Distributed power transmission line fault diagnosis method based on ground potential end transient state information perception

Abstract

The invention is suitable for the technical field of fault diagnosis, and provides a distributed power transmission line fault diagnosis method based on ground potential end transient information perception, and the method comprises the steps: collecting a broadband transient signal of each node ground potential end of a power transmission line; synchronously acquiring an electromagnetic transient signal and an environmental parameter through the composite sensing unit; performing dynamic filtering correction on the broadband transient signal according to a preset transmission line impedance matching model; a deep learning model is adopted to perform hybrid processing on the corrected broadband transient signals and electromagnetic transient signals, traveling wave head time-frequency features and power frequency abrupt change features are extracted, and fault probability and type classification results are output; verifying a fault positioning result through a multi-stage verification mechanism; the extracted characteristic quantity is preprocessed at the edge side of the detection terminal, and the processed data is transmitted to a master station system through an encryption protocol, so that the fault diagnosis accuracy of the distributed power transmission line is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and particularly relates to a distributed transmission line fault diagnosis method based on transient information perception at the ground potential terminal. Background Art

[0002] As a core component of the power system, the safe and stable operation of transmission lines is crucial for the reliability of the power grid. However, with the expansion of the power grid scale and the complexity of the operating environment, transmission line faults occur frequently, especially faults such as lightning strikes, short circuits, and insulation aging, which seriously threaten the safety of the power system. At present, traditional traveling wave positioning is vulnerable to high-frequency signal attenuation and environmental noise interference. Especially under complex meteorological conditions, the stability of the signal propagation speed is poor, resulting in positioning errors exceeding the engineering allowable range. In addition, high-frequency traveling wave signals are vulnerable to environmental noise and attenuation. Due to the lack of a multi-modal signal collaborative processing mechanism, the complementary information between high-frequency traveling waves and power frequency mutations is not fully utilized, resulting in low fault recognition efficiency. At the same time, traditional methods do not correct the influence of environmental parameters on signal propagation in real time, and the model mismatch problem is serious, further reducing the accuracy of fault diagnosis. Summary of the Invention

[0003] An embodiment of the present application provides a distributed transmission line fault diagnosis method based on transient information perception at the ground potential terminal, which is used to solve the problem of low accuracy of fault diagnosis.

[0004] A first aspect of an embodiment of the present application provides a distributed transmission line fault diagnosis method based on transient information perception at the ground potential terminal, including:

[0005] Collecting broadband transient signals at the ground potential terminals of each node of the transmission line through distributed detection terminals, where the broadband transient signals include traveling wave components, power frequency voltage mutation signals, and power frequency current mutation signals in the range of 0.1 MHz to 10 MHz;

[0006] Synchronously obtaining electromagnetic transient signals and environmental parameters through a composite sensing unit, where the composite sensing unit includes an electromagnetic induction sensor, a capacitive voltage divider, and an environmental monitoring module;

[0007] Performing dynamic filtering correction on the broadband transient signals according to a preset transmission line impedance matching model, where the dynamic filtering correction is to adjust the propagation speed of the broadband transient signals and compensate for the influence of environmental temperature on the relative dielectric constant of the medium;

[0008] Using a deep learning model to perform hybrid processing on the corrected broadband transient signals and the electromagnetic transient signals, extracting the time-frequency characteristics of the traveling wave head and the power frequency mutation characteristics, and outputting the fault probability and type classification results;

[0009] Verify the fault location result through a multi-level verification mechanism;

[0010] Preprocess the extracted feature quantities on the edge side of the detection terminal, and transmit the processed data to the master station system through an encryption protocol.

[0011] Furthermore, the electromagnetic transient signal and environmental parameters are synchronously acquired through the composite sensing unit. The composite sensing unit includes an electromagnetic induction sensor, a capacitive voltage divider, and an environmental monitoring module, including:

[0012] The electromagnetic induction sensor adopts a dual-redundancy Rogowski coil structure with a frequency response range of 0.1 MHz - 15 MHz;

[0013] The capacitive voltage divider adopts a composite structure of ceramic capacitors and grading rings, and is configured with a temperature compensation circuit to suppress the change of dielectric loss;

[0014] The environmental monitoring module integrates a temperature and humidity sensor, a salt spray concentration sensor, and a three-dimensional vibration sensor, and the surface of the sensor probe is coated with a nano-hydrophobic coating.

