Method, device and electronic equipment for determining a line space topology

Abstract

This invention discloses a method, apparatus, and electronic device for determining a line spatial topology map. The method includes: acquiring multimodal sensor data; determining component instance information based on the multimodal sensor data; determining a global three-dimensional point cloud map based on visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data; obtaining device nodes representing energy-saving equipment and three-dimensional vector edges representing conductors based on the global three-dimensional point cloud map and the component categories and contour masks of power component instances in the component instance information; and performing topology connection operations based on the device nodes and three-dimensional vector edges to obtain a line spatial topology map. This invention solves the technical problems of delayed updates to line spatial topology information and missing topology connection relationships in related technologies.

Description

Technical Field

[0001] This invention relates to the field of data processing, and more specifically, to a method, apparatus, and electronic device for determining a line space topology map. Background Technology

[0002] As a crucial component of the power system, the 10kV distribution network is characterized by its wide-ranging lines and complex environment. Traditional GIS systems suffer from data lag and inaccurate topological relationships, making it difficult to meet the demands of digital operation and maintenance and intelligent dispatching. While existing drone inspection technologies can collect line images and point cloud data, they also suffer from technical issues such as delayed updates to line spatial topology information and missing topological connection relationships.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This invention provides a method, apparatus, and electronic device for determining a line spatial topology map, in order to at least solve the technical problems of delayed updates of line spatial topology information and missing topology connection relationships in related technologies.

[0005] According to one aspect of the present invention, a method for determining a line spatial topology map is provided, comprising: acquiring multimodal sensor data, wherein the multimodal sensor data is collected by a sensor system mounted on a UAV, including: visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data; determining component instance information based on the multimodal sensor data, wherein the component instance information includes the three-dimensional spatial location of a power component instance, component category, and contour mask, the component category including conductors; and determining global three-dimensional points based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data. The cloud map comprises: a global 3D point cloud map obtained based on the global UAV pose, which is obtained based on a pose optimization function and an initial 3D point cloud map; the initial 3D point cloud map obtained based on an error correction function; the pose optimization function is used to eliminate the cumulative drift error of the UAV pose; and the error correction function is used to correct the measurement error of the sensor. Based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information, device nodes representing energy-saving equipment and 3D vector edges representing conductors are obtained. Based on the device nodes and the 3D vector edges, a topology connection operation is performed to obtain a line space topology map.

[0006] Optionally, determining a global 3D point cloud map and a global UAV pose based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data includes: determining an initial 3D point cloud map and a corrected UAV pose based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data; and performing a global pose optimization operation based on the initial 3D point cloud map and the corrected UAV pose to determine the global 3D point cloud map and the global UAV pose, wherein the global pose optimization operation is used to eliminate accumulated drift errors.

[0007] Optionally, based on the initial 3D point cloud map and the corrected pose of the UAV, a global pose optimization operation is performed to determine the global 3D point cloud map and the global UAV pose. This includes: comparing the visual imaging data with historical keyframes to obtain a comparison result, wherein the historical keyframes are visual imaging frames collected and stored by the UAV in the explored area; if the comparison result indicates that the similarity is higher than a preset similarity threshold, a closed loop is identified, wherein the closed loop indicates that the current spatial position of the UAV is the same spatial position as the historical spatial position corresponding to the historical keyframe; if the closed loop is identified, a global pose optimization operation is performed based on the initial 3D point cloud map and the corrected pose of the UAV to determine the global 3D point cloud map and the global UAV pose.

[0008] Optionally, based on the initial 3D point cloud map and the corrected UAV pose, a global pose optimization operation is performed to determine the global 3D point cloud map and the global UAV pose. This includes: constructing a pose graph using the UAV poses corresponding to multiple time steps in the corrected UAV pose as pose nodes and the relative pose constraints between multiple pose nodes as edges; invoking a pose optimization function, wherein the pose optimization function aims to minimize the weighted sum of squared errors of the relative pose constraints between pose nodes; obtaining the global UAV pose based on the poses corresponding to multiple pose nodes in the pose graph and the pose optimization function; and updating the initial 3D point cloud map based on the global UAV pose to obtain the global 3D point cloud map.

[0009] Optionally, determining an initial 3D point cloud map and a corrected UAV pose based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data includes: constructing an initial state vector based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data, wherein the initial state vector includes initial UAV pose state quantities corresponding to multiple time steps and initial 3D coordinates of multiple map points, the multiple map points representing points constituting the environmental surface in the 3D point cloud map; retrieving an error correction function, wherein the error correction function includes multiple error terms and weight coefficients corresponding to the multiple error terms, the multiple error terms including visual error terms, lidar error terms, inertial measurement unit error terms, and positioning error terms, the weight coefficients corresponding to the multiple error terms being negatively correlated with the measurement uncertainty of the corresponding sensors; obtaining a target state vector based on the error correction function and the initial state vector; determining the corrected UAV pose based on the UAV pose state quantities corresponding to multiple time steps in the target state vector, and determining the initial 3D point cloud map based on the target 3D coordinates of the multiple map points in the target state vector.

[0010] Optionally, based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information, the device nodes representing the energy-saving equipment and the 3D vector edges representing the conductors are obtained, including: performing point cloud clustering operations based on spatial distance and geometric features on multiple map points in the global 3D point cloud map to obtain multiple device point cloud clusters, wherein the device point cloud cluster represents a set of map points belonging to the same device instance; determining the device type corresponding to each of the multiple device point cloud clusters; performing a parametric model fitting operation on the multiple device point cloud clusters to obtain device pose information of multiple device instances, wherein the device pose information includes 3D spatial coordinates and device posture, and the parametric model fitting operation is to fit the device point cloud clusters into a parametric 3D model corresponding to the device type to determine the 3D spatial coordinates and device posture of the device instance; determining the multiple device instances as nodes representing the energy-saving equipment, and assigning the device attribute information of the multiple device instances to the corresponding nodes, wherein the device attribute information includes the device type and the device pose information.

[0011] Optionally, based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information, the node attribute information representing the energy-saving device and the 3D vector edge representing the conductor are obtained, including: determining the conductor point set based on the contour mask of the conductor instance in the component instance information and the global 3D point cloud map, wherein the conductor point set is the set of map points belonging to the conductor in the global 3D point cloud map; performing a 3D space curve fitting operation on the conductor point set to obtain the continuous 3D vector line corresponding to the conductor, wherein the continuous 3D vector line represents the continuous trajectory of the conductor in 3D space, and the 3D space curve fitting operation includes at least one of the following: catenary equation, polynomial fitting algorithm; and determining the 3D vector lines corresponding to multiple conductors as the 3D vector edges representing the conductor.

