Terahertz gas insulated switch measuring equipment and method based on artificial intelligence

Abstract

The invention discloses a terahertz gas insulated switch measuring device and method based on artificial intelligence, the terahertz gas insulated switch measuring device comprises a data imaging module and a data processing module, the data imaging module comprises a terahertz wave emission source, a lock-in amplifier module, a quasi-optical system and a high-sensitivity detector, and the data processing module comprises an upper computer; when the insulation degree of the gas insulated switch is measured, the data imaging module terahertz wave emission source generates terahertz waves, the terahertz waves are focused through the quasi-optical system and transmitted to the gas insulated switch equipment, the high-sensitivity detector receives the terahertz waves passing through the gas insulated switch equipment and converts the terahertz waves into electric signals, and the electric signals are transmitted to the gas insulated switch equipment. Further, the lock-in amplifier module amplifies and filters the electric signals output by the high-sensitivity detector, the data processing module receives the electric signals output by the lock-in amplifier module, defects are recognized through a noise reduction algorithm and a deep learning model, and a data processing result is displayed; the problems that traditional detection is low in precision, small in range, low in efficiency and the like are solved.

Description

Technical Field

[0001] The present invention relates to the field of new power system detection, and specifically relates to a terahertz gas-insulated switch determination device based on artificial intelligence, and also relates to a terahertz gas-insulated switch determination method based on artificial intelligence. Background Art

[0002] Terahertz waves (abbreviated as THz waves) refer to electromagnetic waves with frequencies in the range of 0.1 THz to 10 THz, which are between microwaves and infrared light, and have unique penetration ability, capable of penetrating non-polar substances (such as plastics, papers, fabrics, etc.). Terahertz waves belong to non-ionizing radiation and have high safety, and are suitable for fields such as detection imaging, including security inspection, non-destructive testing, medical imaging, high-speed communication, spectral analysis, and astronomy.

[0003] A gas-insulated switch is a high-voltage electrical device and a key switch device in the power system. The gas pressure change, particle contamination, and insulation surface inside the device directly affect the insulation degree of the device, which may lead to poor breaking performance and safety hazards. Existing detection methods (such as ultrasonic detection, gas component testing method, etc.) have certain limitations in detection accuracy, range, and efficiency, are highly dependent on the environment, and cannot achieve real-time and continuous monitoring of gas-insulated switch devices.

[0004] The deficiencies of the existing technology include: 1. Limited detection accuracy: Existing detection methods have certain limitations in detection accuracy and are difficult to detect tiny defects and faults; 2. Limited detection range: Some detection methods can only detect local areas of the device and cannot comprehensively cover the insulation degree detection of the entire device; 3. Low detection efficiency: Some detection methods require a long detection period and cannot quickly obtain the detection results of the insulation degree; 4. High dependence on the environment: Some detection methods are greatly affected by environmental factors (such as temperature, humidity, etc.), resulting in unstable detection results; 5. Unable to perform real-time monitoring: Most existing detection methods are periodic detection or live detection, and cannot achieve real-time and continuous monitoring of the insulation degree of gas-insulated switch devices. Summary of the Invention

[0005] The purpose of the present invention is to provide a terahertz gas-insulated switch determination device and method based on artificial intelligence to achieve real-time, continuous, and efficient monitoring of gas-insulated switch determination devices.

[0006] To achieve the above functions, the present invention designs a terahertz gas-insulated switch determination device based on artificial intelligence. The terahertz gas-insulated switch determination device includes a data imaging module and a data processing module; Among them, the data imaging module includes a terahertz wave emission source, a lock-in amplifier module, a quasi-optical system, and a high-sensitivity detector; the data processing module consists of a host computer, including a central processing unit, a liquid crystal display unit, a data transmission unit, and a data storage unit; the terahertz wave emission source is connected to the quasi-optical system, and the terahertz wave emission source emits terahertz waves to the gas-insulated switchgear through the quasi-optical system; the high-sensitivity detector, the lock-in amplifier module, and the host computer are connected in sequence. The high-sensitivity detector receives terahertz waves and transmits them to the host computer through the lock-in amplifier module; the host computer stores the defect identification and measurement programs, software interfaces, and execution parameters of the gas-insulated switchgear accordingly.

