Interference-based scanning target verification method and system
By employing terahertz wave scanning and multiphysics modeling techniques, the problem of inaccurate verification of online surge arrester monitoring devices under phase-to-phase interference was solved, enabling accurate monitoring of surge arrester status and defect identification, thereby improving verification accuracy and reliability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGDONG UNIV OF TECH
- Filing Date
- 2026-04-14
- Publication Date
- 2026-07-28
AI Technical Summary
In the existing technology, the online monitoring device for surge arresters does not provide accurate verification results when facing phase-to-phase interference, which can mislead maintenance personnel in their judgment of the surge arrester's status.
Terahertz wave scanning is used to acquire three-dimensional structural and electromagnetic property data. Spatial electromagnetic field modeling is performed by combining finite element analysis, wave equation and Kalman filtering to generate interphase interference compensation data. Defects are identified through deep learning network and the measurement results of monitoring meters are corrected in real time.
It improves the accuracy of surge arrester verification results, effectively eliminates the impact of phase-to-phase interference on monitoring data, achieves accurate monitoring of surge arrester leakage current and lightning strike count, and can identify equipment defects.
Smart Images
Figure CN122469271A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of verification technology, and in particular to a scanning target verification method and system based on interphase interference. Background Technology
[0002] Online surge arrester monitoring devices are used to monitor the health status of surge arresters, primarily recording leakage current and the number of lightning strikes. These measurements are crucial for power system maintenance and operation. However, in actual power systems, electromagnetic interference between different phases (phase-to-phase interference) can affect the accuracy of the monitoring devices, leading to errors in the measurement results and misleading maintenance personnel's judgment of the surge arrester's condition.
[0003] The invention patent application CN120742214A proposes a self-calibration method and system for a surge arrester leakage current meter. The method involves disconnecting the main leakage current meter from the external main measurement circuit and closing the calibration circuit between the redundant calibration sensor and the reference current source after a preset delay. The reference current source is controlled to output standard signals of different current magnitudes. The redundant calibration sensor collects the standard signals and extracts the effective signal components, performs environmental compensation, and iteratively calculates the relative error between the effective signal components and the standard signals to calibrate the current meter. However, this method does not consider the influence of inter-phase interference, resulting in inaccurate calibration results. Summary of the Invention
[0004] Therefore, this invention addresses the problem of inaccurate verification results in existing technologies by providing a scanning target verification method and system based on interphase interference, which can improve the accuracy of verification results. The specific technical solution is as follows: In a first aspect, the present invention proposes a method for generating interphase interference compensation data for scanning targets, comprising the following steps: The electromagnetic signals of the scanned target are preprocessed to obtain three-dimensional structural data and electromagnetic property data; Based on the three-dimensional structural data and electromagnetic property data, the spatial electromagnetic field of the scanned target is modeled, and the electromagnetic interference signal characteristics between each phase of the scanned target are calculated. Interphase interference compensation data is generated based on the characteristics of the electromagnetic interference signal.
[0005] Furthermore, the electromagnetic signal is obtained through terahertz wave scanning; the three-dimensional structural data includes geometric shape and size information, and the electromagnetic property data includes dielectric constant and permeability.
[0006] Furthermore, the process of modeling the spatial electromagnetic field of the scanned target includes finite element analysis, wave equation solving, and Kalman filtering; the electromagnetic interference signal characteristics include amplitude, frequency, and phase.
[0007] Secondly, this invention proposes a scanning target verification method based on inter-phase interference, comprising the following steps: S1: Generate a digital signal for the verification scan target according to the verification requirements, and convert the digital signal into an analog signal; S2: Apply the analog signal to the input port of the scanning target to obtain the actual measurement data; S3: Compare the actual measurement data with the preset standard value. If there is an error, perform numerical calibration on the scanned target so that the scanned target is displayed as the preset standard value. S4: Correct the leakage current displayed on the scanned target based on the inter-phase interference compensation data, so that the scanned target displays the accurate value of the actual leakage current.
[0008] Further, the analog signal in step S2 includes a power frequency AC current signal and a simulated lightning strike pulse signal. The power frequency AC current signal is used to verify the leakage current monitoring function of the scanned target and is applied to the leakage current monitoring input terminal. The simulated lightning strike pulse signal is used to verify the lightning strike count recording function of the scanned target and is applied to the lightning strike count recording input terminal.