[0012] According to one aspect of the present invention, a device for determining a line spatial topology map is provided, comprising: an acquisition module, configured to acquire multimodal sensor data, wherein the multimodal sensor data is collected by a sensor system mounted on a UAV, including: visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data; a first determination module, configured to determine component instance information based on the multimodal sensor data, wherein the component instance information includes the three-dimensional spatial position of a power component instance, component category, and contour mask, the component category including conductors; and a second determination module, configured to determine the entire... A global 3D point cloud map is obtained based on the global UAV pose, which is obtained based on a pose optimization function and an initial 3D point cloud map. The initial 3D point cloud map is obtained based on an error correction function. The pose optimization function is used to eliminate the cumulative drift error of the UAV pose, and the error correction function is used to correct the measurement error of the sensor. A third determining module is used to obtain device nodes representing energy-saving equipment and 3D vector edges representing conductors based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information. A fourth determining module is used to perform topology connection operations based on the device nodes and the 3D vector edges to obtain a line space topology map.

[0013] According to one aspect of the present invention, an electronic device is provided, comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to execute the instructions to implement the method for determining a line space topology as described in any of the preceding claims.

[0014] According to one aspect of the present invention, a computer-readable storage medium is provided, wherein when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform the method for determining the line space topology as described above.

[0015] In this embodiment of the invention, multimodal sensor data is acquired, which is collected by a sensor system mounted on a UAV, including: visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data. Based on the multimodal sensor data, component instance information is determined, including the three-dimensional spatial position of the power component instance, the component category, and the contour mask. The component category includes wires. Based on the visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data, a global three-dimensional point cloud map is determined, which is obtained based on the global UAV pose. The global UAV pose is obtained based on a pose optimization function and an initial three-dimensional point cloud map. The initial three-dimensional point cloud map is obtained based on an error correction function. The pose optimization function is used to eliminate the cumulative drift error of the UAV pose, and the error correction function is used to correct the measurement error of the sensor. Based on the global three-dimensional point cloud map and the component category and contour mask of the power component instance in the component instance information, device nodes representing energy-saving equipment and three-dimensional vector edges representing wires are obtained. Based on the device nodes and three-dimensional vector edges, a topology connection operation is performed to obtain a line space topology map. By employing a method that integrates multi-source sensor fusion mapping with topology reasoning, a global 3D point cloud map is constructed using error correction and pose optimization functions. Device nodes and conductor vector edges are extracted from the global 3D point cloud map, and topology connection operations are performed. This achieves the goal of automatically generating a line spatial topology map, thereby improving the efficiency of line spatial topology information updates and the accuracy of topology connection relationships. It also solves the technical problems of lagging line spatial topology information updates and missing topology connection relationships in related technologies. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0017] Figure 1 This is a flowchart of a method for determining a line space topology according to an embodiment of the present invention;

[0018] Figure 2 This is a flowchart of a method provided by an optional embodiment of the present invention;

[0019] Figure 3 This is a structural block diagram of a device for determining a line space topology according to an embodiment of the present invention. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] Example 1

[0023] According to an embodiment of the present invention, an embodiment of a method for determining a line space topology map is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0024] Figure 1 This is a flowchart of a method for determining a line space topology map according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps:

[0025] Step S102: Acquire multimodal sensor data, wherein the multimodal sensor data is collected by the sensor system carried by the UAV, including: visual imaging data, spatial depth point cloud data, attitude inertial sequence data and position coordinate data;

[0026] Among them, multimodal sensor data refers to the heterogeneous data set collected simultaneously by various types of sensors carried by UAVs in the same inspection mission. It can provide multi-dimensional and complementary panoramic perception information for power line inspection. For example, visual imaging data provides equipment appearance information, and spatial depth point cloud data provides three-dimensional geometric structure information.

[0027] Among them, visual imaging data refers to image data collected by high-definition visible light cameras and infrared thermal imaging cameras, which can clearly present the appearance, color, markings and temperature distribution information of power equipment.

[0028] Among them, spatial depth point cloud data refers to a set of three-dimensional spatial points generated by lidar by emitting laser beams and receiving reflected echoes. It can accurately measure the distance between drones and power lines, towers, and vegetation, providing centimeter-level precision spatial information for three-dimensional mapping.

[0029] Among them, attitude inertial sequence data refers to the time series of three-axis acceleration and three-axis angular velocity of the UAV output in real time by the inertial measurement unit, which can provide prior information on the UAV's body motion and assist in attitude estimation.

[0030] Among them, location coordinate data refers to the global latitude, longitude and elevation coordinates of the UAV obtained by the GPS receiver, which can provide absolute geographical reference for the UAV and support the correlation between positioning and geographical coordinates.

[0031] In this step, the sensor system onboard the UAV simultaneously collects raw data in four modalities while flying along a preset initial flight path. This step provides a time-aligned and spatially unified multimodal raw data foundation for subsequent component instance identification, 3D point cloud map construction, and topological relationship reasoning.

[0032] Step S104: Based on the multimodal sensor data, determine the component instance information, wherein the component instance information includes the three-dimensional spatial location of the power component instance, the component category, and the contour mask, and the component category includes wires;

[0033] Among them, component instance information refers to the structured data set containing individual attribute information of each power component, which is output after sensing and recognizing multimodal sensor data. It can provide structured input with category and spatial location for subsequent semantic extraction and topological relationship parsing, including the three-dimensional spatial location of power component instances, component category, and contour mask.

[0034] Among them, a power component instance refers to a specific equipment individual with a unique spatial location and individual attributes that is actually identified by a perception algorithm in a specific inspection scenario. For example, a certain tower instance has a unique three-dimensional coordinate, and a certain conductor segment instance has a unique spatial orientation.

[0035] Among them, the component category refers to the equipment type label to which the identified power component belongs, which can distinguish different types of power equipment. For example, the category label of the conductor indicates that the component is a linear device that connects to the tower.

[0036] Among them, the contour mask refers to the binary mask image output after pixel-level segmentation of power component instances. It can accurately depict the outline of the component at the pixel level, providing precise spatial constraints for subsequent segmentation of conductor point sets from point clouds.

[0037] In this step, the multimodal sensor data acquired in step S102 is sensed and identified to determine the three-dimensional spatial location, component category, and contour mask of each power component instance. Through this step, the system extracts structured component information from the raw sensor data that can be used for topology mapping, providing semantic guidance for the subsequent accurate extraction of conductor point sets and location of tower equipment from the point cloud map.

[0038] Step S106: Based on visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data, determine a global three-dimensional point cloud map. The global three-dimensional point cloud map is obtained based on the global UAV pose. The global UAV pose is obtained based on the pose optimization function and the initial three-dimensional point cloud map. The initial three-dimensional point cloud map is obtained based on the error correction function. The pose optimization function is used to eliminate the cumulative drift error of the UAV pose, and the error correction function is used to correct the measurement error of the sensor.