[0015] Furthermore, the wideband transient signal is dynamically filtered and corrected according to the preset transmission line impedance matching model. The dynamic filtering and correction is to adjust the propagation speed of the wideband transient signal and compensate for the influence of environmental temperature on the relative dielectric constant of the medium, including:

[0016] Divide the transmission line into several equal-length sections, and then determine the resistance parameters, inductance parameters, conductance parameters, and capacitance parameters of each section. The section length is determined according to the traveling wave propagation delay;

[0017] Correct the resistance parameters, inductance parameters, conductance parameters, and capacitance parameters of each section based on the temperature and humidity data collected by the environmental monitoring module;

[0018] Calculate the characteristic impedance of each section according to the corrected parameters, and at the same time determine the propagation speed of the wideband transient signal based on the characteristic impedance of each section.

[0019] Furthermore, the correction of the resistance parameters, inductance parameters, conductance parameters, and capacitance parameters of each section based on the temperature and humidity data collected by the environmental monitoring module includes:

[0020] R′ = R0·[1 + β R (T ― T0)]

[0021] L′ = L0·[1 + β L (T ― T0)]

[0022]

[0023] C′ = C0·[1 + βC (T – T0)]

[0024] Where: R′, L′, G′, and C′ are the modified resistance, inductance, conductance, and capacitance per unit length respectively, and R0, L0, G0, and C0 are the per unit length parameters at the reference temperature T0, and β R , β L and β C are the temperature coefficients of resistance, inductance, and capacitance respectively, k H is the conductance correction factor for temperature, T is the measured value of the ambient temperature sensor, and H is the relative humidity.

[0025] Furthermore, calculating the characteristic impedance of each section according to the modified parameters, and determining the propagation speed of the broadband transient signal based on the characteristic impedance of each section, includes:

[0026] Calculating the reference value of the theoretical traveling wave transmission speed based on the inductance parameter and capacitance parameter, and calculating the characteristic impedance of each section based on the resistance parameter, inductance parameter, conductance parameter, and capacitance parameter;

[0027] Correcting the reference value through a preset temperature and dielectric constant compensation factor to obtain the corrected propagation speed;

[0028] Calculating the target propagation speed according to the corrected propagation speed, the characteristic impedance of each section, and a preset fault point impedance deviation amount, where the preset fault point impedance deviation amount is used to reflect the mismatch degree between the fault point impedance and the line characteristic impedance.

[0029] Furthermore, calculating the target propagation speed according to the corrected propagation speed, the characteristic impedance of each section, and a preset fault point impedance deviation amount, where the preset fault point impedance deviation amount is used to reflect the mismatch degree between the fault point impedance and the line characteristic impedance, includes:

[0030] The calculation formula for the characteristic impedance Z c of each section:

[0031]

[0032] Where: j is the imaginary unit, describing the phase relationship, and ω is the angular frequency, reflecting the signal frequency characteristic;

[0033] The calculation formula for the preset fault point impedance deviation amount ΔZ:

[0034]

[0035] Where: N is the total number of monitoring points, V n and I nThey are the fundamental frequency component ratios after FFT transformation of the voltage and current time-domain waveforms at the nth monitoring point, respectively, Z ref is the line design reference impedance, γ is the empirical value of the traveling wave attenuation coefficient, x n is the initial positioning distance of the monitoring point n from the theoretical fault point;

[0036] The target propagation speed v actual Calculation formula:

[0037]

[0038] Among them: v T is the corrected propagation speed.

[0039] Furthermore, the deep learning model is used to perform hybrid processing on the corrected wideband transient signal and the electromagnetic transient signal, extract the time-frequency characteristics of the traveling wave head and the power frequency mutation characteristics, and output the fault probability and type classification results, including:

[0040] Normalize the corrected wideband transient signal, extract vectors from the electromagnetic transient signal to obtain the trajectory characteristics in the dq coordinate system, and the trajectory characteristics are used to describe the amplitude and phase change characteristics of voltage and current;

[0041] Construct the input architecture of the deep learning model including a high-frequency channel and a power-frequency channel. The high-frequency channel uses a two-dimensional convolutional neural network to process the time-frequency characteristic matrix of the traveling wave head, and the power-frequency channel uses a long short-term memory network to analyze the trajectory characteristics, voltage mutation time series characteristics, and current mutation time series characteristics in the dq coordinate system;

[0042] Weightedly splice the time-frequency feature vector output by the high-frequency channel and the time series feature vector output by the power-frequency channel to generate a target feature vector;

[0043] Calculate the fault probability and classify according to the target feature vector.

[0044] Furthermore, the normalization process of the corrected wideband transient signal and the extraction of vectors from the electromagnetic transient signal to obtain the trajectory characteristics in the dq coordinate system, and the trajectory characteristics are used to describe the amplitude and phase change characteristics of voltage and current, including:

[0045] Perform Park transformation on the three-phase voltage signal and the three-phase current signal in the electromagnetic transient signal. The formula is:

[0046]

[0047] Among them: v d 、v q are the active component and reactive component in the dq coordinate system, va 、v b 、v c are the instantaneous values of the three-phase voltage, and i a 、i b 、i c are the instantaneous values of the three-phase current, θ = wt is the rotation angle, w = 2π × 50Hz, cosθ, is the cosine function term for projecting the three-phase signal onto the d-axis, -sinθ, is the negative sine function term for projecting the three-phase signal onto the q-axis;

[0048] Calculate the amplitude and phase angle to generate a time-varying trajectory matrix.

