Distribution network equipment health assessment method and system based on point cloud model, equipment and medium

Abstract

The invention relates to the technical field of distribution network monitoring and evaluation, in particular to a distribution network equipment health evaluation method, system, equipment and medium based on a point cloud model, and the method comprises the steps: collecting data of a distribution network and a surrounding environment, constructing a four-dimensional space-time model, and calculating an activity envelope according to an actual process model of a dynamic object based on the constructed four-dimensional space-time model. Generating a static safety buffer area distributed along the power line, converting a space-time range jointly defined by the static safety area and a preset time window into a discrete four-dimensional grid based on the active envelope body, calculating the occupancy probability of a dynamic object based on the discrete four-dimensional grid, and calculating the occupancy probability of the dynamic object based on the discrete four-dimensional grid. A prospective risk value generated by a dynamic object for a power line is quantified through a first-level algorithm, the prospective risk value and a four-dimensional space-time model are subjected to data association and visualized presentation, a complex risk assessment result is converted into visual decision support information, and the decision support information is displayed in a real-time manner. Therefore, the defects of the existing method in the aspects of data dimension, risk quantification capability and information presentation mode are overcome.

Description

Technical Field

[0001] This invention relates to the field of power distribution network monitoring and evaluation technology, and in particular to a method, system, equipment and medium for evaluating the health of power distribution network equipment based on a point cloud model. Background Technology

[0002] The operation and maintenance and risk management of modern power distribution networks are shifting from the traditional post-fault repair and periodic maintenance model to a more advanced condition-based maintenance and proactive early warning model. The key to this transformation lies in the accurate assessment of the health status of power distribution equipment and the use of this assessment results to guide operation and maintenance decisions. However, the current technical means used to support this advanced concept have some inherent defects in practical applications, corresponding to the technical problems that this invention aims to solve: Existing health assessments have certain limitations in terms of data dimensions. The commonly used two-dimensional geographic information systems, namely GIS or static 3D models, are essentially snapshots of the environment at a certain moment. Such static data structures cannot record and analyze the dynamic evolution of the environment, such as the erosion of safe distances caused by seasonal vegetation growth. When risks accumulate or evolve over time, existing technologies struggle to conduct effective trend analysis, resulting in health assessment models lacking consideration of environmental evolution factors.

[0003] The accuracy and quantification capabilities of risk alerts are insufficient. Traditional risk alerts are mostly qualitative or based on a single threshold, such as simple boundary crossing alerts. This binary alert mechanism cannot accurately distinguish the severity of different risk events. When conducting health assessments, this means that external risks cannot be input into the model as a quantifiable parameter, which greatly affects the accuracy of the assessment results.

[0004] The way information is presented is difficult to support intuitive decision-making. The results of risk analysis are usually presented in the form of independent text messages or data reports. This fragmented information makes it impossible for operations and maintenance personnel to quickly and comprehensively understand the overall situation of the risk, which affects the efficiency of making rapid and accurate decisions based on the health status. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, this invention provides a method, system, device, and medium for health assessment of distribution network equipment based on point cloud models to address the limitations of existing health assessments in terms of data dimensions, insufficient accuracy and quantification of risk alarms, and difficulty in providing information presentation methods to support intuitive decision-making.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for health assessment of distribution network equipment based on a point cloud model, comprising: Data on the power distribution network and its surrounding environment are collected to construct a four-dimensional spatiotemporal model. Based on the constructed four-dimensional spatiotemporal model, the active envelope is calculated according to the actual process model of the dynamic object, and a static safety buffer zone distributed along the power line is generated. Based on the active envelope, the spatiotemporal range defined by the static safety zone and the preset time window is transformed into a discrete four-dimensional grid. The occupancy probability of dynamic objects is calculated based on the discrete four-dimensional grid, and the forward-looking risk value of dynamic objects to power lines is quantified through a first-level algorithm. The forward-looking risk value and the four-dimensional spatiotemporal model are correlated and visualized.

[0008] As a preferred embodiment of the point cloud model-based health assessment method for distribution network equipment described in this invention, the method includes: collecting data on the distribution network and its surrounding environment, and constructing a four-dimensional spatiotemporal model, including: Collect data on the power distribution network and its surrounding environment, obtain a baseline 3D point cloud model, and construct a spatial partitioning structure; Based on the spatial partitioning structure, the transformed cells of the spatial partitioning structure are identified through differential point cloud data, and the data of the transformed cells are updated. The cells with updated data are assigned new timestamps, while cells that have not changed remain unchanged, thus completing the construction of the four-dimensional spatiotemporal model.

[0009] As a preferred embodiment of the point cloud model-based health assessment method for distribution network equipment described in this invention, the method includes: based on the constructed four-dimensional spatiotemporal model, calculating the active envelope according to the actual process model of the dynamic object, and generating a static safety buffer region distributed along the power line, including: Based on the constructed four-dimensional spatiotemporal model, the kinematic parameters of the dynamic object are entered, and geometric operations are performed according to the kinematic parameters of the dynamic object. A set of spatial points is generated through geometric operations, and an active envelope is constructed based on the generated set of spatial points. Based on the active envelope, a static safety buffer zone is generated distributed along the power line.

[0010] As a preferred embodiment of the point cloud model-based health assessment method for distribution network equipment described in this invention, the method includes: based on the active envelope, transforming the spatiotemporal range jointly defined by the static safety region and the preset time window into a discrete four-dimensional grid; calculating the occupancy probability of dynamic objects based on the discrete four-dimensional grid; and quantifying the forward-looking risk value of dynamic objects on power lines using a first-level algorithm, including: Based on the active envelope, the spatiotemporal range jointly defined by the static safety region and the preset time window is transformed into a discrete four-dimensional grid; Based on a discrete four-dimensional grid, the method for calculating the occupancy probability is selected according to the dynamic object, and the occupancy probability is calculated based on the selected calculation method. Based on the calculated occupancy probability, a first-level algorithm is used to quantify the forward-looking risk value of dynamic objects to power lines.

