Cable fault diagnosis and positioning method and system based on DTS and DAS fusion

Abstract

The invention discloses a DTS and DAS fusion-based cable fault diagnosis and positioning method and system, and the method comprises the steps: synchronously collecting a distributed temperature sensing signal and a distributed sound wave sensing signal through a single sensing optical fiber which is laid together with a cable, and carrying out the time-space synchronization processing, obtaining temperature field data and phase disturbance field data which are aligned under the same space-time coordinate; calculating a cross-modal coherence spectrum of the temperature field data and the phase disturbance field data, quantifying physical coupling strength of vibration and temperature, and screening out potential fault candidate regions with coupling correlation; for the potential fault candidate region, adopting a dynamic attention mechanism of adaptive adjustment of a temperature change rate to perform nonlinear semantic fusion on the temperature feature and the vibration time-frequency feature to generate a fusion feature manifold; and calculating an entropy anomaly index reflecting a feature concentration ratio based on the generated fusion feature manifold, performing weighted calculation in combination with the coupling strength of the cross-modal coherence spectrum, and outputting an accurate spatial position and a fault type of a cable fault.

Description

Technical Field

[0001] This invention relates to the field of power cable fault diagnosis and location technology, and in particular to a cable fault diagnosis and location method and system based on the integration of DTS and DAS. Background Technology

[0002] As the core transmission and distribution carrier of urban power grids, the operating status of power cables directly affects the reliability of power supply. However, since cables are mostly laid in underground tunnels, utility corridors, or buried directly in the soil, locating and identifying faults in them is far more difficult than for overhead lines, placing extremely high demands on the accuracy and reliability of monitoring and location technologies.

[0003] Currently, the mainstream technologies applied to cable fault location and diagnosis mainly include traveling wave fault location method, distributed optical fiber sensing technology (DTS, DAS), or a "simple superposition" fusion scheme of the two, but they all have significant limitations:

[0004] 1. Traveling wave fault location method and its limitations

[0005] Traveling wave location method has good application results in overhead lines, but its location accuracy and reliability are seriously reduced in cable lines. The main reasons are: (1) Complex propagation characteristics of traveling waves: The distributed capacitance of cables is much larger than that of overhead lines, which leads to severe attenuation and waveform distortion of traveling waves during propagation. Especially for high-frequency fault traveling waves, their wave velocity is unstable and affected by various factors such as cable structure, insulation materials, and laying methods, making it difficult to calculate accurately. (2) Difficulty in calibrating wave velocity: The core of traveling wave location is accurate wave velocity. The wave velocity in cables is not constant. Cables from different manufacturers, different batches, and even different laying environments have different wave velocities. When converting the theoretical wave velocity or a certain calibration value into the actual cable length, unavoidable errors will occur, and the situation of "accurately determining the wave head but miscalculating the position" often occurs. (3) Inability to provide early warning: The traveling wave method is usually used to locate the fault point after a fault has occurred and caused a power outage. It does not have the function of condition monitoring and early warning, and belongs to the "post-event remediation" type of technology, which cannot avoid the losses caused by the fault.

[0006] 2. Monitoring technology based on single distributed optical fiber sensing and its limitations

[0007] Distributed fiber optic sensing technology (DTS, DAS) provides continuous and distributed sensing capabilities for cable condition monitoring, but it has fundamental defects when used independently.

[0008] Limitations of Distributed Temperature Sensing (DTS): (i) Missed alarms: It is ineffective for faults that do not produce significant temperature rises, such as cable sheath damage, initial mechanical damage, and open circuit faults. These faults may not be accompanied by temperature changes at the moment of occurrence. (ii) False alarms: Non-fault factors such as ambient temperature fluctuations, cross-connection loops, and uneven sunlight may cause abnormal temperature alarms, requiring frequent on-site verification by maintenance personnel, resulting in a heavy maintenance burden. (iii) Weak identification capability: Based solely on temperature information, it is impossible to distinguish the type of fault (e.g., it is impossible to distinguish whether it is joint overheating or intermediate section overload).

[0009] Limitations of Distributed Acoustic / Vibration Sensing (DAS): (i) High false alarm rate: In urban environments with abundant background noise sources (such as traffic, construction, and pedestrian flow), the system is prone to misinterpreting non-threatening vibrations as fault alarms. (ii) Missed alarms: It is not sensitive to faults occurring inside cables that do not produce strong mechanical vibrations (such as slow insulation degradation or the initial stage of partial discharge). (iii) Vague diagnosis: It can sense vibrations, but it is difficult to accurately determine the source and nature of the vibrations (excavators, drilling machines, normal vehicle traffic).

[0010] 3. Limitations of existing "simple overlay" fusion solutions

[0011] Currently, there are attempts to deploy DTS and DAS systems on the same cable simultaneously. However, such solutions mostly remain at the level of simple overlay of "independent data acquisition and separate alarms," ​​meaning that the two systems operate independently, displaying temperature alarms and vibration alarms side by side on the monitoring interface, leaving the arduous task of information fusion to the maintenance personnel.