[0039] Among them, the global 3D point cloud map refers to a 3D spatial point set map that has consistency on a global spatial scale after sensor measurement error correction and closed-loop cumulative drift elimination. It can accurately depict the terrain, vegetation and 3D outline of all power facilities in the line corridor, providing a 3D spatial foundation with centimeter-level accuracy for semantic extraction and topological reasoning.

[0040] Among them, the global UAV pose refers to the UAV six-degree-of-freedom pose sequence at each time step obtained after global optimization of the pose graph, which eliminates the cumulative drift error. It can provide an accurate spatial reference benchmark for the global consistency update of the 3D point cloud map.

[0041] The initial 3D point cloud map refers to the 3D point cloud map obtained by fusing multi-sensor observation data and correcting the measurement errors of each sensor, but which has not yet eliminated the long-distance cumulative drift error.

[0042] Among them, the pose optimization function refers to a mathematical function that takes the UAV pose at each time step as a node and minimizes the weighted sum of squared relative pose constraint errors between each node as the optimization objective. It can eliminate the drift error accumulated by the front-end odometry through global optimization, so that the entire trajectory and map remain consistent throughout the entire route.

[0043] Among them, the error correction function is a mathematical function that aims to minimize the weighted sum of squares of the error terms of the inertial measurement unit, the visual error term, the lidar error term, and the global positioning system error term. It can fuse observation data from multiple sensors and jointly correct the measurement errors of each sensor. The weight of each error term is negatively correlated with the measurement uncertainty of the corresponding sensor.

[0044] In this step, an error correction function is used to jointly optimize multi-sensor observations, correcting the measurement errors of each sensor to obtain an initial 3D point cloud map. Then, a pose optimization function is used to eliminate accumulated drift errors to obtain the global UAV pose. Based on the global UAV pose, the initial 3D point cloud map is updated to obtain the global 3D point cloud map. Through this step, the error correction function effectively corrects the measurement errors of a single sensor in a corridor environment with limited GPS signals by jointly optimizing multi-source observation data such as vision, LiDAR, inertial measurement unit, and GPS. The pose optimization function eliminates the drift errors accumulated during long-distance flight through closed-loop detection and global pose map optimization. The final global 3D point cloud map has centimeter-level accuracy and spatial consistency across the entire route, providing a high-precision 3D spatial foundation for subsequent semantic extraction.

[0045] Step S108: Based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information, obtain the device node representing the energy-saving device and the 3D vector edge representing the conductor.

[0046] Among them, equipment nodes refer to the topology graph node data units of power equipment such as poles and towers extracted from the global 3D point cloud map through clustering and template matching. They can carry attribute information such as the type of the corresponding equipment, 3D spatial coordinates and equipment attitude, providing the node foundation for the construction of the topology graph.

[0047] Among them, the three-dimensional vector edge refers to the vector line data that represents the continuous direction of the conductor in three-dimensional space after the conductor point set is segmented from the global three-dimensional point cloud map and then fitted by three-dimensional space curve. It can describe the precise geometric direction and sag shape of the conductor in three-dimensional space, and provide the edge basis for the construction of the topology map.

[0048] In this step, spatial point clustering is performed on the global 3D point cloud map to separate device point cloud clusters. These clusters are then matched with a parametric 3D template library to determine device types, and model fitting is performed to obtain device node attributes. Simultaneously, based on the contour mask of the conductor instances in the component instance information, conductor point sets are segmented from the point cloud map. 3D spatial curve fitting is then performed on these conductor point sets to generate continuous 3D vector edges. This step transforms the dense point cloud into structured vector elements, bridging the gap between a dense point cloud geometric map and a structured object vector representation. Each device node carries precise spatial location and type information, and the conductor vector edges accurately describe the spatial direction and sag parameters of the conductors, providing directly operable structured elements for subsequent topological relationship reasoning.

[0049] Step S110: Based on the device nodes and three-dimensional vector edges, perform topology connection operations to obtain the line space topology map.

[0050] Among them, the topology connection operation refers to the operation of automatically inferring and establishing the connection relationship between each node and each vector edge based on the spatial coordinates of the device node and the spatial endpoint position of the three-dimensional vector edge, and based on spatial proximity. It can organize discrete nodes and edges into a graph structure that describes the electrical connection relationship of the line.

[0051] Among them, the line spatial topology map refers to the graph structure model that describes the spatial connection relationship between various equipment in the overhead line. It adopts a node-edge graph structure, with nodes corresponding to equipment such as poles and towers, and edges corresponding to conductors. It can provide an accurate line topology data foundation for digital operation and maintenance, automated scheduling and intelligent analysis.

[0052] In this step, based on the device nodes and 3D vector edges output in step S108, the system automatically establishes connections by determining the spatial distance between the spatial endpoints of the 3D vector edges and each device node. For cases where multiple 3D vector edges converge at the same device node, they are identified as branch nodes, and a branch topology is generated, constructing a complete line spatial topology map. Through this step, the system upgrades from identifying which devices and conductors to understanding how these devices and conductors are connected, achieving a leap from discrete component identification to structured topological relationship reasoning. The automatically constructed line spatial topology map fundamentally solves the pain points of traditional GIS systems, which rely on manual input for topology information, resulting in delayed updates and poor accuracy.

[0053] Through steps S102-S110 above, multimodal sensor data is acquired. This multimodal sensor data is collected by the sensor system mounted on the UAV and includes: visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data. Based on the multimodal sensor data, component instance information is determined. This component instance information includes the three-dimensional spatial location of the power component instance, the component category, and a contour mask. The component category includes wires. Based on the visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data, a global three-dimensional point cloud map is determined. The local 3D point cloud map is obtained based on the global UAV pose, which in turn is derived from the pose optimization function and the initial 3D point cloud map. The initial 3D point cloud map is then obtained based on the error correction function. The pose optimization function is used to eliminate the cumulative drift error of the UAV pose, while the error correction function is used to correct the measurement error of the sensors. Based on the global 3D point cloud map and the component category and contour mask of the power component instances in the component instance information, the device nodes representing the energy-saving equipment and the 3D vector edges representing the conductors are obtained. Based on the device nodes and the 3D vector edges, a topology connection operation is performed to obtain the line space topology map. By adopting a method that combines multi-source sensor fusion mapping with topology relationship reasoning, a global 3D point cloud map is constructed through the error correction function and the pose optimization function. Device nodes and conductor vector edges are extracted from the global 3D point cloud map, and a topology connection operation is performed. This achieves the goal of automatically generating a line space topology map, thereby improving the technical effect of improving the update efficiency of line space topology information and the accuracy of topology connection relationships. This solves the technical problems of lagging line space topology information updates and missing topology connection relationships in related technologies.

[0054] As an optional embodiment, a global 3D point cloud map and a global UAV pose are determined based on visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data. This includes: determining an initial 3D point cloud map and a corrected UAV pose based on visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data; and performing a global pose optimization operation based on the initial 3D point cloud map and the corrected UAV pose to determine the global 3D point cloud map and the global UAV pose, wherein the global pose optimization operation is used to eliminate accumulated drift errors.

