Power transmission line fault identification method and device based on artificial intelligence

Abstract

The invention discloses a power transmission line fault identification method and device based on artificial intelligence, and relates to the technical field of power transmission line fault identification. Determining line data of the target area, performing data acquisition according to the line data to obtain an original traveling wave signal, and filtering the original traveling wave signal to obtain a time domain signal; performing adaptive variational mode decomposition on the time domain signal to obtain a mode component, and performing calculation according to the mode component to obtain a fault point; and determining an inspection scheme according to the fault point, executing the inspection scheme to obtain a fault point image, and performing alarm judgment on the fault point image through artificial intelligence. Line data and traveling wave signals are accurately collected, noise is filtered by adopting an advanced filtering technology, and the signal quality is improved; adaptive variational mode decomposition is utilized to accurately extract a mode component, and the fault point positioning precision is enhanced; faults are quickly and accurately judged by combining artificial intelligence image recognition, automatic operation and maintenance are realized, and the operation and maintenance efficiency of the power transmission line and the stability of the power system are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of transmission line fault identification, and in particular relates to a transmission line fault identification method and device based on artificial intelligence. Background Art

[0002] With the development of power systems, transmission lines, as a critical component, have become increasingly important for grid stability. However, transmission lines are susceptible to environmental factors, equipment aging, and external damage, leading to faults such as grounding, short circuits, and disconnections. Failure to address these faults promptly and accurately can lead to power outages or even system crashes. Traditional fault identification methods rely on relay protection devices and expert systems. While these methods offer a certain degree of automation, they are highly dependent on fault patterns, making them difficult to adapt to complex fault types. Furthermore, these methods rely on manual experience, resulting in low efficiency and a high risk of misjudgment or omission.

[0003] Patent No.: CN115201632A, discloses a multi-terminal DC transmission line fault identification method, extracting fault characteristics: obtaining voltage traveling wave information of the multi-terminal DC transmission line; extracting characteristics of the voltage traveling wave information to obtain fault characteristics; the fault characteristics include the positive and negative voltage change ratio coefficient, the ratio of the busbar high-frequency energy to the low-frequency energy, and the ratio of the line high-frequency energy to the low-frequency energy; constructing a fault detection model; initializing the hyperparameters of the fault detection model; training the fault detection model to update the hyperparameters; inputting the fault characteristics into the trained fault detection model, and the fault detection model outputs the fault type. The present invention uses wavelet transform to convert the instantaneous voltage transient characteristics of the busbar and the line into wavelet energy proportion and positive and negative energy change ratio, constructing fault characteristics under frequency domain characteristics, which are less affected by high-resistance faults, thereby improving the fault tolerance transition resistance capability.

[0004] Although existing technologies have solved some problems, some problems still exist. For example, when a transmission line fails, the traveling wave will be refracted and reflected at the fault point during propagation. The collected fault current signal may be destroyed by noise, and there will be signal interference in adjacent lines, resulting in a loss of positioning accuracy. Summary of the Invention

[0005] The purpose of the present invention is to solve the problem that when a transmission line fails, the traveling wave will be refracted and reflected at the fault point during propagation, the collected fault current signal may be destroyed by noise, and there will be signal interference in adjacent lines, resulting in a lack of positioning accuracy. A method and device for identifying transmission line faults based on artificial intelligence are proposed.

[0006] In a first aspect of the present invention, a method for identifying power transmission line faults based on artificial intelligence is first proposed, the method comprising: Determine line data for a target area, collect data based on the line data to obtain an original traveling wave signal, and determine fault reflection parameters and fault refraction parameters based on the line data; the line data includes: the number of transmission lines and line relationships; the fault reflection parameters are the reflection intensity and polarity of the traveling wave at the busbar or fault point; and the fault refraction parameters are the ratio and polarity of the traveling wave refracted from the fault line; determining a ratio matrix according to the fault reflection parameter and the fault refraction parameter, and filtering the original traveling wave signal according to the ratio matrix to obtain a time domain signal; Decomposing the time domain signal by adaptive variational modal decomposition to obtain modal components, and calculating the fault point based on the modal components; An inspection plan is determined according to the fault point, an image of the fault point is obtained by executing the inspection plan, and an alarm judgment is performed on the image of the fault point through artificial intelligence.

[0007] Optionally, determining the fault reflection parameter and the fault refraction parameter according to the line data includes: An equivalent resistance and a fault equivalent resistance are calculated according to the number of lines, a fault reflection parameter is calculated according to the equivalent resistance and the fault equivalent resistance, and a fault refraction parameter is calculated according to the fault reflection parameter; Determining a ratio matrix according to the fault reflection parameter and the fault refraction parameter includes: Decomposing and calculating the original traveling wave according to the fault refraction parameter and the fault reflection parameter to obtain a composite traveling wave, and subjecting the original traveling wave and the composite traveling wave to S-transformation to obtain a transformed original traveling wave and a transformed composite traveling wave respectively; Filtering the original traveling wave signal according to the ratio matrix to obtain a time domain signal, comprising: A ratio matrix is ​​calculated based on the transformed original traveling wave and the transformed composite traveling wave, the original traveling wave is fault-screened based on the ratio matrix to obtain a fault wave frequency band, and the fault wave frequency band is converted from a time-frequency matrix through an inverse S transform to obtain a time domain signal.

[0008] Optionally, performing adaptive variational mode decomposition on the time domain signal to obtain modal components, and calculating the fault point according to the modal components includes: Performing phase mode transformation on the time domain signal through a phase mode transformation matrix to obtain a current traveling wave mode component, and decomposing the current traveling wave mode component through an improved VMD algorithm to obtain a modal component; Performing Hilbert transform on the modal component to obtain the instantaneous frequency, and extracting the time nodes of the fault current traveling wave according to the instantaneous frequency; the time nodes include t1 and t2; The fault location is calculated by performing fault location calculation on the time node using a double-terminal traveling wave ranging formula to obtain the fault position.

