Wall-along line insulation damage hidden danger identification method, system, equipment and medium

Abstract

The invention relates to the technical field of insulation damage hidden danger recognition, and discloses an along-wall line insulation damage hidden danger recognition method, system and device and a medium, and the method comprises the steps: obtaining a first historical target image of a target along-wall line, and carrying out the preprocessing, thereby effectively improving the image quality, reducing the noise interference, and improving the recognition accuracy. And a more accurate data basis is provided for subsequent steps. A damage hidden danger identification model is established and is taken as a processing core, so that high-precision positioning of a damage position and efficient identification of a damage category can be realized, and the identification accuracy and efficiency are improved. An alarm judgment mechanism is preset and fused into the output logic of the recognition model, and an alarm can be given out in time when the damage hidden danger is recognized. According to the method, wall-along line insulation damage hidden danger identification is carried out in combination with the first real-time target image, real-time monitoring of the wall-along line insulation state can be realized, powerful technical support is provided for operation and maintenance personnel, the maintenance cost is reduced, and the safety and stability of line operation are improved.

Description

Technical Field

[0001] This invention relates to the field of insulation damage hazard identification technology, and in particular to a method, system, equipment and medium for identifying insulation damage hazards along wall-mounted lines. Background Technology

[0002] In power systems, low-voltage wall-mounted cables are a very common form of power transmission. To ensure the safe operation of these lines, the cables are wrapped with an insulation layer. However, over time, the insulation layer may develop defects due to aging, external environmental factors, or mechanical damage, such as bulging, pinholes, cracks, and localized aging. These problems can lead to electrical short circuits, fires, or electric shocks, endangering equipment and personnel safety.

[0003] Current testing methods mainly include the following:

[0004] Manual inspection: Professional maintenance personnel visually inspect the lines, identifying obvious insulation damage such as detached or broken insulation layers. Limitations: Low efficiency, difficult to cover large areas of lines in a short time, especially lines at high altitudes or in complex geographical locations; effectiveness depends on individual experience and skills, prone to missed or false detections, and cannot guarantee accuracy and comprehensiveness.

[0005] Infrared thermal imaging detection: Using an infrared thermal imager to scan the circuit, analyze the thermal image to identify abnormal hot spots and discover potential faults. Limitations: Its function is limited to detecting temperature anomalies; it is ineffective for non-temperature-related insulation defects such as pinholes and cracks, and has a detection blind spot.

[0006] Fixed sensor detection: Temperature, humidity, or current sensors are installed at key locations to monitor the line's operating status in real time, and the system alarms when abnormalities occur. Insufficient coverage: The layout is limited to specific key areas and cannot fully cover the entire line, making it difficult to monitor the overall health of the line in real time, and potential hazards in areas where sensors are not installed are difficult to detect in a timely manner.

[0007] Handheld device inspection: Using handheld devices such as local infrared imagers and cable inspection equipment, precise inspection can be performed on a small area to detect subtle signs of faults in specific locations. Limitations: Requires personnel to have close contact with the wiring, limiting the scope of application; difficult to implement at heights or in areas with safety risks; cannot comprehensively cover the inspection area. Summary of the Invention

[0008] In view of the aforementioned existing problems, the present invention is proposed.

[0009] Therefore, this invention provides a method, system, equipment, and medium for identifying potential insulation damage hazards along wall-mounted lines. It can solve problems such as low efficiency and easy omissions and false detections in manual inspections, blind spots in infrared thermal imaging detection, insufficient coverage of fixed sensors, and limited application range of handheld devices. It achieves efficient, accurate, and comprehensive identification of potential insulation damage hazards along wall-mounted lines.

[0010] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0011] In a first aspect, the present invention provides a method for identifying potential insulation damage hazards along wall-mounted lines, comprising:

[0012] Acquire a first historical target image along the target's path along the wall, and preprocess the first historical target image;

[0013] The first historical target image is acquired using a preset path planning algorithm;

[0014] The preset path planning algorithm is modified by setting inspection constraints;

[0015] A damage hazard identification model is established, and the input of the damage hazard identification module is the preprocessed first historical target image;

[0016] The damage hazard identification model is used to locate the damage location and identify the damage type;

[0017] A preset alarm determination mechanism is established, and the alarm determination mechanism is integrated into the output logic based on the damage hazard identification model;

[0018] Based on the damage hazard identification model after the integrated alarm judgment mechanism, and combined with the first real-time target image, the potential for insulation damage along the wall is identified.

[0019] As a preferred embodiment of the method for identifying potential insulation damage hazards along wall-mounted lines according to the present invention, the preset path planning algorithm includes:

[0020] Establish the objective function for the shortest inspection path and the inspection constraints;

[0021] The objective function for the shortest inspection path is determined by the flight distance between sampling points and the indicator variables for path selection between sampling points.

[0022] The inspection constraints include inspection sampling point constraints and endurance time constraints.

