Method and system for monitoring and positioning multiple types of faults of submarine direct-current multi-path cable

Abstract

The invention relates to a method and a system for monitoring and positioning multiple types of faults of a submarine direct-current multi-path cable, and the method comprises the steps: selecting m sequences with different code element durations according to the length of a detected cable, and multiplying the m sequences by a chaotic sequence mapped by improved Logistic to obtain an improved chaotic sequence; performing spread spectrum modulation on the improved chaos sequence and a sine wave to form a transmitting signal; transmitting signals are sequentially injected into the test points of the tested cables; collecting reflected signals of each path of tested cable, and estimating time delay between the transmitted signals and the reflected signals to judge whether each path of cable has a fault and a corresponding fault type; and based on the time delay and the cable parameters of the corresponding path, realizing accurate fault positioning through a deep learning algorithm. Compared with the prior art, high interference resistance, multi-path cyclic monitoring and automatic fault type judgment can be realized at the same time, and adaptive compensation can be carried out on environmental factors.

Description

Technical Field

[0001] This invention belongs to the field of submarine cable operation and maintenance technology, specifically involving a method and system for monitoring and locating multiple types of faults in submarine DC multi-channel cables based on an improved SSTDR algorithm. It is applicable to real-time status monitoring and accurate fault location of submarine cables in scenarios such as cross-sea power transmission and marine energy development. Background Technology

[0002] Submarine DC cables are a key component of energy systems such as cross-sea power transmission and offshore wind power grid connection. Due to the complex laying environment and maintenance difficulties, real-time and accurate monitoring and location of faults are crucial. Traditional time domain reflection (TDR) methods, when applied to long-distance submarine cables, face problems such as large signal attenuation, weak anti-interference ability, difficulty in distinguishing multiple cable faults, and inability to accurately locate faults.

[0003] Spread-spectrum time-domain reflectometry (SSTDR) improves the aforementioned problems to some extent by transmitting spread-spectrum signals and using correlation processing to enhance anti-interference capabilities. However, traditional SSTDR methods still have the following limitations: First, the autocorrelation characteristics of the pseudo-random sequence used are not excellent enough, and crosstalk is easily generated when multiple parallel detections are performed; second, the time delay estimation accuracy of the reflected signal drops sharply in the context of strong noise; third, fault location heavily depends on the propagation speed of the signal in the cable, which varies due to cable conditions and seabed environment, resulting in large errors in the location results under a fixed propagation speed model; fourth, there is a lack of systematic and efficient monitoring schemes for multiple cables.

[0004] Therefore, there is an urgent need for a high-precision positioning method and system that can simultaneously achieve high anti-interference, multi-channel roving monitoring, automatic fault type identification, and adaptive compensation for environmental factors. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a diagnostic and location method and system with strong anti-interference ability, high positioning accuracy, and the ability to simultaneously monitor multiple submarine DC cable faults.

[0006] The objective of this invention can be achieved through the following technical solutions: A method for monitoring and locating multiple types of faults in submarine DC multi-channel cables, comprising: Based on the length of the cable under test, m-sequences with different symbol durations are selected and multiplied with the chaotic sequence mapped by the improved Logistic to obtain the improved chaotic sequence; the improved chaotic sequence is then spread-spectrum modulated with a sine wave to form the transmitted signal. The transmitted signal is sequentially injected into the test points of each cable under test; Collect the reflected signals of each cable under test, estimate the time delay between the transmitted signal and the reflected signal, and determine whether there is a fault in each cable and the corresponding fault type. Based on the time delay and the corresponding cable parameters, a deep learning algorithm is used to achieve accurate fault location.

[0007] Furthermore, the expression for the improved Logistic mapping is: In the formula, To map the value of the (n+1)th iteration, To map the value of the nth iteration, For mapping parameters, For disturbance terms; In the process of determining chaotic sequences using the improved Logistic mapping, initial values ​​with chaotic state transition capabilities are obtained through multiple experiments. Mapping parameters and disturbance terms .

[0008] Furthermore, the step of sequentially injecting the transmitted signal into the test points of each cable under test includes: The spread spectrum modulated signal is converted into a high-voltage pulse signal. The digital high-voltage pulse signal is then converted into an analog signal according to a preset timing sequence, and then injected into different branches of the cable under test through a multi-channel analog switch array.

