrix in the multimodal fusion response feature matrix to obtain a denoised multimodal fusion response feature matrix; S4, extract the fault-related modal submatrix from the denoised multimodal fusion response feature matrix and construct a three-dimensional signal fingerprint spectrum; based on the three-dimensional signal fingerprint spectrum, classify and analyze the fault events to obtain the fault type and fault location.

Images