On-line identification method and system for broken strand damage of suspension clamp lead

Abstract

The invention relates to the technical field of on-line monitoring of high-voltage transmission lines, and discloses an on-line identification method and system for broken strand damage of a suspension clamp wire. Multi-mode ultrasonic guided waves are spread in the wire, and when the broken strand damage of the suspension clamp or the wire occurs, reflection is generated; the sensor array receives feedback damage reflection echo signals, extracts multi-dimensional features and inputs the features into the damage identification classifier; and the identification result of the broken strand damage position and degree and early warning information are transmitted to a monitoring center. According to the invention, comprehensive coverage of transverse cracks and axial fractures is realized through multi-mode collaborative excitation, and the problem of single-mode detection blind areas is solved. The technical problems that an existing single mode cannot cover various strand breakage conditions, the false alarm rate is ultrahigh, and the state of a wire at a suspension clamp cannot be accurately detected are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of online monitoring of high-voltage transmission lines, and in particular to a method and system for online identification of broken strand damage of a conductor of a suspension clamp. BACKGROUND

[0002] A suspension clamp is a key component of a high-voltage transmission line, and is mainly used to fix a conductor on an insulator string of a straight pole tower, and also bears the functions of suspending a lightning conductor, supporting a transposed conductor, and fixing a jumper, etc. It is directly related to the stable erection of the conductor and the safe transmission of power, and once the conductor clamped by the suspension clamp is broken, it is easy to cause a conductor breakage accident, which poses a serious threat to the safe operation of the entire power transmission system.

[0003] Therefore, the ultrasonic guided wave technology is usually used to detect the state of the conductor at the suspension clamp, which identifies the broken strand defect by analyzing the propagation characteristics of the guided wave. However, the above method has the problems of serious degradation of signal-to-noise ratio, reduction of ultrasonic coupling efficiency due to the oxidation layer on the surface of the conductor, superposition of wind noise, raindrop impact and other interferences, and the effective signal is often submerged, and the damage sensitivity is insufficient, a single mode cannot cover various broken strand conditions, and the false positive rate is too high, which cannot accurately detect the state of the conductor at the suspension clamp. SUMMARY

[0004] The present application provides a method and system for online identification of broken strand damage of a conductor of a suspension clamp, which solves the technical problem that the existing single mode cannot cover various broken strand conditions, the false positive rate is too high, and the state of the conductor at the suspension clamp cannot be accurately detected.

[0005] The present application provides a method and system for online identification of broken strand damage of a conductor of a suspension clamp, which solves the technical problem that the existing single mode cannot cover various broken strand conditions, the false positive rate is too high, and the state of the conductor at the suspension clamp cannot be accurately detected.

[0006] The pulse electrical signal of the multi-frequency pulse generator drives the sensor array to excite multi-modal ultrasonic guided waves;

[0007] The multi-modal ultrasonic guided waves are propagated on the conductor of the to-be-detected suspension clamp, and the damage reflection echo signal fed back by the to-be-detected suspension clamp and the conductor is received;

[0008] The multi-dimensional features of the damage reflection echo signal are extracted, and the multi-dimensional features are input into the damage identification classifier;

[0009] The damage identification classifier identifies the broken strand damage of the to-be-detected suspension clamp and the conductor, generates a damage identification result, and outputs the damage identification result and corresponding warning information.

[0010] Optionally, the pulse electrical signal passing through the multi-frequency pulse generator drives the sensor array to excite multi-modal ultrasonic guided waves, comprising:

[0011] Obtaining the conductor type corresponding to the to-be-tested suspension clamp, and determining the preset pulse parameters of the multi-frequency pulse generator;

[0012] Based on the preset pulse parameters, modulate the pulse electrical signal of the multi-frequency pulse generator to generate a wideband pulse electrical signal;

[0013] Amplify the wideband pulse electrical signal, and transmit the amplified wideband pulse electrical signal to the excitation unit of the sensor array;

[0014] Trigger the excitation unit to excite multi-modal ultrasonic guided waves.

[0015] Optionally, the multi-modal ultrasonic guided waves are propagated on the conductor of the to-be-tested suspension clamp, and the damage reflection echo signals fed back by the to-be-tested suspension clamp and the conductor, comprising:

[0016] The multi-modal ultrasonic guided waves are propagated on the conductor of the to-be-tested suspension clamp, and when the multi-modal ultrasonic guided waves pass through the to-be-tested suspension clamp or the area with broken strand damage during the propagation process, the reflection echo signal is generated, and the damage reflection echo signal is generated;

[0017] The damage reflection echo signal is received by the receiving unit of the sensor array.

[0018] Optionally, the multi-dimensional features of the damage reflection echo signal are extracted, and the multi-dimensional features are input into the damage identification classifier, comprising:

[0019] The time reversal noise reduction processing is performed on the damage reflection echo signal to generate an updated damage reflection echo signal;

[0020] The multi-modal separation processing is performed on the updated damage reflection echo signal to generate a first damage reflection echo signal and a second damage reflection echo signal;

[0021] The time domain feature, frequency domain feature and time-frequency domain feature of the first damage reflection echo signal and the second damage reflection echo signal are extracted respectively to generate multi-dimensional features;

[0022] The multi-dimensional features are input into the damage identification classifier.

[0023] Optionally, the multi-dimensional features are input into the damage identification classifier, comprising:

[0024] Each sensor in the sensor array is set as a node in a graph structure;

[0025] An adjacency matrix is ​​generated using the Euclidean distance between each node and the multi-dimensional features of the damage reflection echo signal corresponding to each node.

[0026] The spatial and feature information in the adjacency matrix are fused using multi-layer convolution to generate multi-layer fused features, and all layers of the fused features are input into the damage recognition classifier.

[0027] Optionally, the step of identifying strand breakage damage to the suspension clamp and the conductor using the damage identification classifier, generating a damage identification result, and outputting the damage identification result and corresponding warning information includes:

[0028] The damage recognition classifier embeds the fusion features corresponding to the last layer into a category probability distribution.

[0029] If the category probability distribution is greater than or equal to a preset confidence threshold, then it is determined that the node corresponding to the category probability distribution is damaged.

[0030] Spatial consistency detection is performed on the adjacent nodes of the node to identify multiple nodes with broken strands;

[0031] Set the cluster center of all nodes with broken strands as the damage core location;

[0032] Based on the guided wave time difference of arrival algorithm, the location of the damage core is located, and the location of the broken strand damage of the suspension clamp under test and the conductor is determined.

[0033] By combining the damage degree of the category probability distribution, the location of the broken strand damage of the suspension clamp under test and the conductor, a damage identification result is generated;

[0034] Using the damage identification results, the device model corresponding to the suspension clamp under test, the timestamp, and the damage parameters, an early warning information is generated.

[0035] The damage identification results and the early warning information are sent to the operation and maintenance center.

