A Cable Status Identification and Early Warning Method Based on Multi-Sensor Fusion Quadruped Robot Inspection

Abstract

This invention discloses a cable condition identification and early warning method based on a quadruped robot for multi-sensor fusion inspection. The invention utilizes a quadruped robot to scan the tunnel environment in real time, navigating along a preset inspection route. Visible light and infrared thermal images are simultaneously acquired using a dual-spectrum gimbal, and pixel-level alignment is achieved after image enhancement and ORB feature registration. The visible light image is segmented using a U-Net network, and the cable region temperature matrix is ​​extracted from the infrared thermal image, with the average temperature calculated. Anomaly detection is performed based on multi-level temperature criteria. When an abnormal temperature is detected, a partial discharge sensor mounted on a six-axis robotic arm is activated for fixed-point re-testing, detecting discharge amplitude and frequency for multi-level early warning. A dynamic threshold update mechanism is introduced to adjust the judgment threshold. This invention accurately identifies defects such as cable overheating and insulation degradation, providing timely early warning of potential faults and effectively ensuring the stable operation of high-voltage cable lines.

Description

Technical Field

[0001] This invention relates to the field of power equipment testing technology, specifically a cable condition identification and early warning method based on multi-sensor fusion quadruped robot inspection. Background Technology

[0002] Currently, tunnel cable inspection mainly relies on manual methods or fixed monitoring equipment. Manual inspection suffers from low efficiency, high risk, and strong subjectivity, especially in long-distance tunnels with complex environments, making it difficult to achieve comprehensive coverage and continuous monitoring. Fixed monitoring equipment has limited coverage and is installed in fixed locations, making it unable to flexibly respond to emergencies or detailed inspections of key areas.

[0003] Existing mobile robot platforms mostly employ wheeled or tracked structures, resulting in insufficient mobility and adaptability in uneven tunnel environments, and a lack of high-precision autonomous positioning and environmental modeling capabilities. In terms of detection, traditional methods typically rely on single-source methods such as visible light or infrared detection, failing to achieve multi-source information fusion and easily leading to false alarms or missed detections. Partial discharge detection requires manual operation with equipment close to the site, which is dangerous and inefficient, and lacks a linkage mechanism with temperature monitoring.

[0004] Therefore, there is an urgent need for an advanced inspection solution that can adapt to complex tunnel environments, achieve automatic acquisition and intelligent analysis of multimodal data, and possess autonomous decision-making capabilities. Summary of the Invention

[0005] The technical problem to be solved by this invention is to overcome the defects of the existing technology and provide a cable condition identification and early warning method based on a multi-sensor fusion quadruped robot for inspection. It aims to break through the limitations of traditional high-voltage cable line inspection methods, and achieve autonomous operation of live-line inspection by using intelligent robots equipped with cutting-edge sensing technology, thereby significantly improving the accuracy and efficiency of high-voltage cable line condition monitoring. By accurately identifying defects such as cable overheating and insulation deterioration, it can provide timely early warning of potential faults and effectively ensure the stable operation of high-voltage cable lines.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection, comprising the following steps: The S100 uses a quadruped robot to scan the tunnel environment in real time, build a high-precision map and achieve self-localization. It is also equipped with a high-precision dual-spectrum gimbal and a six-axis robotic arm to cruise along a preset inspection route. S200, the high-precision dual-spectrum gimbal is equipped with a visible light camera and an infrared thermal imaging camera. The visible light camera acquires visible light images of the cable; the infrared thermal imaging camera simultaneously acquires infrared thermal images of the same area and converts them into an infrared temperature matrix. S300, the visible light image and the infrared thermal image are aligned using the ORB algorithm; the cable segmentation mask is obtained by registering the visible light image, the cable area temperature matrix is ​​obtained by indexing the infrared temperature matrix, and then the average temperature of each matrix is ​​extracted. S400 uses a multi-level criterion to determine the average temperature based on a set threshold. When the first-level criterion is met, the quadruped robot continues to cruise. When the second-level criterion is met, the temperature is retested at a fixed point. When the third-level criterion is met, a maintenance signal is issued. When using the S500 and secondary criteria, after the six-axis robotic arm transforms its coordinates, it attaches the partial discharge sensor to the cable insulation layer to collect signals, extracts key indicators such as discharge amplitude and discharge frequency, and records the phase. S600 sets the discharge amplitude threshold and discharge frequency threshold and makes a joint judgment. When the first-level warning occurs, the quadruped robot records the position and marks it. When the second-level warning occurs, it issues a maintenance signal. S700 dynamically updates and corrects temperature thresholds, calculates and updates confidence levels and handles them as appropriate. In case of anomalies, it rolls back to the most recently effective threshold saved in the system and sets up a manual review process.

