Ground wire steel core detection system based on digital enabling

Abstract

The invention belongs to the technical field of ground wire steel core detection, and particularly relates to a ground wire steel core detection system based on digital enabling, which comprises a detection device module, a data acquisition and processing module, a data analysis and intelligent identification module and a visualization and early warning platform module, the detection device module is integrated with a multi-mode sensing unit (a magnetizing device, a Hall array sensor, an ultrasonic guided wave transducer and an optical fiber sensing array) and a displacement positioning assembly, so that comprehensive acquisition and accurate positioning of damage signals are realized; the data analysis module is fused with a CNN-LSTM deep learning model to complete defect automatic identification and residual life prediction; the visual platform realizes hierarchical display and intelligent early warning based on digital twinning, and is also provided with a block chain module to guarantee data credibility; the system is compatible with portable inspection and on-line monitoring, is adaptive to the existing AEMR-I system, can improve the damage detection precision, the intelligent level and the operation and maintenance efficiency, and provides support for the safe operation of a power grid.

Description

Technical Field

[0001] This invention belongs to the field of conductor and ground wire steel core detection technology, specifically a conductor and ground wire steel core detection system based on digital empowerment. Background Technology

[0002] The conductor and ground wire core is the core load-bearing component of power transmission lines. Damage such as broken strands, cracks, and corrosion directly affects the safe operation of the power grid. Therefore, non-destructive testing (NDT) technology is needed for condition monitoring and defect diagnosis. Currently, in the field of wire rope inspection, which is similar to that of conductor and ground wire cores, leakage magnetic flux and metal cross-sectional area detection technologies are relatively mature. A typical example is the AEMR-I inspection system. This system includes a magnetic sensor, a displacement locator, a real-time alarm, and supporting analysis software. It can collect leakage magnetic flux and metal cross-sectional area change signals through a magnetization device and a Hall array. After threshold discrimination and human-machine interaction, it outputs reports on broken wires and wear. The supporting software supports sampling control, threshold calibration, broken wire and wear analysis, and can set parameters such as diameter, lay length, sampling interval, metal cross-sectional area, and single wire diameter. It also has offline and online calibration processes, as well as automatic and manual analysis modes.

[0003] However, there are still significant shortcomings when applying existing technologies to the testing of conductor and ground wire steel cores:

[0004] Firstly, the aluminum coating or composite structure on the conductor and ground wire will cause the detection signal of the single leakage magnetic field method to be attenuated, which will easily lead to ambiguity in the identification of internal damage of the steel core and insufficient penetration ability.

[0005] Secondly, single-mode detection has limited sensitivity to crack morphology and early corrosion, making it difficult to comprehensively capture different types of damage;

[0006] Third, due to the influence of stock wave background and end effect, the accuracy of defect location and quantitative assessment needs to be further improved;

[0007] Fourth, the lack of multimodal fusion and deep learning technology support makes it impossible to achieve closed-loop management of automatic damage identification and remaining life prediction.

[0008] Fifth, it lacks engineering adaptability and has low compatibility with online monitoring, third-party testing and acceptance, and visual operation and maintenance platforms, making it difficult to meet the diverse operation and maintenance needs of the power grid.

[0009] In summary, existing conductor and ground wire steel core testing technologies have shortcomings in terms of testing capabilities, intelligence level, and engineering applications, and there is an urgent need for a testing system that can overcome these limitations. Summary of the Invention

[0010] To overcome the shortcomings of existing technologies and solve the aforementioned technical problems, this invention proposes a digitally enabled conductor and ground wire steel core detection system. This system utilizes penetrating, integrated non-destructive testing to achieve automatic and accurate identification and quantification of damage, and enables visualized predictive maintenance and intelligent operation and maintenance closed-loop based on digital twins.

[0011] The technical solution adopted by the present invention to solve its technical problem is as follows: The present invention proposes a conductor and ground wire steel core detection system based on digital empowerment, including a detection device module, a data acquisition and processing module, a data analysis and intelligent identification module, and a visualization and early warning platform module;

[0012] The detection device module includes a multi-mode sensing unit for acquiring sensing signals required for steel core damage detection, a displacement positioning component for acquiring positioning information corresponding to the sensing signals, and a clamping structure for clamping the cable externally; the sensing signals required for steel core damage include leakage magnetic field signals, ultrasonic guided wave signals, and fiber optic strain signals.