[0012] Existing technologies either fail to achieve state monitoring and early warning (such as the traveling wave method), or suffer from high false alarms and high false negatives (such as a single DTS / DAS) due to the single sensing dimension, or fail to achieve true information fusion at the data processing level (such as a simple overlay scheme), resulting in a low level of intelligence. Summary of the Invention

[0013] To address the shortcomings and deficiencies of existing technologies, this invention provides a cable fault diagnosis and location method and system based on distributed optical fiber sensor fusion, aiming to solve the technical problems of high false alarm and false alarm rates in single-mode sensing technology, insufficient intelligence in simple dual-mode superposition schemes, and lack of early warning capabilities in traditional methods.

[0014] This method uses a single-mode optical fiber laid alongside the cable to simultaneously acquire distributed temperature and acoustic wave sensing signals. Through spatiotemporal synchronization processing, it achieves sub-meter spatial alignment and millisecond-level time synchronization within the same spatiotemporal coordinates, breaking down data silos between two modes. By calculating the cross-modal coherence spectrum to quantify the physical coupling strength between vibration and temperature, it accurately screens potential fault candidate regions and effectively eliminates environmental noise interference. For candidate regions, a dynamic attention mechanism with adaptive temperature change rate adjustment is employed to perform nonlinear semantic fusion of temperature features and vibration time-frequency features, generating a fused feature manifold containing multi-dimensional correlation information. Based on this fused feature manifold, the information entropy reflecting feature concentration is calculated, and weighted positioning is performed using the coupling strength of the cross-modal coherence spectrum, achieving accurate spatial location and type identification of cable faults. Simultaneously, a position adaptive correction factor is introduced to optimize temperature measurement accuracy, achieving an accuracy of ±0.3℃ under average conditions over hundreds of meters, with a dynamic range covering -40℃ to +150℃. For single-mode anomaly scenarios, enhanced monitoring is initiated by dynamically adjusting the sampling period or spatial resolution, enabling early warning several minutes before fault occurrence.

[0015] This invention significantly improves the accuracy, reliability, and timeliness of cable fault diagnosis through an integrated design of hardware reuse, cross-modal intelligent fusion, and single-sided anomaly early warning, providing efficient technical support for proactive operation and maintenance of power cables.

[0016] The present invention specifically adopts the following technical solution:

[0017] A cable fault diagnosis and location method based on the fusion of DTS and DAS includes:

[0018] Distributed temperature sensing signals and distributed acoustic wave sensing signals are synchronously acquired through a single sensing optical fiber laid in the same location as the cable, and spatiotemporal synchronization processing is performed to obtain temperature field data and phase perturbation field data aligned under the same spatiotemporal coordinates.

[0019] The cross-modal coherence spectrum of the temperature field data and the phase perturbation field data is calculated to quantify the physical coupling strength between vibration and temperature, and to screen out potential fault candidate regions with coupling correlation.

[0020] For the potential fault candidate region, a dynamic attention mechanism with adaptive temperature change rate is adopted to perform nonlinear semantic fusion of temperature features and vibration time-frequency features to generate a fused feature manifold.

[0021] Based on the generated fusion feature manifold, the entropy anomaly index reflecting the feature concentration is calculated. Combined with the coupling strength of the cross-modal coherence spectrum, a weighted calculation is performed to output the precise spatial location and fault type of the cable fault.

[0022] Furthermore, the single sensing fiber is a single-mode fiber, and synchronous acquisition is achieved by injecting a narrow-linewidth dual-wavelength pulse sequence; the dual-wavelength pulse sequence includes a first pulse adapted to Raman scattering and a second pulse adapted to Rayleigh scattering, and the parameters of the two pulses satisfy a timing offset relationship, wherein the timing offset is not greater than the product of the propagation delay of the total length of the fiber corresponding to the cable under test and the safety factor.

[0023] Furthermore, the spatiotemporal synchronization processing is achieved by embedding timestamps, which include second-level time references provided by GPS, nanosecond-level time corrections provided by a precise time protocol, and period compensation from a local clock.

[0024] Furthermore, the cross-modal coherence spectrum is a wavelet coherence spectrum, which is calculated by performing time-frequency dual-dimensional smoothing on the dynamic characteristics of the temperature field data and the time-frequency characteristics of the phase perturbation field data. An adaptive coherence threshold is used when screening potential fault candidate regions. The adaptive threshold is dynamically calculated based on the statistical mean and standard deviation of the cross-modal coherence spectrum within a sliding window. When the cross-modal coherence spectrum value is greater than the adaptive threshold, it is determined to be a potential fault candidate region.

[0025] Furthermore, the nonlinear semantic fusion process of the dynamic attention mechanism includes: projecting temperature features and vibration time-frequency features onto a preset dimension latent space, calculating the similarity between the two types of features through a dynamic Gram kernel function, allocating attention weights through Softmax normalization, and finally combining the vibration time-frequency features and temperature features after convolutional dimensionality reduction to output a fused feature manifold; the sensitivity of the attention mechanism is adaptively adjusted with the temperature change rate, and the greater the temperature change rate, the higher the semantic alignment sensitivity.