[0009] Thirdly, the present invention proposes a method for detecting defects in a scanning target, characterized by comprising the following steps: The obtained three-dimensional structural data of the scanned target are preprocessed; Feature extraction is performed on the preprocessed 3D structural data to obtain a feature map; Defect detection is performed on the extracted feature maps using a pre-set deep learning network to obtain defect types and candidate boxes; Post-processing of the candidate boxes yields the final set of candidate boxes.
[0010] Furthermore, the specific content of the preprocessing is as follows: The three-dimensional structural data is cleaned and denoised using the SOR algorithm to remove outliers that are far from the average distance. The expression is:
[0011] in, It is the filtered point cloud. It is a point The average value of the surrounding neighborhood. It is the standard deviation. It is a user-defined constant; The specific content of the feature extraction is as follows: geometric feature extraction is performed on the three-dimensional structural data to obtain a feature map. The geometric features include normal vectors and curvature information. The calculation process of the normal vector is as follows: for each point in the three-dimensional structural data... By performing plane fitting using the least squares method on its neighborhood points, the normal vector is obtained: ,in, It is a point The set of neighborhood points, It is a point The normal vector; The specific content of the defect detection involves using a Faster R-CNN network to identify and analyze feature maps. The Faster R-CNN network includes a region proposal network module, a RoI Pooling module, and a classification and regression module. The region proposal network module generates multiple anchor boxes on the feature map using a sliding window to obtain candidate regions of interest. The RoI Pooling module converts the feature map of each candidate region into a fixed-size feature map and feeds it into a fully connected layer for classification. The expression is:
[0012] in, It is a candidate region. These are the pixels within the candidate region, and `max` represents the maximum value selected in each pooling unit. The classification and regression module is used to classify the defect type of each candidate region using a fully connected layer and output the defect category, expressed as:
[0013] in, It is the output defect category. It is a weight matrix. It is an eigenvector. This is the bias term. Softmax is the activation function used for multi-class classification. After classification, bounding box regression is performed on each candidate region to refine the defect location, resulting in candidate boxes. The expression is:
[0014] in, These are the coordinates of the actual bounding box. These are the predicted bounding box coordinates. It is a loss function; The post-processing specifically involves: calculating the intersection-union ratio (IUR) of the candidate box with the highest confidence level with other candidate boxes; when the IUR is greater than a preset threshold, removing candidate boxes with lower confidence levels using non-maximum suppression, and outputting the final defect detection box set.
[0015] Fourthly, the present invention proposes a scanning target verification system based on interphase interference, characterized in that it includes a terahertz imaging device, an AC signal generator, and a host computer, used to scan the target to obtain three-dimensional structural data and electromagnetic characteristic data. The terahertz imaging device is electrically connected to the input terminal of the host computer, and the output terminal of the host computer is electrically connected to the input terminal of the AC signal generator.
[0016] Furthermore, the terahertz imaging device includes a terahertz wave source and a terahertz detector array. The terahertz wave source is a solid-state laser source capable of emitting electromagnetic waves at a preset frequency. The terahertz detector array consists of multiple detector units used to receive the electromagnetic waves emitted by the terahertz wave source and convert them into electrical signals. The AC signal generator is used to generate an analog signal, and the analog signal is used to verify the leakage current monitoring function and lightning strike count recording function of the scanned target to obtain actual measurement data. The AC signal generator includes a signal generator, an amplifier circuit, a frequency control circuit and an output isolation protection circuit. The host computer includes a space electromagnetic field modeling module, a defect detection module, and a verification and correction module. The space electromagnetic field modeling module is used to execute the method for generating interphase interference compensation data of the scanning target as described in this invention. The verification and correction module is used to execute the scanning target verification method based on interphase interference as described in this invention. The defect detection module is used to execute a method for detecting defects in the scanning target as described in this invention.
[0017] Furthermore, the terahertz imaging device scans the target to obtain initial data, and then filters, reduces noise, and enhances the signal to obtain three-dimensional structural data and electromagnetic property data.