[0007] As a preferred technical solution of the present invention: the lock-in amplifier module includes a signal channel, a reference signal channel, a phase-sensitive detector, and a low-pass filter.

[0008] As a preferred technical solution of the present invention: the liquid crystal display unit, the data transmission unit, and the data storage unit of the host computer are connected to the central processing unit. The data transmission unit includes one or more of an RJ45 network port and a debugging serial port, and the data storage unit includes DDR and EMMC.

[0009] As a preferred technical solution of the present invention: the signal-to-noise ratio of the terahertz wave emission source is ≥1000 dB.

[0010] As a preferred technical solution of the present invention: the software interface of the host computer is written in C#, and the software interface includes the following buttons: File, Settings, Help, Parameter Configuration, Data Output, Signal Processing, Model Inference, Real-time Detection Image Data, Start, and Stop.

[0011] As a preferred technical solution of the present invention: the data processing module communicates with the data imaging module through the I2C bus. In the communication protocol, the function code 0x01 represents the terahertz wave emission source, 0x10 represents the lock-in amplifier, and 0x11 represents the high-sensitivity detector.

[0012] The present invention also designs a terahertz gas-insulated switch measurement method based on artificial intelligence. Based on the above-mentioned terahertz gas-insulated switch measurement device based on artificial intelligence, the following steps S1 - step S4 are executed to complete the defect identification and measurement of the gas-insulated switchgear: Step S1: Debug the terahertz wave emission source and the quasi-optical system. According to the technical parameter requirements of the gas-insulated switchgear to be measured, adjust the placement range of the gas-insulated switchgear to be measured until the terahertz wave emission source is directly facing the gas-insulated switchgear, and the terahertz wave emission source emits terahertz waves to the gas-insulated switchgear. Step S2: Perform grid scanning on the gas-insulated switchgear, collect the time-domain spectral signals reflected or transmitted by the gas-insulated switchgear, perform noise reduction processing on the time-domain spectral signals, extract time-domain features and frequency-domain features, and construct a multi-dimensional feature vector; Step S3: Construct a defect recognition model based on the deep learning algorithm, use the multi-dimensional feature vector as the input to identify the defect types of the gas-insulated switchgear, output the defect probability scores corresponding to each type of defect, and perform silhouette coefficient evaluation to determine the number of clusters and divide the defect levels; Step S4: Map the defect types, defect positions, and defect levels of the gas-insulated switchgear into a three-dimensional model and perform visual display on the liquid crystal display unit of the host computer.

[0013] As a preferred technical solution of the present invention: The specific steps of Step S2 are as follows: Step S2.1: Collect the time-domain spectral signals of the gas-insulated switchgear, use a deep learning model to identify the insulation state, and generate an original time-domain electric field signal matrix containing defect features for the gas-insulated switchgear with one or more defects as follows: ; wherein, , represents the time point, is the total number of time points; , represents the points of grid scanning, is the total number of points of grid scanning; represents the amplitude, represents the phase, represents the spectrum; Step S2.2: Use the wavelet decomposition method to decompose the original time-domain electric field signal matrix into high-frequency components and low-frequency components: ; wherein, is the high-frequency noise, is the low-frequency effective signal; Dynamically adjust the decomposition layer number according to the spectral entropy: ; wherein, is the decomposition layer number, is the dynamic adjustment function, is the spectral entropy; Step S2.3: Use the kernel matrix centering method for feature extraction: Let be the wavelet transform result, and extract the kernel matrix : ; ; Among them, is the centralized kernel matrix, is the kernel matrix, and 1 represents the all-ones vector; Step S2.4: Perform eigenvalue decomposition on the centralized kernel matrix: ; Among them, is the eigenvector matrix, is the eigenvalue diagonal matrix; Step S2.5: Concatenate the topological features and physical features into a multi-dimensional feature vector: ; Among them, is the original topological feature data multiplied by the selected n linearly independent eigenvectors to form the topological feature matrix of ; is the original physical feature data multiplied by the selected n linearly independent eigenvectors to form the topological feature matrix of ; Then is the multi-dimensional feature vector of, where each row is the concatenation of the topological features and physical features of the corresponding sample.

