Intelligent evaluation and optimization method and system for safety clearance of power distribution network line

Abstract

The invention discloses an intelligent evaluation and optimization method and system for a safety gap of a power distribution network line. The method comprises the following steps: collecting multi-source sensing data of a target operation area in real time; carrying out fusion processing on the multi-source sensing data, constructing a three-dimensional space environment model containing semantic information of an electrified body, a grounding body and an operation instrument, and extracting real-time geometric gaps among objects; performing dynamic safety risk assessment based on a pre-stored safety standard library and the real-time geometric gap; and in response to a result of the dynamic safety risk assessment, automatically planning at least one recommended operation scheme meeting dynamic safety constraints in the three-dimensional space environment model, the recommended operation scheme comprising a motion path of an operation apparatus and / or an operation mode of an operator. According to the scheme, the crossing from passive warning to active navigation is realized, and the safety, accuracy and efficiency of non-power-cut operation are fundamentally improved.

Description

Technical Field

[0001] This invention relates to the field of power safety operation technology, specifically to a method and system for intelligent assessment and optimization of safety clearances in distribution network lines, and particularly to a method and system for intelligent assessment and optimization of safety clearances in distribution network lines for uninterrupted power supply operations. Background Technology

[0002] With increasing societal demands for power supply reliability, live-line working on power distribution networks has become the mainstream method for equipment maintenance, fault handling, and user connection. The core safety prerequisite for live-line working or working near live conductors is maintaining sufficient safety air gaps to prevent breakdown discharges that could cause personal injury or equipment accidents.

[0003] Currently, on-site clearance assessment mainly relies on two methods: First, manual judgment based on experience and simple tool measurements, such as using insulated distance measuring rods or visual estimation. This method is greatly affected by personnel experience, lighting conditions, and psychological factors, resulting in low accuracy and poor consistency. It is particularly prone to misjudgment in complex environments (such as multi-circuit lines or adjacent buildings), posing significant safety hazards. Second, sensor-assisted methods, such as laser distance meters, are used. While this improves the accuracy of single-point distance measurements, it remains essentially a discrete, static, and isolated measurement method. Workers need to manually aim and measure each suspected risk point, resulting in a huge workload, low efficiency, and heavy reliance on the operator's subjective judgment to select measurement points, easily overlooking critical risk sections. More importantly, it fails to establish a holistic three-dimensional spatial relationship of the work environment, making it difficult to form an effective overall assessment of the safety clearances of distribution network lines, thus hindering the improvement of the overall safety and accuracy of live-line work on the distribution network. Summary of the Invention

[0004] To address the technical challenges of relying on manual experience for work gap assessment, the enormous workload associated with sensor assistance, and the inability to improve the overall safety and accuracy of live-line work in power distribution networks, this invention provides an intelligent assessment and optimization method and system for safety gaps in power distribution networks. This solution integrates multi-source data, including laser and visual data, to construct a semantic three-dimensional work environment model in real time. Combined with dynamically adjusted safety standards, it automatically assesses risks and plans safe work paths and methods, thereby achieving a leap from passive alarms to proactive navigation and fundamentally improving the safety, accuracy, and efficiency of live-line work.

[0005] On the one hand, this invention provides a method for intelligent assessment and optimization of safety clearances in power distribution network lines, comprising the following steps: S1: Real-time acquisition of multi-source sensing data of the target work area, wherein the multi-source sensing data includes at least laser point cloud data and visible light image data; S2: The multi-source sensing data is fused and processed to construct a three-dimensional spatial environment model containing semantic information of all target objects; based on the three-dimensional spatial environment model, the real-time geometric gaps between the target objects are calculated and extracted; S3: Based on the pre-stored security standard library and the real-time geometric gap, a dynamic security risk assessment is performed, and the security thresholds in the security standard library can be dynamically adjusted according to real-time environmental parameters; S4: In response to the results of the dynamic safety risk assessment, at least one recommended operation plan that meets the dynamic safety constraints is automatically planned in the three-dimensional spatial environment model. The recommended operation plan includes the movement path of the operating equipment and / or the operation mode of the operator.

[0006] Furthermore, in step S2, the multi-source perception data is fused and processed to construct a three-dimensional spatial environment model containing semantic information of all target objects, specifically including: S21: Generate an initial three-dimensional point cloud framework based on the laser point cloud data; S22: Perform image recognition on the visible light image data to identify all target objects and assign them corresponding semantic labels, wherein the target objects include at least: conductors, towers, insulators, working equipment and workers; S23: Spatial registration and semantic fusion are performed between the identified target object and its semantic label and the initial 3D point cloud framework to generate the 3D spatial environment model.