[0009] Furthermore, the method estimates the time delay between the transmitted signal and the reflected signal by improving the generalized cross-correlation peak denoising delay estimation algorithm; The improved generalized cross-correlation peak denoising time delay estimation algorithm performs cross-correlation calculations on the transmitted signal and the corresponding reflected signal, then performs a fractional Fourier transform to obtain the cross-power function at the optimal order; then performs an inverse fractional Fourier transform to obtain the improved generalized cross-correlation function, and uses the time difference between the two impulse peaks in the improved generalized cross-correlation function as the time delay.

[0010] Furthermore, the expression for the improved generalized cross-correlation function is as follows: In the formula, For the improved generalized cross-correlation function, The cross-power function at the optimal order. For the fractional Fourier transform at the optimal order The inverse transform kernel.

[0011] Furthermore, determining whether each cable has a fault and the corresponding fault type includes: If the time delay is less than If so, then a fault is determined to exist, where, For cable length, The speed at which signals propagate in the cable; By analyzing the polarity of the two impulse peaks, we can distinguish whether the corresponding fault type is an open-circuit fault or a short-circuit fault.

[0012] Furthermore, the deep learning algorithm uses time delay, cable parameters of the corresponding path, and the cross-power function at the optimal order as input to predict the fault location, and the corresponding expression is: In the formula, Location of the fault. For time delay, L, C, and T represent the distributed inductance, distributed capacitance, and ambient temperature per unit length of cable, respectively. These represent the characteristic frequencies of the cross-power function signal at the optimal order in the fractional-order frequency domain.

[0013] Furthermore, during the acquisition of reflected signals from each cable under test, the method also uses an adaptive filter to separate interference signals caused by multi-cable coupling. The adaptive filter dynamically adjusts its weights based on the least mean square algorithm, and the corresponding weight adjustment expression is: w( n +1)=w( n )+ μ ⋅ e ( n )⋅x( n ) Among them, w( n ) indicates the first n The filter weight vector at time t. μ For step size parameters, e ( n Let x be the error signal, defined as the difference between the desired signal and the filter output. n ) represents the input signal vector; In the process of acquiring the reflected signals of each cable under test, the method also employs continuous wavelet transform to perform joint time-frequency domain analysis on the acquired reflected signals and distributed optical fiber vibration signals, in order to analyze the non-stationary characteristics of the reflected signals and perform feature enhancement. The expression for the continuous wavelet transform is as follows: in, The result is a continuous wavelet transform. x ( t ( ) represents the time-domain reflected signal. ψ ( t) is the mother wavelet function. a For scale parameters, b is the translation parameter, and * denotes complex conjugate.

[0014] Furthermore, before injecting the transmitted signal, the method establishes a cable coupling model based on multi-conductor transmission line theory and performs pre-distortion processing; the cable coupling model uses transmission line equations to pre-adjust the transmitted signal, and the expression of the transmission line equations is: Where V(z,t) and I(z,t) represent the voltage and current vectors along the cable position z and time, respectively. t The distribution of L, R, C, and G represents the inductance, resistance, capacitance, and conductance matrices per unit length, respectively, where the off-diagonal elements characterize the coupling strength between cables.

[0015] This invention also provides a system for monitoring and locating multiple types of faults in submarine DC multi-channel cables, which implements the method for monitoring and locating multiple types of faults in submarine DC multi-channel cables as described above, comprising: An improved SSTDR modulation module is used to select m-sequences with different symbol durations according to the length of the cable under test, and multiply them with the chaotic sequence mapped by the improved Logistic to obtain an improved chaotic sequence; the improved chaotic sequence is then spread spectrum modulated with a sine wave to form a transmitted signal. The signal transmission and acquisition module is used to sequentially inject the transmitted signal into the test points of each cable under test and acquire the reflected signal of each cable under test. The fault detection module is used to estimate the time delay between the transmitted signal and the reflected signal in order to determine whether there is a fault in each cable and the corresponding fault type. The deep learning analysis module is used to accurately locate faults based on the time delay and the corresponding cable parameters using deep learning algorithms.