[0023] This optimized solution ensures that the inspection path is both efficient and economical. By setting the objective function of the shortest inspection path, the flight distance during the inspection process can be minimized, thereby improving inspection efficiency. Simultaneously, the inclusion of inspection constraints, such as sampling point constraints and flight time constraints, ensures the comprehensiveness and safety of the inspection, avoiding safety hazards caused by missed inspections or insufficient equipment power. This path planning algorithm, which comprehensively considers path length and inspection requirements, provides strong technical support for identifying potential insulation damage hazards along wall-mounted lines.

[0024] As a preferred embodiment of the method for identifying potential insulation damage hazards along wall-mounted lines according to the present invention, the step of setting inspection constraints for algorithm correction includes:

[0025] Convert inspection constraints into algorithmic constraints;

[0026] The algorithm constraints include the constraint that each sampling point can only be accessed once, the constraint that duplicate access points are removed, the constraint that unaccessed points are supplemented, and the cruise correction constraint.

[0027] The cruise correction constraint is that if the total path distance exceeds the range limit, the path will be randomly adjusted or a relay point will be introduced to decompose the path.

[0028] As a preferred embodiment of the method for identifying potential insulation damage to wall-mounted lines according to the present invention, the establishment of the potential damage identification model includes:

[0029] A module for acquiring the first damaged region based on frame difference method and threshold segmentation is established;

[0030] Establish a second damaged area acquisition module based on optical flow method;

[0031] The first damaged area acquisition module and the second damaged area acquisition module are connected in series in sequence.

[0032] As a preferred embodiment of the method for identifying potential insulation damage to wall-mounted lines according to the present invention, the step of establishing the potential damage identification model further includes:

[0033] Obtain the hazard results corresponding to the preprocessed first historical target image;

[0034] The hazard results corresponding to the preprocessed first historical target image are used as an output of the damage hazard identification model.

[0035] The output of the second damaged area acquisition module is used as another output of the damage hazard identification model;

[0036] The results of damage hazard identification are obtained from the two outputs.

[0037] As a preferred embodiment of the method for identifying potential insulation damage hazards along wall-mounted lines according to the present invention, the preset alarm judgment mechanism includes:

[0038] Several alarm thresholds are set, and several alarm judgment levels are divided according to the aforementioned alarm thresholds;

[0039] The results of the potential damage identification are compared with the aforementioned alarm thresholds;

[0040] The alarm level is determined based on the comparison results.

[0041] As a preferred embodiment of the method for identifying potential insulation damage hazards along wall-mounted lines according to the present invention, the preset alarm determination mechanism further includes:

[0042] After determining the alarm level, the location and severity of the potential damage, along with suggested repair measures, are sent to the management personnel's terminal in real time.

[0043] Secondly, the present invention provides a system for identifying potential insulation damage hazards along wall-mounted wiring, comprising:

[0044] The data acquisition and processing module is used to acquire a first historical target image along the target wall line and to preprocess the first historical target image.

[0045] The first historical target image is acquired using a preset path planning algorithm;

[0046] The preset path planning algorithm is modified by setting inspection constraints;

[0047] The model building module is used to build a damage hazard identification model. The input of the damage hazard identification module is the preprocessed first historical target image.

[0048] The damage hazard identification model is used to locate the damage location and identify the damage type;

[0049] The early warning mechanism fusion module is used to preset the alarm judgment mechanism and integrate the alarm judgment mechanism into the output logic based on the damage hazard identification model;

[0050] The identification module is used to identify potential insulation damage hazards along the wall lines based on the damage hazard identification model after the fusion alarm judgment mechanism, combined with the first real-time target image.

[0051] Thirdly, the present invention provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0052] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0053] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention proposes a method for identifying potential insulation damage hazards along wall-mounted power lines. First, by acquiring and preprocessing the first historical target image of the target wall-mounted power line, image quality can be effectively improved and noise interference reduced, providing a more accurate data foundation for subsequent steps. Second, by establishing a potential damage hazard identification model and using it as the processing core, high-precision positioning of the damage location and efficient identification of the damage type can be achieved, greatly improving the accuracy and efficiency of identification. Third, by pre-setting an alarm judgment mechanism and integrating it into the output logic of the identification model, an alarm can be issued in a timely manner when a potential damage hazard is identified, enhancing the safety and reliability of the system. Finally, by combining the first real-time target image for identifying potential insulation damage hazards along wall-mounted power lines, real-time monitoring of the insulation status of the wall-mounted power lines can be achieved, providing strong technical support for maintenance personnel, reducing maintenance costs, and improving the safety and stability of line operation. Attached Figure Description

[0054] 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.

[0055] Figure 1 This is a flowchart of a method for identifying potential insulation damage to wall-mounted wiring, provided as an embodiment of the present invention. Detailed Implementation

[0056] 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.

[0057] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a method for identifying potential insulation damage in wall-mounted wiring, including:

[0058] Existing technologies suffer from several drawbacks, such as the low efficiency and susceptibility to missed or false detections associated with manual inspections, blind spots in infrared thermal imaging, insufficient coverage of fixed sensors, and limited application scope of handheld devices. These issues restrict the accuracy and comprehensiveness of identifying potential insulation damage hazards along wall-mounted wiring.

[0059] This invention provides a method that can effectively solve the problems mentioned above. The following will describe in detail how to implement the method for identifying potential insulation damage in wall-mounted lines in conjunction with several embodiments.