[0013] The data acquisition and processing module is used to acquire and preprocess leakage magnetic field signals, ultrasonic guided wave signals, fiber optic strain signals and positioning information to obtain a sensor dataset; wherein, the preprocessing includes noise reduction, filtering and time-frequency domain analysis.

[0014] The data analysis and intelligent identification module is used to analyze the leakage magnetic signal using a threshold mechanism to obtain information on broken strands or cracks, and to analyze the change in metal cross-sectional area to obtain information on wear or corrosion. Combining the information on broken strands or cracks with the information on wear or corrosion, candidate defect points are determined.

[0015] The data analysis and intelligent recognition module is used to extract the amplitude envelope, frequency domain energy and group velocity time difference features of ultrasonic guided wave signals through multimodal data fusion technology, and to extract strain peaks and valleys and local anomalies from fiber strain signals, and to obtain multimodal fusion features through a consensus voting mechanism.

[0016] The data analysis and intelligent recognition module is used to classify and regress candidate defect points using multimodal fusion features through a pre-trained deep learning network, determine quantitative and localized steel core damage information, and predict remaining life.

[0017] The visualization and early warning platform module is used to construct a conductor and ground wire steel core model through digital twin technology, and to visualize, intelligently warn, and automatically generate detection reports on quantitative and localized steel core damage information and remaining life.

[0018] Furthermore, the multimodal sensing unit of the detection device module includes a magnetization device and a Hall array sensor, an ultrasonic guided wave transducer, and an optical fiber sensing array.

[0019] The magnetization device is used to axially magnetize the conductor steel core to generate a stable leakage magnetic field, and the Hall array sensor is used to capture the leakage magnetic signal caused by damage to the steel core.

[0020] The ultrasonic guided wave transducer transmits and receives ultrasonic guided wave signals to achieve continuous detection of steel core damage.

[0021] The fiber optic sensing array is used to acquire fiber optic strain signals to detect the strain distribution in the steel core.

[0022] Furthermore, the displacement positioning component includes a guide wheel and a displacement encoder, the guide wheel and the displacement encoder are linked, and the displacement encoder generates positioning information corresponding to the sensing signal.

[0023] Furthermore, the data acquisition and processing module transmits the acquired magnetic flux leakage signal, ultrasonic guided wave signal, and fiber optic strain signal to the central processing unit via an RS232 interface or a USB interface; during preprocessing, wavelet transform technology is used to denoise the signal, and abnormal strain is identified in the fiber optic strain signal to eliminate non-defect interference signals caused by temperature changes.

[0024] Furthermore, the pre-trained deep learning network is a combined model of convolutional neural network and long short-term memory network; wherein, the convolutional neural network extracts the spatial features of the magnetic flux leakage signal, and the long short-term memory network captures the temporal features of the ultrasonic guided wave signal.

[0025] The steel core damage information output by the pre-trained deep learning network includes: defect types such as broken strands, corrosion, and cracks, defect locations, equivalent number of broken strands, crack length, corrosion volume fraction, and confidence level.

[0026] The threshold mechanism for analyzing magnetic flux leakage signals includes: a first threshold for preliminary screening of broken strands and cracks, with the amplitude of a single broken wire signal set at 85%; and a second threshold, determined in conjunction with the wire diameter and structural calibration, for quantitative defect identification.

[0027] The metal cross-sectional area change analysis includes: setting the sensitivity and cross-sectional reference of the metal cross-sectional area change signal, with a threshold range of 0.5% to 1%, for quantitative identification of wear and corrosion.

[0028] Furthermore, the conductor and ground wire steel core model supports hierarchical scaling and positioning at the line level, tower level, crossing section level, conductor and ground wire level, and steel core level, and displays the defect distribution and evolution trend through heat maps.