[0016] Compared with the prior art, the present invention has the following advantages: (1) The method of the present invention first uses an improved Logistic chaotic mapping to generate an anti-interference detection signal, which has both the noise-like characteristics of chaotic signals and the sharp autocorrelation characteristics of m-sequences. Then, it is injected into the starting test points of the first to Nth cables under test according to a preset time sequence, and the reflection response signal of each cable is collected and stored. Next, based on the improved generalized cross-correlation theory, the time delay between the transmitted signal and the reflected signal is accurately estimated to determine whether there is a fault in the cable and the type of fault. Finally, a cable fault location model is established based on a deep learning algorithm. This model learns the nonlinear mapping relationship between fault characteristics and actual physical location in massive data, adaptively compensates for wave velocity drift caused by environmental changes, and outputs the precise physical location of the fault point.

[0017] (2) This invention employs a spreading code combining chaos and m-sequence, generating a detection signal with lower power spectral density and superior correlation characteristics, greatly improving the signal-to-noise ratio and anti-interference capability in complex seabed electromagnetic environments. The improved generalized cross-correlation algorithm effectively enhances the time delay detection accuracy and reliability of weak reflected signals in low signal-to-noise ratio environments. An innovative deep learning model is introduced to dynamically correct wave propagation speed through data-driven methods, overcoming the impact of environmental factors on positioning accuracy and achieving precise fault ranging. Attached Figure Description

[0018] Figure 1 This application provides a method and system block diagram for multi-type fault monitoring and precise location of submarine DC multi-channel cables; Figure 2 Autocorrelation results of the improved chaotic sequence provided in this application; Figure 3 The flowchart for FPGA-controlled analog switch switching provided in this application; Figure 4a A schematic diagram of open-circuit fault results for the improved generalized cross-correlation algorithm provided in this application; Figure 4b A schematic diagram of short-circuit fault results for the improved generalized cross-correlation algorithm provided in this application; Figure 5 The least squares support vector machine training flowchart provided for this application; Figure 6 This is a flowchart illustrating a method for monitoring and accurately locating multiple types of faults in a submarine DC multi-channel cable, as provided in this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0022] Example 1 Reference Figure 1 As shown in the figure, the method for multi-type fault monitoring and precise location of submarine DC multi-channel cables provided in this embodiment includes: S1: Select m sequences with different symbol durations according to the length of the cable under test, and multiply them with the chaotic sequence mapped by the improved Logistic to obtain the improved chaotic sequence; spread spectrum modulation of the improved chaotic sequence with a sine wave to form the transmitted signal; S2: Inject the transmitted signal sequentially into the test points of each cable under test; S3: Collect the reflected signals of each cable under test, estimate the time delay between the transmitted signal and the reflected signal, and determine whether there is a fault in each cable and the corresponding fault type; S4: Based on time delay and corresponding cable parameters, a deep learning algorithm is used to achieve accurate fault location.

[0023] Specifically, in step S1, a chaotic spread-spectrum time-domain reflectometry method is introduced to improve the Logistic mapping into a chaotic sequence. Based on the length of the cable under test, different symbol durations of sequence m are selected and multiplied to obtain the improved chaotic sequence. This is then modulated with a sinusoidal spread spectrum to obtain a transmitted signal with a wider spectrum, giving the system good anti-interference capabilities. It can detect faults at close range and also allows for a custom ranging range. The expression for the improved Logistic mapping is: in For mapping parameters, To map the value of the nth iteration, This is a perturbation term used to increase the Lyapunov exponent, thereby maximizing the Lyapunov exponent.

[0024] In step S2, the transmitted signal is sequentially injected into the test points of each cable under test, including: The spread spectrum modulated signal is converted into a high-voltage pulse signal. The digital high-voltage pulse signal is then converted into an analog signal according to a preset timing sequence, and then injected into different branches of the cable under test through a multi-channel analog switch array.

[0025] Preferably, the signal transmission and acquisition processes in steps S2 and S3 further include: During the acquisition of reflected signals from each cable under test, an adaptive filter is used to separate interference signals caused by multi-cable coupling. The adaptive filter dynamically adjusts its weights based on the least mean square algorithm, and the corresponding weight adjustment expression is: w( n +1)=w( n )+ μ ⋅ e ( n )⋅x( n ) Among them, w( n ) indicates the first n The filter weight vector at time t. μ For step size parameters, e ( n Let x be the error signal, defined as the difference between the desired signal and the filter output. n ) represents the input signal vector; In the process of acquiring the reflected signals of each cable under test, the method also employs continuous wavelet transform to perform joint time-frequency domain analysis on the acquired reflected signals and distributed optical fiber vibration signals, in order to analyze the non-stationary characteristics of the reflected signals and perform feature enhancement. The expression for continuous wavelet transform is as follows: in, The result is a continuous wavelet transform. x ( t ( ) represents the time-domain reflected signal. ψ ( t ) is the mother wavelet function. a For scale parameters, b is the translation parameter, and * denotes complex conjugate.