[0049] Furthermore, the construction of the input architecture of the deep learning model including a high-frequency channel and a power-frequency channel, where the high-frequency channel uses a two-dimensional convolutional neural network to process the time-frequency feature matrix of the traveling wave head, and the power-frequency channel uses a long short-term memory network to analyze the trajectory features, voltage mutation time series features, and current mutation time series features in the dq coordinate system, includes:

[0050] The structure of the two-dimensional convolutional neural network includes: an input layer for the input time-frequency matrix; a convolutional layer with 3 layers, a filter size of 3×3, a number of channels from 32→64→128, and a ReLU activation function; a pooling layer with 2×2 max pooling; an output layer that obtains a time-frequency feature vector after flattening;

[0051] The structure of the long short-term memory network includes: an input layer for the input time series data; an LSTM layer with 2 layers, hidden units from 128→64, and a Tanh activation function; an output layer of a fully connected layer that generates a time series feature vector.

[0052] Furthermore, the verification of the fault location result through a multi-level verification mechanism includes:

[0053] Calculate the theoretical value and measured value of the time difference of arrival of the traveling waves at the monitoring nodes at both ends of the line. If the deviation meets the first threshold condition, trigger the redundant sensor data to recalculate the fault location;

[0054] Calculate the Pearson correlation coefficient between the extracted power-frequency impedance trajectory and the high-frequency traveling wave waveform. If the Pearson correlation coefficient meets the second threshold condition, start the environmental parameter review and correct the signal attenuation model according to the real-time temperature and humidity data;

[0055] Generate a theoretical fault propagation path through digital twin technology and perform spatio-temporal matching with the measured traveling wave arrival time series. If the positioning error meets the third threshold condition, determine that the fault location result is valid.

[0056] From the above technical solutions, it can be seen that the present invention has the following advantages:

[0057] The present invention collects broadband transient signals and electromagnetic transient signals through distributed detection terminals, and performs collaborative processing and feature extraction of multi-modal signals through dynamic filtering correction and deep learning models, improving the accuracy of fault recognition and the positioning accuracy; through a multi-level verification mechanism and edge-side preprocessing, environmental noise interference and data transmission delay are reduced; a lightweight encryption protocol is adopted to ensure the security of data transmission, solving the defects of traditional methods in terms of computational efficiency, accuracy, and security, and effectively improving the accuracy of transmission line fault diagnosis in complex environments. Brief Description of the Drawings

[0058] Figure 1 It is a schematic flowchart of an embodiment of a distributed transmission line fault diagnosis method based on transient information perception at the ground potential end in the present invention. Detailed Embodiments

[0059] In order to make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0060] In this embodiment, the distributed transmission line fault diagnosis method based on transient information perception at the ground potential end is used to improve the efficiency and accuracy of fault diagnosis. The implementation method in this embodiment can be implemented in a system, can be implemented on a server, or can be implemented on a terminal, and no specific limitation is made.

[0061] Embodiment 1

[0062] Please refer to Figure 1 , an embodiment of a distributed transmission line fault diagnosis method based on transient information perception at the ground potential end in the present invention includes the following steps:

[0063] S11. Collect broadband transient signals at the ground potential end of each node of the transmission line through distributed detection terminals. The broadband transient signals include traveling wave components, power frequency voltage mutation signals, and power frequency current mutation signals in the range of 0.1 MHz to 10 MHz;

[0064] In this embodiment, the detection terminals deployed in a distributed manner are used to collect and process the fault signals of the transmission line: a broadband Rogowski coil and a capacitive voltage divider are used to synchronously collect high-frequency traveling wave components of 0.1 MHz - 10 MHz and power frequency mutation signals, and GPS synchronous sampling is combined to ensure the consistency of the data time sequence of multiple nodes; at the hardware layer, interference is suppressed through shielding design and band-pass filtering, and at the software layer, dynamic sampling rate switching and adaptive filtering are used to improve the signal quality; the detection terminals are deployed at the sensitive nodes of the line, and data is transmitted back through optical fiber or wireless communication, and the transmission load is reduced through edge-side caching and compression, providing a multi-modal transient data basis with high precision and low latency for fault diagnosis.