[0060] Figure 1 A flowchart illustrating a method for identifying potential insulation damage in wall-mounted wiring is shown, including:

[0061] S101, acquire the first historical target image along the target wall line, and preprocess the first historical target image;

[0062] It should be noted that low-voltage wall-mounted lines are the most common form of power transmission in power systems. To ensure the safe operation of these lines, cables are wrapped with an insulation layer. However, over time, this insulation layer may develop defects due to aging, external environmental influences, or mechanical damage, such as bulging, pinholes, cracks, and localized aging. These problems can lead to electrical short circuits, fires, or electric shocks, endangering equipment and personnel safety. Therefore, a method for identifying potential insulation damage in wall-mounted lines is needed.

[0063] In some specific implementations, subsequent pattern and logic analysis can be performed by acquiring relevant data about the target line along the wall, such as the line's geographical location information, historical maintenance records, and environmental factors. This data can provide background information on the line's operational status and potential risks.

[0064] In this embodiment of the invention, the first historical target image is obtained through a preset path planning algorithm;

[0065] In some specific implementations, the path planning algorithm can comprehensively consider factors such as the geographical features of the route, the distribution of obstacles, and the flight performance of the inspection equipment to formulate the optimal inspection path. This algorithm can ensure that the inspection process is both efficient and comprehensive, covering all critical areas without omission, thereby maximizing the identification rate of potential insulation damage.

[0066] For example, when planning inspection routes, the algorithm can prioritize avoiding areas that may interfere with or obstruct the inspection equipment, such as tall buildings and densely wooded areas, while ensuring that the inspection equipment can fly stably along the route and acquire clear image data. Furthermore, the algorithm can be optimized based on parameters such as the inspection equipment's flight speed, altitude, and shooting angle to ensure that the quality of the captured images meets the requirements for subsequent analysis.

[0067] In some specific implementations, the path planning algorithm can take the following steps:

[0068] First, the algorithm collects detailed geographical information about the route and the distribution of obstacles, including the route's length, direction, surrounding terrain, and the location and size of potential obstacles.

[0069] Then, based on this information, the algorithm will generate multiple possible inspection path schemes, each of which will take into account the flight performance of the inspection equipment, such as flight speed, altitude limit and shooting capability.

[0070] Next, the algorithm evaluates these schemes by simulating the inspection process, predicting the interference and obstacles that each scheme may encounter, as well as the quality of the acquired images.

[0071] Finally, the algorithm selects the optimal solution—the path that most efficiently and comprehensively completes the inspection task while ensuring the image data quality meets the requirements for subsequent insulation damage hazard identification. This step not only demonstrates the algorithm's intelligence but also ensures the efficiency and accuracy of the inspection work.

[0072] In this embodiment of the invention, the preset path planning algorithm is modified by setting inspection constraints;

[0073] In this embodiment of the invention, the preset path planning algorithm includes:

[0074] Establish the objective function for the shortest inspection path and the inspection constraints;

[0075] The objective function for the shortest inspection path is determined by the flight distance between sampling points and the indicator variables for path selection between sampling points;

[0076] Inspection constraints include inspection sampling point constraints and endurance time constraints.

[0077] In this embodiment of the invention, setting inspection constraints for algorithm correction includes:

[0078] Convert inspection constraints into algorithmic constraints;

[0079] The algorithm constraints include the constraint that each sampling point can only be visited once, the constraint that duplicate points are removed, the constraint that unvisited points are added, and the cruise correction constraint.

[0080] The cruise correction constraint is that if the total path distance exceeds the range limit, the path will be randomly adjusted or a relay point will be introduced to decompose the path.

[0081] It should be noted that the shortest inspection path objective function refers to a mathematical model or algorithm used when planning inspection routes. Its purpose is to find the shortest flight distance covering all necessary checkpoints (sampling points). This objective function typically considers multiple factors, such as the actual distance between sampling points and the probability of choosing a particular path (indicator variables), to ensure that the inspection can be completed efficiently.

[0082] It should be noted that inspection constraints are specific rules or restrictions used to guide path planning algorithms. For example, the "inspection sampling point constraint" mentioned in this invention means that every location that needs to be inspected must be visited; while the "battery life constraint" refers to the battery life limit of the drone or other inspection equipment, requiring the entire inspection process to be completed within this limit.

[0083] It should be noted that the constraint that each sampling point can only be visited once is to ensure that the same location is not visited repeatedly during the inspection process, thus avoiding wasting time and resources.

[0084] It should be noted that if the initial path generated by the algorithm contains points that are visited repeatedly, the path needs to be adjusted to remove these redundant points.

[0085] It should be noted that the constraint of supplementing unvisited points is used to ensure that all necessary sampling points are included in the final inspection path without any omissions.

[0086] It should be noted that when the calculated total inspection path exceeds the device's maximum range, the path needs to be adjusted. This can be done by randomly changing the order of certain parts or adding relay points (i.e., intermediate charging stations) to divide the long-distance path into several shorter parts, so that the entire inspection task can be completed within the device's capabilities.