[0007] Furthermore, step S100 specifically includes: The quadruped robot is equipped with a SLAM system, which uses LiDAR and an inertial measurement unit (IMU) to scan the tunnel environment in real time and build a high-precision map. It also achieves centimeter-level self-localization within the map, enabling fully autonomous navigation and environmental perception. The quadruped robot is equipped with a high-precision dual-spectral gimbal and a six-axis robotic arm, allowing it to automatically cruise along a pre-set digital inspection route. The high-precision dual-spectral gimbal has an automatic stabilization function to counteract the shaking caused by robot movement, ensuring clear and stable image acquisition.

[0008] Furthermore, in step S200, The visible light camera acquires ultra-high resolution visible light images of the cable and its accessories, and the SLAM system is used to mark the three-dimensional coordinates of the detection area in the world coordinate system, thereby realizing the association between geographic information and visual data. The infrared thermal imaging camera detects the infrared radiation emitted by the surface of an object and simultaneously acquires infrared thermal images of cables in the same area, converting them into an infrared temperature matrix. The infrared temperature matrix is ​​a two-dimensional matrix with the same pixel resolution as the image. Each element in the matrix is ​​a floating-point number representing the absolute temperature value of the object surface corresponding to that pixel.

[0009] Further, step S300 includes: Step S301: Receive visible light and infrared thermal images synchronously acquired by the dual-spectrum gimbal to improve image quality and feature visibility; perform CLAHE enhancement processing on the visible light image and NUC non-uniformity correction processing on the infrared thermal image. Because the optical parameters and installation positions of the two cameras are different, even if they are captured synchronously, the position and shape of the same object in the two images will differ. They must be strictly aligned for subsequent fusion analysis. The ORB algorithm based on feature points is adopted. The ORB algorithm finds the same feature points on the synchronously acquired visible light and infrared thermal images, calculates a homography transformation matrix, and uses the matrix to perform perspective transformation on the infrared thermal image to make it completely aligned with the viewpoint of the other visible light image, achieving pixel alignment and laying the foundation for subsequent accurate temperature analysis and fault diagnosis. Step S302: The registered visible light image is input into the U-Net semantic segmentation network. The U-Net semantic segmentation network is trained with labeled data to accurately identify cable pixels and output a category probability for each pixel in the visible light image. The category with the highest probability for each pixel is selected to generate a cable segmentation mask of the same size as the visible light image. The infrared temperature matrix is ​​indexed using the cable segmentation mask to obtain a cable area temperature matrix that only contains the pixel temperature values ​​of the cable area. Step S303: Extract the average temperature from the temperature matrix of each cable region, wherein the average temperature is: , In the formula, T k T represents the average temperature of the temperature matrix of the k-th cable region, where n represents the total number of pixel temperature points in the temperature matrix of the k-th cable region. k,i This represents the temperature of the i-th pixel in the temperature matrix of the k-th cable region.

[0010] Furthermore, the criterion in step S400 is as follows: , In the formula, a1 and a2 are the first temperature threshold and the second temperature threshold, respectively; when the first criterion is met, the quadruped robot continues to cruise automatically along the preset digital inspection route; when the second criterion is met, the quadruped robot performs fixed-point retesting; when the third criterion is met, the quadruped robot issues a maintenance signal to notify maintenance.

[0011] The average temperature of the temperature matrix for each cable area is divided into multiple criteria, and the response is graded according to the severity of temperature anomalies to optimize inspection efficiency and resource allocation.

