Infrared image-based unmanned aerial vehicle day and night inspection method for overhead current collection line
By combining synchronous infrared and visible light image acquisition during day and night, image registration and weighted fusion, and a dynamic temperature reference model, the reliability problem caused by day-night differences in traditional UAV inspections has been solved, achieving efficient and intelligent inspection of overhead power lines.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-30
- Publication Date
- 2026-04-03
AI Technical Summary
Traditional manual inspection methods are inefficient, have insufficient coverage, and pose safety risks. Existing drone inspection solutions are difficult to balance the reliability of detection under day and night conditions. During the day, they are affected by strong light interference, while at night, the signal-to-noise ratio is high but the thermal effect is weakened, leading to missed detections and misjudgments.
The system employs synchronous infrared and visible light image acquisition during both day and night, adaptively adjusts the inspection distance based on line operating conditions, performs image registration and weighted fusion, uses the YOLO model to identify equipment types, introduces a dynamic temperature benchmark model for anomaly detection, and combines an LSTM time-series compensation algorithm to correct temperature drift.
It significantly improves the accuracy of defect identification during the inspection of overhead power lines, achieves imaging optimization under different operating conditions, avoids overexposure or signal attenuation, and improves the stability and robustness of the inspection.
Smart Images
Figure CN121785362A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of anomaly identification technology, specifically to a method for day and night inspection of overhead power lines using unmanned aerial vehicles (UAVs) based on infrared images. Background Technology
[0002] With the expansion of the scale of new energy power collection lines such as wind farms and photovoltaic power stations, the operating environment of overhead lines is becoming increasingly complex. Line components such as tension clamps, conductor joints, and insulators are subjected to high voltage, high load, and alternating weather conditions for a long time, which makes them prone to defects such as poor contact, surface oxidation, and mechanical fatigue.
[0003] Traditional manual inspection methods are limited by terrain, climate, and safety factors, resulting in low efficiency, insufficient coverage, and high personnel risks. The emergence of drone-based infrared inspection technology has significantly improved operational efficiency and safety; however, existing solutions generally employ single-period inspections, making it difficult to ensure reliable detection under varying day and night conditions. During the day, strong light interference, equipment surface reflection, and electromagnetic interference lead to inflated infrared temperatures. At night, although the signal-to-noise ratio is high, reduced current and weakened thermal effects prevent the detection of some minor defects. Furthermore, changes in drone flight attitude and focal length can cause image registration errors, and fixed threshold judgment methods cannot adapt to temperature distribution differences under various operating conditions, leading to missed and false positives. Summary of the Invention
[0004] In view of the above-mentioned problems, the present invention is proposed.
[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for day and night inspection of overhead power lines by unmanned aerial vehicles based on infrared images, comprising: During the day and at night, infrared and visible light synchronously collect images of the overhead power line components along the same route to obtain inspection image data. Based on the real-time operating conditions of the line, the day and night inspection distance is adaptively adjusted; at the same inspection point, the infrared images of the day and night are registered to realize image pose transformation and image size normalization. When performing image recognition, the infrared information of the target area is weighted and fused in combination with day and night conditions to suppress interference from daytime ambient light; Anomaly detection is achieved by performing corrections on the target area and comparing the corrected results with the baseline range of the infrared image.
[0006] As a preferred embodiment of the UAV day and night inspection method for overhead power collection lines based on infrared images described in this invention, the day and night inspection distance includes determining the day and night inspection interval according to the line current load. When conducting daytime inspections, the current is I0, and at night it is I1. If I0 < I1, the observation resolution during the day is enhanced. If I0 ≥ I1, the distance between the inspection route at night and the inspection object is reduced to 1 / k of the daytime inspection spacing, where k = I0 / I1. While obtaining the infrared images of the equipment by the UAV, combining the visible light image YOLO model and the equipment ledger database, automatically identify the type and number of the target equipment, and load the corresponding temperature rise characteristic curve according to the equipment type. By connecting to the power system data interface, obtain the current load data and operating duration of the equipment.
[0007] As a preferred solution of the UAV day and night inspection method for overhead collector lines based on infrared images according to the present invention, wherein: the registration of the infrared images of the day and night includes performing a perspective transformation on the pixel matrix G(x,y) of the daytime image to align it with the pixel matrix H(x,y) of the nighttime image. Perform two infrared image focal length normalizations according to the focal length change. Let the focal length of the daytime image be f1 and that of the nighttime image be f2, then the scaling factor s = f1 / f2, and scale the daytime image to the size of the nighttime image.