[0007] Furthermore, the real-time environmental parameters on which the dynamic adjustment of the safety threshold in step S3 is based include at least one of the following: real-time wind speed, air temperature and humidity, and atmospheric pressure; the rules for the dynamic adjustment are based on an air gap discharge characteristic model or a pre-established experimental database.

[0008] Furthermore, within the three-dimensional spatial environment model, at least one recommended work plan that satisfies dynamic safety constraints is automatically planned, specifically including: Define a dynamic safety envelope for the working equipment or the limbs of the workers in the three-dimensional spatial environment model; With the constraint that the gap between the dynamic safety envelope and all hazard sources is always not less than the corresponding dynamic safety threshold, a collision-free path search is performed in the three-dimensional spatial environment model. If multiple feasible paths exist, they are sorted according to a preset optimization objective, which includes at least one or more of path length, operational complexity, or overall safety margin. Furthermore, the collision-free path search employs a sampling-based planning algorithm, including a fast exploratory random tree or an improved version thereof.

[0009] Furthermore, the method further includes step S5: By using augmented reality (AR) devices, key information from the recommended work plan is overlaid onto the actual field of vision of the workers; The key information includes: real-time highlighted hazardous areas of live conductors, overlaid recommended navigation paths, and real-time dynamic distance values ​​between the working equipment and live conductors.

[0010] Furthermore, the method further includes the following steps before step S1: Based on historical scenario data or pre-collected data, construct or invoke a digital twin environment; In the digital twin environment, steps S1-S4 are executed to generate multiple pre-selected job schemes; By comparing and analyzing the multiple pre-selected operation plans, optimization decision suggestions are output to guide actual operations.

[0011] Furthermore, the method also includes the following steps: Record multi-source sensing data, actual operation paths, and results during the actual operation process; The actual work path is compared and analyzed with the recommended work plan, and the results of the comparison and analysis are used to optimize the calculation model or algorithm involved in steps S2, S3 and / or S4.

[0012] Based on the comparative analysis results, at least one of the following should be iteratively optimized: a) A perception fusion algorithm used to construct the three-dimensional spatial environment model, specifically a visualization image recognition model; b) A dynamic security threshold adjustment model used to perform the dynamic security risk assessment, specifically a dynamic security threshold adjustment model; c) A planning algorithm used to generate the recommended job scheme.

[0013] On the other hand, the present invention also discloses an intelligent assessment and optimization system for safety clearances of distribution network lines, used to implement the intelligent assessment and optimization method for safety clearances of distribution network lines as described above. The intelligent assessment and optimization system for safety clearances of distribution network lines includes: A multi-source sensing unit is used to collect multi-source sensing data of the target work area in real time. The multi-source sensing data includes at least laser point cloud data and visible light image data. The data processing and fusion unit is used to fuse and process the multi-source sensing data, construct a three-dimensional spatial environment model containing semantic information of charged bodies, grounded bodies and working equipment, and extract the real-time geometric gaps between each object. The safety assessment and planning unit is used to perform dynamic safety risk assessment based on the pre-stored safety standard library and the real-time geometric gap. The safety thresholds in the safety standard library can be dynamically adjusted according to real-time environmental parameters. The interactive output unit is used to output the recommended work plan and risk information.

[0014] Furthermore, the multi-source sensing unit is integrated into a drone, an insulated bucket truck, or a wearable device; the interactive output unit is augmented reality (AR) glasses or an explosion-proof mobile terminal.

[0015] Compared with the prior art, the present invention has at least the following beneficial effects: 1) By fusing laser point cloud data and visible light image data, not only is geometric information with the required accuracy obtained, but also semantic information of the object is obtained, thus constructing a digital twin operation scenario that can be understood and reasoned by computers. This completely eliminates the reliance on human experience and greatly improves the accuracy and reliability of the assessment.

[0016] 2) By incorporating real-time environmental parameters into the safety distance calculation model, the safety threshold can be dynamically adjusted according to weather and working conditions, realizing the transformation from fixed distance to dynamic insulation strength matching. Under the premise of ensuring absolute safety, it can free up more working space and improve operational flexibility.

[0017] 3) It can automatically plan safe work paths, no longer acting as a post-event alarm, but as a pre-event and in-event guide, realizing the leap from early warning to pre-event navigation.

[0018] 4) By using AR technology to perfectly integrate virtual safety boundaries and navigation paths with real-world scenarios, WYSIWYG guidance is achieved, which greatly reduces the cognitive load and mental stress of operators, making complex operations more intuitive and simpler.