[0012] Further, step S500 includes: Step S501: The six-axis robotic arm is equipped with a partial discharge sensor. When the average temperature of the temperature matrix of the cable area is within the secondary criterion, a fixed-point retest is performed. The six-axis robotic arm on the quadruped robot converts the three-dimensional coordinates of the current detection area in the world coordinate system into the coordinates of the robotic arm base coordinate system. Based on the coordinates of the robotic arm base coordinate system, the six-axis robotic arm attaches the partial discharge sensor to the cable insulation layer. The partial discharge sensor is used to detect discharge signals in insulated high-voltage equipment to prevent equipment failure caused by insulation degradation. Step S502: The partial discharge sensor acquires the partial discharge signal of the cable insulation layer at a high-frequency sampling rate and extracts key indicators including the discharge amplitude V. peak and discharge frequency F discharge The phase distribution is recorded; wavelet denoising and adaptive filtering are combined to preprocess the original signal to suppress environmental noise and electromagnetic interference, improve the signal-to-noise ratio and retain detailed features.

[0013] Further, step S600 includes: Step S601: Set a first discharge amplitude threshold V1 and a second discharge amplitude threshold V2, where V1 < V2, and a first discharge frequency threshold F1 and a second discharge frequency threshold F2, where F1 < F2, and perform a joint judgment according to the following rules: ; In step S602, when a Level 1 warning is issued, it indicates that there is a slight discharge abnormality that requires close attention. The quadruped robot records and marks its current position. When a Level 2 warning is triggered, it indicates that there is a serious discharge abnormality and a high risk of failure. The quadruped robot issues a maintenance signal to notify maintenance.

[0014] Further, step S700 includes: Step S701: A dynamic threshold update mechanism is introduced to enable the system to have self-learning and self-adaptive capabilities. By learning historical data in real time, the temperature threshold, discharge amplitude threshold, and discharge frequency threshold are automatically adjusted to adapt to environmental changes, equipment aging, and other influencing factors, thereby improving the system's adaptability and judgment accuracy. The temperature threshold is updated as follows: ; The discharge amplitude and frequency thresholds are updated as follows: ; In the formula, These are the updated first and second temperature thresholds, respectively. , These are the updated first discharge amplitude threshold, second discharge amplitude threshold, first discharge frequency threshold, and second discharge frequency threshold, respectively. These represent the historical average temperature, historical average discharge amplitude, and historical average discharge frequency for all cable area temperature matrices. These represent the standard deviations of the average temperature of the temperature matrix for all cable regions, the standard deviations of the discharge amplitude, and the standard deviations of the discharge frequency, respectively. Step S702: Considering the direct impact of ambient temperature on cable surface temperature measurement, further improve the scientific nature of the threshold setting by individually correcting the updated temperature threshold: ; In the formula, These are the corrected first and second temperature thresholds, respectively. This refers to the deviation between the current tunnel ambient temperature and the standard temperature value. These are correction coefficients for a1 and a2, respectively, and the standard temperature value is the suitable operating temperature for the cable. Step S703: Calculate the threshold update confidence level and provide a threshold rollback function. In case of an anomaly, the system can quickly restore the most recently effective threshold stored in the system. The threshold update confidence level C is: , In the formula, Includes updated temperature threshold, discharge amplitude threshold, and discharge frequency threshold. Includes the temperature threshold, discharge amplitude threshold, and discharge frequency threshold before the update, during calculation. and For thresholds of the same type, calculate the corresponding threshold confidence level by arbitrarily inputting one of the thresholds. When C > c1, update the threshold. When c1 ≤ C ≤ c2, initiate the manual review process for manual review. When C < c2, reject the threshold update.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. Enhance the automation and intelligence of inspection: The quadruped robot relies on SLAM technology to achieve fully autonomous centimeter-level positioning and navigation, replacing manual labor in dangerous and complex tunnel environments, improving inspection efficiency, and reducing the labor intensity and safety risks of personnel.

[0016] 2. Achieve accurate perception and fusion of multimodal information: By simultaneously acquiring visible light and infrared images through a high-precision dual-spectrum gimbal, and combining image enhancement, feature registration and semantic segmentation technologies, accurate identification of cable targets and accurate extraction of temperature data are achieved, eliminating background interference and providing a high-quality data foundation for subsequent analysis.

