Ground wire damage detection method and system based on magnetostrictive guided waves

Abstract

The invention relates to the technical field of ground wire damage detection, and provides a ground wire damage detection method and system based on magnetostrictive guided waves, and the method comprises the following steps: generating a white noise signal through a magnetostrictive transducer to excite a ground wire, and collecting a guided wave response signal of the guided waves; identifying the guided wave response signal through a spectrum analysis model to obtain the inherent frequency characteristic of the ground wire, and obtaining the optimal excitation frequency according to the inherent frequency characteristic and the amplitude frequency mapping relation; adjusting the excitation frequency of the white noise signal according to the optimal excitation frequency to obtain an optimized guided wave signal; identifying the optimized guided wave signal through a defect detection and tension monitoring model to obtain a defect tension detection result; and based on the defect tension detection result, judging the damage level of the ground wire, and generating a detection report. According to the invention, defect detection and tension abnormity monitoring are carried out at the same time, the reliability of ground wire detection is improved, and an effective monitoring means is provided for safe operation of the ground wire.

Description

Technical Field

[0001] The present invention relates to the technical field of conductor and ground wire damage detection, and particularly to a method and system for detecting conductor and ground wire damage based on magnetostrictive guided waves. Background Art

[0002] As an important part of a power transmission line, the safe and reliable operation of a conductor and ground wire is directly related to the stability of the power system and the power supply safety. During long-term operation, the conductor and ground wire are easily affected by factors such as mechanical stress, environmental corrosion, and external force damage, resulting in defects such as broken strands, cracks, and corrosion. At the same time, there may also be problems such as abnormal tension. If these damages are not detected and processed in time, serious accidents such as the breakage of the conductor and ground wire may occur, threatening the safe operation of the power system.

[0003] Existing technologies usually use a single detection method to detect the conductor and ground wire. For example, ultrasonic detection methods are used for defect detection, or stress sensors are used for tension monitoring. Although this single detection method can obtain data information in a certain aspect, it cannot simultaneously achieve defect detection and tension monitoring, and different detection devices and methods need to be used separately, increasing the detection cost and time, resulting in insufficient detection accuracy and reliability. Especially in a complex environment, it is easily interfered with, leading to inaccurate detection results. Summary of the Invention

[0004] In view of this, the present invention proposes a method and system for detecting conductor and ground wire damage based on magnetostrictive guided waves, which solves the problem that existing technologies cannot simultaneously achieve defect detection and tension monitoring when performing damage detection and need to use different detection devices and methods separately.

[0005] The technical solution of the present invention is realized as follows: In the first aspect, the present invention provides a method for detecting conductor and ground wire damage based on magnetostrictive guided waves, which is characterized by including the following steps:

[0006] Generate a white noise signal through a magnetostrictive transducer to excite the conductor and ground wire, so that guided waves propagate in the conductor and ground wire, and collect the guided wave response signal of the guided waves;

[0007] Construct a spectrum analysis model based on a convolutional neural network, identify the guided wave response signal through the spectrum analysis model to obtain the natural frequency characteristics of the conductor and ground wire, and obtain the optimal excitation frequency according to the natural frequency characteristics and the amplitude-frequency mapping relationship;

[0008] Adjust the excitation frequency of the white noise signal according to the optimal excitation frequency to obtain an optimized guided wave signal;

[0009] Construct a defect detection and tensile force monitoring model based on a recurrent neural network, and identify the optimized guided wave signal through the defect detection and tensile force monitoring model to obtain a defect tensile force detection result;

[0010] Based on the defect tensile force detection result, judge the damage level of the ground wire and generate a detection report.

[0011] On the basis of the above technical solution, preferably, generating a white noise signal by a magnetostrictive transducer to excite the ground wire, so that guided waves propagate in the ground wire, and collecting the guided wave response signal of the guided wave specifically includes:

[0012] Generate a white noise signal by a magnetostrictive transducer. The white noise signal has continuous spectrum characteristics, and the frequency range of the white noise signal covers the natural frequency range of the ground wire. The magnetostrictive transducer adopts a non-contact method and converts the white noise signal into mechanical vibration through the magnetostrictive effect, so that guided waves propagate in the ground wire.

[0013] On the basis of the above technical solution, preferably, constructing a spectrum analysis model based on a convolutional neural network, identifying the guided wave response signal through the spectrum analysis model to obtain the natural frequency characteristics of the ground wire, and obtaining the optimal excitation frequency according to the natural frequency characteristics and the amplitude-frequency mapping relationship, specifically includes:

[0014] Construct a spectrum analysis model based on a convolutional neural network. The spectrum analysis model includes: an input layer, a feature extraction layer, an analysis layer, and an output layer; the input layer is used to receive the guided wave response signal under white noise excitation; the feature extraction layer includes multiple convolutional layers and pooling layers, and is used to extract the frequency domain features of the guided wave response signal; the analysis layer includes a fully connected layer, and is used to establish a mapping relationship between frequency and the amplitude of the guided wave; the output layer is used to output the natural frequency characteristics;

[0015] Through time-frequency domain conversion of the guided wave response signal, use the spectrum analysis model to extract the frequency domain features of the signal, determine the change trend of the natural frequency of the ground wire based on the amplitude-frequency relationship, and select the frequency corresponding to the maximum amplitude according to the change trend of the natural frequency to obtain the optimal excitation frequency.