[0019] 5) By recording and feeding back actual operation data, the system can continuously optimize its recognition model, dynamic threshold model, and planning algorithm, achieving iterative performance improvements and becoming more intelligent. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. 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.

[0021] Figure 1 A flowchart illustrating the steps of the intelligent assessment and optimization method for safety gaps in power distribution lines provided in this embodiment of the invention.

[0022] Figure 2 This is a schematic diagram illustrating the principle of constructing a three-dimensional spatial environment model as provided in an embodiment of the present invention.

[0023] Figure 3 The system framework diagram of the intelligent assessment and optimization system for safety gaps in power distribution lines provided in the embodiments of the present invention is shown. Detailed Implementation

[0024] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0025] It should be understood that the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention. Example 1

[0026] refer to Figure 1 This invention illustrates the method of replacing intermediate pole insulators on a 10kV distribution line using an insulated bucket truck as an example. This invention provides an intelligent assessment and optimization method for safety clearances in distribution network lines, comprising the following steps: S1: Real-time acquisition of multi-source sensing data of the target work area, including at least laser point cloud data and visible light image data; S2: Fusion processing of multi-source sensing data to construct a three-dimensional spatial environment model containing semantic information of all target objects; based on the three-dimensional spatial environment model, calculate and extract the real-time geometric gaps between target objects; S3: Based on the pre-stored safety standard library and real-time geometric gaps, dynamic safety risk assessment is performed. The safety thresholds in the safety standard library can be dynamically adjusted according to real-time environmental parameters. S4: In response to the results of dynamic safety risk assessment, automatically plan at least one recommended operation scheme that meets dynamic safety constraints in the three-dimensional spatial environment model. The recommended operation scheme includes the movement path of the operating equipment and / or the operation mode of the operator.

[0027] In this invention, all target objects include at least: one or more live conductors, one or more grounded conductors, working equipment, and workers. The real-time geometric gap between target objects includes the real-time geometric gap between any two of the target objects.

[0028] The core of step S1 is the real-time acquisition of multi-source sensing data of the target work area. This is primarily achieved by integrating the sensing unit (hardware) onto the outer guardrail of the insulated bucket of the insulated boom truck, forming a stable sensing platform. After the boom truck is in position, the system is activated. The operator controls the bucket to rotate slowly, and the sensing unit performs a 360-degree scan of the towers, three-phase conductors, crossarms, insulators, nearby trees, and buildings, continuously acquiring multi-source sensing data of the target work area, including at least laser point cloud data and visible light image data. Preferably, the laser point cloud data acquisition can utilize a 16-line or 32-line mechanical rotating lidar with a horizontal field of view of 360 degrees, a vertical field of view of 30°, a measurement range ≥50 meters, and an accuracy of ±2 cm. Scanning at a frequency of 10Hz, a high-density set of three-dimensional coordinate points on the surfaces of all objects in the work area is obtained. The visible light image data acquisition can utilize a global shutter industrial camera with a resolution of 2 megapixels or higher, equipped with a wide-angle lens, rigidly fixed to the lidar and jointly calibrated. High-definition RGB images are acquired at a frame rate of 15-30fps. During data synchronization, hardware triggering or timestamp synchronization ensures that each frame of image is aligned with the corresponding laser point cloud in time. After the boom truck is in position, the operator slowly rotates the bucket approximately 180 degrees. During this process, the LiDAR and camera work synchronously, completing a complete scan and imaging of the tower, three-phase conductors, insulator strings, jumpers, crossarms, grounding guy wires, and surrounding background (such as trees) within approximately 5-15 seconds, providing a complete data source for subsequent processing.