[0011] As a preferred embodiment of the point cloud model-based health assessment method for distribution network equipment described in this invention, the method includes: collecting data on the distribution network and its surrounding environment, obtaining a baseline three-dimensional point cloud model, and constructing a spatial partitioning structure, including: A drone equipped with a lidar is used to inspect the target area and obtain a baseline 3D point cloud model. The space is divided into an octree structure in a three-dimensional grid; Each internal node in the octree structure has eight child nodes, and the eight child nodes represent the parent node. The three-dimensional space is bisected along the X, Y, and Z axes to form eight octaves. Construct the smallest cube bounding box that surrounds the target point cloud as the root node, and recursively divide the nodes until the preset maximum tree depth is reached and the number of points contained in the leaf nodes is less than a threshold.

[0012] As a preferred embodiment of the point cloud model-based health assessment method for distribution network equipment described in this invention, the method involves: transforming the spatiotemporal range defined by the static security region and a preset time window into a discrete four-dimensional grid based on the active envelope, including: The continuous spatiotemporal range defined by the static safety buffer area (PSB) and the preset time window (T) is transformed into a discrete four-dimensional grid. In three-dimensional space, a three-dimensional voxel grid is constructed with the PSB as the core. In the time dimension, time windows are... Divided into Each time step is of equal length; Any spacetime unit in the grid Defined by a spatial voxel and a time interval; The range covered by spatial voxels is The time interval covers the range of ; in, , , These are the starting coordinates on the X, Y, and Z axes, respectively. These represent the resolution of the voxel mesh in the three spatial directions, respectively. For the starting moment in the time dimension, This represents the temporal resolution of the grid.

[0013] The beneficial effects of this preferred technical solution are as follows: by discretizing a continuous spatiotemporal range into a structured four-dimensional grid, a unified quantitative calculation framework for risk assessment is provided. The spatial and temporal uncertainties of dynamic objects are transformed into the problem of calculating the occupancy probability of discrete spatiotemporal units.

[0014] As a preferred embodiment of the point cloud model-based distribution network equipment health assessment method described in this invention, the method involves: quantifying the forward-looking risk value of dynamic objects on power lines using a first-level algorithm based on the calculated occupancy probability, including: Forward-looking risk value The expression is: in, This is a forward-looking risk value. For the entire future time window Discrete-time integrals over time, For summation, For spacetime units At time step The probability of being occupied by a dynamic object. spatial unit The key weighting factors, For the grid index and time index A four-dimensional spacetime unit; definition The method is inverse distance weighting, and the key weight factor expression is: in, As a key weighting factor, The minimum distance from the center of the cell to the nearest critical device. Scaling factor It is a three-dimensional mesh unit. A spatial linear entity for critical facilities; Categorization assignment assigns discrete weight values ​​based on the type of infrastructure adjacent to the cell.

[0015] The beneficial effects of this preferred technical solution are as follows: by using a risk quantification model, the potential threat of a dynamic object in the future time and space is transformed into a continuous and calculable risk indicator. By introducing key weighting factors, spatial distance and equipment criticality are incorporated into the risk calculation, so that the assessment results can reflect the differences in risk levels at different locations.

[0016] Secondly, the present invention provides a health assessment system for distribution network equipment based on a point cloud model, comprising: The four-dimensional spatiotemporal model construction module collects data on the power distribution network and its surrounding environment, constructs a four-dimensional spatiotemporal model, and calculates the active envelope based on the actual process model of the dynamic object, generating a static safety buffer zone distributed along the power line. The Quantification Forward-Looking Risk Value Module, based on the activity envelope, transforms the spatiotemporal range jointly defined by the static safety zone and the preset time window into a discrete four-dimensional grid. Based on the discrete four-dimensional grid, it calculates the occupancy probability of dynamic objects and quantifies the forward-looking risk value of dynamic objects on power lines through a first-level algorithm. The visualization module links forward-looking risk values ​​with a four-dimensional spatiotemporal model and presents the data visually.

[0017] Thirdly, the present invention provides an electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of a method for health assessment of distribution network equipment based on a point cloud model.

[0018] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the point cloud model-based distribution network equipment health assessment method.

[0019] Compared with existing technologies, the beneficial effects of this invention are as follows: By constructing a four-dimensional spatiotemporal model and an activity envelope, this invention accurately describes the spatiotemporal behavior of dynamic objects, expanding the data dimensions of health assessment; by using discrete four-dimensional grids and occupancy probability calculation, it establishes a forward-looking quantitative assessment method for risks, improving the accuracy and quantifiability of risk warnings; and finally, through data association and visualization, it transforms complex risk assessment results into intuitive decision support information, thereby overcoming the shortcomings of non-intuitive information presentation in existing technologies. Attached Figure Description

[0020] 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.

[0021] Figure 1 This is a schematic diagram of the overall process of a point cloud model-based health assessment method for distribution network equipment according to an embodiment of the present invention. Detailed Implementation

[0022] 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.

[0023] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for health assessment of distribution network equipment based on a point cloud model is provided, comprising: To address the limitations of existing health assessments in terms of data dimensions, insufficient accuracy and quantification of risk alerts, and difficulty in providing intuitive decision-making support through information presentation, this invention provides a health assessment method for distribution network equipment based on a point cloud model.