[0016] On the basis of the above technical solution, preferably, the calculation formula of the spectrum analysis model is:

[0017]

[0018] where f is the input frequency of the white noise signal, F(f) is the response amplitude of the guided wave response signal corresponding to the white noise signal with the input frequency f, w i is the weight of the i-th convolution kernel, n is the number of convolutional layers, σ0(·) is the ReLU activation function, xi is the input feature map of the i-th convolutional layer, b i is the bias term of the i-th convolutional layer, E is the Young's modulus of the ground wire material, ρ is the density of the ground wire material, L is the length of the ground wire, d is the diameter of the ground wire, λ(f) is the wavelength corresponding to f, and f0 is the optimal excitation frequency of the ground wire.

[0019] Based on the above technical solutions, preferably, adjusting the excitation frequency of the white noise signal according to the optimal excitation frequency to obtain an optimized guided wave signal specifically includes:

[0020] Based on the optimal excitation frequency, perform frequency modulation and amplitude optimization on the white noise signal to obtain an optimized guided wave signal;

[0021] Use a magnetostrictive transducer to convert the optimized guided wave signal into mechanical vibration, adopt a dynamic compensation method to eliminate the influence of environmental noise on the mechanical vibration, and amplify and filter the received mechanical vibration signal.

[0022] Based on the above technical solutions, preferably, constructing a defect detection and tension monitoring model according to a recurrent neural network, and identifying the optimized guided wave signal through the defect detection and tension monitoring model to obtain a defect tension detection result, specifically including:

[0023] Extract defect features of the optimized guided wave signal through the defect detection and tension monitoring model, including analyzing the time-domain features and frequency-domain features of the optimized guided wave signal, performing time-domain segmentation on the optimized guided wave signal, extracting the instantaneous amplitude and phase information of the optimized guided wave signal, converting the time-domain signal into a frequency-domain signal using a frequency-domain transform, and extracting the spectral features of the optimized guided wave signal for defect identification;

[0024] Based on the defect detection and tension monitoring model, simultaneously perform defect detection and tension anomaly monitoring on the optimized guided wave signal to obtain a defect tension detection result, and the defect tension detection result includes the defect position and the tension change rate;

[0025] The calculation formula of the defect detection model is:

[0026]

[0027] where D(p) is the data of the p-th defect position on the ground wire, ω j is the weight at the j-th time step, J is the total number of time steps, H(f j ) is the feature extraction result of the j-th frequency component, and b2 is the defect position bias;

[0028] The calculation formula of the tension monitoring model is:

[0029]

[0030] Among them, T(m) is the m-th abnormal tensile force value borne by the ground wire, α3 is the scaling factor, tanh(·) is the hyperbolic tangent activation function, and v k is the weight at the k-th time step, K is the total number of time steps, and H(f k ) is the feature extraction result of the k-th frequency component, and c3 is the tensile bias.

[0031] Based on the above technical solutions, preferably, based on the defect tensile force detection result, determine the damage level of the ground wire and generate a detection report, specifically including:

[0032] Based on the defect detection result and the tensile force monitoring result, comprehensively evaluate the damage state of the ground wire, determine the damage level according to the preset evaluation criteria, establish a grading standard for the severity of defects, including minor damage, moderate damage, and severe damage; establish a grading standard for the degree of abnormal tensile force, including normal, warning, and danger. According to the weighted fusion method, comprehensively consider the evaluation results of the two dimensions of the defect position and the degree of abnormal tensile force to obtain the damage level of the ground wire, and generate a detection report including the damage position, damage type, tensile force state, and maintenance suggestions.

[0033] In a second aspect, the present invention also provides a ground wire damage detection system based on magnetostrictive guided waves, and the system includes:

[0034] A guided wave excitation module for generating a white noise signal through a magnetostrictive transducer to excite the ground wire, so that the guided wave propagates in the ground wire and collect the guided wave response signal of the guided wave;

[0035] A frequency prediction module for constructing a spectrum analysis model based on a convolutional neural network, identifying the guided wave response signal through the spectrum analysis model to obtain the natural frequency characteristics of the ground wire, and obtaining the optimal excitation frequency according to the mapping relationship between the natural frequency characteristics and the amplitude frequency;

[0036] A guided wave optimization module for adjusting the excitation frequency of the white noise signal according to the optimal excitation frequency to obtain an optimized guided wave signal;

[0037] A damage detection module for constructing a defect detection and tensile force monitoring model based on a recurrent neural network, identifying the optimized guided wave signal through the defect detection and tensile force monitoring model to obtain a defect tensile force detection result;

[0038] A report generation module for determining the damage level of the ground wire based on the defect tensile force detection result and generating a detection report.