[0014] As a preferred technical solution of the present invention: The specific steps of Step S3 are as follows: Step S3.1: Build a defect recognition model based on a deep learning algorithm to identify the defect types of gas insulated switchgear, output the defect probability scores corresponding to each type of defect, and output the identified defect types according to the defect probability scores; Step S3.2: Based on the improved hybrid model of Rep-YOLOX, evaluate the silhouette coefficient of the defect probability scores output by the defect recognition model, determine the number of clusters and divide the defect levels; Step S3.3: Map the multi-feature coefficients of the defect location, defect type, and defect level to a three-dimensional model, and perform visual display on the liquid crystal display unit of the host computer through preset color coding.

[0015] As a preferred technical solution of the present invention: For the identified defect types, calculate the multi-feature coefficients of the defect as follows: ; Among them, is the multi - feature coefficient of the defect type i ; is the number of occurrence points of the defect type i in the current grid scanning; is the defect level of the defect type i ; is the number of occurrences of the defect type i within a preset time period; , , are respectively , , 's weight coefficients; Map the multi - feature coefficient of the defect to the 3D model, and through the preset color coding, perform visual display on the liquid crystal display unit of the host computer.

[0016] Advantageous effects: Compared with the prior art, the advantages of the present invention include: The present invention designs a terahertz gas - insulated switch determination device based on artificial intelligence. By designing a data imaging module and a data processing module, it solves the limitations of low traditional detection accuracy, small range, and low efficiency. It also designs a terahertz gas - insulated switch determination method based on artificial intelligence. By constructing a defect recognition model, it solves the problems such as the inability to achieve real - time and continuous monitoring of gas - insulated switch equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a schematic structural diagram of a terahertz gas - insulated switch determination device based on artificial intelligence provided by an embodiment of the present invention; Figure 2 is a schematic diagram of the host computer software interface provided by an embodiment of the present invention; Figure 3 is a schematic diagram of the I2C communication protocol instruction provided by an embodiment of the present invention; Figure 4 is a flowchart of a terahertz gas - insulated switch determination method based on artificial intelligence provided by an embodiment of the present invention; Figure 5 is a block diagram of the algorithm processing flow of a terahertz gas - insulated switch determination method based on artificial intelligence provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and cannot be used to limit the protection scope of the present invention.

[0019] An artificial intelligence-based terahertz gas-insulated switch measurement device provided by an embodiment of the present invention, referring to Figure 1 , the terahertz gas-insulated switch measurement device includes a data imaging module and a data processing module; Among them, the data imaging module includes a terahertz wave emission source, a lock-in amplifier module, a quasi-optical system, and a high-sensitivity detector; the data processing module is composed of a host computer, including a central processing unit, a liquid crystal display unit, a data transmission unit, and a data storage unit; the terahertz wave emission source is connected to the quasi-optical system, and the terahertz wave emission source generates terahertz waves by the optical crystal method, focuses and transmits them to the gas-insulated switch device through the quasi-optical system; the high-sensitivity detector, the lock-in amplifier module, and the host computer are connected in sequence; the high-sensitivity detector can capture weak terahertz signals, and the detection frequency covers 0.1 THz to 1.1 THz. The high-sensitivity detector receives terahertz waves and converts them into electrical signals. The lock-in amplifier module amplifies and filters the electrical signals output by the high-sensitivity detector to improve the signal stability and transmits them to the host computer; the lock-in amplifier module includes a signal channel, a reference signal channel, a phase-sensitive detector, and a low-pass filter.

[0020] Referring to Figure 2 , the host computer stores the defect identification and measurement programs, software interfaces, and execution parameters of the gas-insulated switch device accordingly. The software interface of the host computer is written in C#, and the software interface includes the following buttons: File, Settings, Help, Parameter Configuration, Data Output, Signal Processing, Model Inference, Real-time Detection Image Data, Start, and Stop. Among them, the File button can implement functions such as creating, opening, and saving parameter configurations. The Settings can perform host computer-side settings on the hardware parameters of the data processing module. The Signal Processing is to configure the conversion of terahertz wave signals of the data imaging module. The Parameter Configuration button is to configure the image display window of the current host computer interface. The Model Inference button is to configure the model and parameters of the deep learning processing part. The execution parameters of the defect identification and measurement program of the gas-insulated switch device are configured through the software interface buttons of the host computer. Among them, the Help button is used to call the pre-stored instruction file of the terahertz gas-insulated switch measurement device. Click the Start and Stop buttons to configure the start and stop of the terahertz gas-insulated switch measurement device. After the defect identification and measurement program of the gas-insulated switch device is started, click the Data Output button to obtain the real-time data during the measurement process and display it in the Real-time Detection Image Data to determine the insulation degree measurement result of the measured gas-insulated switch device.