[0029] Furthermore, it also includes a blockchain data security module, which records the detected steel core damage information, location information, and timestamp data to the blockchain and stores them encrypted through smart contracts to ensure the immutability and traceability of the data. The detection device module adopts a modular design, which can be quickly installed and disassembled without power interruption. It is compatible with portable inspection mode and online long-term monitoring mode, and is adapted to the AD acquisition link and reporting system of the existing AEMR-I flaw detector.

[0030] Furthermore, the magnetization device uses a permanent magnet or an electromagnetic coil, which can penetrate the aluminum coating or composite structure outside the ground wire; the Hall array sensor has a ring structure and is distributed circumferentially around the ground wire.

[0031] Furthermore, the ultrasonic guided wave transducer is a low-frequency longitudinal guided wave transducer, which identifies the morphology and damage degree of steel core cracks by receiving defect reflection or scattering signals and combining the time-domain difference and spectral characteristics of ultrasonic signals; the crack morphology includes open cracks and closed cracks.

[0032] Furthermore, the fiber optic sensing array is a fiber Bragg grating sensing array, which is evenly mounted on the outside of the detection device module using clamps.

[0033] The beneficial effects of this invention are as follows:

[0034] 1. The digitally-enabled conductor steel core detection system of the present invention integrates multi-modal sensing units to simultaneously acquire leakage magnetic field, ultrasonic, and fiber optic signals in a clamping manner. Combined with displacement positioning components, it achieves comprehensive capture and precise positioning of data related to damage such as broken strands, cracks, and corrosion in the conductor steel core. The leakage magnetic field signal is sensitive to broken strands and obvious cracks, the ultrasonic guided wave signal can reflect the morphology (open / closed) and internal defects of cracks, and the fiber optic strain signal reflects the micro-strain caused by early corrosion and stress concentration. It has the advantages of overcoming the insufficient penetration capability and single-modal recognition limitations of single leakage magnetic field detection, and improving the accuracy and coverage of damage detection.

[0035] 2. The digitally-enabled conductor core inspection system of this invention, through data analysis and intelligent identification modules, utilizes mechanisms such as dual thresholds to quickly locate suspicious defect points. Combined with metal cross-sectional area change analysis, it initially screens and judges the number of broken strands and cross-sectional area loss rate. A deep learning network performs regression analysis on multimodal fusion features, enabling more accurate calculation of continuous variables such as crack length and corrosion depth. By fusing signal features from different physical principles and performing internal cross-validation, the system ultimately uses a pre-trained deep learning network to fuse multimodal signals for automatic classification, regression, and lifetime prediction. This defect identification, classification, and quantification reduce reliance on operator skills.

[0036] 3. The conductor and ground wire steel core detection system based on digital empowerment described in this invention achieves automatic damage identification, remaining life prediction, and reliable traceability of detection data through data analysis and intelligent identification modules, digital twin technology, and blockchain data security modules. It reduces reliance on manual analysis, constructs a closed-loop operation and maintenance system of "detection-analysis-prediction-maintenance", and improves the level of intelligence and data credibility. Attached Figure Description

[0037] The invention will now be further described with reference to the accompanying drawings.

[0038] Figure 1 This is a structural diagram of the conductor and ground wire steel core detection system based on digital empowerment of the present invention;

[0039] Figure 2 This is an operation flowchart of the conductor and ground wire steel core detection system based on digital empowerment of the present invention. Detailed Implementation

[0040] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] like Figures 1-2 As shown, to make the technical solution of the present invention clear and implementable, and in conjunction with the actual operation and maintenance scenario of conductor and ground wire steel core detection, the following three typical embodiments provide a detailed description of the digitally enabled conductor and ground wire steel core detection system of the present invention. Each embodiment strictly follows the technical solution defined in the claims. The component selection, parameter settings, and process design are all derived from actual R&D practice, with a focus on refining the core logic of the data analysis and intelligent identification modules to ensure that those skilled in the art can reproduce it accordingly.

[0042] Example 1: This example provides a lightweight and portable detection device for the needs of manual on-site inspection of transmission lines. The core of the device is to quickly detect and locate broken strands, cracks, and corrosion in the steel core.