[0039] In a third aspect, the present invention further provides an electronic device, including: at least one processor, at least one memory, a communication interface, and a bus;

[0040] Wherein, the processor, the memory, and the communication interface complete mutual communication through the bus, the memory stores program instructions executable by the processor, and the processor invokes the program instructions to implement the steps of a method for detecting damages of a ground wire based on magnetostrictive guided waves.

[0041] In a fourth aspect, the present invention further provides a computer-readable storage medium, where the computer-readable storage medium stores computer instructions, and the computer instructions enable a computer to implement the steps of a method for detecting damages of a ground wire based on magnetostrictive guided waves.

[0042] The method and system for detecting damages of a ground wire based on magnetostrictive guided waves according to the present invention have the following beneficial effects compared with the prior art:

[0043] (1) By using a magnetostrictive transducer to generate a white noise signal to excite the ground wire, combining with a spectrum analysis model constructed by a convolutional neural network to obtain the optimal excitation frequency, and using a defect detection and tensile force monitoring model constructed by a recurrent neural network to realize the integrated detection of the ground wire, adopting a multi-task learning framework, simultaneously realizing defect detection and tensile force anomaly monitoring, improving the reliability of ground wire detection, and providing an effective monitoring means for the safe operation of the ground wire;

[0044] (2) By adopting a spectrum analysis model constructed by a convolutional neural network, combining with a ReLU activation function and a multi-layer convolutional pooling structure, automatically extracting the frequency-domain features of the guided wave response signal, establishing an accurate mapping relationship between the frequency and the guided wave amplitude, thereby realizing the accurate selection of the optimal excitation frequency and improving the accuracy of natural frequency feature extraction;

[0045] (3) By adopting a recurrent neural network model with a shared hidden layer structure, realizing feature sharing for the defect detection and tensile force monitoring tasks, and designing dedicated output layers for the two tasks respectively, enabling the simultaneous completion of two tasks of defect location detection and tensile force anomaly monitoring, while improving the calculation efficiency and ensuring the detection performance of each task, effectively enhancing the comprehensive performance of ground wire damage detection. Description of the Drawings

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0047] Figure 1 Flow chart of a method for detecting damages of overhead ground wires based on magnetostrictive guided waves according to the present invention;

[0048] Figure 2 Structural diagram of a system for detecting damages of overhead ground wires based on magnetostrictive guided waves according to the present invention. Detailed implementation manners

[0049] Next, in combination with the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] Please refer to Figure 1 , the present invention provides a method for detecting damages of overhead ground wires based on magnetostrictive guided waves, including the following steps:

[0051] Generate a white noise signal through a magnetostrictive transducer to excite the overhead ground wire, so that guided waves propagate in the overhead ground wire, and collect the guided wave response signals of the guided waves;

[0052] Construct a spectrum analysis model based on a convolutional neural network, identify the guided wave response signals through the spectrum analysis model to obtain the natural frequency characteristics of the overhead ground wire, and obtain the optimal excitation frequency according to the mapping relationship between the natural frequency characteristics and the amplitude-frequency. The convolutional neural network is used to establish a relationship model between frequency and guided wave amplitude, and predict the optimal excitation frequency;

[0053] Adjust the excitation frequency of the white noise signal according to the optimal excitation frequency to obtain an optimized guided wave signal;

[0054] Construct a defect detection and tension monitoring model based on a recurrent neural network, identify the optimized guided wave signal through the defect detection and tension monitoring model to obtain a defect tension detection result. The defect detection and tension monitoring model is constructed based on a recurrent neural network, and the defect detection and tension monitoring model adopts a multi-task learning framework to simultaneously perform defect detection and tension anomaly monitoring;

[0055] Based on the defect tension detection result, judge the damage level of the overhead ground wire and generate a detection report.

[0056] Specifically, in this embodiment, a white noise signal is generated by a magnetostrictive transducer to excite the ground wire. A spectrum analysis model constructed by a convolutional neural network is used to obtain the optimal excitation frequency. A defect detection and tension monitoring model constructed by a recurrent neural network is used to achieve integrated detection of the ground wire. A multi-task learning framework is adopted to simultaneously achieve defect detection and abnormal tension monitoring, improving the reliability of ground wire detection and providing an effective monitoring means for the safe operation of the ground wire.

[0057] A white noise signal is generated by a magnetostrictive transducer to excite the ground wire, so that guided waves propagate in the ground wire, and a guided wave response signal of the guided waves is collected. Specifically, it includes:

[0058] A white noise signal is generated by a magnetostrictive transducer. The white noise signal has continuous spectrum characteristics, and the frequency range of the white noise signal covers the natural frequency range of the ground wire, which is used to measure the natural frequency of the ground wire efficiently and accurately. The magnetostrictive transducer adopts a non-contact method and converts the white noise signal into mechanical vibration through the magnetostrictive effect, so that guided waves propagate in the ground wire, realizing the excitation and reception of guided waves.

