A laser-ultrasonic-based insulator internal defect detection method and system

Abstract

The application discloses a kind of insulator internal defect detection method and system based on laser ultrasound, which comprises: pulse laser excitation light path and continuous laser interference detection light path are aligned on the same side and adjusted to concentric overlap;With the energy density lower than damage threshold, the body wave is excited, and multi-directional surface vibration signals are collected;Signal is subjected to time-frequency domain joint transformation and adaptive matching tracking decomposition, and energy distribution characteristic matrix is constructed, and defect echo discrimination threshold is adaptively calculated in combination with fractal dimension analysis;Based on the threshold, candidate defect echo is identified, and its time-frequency characteristic parameters are recorded;Candidate echo under different directions is associated according to spatial position, adjacent direction characteristic similarity is calculated and coherent enhancement is carried out, and multi-directional spatial coherence atlas is constructed;The coherent peak region in atlas is detected, and if it exists, it is judged that real defect exists and depth is estimated.The application realizes the high sensitivity, high reliability non-contact detection of small defect in insulator.

Description

Technical Field

[0001] This invention belongs to the field of non-destructive testing technology for power equipment, and particularly relates to a method and system for detecting internal defects in insulators based on laser ultrasound. Background Technology

[0002] Insulators are critical components in high-voltage transmission lines and substation equipment, and their insulation performance and mechanical strength directly affect the safe and stable operation of the power system. During manufacturing, transportation, installation, and long-term service, insulators may develop internal defects such as pores, cracks, inclusions, and delamination within the ceramic or composite material body due to factors such as raw material fluctuations, defects in molding and sintering processes, assembly impacts, temperature cycling, and electro-thermal-mechanical coupling stress. These internal defects are prone to expansion and evolution under long-term electric fields and mechanical stress, potentially leading to flashover, breakdown, or fracture faults, causing line outages or even large-scale power outages. Therefore, regular and reliable non-destructive testing of insulators to promptly detect potential internal defects is of great significance for ensuring power grid safety.

[0003] Currently, insulator inspection methods mainly include visual inspection, infrared thermography, and ultrasonic testing. Visual inspection can usually only detect obvious surface defects and is difficult to identify minute defects embedded inside the material. Infrared thermography mainly relies on temperature field anomalies to indirectly infer defects, which is easily affected by ambient temperature, sunlight, wind speed, and load fluctuations, and has limited sensitivity to early internal defects that do not produce significant thermal anomalies. Traditional ultrasonic testing of piezoelectric transducers requires the probe to contact the surface being tested and use a coupling agent, which is cumbersome in field operation and makes it difficult to achieve truly non-contact and long-distance testing. At the same time, insulators generally have curved surfaces, sheds, or complex shapes, and the probe contact and coupling conditions are prone to fluctuation, resulting in poor echo consistency and insufficient repeatability, thus affecting the accuracy and reliability of the test.

[0004] Laser ultrasonic testing, as an emerging non-destructive testing technology, uses pulsed lasers to excite ultrasonic waves on the material surface and then employs optical methods to receive the surface vibration signals non-contactly, enabling long-distance, non-contact, and high-bandwidth defect detection. However, under the thermoelastic mechanism, to avoid damaging the insulator surface, the laser energy density must be below the damage threshold, resulting in weak ultrasonic volume wave signals. Defect echoes are easily drowned out by structural echoes and environmental noise, leading to a low signal-to-noise ratio and difficulty in detecting internal micro-defects. Furthermore, existing methods are mostly designed for general materials or surface / near-surface defects, lacking optimized configurations for internal insulator defects and defect signal authenticity verification mechanisms. This makes the detection results susceptible to structural echoes, curvature changes, and environmental disturbances, resulting in insufficient reliability.

[0005] Therefore, there is an urgent need to provide a method for detecting internal defects in insulators that can effectively improve the signal-to-noise ratio of defect echoes and reliably verify suspected defect echoes under non-destructive thermo-elastic conditions. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a laser-ultrasound-based method and system for detecting internal defects in insulators. This invention aims to solve the problems of weak defect echo, low signal-to-noise ratio, and difficulty in detecting internal micro-defects under thermo-elastic non-destructive testing conditions, and to improve the reliability of the detection results.

[0007] In a first aspect, the present invention provides a method for detecting internal defects in insulators based on laser ultrasound, comprising: Obtain surface reference information of the insulator under test in its in-service or installed state, and align the pulsed laser excitation optical path and the continuous laser interferometric detection optical path to the test area based on the surface reference information; Adjust the pulsed laser excitation optical path and the continuous laser interference detection optical path so that the excitation spot and the detection spot are concentrically aligned on the surface of the area to be measured. The pulsed laser is controlled to irradiate the test area of ​​the insulator with an energy density lower than the damage threshold of the insulator material. Under the thermoelastic mechanism, a volume wave propagating into the interior of the insulator is excited. Based on the synchronous trigger signal, multiple sets of surface vibration signals under different detection orientations are collected by a continuous laser interferometer to obtain an initial ultrasonic signal sequence corresponding to each detection orientation. The surface vibration signal contains the defect echo component formed by the reflection or scattering of the volume wave after it encounters a defect inside the insulator. A joint time-frequency domain transformation is performed on each initial ultrasonic signal in the initial ultrasonic signal sequence to extract the energy distribution feature matrix of each signal within a preset time-frequency window. Based on the energy distribution feature matrix and the statistical characteristics of the background noise, the defect echo discrimination threshold corresponding to each initial ultrasonic signal is adaptively calculated. Based on the defect echo discrimination threshold, candidate defect echoes are identified in each initial ultrasonic signal, and the characteristic parameters of each candidate defect echo are recorded. Candidate defect echoes identified under different detection orientations are correlated according to their spatial location to construct a multi-directional spatial coherence map. The horizontal axis of the map is the detection azimuth angle, the vertical axis is the echo arrival time, and the map intensity is the peak amplitude of the corresponding candidate defect echo. Determine whether there is a coherent peak region in the multi-directional spatial coherence map. The coherent peak region is defined as a region that has a consistent echo arrival time in multiple consecutive detection directional positions and whose peak amplitude exceeds a preset coherence threshold. If the coherent peak region exists, it is determined that there is a real defect inside the insulator corresponding to the coherent peak region, and the depth range of the defect is analyzed based on the echo arrival time of the coherent peak region and the preset sound velocity model.

[0008] Secondly, the present invention provides a laser-ultrasound-based insulator internal defect detection system, comprising: The acquisition module is configured to acquire surface reference information of the insulator under test area in service or installation state, and align the pulsed laser excitation optical path and the continuous laser interferometric detection optical path to the test area based on the surface reference information; The adjustment module is configured to adjust the excitation optical path and the detection optical path so that the excitation spot and the detection spot coincide concentrically on the surface of the area to be measured. The acquisition module is configured to control a pulsed laser to irradiate the test area of ​​the insulator with an energy density lower than the damage threshold of the insulator material, thereby exciting a body wave that propagates into the interior of the insulator under a thermoelastic mechanism. Based on a synchronous trigger signal, multiple sets of surface vibration signals at different detection orientations are acquired using a continuous laser interferometer to obtain an initial ultrasonic signal sequence corresponding to each detection orientation. The surface vibration signal includes a defect echo component formed by the reflection or scattering of the body wave after it encounters a defect during propagation inside the insulator. The extraction module is configured to perform a joint time-frequency domain transformation on each initial ultrasound signal in the initial ultrasound signal sequence, extract the energy distribution characteristics of each signal within a preset time-frequency window, and adaptively calculate the defect echo discrimination threshold corresponding to each initial ultrasound signal based on the energy distribution characteristics and the statistical characteristics of the background noise. The identification module is configured to identify candidate defect echoes in each initial ultrasonic signal based on the defect echo discrimination threshold, and record the characteristic parameters of each candidate defect echo. The association module is configured to associate candidate defect echoes identified under different detection orientations according to their spatial locations to construct a multi-directional spatial coherence map, wherein the horizontal axis of the map is the detection azimuth angle, the vertical axis is the echo arrival time, and the map intensity is the peak amplitude of the corresponding candidate defect echo. The judgment module is configured to determine whether there is a coherent peak region in the multi-directional spatial coherence map. The coherent peak region is defined as a region that has a consistent echo arrival time in multiple consecutive detection directions and whose peak amplitude exceeds a preset coherence threshold. The analysis module is configured to determine that if the coherent peak region exists, there is a real defect inside the insulator corresponding to the coherent peak region, and to analyze the depth range of the defect based on the echo arrival time of the coherent peak region and a preset sound velocity model.

[0009] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the laser-ultrasound-based insulator internal defect detection method according to any embodiment of the present invention.

[0010] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the laser-ultrasound-based insulator internal defect detection method according to any embodiment of the present invention.