[0065] S12. Synchronously obtain the electromagnetic transient signal and environmental parameters through the composite sensing unit, and the composite sensing unit includes an electromagnetic induction sensor, a capacitive voltage divider and an environmental monitoring module;

[0066] Step S12 includes the following:

[0067] The electromagnetic induction sensor adopts a dual-redundancy Rogowski coil structure with a frequency response range of 0.1 MHz - 15 MHz;

[0068] The capacitive voltage divider adopts a composite structure of ceramic capacitors and grading rings, and is configured with a temperature compensation circuit to suppress the change of dielectric loss;

[0069] The environmental monitoring module integrates a temperature and humidity sensor, a salt spray concentration sensor and a three-dimensional vibration sensor, and the surface of the sensor probe is coated with a nano-hydrophobic coating.

[0070] Specifically, the dual-redundancy Rogowski coil structure consists of two independently wound hollow Rogowski coils arranged orthogonally. Signals are output through a differential amplifier circuit for redundancy detection and common-mode noise suppression respectively. The coil skeleton is made of polytetrafluoroethylene material with low dielectric loss, and the turn spacing is optimized to reduce distributed capacitance, ensuring that the amplitude-frequency characteristic fluctuation within 15 MHz is less than ±1 dB. The capacitive voltage divider adopts a composite structure of ceramic capacitors and grading rings. Its high-voltage arm is composed of multiple chip ceramic capacitors connected in series, and the low-voltage arm is configured with precision thin-film capacitors. The grading ring adopts an annular aluminum fluid guide, and the edge field interference is eliminated through equipotential design. The temperature compensation circuit integrates a thermistor network at the output end of the voltage divider to compensate the temperature drift of the dielectric constant of the ceramic capacitor in real time, controlling the temperature drift of the voltage division ratio within ±0.5%. The temperature and humidity sensor uses a digital SHT35 chip, with a measurement range of -40°C to 125°C and 0 - 100% RH, and an accuracy of ±0.3°C / ±2% RH. The salt spray concentration sensor is based on the electrochemical principle to detect the chloride ion concentration, and the output signal is transmitted through the I2C interface. The three-dimensional vibration sensor integrates a MEMS three-axis accelerometer to monitor the mechanical vibration amplitude and frequency characteristics of the tower pole. The surface of the sensor probe is coated with a nano-hydrophobic coating, which uses polydimethylsiloxane material doped with silicon dioxide nanoparticles to prevent salt spray condensation and dust adhesion, improving the long-term reliability in harsh environments.

[0071] S13. Dynamically filter and correct the wideband transient signal according to the preset transmission line impedance matching model. The dynamic filtering and correction is to adjust the propagation speed of the wideband transient signal and compensate for the influence of the ambient temperature on the relative dielectric constant of the medium.

[0072] Step S13 includes the following:

[0073] S131. Divide the transmission line into several equal-length sections, and then determine the resistance parameter, inductance parameter, conductance parameter, and capacitance parameter of each section. The section length is determined according to the traveling wave propagation time delay.

[0074] The section length Δx determined according to the traveling wave propagation time delay includes:

[0075] Δx = v base ·Δt

[0076] Where: v base is the initial traveling wave speed, and Δt is the preset time window. The preset time is not limited.

[0077] S132. Correct the resistance parameter, inductance parameter, conductance parameter, and capacitance parameter of each section based on the temperature and humidity data collected by the environmental monitoring module.

[0078] R′ = R0·[1 + β R (T – T0)]

[0079] L′ = L0·[1 + βL (T – T0)]

[0080]

[0081] C′ = C0·[1 + β C (T – T0)]

[0082] Where: R′, L′, G′, and C′ are the resistance, inductance, conductance, and capacitance per unit length after correction, respectively; R0, L0, G0, and C0 are the per-unit-length parameters at the reference temperature T0, respectively; β R , β L and β C are the temperature coefficients of resistance, inductance, and capacitance, respectively; k H is the conductance correction factor for temperature; T is the measured value of the ambient temperature sensor; H is the relative humidity.

[0083] S133. Calculate the characteristic impedance of each section based on the corrected parameters, and determine the propagation speed of the broadband transient signal based on the characteristic impedance of each section.

[0084] 1. Calculate the reference value of the theoretical traveling wave transmission speed based on the inductance parameter and capacitance parameter, and calculate the characteristic impedance of each section based on the resistance parameter, inductance parameter, conductance parameter, and capacitance parameter;

[0085] Formula for the reference value v0:

[0086]

[0087] Characteristic impedance Z c of each section, formula:

[0088]

[0089] Where: j is the imaginary unit, describing the phase relationship; ω is the angular frequency, reflecting the signal frequency characteristics.

[0090] 2. Correct the reference value through the preset temperature and dielectric constant compensation factor to obtain the corrected propagation speed;

[0091] v T = v0·[1 + α·(T – T0)]

[0092] Where: α is the medium temperature coefficient.