[0059] Specifically, in this embodiment, by adopting a white noise signal with continuous spectrum characteristics and making its frequency range cover the natural frequency range of the ground wire, the natural frequency of the ground wire can be measured efficiently and accurately. At the same time, a non-contact magnetostrictive transducer is adopted to realize the conversion between the signal and mechanical vibration through the magnetostrictive effect, which not only avoids the damage that traditional contact detection may cause to the ground wire, but also improves the safety of detection, providing a more stable implementation method for the excitation and reception of guided waves.

[0060] A spectrum analysis model is constructed based on a convolutional neural network. The guided wave response signal is identified through the spectrum analysis model to obtain the natural frequency characteristics of the ground wire. According to the natural frequency characteristics and the amplitude-frequency mapping relationship, the optimal excitation frequency is obtained. Specifically, it includes:

[0061] A spectrum analysis model is constructed based on a convolutional neural network. The spectrum analysis model includes: an input layer, a feature extraction layer, an analysis layer, and an output layer. The input layer is used to receive the guided wave response signal under white noise excitation. The feature extraction layer contains multiple convolutional layers and pooling layers, which are used to extract the frequency domain characteristics of the guided wave response signal. The analysis layer contains fully connected layers, which are used to establish the mapping relationship between frequency and the amplitude of guided waves. The output layer is used to output the natural frequency characteristics.

[0062] By performing time-frequency domain conversion on the guided wave response signal, using the spectrum analysis model to extract the frequency domain characteristics of the signal, determining the changing trend of the natural frequency of the ground wire based on the amplitude-frequency relationship, and selecting the frequency corresponding to the maximum amplitude according to the changing trend of the natural frequency, the optimal excitation frequency is obtained.

[0063] The calculation formula of the spectrum analysis model is as follows:

[0064]

[0065] Among them, f is the input frequency of the white noise signal, F(f) is the response amplitude of the guided wave response signal corresponding to the white noise signal with the input frequency f, w i is the weight of the i-th convolution kernel, n is the number of convolutional layers, σ0(·) is the ReLU activation function, x i is the input feature map of the i-th convolutional layer, b i is the bias term of the i-th convolutional layer, E is the Young's modulus of the ground wire material, ρ is the density of the ground wire material, L is the length of the ground wire, d is the diameter of the ground wire, λ(f) is the wavelength corresponding to f, f0 is the optimal excitation frequency of the ground wire, is the solution function of the value of f when the value of F(f) is the largest.

[0066] Specifically, by adopting the spectrum analysis model based on the convolutional neural network, the automatic feature extraction and analysis of the guided wave response signal are realized. Among them, the multi-layer convolution and pooling structure can effectively extract the frequency domain features of the signal, and the fully connected layer establishes an accurate mapping relationship between the frequency and the guided wave amplitude.

[0067] By performing time-frequency domain analysis on the guided wave response signal, this model grasps the changing trend of the natural frequency of the ground wire, thereby realizing the automatic selection of the optimal excitation frequency, providing a more reliable signal basis for defect detection, and improving the accuracy of detection.

[0068] In a specific embodiment, the spectrum analysis model is realized through the following steps:

[0069] 1) Input processing:

[0070] Receive the input frequency f of the white noise signal and the corresponding guided wave response signal amplitude A(f), and consider the material parameters (Young's modulus E, density ρ) and geometric parameters (length L, diameter d) of the ground wire.

[0071] 2) Convolutional layer calculation:

[0072] Use n convolutional kernels to extract features from the input signal, and perform non-linear transformation using the ReLU activation function. Each convolutional layer contains the input feature map x i and the bias term b i .

[0073] 3) Optimal frequency solution:

[0074] Based on the wavelength λ(f), the relationship between frequency and guided wave characteristics is established. By solving the function arg max F(f), the frequency value f that maximizes the amplitude of the guided wave response signal is determined, and the optimal excitation frequency f0 is output.

[0075] The layer convolution structure sampled in this embodiment can automatically extract multi-scale features of the signal. The ReLU activation function provides good non-linear mapping ability. By considering the influence of material and geometric parameters, the model is made more universal.