[0017] 3. Establish a tiered early warning and collaborative detection mechanism: A multi-level judgment strategy of "preliminary temperature screening - partial discharge retest confirmation" is adopted. When a potential temperature anomaly is detected, the partial discharge sensor mounted on the robotic arm is automatically triggered to perform a close-range, precise retest. The discharge amplitude and frequency are combined for joint judgment, which improves the accuracy and reliability of fault diagnosis and effectively avoids false alarms and missed alarms.

[0018] 4. Possesses adaptive and self-optimizing capabilities: It introduces a dynamic threshold update mechanism, which can automatically learn and adjust the temperature and discharge criterion thresholds based on historical data, overcoming the problem of inaccurate judgment caused by fixed thresholds due to environmental changes, equipment aging, and other factors; at the same time, it adds environmental temperature correction and confidence-based threshold update review and rollback functions to ensure the adaptability of the system in long-term operation and the robustness of decision-making. Attached Figure Description

[0019] Figure 1 This is a flowchart of the cable condition identification and early warning method based on multi-sensor fusion quadruped robot inspection according to the present invention; Figure 2 This is a schematic diagram of the cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to the present invention; Figure 3 This is a schematic diagram illustrating the environmental perception and data localization of the cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to the present invention. Figure 4 This is a schematic diagram of image processing and temperature analysis for the cable condition identification and early warning method based on multi-sensor fusion quadruped robot inspection according to the present invention. Figure 5 This is a schematic diagram of the fixed-point retesting and adaptive optimization of the cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to the present invention. Detailed Implementation

[0020] 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, and 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.

[0021] like Figures 1 to 5 As shown in the figure, this embodiment is a cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection, and its steps are as follows: In step S100, the quadruped robot scans the tunnel environment in real time to build a high-precision map and achieve self-localization. It is also equipped with a high-precision dual-spectrum gimbal and a six-axis robotic arm to cruise along a preset inspection route.

[0022] The quadruped robot is equipped with a SLAM system, which uses LiDAR and an inertial measurement unit (IMU) to scan the tunnel environment in real time and build a high-precision map. It also achieves centimeter-level self-localization within the map, enabling fully autonomous navigation and environmental perception. The quadruped robot is equipped with a high-precision dual-spectral gimbal and a six-axis robotic arm, allowing it to automatically cruise along a pre-set digital inspection route. The high-precision dual-spectral gimbal has an automatic stabilization function to counteract the shaking caused by robot movement, ensuring clear and stable image acquisition.

[0023] In step S200, the high-precision dual-spectrum gimbal is equipped with a visible light camera and an infrared thermal imaging camera. The visible light camera acquires visible light images of the cable; the infrared thermal imaging camera simultaneously acquires infrared thermal images of the same area and converts them into an infrared temperature matrix.

[0024] In step S200, the visible light camera acquires ultra-high resolution visible light images of the cable and its accessories, and uses the SLAM system to label the three-dimensional coordinates of the detection area in the world coordinate system, thereby realizing the association between geographic information and visual data. The infrared thermal imaging camera detects the infrared radiation emitted by the surface of an object and simultaneously acquires infrared thermal images of cables in the same area, converting them into an infrared temperature matrix. The infrared temperature matrix is ​​a two-dimensional matrix with the same pixel resolution as the image. Each element in the matrix is ​​a floating-point number representing the absolute temperature value of the object surface corresponding to that pixel.

[0025] Step S300: Align the visible light image and the infrared thermal image using the ORB algorithm; register the visible light image to obtain the cable segmentation mask, index the infrared temperature matrix to obtain the cable region temperature matrix, and then extract the average temperature of each matrix.