[0093] 3. Calculate the target propagation speed based on the corrected propagation speed, the characteristic impedance of each section, and the preset impedance deviation of the fault point. The preset impedance deviation of the fault point is used to reflect the mismatch degree between the fault point impedance and the line characteristic impedance.

[0094] The calculation of the impedance deviation is achieved through the following steps:

[0095] Multiple monitoring points are distributed along the transmission line, and the voltage and current time-domain waveform data during fault occurrence are synchronously collected; perform fast Fourier transform (FFT) on the voltage and current waveforms of each monitoring point to extract the amplitude and phase information of the power frequency fundamental frequency components; for a certain monitoring point, calculate the ratio of the voltage fundamental frequency component to the current fundamental frequency component to characterize the equivalent impedance of this point; compare the impedance ratio of each monitoring point with the designed reference impedance of the line to obtain the single-point impedance deviation value; measure the initial positioning distance between the monitoring point and the fault point, and use the exponential decay model to perform weighted correction on the single-point impedance deviation value; sum and average the corrected deviation values of all monitoring points to obtain the final impedance deviation ΔZ:

[0096]

[0097] where: N is the total number of monitoring points, V n and I n are respectively the fundamental frequency component ratios after FFT transformation of the voltage and current time-domain waveforms of the nth monitoring point, Z ref is the designed reference impedance of the line, γ is the empirical value of the traveling wave attenuation coefficient, and x n is the initial positioning distance of the nth monitoring point from the theoretical fault point.

[0098] The calculation formula for the target propagation speed v actual is:

[0099]

[0100] where: v T is the corrected propagation speed.

[0101] S14. Use a deep learning model to perform hybrid processing on the corrected wideband transient signal and electromagnetic transient signal, extract the time-frequency characteristics of the traveling wave head and the power frequency mutation characteristics, and output the fault probability and type classification results;

[0102] Step S14 includes the following:

[0103] S141. Normalize the corrected wideband transient signal, and extract the trajectory characteristics in the dq coordinate system from the electromagnetic transient signal. The trajectory characteristics are used to describe the amplitude and phase change characteristics of the voltage and current;

[0104] The specific formula for normalizing the corrected wideband transient signal is:

[0105]

[0106] Where: μ is the signal mean, σ is the signal standard deviation, and the normalization coefficient 3 covers 99.7% of the data distribution based on the 3σ principle.

[0107] Perform Park transformation on the three-phase voltage signal and three-phase current signal in the electromagnetic transient signal. The formula is:

[0108]

[0109] Where: v d and v q are the active and reactive components in the dq coordinate system, v a and v b and v c are the instantaneous values of the three-phase voltages, i a and i b and i c are the instantaneous values of the three-phase currents, θ = wt is the phase angle of the grid voltage, w = 2π × 50Hz, cosθ, is the cosine function term used to project the three-phase signal onto the d axis, -sinθ, is the negative sine function term used to project the three-phase signal onto the q axis;

[0110] Calculate the amplitude and phase angle to generate a time-varying trajectory matrix.

[0111] S142. Construct the input architecture of a deep learning model that includes a high-frequency channel and a power-frequency channel. The high-frequency channel uses a two-dimensional convolutional neural network to process the time-frequency feature matrix of the traveling wave head, and the power-frequency channel uses a long short-term memory network to analyze the trajectory features, voltage mutation time series features, and current mutation time series features in the dq coordinate system;

[0112] High-frequency channel (CNN processes the time-frequency features of the traveling wave head), input: Time-frequency feature matrix of the traveling wave head (generated by improving the S transform):

[0113]

[0114] Where: Window function k is an adjustable bandwidth factor.

[0115] The structure of the two-dimensional convolutional neural network includes: an input layer for the input time-frequency matrix; a convolutional layer with 3 layers, a filter size of 3×3, a number of channels from 32→64→128, and a ReLU activation function; a pooling layer with 2×2 max pooling; an output layer that obtains the time-frequency feature vector after flattening.

[0116] Power-frequency channel (LSTM analyzes dq trajectory and time series features), input: dq trajectory feature matrix (size: time × 2, including v d and v q) The voltage mutation time - series feature ΔV(t)=V(t)−V(t−1); the current mutation time - series feature ΔI(t)=I(t)−I(t−1). The structure of the long - short - term memory network includes: an input layer for inputting time - series data; an LSTM layer with 2 layers, 128→64 hidden units and the activation function Tanh; an output layer of a fully - connected layer for generating a time - series feature vector.

[0117] S143. Weightedly splice the time - frequency feature vector output by the high - frequency channel and the time - series feature vector output by the power - frequency channel to generate a target feature vector;

[0118] The weighted splicing is achieved through the following formula:

[0119]

[0120] where: W1, W2 are trainable weight matrices (dimensions: 256×128, 64×128), and the initialization method is the Xavier normal distribution, The splicing operation generates a joint feature vector F joint .