[0029] In embodiments of the present invention, such as Figure 2As shown, the core of step S2 is to fuse and process multi-source sensing data, construct a semantic 3D spatial environment model, and extract real-time geometric gaps. This step S2 can be completed in a vehicle-mounted ruggedized computer (equipped with a GPU). The fusion and processing of multi-source sensing data includes point cloud preprocessing, image recognition, data fusion and modeling, and gap calculation. During point cloud preprocessing, the original point cloud is filtered (removing isolated noise points), downsampled (reducing the data volume while preserving features), and its coordinate system is normalized. During image recognition, a pre-trained deep learning model is used to process the image. For example, an instance segmentation model based on the Transformer architecture (such as Mask2Former) is adopted, which has been trained and fine-tuned on a massive power equipment image dataset. The instance segmentation model takes a frame of image as input and outputs the category label of each target in the image (e.g., "10kV conductor", "composite insulator", "angle iron crossarm", "insulated bucket truck work bucket", "worker's arm") and its precise pixel-level mask. During data fusion and modeling, a pre-calibrated camera-LiDAR extrinsic parameter matrix is ​​used to back-project each target mask identified in the image onto a 3D point cloud space. For example, pixel regions in the image belonging to "Phase A conductor" are mapped to a series of corresponding points in the point cloud. Semantic labels obtained from image recognition (such as "Phase A conductor") are assigned to these mapped point cloud points. This process is repeated for all identified targets. Ultimately, a "semantic point cloud" is obtained. Each point contains not only (x, y, z) coordinates but also its semantic label. This semantic point cloud is the core data representation of the 3D spatial environment model. When calculating gaps, based on this 3D spatial environment model, the minimum spatial interval between any two different semantic object point sets is calculated. For example, the shortest distance between the point set labeled "work bucket surface" and the point set labeled "Phase B conductor" is calculated, which represents the real-time geometric gap between them (e.g., 1.85 meters).

[0030] In typical complex scenarios, such as replacing insulator strings on tension poles, the system successfully segmented jumpers from images and identified their categories. Through fusion, it accurately distinguished between "jumper" point clouds and "dominant line" point clouds in the 3D point cloud. Subsequently, the system calculated the minimum distance between the current contour of the working bucket and the nearest jumper, thus identifying jumpers as the primary gap risk point when a straight line approaches the insulator string. This step solves the core technical problem of how to enable computers to understand "what objects are in a 3D scene" and "how far apart they are," which is the foundation for realizing intelligent assessment of safety gaps in distribution network lines.

[0031] Preferably, step S2 specifically includes: S21: Preprocess the acquired laser point cloud data to generate an initial 3D point cloud framework for the target work area. The preprocessing process includes filtering, registration, and other steps, and adopts a "tightly coupled" fusion strategy, utilizing multiple feature information such as the geometric coordinates and reflection intensity of the points simultaneously during the point cloud registration process.

[0032] S22: For visible light image data, a deep learning object detection and segmentation model is used to identify target objects such as conductors, insulators, towers, trees, buildings, insulated bucket trucks, workers, and their limbs, and their pixel-level contours are extracted. In the image recognition step S22, in addition to the instance segmentation model (such as Mask2Former), a lightweight network can be connected in parallel to specifically identify power lines (due to their slender and prominent features).

[0033] S23: Using camera calibration parameters and a point cloud-image registration algorithm, the target objects and their semantic labels identified in step S22 are mapped to the initial 3D point cloud framework in step S21, forming a dense 3D point cloud with semantic labels, i.e., a 3D spatial environment model. During the fusion process in step S23, a variant of the Iterative Closest Point (ICP) algorithm based on features (such as point cloud edges and image edges) is used for fine registration, and sub-pixel optimization is performed on the boundaries of image segmentation to improve the reconstruction accuracy of small targets such as wires in the 3D model.

[0034] In addition, for conductors that are partially occluded in the image, the system uses the point cloud features with high reflectivity and contextual information (parallel to other conductors) to perform reasonable completion in the 3D model, ensuring the integrity of the gap calculation benchmark.

[0035] The core of step S3 is to conduct a dynamic safety risk assessment based on a pre-stored safety standard library and real-time geometric gaps. The safety standard library is a structured database storing basic safety distance tables (such as 0.7 meters for 10kV ground potential operations in the "Safety Regulations"), various correction coefficient models, or lookup tables. Preferably, the real-time environmental parameters upon which the dynamic adjustment of the safety threshold in step S3 is based include at least one of the following: real-time wind speed, air temperature and humidity, and atmospheric pressure. The rules for dynamic adjustment are based on air gap discharge characteristic models or pre-established experimental databases. For example, the safety standard library stores a multi-level correction model. The first layer queries the national standard basic table; the second layer accesses the real-time meteorological data interface API; and the third layer embeds a lightweight neural network model. This multi-level correction model takes wind speed, humidity, and temperature as inputs and correction coefficients as outputs. This network has been trained on gap breakdown experimental data under different meteorological conditions in a high-voltage laboratory and can capture complex nonlinear relationships.

[0036] For example, the micro-weather station integrated into the sensing platform measures wind speed (e.g., 4 m / s), temperature (e.g., 30℃), relative humidity (e.g., 70%), and atmospheric pressure in real time. The calculation or adjustment of dynamic safety thresholds is based on the following formulas or lookup tables. For example, the dynamic correction model in the safety standard library is invoked using the following formula:

[0037] in, The basic safety distance (obtained by referring to a table based on voltage level and working method). This is a correction factor determined based on experimental data or industry standards. The system dynamically calculates a safety threshold of 0.7 meters. In the hot and humid environment before summer thunderstorms, the system calculates a higher correction factor using a neural network model, automatically increasing the dynamic safety threshold by 15%. The current dynamic safety threshold is then calculated. .