[0018] Beneficial effects: This invention proposes a scanning target verification method and system based on interphase interference. It obtains three-dimensional structure and electromagnetic data by using terahertz wave scanning, and combines finite element analysis, wave equation and Kalman filtering to model the spatial electromagnetic field to obtain interphase interference. Based on interphase interference, the monitoring meters are corrected in real time, so that the monitoring meters finally display the accurate value of the actual leakage current, which improves the accuracy of digital meter verification of surge arresters. At the same time, this invention can realize defect identification and location of surge arresters and their online monitoring meters based on deep learning. Attached Figure Description
[0019] Figure 1 This is a flowchart of a scanning target verification method based on inter-phase interference in an embodiment of the present invention; Figure 2 This is a schematic diagram of the architecture of a scanning target verification system based on inter-phase interference in an embodiment of the present invention. Detailed Implementation
[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0021] Example 1 This embodiment provides a scanning target verification method based on interphase interference, and the flowchart of the method is as follows. Figure 1 As shown, it includes the following specific content: S1: Generate a digital signal for the verification scan target according to the verification requirements, and convert the digital signal into an analog signal; It should be noted that the simulated signal includes a power frequency AC current signal and a simulated lightning strike pulse signal. The power frequency AC current signal is used to verify the leakage current monitoring function of the scanned target and is applied to the leakage current monitoring input terminal. The simulated lightning strike pulse signal is used to verify the lightning strike count recording function of the scanned target and is applied to the lightning strike count recording input terminal.
[0022] S2: Apply the analog signal to the input port of the scanning target to obtain the actual measurement data; S3: Compare the actual measurement data with the preset standard value. If there is an error, perform numerical calibration on the scanned target so that the scanned target is displayed as the preset standard value. S4: Correct the leakage current displayed on the scanned target based on the inter-phase interference compensation data, so that the scanned target displays the accurate value of the actual leakage current.
[0023] It should be noted that the method for obtaining the interphase interference compensation data is as follows: The electromagnetic signals of the obtained scanning target are preprocessed to obtain three-dimensional structural data and electromagnetic property data. The electromagnetic signals are obtained by terahertz wave scanning. The three-dimensional structural data includes geometric shape and size information, and the electromagnetic property data includes dielectric constant and permeability. Based on the three-dimensional structural data and electromagnetic property data, the spatial electromagnetic field of the scanned target is modeled, and the electromagnetic interference signal characteristics between each phase of the scanned target are calculated. Interphase interference compensation data is generated based on the characteristics of the electromagnetic interference signal.
[0024] Example 2 This embodiment provides a practical application of the scanning target verification method based on interphase interference, including the following specific contents: Terahertz waves were used to scan the surge arrester and its online monitoring instruments to obtain three-dimensional structural data and electromagnetic characteristic data. In one specific embodiment, the frequency of the terahertz wave is 0.1-10 THz.
[0025] The electromagnetic field distribution and interphase interference are calculated based on the three-dimensional structural data and electromagnetic property data. It should be noted that the three-dimensional structural data includes geometric shape and size information, and the electromagnetic property data includes dielectric constant and permeability. The specific details of step S2 are as follows: Based on the aforementioned three-dimensional structural data, Maxwell's equations were discretized using the finite element method, yielding the following matrix equations:
[0026] in, It is a system matrix obtained by assembling the finite element discretization of the electric and magnetic fields. It is the vector of unknown quantities in the electric and magnetic fields. It is the source term vector corresponding to the external excitation and boundary conditions. The matrix equation is a general expression of finite element discretization and is used for meshing three-dimensional structural data and mapping material parameters or boundary conditions. To obtain the dynamic characteristics of the electric field's time-domain evolution and interphase interference, the wave equation is used to distribute the electric field in the time domain, and this is iteratively achieved using the finite difference time-domain method. For time-varying electric fields... The wave equation is expressed as:
[0027] in, The magnetic permeability of the medium, The dielectric constant of the medium. For electrical conductivity, For the source term function corresponding to the applied excitation, the iterative update relationship of the electric field based on the finite difference time domain method is expressed as:
[0028] in, For discrete time steps, For time step, The magnetic field strength, Equivalent source term; Initial interphase interference This can be represented by the effect of the leakage current source on the electric field:
[0029] in, It is the spatial distribution function of the electric field in the i-th phase. It is a leakage current signal; Kalman filtering is introduced to recursively correct the initial interphase interference, resulting in interphase interference that can be directly used for compensation. The specific details are as follows: Constructing a dynamic state-space model of interphase interference: The state vector is denoted as... The initial interphase interference measurement value is recorded as The dynamic state-space model can be represented as
[0030]
[0031] in, It is the state transition matrix. It is a control input matrix. It is a control input. It is a measurement matrix. The Kalman gain is calculated for noise measurement.