[0026] Furthermore, the entropy anomaly index is obtained by calculating the information entropy of the fused feature manifold; the weighted calculation uses the maximum value of the entropy anomaly index and the cross-modal coherence spectrum as weighting factors, and the centroid algorithm is used to solve for the precise spatial location of the fault.

[0027] Furthermore, a position adaptive correction factor is introduced when demodulating the temperature field data. This correction factor is used to offset the attenuation differences and wavelength effects at different fiber locations.

[0028] Furthermore, based on the precise spatial location of cable faults, a single-mode anomaly early warning step is also included: when only a slow temperature rise in the temperature field data is detected and there is no vibration coupling correlation, the sampling period is shortened and thermal latency enhancement monitoring is initiated; when only continuous vibration in the phase disturbance field data is detected and there is no significant temperature change, the spatial resolution is improved and external damage enhancement monitoring is initiated.

[0029] Furthermore, the fault type identification is achieved through a lightweight multi-task classification model: taking the fused feature manifold and entropy anomaly index as input, the fault type and risk level are output simultaneously; the temperature feature is a four-dimensional dynamic fingerprint vector containing thermal equilibrium value, heat flux change rate, thermal diffusion acceleration and thermal conduction gradient; the vibration time-frequency feature is obtained by calculating the local power spectral density after performing complex Morlet continuous wavelet transform on the phase perturbation field data.

[0030] And, a cable fault diagnosis and location method based on the integration of DTS and DAS, including:

[0031] The dual-mode signal acquisition and spatiotemporal synchronization module is used to synchronously acquire distributed temperature sensing signals and distributed acoustic wave sensing signals through a single sensing optical fiber laid in the same way as the cable, and to perform spatiotemporal synchronization processing on the two types of signals to output temperature field data and phase perturbation field data aligned under the same spatiotemporal coordinates.

[0032] The cross-modal coherence analysis and candidate region screening module is used to calculate the cross-modal coherence spectrum of the temperature field data and the phase perturbation field data, quantify the physical coupling strength between vibration and temperature, and screen out potential fault candidate regions with coupling correlation.

[0033] The dynamic attention feature fusion module is used to perform nonlinear semantic fusion of temperature features and vibration time-frequency features for the potential fault candidate region using a dynamic attention mechanism that adaptively adjusts the temperature change rate, and to generate a fused feature manifold.

[0034] The fault location and type identification module is used to calculate the entropy anomaly index reflecting the feature concentration based on the fused feature manifold, and to perform weighted calculation by combining the coupling strength of the cross-modal coherence spectrum to output the accurate spatial location and fault type of the cable fault.

[0035] And a computer device including a processor and a memory, the memory storing a computer program that, when executed by the processor, implements the steps of the method described above.

[0036] A non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0037] Compared with the prior art, the present invention and its preferred embodiments have at least the following beneficial effects:

[0038] This solution effectively overcomes the limitations of traditional single-mode sensing technology, which relies on a single sensing dimension. It achieves simultaneous acquisition and spatiotemporal alignment of temperature and acoustic signals using a single optical fiber, avoiding the complex structures of multi-fiber deployments. It also breaks down the data silos problem inherent in simple dual-mode overlay schemes, enabling effective correlation of multi-dimensional sensing information. By leveraging cross-modal coherence analysis and dynamic attention fusion mechanisms, the system significantly improves the ability to distinguish fault signals from environmental noise, substantially reducing false alarm and false negative rates, and making fault feature extraction and fusion more targeted and accurate. Through weighted calculation of fused features and coherence intensity, it achieves precise fault location and effective fault type identification, solving the problems of insufficient accuracy and ambiguous diagnosis inherent in traditional positioning technologies. Furthermore, the design of the single-mode anomaly early warning mechanism shifts the focus from "post-event remediation" to "pre-event prevention," enabling early intervention in cable faults and reducing losses. The overall solution, through deep coupling of hardware reuse and intelligent algorithms, simplifies system deployment while enhancing the intelligence level of diagnosis, providing reliable support for the efficient operation and maintenance of power cables. Attached Figure Description

[0039] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0040] Figure 1 This is a diagram illustrating the architecture of a cable fault diagnosis and location system based on the fusion of DTS and DAS, as described in an embodiment of the present invention.

[0041] Figure 2 This is a timing diagram of dual-wavelength pulse emission according to an embodiment of the present invention. Detailed Implementation

[0042] In the following, specific embodiments of this application will be described in detail with reference to the accompanying drawings. Based on these detailed descriptions, those skilled in the art will be able to clearly understand and implement this application. Without departing from the principles of this application, features from various embodiments can be combined to obtain new implementations, or certain features from some embodiments can be substituted to obtain other preferred implementations.

[0043] To make the features and advantages of the present invention more apparent and understandable, specific embodiments are described below in conjunction with the accompanying drawings:

[0044] This invention provides a cable fault diagnosis and location method and system based on the fusion of DTS and DAS, which can deeply integrate multi-dimensional information to achieve an integrated intelligent diagnosis solution for accurate cable fault location, type identification and early warning.