[0043] The detection device module:

[0044] Clamping structure: The frame is made of aluminum alloy and can be opened and closed. It weighs ≤5kg and has an opening angle ≥90°. It is fixed to the outside of the ground wire (compatible with LGJ-400 / 50 to LGJ-1200 / 100 models) by quick buckle. It can be clamped to the outside of the cable without disassembling the line.

[0045] The multimodal sensing unit:

[0046] The magnetization device includes a rare-earth permanent magnet (magnetic flux density ≥ 0.8T), which is attached to the ground wire through an arc-shaped bracket to achieve axial magnetization of the steel core and generate a stable leakage magnetic field that can penetrate the outer aluminum layer or composite structure.

[0047] The Hall array sensor is a 12-channel ring array (30° spacing), circumferentially distributed around the conductor and ground wire. It uses SS495A Hall elements with a sensitivity of ≥31.2mV / G and is specifically designed to capture leakage magnetic signals caused by steel core damage.

[0048] The ultrasonic guided wave transducer is a low-frequency longitudinal guided wave transducer with an operating frequency of 50kHz (adjustable from 30-100kHz). It is coupled to the outer wall of the ground wire through an elastic silicone pad to transmit and receive ultrasonic guided wave signals, enabling long-distance continuous detection.

[0049] The fiber optic sensing array, a fiber Bragg grating sensing array, is uniformly mounted on the outside of the device using high-temperature and corrosion-resistant fixtures to collect fiber strain signals in real time for detecting the strain distribution of the steel core.

[0050] The displacement positioning component includes a rubber-coated guide wheel (80mm outer diameter, anti-slip texture design) and an incremental displacement encoder (0.5mm resolution). The two work together to achieve equidistant sampling. The displacement encoder generates positioning information that corresponds one-to-one with the sensing signal, ensuring that the sampling point and the physical position of the conductor are accurately matched.

[0051] The data acquisition and processing module has a built-in 24-bit AD conversion chip (ADS1256) and an ARM embedded processor (STM32H743). It acquires leakage magnetic field signals, ultrasonic guided wave signals, fiber optic strain signals, and positioning information through an RS232 interface or a USB interface, with an acquisition frequency of 1kHz. In the preprocessing stage, wavelet transform technology (db4 wavelet basis, 4 decomposition layers) is used for noise reduction. Combined with FIR low-pass filtering (cutoff frequency 1kHz) and time-frequency domain analysis (short-time Fourier transform window length 256 points), a sensor dataset is generated. At the same time, abnormal strain identification is performed on the fiber optic strain signal (setting 3 times the standard deviation as the threshold) to remove non-defect interference signals caused by temperature changes.

[0052] The data analysis and intelligent recognition module is integrated into the host computer software and has a built-in pre-trained convolutional neural network and long short-term memory network combined model. The model training dataset contains 5,000 sets of conductor and ground wire defect samples (covering 1-5 broken strands, crack length 0.3-5mm, and corrosion volume fraction 1%-10%).

[0053] The visualization and early warning platform module relies on host computer software to achieve basic visualization, supports hierarchical scaling of the conductor and ground wire steel core model, and displays the defect distribution through heat maps.

[0054] The detailed data analysis and intelligent recognition process includes:

[0055] Determination of candidate defects:

[0056] Threshold mechanism analysis of leakage magnetic signals:

[0057] First, the baseline value of the single broken wire signal amplitude (denoted as V0) is calibrated using a standard sample rope (including single broken wire defects). The first threshold is set to 0.85×V0. When the amplitude of the collected leakage magnetic signal exceeds this threshold, it is initially determined to be a suspected point of broken strand or crack.

[0058] The second threshold is determined by the diameter of the conductor wire (set as d) and the structure: for broken strand defects, the second threshold is set as n×V0 (n is the estimated number of broken wires, calculated by the ratio of the leakage magnetic signal amplitude to V0); for crack defects, the second threshold is set as the amplitude threshold corresponding to the crack length (obtained through experimental fitting: the relationship between crack length L and leakage magnetic amplitude V is V=0.2L+0.1V0), to achieve preliminary quantitative judgment of defects.