[0026] Specifically, step S300 includes the following steps: Step S301: Receive visible light and infrared thermal images synchronously acquired by the dual-spectrum gimbal to improve image quality and feature visibility; perform CLAHE enhancement processing on the visible light image and NUC non-uniformity correction processing on the infrared thermal image. Because the optical parameters and installation positions of the two cameras are different, even if they are captured synchronously, the position and shape of the same object in the two images will differ. They must be strictly aligned for subsequent fusion analysis. The ORB algorithm based on feature points is adopted. The ORB algorithm finds the same feature points on the synchronously acquired visible light and infrared thermal images, calculates a homography transformation matrix, and uses the matrix to perform perspective transformation on the infrared thermal image to make it completely aligned with the viewpoint of the other visible light image, achieving pixel alignment and laying the foundation for subsequent accurate temperature analysis and fault diagnosis. Step S302: The registered visible light image is input into the U-Net semantic segmentation network. The U-Net semantic segmentation network is trained with labeled data to accurately identify cable pixels and output a category probability for each pixel in the visible light image. The category with the highest probability for each pixel is selected to generate a cable segmentation mask of the same size as the visible light image. The infrared temperature matrix is ​​indexed using the cable segmentation mask to obtain a cable area temperature matrix that only contains the pixel temperature values ​​of the cable area. Step S303: Extract the average temperature from the temperature matrix of each cable region to comprehensively characterize the overall heat generation of that cable segment. The average temperature is: , In the formula, T k T represents the average temperature of the temperature matrix of the k-th cable region, where n represents the total number of pixel temperature points in the temperature matrix of the k-th cable region. k,i This represents the temperature of the i-th pixel in the temperature matrix of the k-th cable region.

[0027] In step S400, the average temperature is judged according to a set threshold in multiple levels. When the first level of judgment is applied, the quadruped robot continues to cruise. When the second level of judgment is applied, the fixed-point retest is performed. When the third level of judgment is applied, a maintenance signal is issued.

[0028] The specific criterion in step S400 is as follows: , In the formula, a1 and a2 are the first temperature threshold and the second temperature threshold, respectively; when the first criterion is met, the quadruped robot continues to cruise automatically along the preset digital inspection route; when the second criterion is met, the quadruped robot performs fixed-point retesting; when the third criterion is met, the quadruped robot issues a maintenance signal to notify maintenance.

[0029] The average temperature of the temperature matrix for each cable area is divided into multiple criteria, and the response is graded according to the severity of temperature anomalies to optimize inspection efficiency and resource allocation.

[0030] In step S500, during the secondary criterion, after the six-axis robotic arm transforms its coordinates, it attaches the partial discharge sensor to the cable insulation layer to collect signals, extracts key indicators such as discharge amplitude and discharge frequency, and records the phase.

[0031] Specifically, step S500 includes: Step S501: The six-axis robotic arm is equipped with a partial discharge sensor. When the average temperature of the temperature matrix of the cable area is within the secondary criterion, a fixed-point retest is performed. The six-axis robotic arm on the quadruped robot converts the three-dimensional coordinates of the current detection area in the world coordinate system into the coordinates of the robotic arm base coordinate system. Based on the coordinates of the robotic arm base coordinate system, the six-axis robotic arm attaches the partial discharge sensor to the cable insulation layer. The partial discharge sensor is used to detect discharge signals in insulated high-voltage equipment to prevent equipment failure caused by insulation degradation. Step S502: The partial discharge sensor acquires the partial discharge signal of the cable insulation layer at a high-frequency sampling rate and extracts key indicators including the discharge amplitude V. peak and discharge frequency F discharge The phase distribution is recorded; wavelet denoising and adaptive filtering are combined to preprocess the original signal to suppress environmental noise and electromagnetic interference, improve the signal-to-noise ratio and retain detailed features.

[0032] Step S600: Set the discharge amplitude threshold and discharge frequency threshold and make a joint judgment. When the first-level warning occurs, the quadruped robot records the position and marks it. When the second-level warning occurs, a maintenance signal is issued.

[0033] Specifically, step S600 includes the following steps: Step S601: Set a first discharge amplitude threshold V1 and a second discharge amplitude threshold V2, where V1 < V2, and a first discharge frequency threshold F1 and a second discharge frequency threshold F2, where F1 < F2, and perform a joint judgment according to the following rules: ; In step S602, when a Level 1 warning is issued, it indicates that there is a slight discharge abnormality that requires close attention. The quadruped robot records and marks its current position. When a Level 2 warning is triggered, it indicates that there is a serious discharge abnormality and a high risk of failure. The quadruped robot issues a maintenance signal to notify maintenance.