[0045] Its improvements include:

[0046] 1. Achieve integrated "precise location" and "type identification" of cable faults.

[0047] The system automatically associates and outputs the spatial location information of the fault with the nature of the fault (whether it is a short circuit, external damage, or overheating), thereby providing maintenance personnel with a direct and clear basis for emergency repair decisions.

[0048] 2. Reduce the false alarm and false negative rates in cable monitoring systems.

[0049] By employing an effective mechanism to cross-validate temperature and vibration signals, multi-dimensional information is used to filter false alarms and capture latent faults with only a single characteristic, thereby significantly improving the reliability of alarms and the reliability of the system.

[0050] 3. Break down the "data silos" between DTS and DAS systems to achieve true intelligent diagnosis.

[0051] Design a set of data fusion rules or models at the algorithm level to achieve automatic correlation and comprehensive analysis of temperature and vibration characteristics from the same cable and the same spatiotemporal coordinates. This will enable the system to autonomously identify specific fault modes such as "sudden temperature increase + current impact vibration", thereby freeing maintenance personnel from tedious manual information correlation work.

[0052] 4. Achieve a shift from "post-event remediation" to "pre-event early warning".

[0053] Traveling wave method is a reactive method for locating faults, while many faults have warning signs before they occur. This invention utilizes fused information from DTS and DAS to achieve trend judgment and early risk warning of cable operating status. For example, it can provide early warning by observing the slow temperature rise trend before cable joints overheat and burn out; and it can provide early warning by observing the vibration characteristics of continuous digging before power outages caused by external damage, thus transforming passive repair into proactive prevention.

[0054] like Figure 1 , Figure 2 As shown, the overall construction and implementation process of the present invention will be specifically demonstrated and introduced through the following embodiments:

[0055] 1. Fiber optic multiplexing and signal demodulation

[0056] This invention constructs a distributed sensing system with single-end injection and synchronous excitation via a dual-scattering mechanism on a single G.652.D single-mode optical fiber pre-bundled in a cable. Rayleigh phase field is achieved through precise timing arrangement of narrow-linewidth dual-wavelength pulse sequences and coherent / incoherent joint demodulation of backscattered light. Raman photon occupancy temperature field It achieves sub-meter spatial alignment and millisecond-level time synchronization, breaking down data silos in DTS / DAS. The specific implementation process includes:

[0057] (1) Construct a unified physical model for optical pulse excitation

[0058] To achieve a precise correspondence between fiber position and time, the actual propagation speed of light within the fiber must first be determined. Let the fiber's refractive index be... speed of light The speed of light in the fiber is:

[0059]

[0060] To simultaneously achieve temperature sensing (DTS) and vibration sensing (DAS) in the same optical fiber, this invention designs a dual-wavelength pulse emission rule and defines a dual-wavelength pulse emission timing function:

[0061]

[0062] in:

[0063] First pulse (DTS function): ;

[0064] Second pulse (DAS function): ;

[0065] rect() is a rectangular pulse function. () is the wavelength selection function;

[0066] Timing offset (i.e., transmission time difference) between the two pulses: , As a safety factor (to avoid the signals of two pulses overlapping in the optical fiber);

[0067] Among them, L max This represents the total length of the optical fiber corresponding to the cable being tested.

[0068] Theoretical limits of spatial resolution:

[0069]

[0070] (2) Construct a unified time-domain representation of the backscattered signal, including:

[0071] Backscattered power density at the fiber location:

[0072]

[0073] in, It is the normalized power at the incident end. It is the average fiber attenuation coefficient. It is the local scattering coefficient. It is the round-trip time delay (derived from equation (1)).

[0074] The above formula is a universal formula for all fiber backscattered signals, binding position z and time t to describe the power density of the scattered signal. Based on this, respectively:

[0075] For Rayleigh scattering (the core scattering mechanism of DAS), the general formula is transformed into a phase signal that can detect vibrations;

[0076] For Raman scattering (the core scattering mechanism of DTS), the general formula is transformed into a photon number ratio signal that can detect temperature;

[0077] ① Rayleigh coherent scattering (DAS) – Phase-sensitive OTDR (φ-OTDR)

[0078] DAS (Vibration Sensing) senses vibrations through phase changes, thus converting the signal from power to electric field phase:

[0079]

[0080] in, The power density in the general formula is converted into the electric field amplitude term. It is phase. It is a vibration-induced phase disturbance.

[0081] To convert phase perturbation into a measurable vibration quantity, based on existing φ-OTDR technology, this invention further derives the phase perturbation of adjacent pulse interference as follows:

[0082]

[0083] in, It is the pulse repetition period (the time interval between two transmitted pulses). It is the local strain of the optical fiber. To induce optical path difference due to strain, The conjugate signal of the electric field is represented by equation (4). By binding the phase difference and the fiber strain, the vibration intensity can be calculated by measuring the phase difference.