[0055] The initial 3D point cloud map refers to the 3D point cloud map obtained by fusing observation data from multiple sensors and correcting the measurement errors of each sensor, but which has not yet eliminated the long-distance cumulative drift error. It can provide an initial spatial geometric reference for subsequent global pose optimization.

[0056] Among them, UAV pose correction refers to the joint optimization of the UAV six-degree-of-freedom pose sequence at each time step obtained after multi-sensor observation through error correction function. It can provide initial pose node information for global pose optimization and serve as input for pose graph construction and global optimization.

[0057] Among them, global pose optimization operation refers to the optimization operation of making global consistency adjustments to the pose of the UAV to eliminate the cumulative drift error of long-distance flight, which can make the entire flight trajectory and the 3D point cloud map maintain spatial consistency throughout the entire route.

[0058] In this embodiment, an initial 3D point cloud map and a corrected UAV pose are first obtained by jointly optimizing multi-sensor data through an error correction function. Then, a global pose optimization operation is used to eliminate accumulated drift errors, resulting in a global 3D point cloud map and a global UAV pose. This approach divides the mapping process into two progressive stages: front-end error correction and back-end global optimization. The front-end integrates observation data from multiple sensors to correct the measurement errors of each sensor, while the back-end eliminates the drift errors accumulated by the front-end odometer through global pose optimization. The final generated global 3D point cloud map has centimeter-level accuracy and spatial consistency across the entire route.

[0059] As an optional embodiment, based on the initial 3D point cloud map and the corrected UAV pose, a global pose optimization operation is performed to determine the global 3D point cloud map and the global UAV pose. This includes: comparing the similarity between visual imaging data and historical keyframes to obtain a comparison result, wherein the historical keyframes are visual imaging frames collected and stored by the UAV in the explored area; if the comparison result indicates that the similarity is higher than a preset similarity threshold, a closed loop is identified, wherein a closed loop indicates that the current spatial position of the UAV is the same as the historical spatial position corresponding to the historical keyframe; if a closed loop is identified, a global pose optimization operation is performed based on the initial 3D point cloud map and the corrected UAV pose to determine the global 3D point cloud map and the global UAV pose.

[0060] Among them, historical keyframes refer to representative visual imaging frames collected and stored by the UAV when it flies through the explored area. They can serve as matching reference images for closed-loop detection and provide visual feature basis for determining whether the UAV has returned to the location it has passed through.

[0061] Among them, similarity comparison refers to the operation of matching features between the currently acquired visual imaging data and historical keyframes and calculating image similarity. It can quantify the visual similarity between the current scene and the historical scene, and provide an objective basis for closed-loop recognition.

[0062] The preset similarity threshold refers to the minimum similarity value set in advance to determine whether a closed loop has been formed. When the similarity is higher than the threshold, it can be confirmed that the current spatial location of the drone is the same as the historical spatial location corresponding to the historical key frame.

[0063] Among them, closed loop refers to the state where the current spatial position of the UAV is the same as the historical spatial position corresponding to the historical key frame. It can provide spatial position constraints between two time steps with a large interval for global pose optimization, and serve as key constraint information to eliminate cumulative drift error.

[0064] In this embodiment, the similarity between the current visual imaging data and historical keyframes is compared. When the similarity exceeds a preset similarity threshold, a loop closure is identified, and global pose optimization is only performed after a loop closure is detected. This method provides strong spatial constraints between discontinuous time steps for global pose optimization. When the UAV returns to a previously visited location, the loop closure constraint binds the current pose to the historical pose, adding long-distance loop constraint edges to the pose graph. This allows the optimization algorithm to evenly distribute the drift error accumulated by the front-end odometry along the entire pose graph, effectively eliminating drift and ensuring the global consistency of the final map.

[0065] As an optional embodiment, based on the initial 3D point cloud map and the corrected UAV pose, a global pose optimization operation is performed to determine the global 3D point cloud map and the global UAV pose. This includes: constructing a pose graph using the UAV poses corresponding to multiple time steps in the corrected UAV pose as pose nodes and the relative pose constraints between multiple pose nodes as edges; invoking a pose optimization function, wherein the pose optimization function aims to minimize the weighted sum of squared errors of the relative pose constraints between pose nodes; obtaining the global UAV pose based on the poses corresponding to multiple pose nodes in the pose graph and the pose optimization function; and updating the initial 3D point cloud map based on the global UAV pose to obtain the global 3D point cloud map.

[0066] Among them, pose nodes refer to graph nodes with the six-degree-of-freedom pose of the UAV at each time step in the UAV pose correction as the unit. They can serve as the basic variable units for pose graph optimization. Each pose node carries the position and attitude information of the UAV at that moment.

[0067] Among them, the pose graph refers to a graph structure model constructed with the pose of the UAV at each moment as nodes and the relative pose constraints between the nodes as edges. It can express the spatial constraint relationship between the poses at each moment in the entire flight trajectory of the UAV in the form of a graph.

[0068] Among them, relative pose constraint refers to the relative transformation relationship of the UAV pose between two time points represented by the edge connecting two pose nodes in the pose graph. It can provide spatial constraint information between nodes for pose optimization, including adjacent frame constraints provided by odometry and long-distance loop closure constraints provided by loop closure detection.

[0069] The pose optimization function is a mathematical function that aims to minimize the weighted sum of squared errors of relative pose constraints between each pose node. It can minimize the total error of all relative pose constraints by globally adjusting the pose values ​​of each pose node.

[0070] In this embodiment, a pose graph is constructed using the UAV pose at each time step in the UAV pose correction process as nodes and the relative pose constraints between nodes as edges. A pose optimization function is invoked to globally optimize the poses of each node, resulting in the global UAV pose. Then, the initial 3D point cloud map is updated based on the global UAV pose to obtain the global 3D point cloud map. In this way, the pose graph models the entire flight trajectory as a graph structure. The odometry constraints carried by each edge ensure the local smoothness of the trajectory, while the long-distance constraints provided by closed-loop edges ensure the global consistency of the trajectory. The pose optimization function, through a global optimization framework, simultaneously satisfies all local and global constraints, ultimately outputting a globally consistent UAV pose trajectory and a global 3D point cloud map.

[0071] As an optional embodiment, based on visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data, an initial 3D point cloud map and a corrected UAV pose are determined, including: constructing an initial state vector based on visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data, wherein the initial state vector includes the initial UAV pose state quantities corresponding to multiple time steps and the initial 3D coordinates of multiple map points, the multiple map points representing points constituting the environmental surface in the 3D point cloud map; retrieving an error correction function, wherein the error correction function includes multiple error terms and corresponding weight coefficients for each of the multiple error terms, the multiple error terms including visual error terms, lidar error terms, inertial measurement unit error terms, and positioning error terms, the weight coefficients corresponding to each of the multiple error terms being negatively correlated with the measurement uncertainty of the corresponding sensor; obtaining a target state vector based on the error correction function and the initial state vector; determining the corrected UAV pose based on the UAV pose state quantities corresponding to multiple time steps in the target state vector, and determining the initial 3D point cloud map based on the target 3D coordinates of the multiple map points in the target state vector.