[0026] Before injecting the transmission signal, a cable coupling model is established based on multi-conductor transmission line theory to perform pre-distortion processing. The cable coupling model uses transmission line equations to pre-adjust the transmission signal. The expression of the transmission line equations is as follows: Where V(z,t) and I(z,t) represent the voltage and current vectors along the cable position z and time, respectively. t The distribution of L, R, C, and G represents the inductance, resistance, capacitance, and conductance matrices per unit length, respectively, where the off-diagonal elements characterize the coupling strength between cables.

[0027] In step S4, the time delay between the transmitted signal and the reflected signal is estimated by improving the generalized cross-correlation peak denoising delay estimation algorithm. The improved generalized cross-correlation peak denoising time delay estimation algorithm performs cross-correlation calculation on the transmitted signal and the corresponding reflected signal, and then performs fractional Fourier transform to obtain the cross-power function at the optimal order; then, it performs inverse fractional Fourier transform to obtain the improved generalized cross-correlation function, and uses the time difference between the two impulse peaks in the improved generalized cross-correlation function as the time delay.

[0028] The expression for the improved generalized cross-correlation function is: In the formula, For the improved generalized cross-correlation function, The cross-power function at the optimal order. For the fractional Fourier transform at the optimal order The inverse transform kernel.

[0029] Determine if there are faults in each cable and the corresponding fault types, including: If the time delay is less than If so, then a fault is determined to exist, where, For cable length, The speed at which signals propagate in the cable; By analyzing the polarity of the two impulse peaks, we can distinguish whether the corresponding fault type is an open-circuit fault or a short-circuit fault.

[0030] The deep learning algorithm uses time delay, cable parameters of the corresponding path, and cross-power function at the optimal order as input to predict fault location. The corresponding expression is: In the formula, Location of the fault. For time delay, L, C, and T represent the distributed inductance, distributed capacitance, and ambient temperature per unit length of cable, respectively. These represent the characteristic frequencies of the cross-power function signal at the optimal order in the fractional-order frequency domain.

[0031] Specifically, this invention selects four submarine DC cables as the research object, and the parameters of the cables to be tested are shown in Table 1.

[0032] Table 1 Improved expression for Logistic mapping: in For mapping parameters, To map the value of the nth iteration, This is a perturbation term used to increase the Lyapunov exponent, maximizing it. As can be seen from the formula, different initial values... and mapping parameters Mapping The value of will have different effects. After conducting multiple experiments by controlling the variables, when , The system then enters a chaotic state. This leads to the determination of the expression for the chaotic mapping: An improved chaotic sequence is obtained by multiplying an m-sequence with a symbol duration of 1 microsecond with a chaotic sequence. Its autocorrelation characteristics are then analyzed. Figure 2 As shown in the figure, the signal exhibits good autocorrelation performance and a distinct main peak, making it suitable as a test signal. Binary phase shift keying (BPSK) modulation quantizes the original unipolar chaotic sequence into a bipolar signal, which is then multiplied and modulated with a sine wave to obtain a chaotic spread spectrum signal.

[0033] The signal transmission and acquisition module controls a multi-channel analog switch to sequentially inject the transmitted signal into the test points of cables 1 through 4. It includes a DA converter module, a multi-channel analog switch module, a signal transmission module, and a signal acquisition module. The signal transmission module converts the BPSK modulated signal into a high-voltage pulse signal, converts the digital signal into an analog signal according to a preset timing sequence, and injects it into different DC cable paths after passing through the multi-channel analog switch array. The signal acquisition module uses a high-speed data acquisition card to acquire reflected signals and distributed fiber optic vibration signals at a sampling rate of ≥100MS / s, and obtains the acquired signal after filtering and signal enhancement.