[0076] Adjust the excitation frequency of the white noise signal according to the optimal excitation frequency to obtain an optimized guided wave signal, specifically including:

[0077] Based on the optimal excitation frequency, frequency modulation and amplitude optimization are performed on the white noise signal to obtain an optimized guided wave signal. Using frequency modulation technology, the center frequency of the white noise signal is adjusted to the optimal excitation frequency, and the modulated signal is filtered using a band-pass filter. According to the material characteristics of the ground wire, the signal amplitude is adaptively adjusted;

[0078] The calculation formula for the frequency modulation is:

[0079] S mod (t) = A(t)·sin(2πf opt t + φ(t))·M(t);

[0080] Among them, S mod (t) is the guided wave signal after frequency modulation at time step t, A(t) is the amplitude of the white noise signal at time step t, f opt t is the optimal excitation frequency at time step t, φ(t) is the phase modulation function, and M(t) is the modulation depth control function;

[0081] The calculation formula for the amplitude optimization is:

[0082] S opt (t) = G1·S mod (t)·exp(-α1d1)·F(f);

[0083] Among them, S opt (t) is the optimized guided wave signal at time step t, G1 is the gain coefficient, S mod (t) is the guided wave signal after frequency modulation at time step t, exp(·) is the exponential function, α1 is the attenuation coefficient, d1 is the guided wave propagation distance, and F(f) is the response amplitude of the guided wave response signal corresponding to the white noise signal of input frequency f;

[0084] The optimized guided wave signal is converted into mechanical vibration by a magnetostrictive transducer, and the dynamic compensation method is used to eliminate the influence of environmental noise on the mechanical vibration, and the received mechanical vibration signal is amplified and filtered.

[0085] In a specific embodiment, taking the detection of the ground wire of a 500 kV transmission line as an example, the ground wire adopts a steel-cored aluminum stranded wire structure, and the specific implementation steps are as follows:

[0086] 1) Frequency modulation process:

[0087] The calculation formula is:

[0088] S mod (t) = A(t)·sin(2πf opt t + φ(t))·M(t);

[0089] Parameter setting: The amplitude of the white noise signal of A(t) is set within the range of 0.5 - 2 V, f opt t is set according to the optimal excitation frequency obtained from the previous spectrum analysis (usually in the range of 20 - 100 kHz), φ(t) is set as the phase modulation of 0 - 2π, and M(t) is set as the modulation depth range of 0.8 - 1.2 m.

[0090] 2) Amplitude optimization process:

[0091] The calculation formula is:

[0092] S opt (t) = G1·S mod (t)·exp(-α1d1)·F(f);

[0093] Parameter setting: The gain coefficient G1 is set to 2 - 5, the attenuation coefficient α1 is set to 0.1 - 0.3 db / m, and the guided wave propagation distance d1 is set to 100 m.

[0094] 3) Signal processing

[0095] A transducer made of nickel-based magnetostrictive material is used, with a working frequency range of 20 - 100 kHz. The dynamic compensation adopts an adaptive filtering algorithm to suppress the 50 Hz power frequency interference. The signal amplification factor is set to 40 dB, and a band-pass filter with a bandwidth of 1 - 200 kHz is used.

[0096] According to the above-mentioned conductor and ground wire detection scheme, the following can be achieved: the signal-to-noise ratio is increased by 15 - 20 dB, the effective detection distance is increased from the original 50 m to more than 100 m, the environmental noise suppression rate reaches more than 85%, the defect location accuracy is improved to ±0.5 m, the detection sensitivity can identify damages with a cross-sectional area of more than 2%, the signal acquisition time is shortened by 50%, it can adapt to the working temperature range of -20°C to +60°C, the ability to resist 50 Hz power frequency interference is enhanced, the system stability is improved, and the continuous working time can reach 72 hours.

[0097] Construct a defect detection and tension monitoring model based on a recurrent neural network, and identify the optimized guided wave signal through the defect detection and tension monitoring model to obtain a defect tension detection result, specifically including:

[0098] Extract defect features from the optimized guided wave signal through the defect detection and tension monitoring model, including analyzing the time-domain features and frequency-domain features of the optimized guided wave signal, performing time-domain segmentation on the optimized guided wave signal, extracting the instantaneous amplitude and phase information of the optimized guided wave signal, converting the time-domain signal into a frequency-domain signal using frequency-domain transformation, and extracting the spectral features of the optimized guided wave signal for defect identification;

[0099] Based on the defect detection and tension monitoring model, simultaneously perform defect detection and tension anomaly monitoring on the optimized guided wave signal to obtain a defect tension detection result. The defect tension detection result includes the defect position and the tension change rate. Use a shared hidden layer structure to achieve feature sharing for the defect detection and tension monitoring tasks, and design dedicated output layers for the defect detection and tension monitoring tasks respectively to optimize their respective detection performances;

[0100] The calculation formula of the defect detection model is:

[0101]

[0102] where D(p) is the data of the p-th defect position on the conductor and ground wire, ω j is the weight at the j-th time step, J is the total number of time steps, H(f j ) is the feature extraction result of the j-th frequency component, and b2 is the defect position bias;

[0103] The calculation formula of the tension monitoring model is:

[0104]

[0105] where T(m) is the m-th abnormal tension value borne by the conductor and ground wire, α3 is the scaling factor, tanh(·) is the hyperbolic tangent activation function, v k is the weight at the k-th time step, K is the total number of time steps, H(f kis the feature extraction result of the k-th frequency component, and c3 is the tension offset.