[0072] The initial state vector refers to the multidimensional state variables to be estimated in the error correction optimization problem. It can simultaneously contain the UAV pose state variables at each time step and the three-dimensional coordinates of multiple map points, serving as the set of variables for error correction function optimization.

[0073] Among them, the initial pose state quantity of the UAV refers to the estimated motion state quantity of the UAV at each time step in the initial state vector, which can describe the spatial position, velocity, attitude and inertial measurement unit bias of the UAV at that moment.

[0074] Among them, map points refer to the three-dimensional coordinate points that constitute the surface of the environment in the three-dimensional point cloud map. They can be used as one of the optimization variables of the error correction function. During the optimization process, their three-dimensional coordinates are adjusted together with the attitude of the unmanned vehicle.

[0075] Among them, the error correction function is a mathematical function that aims to minimize the weighted sum of squares of the error terms of the inertial measurement unit, the visual error term, the lidar error term, and the global positioning system error term. The weight coefficient of each error term is negatively correlated with the measurement uncertainty of the corresponding sensor. It can fuse observation data from multiple sensors and jointly correct the measurement errors of each sensor.

[0076] The target state vector refers to the optimal state vector obtained after optimization by the error correction function, which includes the optimized UAV pose state variables and the three-dimensional coordinates of map points.

[0077] In this embodiment, an initial state vector is constructed, comprising the initial pose state of the UAV and the initial 3D coordinates of map points. An error correction function is invoked to jointly optimize this initial state vector to obtain the target state vector. The corrected UAV pose and the initial 3D point cloud map are then extracted from the target state vector. In this way, the error correction function unifies the observation constraints of multiple sensors, such as the inertial measurement unit, vision, lidar, and global positioning system, into a joint optimization framework. The weights of each error term are adaptively adjusted according to the measurement uncertainties of the corresponding sensor. The contribution of sensor observations with high uncertainties to the optimization is automatically reduced, thus achieving optimal state estimation results under different environmental conditions.

[0078] As an optional embodiment, based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information, the device nodes representing the energy-saving equipment and the 3D vector edges representing the conductors are obtained. This includes: performing point cloud clustering operations based on spatial distance and geometric features on multiple map points in the global 3D point cloud map to obtain multiple device point cloud clusters, where each device point cloud cluster represents a set of map points belonging to the same device instance; determining the device type corresponding to each of the multiple device point cloud clusters; performing parametric model fitting operations on the multiple device point cloud clusters to obtain the device pose information of multiple device instances, where the device pose information includes 3D spatial coordinates and device posture, and the parametric model fitting operation is to fit the device point cloud clusters into a parametric 3D model corresponding to the device type to determine the 3D spatial coordinates and device posture of the device instances; determining the multiple device instances as nodes representing the energy-saving equipment, and assigning the device attribute information of the multiple device instances to the corresponding nodes, where the device attribute information includes device type and device pose information.

[0079] Among them, the device point cloud cluster refers to the spatial collection of map points belonging to the same device instance separated from the global 3D point cloud map through spatial clustering. It can organize scattered point clouds into independent data units according to device affiliation.

[0080] Among them, the parametric model fitting operation refers to the operation of registering and fitting the spatial point distribution of the device point cloud cluster with the parametric 3D model of the corresponding device type, which can determine the precise position coordinates and attitude parameters of the device instance in 3D space.

[0081] Among them, device pose information refers to the spatial positioning and orientation information of the device instance obtained through parametric model fitting operation, which can accurately describe the position and attitude of the device in three-dimensional space, including three-dimensional spatial coordinates and device attitude.

[0082] Among them, device attribute information refers to the set of structured information assigned to each device node that describes the device's identity and spatial state. It can provide a complete attribute description for the nodes in the topology graph, including device type and device pose information.

[0083] In this embodiment, point cloud clustering based on spatial distance and geometric features is performed on each map point in the global 3D point cloud map to obtain multiple device point cloud clusters. After determining the device type corresponding to each device point cloud cluster, parametric model fitting is performed to obtain the device pose information of each device instance. Each device instance is then identified as a topology graph node and assigned device attribute information. In this way, point cloud clustering separates dense point clouds into independent data units according to device affiliation. Parametric model fitting accurately extracts the 3D spatial coordinates and pose of each device through template matching and spatial registration, forming structured device nodes with accurate spatial positioning and type information, providing a high-quality node foundation for subsequent topology relationship reasoning.

[0084] As an optional embodiment, based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information, the node attribute information representing the energy-saving device and the 3D vector edge representing the conductor are obtained, including: determining the conductor point set based on the contour mask of the conductor instance in the component instance information and the global 3D point cloud map, wherein the conductor point set is the set of map points belonging to the conductor in the global 3D point cloud map; performing a 3D space curve fitting operation on the conductor point set to obtain the continuous 3D vector line corresponding to the conductor, wherein the continuous 3D vector line represents the continuous trajectory of the conductor in 3D space, and the 3D space curve fitting operation includes at least one of the following: catenary equation, polynomial fitting algorithm; and determining the 3D vector lines corresponding to multiple conductors as the 3D vector edges representing the conductor.

[0085] The traverse point set refers to the spatial set of map points belonging to the traverse extracted from the global 3D point cloud map based on the contour mask of the traverse instance, which can provide original spatial point data for 3D spatial curve fitting of the traverse.

[0086] Among them, continuous three-dimensional vector lines refer to vector lines that are continuous and smooth in three-dimensional space after spatial curve fitting of the traverse point set, which can accurately describe the geometric direction and morphological characteristics of the traverse in three-dimensional space.

[0087] Among them, the three-dimensional space curve fitting operation refers to the operation of continuously approximating a discrete set of traverse points using the catenary equation or polynomial fitting algorithm, which can transform discrete spatial points into continuous functional vector representations.

[0088] In this embodiment, the conductor point set is segmented from the global 3D point cloud map based on the contour mask of the conductor instance in the component instance information. A 3D spatial curve fitting operation using the catenary equation or polynomial fitting is performed on the conductor point set to generate continuous 3D vector lines as the 3D vector edges of the conductor. In this way, the contour mask provides accurate spatial constraints for the conductor region in the point cloud segmentation, avoiding interference from non-conductor points such as towers and vegetation on the fitting accuracy. The catenary equation, based on the natural stress shape of the conductor, makes the generated vector lines physically closer to the actual sag shape of the conductor. Polynomial fitting provides more flexible expressive power under non-ideal conductor shapes. The two fitting algorithms complement each other, and the final generated continuous 3D vector lines accurately describe the spatial direction and sag characteristics of the conductor, providing high-precision geometric data for the construction of edges and the analysis of sag parameters in the topology graph.