[0036] The second aspect of this invention provides an online identification system for conductor strand breakage damage in suspension clamps, applied to an online identification device. The device includes a multi-frequency pulse generator, a sensor array, and a damage identification classifier. The system includes:

[0037] The driving module is used to drive the sensor array to excite multimodal ultrasonic guided waves through the pulse electrical signal of the multi-frequency pulse generator;

[0038] The receiving module is used to propagate the multimodal ultrasonic guided wave onto the conductor on the suspension clamp under test, and to receive the damage reflection echo signal fed back by the suspension clamp under test and the conductor.

[0039] an extraction module configured to extract a multi-dimensional feature of the damage reflection echo signal and input the multi-dimensional feature into the damage identification classifier;

[0040] an identification module configured to identify the broken strand damage of the suspension clamp and the conductor wire by the damage identification classifier, generate a damage identification result, and output the damage identification result and corresponding early warning information.

[0041] The third aspect of the present application provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, causes the processor to perform the steps of the suspension clamp conductor wire broken strand damage online identification method according to any one of the preceding aspects.

[0042] The fourth aspect of the present application provides a computer readable storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the suspension clamp conductor wire broken strand damage online identification method according to any one of the preceding aspects.

[0043] The fifth aspect of the present application provides a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, wherein the computer program comprises program instructions, and when the program instructions are executed by a computer, the computer performs the suspension clamp conductor wire broken strand damage online identification method according to any one of the preceding aspects.

[0044] From the above technical solutions, the present application has the following advantages:

[0045] The present application generates a pulse electrical signal through a multi-frequency pulse generator to drive a sensor array arranged on both sides of the conductor wire of the suspension clamp to excite multi-modal ultrasonic guided waves. The multi-modal ultrasonic guided waves propagate in the conductor wire and produce reflection and scattering when encountering the suspension clamp or the conductor wire broken strand damage. The sensor array receives the feedback damage reflection echo signal and extracts a multi-dimensional feature, which is input into the damage identification classifier. Finally, the identification result of the broken strand damage position and degree is output, and the damage early warning module generates early warning information containing the device ID and the time stamp in combination with the result and transmits it to the monitoring center. The present application realizes comprehensive coverage of transverse cracks and axial fractures through multi-modal collaborative excitation, solves the problem of single modal detection blind area, and further meets the online real-time monitoring and provides accurate data support for operation and maintenance, completely solves the multiple defects of traditional technology in modal coverage and feature fusion, and realizes high-precision and high-reliability online identification and early warning of the suspension clamp conductor wire broken strand damage. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings described below are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0047] Figure 1 A step flow chart of a suspension clamp conductor broken strand damage online identification method provided for the first embodiment of the present application.

[0048] Figure 2 A sensor arrangement schematic diagram provided for the first embodiment of the present application.

[0049] Figure 3 A fusion schematic diagram of multi-layer convolution fusion provided for the first embodiment of the present application.

[0050] Figure 4 A structural block diagram of a suspension clamp conductor broken strand damage online identification system provided for the second embodiment of the present application.

[0051] Figure 5 A structural block diagram of a computer device provided for the third embodiment of the present application.

[0052] Among them, the meaning of the reference signs is as follows:

[0053] 1, sensor; 2, conductor; 3, dead weight; 301, to-be-measured suspension clamp; 302, anti-skid rubber pad. DETAILED DESCRIPTION

[0054] The embodiments of the present application provide a suspension clamp conductor broken strand damage online identification method and system, which are used to solve the technical problems that the existing single mode cannot cover various broken strand conditions, the false positive rate is too high, and the conductor state at the suspension clamp cannot be accurately detected.

[0055] In order to make the purposes, features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the embodiments described below are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0056] Please refer to Figures 1 to 3 , Figure 1 A step flow chart of a suspension clamp conductor broken strand damage online identification method provided for the first embodiment of the present application.

[0057] The application provides an online identification method for broken strand damage of a suspension clamp conductor, which is applied to an online identification device.

[0058] In step 101, a multi-mode ultrasonic guided wave is excited by driving the sensor array with a pulse electrical signal of the multi-frequency pulse generator.

[0059] In the embodiment of the application, the online identification device refers to a hardware system for realizing the function of online identification of broken strand damage of a suspension clamp conductor, and at least contains a multi-frequency pulse generator, a sensor array and a damage identification classifier, and can further integrate a signal processor, a damage early warning module and other components.

[0060] The multi-frequency pulse generator refers to a core excitation device in the online identification device, which is used for generating a pulse electrical signal with a wide frequency band of 0.1-2 MHz, and is electrically connected to the sensor array.

[0061] The sensor array, i.e., a ring-shaped piezoelectric sensor array, is composed of 4-8 piezoelectric ceramic sheets, and is physically attached to the conductor 2 at a position 5-10 cm away from the suspension clamp 301 on both sides of the suspension clamp 301 in a ring-shaped and angularly spaced 90°-120° manner through a coupling agent.

[0062] The damage identification classifier refers to an intelligent analysis core in the online identification device, and in the application, is specifically a classifier based on a graph convolutional network (GCN, Graph Convolutional Network), which is electrically connected to the signal processor and the damage early warning module.

[0063] The pulse electrical signal refers to an electrical signal generated by the multi-frequency pulse generator.

[0064] The multi-mode ultrasonic guided wave refers to an ultrasonic guided wave containing two or more vibration modes propagating in the conductor 2.

[0065] On the conductor 2 at a position 5-10 cm away from the suspension clamp 301 on both sides of the suspension clamp 301, the ring-shaped piezoelectric sensor array is physically attached and fixed by high-strength epoxy resin glue through a coupling agent to ensure efficient coupling of the ultrasonic guided wave; then, an operation and maintenance personnel presets parameters to the multi-frequency pulse generator through a system configuration interface, after the multi-frequency pulse generator receives a trigger instruction of an embedded processing unit, a wide frequency pulse electrical signal modulated by a Hanning window is generated by an internal chip, the signal is transmitted to a preset excitation unit in the ring-shaped piezoelectric sensor array through a coaxial cable, the excitation unit converts the pulse electrical signal into mechanical vibration based on the inverse piezoelectric effect, the vibration is transmitted to the conductor 2 and synchronously excites a multi-mode ultrasonic guided wave in the conductor 2.

[0066] Further, step 101 includes the following sub-steps:

[0067] S11, acquire the conductor model corresponding to the to-be-tested suspension clamp 301, and determine the preset pulse parameters of the multi-frequency pulse generator.

[0068] In the embodiment of the present application, the conductor model refers to a parameter for identifying the specifications and material characteristics of the conductor 2, usually including the structure, cross-sectional area, core material, and core cross-sectional area of the conductor 2.

[0069] The preset pulse parameters refer to parameters determined according to the conductor model corresponding to the to-be-tested suspension clamp 301, which are used to configure the output signal of the multi-frequency pulse generator.