[0038] Risk assessment level based on minimum real-time interval With the current dynamic security threshold The ratio is used to divide the data, for example: the extracted real-time geometric gap (e.g., the working bucket-jump wire distance of 1.85 meters) is compared with... Comparison, Low risk (green) Medium risk (yellow) High risk (red). In this example, 1.85 meters > 1.23 meters (1.5 × 0.82), so it is assessed as low risk.

[0039] Due to increased humidity and slightly stronger winds in the afternoon before summer thunderstorms, the system dynamically calculated the safety threshold, raising it from the standard 0.7 meters to 0.82 meters. This scientifically reflects the impact of the environment on insulation strength. The system determines the current working bucket position is safe, but if subsequent path planning causes the gap to shrink to 1.0 meter ( If the area is near a medium-risk zone, a yellow alert will be triggered. This step solves the technical problem of how to scientifically and dynamically define the "safety boundary" based on the actual environment, overcoming the limitations of fixed thresholds.

[0040] The core of step S4 is to automatically plan a recommended operation scheme that meets dynamic safety constraints within the three-dimensional spatial environment model. The boom (work bucket) of the bucket truck is simplified as a rigid body moving in three-dimensional space. Its movement is controlled by several joint angles (rotation, pitch, extension), which constitute a high-dimensional configuration space.

[0041] Preferably, step S4 specifically includes: S41: Based on the geometric and kinematic models of the working equipment (such as the boom of an insulated bucket truck), define a dynamic safety envelope that varies with its posture in the three-dimensional spatial environment model. This envelope adds a dynamic safety threshold to the basic geometric model. As an offset. For example, expanding outwards from the geometry of the working bucket by a... A distance of 0.82 meters (e.g.) is used to generate a larger virtual contour, namely the dynamic safety envelope. The planning must ensure that this dynamic safety envelope does not intersect with any charged body semantic model.

[0042] S42: The path planning problem is transformed into searching for a collision-free path from the initial state to the target state in a high-dimensional configuration space (including position and orientation). A "collision" is defined as a state where the gap between the dynamic safety envelope and all charged and grounded body models is less than zero (i.e., interference occurs). A sampling-based motion planning algorithm, such as the Fast Exploratory Random Tree Augmentation (RRT*) algorithm, is used to search in this configuration space. For example, with the current configuration of the working bucket as the root node of the search tree, a target configuration (e.g., the position and orientation of a string of faulty insulators) is randomly sampled in the configuration space, and the node closest to the sampled point is found in the existing search tree. A new node is generated by extending a small step from the nearest node towards the sampled point. The minimum distance between the dynamic safety envelope corresponding to the new node and all "charged body" models (conductors, jumpers) in the 3D environment model is calculated. If the distance is less than 0 (i.e., interference occurs), the node is discarded. If the distance is greater than 0, the node is accepted and added to the tree, and its actual minimum gap value is recorded.

[0043] S43: When expanding a new node using the RRT* algorithm, not only is the node's state checked for "collision," but the minimum clearance between the dynamic safety envelope and all hazards in that state is also calculated. Will As part of the path cost or as an attraction / repulsion force for expansion, the tree is guided to grow towards safer (larger gaps) regions. For example, for a new node, it is searched to see if there are other parent nodes in its neighborhood that make the path cost from the root node to the new node (such as the weighted sum of path length and safety margin) lower, thereby optimizing the tree structure and asymptotically approaching the optimal path. Repeat the previous steps until a node that reaches the target working area (near the insulator string) is generated. Tracing back from the target node to the root node, a motion path consisting of a series of configurations is obtained. This path is converted into a series of boom truck operation commands (e.g., first retract the boom by X meters, then rotate counterclockwise by Y degrees), which constitutes a recommended operation plan.

[0044] S44: After generating one or more feasible paths, based on the total path length and average safety margin ( The algorithm comprehensively evaluates factors such as average value and operational complexity (e.g., attitude change frequency) to output the optimal path as the recommended job path. For example, Cost = α × path length + β × (1 / average safety margin) can be used as the path cost function of the RRT* algorithm. Here, α and β are adjustable weights, which can be adjusted to implement different strategies such as shortest path priority or safest path priority. Finally, based on the comprehensive evaluation, the optimal path is output as the recommended job path.