[0032] in, It is the error covariance matrix. It measures the noise covariance matrix, and its recursive expression is as follows:
[0033] The corrected expression for interphase interference is:
[0034] in, It is the predicted interference electric field after correction by the Kalman filter.
[0035] The leakage current monitoring function and lightning strike count recording function of the online monitoring meter were verified using analog signals to obtain actual measurement data; It should be noted that the simulated signal includes a power frequency AC current signal and a simulated lightning strike pulse signal. The power frequency AC current signal is used to verify the leakage current monitoring function of the online monitoring meter, and the simulated lightning strike pulse signal is used to verify the lightning strike count recording function of the online monitoring meter.
[0036] The actual measurement data is compared with the preset standard value. If there is an error, the online monitoring meter is calibrated. It should be noted that if the actual measurement data is inconsistent with the preset standard value, it indicates that there is an error, and the value of the online monitoring meter should be calibrated to be consistent with the preset standard value.
[0037] The leakage current monitored by the online monitoring meter is corrected based on the interphase interference.
[0038] It should be noted that the leakage current monitored by the online monitoring meter is the vector sum of the phase-to-phase interference current and the actual leakage current. The online monitoring meter is corrected based on the phase-to-phase interference to make the online monitoring meter display the accurate value of the actual leakage current.
[0039] This invention uses terahertz wave imaging technology to perform non-contact three-dimensional structural scanning of surge arresters and their digital meters. It can not only achieve high-resolution imaging of surface and local internal defects of the equipment, but also effectively avoid the interference or damage caused to the equipment by traditional contact-type detection, significantly improving the accuracy of structural identification and equipment adaptability. Multiphysics modeling methods (including finite element analysis, wave equation, Kalman filtering, etc.) are used to simulate and calculate the three-phase electromagnetic coupling characteristics of surge arresters under operating conditions. This can quantitatively extract the amplitude, frequency, and phase characteristics of interphase interference, compensate for the impact of interference on meter measurement results, and solve the problem that electromagnetic coupling errors cannot be eliminated in traditional calibration.
[0040] Example 3 This embodiment provides a scanning target verification system based on phase-to-phase interference, applied to the verification of surge arresters and their online monitoring meters. The system architecture diagram is shown below. Figure 2 As shown, it includes a terahertz imaging device, a host computer, and an AC signal generating device. The host computer includes a space electromagnetic field modeling module, a defect detection module, and a verification and correction module. The terahertz imaging device is connected to the space electromagnetic field modeling module via a shielded twisted-pair cable. The terahertz imaging device includes a terahertz wave source and a terahertz detector array, which are used to scan the surge arrester and its online monitoring meters using terahertz waves to obtain three-dimensional structural data and electromagnetic characteristic data. It should be noted that the terahertz wave source is a solid-state laser source capable of emitting electromagnetic waves at a preset frequency. The terahertz detector array consists of multiple detector units used to receive the electromagnetic waves emitted by the terahertz wave source and convert them into electrical signals. The three-dimensional structural data includes the geometric shape and size information of each part of the device, and the electromagnetic property data of the device material includes the dielectric constant and permeability.