[0070] The operation and maintenance personnel input the basic information (such as line number and tower number) of the power transmission line where the to-be-tested suspension clamp 301 is located through the man-machine interface of the online identification device or the remote monitoring platform, the system calls the built-in power transmission line equipment database, automatically matches and acquires the model (such as LGJ-400 / 35 steel core aluminum stranded conductor) of the conductor 2 clamped by the to-be-tested suspension clamp 301; then the system retrieves the ultrasonic guided wave dispersion data of the conductor 2 of this model according to the pre-stored correspondence relationship of “conductor model-guided wave dispersion curve”, analyzes and determines the optimal center frequency range that can synchronously and efficiently excite Lamb wave and Torsional wave, and further calculates and determines the preset pulse parameters of the multi-frequency pulse generator in combination with the diameter parameters and ultrasonic coupling characteristics of the conductor 2, including the center frequency, pulse cycle number, modulation window function type, and signal amplitude of the pulse signal, and finally transmits these preset pulse parameters to the multi-frequency pulse generator through the embedded processing unit to complete parameter configuration.

[0071] S12, modulate the pulse electrical signal of the multi-frequency pulse generator based on the preset pulse parameters to generate a wideband pulse electrical signal.

[0072] In the embodiment of the present application, the wideband pulse electrical signal refers to the pulse electrical signal after window function modulation and power amplification.

[0073] The preset pulse parameter configuration instruction is sent to the signal generation module of the multi-frequency pulse generator; the chip inside the multi-frequency pulse generator first generates a basic sinusoidal pulse signal according to the preset center frequency and pulse cycle number, then calls the built-in window function generation algorithm to generate the corresponding Hanning window time domain waveform according to the preset Hanning window modulation type, then performs point multiplication operation on the basic sinusoidal pulse signal and the Hanning window time domain waveform through the pulse modulation module to complete signal modulation, effectively compresses the signal frequency band and suppresses the frequency domain sidelobe interference, then transmits the modulated signal to the power amplification module, amplifies the signal to tens to one hundred volts according to the preset signal amplitude requirement, and finally generates a wideband pulse electrical signal that meets the multi-modal ultrasonic guided wave excitation requirement.

[0074] S13, amplifying the wideband pulse electrical signal, and transmitting the amplified wideband pulse electrical signal to an excitation unit of the sensor array.

[0075] In the embodiment of the present application, the amplification process refers to the process of lifting the amplitude of the wideband pulse electrical signal from millivolt level to tens to one hundred volts through the power amplifier module.

[0076] The excitation unit refers to a PZT (Lead Zirconate Titanate) piezoelectric ceramic sheet in the sensor array that is pre-designated to convert the amplified wideband pulse electrical signal into mechanical vibration, and realizes the conversion of electrical energy into mechanical energy through the inverse piezoelectric effect, which is the direct excitation source of multi-modal ultrasonic guided waves.

[0077] The wideband pulse electrical signal is transmitted to the power amplifier module integrated in the multi-frequency pulse generator. The power amplifier module lifts the amplitude of the wideband pulse electrical signal through a high-gain operational amplifier according to the signal amplitude threshold in the pre-set pulse parameters that matches the driving requirements of the sensor array, while the built-in filter circuit simultaneously filters out the high-frequency noise introduced during the amplification process, ensuring that the amplified wideband pulse electrical signal has stable amplitude and no additional interference. After the amplification process is completed, the multi-frequency pulse generator transmits the signal to the pre-designated excitation unit in the sensor array through a shielded coaxial cable.

[0078] S14, triggering the excitation unit to excite multi-modal ultrasonic guided waves.

[0079] In the embodiment of the present application, after the excitation unit of the sensor array receives the amplified wideband pulse electrical signal transmitted through the shielded coaxial cable, the PZT piezoelectric ceramic sheet inside it converts the electrical energy of the electrical signal into mechanical vibration of the same frequency based on the inverse piezoelectric effect. The mechanical vibration is efficiently transmitted to the inside of the corresponding conductor 2 of the measured suspension clamp 301 through the coupling interface between the high-strength epoxy resin glue and the surface of the conductor 2. Since the frequency of the wideband pulse electrical signal has been matched to the optimal excitation frequency band according to the conductor model, the mechanical vibration will simultaneously excite two multi-modal ultrasonic guided waves with complementary characteristics in the conductor 2, i.e. Lamb waves with elliptical particle vibration trajectories and sensitivity to transverse cracks, and Torsional waves with tangential particle vibration trajectories and sensitivity to axial fractures.

[0080] Step 102, propagating the multi-modal ultrasonic guided waves on the conductor 2 of the measured suspension clamp 301, and receiving the damage reflection echo signals fed back by the measured suspension clamp 301 and the conductor 2.

[0081] In the embodiment of the present application, the measured suspension clamp 301 refers to the core mechanical component for fixing and clamping the conductor 2, which is also a high-risk area of conductor 2 breakage damage, and is provided with an anti-skid rubber pad 302, as shown in Figure 2 ​

[0082] The wire 2, i.e. the power transmission wire 2, is an ultrasonic guided wave propagation medium, on which the plurality of sensors 1 of the sensor array, the excitation unit and the dead end anchor 3 are mounted, and the dead end anchor 3 clamps the wire 2 through the to-be-measured suspension clamp 301.

[0083] The damage reflection echo signal refers to a signal reflected when the multimodal ultrasonic guided wave encounters a broken strand damage of the wire 2 and a structure of the to-be-measured suspension clamp 301.

[0084] The multimodal ultrasonic guided wave (including Lamb wave and Torsional wave) stably propagates along the wire 2 corresponding to the to-be-measured suspension clamp 301 in the axial direction, and in the propagation process, the two modal guided waves maintain inherent vibration characteristics due to the uniformity of the material of the wire 2. When the guided wave reaches the to-be-measured suspension clamp 301 region or encounters a broken strand damage of the wire 2, reflection and scattering phenomena occur due to the change of medium impedance, and the broken strand damage causes part of the guided wave energy to be reflected to form a damage reflection echo signal, and the metal structure of the suspension clamp 301 and the anti-skid rubber pad 302 also generate a structural reflection signal. At this time, the pre-set receiving unit in the sensor array switches to the receiving mode, and through the positive piezoelectric effect, the guided wave vibration is converted into a weak electric signal. The electric signal is first preliminarily amplified by the preamplifier built-in the receiving unit, and then the environmental electromagnetic noise and high-frequency clutter are filtered out through the analog band-pass filter, and then transmitted to the signal acquisition module for quantization sampling, and finally the damage reflection echo signal containing the damage information is obtained.

[0085] Further, the step 102 comprises the following sub-steps:

[0086] S21, propagate the multimodal ultrasonic guided wave on the wire 2 on the to-be-measured suspension clamp 301, and when the multimodal ultrasonic guided wave passes through the to-be-measured suspension clamp 301 or the region with a broken strand damage in the propagation process, a reflection echo signal is generated, and a damage reflection echo signal is generated.

[0087] In the embodiment of the application, the broken strand damage refers to a steel wire fracture defect caused by fatigue, external force damage or fretting wear of the wire 2, including transverse cracks and axial fractures.

[0088] The damage reflection echo signal refers to a digital signal formed after the reflection echo signal is converted into an electric signal by the receiving unit of the sensor array, and then preamplified, band-pass filtered and quantization sampled.