[0045] In this invention, a feasible path specifically refers to a series of continuous machine states (positions and attitudes) generated by a path search algorithm (such as RRT*) within the configuration space corresponding to the three-dimensional spatial environment model. This sequence is considered "feasible" only if it meets all of the following technical conditions: 1) Geometric Feasibility: The minimum spatial interval between the dynamic safety envelope corresponding to each state on the path and the three-dimensional model of all hazards (energized bodies, grounded bodies) is greater than or equal to the dynamic safety threshold calculated in the current state (i.e., the gap ≥ 1 / 2). This ensures there is no risk of collision.

[0046] 2) Kinematic feasibility: The transition between adjacent states on the path must be within the constraints of the actual kinematic model of the working equipment (such as an insulated bucket truck), such as the range of joint angle changes, extension length, rotation speed, etc., to ensure that the path can be actually executed by the physical equipment.

[0047] 3) Task accessibility: The destination of the path must enable the working equipment or personnel to effectively perform the target task (such as reaching the operation position for replacing insulators).

[0048] Therefore, during the planning process, the system verifies each searched path against the aforementioned conditions. Only paths that simultaneously meet all three conditions are deemed "feasible" and proceed to the subsequent optimization and ranking stage. If no path meets all conditions, the system will output a planning failure warning and highlight the main constraint conflicts.

[0049] To address the issue of jumper obstruction on tension poles, the RRT* algorithm might plan a path that first instructs the working bucket to descend and move slightly backward, bypassing the space beneath the jumper, and then approaches the insulator string from the side of the line, rather than attempting to directly cross the narrow gap between the jumper and the main conductor. During planning, the algorithm ensures that the distance between the safety envelope and the jumper / main conductor is greater than 0.82 meters at each extension node, and ultimately selects a path with a shorter total travel distance and a higher average safety margin. This step is the core of the invention, solving the problem of how to automatically calculate a safe (meeting dynamic gap constraints) and efficient (approximately optimal) sequence of operational actions—a problem that existing technologies cannot address.

[0050] Preferably, the method further includes step S5: providing visual guidance through augmented reality (AR) devices. The AR glasses use built-in SLAM technology to locate their position and orientation in physical space in real time. The system renders the 3D spatial environment model and the planned path on the AR glasses' display screen in real time through a unified coordinate system transformation, and accurately overlays them with the real scene constructed by SLAM. When the actual movement deviates from the planned path beyond a threshold, or when the real-time calculated gap enters the yellow warning zone, the virtual path in the AR interface flashes and is superimposed with a prominent arrow for correction guidance. In embodiments of the present invention, information such as the dangerous area of ​​charged bodies, recommended work path, and real-time dynamic distance in the 3D spatial environment model is accurately overlaid onto the operator's real field of vision using spatial mapping technology. Through the AR glasses, the operator can see a glowing blue light strip extending from the front of the work bucket, winding towards the insulator. When the operator operates the bucket truck, a green "ghost bucket" phantom moves along the blue light strip, indicating the next position the operator should reach. If the operation is too fast and causes the ghost bucket to approach the red danger zone, the phantom will turn orange and issue an alarm.

[0051] In an embodiment of the invention, preferably, a separate job simulation module is developed in the background. This job simulation module calls the same model and algorithm as the onboard system, but allows users to interactively set the start and end points of the job and environmental parameters in a digital twin environment. This job simulation module based on the digital twin environment allows operators to run multiple plans in batches, statistically analyze the success rate, average time, and risk indicators of different schemes, and generate comparative reports for operators to make decisions. For example, before a complex urban cable head fabrication operation, operators use the job simulation module in the office. Operators try three different initial parking positions for the boom truck, and the system quickly provides three feasible paths and evaluations for each position. Operators select the scheme with the highest overall score and send this scheme to the onboard system of the corresponding boom truck with one click as the default recommended scheme for this operation.

[0052] In a preferred embodiment of the present invention, the method further includes the following steps: Record multi-source sensing data, actual operation paths, and results during the actual operation process; The actual work path is compared and analyzed with the recommended work plan, and the analysis results are used to optimize the image recognition model, dynamic safety threshold adjustment model, or path planning algorithm.