[0121] S144. Calculate the fault probability and classify according to the target feature vector.

[0122] The fault probability P fault Calculation formula:

[0123] P fault =σ(W p ·F joint +b p )

[0124] where: σ is the Sigmoid function, outputting the fault probability (0 - 1), W p and b p are trainable parameters.

[0125] The structure of the classifier: The first fully - connected layer: 128 nodes, ReLU activation, dropout rate 0.3; the second fully - connected layer: 64 nodes, LeakyReLU activation; the output layer: the Softmax function generates the fault - type probability distribution P = [p 雷击 ,p 短路 ,p 接地 .

[0126] S15. Verify the fault - location result through a multi - level verification mechanism;

[0127] Step S15 includes the following:

[0128] S151. Calculate the theoretical and measured time differences of traveling wave arrival times at the monitoring nodes at both ends of the line. If the deviation meets the first threshold condition, trigger the recalculation of the fault location using redundant sensor data;

[0129] The first threshold condition is:

[0130] |Δt 实测 ―Δt 理论 | > 50μs

[0131] Based on the threshold judgment (50μs) of the traveling wave propagation time difference, quickly eliminate the positioning deviation caused by sensor noise or synchronization error, trigger the recalculation of redundant data, and improve the anti-interference ability.

[0132] S152. Calculate the Pearson correlation coefficient between the extracted power frequency impedance trajectory and the high-frequency traveling wave waveform. If the Pearson correlation coefficient meets the second threshold condition, initiate the environmental parameter review and correct the signal attenuation model according to the real-time temperature and humidity data;

[0133] The second threshold condition:

[0134] ρ < 0.85

[0135] Verify the physical correlation between the power frequency and high-frequency signals through the correlation coefficient, avoid misjudgment, and dynamically correct the signal attenuation amount during the environmental parameter review.

[0136] S153. Generate the theoretical fault propagation path through digital twin technology and perform spatio-temporal matching with the measured traveling wave arrival time series. If the positioning error meets the third threshold condition, determine that the fault location result is valid.

[0137] The third threshold condition is:

[0138]

[0139] Where: x is the theoretical position of the fault point, x i is the known position coordinate of the i-th monitoring point, v eff is the effective propagation speed of the traveling wave in the transmission line, which has been corrected by dynamic filtering, t i and t0 are the measured time of the traveling wave arriving at the i-th monitoring point and the initial time of the fault occurrence, N is the total number of monitoring points participating in the verification, and ∈ is the maximum allowable threshold of the positioning error.

[0140] Combine the theoretical path of the digital twin model with the measured data, and solve the fault location through an optimization algorithm; the error threshold covers the positioning accuracy requirements of typical overhead lines. Combine the fault propagation path generated by the digital twin model to verify the matching degree between the measured data and the theoretical path, and ensure that the positioning error is less than the preset value.

[0141] S16. Preprocess the extracted feature quantities on the edge side of the detection terminal, and transmit the processed data to the master station system through an encryption protocol.

[0142] Perform the following operations on the edge side of the detection terminal:

[0143] 1. Feature preprocessing: Standardize and reduce the dimension of the extracted time-frequency features of the traveling wave head and power frequency mutation features to reduce the data transmission volume; use a sliding window mechanism (window length 100ms, overlap rate 50%) to segment the time-series data and generate a structured feature matrix.

[0144] 2. Encrypted transmission: Encrypt the feature matrix using a lightweight encryption protocol to ensure data integrity and confidentiality; transmit the encrypted data to the master station system through a LoRa / 5G wireless communication module, and the transmission period is dynamically adjusted (1 minute per time in the steady state, and real-time transmission when a fault is triggered).

[0145] This step realizes end-to-end secure transmission under low-power conditions through a lightweight encryption protocol, and resists replay attacks and data tampering.

[0146] The above embodiments improve the accuracy of fault identification and the positioning accuracy; ensure the real-time performance of data processing and transmission security, and provide an efficient and reliable solution for the fault diagnosis of transmission lines in complex environments.

[0147] Those of ordinary skill in the art can realize that the units of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the components of each example have been generally described according to their functions in the above description. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0148] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc. In addition, the functional units in each embodiment of the present invention can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0149] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, read-only memories (ROMs), random access memories (RAMs), mobile hard disks, magnetic disks, or optical discs that can store program codes.

[0150] It can be understood that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the various embodiments of the present invention, and they should all be covered within the scope of the claims and the specification of the present invention.