[0089] The multi-modal ultrasonic guided waves (including Lamb waves and torsional waves) propagate along the conductor 2 axially held by the to-be-detected suspension clamp 301 uniformly, maintain stable vibration forms in the propagation process in reliance on the elastic properties of the metal material of the conductor 2, and the two modes of guided waves each maintain a natural propagation speed; when the multi-modal ultrasonic guided waves propagate to the region of the to-be-detected suspension clamp 301, due to the impedance difference between the metal clamping structure of the suspension clamp 301 and the material of the conductor 2, part of the guided wave energy will be reflected at the interface to form a structure reflection signal; if the conductor 2 has a transverse crack, an axial fracture or other broken strand damage, the cross-section integrity of the conductor 2 in the damage region is destroyed, causing a sudden change in the local medium impedance, and the multi-modal ultrasonic guided waves will be scattered and reflected at this position, wherein the reflected energy carrying the damage information forms a damage characteristic signal; the structure reflection signal and the damage characteristic signal will be naturally superimposed in the propagation process, and jointly constitute a damage reflection echo signal containing the structure information of the suspension clamp 301 and the damage information of the conductor 2.

[0090] S22, receiving the damage reflection echo signal by a receiving unit of the sensor array.

[0091] In the embodiment of the application, the receiving unit refers to a PZT piezoelectric ceramic piece group in the sensor array responsible for receiving the damage reflection echo signal.

[0092] When the damage reflection echo signal reversely propagates to the position of the sensor array along the conductor 2, the receiving unit in the sensor array that works in time division with the excitation unit immediately switches to the receiving mode, and the PZT piezoelectric ceramic piece inside the receiving unit converts the mechanical vibration of the echo signal into a weak electric signal through the positive piezoelectric effect; the weak electric signal is first preliminarily amplified by the preamplifier integrated in the receiving unit to avoid the signal being overwhelmed by noise in transmission, and then filtered to remove environmental electromagnetic interference and high-frequency clutter by the band-pass filter with a center frequency matching the multi-modal ultrasonic guided wave frequency band, and then transmitted to the signal acquisition module; the signal acquisition module quantitatively samples the filtered electric signal by using a high-speed ADC (Analog-to-Digital Converter), converts the analog signal into a digital signal, and embeds timestamp information at the same time, and finally generates the damage reflection echo signal containing complete damage information.

[0093] Step 103, extracting multi-dimensional features of the damage reflection echo signal, and inputting the multi-dimensional features into a damage recognition classifier.

[0094] In the embodiment of the application, the multi-dimensional features refer to a multi-domain feature set reflecting the damage characteristics extracted from the damage reflection echo signal, which includes time domain, frequency domain and time-frequency domain features.

[0095] The Lamb wave and the torsional wave corresponding to the signal components of the damage reflection echo signal are separated; multi-dimensional features are extracted for the two modal signals after separation; then the features of different modalities and different domains are fused to form a multi-dimensional feature vector, which is normalized and transmitted to the input layer of the damage identification classifier through a data interface.

[0096] Further, the step 103 comprises the following sub-steps:

[0097] S31, time reversal noise reduction processing is performed on the damage reflection echo signal to generate an updated damage reflection echo signal.

[0098] In the embodiment of the application, the time reversal noise reduction processing refers to a signal noise reduction method based on the reversibility of sound wave propagation, which constructs a time reversal operator by collecting environmental noise, performs phase conjugation processing on the damage reflection echo signal, so that the noise signals cancel each other due to randomness and the effective signals are enhanced due to deterministic path, thereby improving the signal-to-noise ratio.

[0099] The updated damage reflection echo signal refers to the signal obtained after time reversal noise reduction processing.

[0100] When the damage reflection echo signal is transmitted to the embedded processing unit, the device performs convolution operation on the signal and the time reversal operator, and through the phase conjugation principle, the noise signals cancel each other in the time reversal process, while the effective echo signals carrying damage information are phase enhanced due to the deterministic propagation path; after processing, the root mean square error method is used to evaluate the noise reduction effect, if the noise suppression rate does not reach the preset threshold, the time reversal operator is iteratively optimized and the above process is repeated until the updated damage reflection echo signal with signal-to-noise ratio improved to more than 20dB is generated.

[0101] It is worth mentioning that the calculation formula of the damage reflection echo signal is:

[0102]

[0103] In the formula, is the original damage reflection echo signal; is the effective echo signal without noise, is the noise.

[0104] The calculation formula of the time reversal processing is:

[0105]

[0106] In the formula, is the time reversal operator, is the effective echo is the signal after time reversal, is the noise time-reversed signal.

[0107] The time-reversed signal is then convolved with the original signal, i.e.

[0108]

[0109] wherein, is the updated damage reflection echo signal; is the original damage reflection echo signal; is the integral variable.

[0110] Substituting and into the expression,

[0111]

[0112] wherein, is the updated damage reflection echo signal; is the original damage reflection echo signal; is the integral variable; is the effective echo signal without noise; is the noise.

[0113] Since the effective echo signal has a deterministic propagation path (satisfying the conjugate symmetry property), while the noise is a random signal (satisfying i.e. the noise autocorrelation tends to zero), the above equation can be simplified as:

[0114]

[0115] wherein, is the updated damage reflection echo signal; is the integral variable; is the effective echo signal without noise; is the noise.

[0116] The output updated damage reflection echo signal is the autocorrelation result of the effective signal, and the noise component is greatly suppressed, with the signal-to-noise ratio improved to more than 20 dB, wherein the noise suppression rate is calculated by the following formula:

[0117]

[0118] wherein, represents the root mean square value of the signal, is the root mean square value of the noise in the original signal, is the root mean square value of the residual noise in the noise-reduced signal; is the noise suppression rate, and needs to be ≥40% to meet the subsequent feature extraction requirements.

[0119] S32, the updated damage reflection echo signal is processed by multi-modal separation to generate a first damage reflection echo signal and a second damage reflection echo signal.

[0120] In the embodiments of the present application, multi-modal separation processing refers to the process of decomposing the updated damage reflection echo signal into a single modal signal through a band-pass filter based on the characteristic frequency difference of different modal ultrasonic guided waves.

[0121] The first damage reflection echo signal refers to the damage reflection echo signal dominated by Lamb waves after multi-modal separation processing.

[0122] The second damage reflection echo signal refers to the damage reflection echo signal dominated by Torsional waves after multi-modal separation processing.

[0123] The updated damage reflection echo signal after time reversal noise reduction is processed by multi-modal separation, first based on the guided wave dispersion curve corresponding to the wire type to determine the characteristic frequency range of Lamb waves and Torsional waves, and then through two center frequency matching band-pass filters in the above range to separate the updated damage reflection echo signal. The signal with Lamb wave as the main component is separated out by the first band-pass filter, and the first damage reflection echo signal is generated after amplitude calibration, and the signal with Torsional wave as the main component is separated out by the second band-pass filter, and the second damage reflection echo signal is generated after amplitude calibration. The two signals respectively retain the damage reflection characteristics of the corresponding modal.

[0124] S33, the time domain features, frequency domain features and time-frequency domain features of the first damage reflection echo signal and the second damage reflection echo signal are extracted respectively to generate multi-dimensional features.

[0125] In the embodiments of the present application, the time domain feature refers to the feature extracted from the signal time domain waveform.