[0053] In implementation, the system establishes a job log database to structurally record each job's: environmental parameters, raw perception data (after anonymization), planned path, actual control path, all real-time intermittent records, and alarm events. Periodically (e.g., weekly), new log data is used to incrementally train the image recognition model, Bayesian updates are performed on the parameters of the dynamic threshold model, and actual path data is used to correct kinematic model errors, thus forming a closed-loop optimization system. For example, the system discovered that under certain lighting conditions, the image recognition model's accuracy in segmenting older ceramic insulators decreased. By collecting job log data from multiple such scenarios and performing targeted retraining, the next-generation model improved its recognition accuracy by 25% in this scenario. Similarly, by analyzing a large amount of actually safely completed job data, the system fine-tuned the dynamic correction coefficients for specific wind speed ranges, making the threshold settings more closely match the actual safety boundaries. Example 2

[0054] In another embodiment of the present invention, the sensing unit is mounted on a drone for automatically inspecting operating power lines, assessing the gaps between the lines and trees, and planning safe drone inspection routes. In this second embodiment, compared to embodiment 1, the intelligent assessment and optimization method for safety gaps in power distribution network lines of the present invention mainly includes the following differences: Step S1+: The drone flies along a pre-set general route, and its onboard lidar and binocular camera collect data simultaneously.

[0055] Step S2+: The airborne or ground station computer reconstructs a three-dimensional model of the line corridor in real time and identifies the conductors and trees.

[0056] Step S3+: Based on real-time wind speed and tree sway model, dynamically assess the minimum possible distance between the tree and the guide wire under wind deflection. If a tree is found to be only 0.5 meters away from the guide wire under maximum wind deflection, which is below the dynamic safety threshold of 1.2 meters, the system immediately marks it as a "high-risk tree obstacle point".

[0057] Step S4+: The system not only reports the risk points, but also plans a new return or continued inspection path for the drone. This path ensures that the drone itself maintains a safe distance from all live wires, while taking photos of the risk points from the best angle for evidence collection.

[0058] Step S5+: Ground operators can see the highlighted risk points in the 3D channel model and the safe flight path planned by the drone through AR glasses or a screen. Example 3

[0059] This invention also discloses an intelligent assessment and optimization system for safety clearances in power distribution networks, used to implement the aforementioned intelligent assessment and optimization method for safety clearances in power distribution networks. Figure 3 As shown, the intelligent assessment and optimization system for safety gaps in power distribution networks mainly includes: a multi-source sensing unit and a data processing and fusion unit at the hardware layer; a safety assessment and planning unit and an interactive output unit at the processing layer. Additionally, a work simulation module for operators is provided at the interactive layer. This work simulation module, based on a digital twin environment, allows operators to run multiple plans in batches, statistically analyze the success rate, average time, and risk indicators of different schemes, and generate comparative reports for operators to make decisions.

[0060] Among them, the multi-source sensing unit is used to collect multi-source sensing data of the target operation area in real time. The multi-source sensing data includes at least laser point cloud data and visible light image data.

[0061] The data processing and fusion unit is used to fuse and process multi-source sensing data, construct a three-dimensional spatial environment model containing semantic information of charged bodies, grounded bodies and working equipment, and extract the real-time geometric gaps between each object.

[0062] The safety assessment and planning unit is used to perform dynamic safety risk assessment based on a pre-stored safety standard library and real-time geometric gaps. The safety thresholds in the safety standard library can be dynamically adjusted according to real-time environmental parameters.

[0063] The interactive output unit is used to output recommended work plans and risk information.

[0064] Preferably, the multi-source sensing unit is integrated into a drone, an insulated bucket truck, or a wearable device.

[0065] Preferably, the interactive output unit is augmented reality (AR) glasses or an explosion-proof mobile terminal.

[0066] Preferably, the system perception unit is mounted on a drone to automatically inspect the running line, assess the gap between the line and the trees, and plan a safe drone inspection route.

[0067] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0068] In this embodiment of the invention, the term "and / or" describes the relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The character " / " generally indicates that the preceding and following associated objects have an "or" relationship.

[0069] The various embodiments in this specification are described in a related manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0070] In particular, the device embodiments are basically similar to the method embodiments, so they are described in a simpler way. For relevant details, please refer to the description of the method embodiments.

[0071] For ease of description, the above apparatus is described by dividing it into various functional units / modules. Of course, in implementing this invention, the functions of each unit / module can be implemented in one or more software and / or hardware.

[0072] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0073] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for intelligent assessment and optimization of safety clearances in power distribution network lines, characterized in that, Includes the following steps: S1: Real-time acquisition of multi-source sensing data of the target work area, wherein the multi-source sensing data includes at least laser point cloud data and visible light image data; S2: The multi-source sensing data is fused and processed to construct a three-dimensional spatial environment model containing semantic information of all target objects; based on the three-dimensional spatial environment model, the real-time geometric gaps between the target objects are calculated and extracted; S3: Based on the pre-stored security standard library and the real-time geometric gap, a dynamic security risk assessment is performed, and the security thresholds in the security standard library can be dynamically adjusted according to real-time environmental parameters; S4: In response to the results of the dynamic safety risk assessment, at least one recommended operation plan that meets the dynamic safety constraints is automatically planned in the three-dimensional spatial environment model. The recommended operation plan includes the movement path of the operating equipment and / or the operation mode of the operator.