[0008] As a preferred solution of the UAV day and night inspection method for overhead collector lines based on infrared images according to the present invention, wherein: the weighted fusion of the infrared information of the target area in combination with the day and night working conditions includes: generating a target area mask: Generate a binary mask using the YOLO segmentation result , representing the area of the component to be inspected; For the scaled daytime infrared image , shield the non-target area; Weighted fusion formula:
[0009] Output the fused infrared image I(x,y).
[0010] As a preferred solution of the UAV day and night inspection method for overhead collector lines based on infrared images according to the present invention, wherein: through the trained regression model, construct the benchmark value interval for each device respectively; Input into the regression model: the current load data and operating duration of the equipment, as well as the environmental parameters and equipment attributes output by the YOLO model; predict the surface temperature of each device in the image under normal conditions; Based on the historical temperature distribution under the same working conditions, statistically calculate the standard deviation; The interval length is obtained by multiplying the standard deviation by a fixed coefficient; the interval is captured by taking the surface temperature under normal conditions as the center of the interval, and the reference value interval is obtained.
[0011] As a preferred embodiment of the UAV-based day and night inspection method for overhead power line collectors based on infrared images described in this invention, the correction of the target area includes precise ROI positioning and reconstruction: By combining visible light and fused infrared images and performing cross-modal alignment, the precise ROI area of each key part of the device in the target area is recalibrated; For the set of temperature values of all pixels within the ROI, apply the 3σ criterion to remove outliers; Dynamic temperature compensation: For the LSTM time-series compensation model pre-trained based on historical temperature deviation sequences and corresponding operating conditions, the current operating condition change trend is input, and the predicted compensation amount is output. The actual temperature after compensation is obtained by adding the mean temperature in the precise ROI region after outlier removal to the predicted compensation amount.
[0012] As a preferred embodiment of the UAV day and night inspection method for overhead power line based on infrared images described in this invention, the anomaly determination includes comparing the compensated actual temperature with the reference value range; if it is within the range, it is determined to be without anomaly; otherwise, it is determined to be abnormal. When an anomaly is identified, a Bayesian network analysis is performed based on the input parameters of the regression model to generate the probability of each anomaly.
[0013] A day and night inspection system for overhead power lines using infrared images based on the method described in this invention, wherein: The data acquisition unit simultaneously acquires infrared and visible light data of the overhead power line components along the same route during the day and night to obtain inspection image data. The processing unit adaptively adjusts the day and night inspection distance based on the real-time operating conditions of the line; at the same inspection point, it registers the infrared images from daytime and nighttime to achieve image pose transformation and normalize the image size. The recognition unit, when performing image recognition, combines day and night conditions to perform weighted fusion of infrared information of the target area to suppress interference from daytime ambient light; The output unit performs corrections on the target area and compares the corrected results with the infrared image reference value range to achieve anomaly detection.
[0014] A computer device includes: a memory and a processor; the memory stores a computer program, wherein: when the processor executes the computer program, it implements the steps of the method described in any one of the present invention.
[0015] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method described in any one of the present invention.
[0016] The beneficial effects of this invention are as follows: By fusing and analyzing images from both day and night, this invention combines the advantages of obvious anomalies during the day and a clean background at night, significantly improving the accuracy of defect identification. Adaptive inspection parameter adjustment based on current load optimizes imaging under different operating conditions, avoiding overexposure or signal attenuation. The fusion of YOLO target recognition and infrared data registration effectively solves the ROI misalignment problem caused by UAV attitude deviations. A dynamic temperature benchmark model based on operating conditions and environment is introduced to replace the traditional fixed threshold, improving the adaptability and robustness of diagnosis. A time-series compensation algorithm is used to correct temperature drift, further enhancing the stability and reliability of inspection results, ultimately achieving efficient and intelligent inspection and risk warning for key components of overhead power lines. Attached Figure Description
[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 The first embodiment of the present invention provides an overall flowchart of a day and night inspection method for overhead power line collectors using unmanned aerial vehicles based on infrared images. Detailed Implementation
[0019] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.
[0020] Reference Figure 1 As one embodiment of the present invention, a method for day and night inspection of overhead power lines using unmanned aerial vehicles based on infrared images is provided, comprising: S1: During the day and at night, infrared and visible light are simultaneously collected on the components of the overhead power line along the same route to obtain inspection image data.
[0021] Use drones to inspect the key components of overhead power collection lines during the day and at night respectively. The key components mainly include fittings such as strain clamps, insulators, conductors, and other various metal or insulating components, and collect infrared inspection pictures.