[0084] The vibration detection sensitivity of DAS is defined as:

[0085]

[0086] .

[0087] ② Raman incoherent scattering (DTS) – photon statistical temperature demodulation to convert the general backscattered signal into a photon number ratio signal that reflects temperature:

[0088] Existing Raman distributed temperature sensing (DTS) uses the anti-Stokes (AS) to Stokes (S) photon number ratio for temperature measurement:

[0089]

[0090] in:

[0091] (Silicon-oxygen Raman shift). (Differential attenuation).

[0092] Existing formulas are limited in accuracy due to fiber attenuation and location differences. Therefore, this invention incorporates a location-adaptive correction factor to provide an optimized temperature inversion formula:

[0093]

[0094] The position adaptive correction factor is:

[0095]

[0096] This is used to offset the attenuation difference at different locations and the influence of wavelength terms, making temperature calculations more accurate.

[0097] The optimized results are as follows: Temperature measurement accuracy: ±0.3℃ (100m average), Dynamic range: −40℃~+150℃.

[0098] (3) Mathematical modeling of spatiotemporal synchronization

[0099] The aforementioned DTS (temperature measurement) and DAS (vibration measurement) are two independent sensing channels, each with different spatial sampling intervals and time acquisition cycles. Therefore, they need to be bound to the same "position-time" grid so that temperature and vibration signals at any time and any location can correspond one-to-one.

[0100] Therefore, this invention defines a unified spatiotemporal grid:

[0101]

[0102] in, It is spatial resolution. This is the acquisition period. This allows the temperature signal from the DTS and the vibration signal from the DAS to be mapped to this unified "position z". i +time t k "On the grid, dual signal association at the same spatiotemporal point is achieved."

[0103] Accuracy is characterized by the upper bound of the spatiotemporal alignment error:

[0104]

[0105] Among them, the sources of error It's a clock jitter. It is a dual-wavelength timing offset. Therefore, the spatiotemporal alignment error between the DTS and DAS signals is at the sub-meter level, which can meet the accuracy requirements for cable fault location.

[0106] (4) Data Frames and Synchronization Tag Protocol

[0107] To ensure precise synchronization of DTS and DAS signals in terms of timestamps, this invention provides a design that embeds quantum-level timestamps into each frame of data:

[0108]

[0109] in, Provides a second-level time reference for GPS. Nanosecond-level time correction provided for PTP (Precision Time Protocol). This is for local clock cycle compensation.

[0110] By embedding high-precision timestamps into each frame of DTS and DAS data, the time stamps are perfectly aligned, ensuring that the data from the same spatiotemporal point in both channels are perfectly aligned.

[0111] In this embodiment, IEEE 1588v2 + White Rabbit fusion synchronization is preferably used to achieve end-to-end jitter <±8ns.

[0112] The implementation process of the above solution is as follows: Figure 2 As shown.

[0113] 2. Multimodal feature extraction and alignment:

[0114] To transform the raw, spatiotemporally synchronized sensor signals into correlated and fusionable multi-dimensional features, thus providing a foundation for subsequent fault location, identification, and early warning, this chapter uses the synchronization data demodulated in the previous chapter, including Rayleigh phase perturbations. With Raman temperature Using this as input, a time-frequency-spatial multimodal feature extraction and dynamic semantic alignment mechanism is constructed to achieve accurate characterization, localization, and early warning of gradual changes in fault events.

[0115] The final output is structured multimodal features and potential fault candidate regions, providing high-quality input for subsequent dynamic fusion (DGA-Net) and fault diagnosis.

[0116] Specifically, it includes:

[0117] (1) Time-frequency decomposition of DAS phase perturbation

[0118] For each position Applying complex Morlet continuous wavelet transform (CWT) to the phase signal:

[0119]

[0120] in, It is a wavelet scale ( Corresponding frequency ), Corresponding to the mother wavelet.

[0121] The local power spectral density is then calculated using the wavelet transform results.

[0122]

[0123] Obtain the DAS time-frequency-space feature tensor This provides a basis for identifying the source of vibration.

[0124] (2) Extraction of dynamic temperature features of DTS

[0125] Constructing a 4D dynamic fingerprint vector These correspond to the fourth-order dynamics of thermal equilibrium, heat flow, heat diffusion, and heat conduction, respectively.

[0126]

[0127] Among them, the thermal equilibrium T corresponds to the original temperature value as the basic judgment standard, and the heat flow... The temperature difference between adjacent time points is used to distinguish between rapid fault temperature rise and thermal diffusion. Corresponding to the acceleration of temperature rise, in order to detect latent fault trends and heat conduction The temperature difference between adjacent locations is used to locate local high-temperature sources. This avoids false alarms based on a single temperature threshold and provides dynamic temperature characteristics for subsequent vibration-temperature correlation.

[0128] (3) Cross-physics coherence modeling

[0129] Define the DAS-DTS cross-modal wavelet coherence spectrum to quantify the event coupling strength between vibration and temperature at the same spatiotemporal scale:

[0130]

[0131] in The time-frequency two-dimensional smoothing operator is preferred (in this embodiment, double exponential kernel convolution is preferred, with a time window of 30s and a frequency window of order 3).