[0087] Specifically, this invention is equipped with a high-resolution camera that supports high-definition visible light shooting and infrared imaging, enabling it to acquire clear images even in low-light conditions. It also features obstacle avoidance sensors and a path planning module, ensuring the drone can flexibly navigate complex environments and avoid collisions with power lines or walls.

[0088] Furthermore, drone inspections need to cover low-voltage lines along walls and avoid duplicate paths and uncovered areas as much as possible. The following mathematical model is used to plan the drone path:

[0089] The inspection area is modeled as a graph G(V,E), where V represents the set of sampling points and E represents the flight paths between sampling points. The objective function for the shortest path problem can be as follows:

[0090]

[0091] Where, d ijx represents the flight distance from sampling point i to j. ij Indicator variables for path selection:

[0092]

[0093] Inspection constraints include the requirement that the drone must pass through all target sampling points:

[0094]

[0095] Battery life limitations:

[0096]

[0097] Among them, T max This represents the maximum flight time of the drone.

[0098] In this embodiment of the invention, for the path planning problem, a particle swarm optimization algorithm (PSO) is introduced on the basis of the existing mathematical model: under the above constraints, the path planning problem is optimized to find the optimal inspection path.

[0099] Step 1: Each particle represents a possible path planning solution. Assuming n is the number of sampling points, the positions of the particles can be represented as a sequence of length n:

[0100] x = [v1, v2, ..., v n ]

[0101] Where: v k This represents the k-th sampling point.

[0102] Step 2: The particle's fitness function corresponds to the objective function of path planning, defined as:

[0103]

[0104] Where: d(v) k ,v k+1 () indicates the distance of a path segment.

[0105] Step 3, Particle Update Rules

[0106] 1. Speed ​​update formula:

[0107]

[0108] in, Let ω represent the velocity of particle i in the t-th iteration; ω represent the inertia weight, controlling the balance between global search and local exploitation; c1 and c2 represent acceleration factors, weighing the influence of the individual optimal position and the global optimal position; and r1 and r2 represent random numbers between [0,1]. G represents the historical best position of particle i. best This represents the global optimal position of the group.

[0109] 2. Position update formula:

[0110]

[0111] Step 4: Convert the inspection constraints into algorithm constraints;

[0112] To satisfy the inspection constraints, the particle solution needs to be modified to obtain the constraints that each sampling point can only be visited once, the constraints that repeatedly visited points are removed, the constraints that unvisited points are added, and the cruise correction constraints.

[0113] If the total distance of the path exceeds the range limit, the path will be randomly adjusted or relay points will be introduced to decompose the path.

[0114] In some specific implementations, path optimization operations can be performed, wherein:

[0115] Step A: Randomly generate a swarm of particles, with each particle corresponding to a possible path;

[0116] Initialize the particle's velocity v i .

[0117] Calculate the fitness value of each particle and record the initial fitness value. and g best .

[0118] Step B, Iterative Update:

[0119] a. Calculate the velocity of each particle according to the velocity update formula.

[0120] b. Update the particle's position;

[0121] c. Modify the particle solution to satisfy the constraint conditions;

[0122] d. Calculate the new fitness value and update. and g best .

[0123] Step C: The termination condition is to terminate when the maximum number of iterations is reached or the global optimal solution converges.

[0124] It should also be noted that the drone conducts periodic inspections according to a pre-set flight path, collecting image or video data along the wall. The inspection frequency and data collection angle can be adjusted based on the importance of the route.

[0125] During inspections, drones collect continuous images. t or video stream Mathematical expression of image acquisition:

[0126] I t =f cam (p t ,φ t )

[0127] Where: p t It is the spatial position of the drone at time t, φ t It refers to the camera's orientation; f cam This represents the imaging model of the camera.

[0128] It should be noted that preprocessing includes removing noise from the image using Gaussian filtering or median filtering. The formula for removing image noise using Gaussian filtering is:

[0129]

[0130] in:

[0131]

[0132] It is the Gaussian kernel function; k is half the size of the filter window, and σ is the Gaussian standard deviation.

[0133] Furthermore, the image is converted into a grayscale image to facilitate subsequent feature extraction.

[0134] Convert a color image to a grayscale image:

[0135] I gray (x,y)=0.299·I R (x,y)+0.587·I G (x,y)+0.114·I B (x,y)

[0136] Among them I R ,I G ,I B These are the pixel values ​​for the red, green, and blue channels, respectively.

[0137] Furthermore, normalizing image pixel values ​​to the [0,1] range improves the training and recognition performance of deep learning models.

[0138]

[0139] Furthermore, morphological operations can be performed, which can effectively extract structural features such as trachoma and cracks.

[0140] Etching operation (used to highlight small defects, such as cracks):

[0141]

[0142] Where u and v represent the offset of a pixel position (x, y) in the image, and K represents the structuring element, which determines which pixel values ​​to compare to determine the new value of the target pixel.

[0143] Expansion operation (used to fill damaged areas):

[0144]

[0145] Closed operation (expansion followed by corrosion):

[0146] I closed =Erode(Dilate(I))

[0147] This operation is used to eliminate trachoma and smooth out minor noise.

