An ultrasonic nondestructive flaw detection system for insulating sub used in live-line work
Patent Information
- Application Number
- CN202610525708.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-21
- Publication Date
- 2026-08-18
AI Technical Summary
[0002]绝缘子是发电厂和变电站等电网系统的重要设备部件，被广泛应用于高压输电、开关柜、闸刀等现代电网系统中，起着支撑输电导线和绝缘的作用，由于绝缘子本身脆性大、工作环境恶劣及自然老化等问题，在其自身结构和外界环境的影响下，容易发生机械损伤，从而产生裂纹或气泡等缺陷，在受到应力作用时，裂纹容易迅速扩展，显著降低绝缘子的可靠性，使得绝缘子极易发生突然断裂失效，引发设备故障，造成安全隐患，严重影响电网系统的安全运行，对绝缘子进行无损缺陷检测是预防绝缘子断裂失效、判断使用寿命和剩余寿命的重要措施
[0018] (1) By using a laser excitation source to excite ultrasonic waves on the surface of the insulator, when the ultrasonic waves propagate inside the insulator, they encounter typical defects such as interface cracks, air gaps in the sheath, and delamination between the core rod and the sheath. The propagation path, waveform, amplitude, and phase characteristics will change significantly. These changes in propagation signals caused by defects are captured by a high-sensitivity receiving device and converted into electrical signals, so as to realize non-power-off, non-contact, non-destructive, and efficient defect location detection of live insulators.
Smart Images
Figure CN122591809A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of non-destructive testing technology for insulators, specifically relating to an ultrasonic non-destructive testing system for insulators suitable for live-line work. Background Technology
[0002] Insulators are crucial components of power grid systems, including power plants and substations. They are widely used in modern power grid systems such as high-voltage transmission lines, switchgear, and disconnectors, serving to support transmission lines and provide insulation. Due to their inherent brittleness, harsh working environments, and natural aging, insulators are prone to mechanical damage under the influence of their own structure and the external environment, resulting in defects such as cracks or bubbles. When subjected to stress, these cracks can propagate rapidly, significantly reducing the reliability of the insulator and making it highly susceptible to sudden fracture failure. This can lead to equipment malfunctions, safety hazards, and seriously affect the safe operation of the power grid system. Therefore, non-destructive testing of insulators is an important measure to prevent insulator fracture failure and to determine their service life and remaining life.
[0003] Existing technologies include traditional detection methods such as penetrant testing, magnetic particle testing, electromagnetic testing, X-ray testing, and ultrasonic testing. These methods mainly suffer from the following problems: First, due to the presence of various forms of electromagnetic interference around high-voltage live equipment, with field strengths far exceeding those of ordinary industrial environments, traditional detection methods lack sufficient resistance to electromagnetic interference in strong magnetic environments, leading to noise and misjudgments. Second, they cannot effectively detect hidden damage inside insulators. Such defects typically do not cause significant temperature rises in the early stages, and their heat generation is small or dissipates rapidly, making it difficult to create a temperature difference on the surface that can be captured by infrared equipment, resulting in a high risk of missed detection. Finally, traditional contact ultrasonic testing methods face many practical difficulties in insulator inspection due to the requirement of coupling agent and physical contact. Insulators are often slender, with surfaces made of silicone rubber composite materials and complex skirt structures, making it difficult for probes to achieve uniform and stable coupling. Conducting contact testing under high-voltage live conditions is not only extremely dangerous and prone to electric shock, but also requires power outages or the installation of complex insulation barriers, significantly increasing maintenance costs and time.
[0004] Therefore, in order to solve the above problems, it is necessary to develop an ultrasonic non-destructive testing system for insulators suitable for live-line work. Summary of the Invention
[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide an ultrasonic non-destructive testing system for insulators suitable for live-line work. It adopts non-contact ultrasonic non-destructive testing technology to achieve uninterrupted, non-destructive, and efficient diagnosis of live insulators, ensuring reliable operation of the power grid and improving the level of equipment health management.
[0006] The objective of this invention is achieved as follows: an ultrasonic non-destructive testing system for insulators suitable for live-line work, comprising a laser excitation unit, a scanning motion unit, an ultrasonic receiving unit, a data acquisition unit, a data processing unit, a data storage unit, a data analysis unit, a defect image visualization unit, and a data transmission unit;
[0007] The system comprises the following components: a laser excitation unit that emits pulsed laser light and scans the insulator via a scanning motion unit; an ultrasonic receiving unit that receives ultrasonic signals excited on the insulator; a data acquisition unit that rapidly acquires and digitizes the analog ultrasonic signals received by the ultrasonic receiving unit; a data processing unit that processes the data, including noise reduction and feature parameter extraction, and stores the data via a data storage unit; a data analysis unit that classifies defects based on the processed data to obtain insulator defect results; a defect image visualization unit that images internal defects of the insulator based on ultrasonic imaging theory, achieving high-resolution visualization of internal defects; and a data transmission unit that supports real-time remote transmission and control of detection data.