2. The intelligent assessment and optimization method for safety clearances of power distribution lines according to claim 1, characterized in that, In step S2, the multi-source sensing data is fused and processed to construct a three-dimensional spatial environment model containing semantic information of all target objects, specifically including: S21: Generate an initial three-dimensional point cloud framework based on the laser point cloud data; S22: Perform image recognition on the visible light image data to identify all target objects and assign them corresponding semantic labels, wherein the target objects include at least: conductors, towers, insulators, working equipment and workers; S23: Spatial registration and semantic fusion are performed between the identified target object and its semantic label and the initial 3D point cloud framework to generate the 3D spatial environment model.

3. The intelligent assessment and optimization method for safety clearances of power distribution lines according to claim 1, characterized in that, The real-time environmental parameters on which the dynamic adjustment of the safety threshold in step S3 is based include at least one of the following: real-time wind speed, air temperature and humidity, and atmospheric pressure; the rules for the dynamic adjustment are based on an air gap discharge characteristic model or a pre-established experimental database.

4. The intelligent assessment and optimization method for safety clearances of power distribution lines according to claim 1, characterized in that, In the three-dimensional spatial environment model, at least one recommended operation plan that satisfies dynamic safety constraints is automatically planned, specifically including: Define a dynamic safety envelope for the working equipment or the limbs of the workers in the three-dimensional spatial environment model; With the constraint that the gap between the dynamic safety envelope and all hazard sources is always not less than the corresponding dynamic safety threshold, a collision-free path search is performed in the three-dimensional spatial environment model. If multiple feasible paths exist, the multiple feasible paths are sorted according to a preset optimization objective, which includes at least one or more of path length, operational complexity, or comprehensive safety margin.

5. The intelligent assessment and optimization method for safety clearances of distribution network lines according to claim 4, characterized in that, The collision-free path search employs a sampling-based planning algorithm, which includes a fast exploratory random tree or an improved version thereof.

6. The intelligent assessment and optimization method for safety clearances of distribution network lines according to claim 1, characterized in that, The method further includes step S5: By using augmented reality (AR) devices, key information from the recommended work plan is overlaid onto the actual field of vision of the workers; The key information includes: real-time highlighted hazardous areas of live conductors, overlaid recommended navigation paths, and real-time dynamic distance values ​​between the working equipment and live conductors.

7. The intelligent assessment and optimization method for safety clearances of distribution network lines according to claim 1, characterized in that, The method further includes the following steps before step S1: Based on historical scenario data or pre-collected data, construct or invoke a digital twin environment; In the digital twin environment, steps S1-S4 are executed to generate multiple pre-selected job schemes; By comparing and analyzing the multiple pre-selected operation plans, optimization decision suggestions are output to guide actual operations.

8. The intelligent assessment and optimization method for safety clearances of distribution network lines according to claim 7, characterized in that, The method further includes the following steps: Record multi-source sensing data, actual operation paths, and results during the actual operation process; The actual work path is compared and analyzed with the recommended work plan, and the results of the comparison and analysis are used to optimize the calculation model or algorithm involved in steps S2, S3 and / or S4.

9. A smart assessment and optimization system for safety clearances of distribution network lines, used to implement the smart assessment and optimization method for safety clearances of distribution network lines as described in any one of claims 1-8, characterized in that, include: A multi-source sensing unit is used to collect multi-source sensing data of the target work area in real time. The multi-source sensing data includes at least laser point cloud data and visible light image data. The data processing and fusion unit is used to fuse and process the multi-source sensing data, construct a three-dimensional spatial environment model containing semantic information of charged bodies, grounded bodies and working equipment, and extract the real-time geometric gaps between each object. The safety assessment and planning unit is used to perform dynamic safety risk assessment based on the pre-stored safety standard library and the real-time geometric gap. The safety thresholds in the safety standard library can be dynamically adjusted according to real-time environmental parameters. The interactive output unit is used to output the recommended work plan and risk information.

10. The intelligent assessment and optimization system for safety clearances of power distribution lines according to claim 9, characterized in that, The multi-source sensing unit is integrated into a drone, an insulated bucket truck, or a wearable device; the interactive output unit is augmented reality (AR) glasses or an explosion-proof mobile terminal.

Images