[0041] In one specific embodiment, a terahertz imaging device is installed near the surge arrester under test. A built-in solid-state laser source emits terahertz waves with a frequency range of 0.1-10 THz to perform a comprehensive non-contact scan of the surge arrester and online monitoring meters, acquiring spatial electromagnetic field distribution spectral data. By utilizing the penetration and spectral characteristics of the terahertz waves, electromagnetic property data of the equipment materials, such as dielectric constant and permeability, are obtained. A high-sensitivity terahertz detector array receives terahertz wave signals reflected from the surface and interior of the target equipment. The electromagnetic signals captured by the terahertz detector array are filtered, denoised, and converted by its built-in signal processing module to obtain clear and accurate three-dimensional structural data and electromagnetic property data. This data provides the foundation for subsequent electromagnetic field modeling and defect detection, helping to analyze the electromagnetic environment of the equipment, identify potential defects (such as aging, moisture, cracks, etc.), and provide accurate input for electric field distribution calculation and interphase interference analysis. The obtained three-dimensional data and electromagnetic property data are transmitted to the spatial electromagnetic field modeling module of the host computer to further support equipment verification and monitoring.
[0042] The space electromagnetic field modeling module is used to receive three-dimensional structural data and electromagnetic property data transmitted by the terahertz imaging device through shielded twisted pair cable, and calculate the electromagnetic field distribution and interphase interference based on the three-dimensional structural data and electromagnetic property data. In one specific embodiment, data transmitted from a terahertz imaging device is received and combined with voltage and current signals from each phase of the power system. Spatial electromagnetic field modeling is then performed using finite element analysis (FEM), wave equation solving, Kalman filtering, and multiphysics coupling algorithms. First, based on the acquired data, the electromagnetic field distribution of the surge arrester and its surrounding equipment is calculated, including amplitude, frequency, and phase, simulating the electric and magnetic fields of the equipment under the voltage of each phase. Next, the electromagnetic field distribution between different phases is calculated, and the interaction effect of the interphase electric fields is analyzed to obtain the influence of different phase electric fields on the surge arrester. By calculating the electric field distribution of the surge arrester under the influence of other phase electric fields, the interphase interference current is obtained, providing crucial data support for subsequent interference compensation.
[0043] The AC signal generator is used to generate analog signals, and the analog signals are used to verify the leakage current monitoring function and lightning strike count recording function of the online monitoring meter to obtain actual measurement data. It should be noted that the simulated signal includes a power frequency AC current signal and a simulated lightning strike pulse signal. The power frequency AC current signal is used to verify the leakage current monitoring function of the online monitoring meter, and the simulated lightning strike pulse signal is used to verify the lightning strike count recording function of the online monitoring meter. In one specific embodiment, the host computer generates a digital signal to simulate leakage current or the number of lightning strikes based on the verification requirements. The frequency, amplitude, and other parameters of the generated signal can be adjusted by software programming of the host computer according to the actual verification requirements to adapt to different types of surge arrester monitoring devices. The digital signal generated by the host computer is transmitted to an AC signal generator, and the digital signal is converted into the corresponding analog signal in real time through a digital-to-analog converter (DAC).
[0044] The AC signal generating device includes a signal generator, an amplifier circuit, a frequency control circuit, and an output isolation protection circuit. The signal generator includes an AC signal generating circuit and a pulse signal generating circuit. The AC signal generating circuit uses a crystal oscillator circuit or a function generator chip to generate a power frequency AC current signal (50Hz sine wave). The pulse signal generating circuit generates a simulated lightning strike pulse signal through a monostable trigger circuit or a capacitor discharge circuit.
[0045] The amplification circuit includes a power amplifier and an adjustable amplifier. The power amplifier is used to amplify the generated power frequency AC current signal and the simulated lightning pulse signal to achieve preset voltage and current levels. The adjustable amplifier is used to adjust the amplitude of the signal according to the actual verification requirements, and the gain is adjusted by means of a digital adjustable gain amplifier or an analog switch control.
[0046] The frequency control circuit is implemented through a frequency synthesizer, which includes a phase-locked loop circuit or a direct digital frequency synthesizer, for providing a stable power frequency reference and adjusting the frequency and rise time of the pulse signal, and for programmably controlling the output frequency.
[0047] The output isolation protection circuit includes an isolation transformer and an overcurrent protection circuit. The isolation transformer is used to ensure electrical isolation between the signal generating device and the high-voltage equipment. The overcurrent protection circuit detects and protects the output current through a current sensor or an automatic reset fuse.