[0148] Furthermore, morphological gradients are used to extract the edges of the circuit surface:

[0149] I gradient =I dilated -I eroded

[0150] Furthermore, technologies such as brightness adjustment and contrast enhancement are used to improve image clarity.

[0151] It should be noted that acquiring the first historical image of the target line along the wall and preprocessing it can improve image quality, providing more accurate and reliable input for subsequent feature extraction and insulation damage identification. Preprocessing steps may include, but are not limited to, image denoising, contrast enhancement, and distortion correction, aiming to optimize the image data to make it more suitable for machine learning or computer vision algorithms. Furthermore, a pre-defined path planning algorithm can be used to acquire image data of the target line along the wall, ensuring the comprehensiveness and accuracy of the image data. The application of this pre-defined path planning algorithm not only improves inspection efficiency but also ensures the quality of image data, laying a solid foundation for subsequent feature extraction and insulation damage identification.

[0152] S102, Establish a damage hazard identification model. The input of the damage hazard identification module is the preprocessed first historical target image.

[0153] In some specific implementations, the insulation damage hazard identification model can use deep learning frameworks, such as convolutional neural networks (CNNs), to automatically extract features from images. These features may include the texture, color, and shape of the circuit, which are closely related to the insulation damage state. The model is trained using a large amount of labeled historical image data to learn the mapping relationship from images to insulation damage states. During training, the model continuously adjusts its internal parameters to minimize prediction errors and improve identification accuracy. Once training is complete, the model can quickly and accurately identify insulation damage hazards on new image data. Furthermore, to improve the model's generalization ability, data augmentation techniques, such as rotation, scaling, and flipping, can be used to increase the diversity of the training data.

[0154] In this embodiment of the invention, the damage hazard identification model is used to locate the damage location and identify the damage type;

[0155] In this embodiment of the invention, establishing a damage hazard identification model includes:

[0156] A module for acquiring the first damaged region based on frame difference method and threshold segmentation is established;

[0157] Establish a second damaged area acquisition module based on optical flow method;

[0158] The first damaged area acquisition module and the second damaged area acquisition module are connected in series in sequence.

[0159] In this embodiment of the invention, establishing a damage hazard identification model further includes:

[0160] Obtain the hazard results corresponding to the preprocessed first historical target image;

[0161] The hazard results corresponding to the preprocessed first historical target image are used as an output of the damage hazard identification model.

[0162] The output of the second damaged area acquisition module is used as another output of the damage hazard identification model;

[0163] The results of damage hazard identification are obtained from the two outputs.

[0164] Specifically, the frame difference method can detect dynamic changes in video streams and is used to detect significant changes on the surface of lines (such as damage and aging features).

[0165] The frame difference formula can be as follows:

[0166] D t (x,y)=|I t (x,y)-I t-1 (x,y)|

[0167] Among them: I t (x,y) represents the frame at time t; I t-1 (x,y) represents the previous frame; D t (x,y) is the difference map of the current frame.

[0168] Threshold segmentation (extracting regions of significant change):

[0169]

[0170] Where T is the set difference threshold.

[0171] Furthermore, optical flow is used to analyze the motion of pixels in video, and is suitable for detecting minute changes or defects on lines (such as bulging or crack propagation on insulation surfaces).

[0172] Optical flow estimation formula (based on dense optical flow methods, such as the Farneback algorithm):

[0173] I(x+u,y+v,t+1)=I(x,y,t)

[0174] Expanding through Taylor:

[0175]

[0176] Where u and v represent the horizontal and vertical components of the optical flow field; These represent the spatial and temporal components of the image gradient. By solving the optical flow field, minute movements on the line surface can be tracked, further enabling the detection of damaged areas.

[0177] In this embodiment of the invention, target detection and segmentation based on convolutional neural networks (CNN) are selected to achieve damaged region detection and segmentation, respectively, wherein:

[0178] The objective function for the YOLO (YouOnlyLookOnce) object detection task is:

[0179]

[0180] Where: S is the number of grid divisions; B is the number of bounding boxes predicted for each grid. is the indicator function, indicating whether the i-th grid contains the target; (x,y,w,h) are the center coordinates, width, and height of the prediction box; C is the confidence score.

[0181] Furthermore, image segmentation can employ techniques such as UNet or Mask R-CNN to segment the boundaries of damaged regions.

[0182] The UNet model is used for accurate segmentation of damaged regions. The objective function is a weighted sum of cross-entropy loss and Dice loss.

[0183]

[0184] Where: y i It is the true label of the i-th pixel; p i It is the predicted probability of the i-th pixel.

[0185] In some specific implementations, simulated annealing is introduced to optimize hyperparameters, as the performance of deep learning models is significantly affected by hyperparameters (such as learning rate, batch size, regularization coefficient, etc.). Simulated annealing (SA) avoids getting trapped in local optima by simulating the physical annealing process and probabilistically accepting suboptimal solutions.

[0186] ① To optimize the validation set performance of the model, the objective function could be:

[0187] Minimize the validation set loss:

[0188]

[0189] Or maximize the accuracy of the validation set:

[0190] fitness(θ) = -accuracy val

[0191] ② Temperature update formula:

[0192] T k+1 =αT k

[0193] Wherein: T k : Temperature in the k-th iteration; α: Cooling coefficient, usually taken as 0.8 ≤ α ≤ 0.993. Acceptance probability formula:

[0194]

[0195] Where: ΔE: the difference between the target value of the new solution and the current solution; T: the current temperature.