[0009] Optionally, the current traveling wave mode components are decomposed using an improved VMD algorithm to obtain modal components including: The parameter combination of the VMD algorithm is used as a population individual to initialize the preset algorithm parameters; the parameter combination includes: minimum envelope entropy, modal decomposition layer number K and penalty factor; the parameters include the value range of the population size, the maximum number of iterations, the number of decomposition modal layers, and the value range of the penalty factor; Calculating the fitness of individuals in the population, determining the optimal individual according to the fitness, determining the optimal parameter combination corresponding to the optimal individual, and updating the VMD algorithm according to the optimal parameter combination to obtain an improved VMD algorithm; Execute the improved VMD algorithm to decompose and obtain multiple modal components, determine the envelope entropy of each modal component through the objective function, and if the envelope entropy is less than the envelope entropy threshold, update the envelope entropy threshold; otherwise, determine the number of iterations; Objective function: , in, is the envelope entropy, N is the number of signal sampling points, is the envelope amplitude sequence of the signal after Hilbert demodulation, for The normalized envelope sequence of ; If the current number of iterations is less than the maximum number of iterations, the positions of the individuals in the population are updated; otherwise, if the current number of iterations is greater than or equal to the maximum number of iterations, the optimal parameter combination is output and the VMD algorithm is updated to obtain an improved VMD algorithm.

[0010] Optionally, performing alarm determination on the fault point image by artificial intelligence includes: If the number of fault areas in the fault point image is greater than or equal to the fault threshold, the transmission line is determined to be at high risk; If the number of fault areas in the fault point image is less than the fault threshold, the transmission line is determined to be of low risk.

[0011] In a second aspect of the present invention, a power transmission line fault identification device based on artificial intelligence is proposed, comprising: a data preprocessing module, a fault point determination module, and an alarm judgment module: The data preprocessing module is configured to determine line data in a target area, acquire raw traveling wave signals based on the line data, and filter the raw traveling wave signals to obtain time domain signals. The line data includes: the number of transmission lines and line relationships; the fault reflection parameter is the reflection intensity and polarity of the traveling wave at the busbar or fault point; and the fault refraction parameter is the ratio and polarity of the traveling wave refracted from the fault line. The fault point determination module is configured to obtain modal components by performing adaptive variational modal decomposition on the time domain signal, and calculate the fault point based on the modal components; The alarm judgment module is used to determine an inspection plan based on the fault point, execute the inspection plan to obtain a fault point image, and perform alarm judgment on the fault point image through artificial intelligence.

[0012] Optionally, the data preprocessing module includes: a parameter calculation module, a traveling wave transformation module and a signal conversion module: The parameter calculation module is configured to calculate an equivalent resistance and a fault equivalent resistance according to the number of lines, calculate a fault reflection parameter according to the equivalent resistance and the fault equivalent resistance, and calculate a fault refraction parameter according to the fault reflection parameter; The traveling wave transformation module is used to decompose and calculate the original traveling wave according to the fault refraction parameter and the fault reflection parameter to obtain a composite traveling wave, and respectively perform S-transformation on the original traveling wave and the composite traveling wave to obtain a transformed original traveling wave and a transformed composite traveling wave; The signal conversion module is used to calculate a ratio matrix based on the transformed original traveling wave and the transformed composite traveling wave, perform fault screening on the original traveling wave according to the ratio matrix to obtain a fault wave frequency band, and convert the fault wave frequency band from a time-frequency matrix through an inverse S transform to obtain a time domain signal.

[0013] Optionally, the fault point determination module includes: a modal decomposition module, a time node module and a fault location module: The modal decomposition module is used to perform phase mode transformation on the time domain signal through a phase mode transformation matrix to obtain a current traveling wave mode component, and decompose the current traveling wave mode component through an improved VMD algorithm to obtain a modal component; The time node module is used to perform Hilbert transform on the modal component to obtain the instantaneous frequency, and extract the time node of the fault current traveling wave according to the instantaneous frequency; the time node includes t1 and t2; The fault location module is used to perform fault location calculation on the time node using a double-terminal traveling wave ranging formula to obtain the fault position.

[0014] Optionally, the modal decomposition module includes: a parameter initialization module, a fitness calculation module, an envelope entropy calculation module and an iterative update module The parameter initialization module is used to initialize the preset algorithm parameters using the parameter combination of the VMD algorithm as a population individual; the parameter combination includes: minimum envelope entropy, modal decomposition layer number K and penalty factor; the parameters include the value range of the population size, the maximum number of iterations, the decomposition modal layer number, and the value range of the penalty factor; The fitness calculation module is used to calculate the fitness of individuals in the population, determine the optimal individual according to the fitness, determine the optimal parameter combination corresponding to the optimal individual, and update the VMD algorithm according to the optimal parameter combination to obtain an improved VMD algorithm; The envelope entropy calculation module is used to execute the improved VMD algorithm to decompose and obtain multiple modal components, determine the envelope entropy of each modal component through the objective function, and update the envelope entropy threshold if the envelope entropy is less than the envelope entropy threshold; otherwise, determine the number of iterations; Objective function: , in, is the envelope entropy, N is the number of signal sampling points, is the envelope amplitude sequence of the signal after Hilbert demodulation, for The normalized envelope sequence of ; The iterative update module is used to update the position of the population individuals if the current number of iterations is less than the maximum number of iterations; otherwise, if the current number of iterations is greater than or equal to the maximum number of iterations, output the optimal parameter combination and update the VMD algorithm to obtain an improved VMD algorithm.

[0015] Optionally, the alarm judgment module includes: The first judgment module is configured to judge the transmission line as high risk if the number of fault areas in the fault point image is greater than or equal to a fault threshold; The second judgment module is configured to judge the transmission line as having low risk if the number of fault areas in the fault point image is less than a fault threshold.

[0016] Beneficial effects of the present invention: This invention proposes an artificial intelligence-based method for identifying transmission line faults. The method determines the line data in the target area, collects data based on the line data to obtain a raw traveling wave signal, and filters the raw traveling wave signal to obtain a time-domain signal. The time-domain signal is then decomposed using adaptive variational modal analysis to obtain modal components, and the fault point is calculated based on the modal components. An inspection plan is determined based on the fault point, and the inspection plan is executed to obtain an image of the fault point. This image is then used to perform an alarm determination based on the fault point. The method accurately collects line data and traveling wave signals, employs advanced filtering techniques to filter noise and improve signal quality. Adaptive variational modal decomposition is used to precisely extract modal components, enhancing the accuracy of fault location. Combined with artificial intelligence image recognition, the method quickly and accurately determines faults, enabling automated operation and maintenance, and improving transmission line operation and maintenance efficiency and power system stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be further described below with reference to the accompanying drawings.