[0048] The verification and correction module is used to compare the actual measurement data with the preset standard value. If there is an error, the online monitoring meter is numerically calibrated, and the leakage current monitored by the online monitoring meter is corrected according to the interphase interference.
[0049] In one specific embodiment, a power frequency AC signal is directly applied to the leakage current monitoring input of the meter to verify its measurement function; a simulated lightning pulse signal is applied to the lightning strike count recording input of the surge arrester to verify the accuracy of the meter's recording of lightning events. After the verification signal is applied to the surge arrester meter, the user compares the actual measurement data displayed by the meter with the preset standard value. If there is a difference between the leakage current or lightning strike count displayed by the meter and the standard value, it indicates that there is an error in the meter's measurement result, and therefore the meter needs to be numerically calibrated. After the numerical calibration is completed, since the leakage current detected by the surge arrester monitoring meter is actually the vector sum of the phase-to-phase interference current and the actual leakage current, it is necessary to use the calculated phase-to-phase interference to correct the monitoring meter in real time, so that the monitoring meter ultimately displays the accurate value of the actual leakage current, thereby effectively eliminating the influence of phase-to-phase interference on the surge arrester monitoring data and improving the verification accuracy and reliability.
[0050] The defect detection module is used to detect defects in surge arresters and their online monitoring meters based on 3D structural data scanned by a terahertz imaging device, using a pre-set deep learning network. The specific process includes four steps: data preprocessing, feature extraction, defect detection, and post-processing. Specifically, the module preprocesses the high-resolution 3D structural data obtained from the terahertz imaging device, including denoising and feature extraction, and then uses a target detection algorithm (Faster R-CNN network) based on deep learning for automatic identification. This target detection algorithm can effectively identify and classify various types of potential defects in complex 3D data, including but not limited to cracks, material aging, internal air gaps, bubbles, and material moisture. Furthermore, the module incorporates non-maximum suppression (NMS) and bounding box regression algorithms, enabling precise location and classification of defect areas, effectively improving detection accuracy and reliability.
[0051] It should be noted that the specific content of the data preprocessing is as follows: The three-dimensional structural data is cleaned and denoised using the SOR algorithm to remove outliers that are far from the average distance. The expression is:
[0052] in, It is the filtered point cloud. It is a point The average value of the surrounding neighborhood. It is the standard deviation. It is a user-defined constant; The specific content of the feature extraction is as follows: geometric feature extraction is performed on the three-dimensional structural data to obtain a feature map. The geometric features include normal vectors and curvature information. The calculation process of the normal vector is as follows: for each point in the three-dimensional structural data... By performing plane fitting using the least squares method on its neighborhood points, the normal vector is obtained: ,in, It is a point The set of neighborhood points, It is a point The normal vector; the specific content of the defect detection is to use the Faster R-CNN network to analyze the feature map and obtain the defect location.
[0053] The Faster R-CNN network includes a region proposal network module, a RoI Pooling module, and a classification and regression module. The region proposal network module generates multiple anchor boxes on the feature map using a sliding window to obtain candidate regions of interest. The RoI Pooling module converts the feature map of each candidate region into a fixed-size feature map and feeds it into a fully connected layer for classification. The expression is as follows:
[0054] in, It is a candidate region. These are the pixels within the candidate region, and `max` represents the maximum value selected in each pooling unit. The classification and regression module is used to classify the defect type of each candidate region using a fully connected layer and output the defect category, expressed as:
[0055] in, It is the output defect category. It is a weight matrix. It is an eigenvector. This is the bias term. Softmax is the activation function used for multi-class classification. After classification, bounding box regression is performed on each candidate region to refine the defect location, resulting in candidate boxes. The expression is:
[0056] in, These are the coordinates of the actual bounding box. These are the predicted bounding box coordinates. It is the loss function; the specific content of the post-processing is: calculate the intersection-union ratio (IUR) of the candidate box with the highest confidence and other candidate boxes. When the IUR is greater than a preset threshold, remove the candidate boxes with lower confidence by non-maximum suppression method and output the final defect detection box set.