[0089] Based on the above embodiments and optional embodiments, an optional implementation method is provided, which is described in detail below.

[0090] In related technologies, the 10kV distribution network, as an important component of the power system, has widely distributed lines and complex environments. Traditional GIS systems suffer from lagging data updates and inaccurate topological relationships, making it difficult to meet the needs of digital operation and maintenance and intelligent dispatching. While existing UAV inspection technology can collect line images and point cloud data, it suffers from the following drawbacks: 1. Insufficient mapping accuracy: In corridor environments with limited GPS signals (such as dense forests and valleys), relying solely on GNSS positioning easily leads to map drift, making it difficult to achieve centimeter-level accuracy in 3D reconstruction. 2. Lack of topological relationships: Existing technologies mostly remain at the level of point cloud collection and visualization, lacking automatic identification and structured expression of the connection relationships between line components (towers, conductors, branch points), generating only "visual maps" rather than "semantic maps." 3. Data silos: Inspection data (defects, temperature measurements, etc.) are stored separately from the line geometric model, making attribute association impossible and hindering digital twin applications. 4. High reliance on manual labor: Key topological information such as line routing, tower locations, and conductor connection relationships still rely on manual surveying and input, which is inefficient and prone to errors.

[0091] In view of this, an optional embodiment of the present invention provides a method for determining the spatial topology of a power line, which can also be called a high-precision 3D mapping and topological relationship reasoning method for 10kV lines. Through UAV automatic tracking technology, massive amounts of multimodal sensing data are transformed into an accurate, complete, and semantically rich digital line model, achieving high-precision 10kV line map construction. The optional embodiment of the present invention, through a progressive technical route from "multi-source fusion SLAM (localization and geometry construction) to semantic extraction and vectorization (objectification) to topological relationship reasoning (structuring)," successfully transforms the raw sensing data from UAVs into a digital twin model of power infrastructure that combines geometric accuracy, topological correctness, and semantic richness. This model fundamentally solves the problems of lagging data updates and inaccurate topology in traditional GIS systems, providing a unique and reliable data foundation for the digital operation and maintenance, automated scheduling, and intelligent analysis of distribution networks. Figure 2 This is a flowchart of a method provided by an optional embodiment of the present invention, such as... Figure 2 The diagram illustrates the five-stage core process of the method of this invention: multi-source data acquisition: simultaneous acquisition of LiDAR, visual, IMU, and GNSS data; multi-source fusion SLAM mapping: tightly coupled SLAM and closed-loop optimization to generate a high-precision point cloud map; semantic extraction and vectorization: tower identification and conductor fitting to extract vector representations of power equipment; topological relationship reasoning: constructing a node-edge graph structure to identify complex topologies such as branch points; standardized output and integration: CIM / GeoJSON format output for seamless integration with GIS systems.

[0092] The technical solution of the present invention is as follows:

[0093] (1) SLAM with multi-source data fusion and high-precision map construction:

[0094] To address the challenge of accurate positioning and mapping in corridor environments where GPS signals may be unavailable, the system employs a tightly coupled multi-sensor fusion SLAM framework.

[0095] Vision-Inertial-LiDAR Tightly Coupled SLAM: Using visual feature points, line segment features (especially applicable to wires) extracted by the front-end perception module and LiDAR point cloud features as front-end association information, combined with the body motion prior provided by the IMU, joint state estimation is performed through graph optimization or filtering algorithms.

[0096] The system state vector is defined as:

[0097]

[0098] in, Let the position, velocity, attitude, and IMU bias of the UAV at time k be given. Let be the three-dimensional coordinates of the j-th map point.

[0099] Where X: the state vector of the entire system, which includes the state of the UAV at all times and the coordinates of all map points, is the set of variables to be estimated in graph optimization.

[0100] xk: The local state vector of the UAV at time k. Where:

[0101] pk: The three-dimensional position of the UAV in the global coordinate system (usually x, y, z), in meters.

[0102] vk: The three-dimensional velocity of the drone (vx, vy, vz), in meters per second.

[0103] qk: The attitude quaternion (qw, qx, qy, qz) of the UAV represents the rotation from the machine system to the global system, used to describe pitch, roll, and yaw angles.

[0104] ba: Accelerometer bias (triaxial), in meters per second², used to compensate for systematic errors in IMU acceleration measurements.

[0105] bg: Gyroscope bias (three-axis), in radians per second, used to compensate for zero bias in angular velocity measurements.

[0106] λj: The 3D coordinates (Xj, Yj, Zj) of the j-th map point, in meters. These points are environmental feature points reconstructed by LiDAR or visual triangulation, forming a sparse or dense point cloud map.

[0107] k=0,1,…,n: represents the n+1 drone states from the initial time 0 to the current time n.

[0108] j=0,1,…,m: indicates that there are m+1 landmarks in the map.

[0109] Based on the aforementioned visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data, an initial state vector is constructed. The initial state vector includes the initial pose state of the UAV corresponding to multiple time steps and the initial three-dimensional coordinates of multiple map points. The multiple map points represent the points that constitute the surface of the environment in the three-dimensional point cloud map.

[0110] The objective function to be optimized is:

[0111]

[0112] In the above formula, the multiple terms are IMU error term, visual error term, lidar error term, and GPS error term, respectively. The weight matrix Σ represents Mahalanobis distance, which indicates the uncertainty of sensor measurement. The greater the uncertainty of the sensor (such as when the GPS signal is poor), the larger its Σ is, and the smaller the weight of this term in the optimization.

[0113] By solving the above optimization problem in real time, the system can calculate the six-degree-of-freedom pose of the UAV and simultaneously generate a three-dimensional point cloud map with centimeter-level accuracy, clearly depicting the three-dimensional outline of the terrain, vegetation and all power facilities of the route corridor.

[0114] Similar to the above, the error correction function is retrieved, which includes multiple error terms and corresponding weight coefficients for each error term. The multiple error terms include visual error terms, lidar error terms, inertial measurement unit error terms, and positioning error terms. The weight coefficients for each error term are negatively correlated with the measurement uncertainty of the corresponding sensor. Based on the error correction function and the initial state vector, the target state vector is obtained. Based on the UAV pose state variables corresponding to multiple time steps in the target state vector, the corrected UAV pose is determined, and based on the target 3D coordinates of multiple map points in the target state vector, the initial 3D point cloud map is determined.

[0115] Global pose graph optimization and loop closure detection: To eliminate accumulated drift errors, a loop closure detection mechanism is introduced. When the UAV revisits an explored area, a visual bag-of-words model is used to compare the similarity between the current scene and historical keyframes to identify loops.