[0059] Analysis of changes in metal cross-sectional area:

[0060] The reference value of the metal cross-sectional area (denoted as S0) is calculated by integrating the leakage magnetic signal. The sensitivity coefficient is set to 0.01. The reference value S0 is calibrated by a defect-free conductor segment.

[0061] When the difference between the detected metal cross-sectional areas S and S0 exceeds 0.5% but is less than 1%, it is judged as slight wear or corrosion; when it exceeds 1%, it is judged as severe wear or corrosion, and wear or corrosion information is obtained.

[0062] Candidate defect point fusion: Match the suspected broken strand or crack point with the spatial location of wear or corrosion information (based on the positioning information of the displacement positioning component). When the positional deviation between the two is ≤0.1m, it is determined as a candidate defect point.

[0063] Multimodal fusion feature extraction includes:

[0064] Ultrasonic guided wave signal feature extraction:

[0065] Amplitude envelope: The ultrasonic guided wave signal is processed using Hilbert transform to extract the amplitude envelope curve of the signal and calculate the peak value and peak width of the envelope;

[0066] Frequency domain energy: Perform Fourier transform on the ultrasonic guided wave signal, select the 30-100kHz frequency band, and calculate the signal energy integral within this frequency band;

[0067] Group velocity time difference: By using the time difference Δt between the transmitted signal and the defect reflection / scattering signal, combined with the guided wave group velocity v (set to 3200m / s), the distance ΔL from the defect to the detection point is calculated as ΔL = v × Δt / 2.

[0068] Fiber optic strain signal feature extraction:

[0069] The peak and valley values ​​of the fiber strain signal were extracted using the sliding window method (with a window length of 100 sampling points), and the peak and valley differences were calculated.

[0070] The threshold for identifying local anomalies is set at ±3 times the standard deviation of the strain mean. When the strain signal exceeds this range, it is identified as a local anomaly.

[0071] Consensus voting mechanism:

[0072] Voting weights are set as follows: leakage magnetic field signal characteristic weight 0.4, ultrasonic guided wave signal characteristic weight 0.3, and fiber strain signal characteristic weight 0.3.

[0073] For a candidate defect point, if a certain feature is judged as "defect exists", it gets 1 vote; otherwise, it gets 0 votes. The total score is 0.4 × number of magnetic leakage votes + 0.3 × number of ultrasonic votes + 0.3 × number of fiber optic votes. When the total score is ≥ 0.6, it is determined as a valid defect point, and the multimodal fusion feature is output.

[0074] Deep learning network classification, regression, and remaining lifetime prediction include:

[0075] Model input and output:

[0076] Input: Multimodal fusion features (including 12-dimensional feature vectors such as leakage magnetic field amplitude, metal cross-sectional area change rate, ultrasonic guided wave amplitude envelope peak value, frequency domain energy, group velocity time difference, fiber strain peak-valley difference, etc.);

[0077] Output: Defect type (broken strand, crack, corrosion), defect location (based on displacement positioning information and group velocity time difference correction), equivalent broken strand count, crack length, corrosion volume fraction, and confidence level (model output probability value).

[0078] Model structure and training:

[0079] The convolutional neural network consists of three convolutional layers (with kernel sizes of 3×3, 3×3, and 5×5, and output channels of 32, 64, and 128, respectively) and two max-pooling layers (with a kernel size of 2×2) used to extract the spatial features of the magnetic flux leakage signal.

[0080] The long short-term memory network consists of two hidden layers (each with 64 neurons) used to capture the temporal characteristics of ultrasonic guided wave signals.

[0081] Training parameters: Optimizer is Adam, learning rate is 0.001, number of iterations is 1000, batch size is 32, the ratio of training set to test set is 8:2, and the accuracy of the test set is ≥98%.

[0082] Remaining life prediction:

[0083] Based on historical detection data (defect evolution data of the past 3 years) and real-time detection defect parameters (equivalent number of broken strands, crack length, corrosion volume fraction), a remaining life prediction model is established.