[0034] Step S700: Dynamically update the threshold and correct the temperature threshold, calculate the updated confidence level and handle it according to the situation. In case of an anomaly, restore to the most recently effective threshold saved in the system, and set up a manual review process.

[0035] Specifically, step S700 includes the following steps: Step S701: A dynamic threshold update mechanism is introduced to enable the system to have self-learning and self-adaptive capabilities. By learning historical data in real time, the temperature threshold, discharge amplitude threshold, and discharge frequency threshold are automatically adjusted to adapt to environmental changes, equipment aging, and other influencing factors, thereby improving the system's adaptability and judgment accuracy. The temperature threshold is updated as follows: ; The discharge amplitude and frequency thresholds are updated as follows: ; In the formula, These are the updated first and second temperature thresholds, respectively. , These are the updated first discharge amplitude threshold, second discharge amplitude threshold, first discharge frequency threshold, and second discharge frequency threshold, respectively. These represent the historical average temperature, historical average discharge amplitude, and historical average discharge frequency for all cable area temperature matrices. These represent the standard deviations of the average temperature of the temperature matrix for all cable regions, the standard deviations of the discharge amplitude, and the standard deviations of the discharge frequency, respectively. Step S702: Considering the direct impact of ambient temperature on cable surface temperature measurement, further improve the scientific nature of the threshold setting by individually correcting the updated temperature threshold: ; In the formula, These are the corrected first and second temperature thresholds, respectively. This refers to the deviation between the current tunnel ambient temperature and the standard temperature value. These are correction coefficients for a1 and a2, respectively, and the standard temperature value is the suitable operating temperature for the cable. Step S703: Calculate the threshold update confidence level and provide a threshold rollback function. In case of an anomaly, the system can quickly restore the most recently effective threshold stored in the system. The threshold update confidence level C is: , In the formula, Includes updated temperature threshold, discharge amplitude threshold, and discharge frequency threshold. Includes the temperature threshold, discharge amplitude threshold, and discharge frequency threshold before the update, during calculation. and For thresholds of the same type, calculate the corresponding threshold confidence level by arbitrarily inputting one of the thresholds. When C > c1, update the threshold. When c1 ≤ C ≤ c2, initiate the manual review process for manual review. When C < c2, reject the threshold update.

[0036] The cable condition identification and early warning method based on multi-sensor fusion quadruped robot inspection, as described above, is applied as follows: For the daily inspection needs of underground high-voltage cable tunnels in a certain city, a quadruped robot equipped with multiple sensors is used to achieve full-process automation. The robot is set to automatically cruise along a preset path. The first temperature threshold a1 and the second temperature threshold a2 are set to 40℃ and 60℃, respectively. The first discharge amplitude threshold and the second discharge amplitude threshold are set to V1=5mV and V2=10mV, respectively. The first discharge frequency threshold and the second discharge frequency threshold are set to F1=50 times / minute and F2=100 times / minute, respectively. α=0.01, β=0.02, the standard temperature value is 20℃, and the confidence thresholds are c1=0.8 and c2=0.6. The quadruped robot initiates the SLAM system, navigating along the tunnel, constructing a centimeter-level accurate map in real time and achieving self-localization. A visible light camera acquires high-resolution visible light images of the cable and labels its 3D coordinates. Simultaneously, an infrared thermal imaging camera acquires infrared thermal images, generating a temperature matrix. The average temperature T5 of the fifth cable region is calculated to be 55℃, triggering a secondary criterion and initiating a fixed-point re-measurement. Partial discharge signals are acquired, and V is extracted after wavelet denoising. peak =8 mV, F discharge =70 times / minute, due to V peak >V1 and F discharge >F1 triggers a Level 1 warning; the robot records its location and marks it as a "key observation area." The threshold is updated, and historical data statistics are set: [Settings to be added] The settings are 35℃, 4mV, and 40 times / minute, respectively. The values ​​are 5℃, 1mV, and 10 times / minute, respectively. Updated thresholds: , , , , , Ambient temperature calibration: Set the current tunnel temperature to 18℃. Temperature after calibration =25×(1+0.01×( 2))=24.5℃, =45×(1+0.02×( 2))=43.2℃; Temperature threshold confidence check: Calculate the temperature threshold confidence level. Since C < c2, the current temperature threshold update is rejected, and the system is restored to the most recently effective threshold stored in the system.