[0022] Daytime inspection: The drone is equipped with an infrared thermal imager and a visible light camera. Inspect key components: strain clamps, insulators, conductor joints, and other metal / insulating components. Record the image timestamp, GPS coordinates, and camera attitude angles. Synchronously obtain environmental data (ambient temperature, light intensity, wind speed, etc.).
[0023] Nighttime inspection: Fly repeatedly on the same route. Adjust the inspection strategy according to the real-time current. Reduce the interference of ambient light and improve the signal-to-noise ratio of the heating characteristics.
[0024] S2: Based on the real-time operating conditions of the line, adaptively adjust the day and night inspection distances; at the same inspection point, register the infrared images of the day and night, realize the image attitude transformation, and perform image size normalization.
[0025] To improve the sensitivity of detecting weak heating under small currents at night. According to the change in the current passing through the circuit to be detected at night, assuming the current during daytime inspection is I0 and at night is I1, then the distance between the night inspection route and the detection object is reduced to 1 / k of the daytime inspection spacing (k = I0 / I1). At the same time, adjust the camera shooting focal length to ensure that the target occupies a sufficient pixel area in the image. Update the drone flight path planning to ensure clear focus and viewing angle consistency.
[0026] If I0 < I1, it is considered that the heat generation is reduced, and the daytime observation resolution needs to be enhanced.
[0027] Nighttime inspection spacing:
[0028]
[0029] Among them, dday is the standard daytime inspection distance.
[0030] While obtaining the infrared images of the equipment through the drone, collect environmental data, combine the YOLO model of visible light images and the equipment ledger database, automatically identify the type and number of the target equipment, and load the corresponding temperature rise characteristic curve according to the equipment type (such as the allowable temperature rise of the conductor joint ≤ 30K). Connect to the power system data interface to obtain the current load data, operating duration and other operating condition data of the equipment. Establish an equipment ledger database in advance to store equipment characteristic information such as equipment type, model, specification, and historical operating data.
[0031] Specifically, visible light image processing: Target detection is performed using a pre-trained YOLOv5 / v8 model. Detection categories include: tension clamps, suspension clamps, insulator strings, conductor connection points, etc. Equipment information matching: The detection results are correlated with the equipment ledger database to obtain the equipment number, model, and rated parameters. The allowable temperature rise curve for the corresponding equipment (e.g., conductor joints ≤30K) is loaded. Operating condition data access: Current current (I), voltage, load rate, continuous operating time, etc., are obtained from the SCADA or EMS system via API.
[0032] To address the spatial misalignment issue caused by changes in UAV attitude and focal length differences, infrared images of the same inspection point acquired during the day and night were registered. Considering the changes in viewpoint caused by factors such as UAV shaking during the inspection process, appropriate image attitude transformation was required. Then, the focal length of the two infrared images was normalized according to the focal length change. Assuming the focal length of the daytime image is f1 and the focal length of the nighttime image is f2, the scaling factor s = f1 / f2 was used to scale the daytime image to the size of the nighttime image.
[0033] Image registration: GFTT / SIFT feature points are extracted, and the RANSAC algorithm is used to estimate the affine transformation matrix M, which describes the linear mapping relationship between daytime and nighttime images, including operations such as rotation, translation, scaling, and tilting.
[0034] Perform a perspective transformation on the daytime image G(x,y) to align it with the nighttime image H(x,y).
[0035] Daytime infrared image after affine transformation and registration with the nighttime image: G aligned =warpAffine(G,M); The `warpAffine(G,M)` function performs a spatial perspective transformation on image G (which is identical to G(x,y)) using the affine matrix M, resulting in an aligned image. This process ensures that the daytime and nighttime images correspond in spatial coordinates.
[0036] Focal length normalization (Scale Normalization): Let the focal length be f1 during the day and f2 at night, calculate the scaling factor: S = f1 / f2; The aligned daytime image is scaled to the same spatial scale as the nighttime image: G norm (x, y) = resize(G aligned ,scale=S); resize(G alignedThe function `scale=S` represents the operation function that scales the aligned daytime image by a ratio `S` to make it consistent with the nighttime image in spatial resolution.
[0037] S3: When performing image recognition, the infrared information of the target area is weighted and fused in combination with day and night conditions to suppress interference from daytime ambient light.
[0038] Generate a binary mask using YOLO segmentation results Only the area of the component to be inspected is retained.
[0039] Scaled daytime infrared image , shield non-target areas.
[0040] Weighted fusion formula:
[0041] Output the fused infrared image I(x,y).