[0008] Furthermore, the scanning motion unit adopts a two-dimensional scanning galvanometer device, specifically including X and Y scanning mirrors and X and Y drive motors for driving the X and Y scanning mirrors to rotate. The motion control module controls the X and Y drive motors to drive the X and Y scanning mirrors to rotate respectively, and the focusing system focuses to achieve precise positioning of the laser scanning path.
[0009] Furthermore, the ultrasonic receiving unit uses a laser as the receiving laser source to emit a linear laser beam to receive the signal, and synchronizes the received signal to the data acquisition unit for processing.
[0010] Furthermore, the data processing unit specifically employs a graph neural network (GNN) noise reduction algorithm to perform noise reduction processing on the signal.
[0011] Furthermore, the graph neural network (GNN) denoising algorithm for signal denoising specifically includes the following steps: (1) constructing a graph dataset by calculating the similarity of ultrasonic signals in the frequency domain; specifically, ① nodes and The similarity between them is calculated by their Euclidean distance in the frequency domain: In the formula, and They represent signals respectively and In the The amplitude of each frequency component ① Represents the total number of frequency components of the signal; ② For each node, select the node closest to it based on the calculated frequency domain Euclidean distance. ③ Then the graph dataset is represented as: (Edge connections are established using signals;) In the formula, The nodes in the diagram represent a single ultrasonic signal. An adjacency matrix is a two-dimensional array used to represent the connection relationships between signals. Indicates the total number of signals; if the signal and If there is a connection between them, then Assign a value based on the similarity between the two, and vice versa. (2) The ultrasonic time-domain signal is denoised using a GNN algorithm based on an encoder-decoder structure. The encoder extracts the main features of the signal through a graph structure and generates a denoised signal. The decoder reconstructs the original signal through the features retained in the denoised signal. The principle is expressed as follows: In the formula: It is an adjacency matrix. It is the identity matrix. For parameter matrices, The activation function is used to introduce a nonlinear transformation; and by minimizing the loss function, the network parameters are optimized to preserve the accuracy of signal reconstruction. Specifically, the mean square error (MSE) is used as the loss function on the parameter matrix. Optimize: In the formula, Represents the original signal. Indicates the reconstructed signal. Indicates the total number of signals.
[0012] Furthermore, the data processing unit specifically employs the empirical mode decomposition method to extract feature parameters from the signal.
[0013] Furthermore, the empirical mode decomposition method for extracting feature parameters of the signal specifically includes the following steps: ① performing cubic spline interpolation fitting on the local maxima and minima of the signal to obtain the upper envelope and lower envelope; ② calculating the mean of the upper and lower envelopes. ;③Use the original signal minus The value obtained is : ④ Repeat the above steps until... Satisfying the IMF conditions; assuming IMF1 is: Then we have: The difference was calculated ⑤ Repeat the decomposition steps for the remaining signal to solve for the eigenmode functions of each order in turn. The decomposition ends when the residual is sufficiently small to be a monotonic function; at this point, the original signal is represented as the sum of each order component and the residual term: In the formula, Represents the IMF components at each order. This represents the residual term.
[0014] Furthermore, the data analysis unit specifically employs a support vector machine classifier to classify insulator defects and uses a particle swarm optimization algorithm to optimize the parameters of the support vector machine classifier.
[0015] Furthermore, the defect image visualization unit specifically employs synthetic aperture focusing to achieve imaging of internal defects in the insulator.
[0016] Furthermore, the synthetic aperture focusing method for imaging internal defects in insulators specifically includes the following steps: assuming the coordinates of a certain focal point in the imaging region are... Then the imaging area The process of reconstructing points is represented as follows: In the formula, Indicates the total number of array cells. This represents the ultrasonic time-domain signal of each receiving unit. This represents the time delay between the array element and the target point, i.e., the propagation time of the reflected wave. This time delay is calculated by the following formula: In the formula, Indicates the wave velocity of longitudinal waves. This indicates the wave velocity of the longitudinal wave reflected.