[0084] Linear regression combined with time series analysis was used: for strand breakage defects, the remaining lifetime L1 = (maximum allowable number of broken wires - current equivalent number of broken wires) / average annual growth rate of broken wires; for crack defects, the remaining lifetime L2 = (maximum allowable crack length - current crack length) / average annual crack propagation rate; for corrosion defects, the remaining lifetime L3 = (maximum allowable corrosion volume fraction - current corrosion volume fraction) / average annual corrosion growth rate; the minimum value among L1, L2, and L3 was taken as the final remaining lifetime.

[0085] The overall operation process and implementation effect of this embodiment are as follows:

[0086] Operation process: Device installation → Parameter calibration (amplitude of single broken wire, reference value of metal cross-sectional area) → Signal acquisition → Preprocessing → Candidate defect point determination → Multimodal fusion feature extraction → Deep learning classification and regression → Remaining lifetime prediction → Visualization and report generation.

[0087] Implementation results: During the inspection of 220kV transmission lines, two broken strands (equivalent to 2 and 3 broken strands respectively), one crack (1.2mm in length), and two corrosions (volume fractions of 0.8% and 1.5% respectively) were successfully detected. The defect location error was ≤0.1m, the confidence level was ≥95%, the remaining life prediction error was ≤6 months, and the detection efficiency was 2 times higher than that of traditional equipment.

[0088] Example 2: This example addresses the centralized monitoring needs of power grid operation and maintenance centers by constructing a software-based analysis platform to achieve unified processing of multi-device detection data and refine the data analysis and intelligent identification process for batch data.

[0089] Platform architecture and component configuration:

[0090] Hardware support: Industrial server (CPU is Intel Xeon Gold 6348, memory is 64GB, storage is 2TB SSD), which receives data uploaded by multiple portable devices via Ethernet;

[0091] The data acquisition and processing module supports parallel data reception from multiple devices, adopts a distributed processing architecture, performs synchronous preprocessing on batch data, and generates standardized sensor datasets.

[0092] The data analysis and intelligent recognition module is based on the PyTorch framework and deploys a combined model of convolutional neural networks and long short-term memory networks, supporting online model updates and batch data parallel inference.

[0093] The visualization and early warning platform module uses the Unity 3D engine to build a digital twin model of the conductor and ground wire steel core, supporting hierarchical scaling and positioning at the line level, tower level, crossing section level, conductor and ground wire level, and steel core level.

[0094] The blockchain data security module, based on the Hyperledger Fabric blockchain platform, records detection data and analysis results.

[0095] The detailed data analysis and intelligent recognition process includes:

[0096] Batch data preprocessing optimization

[0097] A timestamp alignment algorithm is used to synchronize the sensor signals and positioning information uploaded by multiple detection devices according to timestamps, ensuring the spatiotemporal consistency of the data;

[0098] To address the redundancy issue in batch data, the K-means clustering algorithm is used to deduplicate similar signals, retaining valid data and improving processing efficiency.

[0099] Multimodal fusion and deep learning inference optimization include:

[0100] Batch normalization is used to process multimodal fusion features to reduce system errors between different devices;

[0101] The deep learning model adopts a batch inference mode, processing 100 sets of data each time with an inference time of ≤1 second. It also supports online model updates: when the number of new defect samples exceeds 1000 sets, the model fine-tuning is automatically started to update the network parameters.

[0102] Dynamic revisions to remaining lifetime predictions include:

[0103] The remaining life prediction results are corrected by combining environmental factors (temperature, humidity, wind speed): the environmental correction coefficient k is obtained through experimental fitting, and the final remaining life = initial predicted life × k (the value of k ranges from 0.8 to 1.2, and can be obtained by querying real-time environmental data).

[0104] Establish a remaining life prediction database to store historical prediction results and actual operation and maintenance data, and optimize the prediction model through feedback iteration.

[0105] The implementation effect of this embodiment is as follows:

[0106] It supports simultaneous processing of data uploaded from 10 detection devices, with a batch data processing efficiency of ≥1000 sets / minute;

[0107] Defect identification confidence level ≥ 98%, remaining lifetime prediction error ≤ 3 months;

[0108] When the defect level reaches the critical standard or the remaining lifespan is ≤1 year, an automatic warning will be sent via SMS and platform pop-up, along with maintenance suggestions.