[0034] The control of the multi-channel cable signal transmission and acquisition analog switch is simulated using FPGA control. The flowchart of the analog switch on / off process is as follows: Figure 3 As shown, time-division control of the detection signal can realize multi-channel cable detection.

[0035] The fault detection module, through an improved generalized cross-correlation peak denoising delay estimation algorithm, reduces the interference from multi-path cables, improves fault detection accuracy, and enables preliminary judgment of fault presence and type. Assuming the transmitted signal is... The received signal is To accurately obtain the time delay estimates for these two signals, the influence of irrelevant noise needs to be filtered out. , After performing cross-correlation, a fractional Fourier transform (FrFT) is performed to obtain the cross-power function at the optimal order. Then, an inverse fractional Fourier transform is performed to obtain the improved generalized cross-correlation function: in For the fractional Fourier transform at the optimal order The inverse transform kernel obtained The time difference between the two impulse peaks is the estimated time delay d. If d is less than... By analyzing the polarity (positive or negative) of the cross-correlation peaks, it is possible to determine whether the fault is an open circuit or a short circuit.

[0036] Since multi-channel cable fault detection uses the same detection device, the transmitted signal is sent to the cable under test through an impedance matching resistor. Reflected signals from other channels can interfere with the cable under test, resulting in multiple signal reflections and superposition of signal amplitudes. Therefore, based on the superposition property of the fractional Fourier transform, the characteristic signal of the cable under test is extracted by selecting the optimal order. The simulation sets the impedance matching resistor to 60 ohms, a short-circuit fault at 400 meters in the first channel, and an open-circuit fault at 20 kilometers in the second channel. First, noise with a mean of 0.05 is introduced, and the processing results of the improved generalized cross-correlation algorithm are as follows: Figure 4a and Figure 4b As shown in Table 2, the peak-to-average power ratio, positioning distance, and error results obtained by performing basic cross-correlation and improved generalized cross-correlation on the transmitted and reflected signals, respectively, are presented in Table 2.

[0037] Table 2 Depend on Figure 4a and Figure 4b As shown in the table above, the polarity of the cross-correlation peak is negative when a short-circuit fault occurs, and positive when an open-circuit fault occurs. Compared with the basic cross-correlation algorithm, the improved generalized cross-correlation algorithm has a larger peak-to-average power ratio and a smaller location error, confirming the superiority of the improved generalized cross-correlation algorithm.

[0038] The above analyses are all based on theoretical conditions. In actual seabed environments, the signal propagation speed in cables is affected by environmental factors such as cable structure, aging, temperature, and pressure, and is not a constant value. Using a fixed speed value will lead to positioning errors. To further improve the accuracy of the system in locating fault points, time delay estimation models for multiple submarine DC cables are established based on deep learning algorithms, comprehensively considering the attenuation of signal speed in the cable, and taking into account time delay d, cable parameters, and... Using frequency characteristics as input, an estimation model for the fault location y is established: Where L, C, and T represent the distributed inductance, distributed capacitance, and ambient temperature per unit length of cable, respectively. They represent The signal's characteristic frequencies in the fractional-order frequency domain are identified using principal component analysis (PCA), yielding the two most relevant characteristic frequencies. The system also includes an initial cable database module, storing parameters of all monitored cables in their original, health-state conditions. This provides crucial prior knowledge for the deep learning analysis module, used for model training and labeling.

[0039] Choosing Least Squares Support Vector Machine (LSSVM) as the deep learning modeling method, the estimate of a certain parameter by LSSVM in high-dimensional space can be expressed as: in The weight vector in the feature space. It is a nonlinear function. From input space Mapped to , This is the bias. Therefore, the optimization problem of LSSVM is to solve... and The process.

[0040] Select the time delay d, cable parameters, and under different fault types at different locations. The frequency characteristics are taken as input, with hyperparameters of 1 and 1×10. -3 The LSSVM estimation model is trained, and the training flowchart is as follows: Figure 5 As shown in Table 3, to demonstrate the effectiveness of the proposed deep learning module, the ambient temperature was increased to 60 degrees Celsius under extreme conditions, with open-circuit faults at 500m and 600m and short-circuit faults at 12km. The localization results for different faults at different distances are shown in Table 3.

[0041] Table 3 As can be seen from the table, the proposed algorithm has stronger robustness and anti-interference ability compared with the direct use of the improved generalized correlation algorithm for fault location, and is equally applicable to open circuit faults and short circuit faults.