[0011] The laser-ultrasound-based insulator internal defect detection method and system of this application have the following beneficial effects: By using a large spot excitation, the laser power density is effectively reduced, ensuring that the excitation process is within the thermoelastic mechanism range, avoiding ablation damage to the insulator surface, and ensuring the integrity of the tested component. By adopting a coaxial concentric optical path configuration, the detection point is located at the center of the sound source, effectively receiving the volume wave echo; combined with matched pursuit time-frequency transformation, fractal dimension adaptive threshold and multi-directional coherence enhancement, the signal-to-noise ratio of weak defect echoes is greatly improved, and sub-millimeter-level internal pores, cracks and other minute defects can be detected. The optical path structure, with excitation and detection on the same side and concentric, is simple, easy to align and adjust, and does not require devices to be arranged on both sides of the insulator. It is suitable for non-contact detection of installed insulators on site. By constructing a multi-directional spatial coherence map and performing coherence peak region detection, the signal is enhanced by utilizing the consistency characteristics of real defect echoes in adjacent azimuth angles, while suppressing isolated noise and stray reflections, effectively reducing the false judgment rate. It perfectly matches the same-side testing conditions of insulators and can effectively adapt to complex curved surface structures such as post porcelain insulators and umbrella-shaped insulators, solving the problem of poor adaptability of traditional testing methods to insulator structures. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating a laser-ultrasound-based method for detecting internal defects in insulators, as provided in an embodiment of the present invention; Figure 2This is a structural block diagram of an insulator internal defect detection system based on laser ultrasound, provided in one embodiment of the present invention. Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Please see Figure 1 The diagram shows a flowchart of a laser-ultrasound-based insulator internal defect detection method according to this application.

[0016] like Figure 1 As shown, the laser-ultrasound-based method for detecting internal defects in insulators specifically includes the following steps: Step S101: Obtain surface reference information of the insulator under test area in service or installation state, and align the pulsed laser excitation optical path and the continuous laser interferometric detection optical path to the test area according to the surface reference information.

[0017] In this step, the center coordinates of the area to be tested and the direction of the surface normal are obtained from the surface reference information. Using the center coordinates as the aiming point, the two-dimensional galvanometer is controlled to adjust the incident direction of the pulsed laser so that the deviation between the center of the excitation spot and the center coordinates is less than a preset first deviation threshold. After the excitation spot alignment is completed, the current receiving angle of the detection optical path is obtained, the ideal receiving angle is calculated according to the surface normal direction, and the automatic angle adjustment mechanism of the detection optical path is controlled to adjust the receiving angle so that the deviation between the actual receiving angle and the ideal receiving angle is less than the preset second deviation threshold. A beam of alignment laser with a power less than a preset power threshold is emitted, and spot images of the excitation optical path and the probe optical path on the surface of the area to be tested are acquired respectively. The pixel distance between the centers of the two spots is calculated. If the pixel distance is greater than a preset third deviation threshold, the alignment process is repeated until the deviation requirement is met.

[0018] In one specific embodiment, the detection system first acquires pre-stored or real-time measured surface reference information of the insulator's test area. This surface reference information includes the center coordinates of the test area and the normal direction of the surface. This information can be obtained in the following ways: by installing a three-dimensional laser rangefinder or structured light scanner on the detection device to rapidly scan the insulator surface, reconstructing the local three-dimensional topography of the test area, thereby automatically extracting the center coordinates and normal direction; or, given the insulator type and installation orientation, retrieving the corresponding geometric parameters from a pre-set database.

[0019] After obtaining the center coordinates, the system controls a two-dimensional galvanometer to adjust the incident direction of the pulsed laser. The two-dimensional galvanometer consists of two high-speed oscillating mirrors, which control the deflection of the beam in the horizontal and vertical directions, respectively. The system calculates the required deflection angle based on the center coordinates and drives the galvanometer motor to precisely point the center of the pulsed laser spot to the stated center coordinates. Simultaneously, a position-sensitive detector or CCD camera monitors the actual position of the spot on the insulator surface in real time, forming a closed-loop feedback. When the distance between the center of the excitation spot and the target center coordinates is less than a preset first deviation threshold (e.g., 0.2 mm), the excitation optical path alignment is considered complete.

[0020] Next, the probe optical path is aligned. The system first acquires the current receiving angle of the continuous laser interferometry probe optical path, which can be read by an angle sensor installed in the probe optical path. Then, based on the previously acquired surface normal direction, the ideal receiving angle is calculated—typically, the probe laser beam is required to be incident perpendicularly to the surface of the area to be measured, i.e., the ideal receiving angle is equal to the surface normal direction. The system controls an automatic angle adjustment mechanism (e.g., a two-axis precision turntable driven by a stepper motor) to rotate the entire probe optical path, gradually bringing the actual receiving angle closer to the ideal receiving angle. When the angular deviation between the two is less than a preset second deviation threshold (e.g., 0.5 degrees), the receiving angle of the probe optical path is considered to meet the requirements.

[0021] To ensure that the excitation and probe spots are perfectly concentric on the surface of the area under test, the system also needs to perform a fine coaxial verification and correction. Specifically, the system emits a low-power visible alignment laser (e.g., a red semiconductor laser with a power of less than 1 milliwatt for safety), which propagates along the pulsed laser excitation path to form a clearly visible first guide spot on the insulator surface. Simultaneously, the system also emits another alignment laser of the same wavelength, which propagates along the continuous laser interferometry probe path to form a second guide spot. A high-resolution industrial camera (which can be equipped with a telephoto lens) is used to acquire surface images including both guide spots.

[0022] In the image, the system uses image processing algorithms (such as grayscale centroid method or circle fitting algorithm) to calculate the pixel coordinates of the two spot centers, thereby obtaining the pixel distance between the two centers. If this pixel distance is greater than a preset third deviation threshold (e.g., 3 pixels), the system automatically determines that the current alignment accuracy is insufficient. At this time, based on the relative positional deviation of the two spot centers, the system calculates the fine-tuning amount required for the two-dimensional galvanometer and angle adjustment mechanism, readjusts the directions of the excitation and probe optical paths, and then re-captures the image and calculates the spot center distance. This iterative process continues until the pixel distance between the two spot centers is less than the third deviation threshold. At this point, it is considered that the excitation spot and the probe spot have achieved concentric coincidence on the surface of the area to be measured, and the optical path alignment step is completed.

[0023] To further improve the stability and adaptability of alignment, after completing the above alignment, the system can rotate the excitation and detection optical paths around the surface normal of the area under test by a small angle (e.g., 5 to 10 degrees) and observe the movement trajectory of the two overlapping points. If the overlapping point remains in a flat area with minimal curvature change while moving within the area under test, the current alignment is good. If the overlapping point slides to the edge of the skirt with greater curvature or an uneven area, the system automatically fine-tunes the relative orientation of the two optical paths to move the overlapping point back to the optimal detection position. This optimization step can effectively reduce echo scattering and signal attenuation caused by the curved surface of the insulator.

[0024] Through the above-mentioned multi-level alignment strategy, this invention achieves efficient and accurate concentric alignment of the excitation spot and the detection spot on the surface of complex curved insulators, laying a solid foundation for subsequent high signal-to-noise ratio ultrasonic signal acquisition.

[0025] Step S102: Adjust the pulsed laser excitation optical path and the continuous laser interference detection optical path so that the excitation spot and the detection spot are concentrically aligned on the surface of the area to be measured.

[0026] In this step, the system first activates the coaxial concentric adjustment module installed inside the detection device. This module includes a set of precision optical adjustment elements, specifically an adjustable mirror, a dichroic mirror, and a frame with three-dimensional fine-tuning capabilities. The dichroic mirror is designed to have high reflectivity for pulsed laser wavelengths (e.g., 1064 nm) and high transmittance for continuous laser wavelengths (e.g., 532 nm), or vice versa, thereby enabling two beams of light to propagate along the same optical path.

[0027] In actual operation, the system first emits a low-power visible light guiding laser beam, which propagates along the pulsed laser excitation path to form a first guiding spot on the insulator surface. This guiding laser has a wavelength of 650 nanometers (red) and a power of less than 1 milliwatt, making it easily observable with the naked eye or a camera. Simultaneously, the system also emits another visible light guiding laser beam (e.g., 532 nanometers, green), which propagates along the continuous laser interferometry detection path to form a second guiding spot on the insulator surface. The power of both guiding laser beams is controlled within a safe range, ensuring no damage to the insulator surface.

[0028] The system uses a high-resolution industrial camera (which can be equipped with a zoom lens and filters) to acquire images of the insulator surface, including two guide light spots, from a suitable angle. To eliminate interference from ambient light, the camera can use narrowband filters to perform time-division multiplexing for red and green wavelengths, or a color camera can be used to directly acquire color images and separate the two light spots through color channels.

[0029] In the acquired images, the system runs image processing algorithms to calculate the precise center positions of the red and green light spots, respectively. Commonly used algorithms include: performing binarization segmentation on the light spot region and then calculating the gray-level centroid coordinates, or fitting an ellipse to the edge of the light spot and then determining the center of the ellipse. To improve accuracy, the system can acquire images multiple times and take the average value to reduce the impact of random noise.