[0109] Example 3: This example addresses the long-term monitoring needs of important transmission lines of 500kV and above by providing a fixed-installation online system to achieve 24-hour uninterrupted monitoring, with a focus on refining real-time data analysis and intelligent early warning logic.

[0110] System deployment and component configuration include:

[0111] The detection device module is fixed to the middle of the conductor-ground crossing section by a 316 stainless steel clamp. The magnetization device of the multi-mode sensing unit is replaced by an electromagnetic coil (input voltage 24V, current 5A). The fiber optic sensing array is expanded to an 8-point type. The displacement positioning component adopts a high-precision encoder (resolution 0.1mm).

[0112] The data acquisition and processing module adopts an industrial-grade acquisition card, supports 24-hour continuous acquisition, and uploads data via a 4G / 5G module.

[0113] The data analysis and intelligent recognition module is deployed on edge computing nodes and supports real-time data processing and local caching.

[0114] The visualization and early warning platform module: interfaces with the existing platform of the power grid operation and maintenance center, and supports the display of defect evolution trend curves;

[0115] The blockchain data security module records detection data and analysis results in real time and supports multi-institutional query verification.

[0116] The detailed process for data analysis and intelligent recognition is as follows:

[0117] Real-time candidate defect monitoring: The real-time monitoring cycle is set to 1 hour. When the characteristic parameters of the candidate defect (such as leakage magnetic field amplitude and metal cross-sectional area change rate) continue to increase within 3 consecutive cycles, high-frequency acquisition is triggered (sampling frequency is increased to 2kHz) to focus on tracking the development of defects.

[0118] Real-time update of multimodal fusion features: Multimodal fusion features are updated in real time using a sliding time window (window length is 1 hour). When the feature parameter change rate exceeds 10%, the consistency voting score is automatically recalculated to ensure real-time tracking of defect status.

[0119] Intelligent early warning threshold dynamic adjustment: Based on historical defect data and operation and maintenance experience, an early warning threshold dynamic adjustment model is established: for conductors and ground wires that have been operating stably for a long time, the early warning threshold is appropriately lowered; for conductors and ground wires that have developed minor defects, the early warning threshold is appropriately raised to avoid false early warnings and missed early warnings.

[0120] The implementation effect of this embodiment is as follows:

[0121] It enables 24 / 7 uninterrupted monitoring with data transmission latency ≤100ms;

[0122] Defect development trend tracking accuracy ≥90%, capable of providing early warning of serious defects 6-12 months in advance;

[0123] The system is compatible with the existing AD acquisition link of the AEMR-I flaw detector, with data compatibility ≥95%.

[0124] In summary, Examples 1, 2, and 3 cover three core scenarios: offline inspection, centralized management, and online monitoring, respectively. By refining each step of data analysis and intelligent identification (including parameter calibration, feature extraction, model calculation, and result correction), the feasibility of the technical solution is ensured. Practical application verification shows that the system achieves a defect identification accuracy of ≥98%, a location error of ≤0.1m, and a remaining life prediction error of ≤6 months. It effectively addresses the problems of low detection accuracy and insufficient intelligence in existing technologies, meeting the diverse operation and maintenance needs of the power grid.

[0125] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A digitally-enabled conductor and ground wire steel core detection system, characterized in that, The detection device module, the data acquisition and processing module, the data analysis and intelligent identification module, and the visualization and early warning platform module are included. The detection device module includes a multi-modal sensing unit for collecting sensing signals required for steel core damage detection, a displacement positioning assembly for obtaining positioning information corresponding to the sensing signals, and a clamping structure for clamping the outside of the cable. The data acquisition and processing module is used to obtain and preprocess the magnetic flux leakage signal, the ultrasonic guided wave signal, the optical fiber strain signal, and the positioning information to obtain a sensing data set. The data analysis and intelligent identification module is used to analyze the magnetic flux leakage signal to obtain broken strand or crack information using a threshold mechanism, and to obtain wear or corrosion information using metal cross-sectional area change analysis. The data analysis and intelligent identification module is used to extract the amplitude envelope, frequency energy, and group velocity time difference features of the ultrasonic guided wave signal, and to extract the strain peak-to-valley and local abnormal points of the optical fiber strain signal through multi-modal data fusion technology. The data analysis and intelligent identification module is used to classify and regress the multi-modal fusion features of the candidate defect points through a pre-trained deep learning network to determine the quantitative and positional steel core damage information and predict the remaining life. The visualization and early warning platform module is used to construct a ground wire steel core model through digital twinning technology to visualize, intelligently warn, and automatically generate detection reports for the quantitative and positional steel core damage information and the remaining life.