[0017] Due to the adoption of the above technical solution, the beneficial effects of the present invention are:
[0018] (1) By using a laser excitation source to excite ultrasonic waves on the surface of the insulator, when the ultrasonic waves propagate inside the insulator, they encounter typical defects such as interface cracks, air gaps in the sheath, and delamination between the core rod and the sheath. The propagation path, waveform, amplitude, and phase characteristics will change significantly. These changes in propagation signals caused by defects are captured by a high-sensitivity receiving device and converted into electrical signals, so as to realize non-power-off, non-contact, non-destructive, and efficient defect location detection of live insulators.
[0019] (2) By setting up a data processing unit and using the graph neural network (GNN) noise reduction algorithm to denoise the data, the signal-to-noise ratio of the signal is improved, which facilitates the effective separation and enhancement of weak defect echo signals in composite insulators from strong noise background. At the same time, the characteristic parameters of the denoised signal are extracted by the empirical mode decomposition method, and the probability of it being a defect signal is analyzed and judged, thereby improving detection efficiency and accuracy.
[0020] (3) By setting up a data analysis unit, based on the data processed by the data processing unit, the support vector machine classifier is used to classify the insulator defects, and the particle swarm algorithm is used to optimize the parameters of the support vector machine classifier to obtain the insulator defect results, thereby realizing the rapid determination and classification of insulator defects and effectively improving the defect detection effect;
[0021] (4) By setting up a defect image visualization unit and using the synthetic aperture focusing method, the internal defects of the insulator are visualized, and high-resolution imaging of the internal defects of the insulator is achieved, which facilitates the analysis and processing of the defects and increases the accuracy of defect detection and quantitative measurement. Attached Figure Description
[0022] Figure 1 This is a schematic diagram of the principle of the present invention. Detailed Implementation
[0023] The technical solution of the present invention will be further described in detail below through embodiments and in conjunction with the accompanying drawings.
[0024] like Figure 1 As shown, an ultrasonic non-destructive testing system for insulators suitable for live-line work includes a laser excitation unit, a scanning motion unit, an ultrasonic receiving unit, a data acquisition unit, a data processing unit, a data storage unit, a data analysis unit, a defect image visualization unit, and a data transmission unit.
[0025] The laser excitation unit is used to emit pulsed laser light and scan the insulator through the scanning motion unit.
[0026] Preferably, the scanning motion unit employs a two-dimensional scanning galvanometer device, specifically including X and Y scanning mirrors and X and Y drive motors for driving the rotation of the X and Y scanning mirrors. The motion control module controls the X and Y drive motors to drive the X and Y scanning mirrors to rotate respectively, and focuses them through a focusing system to achieve precise positioning of the laser scanning path. The motion control module coordinates and controls the laser beam to move on the surface of the insulator with a set trajectory and step size based on the set scanning path, scanning resolution, and running speed, ensuring spatial consistency and repeatability of the signal acquisition process, and achieving overlap between the laser beam and the surface of the insulator being tested.
[0027] The ultrasonic receiving unit is used to receive ultrasonic signals excited on the insulator; preferably, the ultrasonic receiving unit uses a laser as the receiving laser source to emit a linear laser beam to receive the signal, and synchronizes the received signal to the data acquisition unit for processing.
[0028] The data acquisition unit is used to acquire the ultrasonic analog signal received by the ultrasonic receiving unit at high speed and perform digital conversion processing.
[0029] The data processing unit is used to process the data, including noise reduction and feature parameter extraction of the signal, and to store it through the data storage unit.
[0030] Preferably, the data processing unit specifically employs a graph neural network (GNN) denoising algorithm to denoise the signal, specifically including the following steps: (1) constructing a graph dataset by calculating the similarity of ultrasonic signals in the frequency domain; specifically, ① nodes and The similarity between them is calculated by their Euclidean distance in the frequency domain: In the formula, and They represent signals respectively and In the The amplitude of each frequency component ① Represents the total number of frequency components of the signal; ② For each node, select the node closest to it based on the calculated frequency domain Euclidean distance. ③ Then the graph dataset is represented as: (Edge connections are established using signals;) In the formula, The nodes in the diagram represent a single ultrasonic signal. An adjacency matrix is a two-dimensional array used to represent the connection relationships between signals. Indicates the total number of signals; if the signal and If there is a connection between them, then Assign a value based on the similarity between the two, and vice versa. (2) The ultrasonic time-domain signal is denoised using a GNN algorithm based on an encoder-decoder structure. The encoder extracts the main features of the signal through a graph structure and generates a denoised signal. The decoder reconstructs the original signal through the features retained in the denoised signal. The principle is expressed as follows: In the formula: It is an adjacency matrix. It is the identity matrix. For parameter matrices, The activation function is used to introduce a nonlinear transformation; and by minimizing the loss function, the network parameters are optimized to preserve the accuracy of signal reconstruction. Specifically, the mean square error (MSE) is used as the loss function on the parameter matrix. Optimize: In the formula, Represents the original signal. Indicates the reconstructed signal. Indicates the total number of signals.