[0024] S1: Collect data on the power distribution network and its surrounding environment, construct a four-dimensional spatiotemporal model, and calculate the active envelope based on the actual process model of the dynamic object, thereby generating a static safety buffer zone distributed along the power line. S2: Based on the active envelope, the spatiotemporal range defined by the static safety zone and the preset time window is transformed into a discrete four-dimensional grid. The occupancy probability of dynamic objects is calculated based on the discrete four-dimensional grid, and the forward-looking risk value of dynamic objects to power lines is quantified through a first-level algorithm. S3: Link forward-looking risk values ​​with four-dimensional spatiotemporal models and present them visually.

[0025] Therefore, by constructing a point cloud model that integrates spatiotemporal dimensions and a dynamic object activity envelope, a panoramic digital representation of the operating environment of distribution network equipment is established; then, through discrete four-dimensional grids and risk quantification algorithms, a forward-looking assessment of dynamic threats is achieved; finally, through the deep integration and visualization of risk data and three-dimensional scenes, complex risk assessment results are transformed into intuitive decision support information, thereby solving the shortcomings of existing methods in terms of data dimension, risk quantification capability, and information presentation method.

[0026] Example 2, refer to Figure 1 As an embodiment of the present invention, based on the above embodiment, a method for health assessment of distribution network equipment based on point cloud model is provided.

[0027] In this embodiment of the application, step S1 involves collecting data on the power distribution network and its surrounding environment, constructing a four-dimensional spatiotemporal model, and, based on the constructed four-dimensional spatiotemporal model, calculating the active envelope according to the actual process model of the dynamic object, thereby generating a static safety buffer zone distributed along the power line. This includes the following steps A1-A6: A1: Collect data on the power distribution network and its surrounding environment, obtain a baseline 3D point cloud model, and construct a spatial partitioning structure.

[0028] A2: Based on the spatial partitioning structure, the transformed cells of the spatial partitioning structure are identified through differential point cloud data, and the data of the transformed cells are updated.

[0029] The system acquires differential point clouds. By comparing the geometric features or point density of the old and new point clouds within the same Octree cell, the system can quickly identify the changed cells. Instead of replacing the entire model, it updates the data in the changed cells, including the point cloud data and the initial timestamp. Stored in the corresponding leaf node cell In the process, the construction of the benchmark model was completed.

[0030] A3: Assign a new timestamp to the cells with updated data, while keeping the cells that have not changed unchanged, thus completing the construction of the four-dimensional spatiotemporal model.

[0031] A4: Based on the constructed four-dimensional spatiotemporal model, input the kinematic parameters of the dynamic object, and perform geometric calculations based on the kinematic parameters of the dynamic object.

[0032] When a new dynamic object is identified in the system, the maintenance personnel or image recognition algorithm will input the key kinematic parameters of the object. Based on the registered kinematic parameters, the system performs geometric calculations in three-dimensional space. Taking the fixed reference point registered in the system as the origin of the three-dimensional space coordinate system and the effective range L of the moving part as the movement radius, combined with the preset maximum active angle range and maximum travel height limit, a continuous spatial region is constructed in three-dimensional space through geometric calculations. This region represents the set of spatial positions that the moving part can reach under all kinematic constraints, and finally forms a three-dimensional geometric sweep body uniquely defined by the kinematic attributes.

[0033] A5: Generate a set of spatial points through geometric operations, and construct an active envelope based on the generated set of spatial points.

[0034] A6: Generate a static safety buffer zone distributed along the power line based on the active envelope.

[0035] A safety buffer zone with a preset safety radius is generated from the digital model along the power line.

[0036] In this embodiment of the application, the spatial partitioning structure constructed in step A1 specifically adopts an octree structure: A drone equipped with a lidar is used to inspect the target area and obtain a baseline 3D point cloud model. The space is divided into an octree structure in a three-dimensional grid; Each internal node in the octree structure has eight child nodes, and the eight child nodes represent the parent node. The three-dimensional space is bisected along the X, Y, and Z axes to form eight octaves. Construct the smallest cube bounding box that surrounds the target point cloud as the root node, and recursively divide the nodes until the preset maximum tree depth is reached and the number of points contained in the leaf nodes is less than a threshold.

[0037] In an alternative implementation, the spatial partitioning structure constructed in step A1 can also adopt a hybrid index structure. For linear corridor areas that need to support fast spatial range queries, a special R-tree index is built on the basis of the octree for the power line corridor, and the spatial units inside and outside the line buffer are managed separately, thereby improving the efficiency of spatial queries for line safety areas.

[0038] In another alternative implementation, the spatial partitioning structure constructed in step A1 can also adopt a time-aware octree structure. In the inspection area that needs to be dynamically updated frequently, a timestamp index is added to each node based on the traditional spatial partitioning.

[0039] In this application's implementation, the construction of the four-dimensional spatiotemporal model in step S1 is specifically manifested as follows: Update the data in the changed cells and assign them a new timestamp. Meanwhile, the data in cells that have not changed remains unchanged. Through this mechanism of differential update and incremental fusion, a four-dimensional (3D spatial + temporal) data model that can record the historical evolution of the environment can be efficiently constructed and maintained.

[0040] In an alternative implementation, the construction of the four-dimensional spatiotemporal model in step S1 can also be achieved by generating a radiation field reference model through three-dimensional Gaussian sputtering technology, and by comparing the real-time image with the expected image rendered by the reference model, environmental changes can be detected and located.

[0041] In another optional implementation, the four-dimensional spatiotemporal model built in step S1 can also adopt a four-dimensional model based on spatiotemporal hashing. Spatial units are mapped to a hash table through a hash function. Each unit contains point cloud data and a timestamp. During differential updates, the changed units can be located quickly directly through hashing, reducing traversal overhead.