[0018] Figure 1 A flowchart of a method for identifying power transmission line faults based on artificial intelligence provided by an embodiment of the present invention; Figure 2 A schematic structural diagram of a power transmission line fault identification device based on artificial intelligence provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention, in conjunction with the accompanying drawings. Obviously, the described embodiments represent only a portion of the embodiments of the present invention, not all of them. The term "and / or" herein simply describes an association relationship between associated objects, indicating that three possible relationships exist. For example, "A" and "B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, references to "first," "second," and so on in the present invention are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined as "first" or "second" may explicitly or implicitly include at least one of these features. Furthermore, the technical solutions of the various embodiments may be combined, but only if they are achievable by a person of ordinary skill in the art. If a combination of technical solutions contradicts or is unachievable, such combination shall be deemed non-existent and outside the scope of protection claimed by the present invention.

[0020] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0021] The embodiment of the present invention provides a method for identifying power transmission line faults based on artificial intelligence. Figure 1 , Figure 1 A flowchart of a method for identifying power line faults based on artificial intelligence is provided in an embodiment of the present invention. The method comprises the following steps: S101, determining line data in a target area, acquiring original traveling wave signals based on the line data, and determining fault reflection parameters and fault refraction parameters based on the line data; S102, determining a ratio matrix according to the fault reflection parameter and the fault refraction parameter, and filtering the original traveling wave signal according to the ratio matrix to obtain a time domain signal; S103, decomposing the time domain signal by adaptive variational modal decomposition to obtain modal components, and calculating the fault point based on the modal components; S104, determining an inspection plan based on the fault point, executing the inspection plan to obtain an image of the fault point, and performing an alarm judgment on the image of the fault point through artificial intelligence.

[0022] Line data includes: the number of transmission lines and line relationships; fault reflection parameters are the reflection intensity and polarity of the traveling wave generated at the busbar or fault point; fault refraction parameters are the proportion and polarity of the traveling wave reflected from the fault line; An artificial intelligence-based transmission line fault identification method provided by an embodiment of the present invention accurately collects line data and traveling wave signals, uses advanced filtering technology to filter noise and improve signal quality; uses adaptive variational modal decomposition to accurately extract modal components and enhance the accuracy of fault point location; combines artificial intelligence image recognition to quickly and accurately determine faults, realize automated operation and maintenance, and improve the operation and maintenance efficiency of transmission lines and the stability of the power system.

[0023] In one implementation, when a fault occurs on a transmission line, a traveling wave signal is generated at the fault point. This traveling wave signal propagates along the line and is recorded by detection devices at both ends. High-precision traveling wave detection devices, such as traveling wave sensors or current transformers, are installed at both ends of the transmission line. These sensors can record the traveling wave signal in real time during the fault and transmit it to a data processing system. The fault reflection parameter describes the reflection intensity and polarity of the traveling wave caused by impedance mismatch at the busbar or fault point (function: a negative reflection coefficient causes the far-end reflected waves of adjacent lines to cancel each other out when superimposed, suppressing interference). The fault refraction parameter describes the ratio and polarity of the traveling wave reflected from the fault line to adjacent lines (function: enhancing the amplitude of the fault traveling wave and improving the signal-to-noise ratio).

[0024] In one implementation, line data is basic information describing the physical structure and electrical characteristics of the target area's power grid, including parameters such as the number of transmission lines, connection relationships (such as busbar topology, parallel / series structure), line type (cable or overhead line), length, wave impedance, and rated voltage, providing a physical basis for fault detection. The original traveling wave signal is a transient current / voltage signal captured by high-precision detection devices (such as traveling wave sensors and current transformers) at both ends of the transmission line at the moment of a fault. It has high frequency (kilohertz to megahertz) and near-light-speed propagation characteristics. Line data determines the deployment location of the detection device and signal processing parameters (such as calculating equivalent impedance and reflection coefficient through wave impedance), supporting the construction of composite traveling waves and the optimization of virtual mode decomposition (VMD). The line type and length directly affect the propagation characteristics of the traveling wave, providing a basis for improving the ranging formula, thereby collaboratively achieving precision and intelligence in the entire process from signal acquisition, noise suppression, to fault location.

[0025] In one implementation, determining the line data for the target area and collecting raw traveling wave signals can provide accurate basic data for subsequent fault detection. Line data includes the number of transmission lines and their relationships. This information provides a comprehensive understanding of the target area's power grid structure, enabling more accurate collection and processing of traveling wave signals. Filtering the raw traveling wave signals to obtain time-domain signals effectively removes noise and interference, extracting clearer and more accurate signal features. This process provides a high-quality data foundation for subsequent fault detection and location, improving the accuracy and reliability of fault identification.

[0026] In one implementation, adaptive variational modal decomposition (AVM) is used to decompose time-domain signals into modal components. The fault point is then calculated based on these modal components, effectively decomposing and extracting features from complex signals. AVM automatically adjusts decomposition parameters based on signal characteristics, better adapting to varying fault signal characteristics. By analyzing the modal components, fault characteristics can be more accurately identified, improving fault location accuracy. This process not only improves fault detection accuracy but also enhances the system's adaptability to different fault types.

[0027] In one implementation, YOLO (You Only Look Once) is a deep learning-based object detection algorithm, a computer vision technology within the field of artificial intelligence. It uses a convolutional neural network (CNN) to directly map image pixels to bounding box coordinates and class probabilities for fast and accurate object detection. The YOLO algorithm's core strengths lie in its high detection speed and accuracy. It processes image data in real time and identifies objects and their locations within the image. YOLO can detect multiple objects simultaneously, making it particularly useful for complex scenarios such as power transmission lines. Within an identified fault area, there may be multiple fault points or types of faults. YOLO can identify and classify all fault points simultaneously. For example, it can simultaneously detect broken insulators, worn conductors, and damaged towers, providing maintenance personnel with comprehensive fault information to avoid missing any potential issues. The YOLO algorithm can be adjusted and optimized for different application scenarios. For example, the network structure or training data can be adjusted to make it more suitable for specific types of transmission line fault detection; using the YOLO algorithm to identify transmission line faults can achieve fast, accurate and efficient fault detection, significantly improve the inspection efficiency and reliability of transmission lines, reduce manual intervention, and maintain high detection accuracy in complex environments.