[0132] Then, potential fault candidate regions are screened using an adaptive coherence threshold.

[0133] Among them, the coherence threshold adaptive threshold is:

[0134]

[0135] when Time triggering potential fault candidate region (This indicates that the vibration at this location / time is highly correlated with temperature, which is likely a fault.)

[0136] The above design quantifies the coupling relationship between vibration and temperature for the first time, solving the problem that existing simple superposition schemes cannot distinguish between associated faults and independent interferences. By pre-screening potential fault candidate areas, subsequent positioning and identification only need to focus on this area, which greatly reduces the amount of computation and improves accuracy.

[0137] (4) Dynamic Gram kernel attention fusion (DGA-Net)

[0138] To achieve nonlinear semantic alignment between vibration spectrum and temperature dynamics, this invention proposes a Dynamic Gram Attention Module (DGA), whose inputs include: DAS power spectral density tensor. and DTS temperature dynamic fingerprint Non-linear semantic alignment is achieved in three steps:

[0139] 1. Latent space projection ( Linear dimensionality reduction: reducing high-dimensional... and Projecting onto a 64-dimensional latent space, matching feature dimensions:

[0140]

[0141] Among them, W Q W K W V All of them are trainable weights.

[0142] 2. Dynamic Gram kernel (for calculating feature similarity):

[0143]

[0144] in, It is heat flow-driven attention tightening. Physical meaning: When the temperature is abnormal, vibration-temperature semantic alignment is more sensitive, enabling the adjustment of attention sensitivity based on heat flow.

[0145] It is the DAS feature projection vector (from ), It is the DTS feature projection vector (from ).

[0146] 3. Normalized Attention and Fusion Output: The weights are normalized using Softmax, and the fused feature manifold is output. :

[0147] Normalized attention weights:

[0148]

[0149] Fusion feature manifold output:

[0150]

[0151] in, It is a 1×1 convolution dimensionality reduction, used to reduce the dimensionality of the data. Dimensional reduction to match; It is the attention weight, which dynamically allocates the contribution of DAS / DTS features. It also includes the time-frequency information of vibration and the dynamic information of temperature, providing a unified feature input for subsequent fault type identification (such as distinguishing between short circuit, external damage, and joint overheating).

[0152] (5) Dual-threshold cross-validation enables precise fault location

[0153] Its input includes fusion features And the maximum value of the coherence spectrum, through three steps to achieve accurate fault location:

[0154] 1. Calculate the local fusion feature entropy to reflect the concentration of fused features and eliminate noise interference:

[0155]

[0156] in,: The lower the value, the more concentrated the fusion characteristics (high consistency between vibration and temperature), thus eliminating random noise interference.

[0157] 2. Calculating the entropy anomaly index to detect latent faults is also one of the core innovations of this invention:

[0158]

[0159] in, Calculate the mean / standard deviation using a sliding window (adaptive).

[0160] Even without severe vibration, slow thermal anomalies lead to reduce, It can also rise, thus triggering a warning.

[0161] 3. Final fault location (weighted centroid), i.e., using and As weights, the fault location is calculated:

[0162]

[0163] in, High / low confidence candidate regions This is the location of the fault.

[0164] 3. Cross-modal feature fusion and intelligent fault type identification

[0165] In previous chapters, fault location has been implemented. Screening of high-confidence fault candidate regions Based on this, the location of the fault is known, but the fault type still needs further determination. This section further constructs a lightweight multi-task recognition model, which integrates the previously output fused feature manifold. With entropy anomaly index As input, achieving accurate fault identification, risk quantification, and interpretable output can provide a direct and clear basis for operation and maintenance decisions.

[0166] Specifically, it includes:

[0167] (1) Input data

[0168] Fusion feature manifold : Fusion features after vibration-temperature semantic alignment (Equation 11).

[0169] Entropy Anomaly Index The entropy index reflects the intensity of latent anomalies. The higher the entropy value, the more significant the characteristic anomaly at that location / time.

[0170] High-confidence fault candidate region High-confidence fault area: 95% of noise has been filtered out. Analyzing this area can effectively avoid irrelevant interference.

[0171] Fusion feature manifold It includes typical manifestations of a fault (such as high-frequency vibrations and localized temperature increases when a cable is compressed), while the entropy anomaly index... This demonstrates how obvious and serious the fault is.

[0172] (2) Entropy-driven global feature aggregation

[0173] High-confidence fault candidate region Local features encompassing multiple spatiotemporal points are easily affected by slight interference if directly used for identification. Therefore, this embodiment uses entropy-driven global feature aggregation to achieve this. Entropy-weighted aggregation is performed on all spatiotemporal points within the region to automatically focus on the most significant anomaly, i.e., the most critical fault area:

[0174]

[0175] in, It is a globally aggregated feature vector; the higher the entropy, the greater the weight. It is an entropy anomaly index. It is a local encoding feature (by a lightweight Transformer) (Obtained through encoding).