[0042] In this embodiment of the application, step S2, based on the active envelope, transforms the spatiotemporal range jointly defined by the static safety region and the preset time window into a discrete four-dimensional grid. The occupancy probability of dynamic objects is calculated based on the discrete four-dimensional grid, and a first-level algorithm is used to quantify the forward-looking risk value of dynamic objects on power lines. This includes the following steps B1-B3: B1: Based on the active envelope, the spatiotemporal range jointly defined by the static security region and the preset time window is transformed into a discrete four-dimensional grid.

[0043] In three-dimensional space, a three-dimensional voxel grid is constructed with the PSB as the core. In the time dimension, time windows are... Divided into Each time step is of equal length; Any spacetime unit in the grid Defined by a spatial voxel and a time interval; The range covered by spatial voxels is The time interval covers the range of ; in, , , These are the starting coordinates on the X, Y, and Z axes, respectively. These represent the resolution of the voxel mesh in the three spatial directions, respectively. For the starting moment in the time dimension, This represents the temporal resolution of the grid.

[0044] B2: Based on a discrete four-dimensional grid, select a method for calculating the occupancy probability according to the dynamic object, and calculate the occupancy probability based on the selected method.

[0045] For each spatiotemporal unit in the grid Calculate the probability that it is occupied by a dynamic object. The choice of calculation method depends on the nature of the dynamic object (DO) prediction model; When the future behavior of a dynamic object follows a deterministic or semi-deterministic work plan, the dynamic object at any future time... Location It is known that the occupancy probability is essentially binary.

[0046] A spatial unit The condition for being occupied is volume and... Centered dynamic object volume When intersection occurs, to account for small execution errors in reality, the geometric model of the dynamic object can be convolved with a small Gaussian kernel representing positional uncertainty. After simplification, the probability can be expressed as: in, For spacetime units At time step The probability of being occupied by a dynamic object. For unit space voxels, For the geometric volume of a dynamic object.

[0047] When the behavior of a dynamic object is random and must be modeled by learning from historical data, that is, when a deep learning model is used to predict the probability distribution of future trajectories; The model outputs a probabilistic description of future states, which can be directly represented as a future grid map or a Gaussian mixture model. The probability that any spacetime unit is occupied is equal to the probability that the dynamic object is at the corresponding moment. Predicted location probability density function In this unit space voxel The integral over is expressed as: in, For unit space voxels, For dynamic objects at any time The probability density function of the predicted location, For spacetime units At time step The probability of being occupied by a dynamic object.

[0048] B3: Based on the calculated occupancy probability, the forward-looking risk value of dynamic objects to power lines is quantified through a first-level algorithm.

[0049] In this embodiment of the application, step B3, which quantifies the forward-looking risk value of the dynamic object on the power line using a first-level algorithm, is specifically manifested as follows: Forward-looking risk value The expression is: in, This is a forward-looking risk value. For the entire future time window Discrete-time integrals over time, For summation, For spacetime units At time step The probability of being occupied by a dynamic object. spatial unit The key weighting factors, For the grid index and time index A four-dimensional spacetime unit; definition The method is inverse distance weighting, and the key weight factor expression is: in, As a key weighting factor, The minimum distance from the center of the cell to the nearest critical device. This is the scaling factor, ranging from 1 to 3, with a commonly used value of 2. The value is determined by... It is a three-dimensional mesh unit. A spatial linear entity for critical facilities; Categorization assignment assigns discrete weight values ​​based on the type of infrastructure adjacent to the cell.

[0050] In an alternative implementation, step B3, which quantifies the forward-looking risk value of dynamic objects on power lines using a first-level algorithm, can also employ occupancy probability calculation based on multi-resolution grids. When the activity range of dynamic objects varies greatly and some areas require higher-precision modeling, high-resolution grids are used for key areas near power lines, while low-resolution grids are used for non-critical areas on the periphery. Occupancy probability calculation can capture the behavior of dynamic objects more accurately in high-resolution grids and be coarsely processed in low-resolution grids.

[0051] In another alternative implementation, the forward-looking risk value of the dynamic object to the power line generated by the first-level algorithm in step B3 can also be quantified by probability prediction based on Monte Carlo sampling. When the behavior of the dynamic object is complex and it is difficult to obtain an analytical probability density function, the movement of the dynamic object is simulated by a large number of trajectory samples within a given time window, and the frequency of each spatiotemporal unit being occupied is counted as the occupancy probability.

[0052] It should be noted that by using a discrete four-dimensional grid to structurally represent the spatiotemporal range and by establishing a differentiated probability calculation model for deterministic and random dynamic objects, a precise quantitative description of potential risks is achieved. Based on this, by introducing key weighting factors and incorporating spatial distance and equipment criticality into the evaluation system, a risk assessment model that integrates spatiotemporal probability and equipment criticality is constructed, forming a systematic and forward-looking risk quantification method.

[0053] In this embodiment of the application, step S3 involves data association between the forward-looking risk value and the four-dimensional spatiotemporal model, followed by visualization, including: The system will calculate The values ​​are correlated with the four-dimensional spacetime model.

[0054] In this embodiment of the application, the visualization in step S3 is specifically manifested as follows: In a four-dimensional visualization and interactive module, operations and maintenance personnel can see a realistic three-dimensional scene. When a value exceeds the warning threshold, the system will visually mark the objects in the model. For example, dangerous road sections will be rendered with different colors (such as yellow, orange, and red) according to their risk level, and will represent spatial areas where risks intersect. It will be highlighted.