[0042] S4: Anomaly detection is achieved by performing corrections on the target area and comparing the corrected results with the infrared image reference value range.
[0043] Furthermore, infrared diagnostic baseline values are determined based on historical data, current operating conditions and environmental data, and equipment characteristic data. The environmental data, operating condition data, and equipment characteristic data acquired during the current inspection are input into a trained baseline value calculation model. The model outputs the infrared image baseline temperature value of the equipment under the current operating conditions. Simultaneously, considering the temperature distribution patterns of key parts of the equipment, the fluctuation range of the baseline temperature value is determined. For example, using the baseline temperature value as the center, the upper and lower limits are determined based on the standard deviation of temperature fluctuations statistically derived from historical data, forming an infrared image baseline value interval.
[0044] Specifically, the benchmark value changes dynamically with the type of equipment, operating conditions, and environment, and is not a fixed threshold.
[0045] Input variable A: Current operating condition: Current I, Voltage V, t run (Running time); Environmental parameters: Ambient temperature T amb Relative humidity (RH), wind speed (v) wind Equipment attributes: material, cross-sectional area, heat dissipation coefficient (from the ledger).
[0046] Baseline temperature prediction model: Use the trained regression model (Such as XGBoost, MLP, or LSTM) Predicts surface temperature under normal conditions:
[0047] Determine the baseline interval: Based on the historical temperature distribution under the same operating conditions, the statistical standard deviation Ω is calculated.
[0048] Define the security scope:
[0049] Or set the maximum allowable temperature rise ΔT according to the regulations. max The upper limit is T. base +ΔT max .
[0050] ROI Precise Location and Reconstruction: Cross-modal alignment is performed by combining visible light and fused infrared images. Based on the fusion of visible light and infrared images, the target area to be identified is precisely segmented, and the precise ROI area of key locations (such as the wire clamp exit) is re-marked. The temperature value set {T} of all pixels within the ROI is then processed. i Applying the 3σ criterion: Eliminate those satisfying |T i -μ∣>3σ anomalies are identified to prevent misjudgment of hotspots.
[0051]
[0052] in, This represents the mean, which is the average value of the temperature data. It is the arithmetic mean of all data points. Standard deviation represents the degree of fluctuation in temperature data. It measures the dispersion of the data.
[0053] The LSTM network is used to combine historical temperature deviation data and current operating condition data to predict the possible supplementary temperature value: For the LSTM time series compensation model pre-trained based on historical temperature deviation sequence and corresponding operating conditions, the current operating condition change trend is input and the predicted compensation amount is output; The actual temperature δT after compensation is obtained by adding the mean temperature in the accurate ROI region after outlier removal to the predicted compensation amount.
[0054]
[0055] Where μ′ is the mean value after denoising.
[0056] The compensated actual temperature is compared with the baseline range; if it falls within the range, it is considered normal; otherwise, it is considered abnormal. When an abnormality is determined, a Bayesian network analysis is performed based on the input parameters of the regression model to generate the probability of each type of abnormality.
[0057] Compare T final With reference interval: If T final >T high → Determined as "overheating".
[0058] Feature point comparison analysis: Analyze temperature gradient, hot spot location (online clamp outlet), and duration; combined with changes in operating conditions (such as whether there is a sudden increase in load or long-term overload).
[0059] The output is sorted by probability, for example: poor contact (75% probability), loosening or oxidation (60% probability), overload (40% probability), external contamination (20% probability).
[0060] On the other hand, this embodiment also provides a day and night inspection system for overhead power line unmanned aerial vehicles based on infrared images, which includes: The data acquisition unit simultaneously acquires infrared and visible light data of the overhead power line components along the same route during the day and night to obtain inspection image data.
[0061] The processing unit adaptively adjusts the day and night inspection distance based on the real-time operating conditions of the line; at the same inspection point, it registers the infrared images from daytime and nighttime to achieve image pose transformation and normalize the image size.
[0062] The recognition unit, when performing image recognition, combines day and night conditions to perform weighted fusion of infrared information of the target area, thereby suppressing interference from daytime ambient light.
[0063] The output unit performs corrections on the target area and compares the corrected results with the infrared image reference value range to achieve anomaly detection.
[0064] If the above functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0065] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.
[0066] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.