[0042] Example 2 This embodiment provides a hardware platform consisting of an FPGA main control board, a multi-channel transmitter board, and a multi-channel acquisition board, used to generate transmitted signals and acquire reflected signals. The host computer module is used for complex calculations such as feature extraction and deep learning modeling, as well as waveform display. The power supply module includes 12V, 5V, and 3.3V power output modules. The software program can implement the above steps, record the execution results of each step, and provide real-time alarms for cable faults.

[0043] Specifically, it includes: An improved SSTDR modulation module is used to select m-sequences with different symbol durations according to the length of the cable under test, and multiply them with the chaotic sequence mapped by the improved Logistic to obtain an improved chaotic sequence; the improved chaotic sequence is then spread spectrum modulated with a sine wave to form a transmitted signal. The signal transmission and acquisition module is used to sequentially inject the transmitted signal into the test points of each cable under test and acquire the reflected signal of each cable under test. The fault detection module is used to estimate the time delay between the transmitted signal and the reflected signal in order to determine whether there is a fault in each cable and the corresponding fault type. The deep learning analysis module is used to achieve accurate fault location based on time delay and corresponding cable parameters through deep learning algorithms.

[0044] The improved SSTDR modulation module obtains an improved chaotic sequence by multiplying an improved Logistic chaotic sequence with an m-sequence of different symbol durations, and then modulates it with a sine wave via BPSK to form the transmitted signal. The signal transmission and acquisition module controls a multi-channel analog switch to sequentially inject the transmission signal into the test points of the first to Nth cables; The fault detection module uses an improved generalized cross-correlation peak denoising delay estimation algorithm to determine whether there is a fault in the current cable. If there is a fault, the fault type is determined to be open circuit fault or short circuit fault based on the polarity of the cross-correlation normalization value. The deep learning analysis module uses machine learning algorithms to achieve precise fault location based on enhanced signal features and parameters in the initial cable database. Repeat the above process until all cables have been tested.

[0045] The improved SSTDR modulation module introduces a chaotic spread-spectrum time-domain reflectometry method. It improves the Logistic mapping to a chaotic sequence, selecting m-sequences with different symbol durations based on the length of the cable under test, and multiplying them to obtain the improved chaotic sequence. This is then spread-spectrum modulated with a sinusoidal wave to obtain a transmitted signal with a wider spectrum, giving the system good anti-interference capabilities. It can detect faults at close range and allows for customized ranging ranges. The expression for the improved Logistic mapping is: in For mapping parameters, To map the value of the nth iteration, This is a perturbation term used to increase the Lyapunov exponent, thereby maximizing the Lyapunov exponent.

[0046] The signal transmission and acquisition module includes a DA digital-to-analog converter module, a multi-channel analog switch module, a signal transmission module, and a signal acquisition module.

[0047] The signal acquisition module further includes an adaptive filter unit for real-time separation of interference signals caused by multi-cable coupling. This adaptive filter unit dynamically adjusts the filter weights based on the Least Mean Square (LMS) algorithm to minimize the correlation between reflected and interference signals, thereby improving the signal-to-noise ratio. Specifically, the weight update formula for the adaptive filter is: w( n +1)=w( n )+ μ ⋅ e ( n )⋅x( n ) Among them, w( n ) indicates the first n The filter weight vector at time t. μ For step size parameters, e ( n Let x be the error signal, defined as the difference between the desired signal and the filter output. n () represents the input signal vector. Adaptive signal processing effectively suppresses cross-interference caused by multipath propagation, improving the quality of the acquired signal.

[0048] The signal acquisition module integrates a time-frequency analysis unit, employing continuous wavelet transform (CWT) to perform joint time-frequency domain analysis on the acquired reflected signals and distributed fiber optic vibration signals. This aims to analyze the non-stationary characteristics of the signals and enhance feature extraction. The wavelet transform formula for the time-frequency analysis unit is: in, x ( t () represents the acquired time-domain signal. ψ ( t ) represents the mother wavelet function. a For scale parameters, b is the translation parameter, and * denotes the complex conjugate. High-precision decomposition of the signal in the time-frequency domain was achieved through multi-resolution analysis, effectively distinguishing between reflected and vibration signals and reducing the impact of coupling interference.