[0196] ③ Optimization process by introducing the annealing algorithm

[0197] a. Initialization: Randomly initialize the hyperparameter combination θ; set the initial temperature T0.

[0198] b. Calculate fitness: Calculate the validation set performance of the current model based on the objective function.

[0199] c. Generate a new solution: Randomly perturb the current hyperparameter θ to generate a new solution θ′;

[0200] For example, the learning rate is adjusted as follows:

[0201]

[0202] Batch size varies randomly: batch_size′=batch_size±Δb

[0203] d. Accept the new solution: Calculate P(ΔE) according to the acceptance probability formula and decide whether to accept the new solution.

[0204] e. Cooling: Update temperature T k+1 =αT k .

[0205] f. Termination condition: When the temperature is below the threshold T min It may terminate when the maximum number of iterations is reached.

[0206] It should be noted that the input to the damage hazard identification model is a pre-processed high-definition image of the circuit. The output is the type of insulation damage (such as bulge, pinhole, crack) and its location coordinates.

[0207] It should be noted that establishing a damage hazard identification model can improve the accuracy and efficiency of identification. By continuously updating location information, the model can more accurately locate potential damage areas on the line, reducing false alarms and missed alarms. At the same time, the location update formula also helps to track the line status in real time, promptly detect and address potential insulation damage hazards, and ensure the safe operation of the line.

[0208] S103, Preset alarm judgment mechanism, integrate the alarm judgment mechanism into the output logic based on the damage hazard identification model;

[0209] It should be noted that the alarm determination mechanism is designed to trigger an alarm promptly upon detecting potential insulation damage, ensuring that relevant personnel can respond quickly and take necessary corrective measures. This mechanism can be based on the output of the damage detection model, setting a series of judgment conditions and thresholds. For example, an alarm is triggered when the area of ​​the damaged area identified by the model exceeds a preset threshold, or when the damage reaches a severe level. Furthermore, the alarm determination mechanism can also combine other factors, such as the importance of the line and current environmental conditions, for a comprehensive judgment, thereby improving the accuracy and reliability of the alarm. Once an alarm is triggered, the system will send the alarm information to relevant personnel via preset communication methods, such as SMS, email, or real-time push notifications, so that they can take timely action.

[0210] In some specific implementations, the alarm determination mechanism can make a judgment based on the output of the damage hazard identification model, combined with preset judgment conditions and thresholds. For example, when the area of ​​the damaged area identified by the model exceeds a certain threshold, or the degree of damage reaches a preset severity level, it is determined that an alarm is required. Furthermore, the alarm determination mechanism can also consider factors such as the importance level of the line, current environmental conditions (such as temperature and humidity), and historical damage records for a comprehensive evaluation to improve the accuracy and relevance of the alarm. If the alarm conditions are met, the alarm mechanism is triggered, and the alarm information is sent to relevant personnel in real time through preset communication channels, ensuring that they can respond quickly and take necessary repair measures, thereby effectively ensuring the safe operation of the lines along the wall.

[0211] In this embodiment of the invention, the preset alarm determination mechanism includes:

[0212] Set several alarm thresholds, and classify several alarm judgment levels based on these alarm thresholds;

[0213] The results of the potential damage identification are compared with several alarm thresholds;

[0214] The alarm level is determined based on the comparison results.

[0215] In this embodiment of the invention, the preset alarm determination mechanism further includes:

[0216] After determining the alarm level, the location and severity of the potential damage, along with suggested repair measures, are sent to the management personnel's terminal in real time.

[0217] In this embodiment of the invention, multiple alarm thresholds are set to distinguish between different degrees of damage. These thresholds can be adjusted according to actual conditions (such as historical data, expert advice, etc.). Wherein:

[0218] Minor damage threshold: When the damaged area or degree of damage is below this threshold, it is defined as minor damage.

[0219] Moderate damage threshold: Damage levels between the minor damage threshold and the severe damage threshold.

[0220] Severe damage threshold: When the damaged area or degree of damage exceeds this threshold, it is defined as severe damage.

[0221] For example, the threshold for minor damage is 5% of the area damaged; the threshold for moderate damage is 20% of the area damaged; and the threshold for severe damage is 50% of the area damaged.

[0222] Furthermore, based on the results output by the damage hazard identification model, the location, type, and severity of the damage are compared with preset alarm thresholds.

[0223] Enter the location of the damaged area, the type of damage (such as cracks, bulges, etc.), and the percentage of the damaged area.

[0224] If the damaged area is less than 5%, it is considered minor damage; if the damaged area is between 5% and 20%, it is considered moderate damage; if the damaged area is greater than 20%, it is considered severe damage.

[0225] Furthermore, the alarm level is determined based on the comparison results. The system will automatically determine which alarm level the currently detected damage belongs to and trigger the corresponding response measures.

[0226] Minor damage: Record the event and generate a low-priority report; regular inspection is recommended.