[0126] The frequency domain feature refers to the feature extracted after the signal is converted to the frequency domain by Fourier transform.

[0127] The time-frequency domain feature refers to the feature extracted by time-frequency analysis methods such as wavelet transform.

[0128] For the separated first damage reflection echo signal and second damage reflection echo signal, a feature extraction algorithm is called respectively to perform multi-domain feature extraction: in time domain, the peak amplitude, rising edge slope and time difference of arrival of the two signals are calculated; in frequency domain, the power spectrum of the two signals is obtained through fast Fourier transform to extract the main peak frequency offset and 3dB bandwidth; in time-frequency domain, Morlet wavelet transform is used for time-frequency analysis of the two signals to extract the wavelet energy entropy and instantaneous frequency variance corresponding to the 1MHz center frequency. Then, the three time domain features, two frequency domain features and two time-frequency domain features of the first signal are integrated with the corresponding features of the second signal, and after normalization, a multi-dimensional feature containing 14 dimensions is generated.

[0129] S34, input the multi-dimensional feature into the damage identification classifier.

[0130] In the embodiment of the application, the multi-dimensional feature containing 14 dimensions is input into the input layer of the damage identification classifier.

[0131] Further, step S34 includes the following sub-steps:

[0132] S341, set each sensor 1 in the sensor array as a node in the graph structure.

[0133] In the embodiment of the application, the graph structure refers to a data structure for representing the association relationship between elements, which is composed of nodes and edges.

[0134] The node refers to the basic element in the graph structure, which corresponds to each sensor 1 in the sensor array one by one in the application, and each node n contains the damage reflection echo signal feature vector v collected by the corresponding sensor 1.

[0135] Referring to Figure 3 , first, the physical layout information of the sensor array is obtained, for example, 4-8 sensors 1 are distributed in a ring on the surface of the wire 2 at an angle interval of 90°-120°, each sensor 1 is defined as an independent node in the graph structure, and the node number corresponds to the actual position of the sensor 1 in the array one by one.

[0136] It is worth mentioning that the sensors 1 are numbered from 1 to N in clockwise direction, and N is the number of sensors 1.

[0137] S342, generate an adjacency matrix using the Euclidean distance of each node and the multi-dimensional feature of the damage reflection echo signal corresponding to each node.

[0138] In the embodiment of the application, the Euclidean distance refers to an index for measuring the spatial position difference of two nodes (sensors 1), which is calculated by the square root of the sum of the differences of the three-dimensional coordinates of the nodes, reflecting the actual distance of the sensors 1 in the physical space.

[0139] The adjacency matrix refers to an N*N matrix (N is the number of nodes) representing the association between nodes in the graph structure, and the matrix element value represents the association weight between the two nodes. The greater the weight, the closer the spatial distance between the nodes or the more similar the features.

[0140] First, the Euclidean distance of each node in the sensor array is calculated, and the multi-dimensional features of the damage reflection echo signal corresponding to each node are extracted, including time domain, frequency domain and time-frequency domain features. Then, the reciprocal of the Euclidean distance between nodes is used as the basic weight, and the cosine similarity of the multi-dimensional features of the two nodes is combined to modify the basic weight, and the association weight value between the nodes is obtained. Finally, the association weight values between all nodes are arranged in order according to the node number to generate an N*N adjacency matrix (N is the total number of nodes), and the matrix element represents the association weight value between node i and node j. , which represents the node itself without association.

[0141] It is worth mentioning that based on the coordinate position of the sensor 1 ring distribution, the Euclidean distance between node i and node j is calculated by the square root of the sum of the squares of the three-dimensional coordinate differences, and the formula is as follows:

[0142] .

[0143] In the formula, is the Euclidean distance between node i and node j, are the x-axis coordinates of node i and node j respectively, are the y-axis coordinates of node i and node j respectively, are the z-axis coordinates of node i and node j respectively.

[0144] S343, the spatial information and feature information in the adjacency matrix are fused by multi-layer convolution to generate multi-layer fusion features and input all layer fusion features into the damage identification classifier.

[0145] In the embodiment of the application, the spatial information refers to the association between the nodes in the adjacency matrix, such as the ring distribution distance of the sensor 1, the angle interval, etc.

[0146] The feature information refers to the multi-dimensional features of the damage reflection echo signal corresponding to each node, such as time domain, frequency domain and time-frequency domain features.

[0147] Multi-layer convolution fusion refers to the process of gradually aggregating the node and its neighborhood features in the graph convolution network through multi-layer convolution operation.

[0148] Multi-layer fusion features refer to the features output by each layer after multi-layer convolution operation, from the local detailed features of the initial layer to the global abstract features of the high layer, integrating more extensive spatial correlation and more profound damage patterns layer by layer, and comprehensively covering different dimensions of damage features.

[0149] The graph convolution network (GCN) of the damage identification classifier first performs multi-graph layer convolution fusion on the input node multi-dimensional features based on the constructed adjacency matrix: the first layer of convolution extracts the local spatial correlation features of each node and its directly adjacent nodes (such as the feature interaction of the nearest 2-3 sensors 1) through the adjacency matrix, and performs element-level weighted fusion with the multi-dimensional features of the node itself to generate initial fusion features containing local spatial information; the second layer of convolution generates fusion features containing global levels based on the initial fusion features output by the first layer, combines larger range of neighborhood node features (such as the nearest 4-5 sensors 1), and enhances the feature nonlinear expression ability through the activation function.

[0150] Step 104, the damage identification classifier is used to identify the broken strand damage of the to-be-tested suspension clamp 301 and the conductor 2, generate a damage identification result, and output the damage identification result and corresponding warning information.

[0151] In the embodiment of the application, the broken strand damage identification refers to the process of determining whether the conductor 2 has a steel wire breakage and the specific attributes (type, degree, position) of the breakage by analyzing the reflection characteristics of the multi-modal ultrasonic guided wave.

[0152] The damage identification result refers to the output result of the damage identification classifier, which includes four core contents of damage existence (yes / no), type (transverse crack / axial breakage), degree (light / medium / heavy), and position (coordinate information).

[0153] The warning information refers to the graded warning information generated based on the damage identification result.

[0154] After receiving the multi-layer fusion features, the damage identification classifier classifies and maps the feature vectors, first determines whether the to-be-tested suspension clamp 301 and the conductor 2 have a broken strand damage; if it is determined that there is a damage, the damage type and the damage degree are further subdivided, and the damage position coordinates are calculated by combining the arrival time of the damage reflection echo signal and the guided wave propagation speed, and the damage identification result containing the damage existence, the type, the degree, and the position is integrated; then, the system matches the preset warning level according to the damage degree, generates the warning information containing the damage details and the maintenance suggestion, and finally synchronously pushes the damage identification result and the warning information to the remote monitoring platform through the wired transmission module.

[0155] Further, step 104 includes the following sub-steps:

[0156] S41, embedding mapping the fusion feature corresponding to the last layer to a class probability distribution by the damage identification classifier.

[0157] In the embodiment of the application, the fusion feature corresponding to the last layer refers to the final feature vector output by the graph convolution network after multi-layer convolution fusion.