[0049] The signal transmission module further includes a coupling compensation unit, which establishes a cable coupling model based on multi-conductor transmission line theory and performs pre-distortion processing before signal transmission to compensate for the mutual coupling effect between multiple cables. The coupling compensation unit uses the transmission line equation to pre-adjust the signal, which is expressed as: Where V(z,t) and I(z,t) represent the voltage and current vectors along the cable position z and time, respectively. t The distribution of L, R, C, and G represents the inductance, resistance, capacitance, and conductance matrices per unit length, respectively, where the off-diagonal elements characterize the coupling strength between cables. By numerically solving the above equations, the compensation unit generates an anti-coupling signal, thereby canceling multipath propagation interference at the transmitting end and improving the integrity of the reflected signal.

[0050] The fault detection module improves the generalized cross-correlation peak denoising delay estimation algorithm, reduces the interference of multi-path cables, improves the accuracy of fault detection, and enables preliminary judgment of the existence and type of fault.

[0051] Assuming the transmitted signal is The received signal is To accurately obtain the time delay estimates for these two signals, the influence of irrelevant noise needs to be filtered out. , After performing cross-correlation, a fractional Fourier transform (FrFT) is performed to obtain the cross-power function at the optimal order. Then, an inverse fractional Fourier transform is performed to obtain the improved generalized cross-correlation function: in For the fractional Fourier transform at the optimal order The inverse transform kernel obtained The time difference between the two impulse peaks is the estimated time delay d. If d is less than... By analyzing the polarity (positive or negative) of the cross-correlation peaks, it is possible to determine whether the fault is an open circuit or a short circuit.

[0052] The deep learning analysis module establishes time delay estimation models for multiple submarine DC cables based on deep learning algorithms, comprehensively considering the signal attenuation at its speed in the cable, and taking into account time delay d, cable parameters, and... Using frequency characteristics as input, an estimation model for the fault location y is established: Where L, C, and T represent the distributed inductance, distributed capacitance, and ambient temperature per unit length of cable, respectively. They represent The two most relevant characteristic frequencies of a signal in the fractional frequency domain are obtained through principal component analysis.

[0053] The deep learning analytics module also includes an initial cable database module, which stores parameters of the original, health status of all monitored cables, providing crucial prior knowledge benchmarks for the deep learning analytics module to be used for model training and labeling.

[0054] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0055] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0056] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0057] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0058] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for monitoring and locating multiple types of faults in submarine DC multi-channel cables, characterized in that, include: Based on the length of the cable under test, select m sequences with different symbol durations and multiply them with the chaotic sequence mapped by the improved Logistic map to obtain the improved chaotic sequence; The improved chaotic sequence is spread-spectrum modulated with a sine wave to form a transmitted signal; The transmitted signal is sequentially injected into the test points of each cable under test; Collect the reflected signals of each cable under test, estimate the time delay between the transmitted signal and the reflected signal, and determine whether there is a fault in each cable and the corresponding fault type. Based on the time delay and the corresponding cable parameters, a deep learning algorithm is used to achieve accurate fault location.

2. The method for monitoring and locating multiple types of faults in a submarine DC multi-channel cable according to claim 1, characterized in that, The expression for the improved Logistic mapping is: In the formula, To map the value of the (n+1)th iteration, To map the value of the nth iteration, For mapping parameters, For disturbance terms; In the process of determining chaotic sequences using the improved Logistic mapping, initial values ​​with chaotic state transition capabilities are obtained through multiple experiments. Mapping parameters and disturbance terms .

3. The method for monitoring and locating multiple types of faults in a submarine DC multi-channel cable according to claim 1, characterized in that, The process of sequentially injecting the transmitted signal into the test points of each cable under test includes: The spread spectrum modulated signal is converted into a high-voltage pulse signal. The digital high-voltage pulse signal is then converted into an analog signal according to a preset timing sequence, and then injected into different branches of the cable under test through a multi-channel analog switch array.

4. The method for monitoring and locating multiple types of faults in a submarine DC multi-channel cable according to claim 1, characterized in that, The method estimates the time delay between the transmitted signal and the reflected signal by improving the generalized cross-correlation peak denoising delay estimation algorithm. The improved generalized cross-correlation peak denoising time delay estimation algorithm performs cross-correlation calculations on the transmitted signal and the corresponding reflected signal, then performs a fractional Fourier transform to obtain the cross-power function at the optimal order; then performs an inverse fractional Fourier transform to obtain the improved generalized cross-correlation function, and uses the time difference between the two impulse peaks in the improved generalized cross-correlation function as the time delay.