[0055] It provides interactive timeline query and backtracking functions. Users can not only query and backtrack to any historical moment's 3D scene snapshot of the power grid by dragging the timeline control, but also view the risk warning status at different future time points, realizing dynamic review and analysis of environmental changes and risk evolution processes.

[0056] In an alternative implementation, the visualization in step S3 can also employ transparent volume rendering. When the risk not only lies along the route but may also involve spatial volume intersections, the four-dimensional risk value is mapped to the transparency or color intensity in the three-dimensional volume, and the time dimension is displayed through animation or a time slider.

[0057] In another optional implementation, the visualization in step S3 can also use dynamic annotations and warning icons. When maintenance personnel need to pay close attention to certain key equipment or dangerous line segments, dynamic icons are added to high-risk units or line segments in the three-dimensional scene, and dynamic annotations of risks changing over time are achieved in conjunction with the time axis.

[0058] In summary, this invention utilizes four-dimensional spatiotemporal modeling and point cloud data fusion to achieve dynamic visualization and risk quantification assessment of the operating status of power distribution network equipment. It effectively solves problems such as lagging health assessment and low risk identification accuracy in traditional inspection methods. By constructing a four-dimensional spatiotemporal model based on UAV LiDAR data, and combining the kinematic process of dynamic objects to generate active envelopes and static safety buffers, it maps these into discrete four-dimensional grids, calculates the occupancy probability of dynamic objects in spatiotemporal units, and quantifies risk values ​​by combining spatial criticality weights, thereby achieving forward-looking prediction and assessment of potential risks to power lines.

[0059] Example 3 discloses a method for health assessment of distribution network equipment based on a point cloud model. The specific implementation steps of this method in a typical urban power grid operation and maintenance scenario are as follows: Step 1: Construction of a four-dimensional spacetime model Data on the power distribution network and its surrounding environment were collected to construct a four-dimensional spatiotemporal model including three-dimensional spatial coordinates and timestamps. Specifically, the model was initialized. A drone equipped with a LiDAR was used to conduct an initial high-precision, full-coverage inspection of the target power distribution network area to obtain a baseline three-dimensional point cloud model. The system divides the space into a three-dimensional grid (preferably an octree structure) and integrates the point cloud data with the initial timestamp. Stored in the corresponding leaf node cell In the process, the construction of the benchmark model was completed.

[0060] The collected data includes a baseline 3D point cloud model obtained using a drone equipped with a lidar system. Differential point cloud data obtained during high-frequency inspections , and multi-view images for three-dimensional Gaussian sputtering schemes.

[0061] The first high-precision, full-coverage inspection of the target area was carried out using a drone equipped with a lidar, and a baseline 3D point cloud model was obtained. The space is divided into a three-dimensional mesh, and the point cloud data and initial timestamps are... Stored in the corresponding leaf node cell In subsequent high-frequency inspections, differential point clouds are acquired. By comparing the geometric features or point density of the old and new point clouds within the same octree cell, changed cells are identified. The system only updates the data in these changed cells and assigns them a new timestamp. Cells whose data has not changed will remain unchanged.

[0062] An octree is a tree-like data structure used to manage sparse 3D data. Its core idea is to recursively and hierarchically partition the 3D space into binary components.

[0063] Each internal node has eight child nodes. The eight child nodes represent the eight octaves formed by bisecting the three-dimensional space volume (called a voxel) represented by the parent node along the X, Y, and Z axes.

[0064] Construction begins with the smallest cube bounding box that can completely enclose the entire target point cloud; this is the root node. Nodes are recursively divided until a preset maximum tree depth is reached, or the number of points contained in a leaf node falls below a certain threshold. The resolution of the leaf nodes is a key parameter determining the depth of the tree and the model's finesse.

[0065] Unlike dense voxel grids, octree structures do not require the allocation of child nodes for storage in empty areas of space, and have extremely high memory efficiency when representing point cloud data that is essentially sparse.

[0066] The model is dynamically and incrementally updated. In subsequent high-frequency inspections, the system acquires the differential point cloud. By comparing the geometric features or point density of the old and new point clouds within the same Octree cell, the system can quickly identify changed cells. Instead of replacing the entire model, it only updates the data in these changed cells and assigns them new timestamps. Meanwhile, the data in cells that have not changed remains unchanged. Through the mechanism of differential update and incremental fusion, a four-dimensional (3D spatial + temporal) data model that records the historical evolution of the environment can be efficiently constructed and maintained.

[0067] As a parallel approach, this can also be achieved using a multi-source vision and 3D Gaussian sputtering (3DGS) model. This involves first training a high-fidelity radiation field baseline model using multi-view images. Subsequently, by comparing real-time images with the expected images rendered from the baseline model at the pixel level, it is possible to quickly identify and locate changing hotspots in the environment, triggering alarms or more precise measurements.

[0068] Step 2: Calculation of the Future Activity Envelope When a new dynamic object is identified in the system (taking a tower crane as an example), maintenance personnel or image recognition algorithms will input the object's key kinematic parameters; these key parameters include: Precise coordinates of the base ( ): The fixed base point position of the tower crane in three-dimensional space.

[0069] arm length ( ): The length of the tower crane boom.

[0070] Work plan constraints: including the maximum allowable slewing angle range and lifting height.

[0071] The system performs geometric calculations in three-dimensional space based on registered kinematic parameters. The calculation process is as follows: taking the tower crane's base... As the origin or reference point, combined with the arm length Using the radius as a guide and limited by its maximum slewing angle and lifting height, a geometry is swept out in three-dimensional space. This geometry represents the set of all spatial points that the tower crane boom can reach within the limits allowed by the work plan.