[0031] Preferably, the data processing unit specifically employs the empirical mode decomposition method to extract feature parameters from the signal, including the following steps: ① performing cubic spline interpolation fitting on the local maxima and minima of the signal to obtain the upper and lower envelopes; ② calculating the mean of the upper and lower envelopes. ;③Use the original signal minus The value obtained is : ④ Repeat the above steps until... Satisfying the IMF conditions; assuming IMF1 is: Then we have: The difference was calculated ⑤ Repeat the decomposition steps for the remaining signal to solve for the eigenmode functions of each order in turn. The decomposition ends when the residual is sufficiently small to be a monotonic function; at this point, the original signal is represented as the sum of each order component and the residual term: In the formula, Represents the IMF components at each order. This represents the residual term.
[0032] The data analysis unit classifies defects based on the data processed by the data processing unit to obtain insulator defect results. Preferably, the data analysis unit specifically uses a support vector machine classifier to classify insulator defects and uses a particle swarm optimization algorithm to optimize the parameters of the support vector machine classifier.
[0033] Preferably, the data analysis unit uses a support vector machine classifier to split insulator defects, specifically including the following steps: For sample points , and These are the input and output feature vectors, respectively. Then its optimal classification surface Represented as: In the formula, Represents the Lagrange operator, The kernel function is represented by the Gaussian radial basis function as follows: , Need to meet ,in, This represents any support vector.
[0034] Preferably, the optimization of the support vector machine classifier parameters using the particle swarm optimization algorithm in the data analysis unit specifically includes the following steps: ① setting the particle swarm population size and the dimension of the search space, initializing particle positions and velocities, and setting the inertia weights in the algorithm. Individual learning factors Social learning factors Parameters; ② Train the support vector machine model for each particle using the current parameter combination, and calculate the initial fitness value using the fitness function; ③ Determine the optimal position for each particle based on the initial fitness. Then from the optimal position of all particles The position with the best fitness was selected as the global optimal position for the population. ④ Update the velocity and optimal position of each particle, where the velocity update formula is expressed as: In the formula, and These represent the speeds before and after the update, respectively. , express Random numbers within the interval; the position of each particle is updated by adding the new velocity value to the current position; ⑤ Determine whether the maximum number of evolution generations has been reached or the termination condition is met. If not, return to step ② and continue; ⑥ After completing the iterative optimization, use the globally optimal position. The optimal combination of parameters is taken as the best solution, and the corresponding optimal fitness value is recorded as the best parameters for the support vector machine model.
[0035] The defect image visualization unit is based on ultrasonic imaging theory to image internal defects of insulators, thereby achieving high-resolution visualization of internal defects of insulators.
[0036] Preferably, the defect image visualization unit specifically employs synthetic aperture focusing to achieve internal defect imaging of the insulator; wherein, it is assumed that the coordinates of a certain focal point in the imaging region are... Then the imaging area The process of reconstructing points is represented as follows: In the formula, Indicates the total number of array cells. This represents the ultrasonic time-domain signal of each receiving unit. This represents the time delay between the array element and the target point, i.e., the propagation time of the reflected wave. This time delay is calculated by the following formula: In the formula, Indicates the wave velocity of longitudinal waves. This indicates the wave velocity of the longitudinal wave reflected.
[0037] The data transmission unit is used to support real-time remote transmission and remote control of detection data.
[0038] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of the claims of the present invention.
Claims
1. An ultrasonic non-destructive testing system for insulators suitable for live-line work, characterized in that: It includes a laser excitation unit, a scanning motion unit, an ultrasonic receiving unit, a data acquisition unit, a data processing unit, a data storage unit, a data analysis unit, a defect image visualization unit, and a data transmission unit; The system comprises the following components: a laser excitation unit that emits pulsed laser light and scans the insulator via a scanning motion unit; an ultrasonic receiving unit that receives ultrasonic signals excited on the insulator; a data acquisition unit that rapidly acquires and digitizes the analog ultrasonic signals received by the ultrasonic receiving unit; a data processing unit that processes the data, including noise reduction and feature parameter extraction, and stores the data via a data storage unit; a data analysis unit that classifies defects based on the processed data to obtain insulator defect results; a defect image visualization unit that images internal defects of the insulator based on ultrasonic imaging theory, achieving high-resolution visualization of internal defects; and a data transmission unit that supports real-time remote transmission and control of detection data.