[0158] Embedding mapping refers to the process of converting high-dimensional fusion features into a low-dimensional vector matching the number of damage categories through a fully connected layer.

[0159] The class probability distribution refers to a probability vector obtained after processing by an activation function, each element corresponding to the occurrence probability of a predetermined damage category.

[0160] After the damage identification classifier obtains the fusion feature output by the last layer of the graph convolution network, it first maps the dimension of the fusion feature to a low-dimensional space matching the number of predetermined damage categories through a fully connected layer, for example, when there are 8 categories including no damage, transverse mild, axial severe, etc., it is mapped to an 8-dimensional vector; then an activation function is called to normalize the low-dimensional vector, converting each element in the vector to a probability value between 0 and 1, and the sum of all element probabilities is 1, finally generating a class probability distribution containing the occurrence probability of each category.

[0161] S42, when the class probability distribution is greater than or equal to the preset confidence threshold, it is determined that the node corresponding to the class probability distribution has damage.

[0162] In the embodiment of the application, the preset confidence threshold refers to a probability threshold for determining whether damage exists, which is set to 0.7 in the application.

[0163] Damage refers to the broken strand defect existing in the area of the conductor 2 or the suspension clamp 301, including transverse cracks (Lamb wave sensitive) and axial fractures (Torsional wave sensitive), which are divided into three levels of light, medium and heavy according to the severity.

[0164] After the damage identification classifier generates the class probability distribution, the category with the highest probability value in the distribution and its corresponding probability value are extracted, which are compared with the preset confidence threshold 0.7; if the highest probability value is greater than or equal to 0.7, it is determined that the node corresponding to the class probability distribution has damage, and the damage type and degree corresponding to the category are recorded; if the highest probability value is less than 0.7, the secondary detection mechanism is triggered, the damage reflection echo signal is reacquired, and the feature extraction and classification process is repeated, if the secondary detection result still does not reach the threshold, it is determined as suspected damage and marked for manual review, in this way, the low confidence results are filtered through the preset threshold, reducing false positives.

[0165] S43, spatial consistency detection is performed on the adjacent nodes of the node to determine multiple nodes with broken strand damage.

[0166] In the embodiment of the present application, the adjacent node refers to the node determined based on the adjacency matrix and closely associated with the target node in space.

[0167] The spatial consistency detection refers to the detection process of verifying whether the damage identification result of the adjacent node matches the determined damage node feature.

[0168] When it is determined that a certain node has a broken strand damage, the list of adjacent nodes of the node in the graph structure is called, and the damage identification result of these adjacent nodes is extracted. The spatial consistency detection is performed, that is, whether the damage type, degree and position of the adjacent node match the spatial correlation of the determined damage node is analyzed. If the damage determination result of a certain node in the adjacent node matches the damage feature of the determined damage node by ≥80%, it is determined that the adjacent node also has a broken strand damage. This process is repeated to verify all adjacent nodes one by one, and finally all nodes that meet the spatial consistency condition are integrated to determine multiple nodes that have a broken strand damage, and the spatial distribution relationship of each damage node is marked.

[0169] S44, the center of the node cluster with a broken strand damage is set as the damage core position.

[0170] In the embodiment of the present application, the node cluster refers to the set of damage nodes with adjacent spatial positions classified by the spatial clustering algorithm.

[0171] The damage core position refers to the broken strand damage key area coordinate determined based on the node cluster center, which is the optimal estimate of the actual damage position and can be simplified as a single point coordinate for intuitive positioning of the damage position.

[0172] The spatial clustering is performed on all nodes that have been determined to have a broken strand damage. The physical coordinates of each node are calculated by the spatial clustering algorithm, the damage nodes with a spatial distance less than a preset clustering threshold are divided into the same cluster, and the weighted average method is used to calculate the geometric center of each cluster. The center coordinate is set as the damage core position. If there are multiple discrete clusters, the core positions of the clusters are determined respectively and sorted according to the damage degree, and finally the result containing the core position coordinate and the number of damage nodes in the corresponding cluster is output.

[0173] S45, based on the guided wave time difference of arrival algorithm, the damage core position is positioned to determine the broken strand damage position of the to-be-tested suspension clamp 301 and the conductor 2.

[0174] In the embodiment of the present application, the guided wave time difference of arrival algorithm refers to the algorithm for calculating the damage position based on the time difference of receiving the damage reflection echo signal by different sensors 1.

[0175] The broken strand damage position refers to a damage key area coordinate determined by a node cluster center where the broken strand damage exists, is a positioning starting point of a time difference of arrival algorithm, and represents a central reference point of the damage concentrated distribution.

[0176] The time of arrival of the damage reflection echo signal received by each sensor 1 in the node cluster corresponding to the damage core position is called, that is, the time when the signal arrives at each sensor 1; in combination with a preset multimodal ultrasonic guided wave propagation speed, the distance from the damage core position to each sensor 1 is calculated through a guided wave time difference of arrival algorithm (distance = propagation speed x time of arrival); then, based on the physical coordinates of each sensor 1, a spatial positioning equation set is constructed based on a multilateration principle, and the three-dimensional coordinates of the damage core position are solved, the coordinates are mapped to the physical structure model of the conductor 2 and the suspension clamp 301, and finally the specific position of the broken strand damage on the to-be-measured suspension clamp 301 and the conductor 2 is determined.

[0177] S46, in combination with the damage degree, the broken strand damage position of the to-be-measured suspension clamp 301 and the conductor 2, a damage identification result is generated.

[0178] In the embodiment of the application, the damage degree refers to the damage severity classification with the highest probability extracted from the category probability distribution, which is classified into light, medium and heavy levels.

[0179] The damage identification result refers to a structured output after integrating the damage type, degree, position and range characteristics.

[0180] The damage degree (light / medium / heavy) with the highest probability in the category probability distribution is extracted, in combination with the damage type (transverse crack / axial fracture) corresponding to the category, and then the broken strand damage position (such as 25 mm left of the suspension clamp 301 and 60° circumferentially of the conductor 2) determined through the time difference of arrival algorithm is associated, and these information is integrated as structured data; subsequently, the number of cluster nodes of the damage core position and the spatial distribution characteristics are supplemented, and a complete damage identification result containing the damage type, degree, accurate position and range characteristics is formed.

[0181] S47, using the damage identification result, the device model corresponding to the to-be-measured suspension clamp 301, the time stamp and the damage parameter, a warning information is generated.

[0182] In the embodiment of the application, the device model refers to the specification identification of the suspension clamp 301 and the conductor 2.

[0183] The time stamp refers to the accurate time of recording damage detection, which contains date and time, is used for tracing the damage occurrence period and evaluating the development speed, and is a time reference for time series analysis and operation scheduling.

[0184] Damage parameters refer to indicators that quantitatively describe the damage state (such as the number of broken strands, crack depth, frequency offset), and supplement the specific physical quantities of the damage degree.