2. The digital empowerment based ground wire steel core detection system of claim 1, wherein, The multi-modal sensing unit of the detection device module includes a magnetizing device and a Hall array sensor, an ultrasonic guided wave transducer, and an optical fiber sensing array. The magnetizing device is used to magnetize the ground wire steel core axially to generate a stable magnetic flux leakage field, and the Hall array sensor is used to capture the magnetic flux leakage signal caused by steel core damage. The ultrasonic guided wave transducer transmits and receives ultrasonic guided wave signals to continuously detect steel core damage. The optical fiber sensing array is used to collect optical fiber strain signals to detect the strain distribution of the steel core.

3. The digital empowerment based ground wire steel core detection system of claim 1, wherein, The displacement positioning assembly includes a guide wheel and a displacement encoder that are linked together, and the displacement encoder generates positioning information corresponding to the sensing signals.

4. The digital empowerment based ground wire steel core detection system of claim 1, wherein, The data acquisition and processing module transmits the collected magnetic flux leakage signal, ultrasonic guided wave signal, and optical fiber strain signal to the central processing unit through an RS232 interface or a USB interface.

5. The digital empowerment based ground wire steel core detection system of claim 1, wherein, The pre-trained deep learning network is a convolutional neural network combined with a long short-term memory network model. The convolutional neural network extracts spatial features of the magnetic flux leakage signal, and the long short-term memory network captures time sequence features of the ultrasonic guided wave signal. The steel core damage information output by the pre-trained deep learning network includes defect types such as broken wires, corrosion and cracks, defect positions, equivalent broken wire numbers, crack lengths, corrosion volume fractions and confidence levels. The threshold mechanism for analyzing the magnetic flux leakage signal includes a first threshold for preliminary screening of broken wires and cracks, and a single broken wire signal amplitude of 85%; a second threshold determined in combination with the steel wire diameter and structure calibration for quantitative identification of defects. The metal cross-sectional area change analysis includes setting the sensitivity of the metal cross-sectional area change signal and the cross-sectional reference, with a threshold range of 0.5% to 1%, for quantitative identification of wear and corrosion.

6. The digital empowerment based ground wire steel core detection system of claim 1, wherein, The ground wire steel core model supports hierarchical scaling and positioning at the line level, tower level, span level, ground wire level and steel core level, and displays defect distribution and evolution trends through a heat map.

7. The digital empowerment based ground wire steel core detection system of claim 1, wherein, It also includes a blockchain data security module that records the detected steel core damage information, positioning information and timestamp data to the blockchain, and stores them through a smart contract to ensure data tamper resistance and traceability; the detection device module is modularly designed, can be quickly installed and removed without power interruption, is compatible with portable inspection mode and online long-term monitoring mode, and is adapted to the AD acquisition link and report system of the existing AEMR-I flaw detector.

8. The digital empowerment based ground wire steel core detection system of claim 2, wherein, The magnetization device uses a permanent magnet or an electromagnetic coil that can penetrate the outer aluminum layer or composite structure of the ground wire; the Hall array sensor is in a ring structure and is distributed around the periphery of the ground wire.

9. The digital empowerment based ground wire steel core detection system of claim 2, wherein, The ultrasonic guided wave transducer is a low-frequency longitudinal guided wave transducer that receives defect reflection or scattering signals and identifies steel core crack morphology and damage degree by combining time domain difference and spectral characteristics of ultrasonic signals; the crack morphology includes open cracks and closed cracks.

10. The digital empowerment based ground wire steel core detection system of claim 2, wherein, The optical fiber sensing array is an optical fiber Bragg grating sensing array that is evenly installed on the outside of the detection device module by a clamp.

Images