[0106] In a specific embodiment, taking the detection of the ground wire of a 220 kV transmission line as an example, the ground wire is of an aluminum-clad steel stranded wire structure, and the specific implementation steps are as follows:

[0107] 1) Defect feature extraction:

[0108] Perform a 100 ms time window segmentation on the optimized guided wave signal, set the sampling frequency to 1 MHz, extract the instantaneous amplitude (amplitude range 0.2 - 3 V) and phase information of the signal, and use the Hilbert transform to extract the phase information; use the fast Fourier transform to analyze the frequency range of 10 - 100 kHz, set the frequency resolution to 50 Hz, and extract the main frequency component and harmonic features for defect identification.

[0109] 2) Defect detection and tension monitoring model technology:

[0110] Defect detection parameter settings: Set the number of time steps J to 120, and the weight ω j ranges from [-1, 1], and the feature extraction result H(f j ) is the normalized spectrum data, and the defect position offset b2 is set to 0.05;

[0111] Tension monitoring parameter settings: Set the scaling factor α3 to 1.2, the number of time steps K to 100, and the weight v k ranges from [-0.5, 0.5], and the feature extraction result H(f k ) is the normalized amplitude data, and the tension offset c3 is set to 0.1.

[0112] According to the above ground wire detection scheme, the following can be achieved: the defect location accuracy reaches ±0.2 m, the minimum detectable defect size is reduced to 1% of the cross-sectional area, the tension monitoring accuracy reaches ±1.5%, the false alarm rate is reduced to less than 2%, feature sharing improves the calculation efficiency, the processing time is shortened by 30%, the accuracy of defect detection and tension monitoring is simultaneously increased by 20%, the system resource utilization rate is increased by 25%, the single detection time is controlled within 1.5 seconds, the data processing delay is reduced to within 80 ms, and the continuous monitoring ability is improved to 48 hours without interruption.

[0113] Based on the defect tension detection results, judge the damage level of the ground wire and generate a detection report, which specifically includes:

[0114] Based on the defect detection results and tension monitoring results, comprehensively evaluate the damage status of the conductor and ground wire, determine the damage level according to the preset evaluation criteria, establish a grading standard for the severity of defects, including minor damage, moderate damage, and severe damage; establish a grading standard for the abnormal degree of tension, including normal, warning, and danger. According to the weighted fusion method, comprehensively consider the evaluation results of two dimensions: the defect location and the abnormal degree of tension, obtain the damage level of the conductor and ground wire, and generate a detection report containing the damage location, damage type, tension status, and maintenance suggestions. The detection report records the basic detection information, including the detection time, location, environmental conditions, etc., details the damage characteristics, including the precise coordinates of the damage location and the specific description of the damage type, gives the evaluation result of the tension status, including the tension value, change trend, and abnormal degree, and provides targeted maintenance suggestions, including repair plans, detection cycles, and preventive measures.

[0115] In a specific embodiment, taking the detection of the conductor and ground wire of a 500 kV UHV transmission line as an example, the conductor and ground wire is a 48 / 7 structure galvanized steel strand. The specific implementation steps are as follows:

[0116] 1) Setting of damage level standards:

[0117] ① Grading of the severity of defects: Minor damage: Cross-sectional area loss < 5%, or surface corrosion depth < 0.5 mm; Moderate damage: Cross-sectional area loss 5% - 15%, or surface corrosion depth 0.5 - 1.5 mm; Severe damage: Cross-sectional area loss > 15%, or surface corrosion depth > 1.5 mm.

[0118] ② Grading of the abnormal degree of tension: Normal: Tension change rate within ±5%; Warning: Tension change rate between ±5% and ±10%; Danger: Tension change rate exceeding ±10%.

[0119] 2) Comprehensive evaluation method:

[0120] Weighted fusion calculation, damage score = 0.6 × defect degree score + 0.4 × tension anomaly score; Defect degree score: Minor (1 - 3 points), Moderate (4 - 7 points), Severe (8 - 10 points); Tension anomaly score: Normal (1 - 3 points), Warning (4 - 7 points), Danger (8 - 10 points).

[0121] 3) Generation of the detection report:

[0122] Recording of basic information: Detection time, accurate to the minute; Geographical location, GPS coordinates (accuracy ±1 m); Environmental conditions, temperature, humidity, wind speed.

[0123] Description of damage characteristics: Damage location, marked tower number + distance from the reference point; Damage type, broken wire, corrosion, wear; Damage size, length, width, depth.

[0124] Maintenance suggestions: repair solutions, replacement, reinforcement or monitoring; inspection cycle, normal (12 months), warning (3 months), dangerous (handle immediately); preventive measures, anti-corrosion treatment, stress relief.