[0185] The damage identification result is called, the device model of the to-be-tested suspension clamp 301 and the conductor 2 is matched, for example, the to-be-tested suspension clamp 301 is an XGU-5 type, to obtain the structural parameters, and the timestamp of the current detection is recorded, which needs to be accurate to seconds and is in the format of YYYY-MM-DD HH:MM:SS, and damage parameters such as the number of broken strands, crack depth, and characteristic frequency offset are extracted; then, the warning level is matched according to the damage degree, that is, a yellow warning corresponds to a slight damage, an orange warning corresponds to a moderate damage, and a red warning corresponds to a severe damage, the operation and maintenance standard corresponding to the device model is combined, for example, the XGU-5 type suspension clamp 301 needs to be overhauled within 72 hours for moderate damage in the axial direction, the timestamp, damage position, parameters, and processing suggestions are integrated, and warning information including the warning level + device information + damage details + time + processing scheme is generated, for example, red warning: there is a severe axial broken strand (6 broken wires) of the conductor 2 at the right side 30 mm of the XGU-5 type suspension clamp 301, the detection time is 2025-10-27 10:15:30, and it is suggested to overhaul immediately.

[0186] S48, the damage identification result and the warning information are sent to the operation and maintenance center.

[0187] In the embodiment of the application, the operation and maintenance center refers to a core institution responsible for the operation and maintenance management of high-voltage transmission line equipment, has the ability to receive, store, and analyze damage and warning information, can generate operation and maintenance work orders and dispatch personnel to carry out maintenance.

[0188] The damage identification result (including damage type, degree, position, and range parameters) and the warning information (including warning level, processing suggestion, and timestamp) are integrated into a standardized data packet, the data packet is encrypted according to a preset encryption protocol through a 4G / 5G wireless communication module or an optical fiber wired transmission channel built in the device, to avoid information leakage or tampering in the transmission process; then, the encrypted packet is sent to a designated receiving server of the operation and maintenance center, the server automatically completes decryption and analysis after receiving the packet, and the damage identification result and the warning information are respectively recorded in a “damage database” and a “warning work order library” of the operation and maintenance management system, and a system pop-up window and a short message notification are triggered to remind the on-duty personnel of the operation and maintenance center to check in time; if it is a red warning (severe damage), the system will also automatically generate an operation and maintenance work order with the highest priority, associate the corresponding device model and historical maintenance records, and synchronously push to the operation and maintenance team terminal responsible for the line, to ensure that the operation and maintenance center can quickly obtain complete information and start the response process.

[0189] Please refer to Figure 4 , Figure 4A structural block diagram of a suspension clamp conductor broken strand damage online identification system provided for embodiment two of the application.

[0190] The application provides a suspension clamp conductor broken strand damage online identification system, which is applied to an online identification device.

[0191] The driving module 201 is used for driving the sensor array to excite multi-modal ultrasonic guided waves through the pulse electrical signal of the multi-frequency pulse generator.

[0192] The receiving module 202 is used for propagating the multi-modal ultrasonic guided waves on the conductor 2 on the to-be-measured suspension clamp 301 and receiving the damage reflection echo signal fed back by the to-be-measured suspension clamp 301 and the conductor 2.

[0193] The extraction module 203 is used for extracting the multi-dimensional features of the damage reflection echo signal and inputting the multi-dimensional features into the damage identification classifier.

[0194] The identification module 204 is used for identifying the broken strand damage of the to-be-measured suspension clamp 301 and the conductor 2 through the damage identification classifier, generating a damage identification result, and outputting the damage identification result and corresponding early warning information.

[0195] Further, the driving module 201 comprises:

[0196] The acquisition sub-module is used for acquiring the conductor model corresponding to the to-be-measured suspension clamp 301 and determining the preset pulse parameters of the multi-frequency pulse generator.

[0197] The modulation sub-module is used for modulating the pulse electrical signal of the multi-frequency pulse generator based on the preset pulse parameters to generate a wideband pulse electrical signal.

[0198] The transmission sub-module is used for amplifying the wideband pulse electrical signal and transmitting the amplified wideband pulse electrical signal to the excitation unit of the sensor array.

[0199] The triggering sub-module is used for triggering the excitation unit to excite the multi-modal ultrasonic guided waves.

[0200] Further, the receiving module 202 comprises:

[0201] The propagation sub-module is used for propagating the multi-modal ultrasonic guided waves on the conductor 2 on the to-be-measured suspension clamp 301.

[0202] The receiving sub-module is used for receiving the damage reflection echo signal through the receiving unit of the sensor array.

[0203] Further, the extraction module 203 comprises:

[0204] a processing submodule, configured to perform time reversal noise reduction processing on the damage reflection echo signal to generate an updated damage reflection echo signal;

[0205] a separation processing submodule, configured to perform multi-modal separation processing on the updated damage reflection echo signal to generate a first damage reflection echo signal and a second damage reflection echo signal;

[0206] an extraction submodule, configured to extract time domain features, frequency domain features and time-frequency domain features of the first damage reflection echo signal and the second damage reflection echo signal respectively to generate multi-dimensional features;

[0207] an input submodule, configured to input the multi-dimensional features into a damage identification classifier.

[0208] Further, the input submodule comprises:

[0209] a node submodule, configured to set each sensor 1 in the sensor array as a node in a graph structure;

[0210] an adjacency matrix submodule, configured to generate an adjacency matrix by using the Euclidean distance of each node and the multi-dimensional features of the damage reflection echo signal corresponding to each node;

[0211] a fusion submodule, configured to perform multi-graph layer convolution fusion on the spatial information and feature information in the adjacency matrix to generate multi-layer fused features and input all the layer fused features into the damage identification classifier.

[0212] Further, the identification module 204 comprises:

[0213] an embedding mapping submodule, configured to embed and map the fused features corresponding to the last layer into a class probability distribution by the damage identification classifier;

[0214] a damage submodule, configured to determine that the node corresponding to the class probability distribution has damage when the class probability distribution is greater than or equal to a preset confidence threshold;

[0215] a detection submodule, configured to perform spatial consistency detection on adjacent nodes of the node to determine a plurality of nodes having strand damage;

[0216] a damage core submodule, configured to set a cluster center of all the nodes having strand damage as a damage core position;

[0217] a positioning submodule, configured to position the damage core position based on a guided wave time difference of arrival algorithm to determine a strand damage position of the to-be-measured suspension clamp 301 and the conductor 2;

[0218] The combining submodule is used for combining the damage degree, the broken strand damage position of the to-be-tested suspension clamp 301 and the conductor 2, and generating a damage identification result;

[0219] The early warning submodule is used for generating early warning information by using the damage identification result, the device model corresponding to the to-be-tested suspension clamp 301, the timestamp and the damage parameter.

[0220] The sending submodule is used for sending the damage identification result and the early warning information to the operation and maintenance center.

[0221] Please refer to Figure 5 , Figure 5 A structural block diagram of a computer device provided for the third embodiment of the present application.

[0222] The electronic device of the embodiment of the present application comprises a memory 401 and a processor 402, the memory 401 stores a computer program; when the computer program is executed by the processor 402, the processor 402 executes the suspension clamp conductor broken strand damage online identification method of any one of the above embodiments.