[0030] After obtaining the center coordinates of the two light spots, the system calculates the pixel distance between them and the deviation components in the horizontal and vertical directions of the image. Based on this deviation information, the system controls the adjustable mirror and three-dimensional fine-tuning frame in the optical path coaxial concentric adjustment module to fine-tune the direction of the excitation or probe optical path. Specifically: If the red spot (representing the excitation light path) is to the left relative to the green spot (representing the detection light path), the system controls the horizontal adjustment motor to rotate the reflector clockwise or counterclockwise by a small angle, causing the red spot to move to the right.

[0031] If the red light spot is too high, the system adjusts the fine-tuning screw in the vertical direction to move the light spot downward.

[0032] If the two light spots are not the same size or are asymmetrical in shape, the system will also adjust the position of the focusing lens to change the size of the light spots so that they match as closely as possible.

[0033] After each fine-tuning, the system reacquires images, recalculates the center positions of the two light spots, and compares them with the previous results. This process is repeated to form a closed-loop control. When the pixel distance between the centers of the two light spots is less than a preset concentricity threshold (e.g., 1 pixel, or, depending on the camera's field of view and detection accuracy requirements, equivalent to an actual distance of less than 0.05 mm), the system considers that the excitation light spot and the detection light spot have achieved high-precision concentric alignment.

[0034] To verify the stability and reliability of concentric overlap, the system can also move the detection device as a whole or change its relative distance to the insulator after the above adjustments are completed, and observe whether the two light spots always remain aligned. If the overlap decreases due to mechanical clearance or optical system errors, the system will automatically initiate a correction program to readjust.

[0035] Furthermore, in some implementations, the system can utilize the continuous probe laser output from the laser interferometer itself to assist alignment. Specifically, a beam splitter is placed in the output optical path of the continuous laser interferometer to guide a small portion of the probe laser to a position-sensitive detector. When the excitation spot and the probe spot are completely overlapped, the ultrasonic vibration signal generated by the excitation pulse laser on the insulator surface is strongest. This signal is received by the interferometer, and the degree of overlap between the two spots can be indirectly characterized by monitoring the intensity of the received signal. The system can combine image recognition and signal intensity information to achieve more robust concentric alignment adjustment.

[0036] Through the aforementioned fine-tuning process, this invention ensures that the excitation and detection spots are precisely concentrically aligned on the insulator surface, thus placing the detection point precisely in the central region of the sound source. This maximizes the reception of radially propagating and reflected bulk wave signals, effectively improving the signal-to-noise ratio and detection sensitivity of the defect echo.

[0037] Step S103: Control the pulsed laser to irradiate the test area of ​​the insulator with an energy density lower than the damage threshold of the insulator material. Excite the body wave propagating into the interior of the insulator under the thermoelastic mechanism. Based on the synchronous trigger signal, collect multiple sets of surface vibration signals at different detection orientations using a continuous laser interferometer to obtain the initial ultrasonic signal sequence corresponding to each detection orientation. The surface vibration signal includes the defect echo component formed by the body wave propagating inside the insulator and being reflected or scattered after encountering a defect.

[0038] In this step, the system first activates the pulsed laser, outputting pulsed laser light with preset parameters. The wavelength of the pulsed laser is typically selected from the near-infrared band, which is relatively safe for the human eye and has good penetration into insulator materials, such as 1064 nanometers. The pulse width is set in the nanosecond range, typically 10 nanoseconds, and the repetition frequency can be set from 10 Hz to 100 Hz depending on the required detection speed. After beam expansion and collimation, the pulsed laser light is processed by a focusing lens or shaping optical element to form a large circular spot with a diameter between 5 mm and 8 mm. This large spot, through the previously adjusted optical path, precisely irradiates the surface of the insulator area to be tested.

[0039] Before formal testing begins, the system needs to undergo energy calibration to ensure that the excitation process remains within a thermoelastic mechanism and does not damage the insulator surface. Specifically, the system starts with a low energy density and gradually increases the output energy of the pulsed laser, while an online energy monitoring probe records the energy value in real time. At each energy level, the system emits several pulses of laser light, and a high-sensitivity optical microscope or surface morphology analyzer is used to observe the insulator surface for any signs of melting, ablation, or discoloration. When surface damage is first observed, the system records the current energy density and determines it as the damage threshold for the insulator material. The system then sets the actual energy density used to be 80% to 90% below this damage threshold to ensure the safety and stability of long-term testing. This calibration process can be performed on insulator samples from the same batch before testing, or periodically on standard samples, without needing to be repeated for each test.

[0040] After energy calibration, the system officially begins signal acquisition. A pulsed laser irradiates the insulator surface with a large spot size. Because the laser energy density is below the damage threshold, the material undergoes transient thermal expansion and contraction, generating elastic stress waves, i.e., ultrasonic waves. This excitation mechanism is called the thermoelastic mechanism, and its greatest advantage is that it does not cause any permanent damage to the surface. Due to the large spot size and wider excitation area, it is beneficial for exciting high-energy volume waves, including longitudinal and transverse waves. These volume waves propagate perpendicular to the surface or at a certain angle into the insulator.

[0041] Simultaneously with the pulsed laser emission, a photodetector is positioned near the insulator to receive the weak pulsed laser signal scattered from the insulator surface. The photodetector typically employs a high-speed response PIN photodiode with a response time of less than 1 nanosecond. Upon receiving the scattered light signal, the detector immediately converts it into an electrical pulse. This pulse is processed by a high-speed comparator and shaping circuitry to form a steep-edge TTL-level trigger signal. This trigger signal is simultaneously sent to the synchronization input of the data acquisition card and the laser interferometer, serving as the time zero point for the entire acquisition system. Due to the extremely fast response of the photodetector, the delay between the trigger signal and the laser excitation moment can be controlled at the nanosecond level, ensuring the accuracy of the echo arrival time in subsequent signal analysis.

[0042] Driven by a synchronous trigger signal, the continuous laser interferometer begins operation. The continuous probe laser output by the interferometer (e.g., green light with a wavelength of 532 nm) is focused into a small spot (approximately 0.1 mm in diameter) through a previously aligned optical path, precisely landing concentrically with the excitation spot. When the bulk wave propagates inside the insulator and reaches the surface, it causes a minute normal displacement of the surface (typically on the order of picometers to nanometers). The interferometer uses the interference principle of the reference light and the signal light to convert the surface displacement change into a change in light intensity, which is then converted into a voltage signal by a built-in photodetector. Due to the use of highly sensitive dual-wave mixing or heterodyne interferometry, the interferometer can achieve high signal-to-noise ratio non-contact measurements under such weak vibrations.

[0043] The acquired raw ultrasonic signals are amplified by a preamplifier and low-pass filtered before being digitized by a high-speed data acquisition card (sampling rate typically not less than 100 Mbps, resolution not less than 12 bits) starting from the rising edge of the synchronous trigger signal. Each trigger pulse corresponds to a complete ultrasonic waveform, and the recording length is estimated based on the insulator wall thickness, typically set to tens of microseconds to ensure that all information from surface waves, defect echoes, and bottom waves is included.

[0044] To obtain multiple sets of signals from different detection azimuths, the system also integrates an azimuth scanning mechanism. This scanning mechanism can be an electric rotary table, a pan-tilt unit, or a handheld angle encoder. During detection, the system controls the azimuth scanning mechanism to rotate the entire optical head (including the pulsed laser excitation module and the interferometric detection module) around the normal direction of the surface of the area to be measured, stopping at each preset azimuth angle and acquiring a set of ultrasonic signals. The interval of the azimuth angles can be set according to the insulator diameter and defect size; for example, one set can be acquired every 15 degrees, for a total of 24 sets per revolution. At each azimuth angle, the system can continuously acquire signals multiple times (e.g., 32 or 64 times) and accumulate and average these signals to suppress random electronic noise and environmental vibration interference, thereby improving the signal-to-noise ratio.

[0045] During the acquisition process, the system monitors the quality of the interferometer signal in real time, for example, by calculating the average amplitude and spectral signal-to-noise ratio of the signal. If the signal quality is poor at a certain azimuth angle (e.g., due to surface dust, poor angle, or ambient light interference), the system can prompt the operator to clean the surface or fine-tune the device and reacquire the signal.

[0046] The signals acquired and averaged at each azimuth angle are stored as an initial ultrasonic signal. Simultaneously, the azimuth parameters, time axis information, and corresponding detection system status parameters (such as laser energy and spot size) of this signal are recorded. All these signals are organized into an initial ultrasonic signal sequence according to azimuth angle, serving as input for subsequent steps.

[0047] In this way, through the above process, the system obtained multiple initial ultrasonic signal sequences from different detection orientations under completely non-contact and non-destructive conditions. Each signal contained reflected or scattered echo components that may be caused by internal defects, laying a high-quality data foundation for subsequent time-frequency analysis, candidate echo identification, and spatial coherence analysis.

[0048] Step S104: Perform time-frequency domain joint transformation on each initial ultrasound signal in the initial ultrasound signal sequence, extract the energy distribution feature matrix of each signal within a preset time-frequency window, and adaptively calculate the defect echo discrimination threshold corresponding to each initial ultrasound signal based on the energy distribution feature matrix and the statistical characteristics of the background noise.