[0227] Moderate damage: Triggers a medium-priority alarm, notifying the maintenance team for further assessment and repair.

[0228] Severe damage: Triggers a high-priority alarm, immediately notifying the emergency repair team for handling.

[0229] Furthermore, once the alarm level is determined, the system will immediately send the relevant information to the relevant administrators or maintenance team.

[0230] It should be noted that pre-setting an alarm determination mechanism and integrating it into the output logic of the damage hazard identification model can significantly improve the efficiency and accuracy of damage hazard identification. By combining the alarm determination mechanism with the output logic of the damage hazard identification model, the system can quickly determine the severity and urgency of a damage hazard while identifying it, thus avoiding the time delay caused by secondary manual judgment. Furthermore, this integration mechanism can ensure the accuracy and consistency of alarm information, reduce false alarms or missed alarms caused by human factors, and provide a more reliable basis for subsequent maintenance and handling.

[0231] S104. Based on the damage hazard identification model after the fusion alarm judgment mechanism, and combined with the first real-time target image, identify the potential insulation damage hazard of the line along the wall.

[0232] In summary, this invention proposes a method for identifying potential insulation damage along wall-mounted power lines. First, by acquiring and preprocessing the first historical target image of the target line along the wall, image quality can be effectively improved and noise interference reduced, providing a more accurate data foundation for subsequent steps. Second, by establishing a potential damage identification model and using it as the core processing unit, high-precision location of the damage and efficient identification of the damage type can be achieved, greatly improving the accuracy and efficiency of identification. Third, by pre-setting an alarm judgment mechanism and integrating it into the output logic of the identification model, an alarm can be issued promptly when a potential damage is identified, enhancing the system's safety and reliability. Finally, by combining the first real-time target image with the identification of potential insulation damage along the wall, real-time monitoring of the insulation status of the line along the wall can be achieved, providing strong technical support for maintenance personnel, reducing maintenance costs, and improving the safety and stability of line operation.

[0233] Example 2: In a preferred embodiment, a real-time monitoring and alarm system can also be constructed, where the alarm determination logic includes the comprehensive confidence level of the detection results.

[0234] C final =αC cv +βC dl

[0235] Where: C cv C represents the detection confidence of a computer vision algorithm. dl Let α and β be the detection confidence of the deep learning model, and let α and β be the weights that control their contributions.

[0236] It's important to note that combining the advantages of two different technologies can improve overall detection accuracy. Computer vision algorithms are typically based on rules and feature extraction, making them suitable for handling structured data and simple pattern recognition; while deep learning models excel at automatically learning complex feature representations from large amounts of data, making them suitable for handling unstructured data and complex pattern recognition. By appropriately setting the weight parameters α and β, the contributions of these two methods can be balanced, resulting in more reliable detection results.

[0237] Alarm conditions include:

[0238]

[0239] This means that the system will trigger an alarm when the overall confidence level exceeds a preset alarm threshold. This mechanism ensures that an alarm is only issued when the detection result has a high degree of confidence, avoiding the problems of false alarms and missed alarms.

[0240] Visual monitoring includes displaying the detection results in real time through a monitoring platform: the damaged area is marked with a red box. The output includes the classification label (such as "trachoma" or "bulging") and confidence level of the detection results.

[0241] In some specific implementations, displaying the detection results in real time through a monitoring platform ensures that operators are always aware of the current status. This includes the following:

[0242] Damaged areas are marked with red boxes so that operators can see the problem at a glance.

[0243] Category Labels: Output category labels for the test results, such as "sand hole" or "bulging belly", to help operators understand the specific damage type.

[0244] Confidence level display: Displays the confidence level of each test result, enabling operators to assess the reliability of the test.

[0245] To achieve the above functions, the design of the monitoring platform should consider the following aspects:

[0246] User-friendly interface: Provides a simple and clear user interface, enabling operators to quickly get started and use it efficiently.

[0247] Real-time updates: Ensure that test results are updated and displayed on the platform in real time so that operators can respond promptly.

[0248] Historical record query: Provides a historical record query function, enabling operators to review past test results for analysis and summarization.

[0249] Alarm information prompts: When an alarm is triggered, the platform should provide prompts through sound, color changes, or other means to ensure that operators do not miss important information.

[0250] Example 3: This example also provides a system for identifying potential insulation damage to wall-mounted wiring, including:

[0251] The data acquisition and processing module is used to acquire the first historical target image along the target wall line and to preprocess the first historical target image;

[0252] The first historical target image is obtained through a preset path planning algorithm;

[0253] The preset path planning algorithm is corrected by setting inspection constraints;

[0254] The model building module is used to build a damage hazard identification model. The input of the damage hazard identification module is the preprocessed first historical target image.

[0255] The damage hazard identification model is used to locate the damage location and identify the damage type;

[0256] The early warning mechanism fusion module is used to preset the alarm judgment mechanism and integrate the alarm judgment mechanism into the output logic based on the damage hazard identification model;

[0257] The identification module is used to identify potential insulation damage hazards along the wall lines based on the damage hazard identification model after the fusion alarm judgment mechanism, combined with the first real-time target image.