Claims

1. A distributed transmission line fault diagnosis method based on transient information perception of the ground potential terminal, characterized in that, Including: Collecting broadband transient signals at the ground potential terminals of each node of the transmission line through distributed detection terminals. The broadband transient signals include traveling wave components, power frequency voltage mutation signals, and power frequency current mutation signals in the range of 0.1 MHz to 10 MHz; Synchronously obtaining electromagnetic transient signals and environmental parameters through a composite sensing unit. The composite sensing unit includes an electromagnetic induction sensor, a capacitive voltage divider, and an environmental monitoring module; Performing dynamic filtering and correction on the broadband transient signals according to a preset transmission line impedance matching model. The dynamic filtering and correction is to adjust the propagation speed of the broadband transient signals and compensate for the influence of environmental temperature on the relative dielectric constant of the medium; Using a deep learning model to perform hybrid processing on the corrected broadband transient signals and the electromagnetic transient signals, extracting the time-frequency characteristics of the traveling wave head and the power frequency mutation characteristics, and outputting the fault probability and type classification results; Verifying the fault location results through a multi-level verification mechanism; Preprocessing the extracted feature quantities on the edge side of the detection terminal and transmitting the processed data to the master station system through an encryption protocol.

2. The distributed transmission line fault diagnosis method based on transient information sensing of the ground potential terminal according to claim 1, characterized in that, The step of synchronously obtaining electromagnetic transient signals and environmental parameters through a composite sensing unit. The composite sensing unit includes an electromagnetic induction sensor, a capacitive voltage divider, and an environmental monitoring module, includes: The electromagnetic induction sensor adopts a dual-redundancy Rogowski coil structure with a frequency response range of 0.1 MHz - 15 MHz; The capacitive voltage divider adopts a composite structure of ceramic capacitors and grading rings, and is configured with a temperature compensation circuit to suppress the change of dielectric loss; The environmental monitoring module integrates a temperature and humidity sensor, a salt fog concentration sensor, and a three-dimensional vibration sensor, and the surfaces of the sensor probes are coated with a nano-hydrophobic coating.

3. The distributed transmission line fault diagnosis method based on transient information perception of the ground potential terminal according to claim 2, wherein The step of performing dynamic filtering and correction on the broadband transient signals according to a preset transmission line impedance matching model. The dynamic filtering and correction is to adjust the propagation speed of the broadband transient signals and compensate for the influence of environmental temperature on the relative dielectric constant of the medium, includes: Dividing the transmission line into several equal-length sections, and then determining the resistance parameters, inductance parameters, conductance parameters, and capacitance parameters of each section. The section length is determined according to the traveling wave propagation time delay; Correcting the resistance parameters, inductance parameters, conductance parameters, and capacitance parameters of each section based on the temperature and humidity data collected by the environmental monitoring module; Calculating the characteristic impedance of each section according to the corrected parameters, and at the same time determining the propagation speed of the broadband transient signals based on the characteristic impedance of each section.

4. The distributed transmission line fault diagnosis method based on transient information perception of the ground potential terminal according to claim 3, characterized in that, The step of correcting the resistance parameters, inductance parameters, conductance parameters, and capacitance parameters of each section based on the temperature and humidity data collected by the environmental monitoring module, includes: R′ = R0·[1 + β R (T – T0)] L′ = L0·[1 + β L (T – T0)] C′ = C0·[1 + β C (T – T0)] Where: R′, L′, G′, and C′ are the corrected resistance, inductance, conductance, and capacitance per unit length, respectively; R0, L0, G0, and C0 are the per-unit-length parameters at the reference temperature T0; β R , β L , and β C are the temperature coefficients of resistance, inductance, and capacitance, respectively; k H is the conductance correction factor for temperature; T is the measured value of the ambient temperature sensor; and H is the relative humidity.

5. The distributed transmission line fault diagnosis method based on transient information perception of the ground potential terminal according to claim 4, wherein, The step of calculating the characteristic impedance of each section according to the corrected parameters, and at the same time determining the propagation speed of the broadband transient signals based on the characteristic impedance of each section, includes: Calculating the reference value of the theoretical traveling wave transmission speed based on the inductance parameters and capacitance parameters, and calculating the characteristic impedance of each section based on the resistance parameters, inductance parameters, conductance parameters, and capacitance parameters; Correcting the reference value through a preset temperature and dielectric constant compensation factor to obtain the corrected propagation speed; Calculate the target propagation speed according to the corrected propagation speed, the characteristic impedance of each section, and a preset impedance deviation of the fault point, where the preset impedance deviation of the fault point is used to reflect the mismatch degree between the fault point impedance and the line characteristic impedance.