[0049] In this step, for any initial ultrasonic signal, the known wall thickness range of the insulator's test area is determined. and the longitudinal wave velocity of materials This allows for the estimation of the time window of the defect echo. The expression is: , , In the formula, This represents the earliest time when the defect echo appears. This is the latest time when the defect echo appears. The minimum known wall thickness of the insulator's test area. The maximum known wall thickness of the insulator's test area; Within the stated time window, the initial ultrasonic signal is subjected to adaptive matched tracking decomposition to construct an overcomplete atom library that matches the wave propagation characteristics of the insulator. Each atom in the atom library... Defined as: , In the formula, For continuous time variables, For the atomic center time, For atomic time-width parameters, The atomic center frequency, The imaginary unit, It is an exponential decay factor. It is an exponential function; An orthogonal matching pursuit algorithm is used to successively extract the signal component that best matches the atom from the initial ultrasonic signal. After each extraction, the residual signal is calculated. The iteration stops when the energy of the residual signal is lower than a preset percentage of the energy of the original signal. A set of matching atoms and a set of time-frequency parameters corresponding to the matching atoms are obtained. The time-frequency parameters include the atom center time, the atom center frequency, the atom time width parameter, and the atom amplitude. The time-frequency parameters of all matching atoms are mapped onto the time-frequency plane to form a discretized energy distribution feature matrix. The position of each non-zero element in the energy distribution feature matrix is ​​determined by the atomic center time and atomic center frequency, and the element value is obtained by weighting the atomic amplitude and atomic energy.

[0050] Furthermore, fractal dimension analysis is performed on the energy distribution feature matrix. For each time-frequency point in the energy distribution feature matrix, a 3×3 neighborhood window is extracted with the time-frequency point as the center, and the box dimension within the neighborhood window is calculated. Statistically analyze the distribution histogram of box dimension for all time-frequency points, classify time-frequency points with box dimension less than a preset fractal threshold as background noise dominant points, and classify time-frequency points with box dimension greater than or equal to as signal dominant points; The 95th percentile of the energy values ​​of all background noise dominant points is used as the basic noise energy, and the 5th percentile of the energy values ​​of all signal dominant points is used as the weak signal energy. The basic noise energy and the weak signal energy are then weighted and averaged to obtain the defect echo discrimination threshold.

[0051] In one specific embodiment, the system first acquires prior geometric information about the insulator's test area for a given initial ultrasonic signal being processed. This information includes the known insulator wall thickness range and the longitudinal wave velocity of the material. The wall thickness range can typically be obtained from the insulator's design drawings or specifications; for example, the wall thickness of a certain type of post porcelain insulator is between 5 mm and 6 mm. The longitudinal wave velocity of the material can be obtained by calibrating and measuring samples from the same batch of defect-free specimens; for porcelain insulators, it is typically between 5500 m / s and 6000 m / s. With the wall thickness range and velocity value, the system estimates the possible time window for defect echoes: the earliest possible echo time is equal to twice the minimum wall thickness divided by the velocity of sound, and the latest possible echo time is equal to twice the maximum wall thickness divided by the velocity of sound. This time window corresponds to the round-trip time range of the bulk wave propagating from the surface to the internal defect and back to the surface. Only echoes occurring within this time window are likely related to internal bulk defects; signals outside the window (such as surface waves, multiple bottom wave echoes) will be ignored in subsequent processing.

[0052] Next, the system enters the core stage of joint time-frequency domain transformation, namely, decomposing the signal within the time window using an adaptive matching pursuit algorithm. The basic idea of ​​matching pursuit is to represent the original signal as a weighted sum of a set of basis functions (called atoms), with each atom representing a time-frequency component. To improve decomposition accuracy, the system first constructs an overcomplete atom library specifically tailored to the characteristics of insulator body waves. This atom library contains a large number of atoms with diverse forms, each mathematically designed to match the actual propagation characteristics of ultrasonic body waves in the insulator material. Specifically, the atoms exhibit a Gaussian envelope in time, with smooth onset and decay processes; they exhibit narrow-band characteristics in frequency, with center frequencies covering the typical frequency range of insulator body waves (e.g., 1 MHz to 5 MHz); furthermore, the atoms also include an exponential decay factor to simulate the energy attenuation phenomenon of ultrasonic waves propagating through the ceramic insulator body. The overcompleteness of the atom library means that the number of atoms is far greater than the minimum number required to represent the signal, providing ample flexibility for accurately matching various time-frequency structures in the signal.

[0053] After the atom library is constructed, the system initiates the iterative decomposition process of the orthogonal matching pursuit algorithm. The algorithm first searches for the atom in the original signal that has the largest inner product with the current residual signal (the residual in the first iteration is the original signal). This atom represents the most significant time-frequency component of the signal. After finding this optimal matching atom, the algorithm records four parameters of the atom: atom center time (reflecting the arrival time of the echo component), atom center frequency (reflecting the dominant frequency of the component), atom duration (reflecting the duration of the component), and atom amplitude (reflecting the intensity of the component). Then, the algorithm subtracts the contribution of this atom to the signal from the residual signal to obtain a new residual signal. After each subtraction of the matching atom, the algorithm checks the energy of the new residual signal. If the residual energy is already below a preset percentage (e.g., 5%) of the original signal energy, it indicates that most of the structure in the signal has been extracted, and the remaining part is mainly unmodelable random noise; at this point, the iteration stops. Otherwise, the algorithm continues to search for the next optimal matching atom in the residual signal, and so on.

[0054] Through the above iterative decomposition, the system extracts a set of matching atoms from the original ultrasonic signal, with each atom corresponding to a physical echo event or signal component. These atoms are arranged in descending order of energy, with larger atoms typically corresponding to surface waves, bottom waves, or strong defect echoes, while smaller atoms may correspond to weak defect echoes or noise.

[0055] Next, the system maps the time-frequency parameters of all matched atoms onto a two-dimensional time-frequency plane, constructing an energy distribution feature matrix. The horizontal axis of the time-frequency plane corresponds to the discretized time points, and the vertical axis corresponds to the discretized frequency points. For each extracted atom, the system determines its position on the time axis based on its atom center time and its position on the frequency axis based on its atom center frequency, then assigns an energy value to that position. This energy value is not simply equal to the atom amplitude, but is obtained by weighting the atom amplitude and the atom energy—the usual practice is to square the atom amplitude and then distribute it to several neighboring time-frequency grids according to the atom time width under a Gaussian envelope shape. After mapping all atoms, a heatmap reflecting the signal energy distribution is formed on the time-frequency plane, where bright areas represent time-frequency components with concentrated energy, and dark areas represent noise background.

[0056] After obtaining the energy distribution characteristic matrix, the system needs to analyze the statistical characteristics of the background noise in order to adaptively calculate the discrimination threshold that distinguishes echoes from noise. Traditional fixed threshold methods are difficult to adapt to changes in different insulator surface conditions and environmental noise levels. This invention introduces an adaptive threshold calculation method based on fractal dimension analysis.

[0057] Specifically, for each time-frequency point in the energy distribution feature matrix, the system extracts a local neighborhood window of size 3x3 (i.e., including the point itself and nine points in the directions above, below, left, right, and diagonally opposite). For this nine-point window, the system calculates its box-dimensionality. The calculation process is as follows: the energy value of each point within the window is considered as a three-dimensional surface (time and frequency on the horizontal and vertical axes, and energy on the vertical axis). This surface is covered with boxes of different sizes. The minimum number of boxes required for coverage at each size is calculated, and then the slope of the line is obtained through linear fitting on logarithmic coordinates. This slope is the box-dimensionality. The box-dimensionality typically ranges from 2 to 3: when the energy distribution within the window is relatively regular and exhibits a clear peak structure, the box-dimensionality will be higher (close to 3); when the energy within the window is completely random and has no obvious structure, the box-dimensionality will be lower (close to 2). Therefore, the box-dimensionality can serve as an effective indicator for distinguishing between "meaningful signal points" and "random noise points."

[0058] The system traverses the entire time-frequency matrix, calculates the box dimension of each time-frequency point, and statistically analyzes the distribution histogram of all box dimensions. Based on a preset empirical fractal threshold (e.g., 1.8), all time-frequency points are divided into two categories: points with a box dimension less than the threshold are considered as background noise-dominated points, which typically correspond to randomly fluctuating noise; points with a box dimension greater than or equal to the threshold are considered as signal-dominated points, which typically correspond to structured echo components.

[0059] Next, the system extracts all energy values ​​from the set of dominant background noise points, sorts them in ascending order, and finds the energy value at the 95th percentile, which is taken as the baseline noise energy. The 95th percentile is used instead of the maximum value to exclude occasional abnormal spike noise. Simultaneously, the system extracts all energy values ​​from the set of dominant signal points and finds the energy value at the 5th percentile, which is taken as the weak signal energy. These two percentiles represent the upper limit of typical noise and the lower limit of typical weak signal, respectively.