[0125] Applying the above solutions can achieve: improving the damage location accuracy to ±0.1 m, increasing the accuracy of damage degree assessment to 95%, increasing the early warning time of abnormal tension to 72 hours, raising the accuracy of maintenance suggestions to 90%, increasing the accuracy of judgment of maintenance priority to 93%, improving the accuracy of life prediction to 85%, shortening the report generation time to within 5 minutes, enhancing the data traceability to 100%, increasing the efficiency of maintenance plan formulation by 40%, reducing the preventive maintenance cost by 30%, extending the equipment life by 15% - 20%, and improving the maintenance work efficiency by 35%.

[0126] Please refer to Figure 2 , the present invention also provides a ground wire damage detection system based on magnetostrictive guided waves, and the system includes:

[0127] A guided wave excitation module, configured to generate a white noise signal through a magnetostrictive transducer to excite the ground wire, so that the guided wave propagates in the ground wire, and collect the guided wave response signal of the guided wave;

[0128] A frequency prediction module, configured to construct a spectrum analysis model based on a convolutional neural network, identify the guided wave response signal through the spectrum analysis model to obtain the natural frequency characteristics of the ground wire, and obtain the optimal excitation frequency according to the natural frequency characteristics and the amplitude-frequency mapping relationship;

[0129] A guided wave optimization module, configured to adjust the excitation frequency of the white noise signal according to the optimal excitation frequency to obtain an optimized guided wave signal;

[0130] A damage detection module, configured to construct a defect detection and tension monitoring model based on a recurrent neural network, identify the optimized guided wave signal through the defect detection and tension monitoring model to obtain a defect tension detection result;

[0131] A report generation module, configured to judge the damage level of the ground wire based on the defect tension detection result and generate a detection report.

[0132] The present invention also discloses an electronic device, including: at least one processor, at least one memory communication interface and a bus: wherein, the processor, the memory and the communication interface complete mutual communication through the bus; the memory stores program instructions executable by the processor, and the processor calls the program instructions to implement a ground wire damage detection method based on magnetostrictive guided waves.

[0133] The present invention also discloses a computer-readable storage medium storing computer instructions that enable a computer to implement all or part of the steps of the method for detecting damages to ground wires based on magnetostrictive guided waves according to an embodiment of the present invention. The storage medium includes various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs.

[0134] The foregoing is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for detecting damage to overhead ground wires based on magnetostrictive guided waves, characterized in that, It includes the following steps: Generate a white noise signal through a magnetostrictive transducer to excite the ground wire, so that guided waves propagate in the ground wire, and collect the guided wave response signals of the guided waves; Construct a spectrum analysis model based on a convolutional neural network, identify the guided wave response signals through the spectrum analysis model to obtain the natural frequency characteristics of the ground wire, and obtain the optimal excitation frequency according to the mapping relationship between the natural frequency characteristics and the amplitude-frequency; Adjust the excitation frequency of the white noise signal according to the optimal excitation frequency to obtain an optimized guided wave signal; Construct a defect detection and tension monitoring model based on a recurrent neural network, identify the optimized guided wave signal through the defect detection and tension monitoring model to obtain a defect tension detection result; Based on the defect tension detection result, judge the damage level of the ground wire and generate a detection report.

2. The method for detecting damage to a ground wire based on magnetostrictive guided waves according to claim 1, characterized in that, The step of generating a white noise signal through a magnetostrictive transducer to excite the ground wire, so that guided waves propagate in the ground wire, and collect the guided wave response signals of the guided waves specifically includes: Generate a white noise signal through a magnetostrictive transducer. The white noise signal has continuous spectrum characteristics, and the frequency range of the white noise signal covers the natural frequency range of the ground wire. The magnetostrictive transducer adopts a non-contact method and converts the white noise signal into mechanical vibration through the magnetostrictive effect, so that guided waves propagate in the ground wire.

3. The method for detecting damages of a ground wire based on magnetostrictive guided waves according to claim 1, characterized in that, The step of constructing a spectrum analysis model based on a convolutional neural network, identifying the guided wave response signals through the spectrum analysis model to obtain the natural frequency characteristics of the ground wire, and obtaining the optimal excitation frequency according to the mapping relationship between the natural frequency characteristics and the amplitude-frequency specifically includes: Construct a spectrum analysis model based on a convolutional neural network. The spectrum analysis model includes: an input layer, a feature extraction layer, an analysis layer, and an output layer; the input layer is used to receive the guided wave response signals under white noise excitation; the feature extraction layer includes multiple convolutional layers and pooling layers, and is used to extract the frequency domain characteristics of the guided wave response signals; the analysis layer includes fully connected layers and is used to establish the mapping relationship between frequency and guided wave amplitude; the output layer is used to output the natural frequency characteristics; Through time-frequency domain conversion of the guided wave response signals, use the spectrum analysis model to extract the frequency domain characteristics of the signals, determine the change trend of the natural frequency of the ground wire based on the amplitude-frequency relationship, and select the frequency corresponding to the maximum amplitude according to the change trend of the natural frequency to obtain the optimal excitation frequency.