5. The method for monitoring and locating multiple types of faults in a submarine DC multi-channel cable according to claim 4, characterized in that, The expression for the improved generalized cross-correlation function is as follows: In the formula, For the improved generalized cross-correlation function, This is the cross-power function at the optimal order. For the fractional Fourier transform at the optimal order The inverse transform kernel.

6. The method for monitoring and locating multiple types of faults in a submarine DC multi-channel cable according to claim 4, characterized in that, The determination of whether each cable has a fault and the corresponding fault type includes: If the time delay is less than If so, then a fault is determined to exist. For cable length, The speed at which signals propagate in the cable; By analyzing the polarity of the two impulse peaks, we can distinguish whether the corresponding fault type is an open-circuit fault or a short-circuit fault.

7. The method for monitoring and locating multiple types of faults in a submarine DC multi-channel cable according to claim 4, characterized in that, The deep learning algorithm uses time delay, cable parameters of the corresponding path, and the cross-power function at the optimal order as input to predict the fault location. The corresponding expression is: In the formula, Location of the fault. For time delay, L, C, and T represent the distributed inductance, distributed capacitance, and ambient temperature per unit length of cable, respectively. These represent the characteristic frequencies of the cross-power function signal at the optimal order in the fractional-order frequency domain.

8. The method for monitoring and locating multiple types of faults in a submarine DC multi-channel cable according to claim 1, characterized in that, During the acquisition of reflected signals from each cable under test, the method also uses an adaptive filter to separate interference signals caused by coupling between multiple cables. The adaptive filter dynamically adjusts its weights based on the least mean square algorithm, and the corresponding weight adjustment expression is: In( n +1)=in( n )+ μ ⋅ e ( n )⋅x( n ) Among them, w( n ) indicates the first n The filter weight vector at time t. μ For step size parameters, e ( n Let x be the error signal, defined as the difference between the desired signal and the filter output. n ) represents the input signal vector; In the process of acquiring the reflected signals of each cable under test, the method also employs continuous wavelet transform to perform joint time-frequency domain analysis on the acquired reflected signals and distributed optical fiber vibration signals, in order to analyze the non-stationary characteristics of the reflected signals and perform feature enhancement. The expression for the continuous wavelet transform is as follows: in, The result is a continuous wavelet transform. x ( t ( ) represents the time-domain reflected signal. ψ ( t ) represents the mother wavelet function. a For scale parameters, b is the translation parameter, and * denotes complex conjugate.

9. The method for monitoring and locating multiple types of faults in a submarine DC multi-channel cable according to claim 1, characterized in that, Before injecting the transmitted signal, the method establishes a cable coupling model based on multi-conductor transmission line theory and performs pre-distortion processing. The cable coupling model uses transmission line equations to pre-adjust the transmitted signal, and the expression of the transmission line equations is as follows: Where V(z,t) and I(z,t) represent the voltage and current vectors along the cable position z and time, respectively. t The distribution of L, R, C, and G represents the inductance, resistance, capacitance, and conductance matrices per unit length, respectively, where the off-diagonal elements characterize the coupling strength between cables.

10. A system for monitoring and locating multiple types of faults in a submarine DC multi-channel cable, implementing the method for monitoring and locating multiple types of faults in a submarine DC multi-channel cable as described in any one of claims 1-9, characterized in that, include: An improved SSTDR modulation module is used to select m-sequences with different symbol durations according to the length of the cable under test, and multiply them with the chaotic sequence mapped by the improved Logistic to obtain an improved chaotic sequence; the improved chaotic sequence is then spread spectrum modulated with a sine wave to form a transmitted signal. The signal transmission and acquisition module is used to sequentially inject the transmitted signal into the test points of each cable under test and acquire the reflected signal of each cable under test. The fault detection module is used to estimate the time delay between the transmitted signal and the reflected signal in order to determine whether there is a fault in each cable and the corresponding fault type. The deep learning analysis module is used to accurately locate faults based on the time delay and the corresponding cable parameters using deep learning algorithms.

Images