[0176] (3) Multi-task identification head (outputs fault type and risk level in one step)

[0177] Based on global aggregation features This invention constructs a lightweight multi-task classification model that simultaneously outputs the fault type and risk level. The identification formula is as follows:

[0178]

[0179] in, It is a probability vector of fault types. ,Pick As a result of the identification, It is the classification weight matrix, which consists of trainable parameters. It is the classification bias vector, which is a trainable parameter.

[0180] During training, the "risk level (low / medium / high)" and "fault type" are bound as multi-task labels (e.g., "connector overheating + high risk" and "environmental interference + low risk"). Therefore, the risk level can be obtained simultaneously when the model outputs p.

[0181] 4. Detection and Early Warning of Latent Anomalies

[0182] Building upon the accurate identification of fault types achieved in the aforementioned three chapters, this section further explores the single-mode abnormal signals obtained in Chapter 2, constructing a "single-sided triggering → early prevention" mechanism. This mechanism initiates an early warning when both modes are not simultaneously abnormal, intervening 5-15 minutes in advance to achieve "prevention before the event".

[0183] (1) Input variables

[0184] Raman temperature field;

[0185] : Rate of temperature change;

[0186] DAS power spectral density;

[0187] Entropy anomaly index;

[0188] Cross-modal coherence spectrum;

[0189] The judgment of the early warning is based on the actual physical laws of cable faults:

[0190] If there is only a slow temperature rise without vibration (DTS slow rise): This corresponds to "latent thermal fault" (such as cable joints slowly aging without causing serious problems, but the temperature is already slowly rising).

[0191] If there is only continuous vibration and no temperature change (DAS continuous vibration): This corresponds to "external force damage risk" (for example, someone is digging near the cable and has not yet damaged the cable, but the vibration has already been transmitted to the cable).

[0192] (2) Enhanced anomaly triggering and dynamic monitoring

[0193] This chapter designs two early warning strategies to automatically enhance monitoring accuracy for different anomaly patterns, ensuring timely detection of potential problems.

[0194] Triggering conditions (start if any one of them is met):

[0195] Strategy 1: DTS gradual increase → enhanced thermal latency monitoring (when "temperature rises quietly but without vibration" is detected, "enhanced thermal latency monitoring" is activated):

[0196] Condition 1: ;

[0197] Condition 2: ;

[0198] Condition 3: DAS silent: .

[0199] Dynamic actions:

[0200]

[0201] in, Enhanced post-sampling period (unit: ms) This is the original sampling period (100ms). It is the acceleration factor (the higher the temperature rise, the denser the sampling).

[0202] Strategy 2: DAS continuous vibration → External rupture enhancement monitoring (When "continuous vibration is detected, but the temperature does not change", "external rupture enhancement monitoring" is activated)

[0203] Condition 1: Duration ≥30 seconds;

[0204] Condition 2: DTS silent: .

[0205]

[0206] in, This is the enhanced spatial resolution (unit: m). This is the original spatial resolution (0.8m). It is the resolution improvement factor (the stronger the vibration, the more precise the positioning).

[0207] The full-process architecture provided in the embodiments of the present invention is as follows: Figure 1 As shown.

[0208] Based on the same inventive concept, this invention also provides a computer device, comprising: one or more processors, and a memory for storing one or more computer programs; the programs include program instructions, and the processor executes the program instructions stored in the memory. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, used to implement one or more instructions, specifically for loading and executing one or more instructions stored in a computer storage medium to implement the above-described method.

[0209] It should be further explained that, based on the same inventive concept, the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, performs the above-described method. This storage medium can be any combination of one or more computer-readable media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0210] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0211] The foregoing has shown and described the basic principles, main features, and advantages of this disclosure. Those skilled in the art should understand that this disclosure is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this disclosure. Various changes and modifications can be made to this disclosure without departing from its spirit and scope, and all such changes and modifications fall within the scope of this disclosure as claimed.

[0212] This invention is not limited to the above-described preferred embodiments. Anyone inspired by this invention can derive other forms of cable fault diagnosis and location methods and systems based on the fusion of DTS and DAS. All equivalent variations and modifications made within the scope of the claims of this invention shall fall within the scope of this invention.

Claims

1. A cable fault diagnosis and positioning method based on fusion of DTS and DAS, characterized in that, The application relates to a cable fault diagnosis method based on a single sensing optical fiber. The method comprises the following steps: collecting distributed temperature sensing signals and distributed acoustic wave sensing signals synchronously through a single sensing optical fiber laid along a cable, and performing time-space synchronous processing to obtain temperature field data and phase disturbance field data aligned in the same time-space coordinate; Calculating cross-modal coherence spectrum of the temperature field data and the phase disturbance field data, quantifying the physical coupling strength of vibration and temperature, and screening out potential fault candidate regions with coupling correlation; For the potential fault candidate regions, a dynamic attention mechanism with temperature change rate self-adaptive adjustment is adopted to perform nonlinear semantic fusion on temperature features and vibration time-frequency features, and a fusion feature manifold is generated; Based on the generated fusion feature manifold, an entropy anomaly index reflecting feature concentration is calculated, and the coupling strength of the cross-modal coherence spectrum is used for weighted calculation to output the precise spatial position and fault type of the cable fault.