[0072] The set of points ultimately generated by geometric operations constitutes the three-dimensional Dynamic Safety Envelope (DSE), denoted in the document as... . This provides a clear geometric object for subsequent risk quantification steps.

[0073] A simplified mobile envelope of a tower crane can be modeled as a sector-shaped cylinder or a portion thereof, and the generation process can be defined by motion parameters: Base coordinates: Arm length (radius): Height range: Rotation angle range: in, For base coordinates, , , These are the X, Y, and Z coordinates of the base location, respectively. , These are the minimum and maximum heights of the tower crane's operating range, respectively. , These are the minimum and maximum slewing angles of the tower crane boom, respectively.

[0074] The three-dimensional envelope can be defined using motion parameters. The boundary.

[0075] Based on the aforementioned four-dimensional spatiotemporal model, a static safety buffer zone distributed along the power line is generated. And based on the kinematic model or work plan of at least one dynamic object, the four-dimensional motion trajectory or activity envelope of the dynamic object within a preset time window is calculated. Specifically, when a new dynamic object (such as a tower crane) is detected in the environment in step one, the system enters the dynamic object modeling and registration process. Maintenance personnel or image recognition algorithms will input the tower crane's key kinematic parameters, including its precise base coordinates. arm length And the maximum range of slewing angles and lifting heights allowed in its work plan.

[0076] Based on the registered kinematic parameters, the system performs geometric calculations in three-dimensional space to calculate the set of all physically accessible points of the tower crane, thereby generating a three-dimensional dynamic safety envelope (DSE), denoted as . . This is a specific instance of the activity envelope, providing a clear geometric object for subsequent risk quantification.

[0077] Geometric calculations are performed based on the kinematic key parameters of a registered dynamic object (such as a tower crane). These key parameters include: the precise coordinates of the base (…). ), arm length ( The system uses these kinematic parameters, along with the maximum allowable slewing angle range and lifting height specified in the work plan, to calculate and determine the set of all physically accessible points of the tower crane in three-dimensional space. This three-dimensional set of all reachable points constitutes the dynamic safety envelope (DSE), denoted as […]. The envelope provides a clear geometric object for subsequent risk quantification.

[0078] Step 3: Quantification of forward-looking risk values The forward-looking risk value of the dynamic object to the power line is quantified by calculating the spatiotemporal overlap, minimum distance, or collision probability between the static safety buffer area and the four-dimensional motion trajectory or active envelope within the preset time window.

[0079] The system first generates a power line safety buffer zone with a preset safety radius (e.g., 5 meters) along the digital model of the power line, denoted as . .

[0080] Risk quantification is generated through calculation step two. and This is achieved through the spatial intersection. The intersection represents the portion of the tower crane's potential operating area that encroaches upon the safety zone of the power line, and its volume is denoted as... .

[0081] To obtain a measurable risk value, a more advanced model is used for calculation: in, It is a quantified instantaneous collision risk index (i.e., a forward-looking risk value). It is the volume of the intersection mentioned above. It is the probability of operating within the intersection area, obtained based on information such as construction plans. and These are the corresponding weighting coefficients.

[0082] Step 4: Visual Presentation The forward-looking risk value is correlated with the objects in the four-dimensional spatiotemporal model and then visualized.

[0083] The system will calculate the result in step three. The values ​​are correlated with the tower cranes and corresponding line segments in the four-dimensional spatiotemporal model.

[0084] In a four-dimensional visualization and interactive module, operations and maintenance personnel can see a realistic three-dimensional scene. When a value exceeds the warning threshold, the system will visually mark the objects in the model. For example, dangerous road sections will be rendered with different colors (such as yellow, orange, and red) according to their risk level, and will represent spatial areas where risks intersect. It will be highlighted.

[0085] In addition, the module provides timeline query and backtracking interactive functions. By dragging the timeline control, users can not only query and backtrack to any historical moment's 3D scene snapshot of the power grid, but also view the risk warning status at different future points in time, realizing dynamic review and analysis of environmental changes and risk evolution.

[0086] To further illustrate the advanced nature of the technical solution of the present invention, more in-depth implementation methods will be provided for some steps in the above embodiments below.

[0087] In some scenarios where higher accuracy of risk assessment is required, the third step, "quantification of forward-looking risk values," can employ the following more refined model.

[0088] This approach aims to transform vague potential risks into specific probabilistic risks.

[0089] The spatiotemporal range formed by the static safety buffer zone and the preset time window is discretized into a four-dimensional probabilistic occupancy grid. Specifically, the system discretizes the entire operating space, especially the area covered by the power line safety buffer zone, into a three-dimensional voxel grid G ​​(e.g., cell size 1m*1m*1m). Simultaneously, the preset future time window T (e.g., the next 24 hours) is discretized into N time steps Δt (e.g., one step every 10 minutes). This forms a four-dimensional (x,y,z,t) probabilistic occupancy grid.

[0090] For each spatiotemporal unit in the grid, based on the job plan or historical operation mode of the dynamic object, the probability value of its occupation at the corresponding time step is calculated. Specifically, for any unit in the grid... and future time steps The system will calculate the probability that it is occupied by a dynamic object (such as a tower crane). The probability value is not simply 0 or 1, but a continuous value between [0,1] calculated based on the imported construction scheduling plan and the equipment's historical operating patterns. For example, cells on the lifting path explicitly specified in the construction plan are assigned a higher probability value (e.g., 0.8), while areas that are only theoretically accessible but not covered in the plan have a lower probability value (e.g., 0.05).