[0223] The memory 401 can be an electronic memory such as a flash memory, an EEPROM (Electrically Erasable Programmable Read-Only Memory), an EPROM, a hard disk or a ROM. The memory 401 has a storage space 403 for program codes 313 for executing any method step in the above method. For example, the storage space 403 for program codes can include various program codes 413 for respectively implementing various steps in the above method. These program codes can be read from or written into one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks. The program codes can be compressed in an appropriate form, for example. These codes, when executed by a computing processing device, cause the computing processing device to perform various steps in the above method. These program codes can be read from or written into one or more computer program products. These computer program products include program code carriers such as hard disks, compact disks (CDs), memory cards or floppy disks. The program codes can be compressed in an appropriate form, for example. These codes, when executed by a computing processing device, cause the computing processing device to perform various steps in the above suspension clamp conductor broken strand damage online identification method.

[0224] The fourth embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed to realize the suspension clamp conductor broken strand damage online identification method of any one of the embodiments of the present application.

[0225] The embodiment five of the present application provides a computer program product, the computer program product comprises a computer program stored on a non-transitory computer readable storage medium, the computer program comprises program instructions, wherein when the program instructions are executed by a computer, the computer executes the overhead line conductor broken strand damage online identification method of any one of the above-mentioned embodiments.

[0226] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-mentioned system, device and unit can refer to the corresponding process in the foregoing method embodiment, and will not be repeated here.

[0227] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the above-mentioned device embodiment is only schematic, for example, the division of the unit is only a logical function division, and actual implementation can have another division mode, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0228] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or they can be distributed on a plurality of network units. According to actual needs, part or all of the units can be selected to achieve the purpose of the embodiment scheme.

[0229] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically independently, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of software functional unit.

[0230] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0231] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features; and these modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for online identification of conductor strand breakage damage in suspension clamps, characterized in that, The method is applied to an online identification device, and the device comprises a multi-frequency pulse generator, a sensor array and a damage identification classifier; the method comprises the following steps: driving the sensor array by a pulse electric signal of the multi-frequency pulse generator to excite multi-modal ultrasonic guided waves; propagating the multi-modal ultrasonic guided waves on a conductor of a to-be-tested suspension clamp, and receiving damage reflection echo signals fed back by the to-be-tested suspension clamp and the conductor; extracting multi-dimensional features of the damage reflection echo signals, and inputting the multi-dimensional features into the damage identification classifier; identifying broken strand damage of the to-be-tested suspension clamp and the conductor by the damage identification classifier, generating damage identification results, and outputting the damage identification results and corresponding warning information.

2. The method of claim 1, wherein the method further comprises: The step of driving the sensor array by a pulse electric signal of the multi-frequency pulse generator to excite multi-modal ultrasonic guided waves comprises the following steps: acquiring a conductor model corresponding to the to-be-tested suspension clamp, and determining preset pulse parameters of the multi-frequency pulse generator; modulating the pulse electric signal of the multi-frequency pulse generator based on the preset pulse parameters to generate a wideband pulse electric signal; amplifying the wideband pulse electric signal, and transmitting the amplified wideband pulse electric signal to an excitation unit of the sensor array; triggering the excitation unit to excite multi-modal ultrasonic guided waves.

3. The method of claim 1, wherein the method further comprises: The step of propagating the multi-modal ultrasonic guided waves on a conductor of a to-be-tested suspension clamp, and receiving damage reflection echo signals fed back by the to-be-tested suspension clamp and the conductor comprises the following steps: propagating the multi-modal ultrasonic guided waves on the conductor of the to-be-tested suspension clamp, and when the multi-modal ultrasonic guided waves pass through the to-be-tested suspension clamp or a region with broken strand damage in a propagation process, reflection echo signals are generated to generate damage reflection echo signals; receiving the damage reflection echo signals by a receiving unit of the sensor array.

4. The method of claim 1, wherein the method further comprises: The step of extracting multi-dimensional features of the damage reflection echo signals, and inputting the multi-dimensional features into the damage identification classifier comprises the following steps: performing time reversal noise reduction processing on the damage reflection echo signals to generate updated damage reflection echo signals; performing multi-modal separation processing on the updated damage reflection echo signals to generate a first damage reflection echo signal and a second damage reflection echo signal; extracting time domain features, frequency domain features and time-frequency domain features of the first damage reflection echo signal and the second damage reflection echo signal respectively to generate multi-dimensional features; inputting the multi-dimensional features into the damage identification classifier.

5. The method of claim 4, wherein the method further comprises: The step of inputting the multi-dimensional features into the damage identification classifier comprises the following steps: setting each sensor in the sensor array as a node in a graph structure; generating an adjacency matrix by using the Euclidean distance of each node and the multi-dimensional features of the damage reflection echo signals corresponding to each node; performing multi-graph layer convolution fusion on spatial information and feature information in the adjacency matrix to generate multi-layer fusion features, and inputting all layers of the fusion features into the damage identification classifier.

6. The method of claim 5, wherein the method further comprises: The damage identification classifier is used to identify the broken strand damage of the to-be-tested suspension clamp and the conductor, generate a damage identification result, and output the damage identification result and corresponding early warning information, including: The fusion features corresponding to the last layer are mapped into a category probability distribution by the damage identification classifier; When the category probability distribution is greater than or equal to a preset confidence threshold, it is determined that the node corresponding to the category probability distribution has damage; The spatial consistency of adjacent nodes of the node is detected to determine multiple nodes having broken strand damage; The cluster center of all nodes having broken strand damage is set as a damage core position; The damage core position is located based on a guided wave time difference of arrival algorithm to determine the broken strand damage position of the to-be-tested suspension clamp and the conductor; The damage identification result is generated in combination with the damage degree of the category probability distribution and the broken strand damage position of the to-be-tested suspension clamp and the conductor; The damage identification result, the device model corresponding to the to-be-tested suspension clamp, the timestamp, and the damage parameter are used to generate early warning information; The damage identification result and the early warning information are sent to an operation and maintenance center.

7. An overhead conductor line suspension clamp broken conductor damage on-line identification system characterized by, The application is applied to an online identification device, and the device includes a multi-frequency pulse generator, a sensor array, and a damage identification classifier. The system includes: A driving module is configured to drive the sensor array to excite multi-modal ultrasonic guided waves through pulse electrical signals of the multi-frequency pulse generator. A receiving module is configured to propagate the multi-modal ultrasonic guided waves on a conductor of a to-be-tested suspension clamp and receive damage reflection echo signals fed back by the to-be-tested suspension clamp and the conductor. An extraction module is configured to extract multi-dimensional features of the damage reflection echo signals and input the multi-dimensional features into the damage identification classifier. An identification module is configured to identify the broken strand damage of the to-be-tested suspension clamp and the conductor through the damage identification classifier, generate a damage identification result, and output the damage identification result and corresponding early warning information.

8. An electronic device, comprising: The computer program is executed to implement the online identification method of the broken strand damage of the suspension clamp conductor according to any one of claims 1-6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed to implement the online identification method of the broken strand damage of the suspension clamp conductor according to any one of claims 1-6.

10. A computer program product, characterised in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer executes the online identification method of the broken strand damage of the suspension clamp conductor according to any one of claims 1-6.

Images