[0028] In one implementation, optimized signal filtering technology further improves the quality of signal processing, allowing fault characteristics to be extracted more clearly. Improved fault location technology enhances the ability to process complex fault signals and achieves higher-precision fault location. The optimization process of parameter adjustment improves the adaptability and robustness of the algorithm, enabling it to better respond to fault detection needs under different working conditions and enhance the performance of the transmission line fault identification system.

[0029] In one embodiment, determining the fault reflection parameter and the fault refraction parameter according to the line data includes: The equivalent resistance and the fault equivalent resistance are calculated based on the number of lines, the fault reflection parameter is calculated based on the equivalent resistance and the fault equivalent resistance, and the fault refraction parameter is calculated based on the fault reflection parameter; A ratio matrix is ​​determined according to the fault reflection parameter and the fault refraction parameter, including: The original traveling wave is decomposed and calculated according to the fault refraction parameter and the fault reflection parameter to obtain a composite traveling wave, and the original traveling wave and the composite traveling wave are respectively subjected to S-transformation to obtain a transformed original traveling wave and a transformed composite traveling wave; The original traveling wave signal is filtered according to the ratio matrix to obtain the time domain signal, including: The ratio matrix is ​​calculated based on the transformation of the original traveling wave and the transformation of the composite traveling wave. The original traveling wave is screened for faults based on the ratio matrix to obtain the fault wave frequency band. The fault wave frequency band is converted from the time-frequency matrix to obtain the time domain signal through the inverse S transform.

[0030] In one implementation, the number of lines in the target area power grid is determined. For example, the target area has N lines, and the resistance of each line is R c In overhead transmission lines, under the conditions of the same voltage level and consistent parameters, the line resistance will not be very different. The resistance of the conductor (line) after parallel connection can be replaced by an equivalent resistance, that is, the resistance is the same. The equivalent resistance can be simplified to; R1=R c / N, the equivalent resistance during a fault (fault equivalent resistance) is R2=R c ; The fault reflection parameter GF is calculated based on the equivalent resistance and the fault equivalent resistance: GF=(R1-R2) / (R1+R2)=(R c / NR c ) / (R c / N+R c )=(1-N) / (1+N); The fault refraction parameter GZ is calculated based on the fault reflection parameter: GZ=1+GF=1+(1-N) / (1+N)=2 / (1+N).

[0031] In one implementation, the original traveling wave is decomposed and calculated according to the fault refraction parameter and the fault reflection parameter to obtain a composite traveling wave: ,in is a composite traveling wave, is the original traveling wave (representing the current traveling wave signal of the nth adjacent fault-free line); a single signal contains the fault traveling wave refraction component (the current wave reflected from the fault traveling wave from the fault line to the busbar and refracted to the nth adjacent line is recorded as tw1) and the far-end reflected wave component (the current wave reflected from the far end of the adjacent line (such as the end is open or grounded) and propagated to the busbar is recorded as tw2). ; Superimpose the signals of N adjacent lines, , where C tw is a composite traveling wave, tw1 is the fault refracted wave, tw2 is the far-end reflected wave, and since GZ>0, all fault refracted waves are superimposed in phase, and the amplitude is enhanced to a single line. times, that is , since GF < 0, the polarities of the reflected waves at the far ends of each line are opposite, and they cancel each other out after superposition, that is, , so the composite traveling wave , the refraction parameters of the GZ fault cause the polarity of the far-end reflected wave to be reversed, which will naturally cancel out after summation.

[0032] In one implementation, the S transform is used to convert the current traveling wave signal from the time domain to the time-frequency domain to analyze the signal's energy distribution at different times and frequencies. By performing the S transform on both the original traveling wave (single-line signal) and the composite traveling wave (multiple-line superimposed signal), corresponding time-frequency matrices are generated. These matrices are then used to filter out the fault's dominant frequency band (regions with amplitude ≥ 1) through ratio calculation (ratio matrix), thereby distinguishing the fault traveling wave from noise or far-end reflected waves. The S transform's high-frequency resolution and time-frequency localization characteristics make the fault signal's sudden changes (such as the traveling wave head) clearly visible in the time-frequency matrix while preserving phase information, providing the basis for subsequent inverse S transforms to reconstruct a pure time-domain signal. This process effectively improves the signal-to-noise ratio of the fault traveling wave, ensuring that modal decomposition (such as VMD) can accurately extract the arrival time of the traveling wave.

[0033] In one embodiment, the time domain signal is decomposed by adaptive variational modal decomposition to obtain modal components, and the fault point is calculated based on the modal components, including: The phase mode transformation matrix is ​​used to transform the time domain signal into a phase mode to obtain the current traveling wave mode component, and the current traveling wave mode component is decomposed into the modal component using the improved VMD algorithm; Perform Hilbert transform on the modal components to obtain the instantaneous frequency, and extract the time nodes of the fault current traveling wave based on the instantaneous frequency; the time nodes include t1 and t2; The fault location is calculated by double-terminal traveling wave ranging formula at the time node to obtain the fault location.

[0034] In one implementation, phase-mode transformation of modal components using a phase-mode transformation matrix effectively separates electromagnetic coupling in transmission lines, thereby obtaining current traveling wave modal components. This transformation method eliminates the interference of interphase coupling on signal analysis, making subsequent signal processing more accurate and efficient. During fault detection, independent modal components are extracted from the current traveling wave signal, allowing for clearer observation of fault characteristics, providing a purer signal foundation for subsequent fault location and improving the accuracy and reliability of fault detection.

[0035] In one implementation, the current traveling wave mode components are decomposed using an improved VMD (Variational Mode Decomposition) algorithm to obtain more accurate modal components. By optimizing parameter selection, such as the number of modal decomposition levels and the penalty factor, the improved VMD algorithm can better adapt to different signal characteristics, thereby improving the quality of signal decomposition. In fault location, this means that more representative and accurate modal components can be extracted from complex traveling wave signals, providing higher-quality data for subsequent fault feature extraction. For example, the improved VMD algorithm can more effectively separate high-frequency and low-frequency components in the fault signal, thereby more accurately reflecting the fault characteristics and improving the accuracy of fault location.