[0258] The above-mentioned unit modules can be embedded in the processor of the electronic device in hardware form or independent of it, or they can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0259] This embodiment also provides an electronic device, which can be a terminal. The electronic device includes a processor, memory, communication interface, display screen, and input device connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for identifying potential insulation damage hazards along wall-mounted wiring. The display screen can be a liquid crystal display (LCD) or an e-ink display. The input device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the device's casing, or an external keyboard, touchpad, or mouse.

[0260] This embodiment also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it performs the following steps:

[0261] Acquire the first historical target image along the target's path along the wall, and preprocess the first historical target image;

[0262] The first historical target image is obtained through a preset path planning algorithm;

[0263] The preset path planning algorithm is corrected by setting inspection constraints;

[0264] A damage hazard identification model is established, and the input of the damage hazard identification module is the preprocessed first historical target image;

[0265] The damage hazard identification model is used to locate the damage location and identify the damage type;

[0266] A preset alarm determination mechanism is established and integrated into the output logic based on the damage hazard identification model.

[0267] Based on the damage hazard identification model after the integrated alarm judgment mechanism, and combined with the first real-time target image, the potential for insulation damage along the wall is identified.

[0268] 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.

[0269] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0270] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for identifying potential insulation damage in wall-mounted wiring, characterized in that, include: Acquire a first historical target image along the target's path along the wall, and preprocess the first historical target image; The first historical target image is acquired using a preset path planning algorithm; The preset path planning algorithm is modified by setting inspection constraints; A damage hazard identification model is established, and the input of the damage hazard identification module is the preprocessed first historical target image; The damage hazard identification model is used to locate the damage location and identify the damage type; A preset alarm determination mechanism is established, and the alarm determination mechanism is integrated into the output logic based on the damage hazard identification model; Based on the damage hazard identification model after the integrated alarm judgment mechanism, and combined with the first real-time target image, the potential for insulation damage along the wall is identified.

2. The method for identifying potential insulation damage in wall-mounted wiring as described in claim 1, characterized in that, The preset path planning algorithm includes: Establish the objective function for the shortest inspection path and the inspection constraints; The objective function for the shortest inspection path is determined by the flight distance between sampling points and the indicator variables for path selection between sampling points. The inspection constraints include inspection sampling point constraints and endurance time constraints.

3. The method for identifying potential insulation damage in wall-mounted wiring as described in claim 2, characterized in that, The algorithm correction by setting inspection constraints includes: Convert inspection constraints into algorithmic constraints; The algorithm constraints include the constraint that each sampling point can only be accessed once, the constraint that duplicate access points are removed, the constraint that unaccessed points are supplemented, and the cruise correction constraint. The cruise correction constraint is that if the total path distance exceeds the range limit, the path will be randomly adjusted or a relay point will be introduced to decompose the path.

4. The method for identifying potential insulation damage in wall-mounted wiring as described in claim 3, characterized in that, The establishment of the potential damage identification model includes: A module for acquiring the first damaged region based on frame difference method and threshold segmentation is established; Establish a second damaged area acquisition module based on optical flow method; The first damaged area acquisition module and the second damaged area acquisition module are connected in series in sequence.

5. The method for identifying potential insulation damage in wall-mounted wiring as described in claim 4, characterized in that, The establishment of the damage hazard identification model also includes: Obtain the hazard results corresponding to the preprocessed first historical target image; The hazard results corresponding to the preprocessed first historical target image are used as an output of the damage hazard identification model. The output of the second damaged area acquisition module is used as another output of the damage hazard identification model; The results of damage hazard identification are obtained from the two outputs.

6. The method for identifying potential insulation damage in wall-mounted wiring as described in claim 5, characterized in that, The preset alarm determination mechanism includes: Several alarm thresholds are set, and several alarm judgment levels are divided according to the aforementioned alarm thresholds; The results of the potential damage identification are compared with the aforementioned alarm thresholds; The alarm level is determined based on the comparison results.

7. The method for identifying potential insulation damage in wall-mounted wiring as described in claim 6, characterized in that, The preset alarm determination mechanism also includes: After determining the alarm level, the location and severity of the potential damage, along with suggested repair measures, are sent to the management personnel's terminal in real time.

8. A system for identifying potential insulation damage to wall-mounted wiring, using the method described in any one of claims 1 to 7, characterized in that, include: The data acquisition and processing module is used to acquire a first historical target image along the target wall line and to preprocess the first historical target image. The first historical target image is acquired using a preset path planning algorithm; The preset path planning algorithm is modified by setting inspection constraints; The model building module is used to build a damage hazard identification model. The input of the damage hazard identification module is the preprocessed first historical target image. The damage hazard identification model is used to locate the damage location and identify the damage type; The early warning mechanism fusion module is used to preset the alarm judgment mechanism and integrate the alarm judgment mechanism into the output logic based on the damage hazard identification model; The identification module is used to identify potential insulation damage hazards along the wall lines based on the damage hazard identification model after the fusion alarm judgment mechanism, combined with the first real-time target image.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method for identifying potential insulation damage hazards along wall-mounted lines as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for identifying potential insulation damage to wall-mounted lines according to any one of claims 1 to 7.

Images