[0060] Finally, the system performs a weighted average of the baseline noise energy and the weak signal energy to obtain the defect echo discrimination threshold. This weighted average is not a simple arithmetic average, but a geometric average with a dynamic weighting coefficient. The value of the weighting coefficient is related to the roughness of the insulator surface: the rougher the surface, the stronger the background scattering noise, and the system adjusts the weighting coefficient to make the threshold more biased towards the baseline noise energy, avoiding misclassification of noise as echo; the smoother the surface, the lower the noise level, and the system adjusts the weighting coefficient to make the threshold more biased towards the weak signal energy, improving the detection sensitivity of weak echoes. This dynamic adaptive mechanism ensures that the threshold can be automatically optimized as the detection object and detection environment change, always maintaining the best discrimination effect.

[0061] Step S105: Based on the defect echo discrimination threshold, identify candidate defect echoes in each initial ultrasonic signal and record the characteristic parameters of each candidate defect echo.

[0062] In this step, all time-frequency points in the energy distribution feature matrix that satisfy the condition that the element value is greater than the defect echo discrimination threshold and the box dimension within the neighborhood window is greater than the preset fractal threshold are marked as echo seed points. Starting from the echo seed point, the echo region is obtained by continuously expanding along the time axis. Specifically, the continuous expansion refers to: for the current echo seed point... If at the same frequency point in the next moment Or the adjacent frequency point at the next moment , The location has an energy value greater than If the target time domain point is included in the same echo region, then the target time domain point will be included in the same echo region. For the k-th discrete time point, For the first Discrete frequency points, For the (k+1)th discrete time point, For the first Discrete frequency points, This is the attenuation tolerance coefficient. The threshold for defect echo discrimination; For each expanded echo region, the time centroid of the echo region is calculated. Echo regions with a time centroid less than the estimated arrival time of the surface wave and echo regions with a time centroid greater than the estimated arrival time of the bottom wave are removed. The signal event corresponding to each remaining echo region is determined as a candidate defect echo, and the characteristic parameters of each candidate defect echo are recorded. The time centroid of the candidate defect echo is calculated. The expression is: , In the formula, The region is a connected region composed of consecutive time-frequency points in the candidate defect echo. Here, k is the time-frequency grid index, and k is the time index. For frequency index, For time and frequency points The energy value at that location; Calculate the center frequency of the candidate defect echo. The expression is: , Calculate the total energy of the candidate defect echo. The expression is: .

[0063] In one specific embodiment, the system first performs a traversal scan of the energy distribution feature matrix corresponding to the current initial ultrasound signal. For each time-frequency point in the matrix, the system simultaneously checks two conditions: whether the energy value of the time-frequency point is greater than the defect echo discrimination threshold calculated in step S104; and whether the box dimension of the local window where the time-frequency point is located is greater than or equal to a preset fractal threshold. Only time-frequency points that simultaneously satisfy both conditions are marked as echo seed points. Here, the energy condition ensures that the point has sufficient intensity, and the box dimension condition ensures that the point has the structural characteristics that the signal should have, rather than random noise. The combination of the two can effectively prevent isolated strong noise from being misjudged as echo seeds.

[0064] After obtaining all echo seed points, the system expands the region along the positive time axis (i.e., the subsequent moment of signal propagation) starting from each seed point. The specific expansion rule is as follows: starting from the current seed point, check the same frequency position after the time index is incremented by one, as well as the adjacent frequency positions after the time index is incremented by one (i.e., the positions of frequency index decrement and frequency index increment). If any of these positions contains a time-frequency point, and the energy value of that point is greater than the defect echo discrimination threshold multiplied by an attenuation tolerance coefficient, then that point is included in the same echo region as the current seed point. The typical value of the attenuation tolerance coefficient is between 0.6 and 0.8, allowing for some attenuation of echo energy during propagation, but not excessively rapid attenuation. This expansion mechanism simulates the continuous form of real defect echoes in the time-frequency plane: because ultrasonic pulses have a certain duration, echo energy is continuously distributed along the time axis, with the energy most concentrated near the center frequency; while random noise often presents as isolated, discontinuous points. Through this continuous expansion along the time axis, the system can aggregate all time-frequency points belonging to the same physical echo event into a connected region. The system repeats this expansion process until no new time-frequency points can be added, thus forming a complete echo region. For different seed points, if the regions they form after expansion are interconnected, they will be merged into the same echo region.

[0065] After obtaining several echo regions, the system needs to perform preliminary screening for each region. First, the system calculates the time centroid of each echo region. The time centroid is calculated by using the energy values ​​of all time-frequency points within the region as weights and then averaging the corresponding time coordinates. Time points with higher energy contribute more to the centroid. The resulting time centroid reflects the position of the echo signal's center of gravity on the time axis. Then, the system compares the time centroid with two predicted values: one is the predicted arrival time of the surface wave, which can be calculated based on the surface distance between the excitation point and the detection point and the surface wave velocity; the other is the predicted arrival time of the bottom wave, which can be calculated based on the insulator wall thickness and the longitudinal wave velocity. Echo regions with a time centroid smaller than the predicted arrival time of the surface wave are identified as surface wave interference and are eliminated; echo regions with a time centroid larger than the predicted arrival time of the bottom wave are identified as bottom wave or multiple echo interference and are also eliminated. Surface waves and bottom waves are inherent structural echoes in insulator detection and are not defect signals, therefore they must be excluded. The remaining echo region after this elimination is then identified by the system as a candidate defect echo, which is a signal to be verified that may originate from an internal defect.

[0066] For each candidate defect echo, the system further calculates its characteristic parameters, including time centroid, center frequency, total energy, and time width. These parameters will serve as the basis for subsequent multi-directional spatial coherence analysis.

[0067] The calculation of the time centroid is as described above: the energy at all time-frequency points within the region is weighted and averaged according to the time coordinate, and the result is the arrival time of the candidate defect echo. This time value will be used for subsequent depth estimation.

[0068] The center frequency is calculated similarly: the energy at all time-frequency points within the region is weighted and averaged according to frequency coordinates to obtain the dominant frequency of the candidate defect echo. This parameter can help distinguish between different types of echoes—defect echoes often have specific frequency characteristics, while the frequency distribution of noise is more random.

[0069] The total energy is calculated by directly summing the energy values ​​at all time-frequency points within the region. The total energy reflects the strength of the echo component and is one of the important indicators for determining the presence of defects.

[0070] The time width is calculated by finding the maximum and minimum values ​​of the time coordinates within the region and then calculating the difference between them. The time width reflects the duration of the echo signal. Real defect echoes generally have a finite and continuous time width, while noise or interference signals are often too narrow or too wide.

[0071] After completing the above calculations, the system records all characteristic parameters of the candidate defect echo—time centroid, center frequency, total energy, and time width—in a data structure and associates it with the corresponding detection azimuth angle, storing the data. This completes the identification and characteristic parameter recording of the candidate defect echo in an initial ultrasonic signal. The system then repeats the entire process for the next initial ultrasonic signal until signals from all detection azimuths have been processed, obtaining a set of candidate defect echo characteristic parameters for different azimuths across the entire test area.

[0072] Step S106: The candidate defect echoes identified under different detection orientations are associated according to their spatial locations to construct a multi-directional spatial coherence map. The horizontal axis of the map is the detection azimuth angle, the vertical axis is the echo arrival time, and the map intensity is the peak amplitude or frequency band energy ratio of the corresponding candidate echo.

[0073] In this step, a two-dimensional graph matrix is ​​initialized. The two-dimensional map matrix The rows correspond to the detection azimuth angle, the columns correspond to the discretized echo arrival time, and the matrix elements are initialized to zero; For each candidate defect echo under each detection azimuth angle, the time column index of the candidate defect echo is determined according to the time centroid of the candidate defect echo, and the total energy of the candidate defect echo is assigned to the element in the two-dimensional spectrum matrix. If multiple candidate defect echoes correspond to the same position, the maximum total energy of the candidate defect echoes is taken. For each time column index, the similarity of feature parameters between adjacent azimuth angles is calculated sequentially. The formula for calculating the similarity is: , In the formula, For the first The azimuth index and the first Discretized index of azimuth angle index at echo arrival time Similarity on This is the preset maximum tolerance value for time difference. , , All are preset weighting coefficients, and satisfy the following conditions: , This is the preset maximum tolerance value for frequency difference. , , The first Time centroid, center frequency, and total energy of candidate defect echoes under each azimuth index. , , The first Time centroid, center frequency, and total energy of candidate defect echoes under each azimuth index; The coherently enhanced target two-dimensional map matrix is ​​calculated based on the similarity. and the target two-dimensional map matrix As the multi-directional spatial coherence map, the target two-dimensional map matrix The formula for calculating elements in the middle is: , In the formula, For the target two-dimensional map matrix Chinese azimuth index Discretized index of echo arrival time At this point, the total energy of the candidate defect echo. Two-dimensional map matrix Chinese azimuth index Discretized index of echo arrival time At this point, the total energy of the candidate defect echo. This is the preset coherence enhancement coefficient.

[0074] Step S107: Determine whether there is a coherent peak region in the multi-directional spatial coherence map. The coherent peak region is defined as a region with consistent echo arrival time and peak amplitude exceeding a preset coherence threshold in multiple consecutive detection directions.