[0036] In one implementation, the instantaneous frequency of the signal can be obtained by performing a Hilbert transform on the modal components. The instantaneous frequency reflects the frequency change of the signal in the time domain and is an important tool for nonlinear and non-stationary signal analysis. Through the Hilbert transform, the time nodes of the fault current traveling wave, such as t1 and t2, can be extracted from the modal components. These time nodes correspond to the initial arrival time and reflection arrival time of the fault traveling wave and are key parameters for fault location. In fault location, the application of the Hilbert transform enables the system to accurately capture the characteristic time points of the fault traveling wave, thereby providing accurate time data for the two-terminal traveling wave ranging formula, improving the accuracy and reliability of fault location.

[0037] In one implementation, a two-terminal traveling wave ranging formula is used to calculate fault location at time nodes, thereby determining the fault location. The two-terminal traveling wave ranging formula uses the propagation time difference of the fault traveling wave at both ends of the transmission line to calculate the fault point's location. This method does not rely on precise measurement of the traveling wave velocity, thus reducing positioning errors caused by traveling wave velocity variations or measurement errors. This means that fault location can be determined more accurately, maintaining high positioning accuracy even in complex transmission line environments. For example, on long-distance transmission lines, the two-terminal traveling wave ranging formula can effectively cope with line parameter variations and environmental interference, providing reliable fault location results and significantly improving the efficiency and reliability of transmission line fault handling. A time node refers to the time at which the fault traveling wave is detected at both ends of the transmission line. t1 typically represents the time it takes for the fault traveling wave to propagate from the fault point to one end of the line, while t2 represents the time it takes for the fault traveling wave to propagate from the fault point to the other end of the line. These two time nodes are key parameters in the two-terminal traveling wave ranging formula. They can be used to calculate the distance from the fault point to both ends of the line, thereby determining the specific location of the fault.

[0038] In one implementation, the double-terminal traveling wave fault location formula is as follows: , where d m is the horizontal distance between the fault point and the M end of the line, t2 is the absolute time when the traveling wave head reaches the M end of the line (detection point), t1 is the absolute time when the traveling wave head reaches the N end of the line (detection point), t0 is the initial absolute time of the fault occurrence (recorded by the protection device), L mn The total horizontal length from end M to end N of the line.

[0039] In one embodiment, decomposing the current traveling wave mode component by using the improved VMD algorithm to obtain the modal components includes: The parameter combination of the VMD algorithm is used as a population individual to initialize the preset algorithm parameters; the parameter combination includes: minimum envelope entropy, modal decomposition layer number K and penalty factor; the parameters include population size, maximum number of iterations, value range of decomposition modal layers, and value range of penalty factor; Calculate the fitness of individuals in the population, determine the optimal individual based on the fitness, determine the optimal parameter combination corresponding to the optimal individual, and update the VMD algorithm based on the optimal parameter combination to obtain an improved VMD algorithm; Execute the improved VMD algorithm to decompose and obtain multiple modal components, determine the envelope entropy of each modal component through the objective function, and if the envelope entropy is less than the envelope entropy threshold, update the envelope entropy threshold; otherwise, determine the number of iterations; Objective function: , in, is the envelope entropy, N is the number of signal sampling points, is the envelope amplitude sequence of the signal after Hilbert demodulation, for The normalized envelope sequence of ; If the current number of iterations is less than the maximum number of iterations, the positions of the individuals in the population are updated; otherwise, if the current number of iterations is greater than or equal to the maximum number of iterations, the optimal parameter combination is output and the VMD algorithm is updated to obtain an improved VMD algorithm.

[0040] In one implementation, the key parameters of the VMD algorithm (modal decomposition level K, penalty factor) are used as optimization variables, combined with the global search capability of the preset algorithm, to dynamically search for the optimal parameter combination corresponding to the minimum envelope entropy. This step solves the limitation of traditional VMD relying on manual experience to set parameters, ensuring that the modal decomposition level and penalty factor adaptively match the time-frequency characteristics of different fault signals, avoiding modal aliasing or under-decomposition problems; preset algorithms include: genetic algorithm (GA), particle swarm optimization (PSO), gray wolf optimizer (GWO), ant colony algorithm (ACO), differential evolution (DE), simulated annealing (SA); objective function, E p It is used to measure the sparsity of signal energy distribution: the smaller the entropy value, the more concentrated the signal energy is at a few points (such as the mutation point of the traveling wave head), indicating that the signal purity is high; the larger the entropy value, the more dispersed the signal energy is (more noise interference); e j is the ratio of the envelope amplitude at point j to the total envelope energy; a(j) reflects the instantaneous energy intensity of the signal in the time domain. For IMF components (such as IMF1), a(j) will show a significant mutation at the traveling wave head.

[0041] In one implementation, the VMD algorithm is updated based on the determined optimal parameter combination, resulting in an improved VMD algorithm. This improvement enables the VMD algorithm to more effectively decompose signals and extract more representative and accurate modal components. The envelope entropy of each modal component is determined through an objective function and dynamically updated based on a comparison of the envelope entropy with a threshold. This further optimizes the selection of modal components and improves the quality and accuracy of signal decomposition. This dynamic adjustment mechanism better adapts to the characteristics of different signals and enhances the robustness of the algorithm.