[0021] The liquid crystal display unit, data transmission unit, and data storage unit of the host computer are connected to the central processing unit. The data transmission unit includes one or more of an RJ45 network port and a debugging serial port. The data storage unit includes DDR and EMMC.

[0022] In one embodiment, the data transmission unit may be an RJ45 network port and / or an RS232 debugging interface, which is connected to the laboratory debugging server through the RJ45 network port and / or the RS232 debugging interface to monitor and remotely control the working state of the terahertz wave gas-insulated switch measuring device.

[0023] In one embodiment, the model of the central processing unit is RK3588J, which can receive the signals transmitted by the data imaging module, call the deep learning algorithm to process the signals, and display them on the liquid crystal display unit. The signal-to-noise ratio of the terahertz wave emission source is ≥1000 dB.

[0024] Refer to Figure 3 , the data processing module communicates with the data imaging module through the I2C bus. In the communication protocol, the function code 0x01 represents the terahertz wave emission source, 0x10 represents the lock-in amplifier, and 0x11 represents the high-sensitivity detector.

[0025] The embodiment of the present invention also provides a method for measuring terahertz gas-insulated switches based on artificial intelligence. Based on the above-mentioned terahertz gas-insulated switch measuring device based on artificial intelligence, refer to Figure 4 , and perform the following steps S1-S4 to complete the defect identification and measurement of the gas-insulated switch device: Step S1: Debug the terahertz wave emission source and the quasi-optical system. According to the technical parameter requirements of the gas-insulated switch device to be measured, adjust the placement range of the gas-insulated switch device to be measured until the terahertz wave emission source is directly facing the gas-insulated switch device, and the terahertz wave emission source emits terahertz waves to the gas-insulated switch device. Step S2: Perform a grid scan on the gas-insulated switch device. The scanning step accuracy is 0.1-1 mm / time. Collect the time-domain spectral signals reflected or transmitted by the gas-insulated switch device, perform noise reduction processing on the time-domain spectral signals, extract the time-domain features and frequency-domain features, and construct a multi-dimensional feature vector. In one embodiment, the above-mentioned time-domain features include peak value and delay time, and the frequency-domain features include spectral amplitude and phase. Refer to Figure 5 , the specific steps of step S2 are as follows: Step S2.1: Collect the time-domain spectral signals of the gas-insulated switch device, use a deep learning model to extract the signal features, and associate them with the insulation state of the gas-insulated switch device (such as the size of the air gap and the degree of aging) to identify the insulation state of the gas-insulated switch device. The deep learning model includes a support vector machine (SVM) or a convolutional neural network (CNN); for a gas-insulated switch device containing one or more defects, generate an original time-domain electric field signal matrix containing defect features As shown in the following formula: ; wherein, , represents the time point, is the total number of time points; , represents the points of grid scanning, is the total number of points of grid scanning; represents the amplitude, represents the phase, represents the spectrum; The time-domain electric field signal in matrix form is: ; Step S2.2: Using the wavelet decomposition method, decompose the original time-domain electric field signal matrix into high-frequency components and low-frequency components: ; wherein, is the high-frequency noise, is the low-frequency effective signal; Dynamically adjust the decomposition level according to the spectral entropy: ; wherein, is the decomposition level, is the dynamic adjustment function, is the spectral entropy; Step S2.3: Using the kernel matrix centering method for feature extraction: Let be the wavelet transform result, and extract the kernel matrix : ; ; wherein, is the centered kernel matrix, is the kernel matrix, and 1 represents the all-ones vector; Step S2.4: Perform eigenvalue decomposition on the centered kernel matrix: ; wherein, is the eigenvector matrix, is the eigenvalue diagonal matrix; Step S2.5: Concatenate the topological features and physical features into a multi-dimensional feature vector: ; wherein, is the original topological feature data multiplied by the selected n linearly independent eigenvectors to form The topological feature matrix of ; is the original physical feature data multiplied by the selected n linearly independent eigenvectors to form the topological feature matrix of ; Then is the multi-dimensional feature vector of, where each row is the concatenation of the topological features and physical features of the corresponding sample.