6. The distributed transmission line fault diagnosis method based on transient information sensing of the ground potential terminal according to claim 5, wherein The calculating the target propagation speed according to the corrected propagation speed, the characteristic impedance of each section, and a preset impedance deviation of the fault point, where the preset impedance deviation of the fault point is used to reflect the mismatch degree between the fault point impedance and the line characteristic impedance, includes: Characteristic impedance Z of each section c Calculation formula: Where: j is the imaginary unit describing the phase relationship, and ω is the angular frequency reflecting the signal frequency characteristic; The calculation formula for the preset impedance deviation ΔZ of the fault point: Where: N is the total number of monitoring points, V n and I n are respectively the fundamental frequency component ratios after FFT transformation of the voltage and current time-domain waveforms at the nth monitoring point, Z ref is the line design reference impedance, γ is the empirical value of the traveling wave attenuation coefficient, x n is the initial positioning distance of the monitoring point n from the theoretical fault point; Target propagation speed v actual Calculation formula: where: v T is the corrected propagation speed.

7. The distributed transmission line fault diagnosis method based on transient information perception of the ground potential terminal according to claim 1, characterized in that The using a deep learning model to perform hybrid processing on the corrected wideband transient signal and the electromagnetic transient signal, extract the time-frequency characteristics of the traveling wave head and the power frequency mutation characteristics, and output the fault probability and type classification result, includes: Perform normalization processing on the corrected wideband transient signal, and extract vectors from the electromagnetic transient signal to obtain the trajectory characteristics in the dq coordinate system, where the trajectory characteristics are used to describe the amplitude and phase change characteristics of voltage and current; Construct an input architecture of a deep learning model including a high-frequency channel and a power-frequency channel. The high-frequency channel uses a two-dimensional convolutional neural network to process the time-frequency characteristic matrix of the traveling wave head, and the power-frequency channel uses a long short-term memory network to analyze the trajectory characteristics, voltage mutation time series characteristics, and current mutation time series characteristics in the dq coordinate system; Perform weighted splicing on the time-frequency feature vector output by the high-frequency channel and the time series feature vector output by the power-frequency channel to generate a target feature vector; Calculate the fault probability and classify according to the target feature vector.

8. The distributed transmission line fault diagnosis method based on transient information sensing of ground potential terminal according to claim 7, characterized in that The performing normalization processing on the corrected wideband transient signal, and extracting vectors from the electromagnetic transient signal to obtain the trajectory characteristics in the dq coordinate system, where the trajectory characteristics are used to describe the amplitude and phase change characteristics of voltage and current, includes: Perform Park transformation on the three-phase voltage signal and the three-phase current signal in the electromagnetic transient signal. The formula is: Where: v d and v q are the active and reactive components in the dq coordinate system, v a , v b , and v c are the instantaneous values of the three-phase voltages, i a , i b , and i c are the instantaneous values of the three-phase currents, θ = wt is the rotation angle, w = 2π × 50 Hz, cosθ, is the cosine function term for projecting the three-phase signals onto the d-axis, -sinθ, is the negative sine function term for projecting the three-phase signals onto the q-axis; Calculate the amplitude and phase angle to generate a time-varying trajectory matrix.

9. The distributed transmission line fault diagnosis method based on transient information perception of ground potential terminal according to claim 7, characterized in that, The constructing an input architecture of a deep learning model including a high-frequency channel and a power-frequency channel. The high-frequency channel uses a two-dimensional convolutional neural network to process the time-frequency characteristic matrix of the traveling wave head, and the power-frequency channel uses a long short-term memory network to analyze the trajectory characteristics, voltage mutation time series characteristics, and current mutation time series characteristics in the dq coordinate system, includes: The structure of the two-dimensional convolutional neural network includes: an input layer for inputting the time-frequency matrix; a convolutional layer with 3 layers, a filter size of 3×3, the number of channels from 32 to 64 to 128, and a ReLU activation function; a pooling layer with 2×2 max pooling; an output layer that obtains a time-frequency feature vector after Flattening; The structure of the long short-term memory network includes: an input layer for inputting time series data; an LSTM layer with 2 layers, 128→64 hidden units, and a Tanh activation function; an output layer of a fully connected layer that generates a time series feature vector.

10. The distributed transmission line fault diagnosis method based on transient information sensing of the ground potential terminal according to claim 1, characterized in that, The verifying the fault location result through a multi-level verification mechanism includes: Calculate the theoretical value and the measured value of the time difference of arrival of traveling waves at the monitoring nodes at both ends of the line. If the deviation meets the first threshold condition, trigger the recalculation of the fault location with redundant sensor data; Calculate the Pearson correlation coefficient between the extracted power frequency impedance trajectory and the high-frequency traveling wave waveform. If the Pearson correlation coefficient meets the second threshold condition, initiate the environmental parameter review and correct the signal attenuation model according to the real-time temperature and humidity data; Generate a theoretical fault propagation path through digital twin technology and perform spatio-temporal matching with the measured traveling wave arrival time series. If the positioning error meets the third threshold condition, determine that the fault location result is valid.

Images