[0042] In one implementation, during the iteration process, the relationship between the current number of iterations and the maximum number of iterations is determined to determine whether to continue optimizing or output the optimal parameter combination. When the current number of iterations is less than the maximum number of iterations, the positions of the individuals in the population are updated, and the search for a better parameter combination continues; when the maximum number of iterations is reached, the optimal parameter combination is output. This iterative optimization mechanism ensures that the algorithm can approach the global optimal solution as closely as possible within limited computing resources, improving the algorithm's convergence speed and optimization effect. By setting the maximum number of iterations, the algorithm is prevented from falling into an infinite loop, ensuring the efficiency and feasibility of the calculation process; the positions of the individuals in the population are updated: Update formula 1: , Update formula 2: , Update formula 3: , Update formula 4: , Update formula 5: ,

[0043] Among them, P i The prey location selected by the i-th individual in the population, X k is the position of the kth individual randomly selected from the population (i≠k), is the new position of the i-th individual in the j-th dimension, The current position of the i-th individual in the j-th dimension, is the position of the prey (kth individual) in the jth dimension, r is a random number in the interval [1,2] (used to control the moving step length), I is a random integer (1 or 2, simulating the randomness of the attack intensity), is the fitness value of the prey (envelope entropy), is the fitness value of the current individual, is the updated individual position of P1, is the updated fitness value of P1, is the new position of the i-th individual in the j-th dimension, R is the search radius (decreasing with the number of iterations ), r: a random number in the interval [0,1] (controls the direction of movement), is the updated individual position of P2, is the updated fitness value of P2.

[0044] In one implementation, the improved VMD algorithm, through the aforementioned optimization process, can more accurately decompose fault signals and extract fault characteristics. In transmission line fault location, this means more precise identification of the fault location and type, reducing the likelihood of misjudgments and missed detections. For example, in complex transmission line environments, the improved VMD algorithm can better handle noise and interference, accurately extracting the characteristics of fault traveling wave signals, and thus achieving highly accurate fault location.

[0045] In one embodiment, using artificial intelligence to determine an alarm on a fault point image includes: If the number of fault areas in the fault point image is ≥ the fault threshold, the transmission line is judged as high risk; If the number of fault areas in the fault point image is less than the fault threshold, the transmission line is judged to be of low risk.

[0046] In one implementation, when an artificial intelligence algorithm determines that the number of faulty areas on a transmission line reaches or exceeds the fault threshold and is judged to be high-risk, emergency measures should be taken immediately to ensure the safe and stable operation of the power system. The system will automatically trigger a high-risk alarm and transmit detailed information (including the fault location, type, and severity) to the operation and maintenance center in real time, notifying maintenance personnel to quickly rush to the site for emergency repairs. The intelligent dispatching system will be used to adjust the grid operation mode, such as reallocating power loads and switching to backup lines, to reduce the impact of the fault on grid operation and avoid larger-scale power outages. For high-risk areas, drones or robots can also be deployed for real-time monitoring, continuously tracking the development of the fault and providing real-time data support for maintenance decisions. The technical effect of this process is to significantly improve the speed of fault response and processing efficiency, minimize the impact of faults on grid operation and user power supply, and enhance the reliability and risk resistance of the power system.

[0047] In one implementation, if an artificial intelligence algorithm determines that the number of faulty areas on a transmission line is below the fault threshold and is judged to be low-risk, while the current fault situation is relatively controllable, a series of measures must still be taken to ensure the long-term stable operation of the line. The system records the fault information and marks it as low-risk, regularly reminding operations and maintenance personnel to conduct routine inspections and maintenance in these areas to promptly identify potential problems and repair them to prevent further deterioration. Historical data and fault development trends can be combined to conduct risk assessments and predictions for low-risk areas, allowing for pre-planned maintenance plans, rational allocation of repair resources, and optimized operations and maintenance strategies. For low-risk areas, monitoring frequency can be appropriately reduced, while maintaining a certain level of intensity to ensure that any changes in fault conditions are promptly detected and addressed. The technical effect of this process is to achieve refined management of transmission lines and rational allocation of resources, improve the targetedness and efficiency of operations and maintenance, reduce operations and maintenance costs, while ensuring the safe and stable operation of the power system and extending the service life of transmission lines.

[0048] Based on the same inventive concept, the embodiment of the present invention also provides a power transmission line fault identification device based on artificial intelligence. Figure 2 , Figure 2 A schematic diagram of the structure of a power transmission line fault identification device based on artificial intelligence provided by an embodiment of the present invention includes: a data preprocessing module, a signal filtering module, a fault point determination module and an alarm judgment module: The data preprocessing module is used to determine the line data in the target area, collect data based on the line data to obtain the original traveling wave signal, and determine the fault reflection parameter and fault refraction parameter based on the line data. The line data includes: the number of transmission lines and the line relationship; the fault reflection parameter is the reflection intensity and polarity generated by the traveling wave at the busbar or fault point; the fault refraction parameter is the ratio and polarity of the traveling wave refracted from the fault line; A signal filtering module is used to determine a ratio matrix according to the fault reflection parameter and the fault refraction parameter, and filter the original traveling wave signal according to the ratio matrix to obtain a time domain signal; The fault point determination module is used to obtain modal components by adaptive variational mode decomposition of the time domain signal and calculate the fault point based on the modal components; The alarm judgment module is used to determine the inspection plan based on the fault point, execute the inspection plan to obtain the fault point image, and perform alarm judgment on the fault point image through artificial intelligence.

[0049] An artificial intelligence-based transmission line fault identification device provided by an embodiment of the present invention accurately collects line data and traveling wave signals, uses advanced filtering technology to filter noise and improve signal quality; uses adaptive variational modal decomposition to accurately extract modal components and enhance the accuracy of fault point location; and combines artificial intelligence image recognition to quickly and accurately determine faults, realize automated operation and maintenance, and improve the operation and maintenance efficiency of transmission lines and the stability of the power system.

[0050] In one embodiment, the data preprocessing module is further configured to calculate an equivalent resistance and a fault equivalent resistance according to the number of lines, calculate a fault reflection parameter according to the equivalent resistance and the fault equivalent resistance, and calculate a fault refraction parameter according to the fault reflection parameter; The signal filtering module also includes: traveling wave transformation module and signal conversion module: The traveling wave transformation module is used to decompose and calculate the original traveling wave according to the fault refraction parameter and the fault reflection parameter to obtain a composite traveling wave, and respectively perform S-transformation on the original traveling wave and the composite traveling wave to obtain a transformed original traveling wave and a transformed composite traveling wave; The signal conversion module is used to calculate the ratio matrix based on the transformed original traveling wave and the transformed composite traveling wave, perform fault screening on the original traveling wave according to the ratio matrix to obtain the fault wave frequency band, and convert the fault wave frequency band from the time-frequency matrix to obtain the time domain signal through the inverse S transform.