[0075] In this step, the system first performs a global scan of the multi-azimuth spatial coherence map to identify all local maxima. A local maximum is defined as a point whose spectral value is greater than the spectral values ​​of all eight of its eight neighboring points within a three-row, three-column neighborhood centered on it. To reduce edge effects, the system only considers points not on the outermost ring of the map. These local maxima represent candidate locations with relatively prominent energy in the map, each potentially corresponding to a defective event. The system records the row and column coordinates (i.e., the corresponding azimuth and time indices) of each local maximum as well as its spectral value.

[0076] Next, the system sorts all detected local maxima points in descending order of their spectral values. Since the coherent peak regions generated by real defects typically have the highest energy intensity, the system selects only the top few points after sorting as candidate coherent centers. The number of selected points can be set according to actual needs, such as selecting the first three to the first five. This avoids misclassifying a large number of low-energy noise points as candidate centers, thus improving detection efficiency.

[0077] For each candidate coherent center, the system uses its location as a starting point to perform region growing to determine whether the center can be expanded to form a continuous region that meets the conditions for a coherent peak region. The specific rules for region growing are as follows: Starting from the candidate coherent center point, examine the adjacent positions in the four directions (up, down, left, and right) (or further examine eight directions). If an adjacent position simultaneously meets two conditions—the spectral value of the position is not less than the spectral value of the candidate coherent center point multiplied by a preset ratio threshold (e.g., 60% to 70%), and the position is within a continuous index range with the center point in the azimuth direction—then the adjacent position is included in the current region. Then, using the newly included point as a new starting point, the system continues to expand outwards until no more points can be added. After the growing is completed, the system obtains a connected region.

[0078] The system evaluates the grown region in two aspects. First, it assesses the region's span in the azimuth direction. Real defects, due to their spatial scale, will have echoes appearing at multiple consecutive azimuth angles. Therefore, the region's continuous span in the azimuth direction should be no less than a preset minimum number of consecutive azimuth angles (e.g., three or four). If the region only covers one or two isolated azimuth indices, it is likely isolated noise and should not be considered. Second, it assesses the consistency of echo arrival times at all points within the region. Since echoes from the same defect should have similar arrival times at different azimuth angles (because the distance from the defect to the surface does not change significantly), the system calculates the standard deviation or range of the time indices for all points within the region. If the time variation range is less than a preset time consistency threshold (e.g., corresponding depth variations not exceeding 1 mm), then the condition for consistent echo arrival times is met.

[0079] For regions that simultaneously meet the conditions of "sufficiently large continuous azimuth span" and "sufficiently good consistency of echo arrival time," the system further checks whether the peak value of the spectral values ​​within that region exceeds a preset coherence threshold. This threshold can be dynamically determined based on the background noise level of the entire spectrum, for example, by multiplying the median of all non-zero values ​​in the spectrum by a coefficient (such as three times), or by taking a pre-set absolute energy threshold. Only when the maximum spectral value within the region exceeds this coherence threshold is the system recognized as a valid coherence peak region.

[0080] If at least one coherent peak region exists that satisfies all the above conditions, the system determines that a coherent peak region exists in the multi-directional spatial coherence map, and thus determines that a real defect exists inside the insulator. Simultaneously, the system records the location information of the coherent peak region, especially the weighted average of the echo arrival times of that region (using the map values ​​of each point within the region as weights), for use in subsequent steps to calculate the defect depth.

[0081] If, after region growth and condition verification of all candidate coherent centers, no region is found that meets the requirements of continuous azimuth span, temporal consistency, and numerical threshold, the system determines that there is no effective coherent peak region in the map and outputs a defect-free detection conclusion.

[0082] Step S108: If the coherent peak region exists, it is determined that there is a real defect inside the insulator corresponding to the coherent peak region, and the depth range of the defect is analyzed based on the echo arrival time of the coherent peak region and the preset sound velocity model.

[0083] In this step, the multi-directional spatial coherence map In the process, all local maxima are detected. A local maximum is defined as a point in a 3×3 neighborhood where the value of a matrix element is greater than the values ​​of all adjacent matrix elements. The detected local maxima points are sorted from largest to smallest according to their corresponding matrix element values, and the top Q local maxima points with the largest matrix element values ​​are selected as candidate coherence centers. For each candidate coherence center, region growth is performed using the candidate coherence center as the seed point under preset growth conditions to obtain the coherence peak region. The preset growth conditions include: any coordinate within the coherent peak region Matrix element values ​​at Not less than the coordinates of the candidate coherence center Matrix element values ​​at of times, of which, The preset ratio threshold; The continuous span of points within the coherent peak region in the azimuth direction is not less than A continuous azimuth index The minimum consecutive number of directions is preset; Calculate the weighted average echo arrival time of all points within the coherent peak region. : , In the formula, The region is a connected region composed of consecutive time-frequency points in the candidate defect echo. For the first The echo arrival time corresponding to each discrete time index; Based on the longitudinal wave velocity of the insulator material The defect depth is calculated using the weighted average echo arrival time. The expression is: , In the formula, The sound path correction caused by the surface curvature of the insulator is pre-calculated using the surface geometric parameters of the area to be measured. Calculate the full width at half maximum (FWHM) of the coherent peak region on the time axis. Using the depth range corresponding to the half-width at half-maximum as the uncertainty of the defect depth, the defect depth range is output. The half-width at half-maximum (WHM) is defined as the time width corresponding to when the matrix element value at each point in the coherent peak region drops to half of the maximum matrix element value in the coherent peak region.

[0084] In summary, the method of this application aligns and adjusts the pulsed laser excitation optical path and the continuous laser interferometric detection optical path on the same side until they are concentrically coincident; it excites volume waves with an energy density lower than the damage threshold and collects multi-directional surface vibration signals; it performs time-frequency domain joint transformation and adaptive matched pursuit decomposition on the signals to construct an energy distribution feature matrix, and adaptively calculates the defect echo discrimination threshold by combining fractal dimension analysis; it identifies candidate defect echoes based on this threshold and records their time-frequency characteristic parameters; it associates candidate echoes in different directions according to their spatial positions, calculates the feature similarity of adjacent directions and performs coherence enhancement to construct a multi-directional spatial coherence map; it detects coherence peak regions in the map, and if they exist, it determines the real defect and estimates its depth; thus, it achieves highly sensitive and reliable non-contact detection of tiny defects inside insulators.

[0085] Please see Figure 2 The diagram shows a structural block diagram of an insulator internal defect detection system based on laser ultrasound according to this application.

[0086] like Figure 2 As shown, the insulator internal defect detection system 200 includes an acquisition module 210, an adjustment module 220, an acquisition module 230, an extraction module 240, an identification module 250, an association module 260, a judgment module 270, and an analysis module 280.

[0087] The module 210 is configured to acquire surface reference information of the insulator under test in its in-service or installed state, and align the pulsed laser excitation optical path and the continuous laser interferometry detection optical path to the test area based on the surface reference information. The adjustment module 220 is configured to adjust the excitation optical path and the detection optical path so that the excitation spot and the detection spot are concentrically coincident on the surface of the test area. The acquisition module 230 is configured to control the pulsed laser to irradiate the insulator under test with an energy density lower than the damage threshold of the insulator material, excite the volume wave propagating into the insulator under the thermoelastic mechanism, and acquire multiple sets of surface vibration signals at different detection orientations based on the synchronous trigger signal and the continuous laser interferometer, thereby obtaining an initial ultrasonic signal sequence corresponding to each detection orientation. The surface vibration signal contains the defect echo component formed by the reflection or scattering of the volume wave after it encounters a defect inside the insulator. The extraction module 240 is configured to perform a time-frequency domain joint transformation on each initial ultrasonic signal in the initial ultrasonic signal sequence, extract the energy distribution characteristics of each signal within a preset time-frequency window, and extract the energy distribution characteristics based on the energy distribution characteristics. The system uses statistical characteristics of background noise and adaptively calculates the defect echo discrimination threshold for each initial ultrasonic signal. The identification module 250 is configured to identify candidate defect echoes in each initial ultrasonic signal based on the defect echo discrimination threshold and record the characteristic parameters of each candidate defect echo. The association module 260 is configured to associate candidate defect echoes identified under different detection orientations according to their spatial locations to construct a multi-directional spatial coherence map, where the horizontal axis of the map represents the detection azimuth angle, the vertical axis represents the echo arrival time, and the map intensity represents the correlation coefficient between the detection azimuth angle and the background noise. The judgment module 270 is configured to determine whether there is a coherent peak region in the multi-directional spatial coherence spectrum, wherein the coherent peak region is defined as a region with consistent echo arrival time and peak amplitude exceeding a preset coherence threshold in multiple consecutive detection directions; the analysis module 280 is configured to determine that if the coherent peak region exists, there is a real defect inside the insulator corresponding to the coherent peak region, and analyze the depth range of the defect based on the echo arrival time of the coherent peak region and a preset sound velocity model.