[0037] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection, characterized in that, Including the following steps: The S100 uses a quadruped robot to scan the tunnel environment in real time, build a high-precision map and achieve self-localization. It is also equipped with a high-precision dual-spectrum gimbal and a six-axis robotic arm to cruise along a preset inspection route. S200, the high-precision dual-spectrum gimbal is equipped with a visible light camera and an infrared thermal imaging camera. The visible light camera acquires visible light images of the cable; the infrared thermal imaging camera simultaneously acquires infrared thermal images of the same area and converts them into an infrared temperature matrix. S300, the visible light image and the infrared thermal image are aligned using the ORB algorithm; the cable segmentation mask is obtained by registering the visible light image, the cable area temperature matrix is ​​obtained by indexing the infrared temperature matrix, and then the average temperature of each matrix is ​​extracted. S400 uses a multi-level criterion to determine the average temperature based on a set threshold. When the first-level criterion is met, the quadruped robot continues to cruise. When the second-level criterion is met, the temperature is retested at a fixed point. When the third-level criterion is met, a maintenance signal is issued. When using the S500 and secondary criteria, after the six-axis robotic arm transforms its coordinates, it attaches the partial discharge sensor to the cable insulation layer to collect signals, extracts key indicators such as discharge amplitude and discharge frequency, and records the phase. S600 sets the discharge amplitude threshold and discharge frequency threshold and makes a joint judgment. When the first-level warning occurs, the quadruped robot records the position and marks it. When the second-level warning occurs, it issues a maintenance signal. S700 dynamically updates and corrects temperature thresholds, calculates and updates confidence levels and handles them as appropriate. In case of anomalies, it restores the system to the most recently effective threshold stored in the system and sets up a manual review process.

2. The cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to claim 1, characterized in that, Step S100 is as follows: The quadruped robot is equipped with a SLAM system, which uses LiDAR and an inertial measurement unit (IMU) to scan the tunnel environment in real time and build a high-precision map. It can also achieve centimeter-level self-positioning on the map. The quadruped robot is equipped with a high-precision dual-spectrum gimbal and a six-axis robotic arm. The quadruped robot can cruise fully automatically along a preset digital inspection route.

3. The cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to claim 2, characterized in that, In step S200, The visible light camera acquires ultra-high resolution visible light images of the cable and its accessories, and the SLAM system is used to mark the three-dimensional coordinates of the detection area in the world coordinate system, thereby realizing the association between geographic information and visual data. The infrared thermal imaging camera detects the infrared radiation emitted by the surface of an object and simultaneously acquires infrared thermal images of cables in the same area, converting them into an infrared temperature matrix. The infrared temperature matrix is ​​a two-dimensional matrix with the same pixel resolution as the image. Each element in the matrix is ​​a floating-point number representing the absolute temperature value of the object surface corresponding to that pixel.

4. The cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to claim 1, characterized in that, Step S300 includes: Step S301: Receive visible light images and infrared thermal images synchronously acquired by the dual-spectrum gimbal. Perform CLAHE enhancement processing on the visible light image and non-uniformity correction (NUC) processing on the infrared thermal image. Employ the ORB algorithm based on feature points. The ORB algorithm finds the same feature points on the synchronously acquired visible light image and infrared thermal image, calculates a homography transformation matrix, and uses the matrix to perform perspective transformation on the infrared thermal image to completely align its viewing angle with that of the other visible light image, thus achieving pixel alignment. Step S302: Input the registered visible light image into the U-Net semantic segmentation network, output a class probability for each pixel in the visible light image, take the class with the highest probability for each pixel, generate a cable segmentation mask of the same size as the visible light image, and index the infrared temperature matrix using the cable segmentation mask to obtain a cable region temperature matrix that only contains the pixel temperature values ​​of the cable region. Step S303: Extract the average temperature from the temperature matrix of each cable region, wherein the average temperature is: ， In the formula, T k T represents the average temperature of the k-th cable region temperature matrix, where n represents the total number of pixel temperature points in the k-th cable region temperature matrix. k,i This represents the temperature of the i-th pixel in the temperature matrix of the k-th cable region.