[0051] In one embodiment, the data preprocessing module includes: a parameter calculation module, a traveling wave transformation module and a signal conversion module: A parameter calculation module is used to calculate an equivalent resistance and a fault equivalent resistance according to the number of lines, calculate a fault reflection parameter according to the equivalent resistance and the fault equivalent resistance, and calculate a fault refraction parameter according to the fault reflection parameter; The traveling wave transformation module is used to decompose and calculate the original traveling wave according to the fault refraction parameter and the fault reflection parameter to obtain a composite traveling wave, and respectively perform S-transformation on the original traveling wave and the composite traveling wave to obtain a transformed original traveling wave and a transformed composite traveling wave; The signal conversion module is used to calculate the ratio matrix based on the transformed original traveling wave and the transformed composite traveling wave, perform fault screening on the original traveling wave according to the ratio matrix to obtain the fault wave frequency band, and convert the fault wave frequency band from the time-frequency matrix to obtain the time domain signal through the inverse S transform.

[0052] In one embodiment, the fault point determination module includes: a modal decomposition module, a time node module and a fault location module: The modal decomposition module is used to perform phase mode transformation on the time domain signal through the phase mode transformation matrix to obtain the current traveling wave mode component, and decompose the current traveling wave mode component through the improved VMD algorithm to obtain the modal component; The time node module is used to perform Hilbert transform on the modal components to obtain the instantaneous frequency, and extract the time node of the fault current traveling wave based on the instantaneous frequency; the time nodes include t1 and t2; The fault location module is used to calculate the fault location of the time node using the double-end traveling wave ranging formula to obtain the fault position.

[0053] In one embodiment, the modal decomposition module includes: a parameter initialization module, a fitness calculation module, an envelope entropy calculation module and an iterative update module The parameter initialization module is used to initialize the preset algorithm parameters using the parameter combination of the VMD algorithm as a population individual; the parameter combination includes: minimum envelope entropy, modal decomposition layer number K and penalty factor; the parameters include the population size, maximum number of iterations, the value range of the decomposition modal layer number, and the value range of the penalty factor; The fitness calculation module is used to calculate the fitness of individuals in the population, determine the optimal individual based on the fitness, determine the optimal parameter combination corresponding to the optimal individual, and update the VMD algorithm based on the optimal parameter combination to obtain an improved VMD algorithm; The envelope entropy calculation module is used to execute the improved VMD algorithm to decompose and obtain multiple modal components, determine the envelope entropy of each modal component through the objective function, and update the envelope entropy threshold if the envelope entropy is less than the envelope entropy threshold; otherwise, the number of iterations is determined; Objective function: , in, is the envelope entropy, N is the number of signal sampling points, is the envelope amplitude sequence of the signal after Hilbert demodulation, for The normalized envelope sequence of ; The iterative update module is used to update the position of the individuals in the population if the current number of iterations is less than the maximum number of iterations; otherwise, if the current number of iterations is greater than or equal to the maximum number of iterations, the optimal parameter combination is output and the VMD algorithm is updated to obtain an improved VMD algorithm.

[0054] In one embodiment, the alarm determination module includes: A first judgment module is configured to judge the transmission line as high risk if the number of fault areas in the fault point image is greater than or equal to a fault threshold; The second judgment module is configured to judge the transmission line as having low risk if the number of fault areas in the fault point image is less than the fault threshold.

[0055] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for identifying power transmission line faults based on artificial intelligence, characterized in that: The method comprises: Determine line data for a target area, collect data based on the line data to obtain an original traveling wave signal, and determine fault reflection parameters and fault refraction parameters based on the line data; the line data includes: the number of transmission lines and line relationships; the fault reflection parameters are the reflection intensity and polarity of the traveling wave at the busbar or fault point; and the fault refraction parameters are the ratio and polarity of the traveling wave refracted from the fault line; determining a ratio matrix according to the fault reflection parameter and the fault refraction parameter, and filtering the original traveling wave signal according to the ratio matrix to obtain a time domain signal; Decomposing the time domain signal by adaptive variational modal decomposition to obtain modal components, and calculating the fault point based on the modal components; An inspection plan is determined according to the fault point, an image of the fault point is obtained by executing the inspection plan, and an alarm judgment is performed on the image of the fault point through artificial intelligence.

2. The method for identifying power transmission line faults based on artificial intelligence according to claim 1, characterized in that: Determining the fault reflection parameter and the fault refraction parameter according to the line data includes: An equivalent resistance and a fault equivalent resistance are calculated according to the number of lines, a fault reflection parameter is calculated according to the equivalent resistance and the fault equivalent resistance, and a fault refraction parameter is calculated according to the fault reflection parameter; Determining a ratio matrix according to the fault reflection parameter and the fault refraction parameter includes: Decomposing and calculating the original traveling wave according to the fault refraction parameter and the fault reflection parameter to obtain a composite traveling wave, and subjecting the original traveling wave and the composite traveling wave to S-transformation to obtain a transformed original traveling wave and a transformed composite traveling wave respectively; Filtering the original traveling wave signal according to the ratio matrix to obtain a time domain signal includes: A ratio matrix is ​​calculated based on the transformed original traveling wave and the transformed composite traveling wave, the original traveling wave is fault-screened based on the ratio matrix to obtain a fault wave frequency band, and the fault wave frequency band is converted from a time-frequency matrix through an inverse S transform to obtain a time domain signal.

3. The method for identifying power transmission line faults based on artificial intelligence according to claim 1, characterized in that: Decomposing the time domain signal by adaptive variational modal decomposition to obtain modal components, and calculating the fault point based on the modal components includes: Performing phase mode transformation on the time domain signal through a phase mode transformation matrix to obtain a current traveling wave mode component, and decomposing the current traveling wave mode component through an improved VMD algorithm to obtain a modal component; Performing Hilbert transform on the modal component to obtain the instantaneous frequency, and extracting the time nodes of the fault current traveling wave according to the instantaneous frequency; the time nodes include t1 and t2; The fault location is calculated by performing fault location calculation on the time node using a double-terminal traveling wave ranging formula to obtain the fault position.