[0057] This invention, based on a deep learning algorithm (Faster R-CNN), performs intelligent defect identification and localization on terahertz imaging data. It can detect typical deterioration problems such as cracks, bubbles, air gaps, and moisture. Combined with the verification process, it realizes equipment operation status assessment and maintenance decision support, enhancing the system's functional integrity and application scalability. The modules communicate through standardized interfaces, possessing good portability and electromagnetic interference resistance. It is especially suitable for online verification tasks in strong electric field areas such as substations and line terminals, overcoming the technical bottleneck that existing static laboratory verification methods cannot adapt to complex field conditions.
[0058] Example 4 This embodiment provides an implementation method for detecting target defects, applied to the defect detection of surge arresters and their online monitoring meters, including the following steps: The terahertz imaging device was installed on-site near the surge arrester under test and its online monitoring meters to ensure that the scanning angle fully covered the device under test. The terahertz imaging device is activated to scan and acquire terahertz wave signals reflected from the surface and interior of the equipment in real time. It should be noted that the terahertz imaging device uses a built-in solid-state laser source to emit terahertz waves with a frequency range of 0.1-10THz, and uses a non-contact method to perform rapid and comprehensive scanning of the equipment, acquiring in real time the terahertz wave signals reflected back from the surface and different internal structural interfaces of the equipment.
[0059] The scanned data is transmitted in real time to the signal processing module via shielded twisted-pair cable and preprocessed, including real-time filtering, noise reduction and signal enhancement, to generate three-dimensional structural point cloud data; It should be noted that the three-dimensional structural point cloud data is transmitted to the host computer in real time via shielded twisted-pair cable.
[0060] After receiving the 3D structural point cloud data generated on-site, the host computer calls the built-in defect detection algorithm in real time to automatically identify and locate the potential defect locations of the on-site surge arrester equipment.
[0061] It should be noted that, firstly, outlier removal (SOR) based on statistical methods is performed on the 3D point cloud data obtained on-site to eliminate obvious outliers. Then, geometric features are extracted by calculating the normal vector and local curvature information of each point. The defect detection algorithm uses a trained deep learning network (Faster R-CNN model) to identify and analyze the extracted feature data in real time, quickly identifying candidate regions (RoIs) for potential defects through a Region Proposal Network (RPN). Subsequently, a convolutional neural network is used to extract deeper local features, and a fully connected layer classifier is used for classification and regression to accurately identify defect types (such as cracks, bubbles, dampness, etc.) and determine the boundary location and size of the defects. Finally, the defect detection algorithm automatically generates the 3D coordinates and detailed marking information of the defect area, which is fed back to the host computer screen in real time to visually indicate the location of potential defects in the surge arrester and its monitoring meters on-site.
[0062] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.
Claims
1. A method for generating interphase interference compensation data for scanned targets, characterized in that, Includes the following steps: The electromagnetic signals of the scanned target are preprocessed to obtain three-dimensional structural data and electromagnetic property data; Based on the three-dimensional structural data and electromagnetic property data, the spatial electromagnetic field of the scanned target is modeled, and the electromagnetic interference signal characteristics between each phase of the scanned target are calculated. Interphase interference compensation data is generated based on the characteristics of the electromagnetic interference signal.
2. The method for generating interphase interference compensation data for scanned targets according to claim 1, characterized in that, The electromagnetic signal is obtained by terahertz wave scanning; the three-dimensional structural data includes geometric shape and size information, and the electromagnetic property data includes dielectric constant and permeability.
3. The method for generating interphase interference compensation data for scanned targets according to claim 1, characterized in that, The modeling of the spatial electromagnetic field of the scanned target includes finite element analysis, wave equation solving, and Kalman filtering; the characteristics of the electromagnetic interference signal include amplitude, frequency, and phase.
4. A scanning target verification method based on inter-phase interference, characterized in that, Includes the following steps: S1: Generate a digital signal for the verification scan target according to the verification requirements, and convert the digital signal into an analog signal; S2: Apply the analog signal to the input port of the scanning target to obtain the actual measurement data; S3: Compare the actual measurement data with the preset standard value. If there is an error, perform numerical calibration on the scanned target so that the scanned target is displayed as the preset standard value. S4: Correct the leakage current displayed on the scanned target based on the inter-phase interference compensation data, so that the scanned target displays the accurate value of the actual leakage current.