2. The ultrasonic non-destructive testing system for insulators suitable for live-line work according to claim 1, characterized in that: The scanning motion unit adopts a two-dimensional scanning galvanometer device, specifically including X and Y scanning mirrors and X and Y drive motors for driving the rotation of the X and Y scanning mirrors. The motion control module controls the X and Y drive motors to drive the X and Y scanning mirrors to rotate respectively, and the focusing system focuses to achieve precise positioning of the laser scanning path.
3. The ultrasonic non-destructive testing system for insulators suitable for live-line work according to claim 1, characterized in that: The ultrasonic receiving unit uses a laser as the receiving laser source to emit a linear laser beam to receive the signal, and synchronizes the received signal to the data acquisition unit for processing.
4. The ultrasonic non-destructive testing system for insulators suitable for live-line work according to claim 1, characterized in that: The data processing unit specifically employs a graph neural network (GNN) noise reduction algorithm to perform noise reduction processing on the signal.
5. The ultrasonic non-destructive testing system for insulators suitable for live-line work according to claim 4, characterized in that: The graph neural network (GNN) denoising algorithm for signal denoising specifically includes the following steps: (1) constructing a graph dataset by calculating the similarity of ultrasonic signals in the frequency domain; specifically, ① nodes and The similarity between them is calculated by their Euclidean distance in the frequency domain: In the formula, and They represent signals respectively and In the The amplitude of each frequency component ① Represents the total number of frequency components of the signal; ② For each node, select the node closest to it based on the calculated frequency domain Euclidean distance. ③ Then the graph dataset is represented as: (Edge connections are established using signals;) In the formula, The nodes in the diagram represent a single ultrasonic signal. An adjacency matrix is a two-dimensional array used to represent the connection relationships between signals. Indicates the total number of signals; if the signal and If there is a connection between them, then Assign a value based on the similarity between the two, and vice versa. (2) The ultrasonic time-domain signal is denoised using a GNN algorithm based on an encoder-decoder structure. The encoder extracts the main features of the signal through a graph structure and generates a denoised signal. The decoder reconstructs the original signal through the features retained in the denoised signal. The principle is expressed as follows: In the formula: It is an adjacency matrix. It is the identity matrix. For parameter matrices, The activation function is used to introduce a nonlinear transformation; and by minimizing the loss function, the network parameters are optimized to preserve the accuracy of signal reconstruction. Specifically, the mean square error (MSE) is used as the loss function on the parameter matrix. Optimize: In the formula, Represents the original signal. Indicates the reconstructed signal. Indicates the total number of signals.
6. The ultrasonic non-destructive testing system for insulators suitable for live-line work according to claim 1, characterized in that: The data processing unit specifically employs the empirical mode decomposition method to extract feature parameters from the signal.
7. The ultrasonic non-destructive testing system for insulators suitable for live-line work according to claim 6, characterized in that: The empirical mode decomposition method for extracting feature parameters from a signal specifically includes the following steps: ① performing cubic spline interpolation fitting on the local maxima and minima of the signal to obtain the upper and lower envelopes; ② calculating the mean of the upper and lower envelopes. ;③Use the original signal minus The value obtained is : ④ Repeat the above steps until... Satisfying the IMF conditions; assuming IMF1 is: Then we have: The difference was calculated ⑤ Repeat the decomposition steps for the remaining signal to solve for the eigenmode functions of each order in turn. The decomposition ends when the residual is sufficiently small to be a monotonic function; at this point, the original signal is represented as the sum of each order component and the residual term: In the formula, Represents the IMF components at each order. This represents the residual term.
8. The ultrasonic non-destructive testing system for insulators suitable for live-line work according to claim 1, characterized in that: The data analysis unit specifically uses a support vector machine classifier to classify insulator defects and employs a particle swarm optimization algorithm to optimize the parameters of the support vector machine classifier.
9. The ultrasonic non-destructive testing system for insulators suitable for live-line work according to claim 1, characterized in that: The defect image visualization unit specifically employs synthetic aperture focusing to achieve imaging of internal defects in insulators.
10. The ultrasonic non-destructive testing system for insulators suitable for live-line work according to claim 9, characterized in that: The synthetic aperture focusing method for imaging internal defects in insulators specifically includes the following steps: assuming the coordinates of a certain focal point in the imaging region are... Then the imaging area The process of reconstructing points is represented as follows: In the formula, Indicates the total number of array cells. This represents the ultrasonic time-domain signal of each receiving unit. This represents the time delay between the array element and the target point, i.e., the propagation time of the reflected wave. This time delay is calculated by the following formula: In the formula, Indicates the wave velocity of longitudinal waves. This indicates the wave velocity of the longitudinal wave reflected.