4. The method for identifying power transmission line faults based on artificial intelligence according to claim 3, characterized in that: The modal components obtained by decomposing the current traveling wave mode components using the improved VMD algorithm include: The parameter combination of the VMD algorithm is used as a population individual to initialize the preset algorithm parameters; the parameter combination includes: minimum envelope entropy, modal decomposition layer number K and penalty factor; the parameters include the value range of the population size, the maximum number of iterations, the number of decomposition modal layers, and the value range of the penalty factor; Calculating the fitness of individuals in the population, determining the optimal individual according to the fitness, determining the optimal parameter combination corresponding to the optimal individual, and updating the VMD algorithm according to the optimal parameter combination to obtain an improved VMD algorithm; Execute the improved VMD algorithm to decompose and obtain multiple modal components, determine the envelope entropy of each modal component through the objective function, and if the envelope entropy is less than the envelope entropy threshold, update the envelope entropy threshold; otherwise, determine the number of iterations; Objective function: , in, is the envelope entropy, N is the number of signal sampling points, is the envelope amplitude sequence of the signal after Hilbert demodulation, for The normalized envelope sequence of ; If the current number of iterations is less than the maximum number of iterations, the positions of the individuals in the population are updated; otherwise, if the current number of iterations is greater than or equal to the maximum number of iterations, the optimal parameter combination is output and the VMD algorithm is updated to obtain an improved VMD algorithm.

5. The method for identifying power transmission line faults based on artificial intelligence according to claim 1, characterized in that: Performing alarm judgment on the fault point image by artificial intelligence includes: If the number of fault areas in the fault point image is greater than or equal to the fault threshold, the transmission line is determined to be at high risk; If the number of fault areas in the fault point image is less than the fault threshold, the transmission line is determined to be of low risk.

6. A power transmission line fault identification device based on artificial intelligence, characterized in that: The device includes: a data pre-processing module, a signal filtering module, a fault point determination module and an alarm judgment module: The data preprocessing module is configured to determine line data in a target area, acquire raw traveling wave signals based on the line data, and determine fault reflection parameters and fault refraction parameters based on the line data; the line data includes the number of transmission lines and their relationships; the fault reflection parameters are the reflection intensity and polarity of the traveling wave at the busbar or fault point; and the fault refraction parameters are the ratio and polarity of the traveling wave refracted from the fault line; The signal filtering module is configured to determine a ratio matrix according to the fault reflection parameter and the fault refraction parameter, and filter the original traveling wave signal according to the ratio matrix to obtain a time domain signal; The fault point determination module is configured to obtain modal components by performing adaptive variational modal decomposition on the time domain signal, and calculate the fault point based on the modal components; The alarm judgment module is used to determine an inspection plan based on the fault point, execute the inspection plan to obtain a fault point image, and perform alarm judgment on the fault point image through artificial intelligence.

7. The power transmission line fault identification device based on artificial intelligence according to claim 6, characterized in that: The data preprocessing module is further configured to calculate an equivalent resistance and a fault equivalent resistance according to the number of lines, calculate a fault reflection parameter according to the equivalent resistance and the fault equivalent resistance, and calculate a fault refraction parameter according to the fault reflection parameter; The signal filtering module also includes: a traveling wave transformation module and a signal conversion module: The traveling wave transformation module is used to decompose and calculate the original traveling wave according to the fault refraction parameter and the fault reflection parameter to obtain a composite traveling wave, and respectively perform S-transformation on the original traveling wave and the composite traveling wave to obtain a transformed original traveling wave and a transformed composite traveling wave; The signal conversion module is used to calculate a ratio matrix based on the transformed original traveling wave and the transformed composite traveling wave, perform fault screening on the original traveling wave according to the ratio matrix to obtain a fault wave frequency band, and convert the fault wave frequency band from a time-frequency matrix through an inverse S transform to obtain a time domain signal.

8. The power transmission line fault identification device based on artificial intelligence according to claim 6, characterized in that: The fault point determination module includes: a modal decomposition module, a time node module and a fault location module: The modal decomposition module is used to perform phase mode transformation on the time domain signal through a phase mode transformation matrix to obtain a current traveling wave mode component, and decompose the current traveling wave mode component through an improved VMD algorithm to obtain a modal component; The time node module is used to perform Hilbert transform on the modal component to obtain the instantaneous frequency, and extract the time node of the fault current traveling wave according to the instantaneous frequency; the time node includes t1 and t2; The fault location module is used to perform fault location calculation on the time node using a double-terminal traveling wave ranging formula to obtain the fault position.

9. The power transmission line fault identification device based on artificial intelligence according to claim 8, characterized in that: The modal decomposition module includes: a parameter initialization module, a fitness calculation module, an envelope entropy calculation module and an iterative update module The parameter initialization module is used to initialize the preset algorithm parameters using the parameter combination of the VMD algorithm as a population individual; the parameter combination includes: minimum envelope entropy, modal decomposition layer number K and penalty factor; the parameters include the value range of the population size, the maximum number of iterations, the decomposition modal layer number, and the value range of the penalty factor; The fitness calculation module is used to calculate the fitness of individuals in the population, determine the optimal individual according to the fitness, determine the optimal parameter combination corresponding to the optimal individual, and update the VMD algorithm according to the optimal parameter combination to obtain an improved VMD algorithm; The envelope entropy calculation module is used to execute the improved VMD algorithm to decompose and obtain multiple modal components, determine the envelope entropy of each modal component through the objective function, and update the envelope entropy threshold if the envelope entropy is less than the envelope entropy threshold; otherwise, determine the number of iterations; Objective function: , in, is the envelope entropy, N is the number of signal sampling points, is the envelope amplitude sequence of the signal after Hilbert demodulation, for The normalized envelope sequence of ; The iterative update module is used to update the position of the population individuals if the current number of iterations is less than the maximum number of iterations; otherwise, if the current number of iterations is greater than or equal to the maximum number of iterations, output the optimal parameter combination and update the VMD algorithm to obtain an improved VMD algorithm.

10. The power transmission line fault identification device based on artificial intelligence according to claim 6, characterized in that: The alarm judgment module includes: The first judgment module is configured to judge the transmission line as high risk if the number of fault areas in the fault point image is greater than or equal to a fault threshold; The second judgment module is configured to judge the transmission line as having low risk if the number of fault areas in the fault point image is less than a fault threshold.

Images