[0116] Similarity comparison is performed between the visual imaging data and historical keyframes as described above to obtain the comparison results. The historical keyframes are visual imaging frames collected and stored by the UAV in the explored area. If the comparison results indicate that the similarity is higher than a preset similarity threshold, a closed loop is identified. A closed loop means that the current spatial position of the UAV is the same as the historical spatial position corresponding to the historical keyframe. If a closed loop is identified, a global pose optimization operation is performed based on the initial 3D point cloud map and the corrected pose of the UAV to determine the global 3D point cloud map and the global UAV pose.

[0117] Then, the pose graph optimization problem is constructed:

[0118]

[0119] Wherein, min: minimization operation. The goal is to find an optimal set of pose variables {p0, p1, ..., pn} such that the weighted sum of squares of all error terms is minimized. p0, p1, ..., pn: UAV pose at keyframes. Each pi typically represents the UAV's position and attitude in the global coordinate system at time i. pi and pj: represent the poses of the i-th and j-th keyframes, respectively. There are usually relative measurement constraints between them. The squared Mahalanobis distance represents the error term between two poses. Consistent high-precision maps and trajectories are obtained through global adjustments.

[0120] Similar to the above, in the process of UAV pose correction, the UAV poses corresponding to multiple time steps are used as pose nodes, and the relative pose constraints between multiple pose nodes are used as edges to construct a pose graph; the pose optimization function is invoked, where the optimization objective of the pose optimization function is to minimize the weighted sum of squared errors of the relative pose constraints between pose nodes; based on the poses corresponding to multiple pose nodes in the pose graph and the pose optimization function, the global UAV pose is obtained; based on the global UAV pose, the initial 3D point cloud map is updated to obtain the global 3D point cloud map.

[0121] (2) Semantic extraction and vector deconstruction of topological elements:

[0122] Based on the high-precision point cloud map and UAV trajectory, the system extracts key semantic information and transforms the dense point cloud into structured vector elements.

[0123] Pole and tower and key equipment instance extraction and location:

[0124] Point cloud clustering and classification: Apply point cloud clustering algorithms (such as DBSCAN) based on Euclidean distance and features to the global point cloud to separate independent device point cloud clusters.

[0125] As described above, the device types corresponding to multiple device point cloud clusters are determined. Parametric model fitting is performed on the multiple device point cloud clusters to obtain the device pose information of multiple device instances. The device pose information includes three-dimensional spatial coordinates and device posture. The parametric model fitting operation is to fit the device point cloud clusters into a parametric three-dimensional model corresponding to the device type to determine the three-dimensional spatial coordinates and device posture of the device instances. The multiple device instances are determined as nodes representing energy-saving devices, and the device attribute information of the multiple device instances is assigned to the corresponding nodes. The device attribute information includes device type and device pose information.

[0126] Model matching and parametric fitting: Utilizing a pre-built parametric 3D template library of equipment such as poles and transformers, each point cloud cluster is identified and model-fitted using an iterative nearest-point algorithm or a deep learning point cloud segmentation network. The final output includes the type, 3D spatial coordinates (latitude, longitude, and elevation), and attitude (e.g., pole tilt) of each device, which are then used as nodes in the topology graph.

[0127] 3D Reconstruction and Vectorization of Conductors and Ground Wires:

[0128] Trajectory extraction of conductor point cloud: Combining visual recognition results, the point sets belonging to conductors and ground wires are segmented from the original point cloud.

[0129] Catenary model fitting: Based on physical knowledge, the catenary equation or polynomial fitting algorithm is used to fit a discrete conductor point cloud in three-dimensional space. This process outputs a continuous, smooth three-dimensional vector line for each conductor / ground wire and calculates key parameters such as the height of the lowest sag point and the amount of sag. These vector lines will serve as edges connecting nodes in the topology graph.

[0130] Based on the contour mask of the traverse instance in the component instance information and the global 3D point cloud map, the traverse point set is determined, wherein the traverse point set is the set of map points belonging to the traverse in the global 3D point cloud map; a 3D space curve fitting operation is performed on the traverse point set to obtain the continuous 3D vector lines corresponding to the traverse, wherein the continuous 3D vector lines represent the continuous trajectory of the traverse in 3D space, and the 3D space curve fitting operation includes at least one of the following: catenary equation, polynomial fitting algorithm; the 3D vector lines corresponding to multiple traverses are determined as the 3D vector edges representing the traverses.

[0131] (3) Topological relationship reasoning and structured map generation:

[0132] This step is crucial for giving the map "topological" properties, organizing discrete nodes and edges into a graph structure that describes the connections between them.

[0133] Spatial Relationship and Connection Logic Reasoning: Based on geometric proximity and electrical connection rules, the system automatically infers the connection relationships between nodes and edges, and constructs a topology graph.

[0134]

[0135] The node set Includes equipment such as poles, transformers, and switches, and is part of a cluster. These are conductor vector lines. Each edge necessarily connects two nodes, forming a connected graph. For branch points (such as T-junctions), multiple conductors converging at the same node are identified, and the branch topology is automatically generated.

[0136] Attribute association and semantic enrichment: All attributes output by the front-end intelligent recognition module (such as device ID, model, identified defect type, infrared temperature measurement data, and acoustic diagnostic results) are associated with corresponding nodes and edges in the topology graph. This upgrades the topology graph from a simple geometric network into a semantic map carrying rich operational information.

[0137] (4) Standardized data output and GIS system integration:

[0138] Ultimately, the system converts the constructed structured topology data into a format that conforms to power industry standards (such as CIM standards or GeoJSON), enabling seamless integration with existing power grid management systems.

[0139] Output: The generated 10kV line spatial topology map contains the following core elements:

[0140] Node layer: A table showing the spatial location and attributes of equipment such as poles, transformers, and switches.

[0141] Edge layer: The three-dimensional spatial orientation and electrical parameters of conductors and ground wires.

[0142] Topology connection table: explicitly describes the connection relationships between nodes and edges.

[0143] Inspection data layer: Real-time inspection data such as defects, temperature measurement, and point cloud data.

[0144] The above optional implementation methods can achieve at least the following beneficial effects:

[0145] (1) High-precision positioning and mapping capability: The present invention adopts a vision-inertial-lidar tightly coupled SLAM framework, which can still achieve centimeter-level accurate positioning in complex corridor environments where GPS signals are limited. Through closed-loop detection and global pose map optimization, it effectively eliminates accumulated drift error and ensures the global consistency and geometric accuracy of long-distance inspection maps.

[0146] (2) Automated semantic extraction and topological reasoning: Point cloud clustering and deep learning segmentation technology are applied to automatically identify power facilities such as poles and conductors; the catenary model fitting algorithm is used to convert discrete point clouds into continuous vector lines, and node-edge topology graph structure is intelligently constructed based on geometric proximity and electrical rules to accurately identify complex topological relationships such as branch points and T-junctions, completely eliminating the dependence on manual surveying.