4. The method for detecting damages of a ground wire based on magnetostrictive guided waves according to claim 3, characterized in that, The calculation formula of the spectrum analysis model is: Among them, f is the input frequency of the white noise signal, F(f) is the response amplitude of the guided wave response signal corresponding to the white noise signal of the input frequency f, w i is the weight of the i-th convolution kernel, n is the number of convolutional layers, σ0(·) is the ReLU activation function, x i is the input feature map of the i-th convolutional layer, b i is the bias term of the i-th convolutional layer, E is the Young's modulus of the ground wire material, ρ is the density of the ground wire material, L is the length of the ground wire, d is the diameter of the ground wire, λ(f) is the wavelength corresponding to f, and f0 is the optimal excitation frequency of the ground wire.

5. The method for detecting damages of ground wires based on magnetostrictive guided waves according to claim 1, characterized in that, The step of adjusting the excitation frequency of the white noise signal according to the optimal excitation frequency to obtain an optimized guided wave signal specifically includes: Based on the optimal excitation frequency, perform frequency modulation and amplitude optimization on the white noise signal to obtain an optimized guided wave signal; Use a magnetostrictive transducer to convert the optimized guided wave signal into mechanical vibration, adopt a dynamic compensation method to eliminate the influence of environmental noise on the mechanical vibration, and amplify and filter the received mechanical vibration signals.

6. The method for detecting damage to a ground wire based on magnetostrictive guided waves according to claim 1, characterized in that, The step of constructing a defect detection and tension monitoring model based on a recurrent neural network, identifying the optimized guided wave signal through the defect detection and tension monitoring model to obtain a defect tension detection result specifically includes: Performing defect feature extraction on the optimized guided wave signal through the defect detection and tensile force monitoring model, including analyzing the time-domain features and frequency-domain features of the optimized guided wave signal, performing time-domain segmentation on the optimized guided wave signal, extracting the instantaneous amplitude and phase information of the optimized guided wave signal, converting the time-domain signal into a frequency-domain signal by using frequency-domain transformation, and extracting the spectral features of the optimized guided wave signal for defect identification; Based on the defect detection and tensile force monitoring model, simultaneously performing defect detection and tensile force anomaly monitoring on the optimized guided wave signal to obtain a defect tensile force detection result, where the defect tensile force detection result includes a defect position and a tensile force change rate; The calculation formula of the defect detection model is: Among them, D(p) is the data of the p-th defect position on the ground wire, ω j is the weight at the j-th time step, J is the total number of time steps, H(f j ) is the feature extraction result of the j-th frequency component, and b2 is the defect position offset; The calculation formula of the tensile force monitoring model is: Among them, T(m) is the m-th abnormal tensile force value borne by the ground wire, α3 is the scaling factor, tanh(·) is the hyperbolic tangent activation function, v k is the weight at the k-th time step, K is the total number of time steps, H(f k ) is the feature extraction result of the k-th frequency component, and c3 is the tensile bias.

7. The method for detecting damages of ground wires based on magnetostrictive guided waves according to claim 1, wherein Based on the defect tensile force detection result, judging the damage level of the ground wire and generating a detection report, specifically including: Based on the defect detection result and the tensile force monitoring result, comprehensively evaluating the damage state of the ground wire, determining the damage level according to the preset evaluation criteria, establishing a defect severity grading standard, including minor damage, moderate damage, and severe damage; establishing a tensile force anomaly degree grading standard, including normal, warning, and danger, and according to the weighted fusion method, comprehensively considering the evaluation results of the two dimensions of the defect position and the tensile force anomaly degree, obtaining the damage level of the ground wire, and generating a detection report including the damage position, damage type, tensile force state, and maintenance suggestions.

8. A detection system for damage of transmission line conductors and ground wires based on magnetostrictive guided waves, characterized in that, The system includes: A guided wave excitation module, configured to generate a white noise signal through a magnetostrictive transducer to excite the ground wire, so that the guided wave propagates in the ground wire, and collect the guided wave response signal of the guided wave; A frequency prediction module, configured to construct a spectral analysis model based on a convolutional neural network, identify the guided wave response signal through the spectral analysis model to obtain the natural frequency characteristics of the ground wire, and obtain the optimal excitation frequency according to the natural frequency characteristics and the amplitude-frequency mapping relationship; A guided wave optimization module, configured to adjust the excitation frequency of the white noise signal according to the optimal excitation frequency to obtain an optimized guided wave signal; A damage detection module, configured to construct a defect detection and tensile force monitoring model based on a recurrent neural network, identify the optimized guided wave signal through the defect detection and tensile force monitoring model to obtain a defect tensile force detection result; A report generation module, configured to judge the damage level of the ground wire based on the defect tensile force detection result and generate a detection report.

9. An electronic device, characterized in that, Including: At least one processor, at least one memory, a communication interface, and a bus; Wherein, the processor, the memory, and the communication interface complete communication with each other through the bus, the memory stores program instructions executable by the processor, and the processor calls the program instructions to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to implement the method according to any one of claims 1 to 7.

Images