5. The cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to claim 4, characterized in that, The specific criterion in step S400 is as follows: ， In the formula, a1 and a2 are the first temperature threshold and the second temperature threshold, respectively; when the first criterion is met, the quadruped robot continues to cruise automatically along the preset digital inspection route; when the second criterion is met, the quadruped robot performs fixed-point retesting; when the third criterion is met, the quadruped robot issues a maintenance signal to notify maintenance.

6. The cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to claim 4, characterized in that, Step S500 includes: Step S501: The six-axis robotic arm is equipped with a partial discharge sensor. When the average temperature of the temperature matrix of the cable area is within the secondary criterion, a fixed-point retest is performed. The six-axis robotic arm on the quadruped robot converts the three-dimensional coordinates of the current detection area in the world coordinate system into the coordinates of the robotic arm base coordinate system. Based on the coordinates of the robotic arm base coordinate system, the six-axis robotic arm attaches the partial discharge sensor to the cable insulation layer. Step S502: The partial discharge sensor acquires the partial discharge signal of the cable insulation layer at a high-frequency sampling rate and extracts key indicators including the discharge amplitude V. peak and discharge frequency F discharge The phase distribution is recorded; wavelet denoising and adaptive filtering are combined to preprocess the original signal.

7. The cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to claim 6, characterized in that, Step S600 includes: Step S601: Set a first discharge amplitude threshold V1 and a second discharge amplitude threshold V2, where V1 < V2, and a first discharge frequency threshold F1 and a second discharge frequency threshold F2, where F1 < F2, and perform a joint judgment according to the following rules: ； In step S602, when a first-level warning is issued, the quadruped robot records and marks its current position. When a second-level warning is triggered, the quadruped robot sends a maintenance signal to notify maintenance.

8. The cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to claim 7, characterized in that, Step S700 includes: Step S701: A dynamic threshold update mechanism is introduced. By learning historical data in real time, the temperature threshold, discharge amplitude threshold, and discharge frequency threshold are automatically adjusted. The temperature threshold is updated as follows: ； The discharge amplitude and frequency thresholds are updated as follows: ； In the formula, These are the updated first and second temperature thresholds, respectively. , These are the updated first discharge amplitude threshold, second discharge amplitude threshold, first discharge frequency threshold, and second discharge frequency threshold, respectively. These represent the historical average temperature, historical average discharge amplitude, and historical average discharge frequency for all cable area temperature matrices. These represent the standard deviations of the average temperature of the temperature matrix for all cable regions, the standard deviations of the discharge amplitude, and the standard deviations of the discharge frequency, respectively.

9. The cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to claim 8, characterized in that, Step S700 also includes: Step S702, individually correct the updated temperature threshold: ； In the formula, These are the corrected first and second temperature thresholds, respectively. This refers to the deviation between the current tunnel ambient temperature and the standard temperature value. These are correction coefficients for a1 and a2, respectively, and the temperature standard value is the suitable operating temperature for the cable.

10. The cable status identification and early warning method based on multi-sensor fusion quadruped robot inspection according to claim 9, characterized in that, Step S700 also includes: Step S703: Calculate the threshold update confidence level and provide a threshold rollback function. In case of an anomaly, the system can quickly restore the most recently effective threshold stored in the system. The threshold update confidence level C is: ， In the formula, Includes updated temperature threshold, discharge amplitude threshold, and discharge frequency threshold. Includes the temperature threshold, discharge amplitude threshold, and discharge frequency threshold before the update, during calculation. and For thresholds of the same type, calculate the corresponding threshold confidence level by arbitrarily inputting one of the thresholds. When C > c1, update the threshold. When c1 ≤ C ≤ c2, initiate the manual review process for manual review. When C < c2, reject the threshold update.

Images