[0091] The probability values ​​of all spatiotemporal units located within the static safety buffer zone are weighted and summed or integrated within the preset time window to obtain the forward-looking risk value. Specifically, the final forward-looking risk value is no longer a simple volume intersection, but rather the risk integral of all grid units within the power line safety buffer zone (PSB) within a future time window T. : in, It is a cell The importance weights are assigned, for example, cells closer to critical conductors or insulators have higher weights. In this way, the system can more precisely depict the distribution and evolution of risks in space and time.

[0092] Example 4 discloses a method for health assessment of distribution network equipment based on point cloud model. This method is mainly applied to complex scenarios where multiple dynamic objects work together.

[0093] The system receives and parses motion commands and their dynamic priorities from multiple collaborative operating devices. A standard data interface is defined for receiving operation commands issued by the central dispatch system within a short future time window. The commands include the motion task of each tower crane and its dynamic priority within that task.

[0094] Based on priority from highest to lowest, the motion trajectory of high-priority devices within the preset time window generates a four-dimensional spatiotemporal forbidden volume, which the system analyzes sequentially according to priority from highest to lowest. For the tower crane with the highest current priority, the four-dimensional motion trajectory within the time window is calculated based on the motion command, and expanded by a safety radius to generate a spatiotemporal trajectory entity that is defined as the forbidden volume for all other low-priority tower cranes within the time window.

[0095] For any low-priority device, the taboo volumes generated by all higher-priority devices are recursively removed from the original active envelope by geometric difference operation to obtain the final active envelope under cooperative constraints. For the tower crane of the next priority, the system first generates an unconstrained original active envelope, and then performs geometric difference operation to remove the taboo volumes from the original active envelope to obtain the real reachable active space after following the cooperative avoidance rules, i.e., the final active envelope.

[0096] The forward-looking risk value is calculated based on the intersection of the final active envelope and the static safety buffer zone. Specifically, the system includes the crane's trajectory entity in the taboo volume and continues to process lower-priority cranes. Finally, the intersection of the trimmed final active envelope of all cranes with the power line safety zone is calculated to obtain a true risk assessment result that excludes interference from the cooperative avoidance logic.

[0097] Furthermore, the steps for constructing a four-dimensional spatiotemporal model specifically include: firstly acquiring a baseline three-dimensional point cloud model and constructing an octree structure; subsequently, identifying changing cells in the octree structure using high-frequency acquired differential point cloud data; incrementally updating only the data of the changing cells and assigning them new timestamps.

[0098] Furthermore, the step of constructing the four-dimensional spatiotemporal model also includes: using multi-view images to generate a radiation field reference model through three-dimensional Gaussian sputtering technology; and detecting and locating environmental changes by comparing real-time images with the expected images rendered by the reference model.

[0099] Furthermore, the visualization presentation steps also include: providing an interactive timeline control for querying and tracing back the scene snapshots and risk states of the four-dimensional spatiotemporal model at any historical moment. Regarding a more refined implementation of the visualization presentation steps, to enhance users' understanding of the risk evolution process, step four, visualization presentation, may include the following interactive functions. At the bottom of the interface of the four-dimensional visualization and interaction module, a draggable timeline control is provided. By operating the control, users can not only query and trace back the three-dimensional scene snapshots of the power grid at any historical moment, such as reviewing how the tower cranes were gradually erected over the past week, but more importantly, users can drag the timeline to future times. At this time, the three-dimensional scene will dynamically display the risk warning states calculated by the system based on the aforementioned quantitative model at different future moments. This interactive method enables dynamic review and forward-looking analysis of environmental changes and risk evolution processes, providing intuitive and traceable data support for operation and maintenance decisions.

[0100] Example 5 illustrates a schematic scheme for a point cloud model-based distribution network equipment health assessment method. It should be noted that the technical solution of this point cloud model-based distribution network equipment health assessment system belongs to the same concept as the technical solution of the point cloud model-based distribution network equipment health assessment method described above. Details not described in detail in the point cloud model-based distribution network equipment health assessment system in this embodiment can be found in the description of the point cloud model-based distribution network equipment health assessment method described above.

[0101] This embodiment also provides a health assessment system for distribution network equipment based on a point cloud model, including: The four-dimensional spatiotemporal model construction module collects data on the power distribution network and its surrounding environment, constructs a four-dimensional spatiotemporal model, and calculates the active envelope based on the actual process model of the dynamic object, generating a static safety buffer zone distributed along the power line. The Quantification Forward-Looking Risk Value Module, based on the activity envelope, transforms the spatiotemporal range jointly defined by the static safety zone and the preset time window into a discrete four-dimensional grid. Based on the discrete four-dimensional grid, it calculates the occupancy probability of dynamic objects and quantifies the forward-looking risk value of dynamic objects on power lines through a first-level algorithm. The visualization module links forward-looking risk values ​​with a four-dimensional spatiotemporal model and presents the data visually.

[0102] This embodiment also provides an electronic device suitable for health assessment of distribution network equipment based on point cloud models, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement a method for health assessment of distribution network equipment based on point cloud models as proposed in the above embodiment.

[0103] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a point cloud model-based health assessment method for distribution network equipment as proposed in the above embodiments.

[0104] The storage medium proposed in this embodiment belongs to the same inventive concept as the method for implementing a point cloud model-based health assessment of distribution network equipment proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0105] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0106] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for health assessment of distribution network equipment based on a point cloud model, characterized in that, include: Data on the power distribution network and its surrounding environment are collected to construct a four-dimensional spatiotemporal model. Based on the constructed four-dimensional spatiotemporal model, the active envelope is calculated according to the actual process model of the dynamic object, and a static safety buffer zone distributed along the power line is generated. Based on the active envelope, the spatiotemporal range defined by the static safety zone and the preset time window is transformed into a discrete four-dimensional grid. The occupancy probability of dynamic objects is calculated based on the discrete four-dimensional grid, and the forward-looking risk value of dynamic objects to power lines is quantified through a first-level algorithm. The forward-looking risk value and the four-dimensional spatiotemporal model are correlated and visualized.