[0088] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0089] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the laser-ultrasound-based insulator internal defect detection method in any of the above method embodiments. In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows: Obtain surface reference information of the insulator under test in its in-service or installed state, and align the pulsed laser excitation optical path and the continuous laser interferometric detection optical path to the test area based on the surface reference information; Adjust the pulsed laser excitation optical path and the continuous laser interference detection optical path so that the excitation spot and the detection spot are concentrically aligned on the surface of the area to be measured. The pulsed laser is controlled to irradiate the test area of ​​the insulator with an energy density lower than the damage threshold of the insulator material. Under the thermoelastic mechanism, a volume wave propagating into the interior of the insulator is excited. Based on the synchronous trigger signal, multiple sets of surface vibration signals under different detection orientations are collected by a continuous laser interferometer to obtain an initial ultrasonic signal sequence corresponding to each detection orientation. The surface vibration signal contains the defect echo component formed by the reflection or scattering of the volume wave after it encounters a defect inside the insulator. A joint time-frequency domain transformation is performed on each initial ultrasonic signal in the initial ultrasonic signal sequence to extract the energy distribution feature matrix of each signal within a preset time-frequency window. Based on the energy distribution feature matrix and the statistical characteristics of the background noise, the defect echo discrimination threshold corresponding to each initial ultrasonic signal is adaptively calculated. Based on the defect echo discrimination threshold, candidate defect echoes are identified in each initial ultrasonic signal, and the characteristic parameters of each candidate defect echo are recorded. Candidate defect echoes identified under different detection orientations are correlated according to their spatial location to construct a multi-directional spatial coherence map. The horizontal axis of the map is the detection azimuth angle, the vertical axis is the echo arrival time, and the map intensity is the peak amplitude of the corresponding candidate defect echo. Determine whether there is a coherent peak region in the multi-directional spatial coherence map. The coherent peak region is defined as a region that has a consistent echo arrival time in multiple consecutive detection directional positions and whose peak amplitude exceeds a preset coherence threshold. If the coherent peak region exists, it is determined that there is a real defect inside the insulator corresponding to the coherent peak region, and the depth range of the defect is analyzed based on the echo arrival time of the coherent peak region and the preset sound velocity model.

[0090] Computer-readable storage media may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application program required for at least one function; the data storage area may store data created based on the use of the laser-ultrasound-based insulator internal defect detection system, etc. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely disposed relative to a processor, which can be connected to the laser-ultrasound-based insulator internal defect detection system via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0091] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3 Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby implementing the laser-ultrasound-based insulator internal defect detection method described in the above embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the laser-ultrasound-based insulator internal defect detection system. The output device 340 may include a display screen or other display device.

[0092] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0093] In one implementation, the above-described electronic device is applied to a laser-ultrasound-based insulator internal defect detection system for a client application, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: Obtain surface reference information of the insulator under test in its in-service or installed state, and align the pulsed laser excitation optical path and the continuous laser interferometric detection optical path to the test area based on the surface reference information; Adjust the pulsed laser excitation optical path and the continuous laser interference detection optical path so that the excitation spot and the detection spot are concentrically aligned on the surface of the area to be measured. The pulsed laser is controlled to irradiate the test area of ​​the insulator with an energy density lower than the damage threshold of the insulator material. Under the thermoelastic mechanism, a volume wave propagating into the interior of the insulator is excited. Based on the synchronous trigger signal, multiple sets of surface vibration signals under different detection orientations are collected by a continuous laser interferometer to obtain an initial ultrasonic signal sequence corresponding to each detection orientation. The surface vibration signal contains the defect echo component formed by the reflection or scattering of the volume wave after it encounters a defect inside the insulator. A joint time-frequency domain transformation is performed on each initial ultrasonic signal in the initial ultrasonic signal sequence to extract the energy distribution feature matrix of each signal within a preset time-frequency window. Based on the energy distribution feature matrix and the statistical characteristics of the background noise, the defect echo discrimination threshold corresponding to each initial ultrasonic signal is adaptively calculated. Based on the defect echo discrimination threshold, candidate defect echoes are identified in each initial ultrasonic signal, and the characteristic parameters of each candidate defect echo are recorded. Candidate defect echoes identified under different detection orientations are correlated according to their spatial location to construct a multi-directional spatial coherence map. The horizontal axis of the map is the detection azimuth angle, the vertical axis is the echo arrival time, and the map intensity is the peak amplitude of the corresponding candidate defect echo. Determine whether there is a coherent peak region in the multi-directional spatial coherence map. The coherent peak region is defined as a region that has a consistent echo arrival time in multiple consecutive detection directional positions and whose peak amplitude exceeds a preset coherence threshold. If the coherent peak region exists, it is determined that there is a real defect inside the insulator corresponding to the coherent peak region, and the depth range of the defect is analyzed based on the echo arrival time of the coherent peak region and the preset sound velocity model.

[0094] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0095] 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 them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting internal defects in insulators based on laser ultrasound, characterized in that, include: Obtain surface reference information of the insulator under test in its in-service or installed state, and align the pulsed laser excitation optical path and the continuous laser interferometric detection optical path to the test area based on the surface reference information; Adjust the pulsed laser excitation optical path and the continuous laser interference detection optical path so that the excitation spot and the detection spot are concentrically aligned on the surface of the area to be measured. The pulsed laser is controlled to irradiate the test area of ​​the insulator with an energy density lower than the damage threshold of the insulator material. Under the thermoelastic mechanism, a volume wave propagating into the interior of the insulator is excited. Based on the synchronous trigger signal, multiple sets of surface vibration signals under different detection orientations are collected by a continuous laser interferometer to obtain an initial ultrasonic signal sequence corresponding to each detection orientation. The surface vibration signal contains the defect echo component formed by the reflection or scattering of the volume wave after it encounters a defect inside the insulator. A joint time-frequency domain transformation is performed on each initial ultrasonic signal in the initial ultrasonic signal sequence to extract the energy distribution feature matrix of each signal within a preset time-frequency window. Based on the energy distribution feature matrix and the statistical characteristics of the background noise, the defect echo discrimination threshold corresponding to each initial ultrasonic signal is adaptively calculated. Based on the defect echo discrimination threshold, candidate defect echoes are identified in each initial ultrasonic signal, and the characteristic parameters of each candidate defect echo are recorded. Candidate defect echoes identified under different detection orientations are correlated according to their spatial location to construct a multi-directional spatial coherence map. The horizontal axis of the map is the detection azimuth angle, the vertical axis is the echo arrival time, and the map intensity is the peak amplitude of the corresponding candidate defect echo. Determine whether there is a coherent peak region in the multi-directional spatial coherence map. The coherent peak region is defined as a region that has a consistent echo arrival time in multiple consecutive detection directional positions and whose peak amplitude exceeds a preset coherence threshold. If the coherent peak region exists, it is determined that there is a real defect inside the insulator corresponding to the coherent peak region, and the depth range of the defect is analyzed based on the echo arrival time of the coherent peak region and the preset sound velocity model.

2. The method for detecting internal defects in insulators based on laser ultrasound according to claim 1, characterized in that, The step of aligning the pulsed laser excitation optical path and the continuous laser interferometry detection optical path to the area to be measured based on the surface reference information includes: The center coordinates of the area to be tested and the direction of the surface normal are obtained from the surface reference information. Using the center coordinates as the aiming point, the two-dimensional galvanometer is controlled to adjust the incident direction of the pulsed laser so that the deviation between the center of the excitation spot and the center coordinates is less than a preset first deviation threshold. After the excitation spot alignment is completed, the current receiving angle of the detection optical path is obtained, the ideal receiving angle is calculated according to the surface normal direction, and the automatic angle adjustment mechanism of the detection optical path is controlled to adjust the receiving angle so that the deviation between the actual receiving angle and the ideal receiving angle is less than the preset second deviation threshold. A beam of alignment laser with a power less than a preset power threshold is emitted, and spot images of the excitation optical path and the probe optical path on the surface of the area to be tested are acquired respectively. The pixel distance between the centers of the two spots is calculated. If the pixel distance is greater than a preset third deviation threshold, the alignment process is repeated until the deviation requirement is met.

3. The method for detecting internal defects in insulators based on laser ultrasound according to claim 1, characterized in that, The step of performing a joint time-frequency domain transformation on each initial ultrasound signal in the initial ultrasound signal sequence to extract the energy distribution feature matrix of each signal within a preset time-frequency window includes: For any initial ultrasonic signal, determine the known wall thickness range of the insulator's test area. and the longitudinal wave velocity of materials This allows for the estimation of the time window of the defect echo. The expression is: ， ， In the formula, This represents the earliest time when the defect echo appears. This is the latest time when the defect echo appears. The minimum known wall thickness of the insulator's test area. The maximum known wall thickness of the insulator's test area; Within the stated time window, the initial ultrasonic signal is subjected to adaptive matched tracking decomposition to construct an overcomplete atom library that matches the wave propagation characteristics of the insulator. Each atom in the atom library... Defined as: ， In the formula, For continuous time variables, For the atomic center time, For atomic time-width parameters, The atomic center frequency, The imaginary unit, It is an exponential decay factor. It is an exponential function; An orthogonal matching pursuit algorithm is used to successively extract the signal component that best matches the atom from the initial ultrasonic signal. After each extraction, the residual signal is calculated. The iteration stops when the energy of the residual signal is lower than a preset percentage of the energy of the original signal. A set of matching atoms and a set of time-frequency parameters corresponding to the matching atoms are obtained. The time-frequency parameters include the atom center time, the atom center frequency, the atom time width parameter, and the atom amplitude. The time-frequency parameters of all matching atoms are mapped onto the time-frequency plane to form a discretized energy distribution feature matrix. The position of each non-zero element in the energy distribution feature matrix is ​​determined by the atomic center time and atomic center frequency, and the element value is obtained by weighting the atomic amplitude and atomic energy.