[0067] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
[0068] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A method for day and night inspection of overhead power lines using unmanned aerial vehicles (UAVs) based on infrared images, characterized in that, It includes: During the day and at night, infrared and visible light are synchronously collected for overhead collector line components along the same route to obtain inspection image data. Based on the real-time working conditions of the line, the inspection distance during day and night is adaptively adjusted. At the same inspection point, infrared images taken during the day and at night are registered to achieve image pose transformation and image size normalization. When performing image recognition, the infrared information of the target area is weighted and fused in combination with day and night working conditions to suppress the interference of daytime ambient light. By performing correction on the target area and comparing the corrected result with the infrared image reference value interval, abnormal determination is achieved.
2. The method for day and night inspection of overhead power lines by UAV based on infrared images as described in claim 1, characterized in that: The day and night inspection distance includes determining the day and night inspection spacing according to the line current load. Suppose the current during daytime inspection is I0 and at night is I1; if I0 < I1, the observation resolution during the day is enhanced. If I0 ≥ I1, the distance between the inspection route at night and the detection object is reduced to 1 / k of the daytime inspection spacing; where k = I0 / I1. While obtaining the infrared image of the device through the drone, in combination with the visible light image YOLO model and the device ledger database, the type and number of the target device are automatically identified, and the corresponding temperature rise characteristic curve is loaded according to the device type. By connecting to the power system data interface, the current load data and operation duration of the device are obtained.
3. The method for day and night inspection of overhead power lines by UAV based on infrared images as described in claim 2, characterized in that: The registration of infrared images taken during the day and at night includes performing perspective transformation on the pixel matrix G(x,y) of the daytime image to align it with the pixel matrix H(x,y) of the nighttime image. According to the change in focal length, the focal lengths of the two infrared images are normalized. Suppose the focal length of the daytime image is f1 and that of the nighttime image is f2, then the scaling factor s = f1 / f2, and the daytime image is scaled to the size of the nighttime image.
4. The method for day and night inspection of overhead power lines by UAV based on infrared images as described in claim 3, characterized in that: The weighted fusion of the infrared information of the target area in combination with day and night working conditions includes: generating a target area mask. Generate a binary mask using YOLO segmentation results , indicating the area of the component to be inspected; Scaled daytime infrared image Shield non-target areas; Weighted fusion formula: Output the fused infrared image I(x,y).
5. The method for day and night inspection of overhead power lines by UAV based on infrared images as described in claim 4, characterized in that: Through the trained regression model, the reference value interval is constructed for each device respectively. Input into the regression model: the current load data and operation duration of the device, as well as the environmental parameters and device attributes output by the YOLO model; predict the surface temperature of each device in the image under normal conditions. Based on the historical temperature distribution under the same working conditions, calculate the standard deviation. Multiply the standard deviation by a fixed coefficient to obtain the interval length. Centered on the surface temperature under normal conditions, perform interval capture to obtain the reference value interval.
6. The method for day and night inspection of overhead power lines by UAV based on infrared images as described in claim 5, characterized in that: The correction of the target area includes: precise ROI positioning and reconstruction. Combining the visible light and the fused infrared image, after cross-modal alignment, accurately calibrate the precise ROI area of each key part of the device in the target area. Apply the 3σ criterion to剔除异常点 the set of pixel temperature values within the ROI to eliminate abnormal points. Temperature dynamic compensation: Input the current working condition change trend into the LSTM time series compensation model pre-trained based on the historical temperature deviation sequence and the corresponding working conditions, and output the predicted compensation amount. Add the mean value of the temperature within the precise ROI area after eliminating abnormal points to the predicted compensation amount to obtain the compensated actual temperature.
7. The method for day and night inspection of overhead power lines by UAV based on infrared images as described in claim 6, characterized in that: The anomaly determination includes comparing the compensated actual temperature with the baseline value range; If it falls within the range, it is considered to be without anomalies; Conversely, it is judged as abnormal; When an anomaly is identified, a Bayesian network analysis is performed based on the input parameters of the regression model to generate the probability of each anomaly.
8. A day and night inspection system for overhead power lines based on infrared images, employing the method described in any one of claims 1-7, characterized in that: The data acquisition unit simultaneously acquires infrared and visible light data of the overhead power line components along the same route during the day and night to obtain inspection image data. The processing unit adaptively adjusts the day and night inspection distance based on the real-time operating conditions of the line. At the same inspection point, infrared images from daytime and nighttime are registered to achieve image pose transformation and image size normalization. The recognition unit, when performing image recognition, combines day and night conditions to perform weighted fusion of infrared information of the target area to suppress interference from daytime ambient light; The output unit performs corrections on the target area and compares the corrected results with the infrared image reference value range to achieve anomaly detection.
9. A computer device, comprising: A memory and a processor; the memory stores a computer program, characterized in that: when the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-7.