2. The cable fault diagnosis and location method based on the fusion of DTS and DAS according to claim 1, characterized in that: The single sensing optical fiber is a single-mode optical fiber, and synchronous collection is realized through narrow-line-width double-wavelength pulse sequence injection; the double-wavelength pulse sequence comprises a first pulse adapted to Raman scattering and a second pulse adapted to Rayleigh scattering, and the parameters of the two pulses satisfy a time sequence offset relationship, and the time sequence offset is not greater than the product of the propagation time delay of the total length of the optical fiber corresponding to the measured cable and a safety factor.

3. The cable fault diagnosis and location method based on the fusion of DTS and DAS according to claim 1, characterized in that: The time-space synchronous processing is realized by embedding a time stamp, and the time stamp comprises a second-level time reference provided by a GPS, a nanosecond-level time correction provided by a precise time protocol and a period compensation of a local clock.

4. The cable fault diagnosis and location method based on the fusion of DTS and DAS according to claim 1, characterized in that: The cross-modal coherence spectrum is a wavelet coherence spectrum, which is calculated after time-frequency double-dimensional smoothing processing is performed on the dynamic features of the temperature field data and the time-frequency features of the phase disturbance field data; When screening the potential fault candidate regions, an adaptive coherence threshold is adopted, and the adaptive threshold is dynamically calculated based on the statistical mean value and the standard deviation of the cross-modal coherence spectrum in a sliding window; when the cross-modal coherence spectrum value is greater than the adaptive threshold, the region is determined as a potential fault candidate region.

5. The cable fault diagnosis and location method based on the fusion of DTS and DAS according to claim 1, characterized in that: The nonlinear semantic fusion process of the dynamic attention mechanism comprises the following steps: projecting the temperature features and the vibration time-frequency features into a preset dimension latent space respectively, calculating the similarity of the two types of features through a dynamic Gram kernel function, assigning attention weights through Softmax normalization, and finally outputting the fusion feature manifold in combination with the vibration time-frequency features and the temperature features after convolution dimension reduction; the sensitivity of the attention mechanism is adaptively regulated according to the temperature change rate; the greater the temperature change rate, the higher the semantic alignment sensitivity.

6. The method according to claim 1, characterized in that: The entropy anomaly index is obtained by calculating the information entropy of the fusion feature manifold; the weighted calculation takes the maximum value of the entropy anomaly index and the cross-modal coherence spectrum as a weighting factor, and the precise spatial position of the fault is solved through a centroid algorithm.

7. The method according to claim 1, characterized in that: When demodulating the temperature field data, a position self-adaptive correction factor is introduced, and the correction factor is used to offset the attenuation difference and wavelength influence of different optical fibers.

8. The method according to claim 1, characterized in that: On the basis of accurate spatial location of cable fault, a single-mode abnormal early warning step is further included: when only slow temperature rise of temperature field data is detected and no vibration coupling correlation is found, the sampling period is shortened and heat latency enhanced monitoring is started; when only continuous vibration of phase disturbance field data is detected and no obvious temperature change is found, the spatial resolution is improved and external damage enhanced monitoring is started.

9. The cable fault diagnosis and location method based on the fusion of DTS and DAS according to claim 1, characterized in that: The fault type identification is realized by a lightweight multi-task classification model: taking the fusion feature manifold and entropy anomaly index as input, the fault type and risk level are synchronously output; the temperature feature is a four-dimensional dynamic fingerprint vector containing heat balance value, heat flow change rate, heat diffusion acceleration and heat conduction gradient, and the vibration time-frequency feature is obtained by calculating the local power spectral density after complex Morlet continuous wavelet transform of the phase disturbance field data.

10. A cable fault diagnosis and location based on fusion of DTS and DAS, characterized in that, The method comprises: a dual-mode signal acquisition and space-time synchronization module, configured to synchronously acquire distributed temperature sensing signals and distributed acoustic wave sensing signals through a single sensing optical fiber laid with the cable, and perform space-time synchronization processing on the two types of signals to output aligned temperature field data and phase disturbance field data in the same space-time coordinate; a cross-modal coherence analysis and candidate area screening module, configured to calculate the cross-modal coherence spectrum of the temperature field data and the phase disturbance field data, quantify the physical coupling strength of vibration and temperature, and screen out potential fault candidate areas with coupling correlation; a dynamic attention feature fusion module, configured to, for the potential fault candidate areas, adopt a dynamic attention mechanism with temperature change rate adaptive adjustment to perform nonlinear semantic fusion on the temperature feature and the vibration time-frequency feature, and generate a fusion feature manifold; a fault location and type identification module, configured to calculate an entropy anomaly index reflecting the feature set density based on the fusion feature manifold, combine the coupling strength of the cross-modal coherence spectrum for weighted calculation, and output the accurate spatial location and fault type of the cable fault.

Images