[0026] Step S3: Construct a defect recognition model based on a deep learning algorithm, use the multi-dimensional feature vector as the input, identify the defect types of gas-insulated switchgear, output the defect probability scores corresponding to each type of defect, and perform a silhouette coefficient evaluation to determine the number of clusters and divide the defect levels; Among them, the defect types include air gaps, delamination, and impurities, and the defect levels are divided into minor, moderate, and severe; The specific steps of Step S3 are as follows: Step S3.1: Construct a defect recognition model based on a deep learning algorithm to identify the defect types of gas-insulated switchgear. The deep learning model includes a support vector machine (SVM) or a convolutional neural network (CNN); output the defect probability scores corresponding to each type of defect, and output the identified defect types according to the defect probability scores; Step S3.2: Based on the improved hybrid model (CNN+Transformer) of Rep-YOLOX, perform a silhouette coefficient evaluation on the defect probability scores output by the defect recognition model to determine the number of clusters and divide the defect levels; Step S3.3: Map the multi-feature coefficients of the defect location, defect type, and defect level to a three-dimensional model, and perform visual display on the liquid crystal display unit of the host computer through a preset color coding.

[0027] For the identified defect types, calculate the multi-feature coefficients of the defect as follows: ; Among them, is the multi-feature coefficient of the defect type i of is the number of occurrence points of the defect type i in the current grid scan, is the defect level of the defect type i of is the number of occurrences of the defect type i within a preset time period, , , respectively and and weight coefficients; Map the multi-feature coefficient of the defect to the 3D model, and perform visual display on the liquid crystal display unit of the host computer through preset color coding.

[0028] Step S4: Map the defect type, defect location, and defect level of the gas-insulated switchgear to a 3D model, and perform visual display on the liquid crystal display unit of the host computer.

[0029] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention within the scope of knowledge possessed by those of ordinary skill in the art.

Claims

1. A terahertz gas-insulated switch measuring device based on artificial intelligence, characterized in that: The terahertz gas-insulated switch measuring device includes a data imaging module and a data processing module; Among them, the data imaging module includes a terahertz wave emission source, a phase-locked amplifier module, a quasi-optical system, and a high-sensitivity detector; the data processing module is composed of a host computer, including a central processing unit, a liquid crystal display unit, a data transmission unit and a data storage unit; the terahertz wave emission source is connected to the quasi-optical system, and the terahertz wave emission source emits terahertz waves to the gas-insulated switchgear through the quasi-optical system; the high-sensitivity detector, the phase-locked amplifier module, and the host computer are connected in sequence, the high-sensitivity detector receives the terahertz wave, and transmits it to the host computer through the phase-locked amplifier module; the defect identification and measurement program, software interface and execution parameters of the gas-insulated switchgear are correspondingly stored in the host computer.

2. The terahertz gas-insulated switch measuring device based on artificial intelligence according to claim 1 is characterized in that: The lock-in amplifier module comprises a signal channel, a reference signal channel, a phase-sensitive detector and a low-pass filter.

3. The terahertz gas-insulated switch measuring device based on artificial intelligence according to claim 1 is characterized in that: The liquid crystal display unit, data transmission unit and data storage unit of the host computer are connected with the central processing unit. The data transmission unit includes one or more of an RJ45 network port and a debugging serial port, and the data storage unit includes DDR and EMMC.

4. The terahertz gas-insulated switch measuring device based on artificial intelligence according to claim 1 is characterized in that: The signal-to-noise ratio of the terahertz wave emission source is ≥1000 dB.

5. The terahertz gas-insulated switch measuring device based on artificial intelligence according to claim 1 is characterized in that: The software interface of the host computer is written in C# and includes the following buttons: File, Settings, Help, Parameter Configuration, Data Output, Signal Processing, Model Inference, Real-time Detection Image Data, Start and Stop.

6. The terahertz gas-insulated switch measuring device based on artificial intelligence according to claim 1 is characterized in that: The data processing module communicates with the data imaging module via an I2C bus. In the communication protocol, the function code 0x01 represents a terahertz wave emission source, 0x10 represents a phase-locked amplifier, and 0x11 represents a high-sensitivity detector.