5. The scanning target verification method based on inter-phase interference according to claim 4, characterized in that, The analog signals include a power frequency AC current signal and a simulated lightning strike pulse signal. The power frequency AC current signal is used to verify the leakage current monitoring function of the scanned target and is applied to the leakage current monitoring input terminal. The simulated lightning strike pulse signal is used to verify the lightning strike count recording function of the scanned target and is applied to the lightning strike count recording input terminal.
6. A method for detecting defects in a scanning target, characterized in that, Includes the following steps: The obtained three-dimensional structural data of the scanned target are preprocessed; Feature extraction is performed on the preprocessed 3D structural data to obtain a feature map; Defect detection is performed on the extracted feature maps using a pre-set deep learning network to obtain defect types and candidate boxes; Post-processing of the candidate boxes yields the final set of candidate boxes.
7. The method for detecting defects in a scanning target according to claim 6, characterized in that, The preprocessing specifically involves using the SOR algorithm to clean and denoise the 3D structural data, removing outliers that are far from the average distance. The expression is: in, It is the filtered point cloud. It is a point The average value of the surrounding neighborhood. It is the standard deviation. It is a user-defined constant; The specific content of the feature extraction is as follows: geometric feature extraction is performed on the three-dimensional structural data to obtain a feature map. The geometric features include normal vectors and curvature information. The calculation process of the normal vector is as follows: for each point in the three-dimensional structural data... By performing plane fitting using the least squares method on its neighborhood points, the normal vector is obtained: ,in, It is a point The set of neighborhood points, It is a point The normal vector; The specific content of the defect detection involves using a Faster R-CNN network to identify and analyze feature maps. The Faster R-CNN network includes a region proposal network module, a RoI Pooling module, and a classification and regression module. The region proposal network module generates multiple anchor boxes on the feature map using a sliding window to obtain candidate regions of interest. The RoI Pooling module converts the feature map of each candidate region into a fixed-size feature map and feeds it into a fully connected layer for classification. The expression is: in, It is a candidate region. These are the pixels within the candidate region, and `max` represents the maximum value selected in each pooling unit. The classification and regression module is used to classify the defect type of each candidate region using a fully connected layer and output the defect category, expressed as: in, It is the output defect category. It is a weight matrix. It is an eigenvector. This is the bias term. Softmax is the activation function used for multi-class classification. After classification, bounding box regression is performed on each candidate region to refine the defect location, resulting in candidate boxes. The expression is: in, These are the coordinates of the actual bounding box. These are the predicted bounding box coordinates. It is a loss function; The post-processing specifically involves: calculating the intersection-union ratio (IUR) of the candidate box with the highest confidence level with other candidate boxes; when the IUR is greater than a preset threshold, removing candidate boxes with lower confidence levels using non-maximum suppression, and outputting the final defect detection box set.
8. A scanning target verification system based on interphase interference, characterized in that, The system includes a terahertz imaging device, an AC signal generator, and a host computer. The terahertz imaging device is used to scan the target to obtain three-dimensional structural data and electromagnetic property data. The terahertz imaging device is electrically connected to the input terminal of the host computer, and the output terminal of the host computer is electrically connected to the input terminal of the AC signal generator. The host computer includes a space electromagnetic field modeling module, a defect detection module, and a verification and correction module. The space electromagnetic field modeling module executes the methods of weights 1 to 3, the verification and correction module executes the method of weight 4 or 5, and the defect detection module executes the method of weight 6 or 7.
9. The scanning target verification system based on inter-phase interference according to claim 8, characterized in that, The terahertz imaging device includes a terahertz wave source and a terahertz detector array. The terahertz wave source is a solid-state laser source that can emit electromagnetic waves of a preset frequency. The terahertz detector array consists of multiple detector units used to receive the electromagnetic waves emitted by the terahertz wave source and convert them into electrical signals.
10. The scanning target verification system based on inter-phase interference according to claim 9, characterized in that, The AC signal generator is used to generate an analog signal. The analog signal is used to verify the leakage current monitoring function and lightning strike count recording function of the scanned target to obtain actual measurement data. The AC signal generator includes a signal generator, an amplifier circuit, a frequency control circuit, and an output isolation protection circuit.