4. The method for detecting internal defects in insulators based on laser ultrasound according to claim 1, characterized in that, The adaptive calculation of the defect echo discrimination threshold corresponding to each initial ultrasonic signal based on the energy distribution feature matrix and the statistical characteristics of the background noise includes: Fractal dimension analysis is performed on the energy distribution feature matrix. For each time-frequency point in the energy distribution feature matrix, a 3×3 neighborhood window is extracted with the time-frequency point as the center, and the box dimension within the neighborhood window is calculated. Statistically analyze the distribution histogram of box dimension for all time-frequency points, classify time-frequency points with box dimension less than a preset fractal threshold as background noise dominant points, and classify time-frequency points with box dimension greater than or equal to as signal dominant points; The 95th percentile of the energy values ​​of all background noise dominant points is used as the basic noise energy, and the 5th percentile of the energy values ​​of all signal dominant points is used as the weak signal energy. The basic noise energy and the weak signal energy are then weighted and averaged to obtain the defect echo discrimination threshold.

5. The method for detecting internal defects in insulators based on laser ultrasound according to claim 4, characterized in that, The step of identifying candidate defect echoes in each initial ultrasonic signal based on the defect echo discrimination threshold and recording the characteristic parameters of each candidate defect echo includes: All time-frequency points in the energy distribution feature matrix that satisfy the condition that the element value is greater than the defect echo discrimination threshold and the box dimension in the neighborhood window is greater than the preset fractal threshold are marked as echo seed points. Starting from the echo seed point, the echo region is obtained by continuously expanding along the time axis. Specifically, the continuous expansion refers to: for the current echo seed point... If at the same frequency point in the next moment Or the adjacent frequency point at the next moment , The location has an energy value greater than If the target time domain point is included in the same echo region, then the target time domain point will be included in the same echo region. For the k-th discrete time point, For the first Discrete frequency points, For the (k+1)th discrete time point, For the first Discrete frequency points, This is the attenuation tolerance coefficient. The threshold for defect echo discrimination; For each expanded echo region, the time centroid of the echo region is calculated. Echo regions with a time centroid less than the estimated arrival time of the surface wave and echo regions with a time centroid greater than the estimated arrival time of the bottom wave are removed. The signal event corresponding to each remaining echo region is determined as a candidate defect echo, and the characteristic parameters of each candidate defect echo are recorded. The time centroid of the candidate defect echo is calculated. The expression is: ， In the formula, The region is a connected region composed of consecutive time-frequency points in the candidate defect echo. This is a time-frequency grid index, where k is the time index. For frequency index, For time and frequency points The energy value at that location; Calculate the center frequency of the candidate defect echo. The expression is: ， Calculate the total energy of the candidate defect echo. The expression is: 。 6. The method for detecting internal defects in insulators based on laser ultrasound according to claim 1, characterized in that, The step of correlating candidate defect echoes identified from different detection orientations according to their spatial locations to construct a multi-directional spatial coherence map includes: Initialize a two-dimensional graph matrix The two-dimensional map matrix The rows correspond to the detection azimuth angle, the columns correspond to the discretized echo arrival time, and the matrix elements are initialized to zero; For each candidate defect echo under each detection azimuth angle, the time column index of the candidate defect echo is determined according to the time centroid of the candidate defect echo, and the total energy of the candidate defect echo is assigned to the element in the two-dimensional spectrum matrix. If multiple candidate defect echoes correspond to the same position, the maximum total energy of the candidate defect echoes is taken. For each time column index, the similarity of feature parameters between adjacent azimuth angles is calculated sequentially. The formula for calculating the similarity is: ， In the formula, For the first The azimuth index and the first Discretized index of azimuth angle index at echo arrival time Similarity on This is the preset maximum tolerance value for time difference. , , All are preset weighting coefficients, and satisfy the following conditions: , This is the preset maximum tolerance value for frequency difference. , , The first Time centroid, center frequency, and total energy of candidate defect echoes under each azimuth index. , , The first Time centroid, center frequency, and total energy of candidate defect echoes under each azimuth index; The coherently enhanced target two-dimensional map matrix is ​​calculated based on the similarity. and the target two-dimensional map matrix As the multi-directional spatial coherence map, the target two-dimensional map matrix The formula for calculating elements in the middle is: ， In the formula, For the target two-dimensional map matrix Chinese azimuth index Discretized index of echo arrival time At this point, the total energy of the candidate defect echo. Two-dimensional map matrix Chinese azimuth index Discretized index of echo arrival time At this point, the total energy of the candidate defect echo. This is the preset coherence enhancement coefficient.

7. The method for detecting internal defects in insulators based on laser ultrasound according to claim 1, characterized in that, The analysis of the depth range of the defect based on the echo arrival time of the coherent peak region and a preset sound velocity model includes: The multi-directional spatial coherence map In the process, all local maxima are detected. A local maximum is defined as a point in a 3×3 neighborhood where the value of a matrix element is greater than the values ​​of all adjacent matrix elements. The detected local maxima points are sorted from largest to smallest according to their corresponding matrix element values, and the top Q local maxima points with the largest matrix element values ​​are selected as candidate coherence centers. For each candidate coherence center, region growth is performed using the candidate coherence center as the seed point under preset growth conditions to obtain the coherence peak region. The preset growth conditions include: any coordinate within the coherent peak region Matrix element values ​​at Not less than the coordinates of the candidate coherence center Matrix element values ​​at of times, of which, The preset ratio threshold; The continuous span of points within the coherent peak region in the azimuth direction is not less than A series of azimuth indices The minimum consecutive number of directions is preset; Calculate the weighted average echo arrival time of all points within the coherent peak region. : ， In the formula, The region is a connected region composed of consecutive time-frequency points in the candidate defect echo. For the first The echo arrival time corresponding to each discrete time index; Based on the longitudinal wave velocity of the insulator material The defect depth is calculated using the weighted average echo arrival time. The expression is: ， In the formula, The sound path correction caused by the surface curvature of the insulator is pre-calculated using the surface geometric parameters of the area to be measured. Calculate the full width at half maximum (FWHM) of the coherent peak region on the time axis. Using the depth range corresponding to the half-width at half-maximum as the uncertainty of the defect depth, the defect depth range is output. The half-width at half-maximum (WHM) is defined as the time width corresponding to when the matrix element value at each point in the coherent peak region drops to half of the maximum matrix element value in the coherent peak region.

8. A laser-ultrasound-based insulator internal defect detection system, characterized in that, include: The acquisition module is configured to acquire surface reference information of the insulator under test area in service or installation state, and align the pulsed laser excitation optical path and the continuous laser interferometric detection optical path to the test area based on the surface reference information; The adjustment module is configured to adjust the excitation optical path and the detection optical path so that the excitation spot and the detection spot coincide concentrically on the surface of the area to be measured. The acquisition module is configured to control a pulsed laser to irradiate the test area of ​​the insulator with an energy density lower than the damage threshold of the insulator material, thereby exciting a body wave that propagates into the interior of the insulator under a thermoelastic mechanism. Based on a synchronous trigger signal, multiple sets of surface vibration signals at different detection orientations are acquired using a continuous laser interferometer to obtain an initial ultrasonic signal sequence corresponding to each detection orientation. The surface vibration signal includes a defect echo component formed by the reflection or scattering of the body wave after it encounters a defect during propagation inside the insulator. The extraction module is configured to perform a joint time-frequency domain transformation on each initial ultrasound signal in the initial ultrasound signal sequence, extract the energy distribution characteristics of each signal within a preset time-frequency window, and adaptively calculate the defect echo discrimination threshold corresponding to each initial ultrasound signal based on the energy distribution characteristics and the statistical characteristics of the background noise. The identification module is configured to identify candidate defect echoes in each initial ultrasonic signal based on the defect echo discrimination threshold, and record the characteristic parameters of each candidate defect echo. The association module is configured to associate candidate defect echoes identified under different detection orientations according to their spatial locations to construct a multi-directional spatial coherence map, wherein the horizontal axis of the map is the detection azimuth angle, the vertical axis is the echo arrival time, and the map intensity is the peak amplitude of the corresponding candidate defect echo. The judgment module is configured to determine whether there is a coherent peak region in the multi-directional spatial coherence map. The coherent peak region is defined as a region that has a consistent echo arrival time in multiple consecutive detection directions and whose peak amplitude exceeds a preset coherence threshold. The analysis module is configured to determine that if the coherent peak region exists, there is a real defect inside the insulator corresponding to the coherent peak region, and to analyze the depth range of the defect based on the echo arrival time of the coherent peak region and a preset sound velocity model.

9. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method according to any one of claims 1 to 7.

Images