2. The method for health assessment of distribution network equipment based on a point cloud model as described in claim 1, characterized in that, The data collected from the power distribution network and its surrounding environment are used to construct a four-dimensional spatiotemporal model, including: Collect data on the power distribution network and its surrounding environment, obtain a baseline 3D point cloud model, and construct a spatial partitioning structure; Based on the spatial partitioning structure, the transformed cells of the spatial partitioning structure are identified through differential point cloud data, and the data of the transformed cells are updated. The cells with updated data are assigned new timestamps, while cells that have not changed remain unchanged, thus completing the construction of the four-dimensional spatiotemporal model.

3. The method for health assessment of distribution network equipment based on a point cloud model as described in claim 2, characterized in that, The constructed four-dimensional spatiotemporal model calculates the active envelope based on the actual process model of the dynamic object, and generates a static safety buffer region distributed along the power line, including: Based on the constructed four-dimensional spatiotemporal model, the kinematic parameters of the dynamic object are entered, and geometric operations are performed according to the kinematic parameters of the dynamic object. A set of spatial points is generated through geometric operations, and an active envelope is constructed based on the generated set of spatial points. Based on the active envelope, a static safety buffer zone is generated distributed along the power line.

4. The method for health assessment of distribution network equipment based on a point cloud model as described in claim 3, characterized in that, The method, based on the active envelope, transforms the spatiotemporal range jointly defined by the static safety region and the preset time window into a discrete four-dimensional grid. Based on this discrete four-dimensional grid, the occupancy probability of dynamic objects is calculated. A first-level algorithm quantifies the forward-looking risk value of dynamic objects on power lines, including: Based on the active envelope, the spatiotemporal range jointly defined by the static safety region and the preset time window is transformed into a discrete four-dimensional grid; Based on a discrete four-dimensional grid, the method for calculating the occupancy probability is selected according to the dynamic object, and the occupancy probability is calculated based on the selected calculation method. Based on the calculated occupancy probability, a first-level algorithm is used to quantify the forward-looking risk value of dynamic objects to power lines.

5. The method for health assessment of distribution network equipment based on a point cloud model as described in claim 4, characterized in that, The process of collecting data on the power distribution network and its surrounding environment, obtaining a baseline 3D point cloud model, and constructing a spatial partitioning structure includes: A drone equipped with a lidar is used to inspect the target area and obtain a baseline 3D point cloud model. The space is divided into an octree structure in a three-dimensional grid; Each internal node in the octree structure has eight child nodes, and the eight child nodes represent the parent node. The three-dimensional space is bisected along the X, Y, and Z axes to form eight octaves. Construct the smallest cube bounding box that surrounds the target point cloud as the root node, and recursively divide the nodes until the preset maximum tree depth is reached and the number of points contained in the leaf nodes is less than a threshold.

6. The method for health assessment of distribution network equipment based on a point cloud model as described in claim 5, characterized in that, The process of transforming the spatiotemporal range defined by the static security region and the preset time window into a discrete four-dimensional grid based on the active envelope includes: The continuous spatiotemporal range defined by the static safety buffer area (PSB) and the preset time window (T) is transformed into a discrete four-dimensional grid. In three-dimensional space, a three-dimensional voxel grid is constructed with the PSB as the core. In the time dimension, time windows are... Divided into Each time step is of equal length; Any spacetime unit in the grid Defined by a spatial voxel and a time interval; The range covered by spatial voxels is The time interval covers the range of ; in, , , These are the starting coordinates on the X, Y, and Z axes, respectively. These represent the resolution of the voxel mesh in the three spatial directions, respectively. For the starting moment in the time dimension, This represents the temporal resolution of the grid.

7. The method for health assessment of distribution network equipment based on a point cloud model as described in claim 6, characterized in that, The step of quantifying the forward-looking risk value of dynamic objects on power lines based on the calculated occupancy probability using a first-level algorithm includes: Forward-looking risk value The expression is: in, This is a forward-looking risk value. For the entire future time window Discrete-time integrals over time, For summation, For spacetime units At time step The probability of being occupied by a dynamic object. spatial unit The key weighting factors, For the grid index and time index A four-dimensional spacetime unit; definition The method is inverse distance weighting, and the key weight factor expression is: in, As a key weighting factor, The minimum distance from the center of the cell to the nearest critical device. Scaling factor It is a three-dimensional mesh unit. A spatial linear entity for critical facilities; Categorization assignment assigns discrete weight values ​​based on the type of infrastructure adjacent to the cell.

8. A health assessment system for distribution network equipment based on a point cloud model, using the method described in any one of claims 1-7, characterized in that, include: The four-dimensional spatiotemporal model construction module collects data on the power distribution network and its surrounding environment, constructs a four-dimensional spatiotemporal model, and calculates the active envelope based on the actual process model of the dynamic object, generating a static safety buffer zone distributed along the power line. The Quantification Forward-Looking Risk Value Module, based on the activity envelope, transforms the spatiotemporal range jointly defined by the static safety zone and the preset time window into a discrete four-dimensional grid. Based on the discrete four-dimensional grid, it calculates the occupancy probability of dynamic objects and quantifies the forward-looking risk value of dynamic objects on power lines through a first-level algorithm. The visualization module links forward-looking risk values ​​with a four-dimensional spatiotemporal model and presents the data visually.

9. A computer 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 according to 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 a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images