7. A terahertz gas insulated switch measurement method based on artificial intelligence, characterized in that: Based on the artificial intelligence-based terahertz gas-insulated switch measuring device according to any one of claims 1 to 6, the following steps S1 to S4 are performed to complete defect identification and measurement of the gas-insulated switch device: Step S1: Debugging the terahertz wave emission source and the quasi-optical system, and adjusting the placement range of the gas-insulated switchgear under test according to the technical parameter requirements of the gas-insulated switchgear under test, until the terahertz wave emission source is facing the gas-insulated switchgear, and the terahertz wave emission source emits terahertz waves to the gas-insulated switchgear; Step S2: performing grid scanning on the gas-insulated switchgear, collecting the time-domain spectrum signal reflected or transmitted by the gas-insulated switchgear, performing noise reduction processing on the time-domain spectrum signal, extracting the time-domain features and the frequency-domain features, and constructing a multi-dimensional feature vector; Step S3: construct a defect recognition model based on a deep learning algorithm, use a multidimensional feature vector as input, identify the defect type of the gas insulated switchgear, output the defect probability score corresponding to each type of defect, and perform a silhouette coefficient evaluation to determine the number of clusters and classify the defect levels; Step S4: Mapping the defect type, defect location, and defect level of the gas-insulated switchgear into a three-dimensional model, and visually displaying it on a liquid crystal display unit of a host computer.

8. The method for measuring a terahertz gas-insulated switch based on artificial intelligence according to claim 7, characterized in that: The specific steps of step S2 are as follows: Step S2.1: Collect the time domain spectrum signal of the gas-insulated switchgear, use the deep learning model to identify the insulation state, and generate the original time domain electric field signal matrix containing the defect characteristics for the gas-insulated switchgear containing one or more defects As follows: ； in, , Indicates a point in time, is the total number of time points; , represents the points of the grid scan, Represents the total number of points in the grid scan; represents the amplitude, Indicates the phase, represents the spectrum; Step S2.2: Using wavelet decomposition method, the original time domain electric field signal matrix Decompose into high-frequency and low-frequency components: ； in, is high frequency noise, It is a low-frequency effective signal; Dynamically adjust the number of decomposition layers according to spectral entropy: ； in, is the number of decomposition layers, To dynamically adjust the function, is the spectrum entropy; Step S2.3: Feature extraction using kernel matrix centering method: set up For the wavelet transform result, extract the kernel matrix : ； ； in, is the centralized kernel matrix, is the kernel matrix, 1 represents a vector of all 1s; Step S2.4: Perform eigenvalue decomposition on the centralized kernel matrix: ； in, is the eigenvector matrix, is the eigenvalue diagonal matrix; Step S2.5: Concatenate topological features and physical features into a multi-dimensional feature vector: ； in, Original topological feature data Multiply by the selected n linearly independent eigenvectors constitute The topological characteristic matrix of ; Original physical characteristic data Multiply by the selected n linearly independent eigenvectors constitute The topological characteristic matrix of ; but for A multi-dimensional feature vector of , where each row is the concatenation of the topological features and physical features of the corresponding sample.

9. The method for measuring a terahertz gas-insulated switch based on artificial intelligence according to claim 7, characterized in that: The specific steps of step S3 are as follows: Step S3.1: construct a defect recognition model based on a deep learning algorithm, identify the defect type of the gas insulated switchgear, output the defect probability score corresponding to each type of defect, and output the identified defect type according to the defect probability score; Step S3.2: Based on the hybrid model improved by Rep-YOLOX, the defect probability scores output by the defect recognition model are evaluated by silhouette coefficient, the number of clusters is determined, and the defect levels are divided; Step S3.3: Map the multi-feature coefficients of defect position, defect type, and defect level into a three-dimensional model, and visualize it on the liquid crystal display unit of the host computer through preset color coding.

10. The method for measuring a terahertz gas-insulated switch based on artificial intelligence according to claim 9, characterized in that: For the identified defect type, the multi-feature coefficient of the defect is calculated as follows: ； in, Defect type i The multi-characteristic coefficients of Defect type i The number of occurrence points in the current grid scan, Defect type i The defect level, Defect type i The number of occurrences within a preset time period, , , They are , , The weight coefficient of The multi-characteristic coefficients of the defect are mapped to the three-dimensional model and visualized on the LCD display unit of the host computer through preset color coding.

Images