[0147] (3) Data fusion and standardized output: Deeply associate multi-source data such as inspection defects, infrared temperature measurement, and acoustic diagnosis with geometric models to achieve data sharing of "one-stop collection and multiple applications"; support industry standard formats such as CIM and GeoJSON, and seamlessly connect with existing GIS systems to fundamentally solve the pain points of traditional GIS data update lag and inaccurate topology connection.

[0148] (4) Full-process automation and engineering value: The entire process from data acquisition and processing to topology generation is automated, requiring no manual intervention, which greatly improves mapping efficiency and supports the rapid digitization of large-scale distribution networks. This invention provides a reliable data foundation for the digital operation and maintenance, automated scheduling and intelligent analysis of distribution networks, and has significant engineering application value and industrial promotion prospects.

[0149] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0150] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. 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 is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0151] Example 2

[0152] According to an embodiment of the present invention, an apparatus for implementing the above-described method for determining a line spatial topology map is also provided. Figure 3 This is a structural block diagram of a device for determining a line space topology according to an embodiment of the present invention, such as... Figure 3 As shown, the device includes: an acquisition module 302, a first determination module 304, a second determination module 306, a third determination module 308, and a fourth determination module 310. The device will be described in detail below.

[0153] The acquisition module 302 is used to acquire multimodal sensor data, which is collected by the sensor system carried by the UAV and includes: visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data. The first determination module 304, connected to the acquisition module 302, is used to determine component instance information based on the multimodal sensor data. The component instance information includes the three-dimensional spatial location of the power component instance, the component category, and a contour mask. The component category includes wires. The second determination module 306, connected to the first determination module 304, is used to determine a global three-dimensional point cloud map based on the visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data. The cloud map is obtained based on the global UAV pose, which is obtained based on the pose optimization function and the initial 3D point cloud map. The initial 3D point cloud map is obtained based on the error correction function. The pose optimization function is used to eliminate the cumulative drift error of the UAV pose, and the error correction function is used to correct the measurement error of the sensor. The third determination module 308, connected to the second determination module 306, is used to obtain the device nodes representing the energy-saving equipment and the 3D vector edges representing the conductors based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information. The fourth determination module 310, connected to the third determination module 308, is used to perform topology connection operation based on the device nodes and the 3D vector edges to obtain the line space topology map.

[0154] It should be noted that the above-mentioned acquisition module 302, first determination module 304, second determination module 306, third determination module 308 and fourth determination module 310 correspond to steps S102 to S110 in the method for determining the spatial topology of the implementation line. The multiple modules and the corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in the above embodiment 1.

[0155] Example 3

[0156] According to another aspect of the present invention, an electronic device is also provided, comprising: a processor; and a memory for storing processor-executable instructions, wherein the processor is configured to execute instructions to implement the method for determining the line space topology map of any of the above embodiments.

[0157] Example 4

[0158] According to another aspect of the present invention, a computer-readable storage medium is also provided, which, when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, enables the electronic device to perform the method for determining the line space topology of any of the above embodiments.

[0159] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0160] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0161] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0162] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0163] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0164] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0165] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for determining a spatial topology map of a railway line, characterized in that, include: Acquire multimodal sensor data, wherein the multimodal sensor data is collected by the sensor system carried by the UAV, including: visual imaging data, spatial depth point cloud data, attitude inertial sequence data, and position coordinate data; Based on the multimodal sensor data, component instance information is determined, wherein the component instance information includes the three-dimensional spatial location of the power component instance, the component category, and the contour mask, and the component category includes wires; Based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data, a global 3D point cloud map is determined. The global 3D point cloud map is obtained based on the global UAV pose, which is obtained based on the pose optimization function and the initial 3D point cloud map. The initial 3D point cloud map is obtained based on the error correction function. The pose optimization function is used to eliminate the cumulative drift error of the UAV pose, and the error correction function is used to correct the measurement error of the sensor. Based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information, the device node representing the energy-saving device and the 3D vector edge representing the conductor are obtained. Based on the device nodes and the three-dimensional vector edges, a topology connection operation is performed to obtain a line space topology map.

2. The method according to claim 1, characterized in that, Based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data, a global 3D point cloud map and a global UAV pose are determined, including: Based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data, an initial 3D point cloud map and a corrected UAV pose are determined. Based on the initial 3D point cloud map and the corrected pose of the UAV, a global pose optimization operation is performed to determine the global 3D point cloud map and the global UAV pose. The global pose optimization operation is used to eliminate cumulative drift error.

3. The method according to claim 2, characterized in that, Based on the initial 3D point cloud map and the corrected pose of the UAV, a global pose optimization operation is performed to determine the global 3D point cloud map and the global UAV pose, including: The visual imaging data is compared with historical keyframes to obtain the comparison results. The historical keyframes are visual imaging frames collected and stored by the UAV in the explored area. If the comparison result indicates that the similarity is higher than a preset similarity threshold, a closed loop is identified, wherein the closed loop indicates that the current spatial location of the UAV is the same spatial location as the historical spatial location corresponding to the historical keyframe. Upon identifying the closed loop, a global pose optimization operation is performed based on the initial 3D point cloud map and the corrected pose of the UAV to determine the global 3D point cloud map and the global UAV pose.

4. The method according to claim 2, characterized in that, Based on the initial 3D point cloud map and the corrected pose of the UAV, a global pose optimization operation is performed to determine the global 3D point cloud map and the global UAV pose, including: Using the UAV poses corresponding to multiple time steps in the UAV pose correction process as pose nodes, and the relative pose constraints between multiple pose nodes as edges, a pose graph is constructed. The pose optimization function is invoked, wherein the pose optimization function aims to minimize the weighted sum of squared relative pose constraint errors between pose nodes. The global UAV pose is obtained based on the poses corresponding to multiple pose nodes in the pose graph and the pose optimization function. Based on the global UAV pose, the initial 3D point cloud map is updated to obtain the global 3D point cloud map.

5. The method according to claim 2, characterized in that, Based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data, an initial 3D point cloud map and a corrected UAV pose are determined, including: Based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data, an initial state vector is constructed. The initial state vector includes the initial pose state of the UAV corresponding to multiple time steps and the initial three-dimensional coordinates of multiple map points. The multiple map points represent points that constitute the surface of the environment in the three-dimensional point cloud map. The error correction function is retrieved, wherein the error correction function includes multiple error terms and weight coefficients corresponding to the multiple error terms respectively. The multiple error terms include visual error terms, lidar error terms, inertial measurement unit error terms and positioning error terms. The weight coefficients corresponding to the multiple error terms are negatively correlated with the measurement uncertainty of the corresponding sensor. Based on the error correction function and the initial state vector, the target state vector is obtained; Based on the UAV pose state quantities corresponding to multiple time steps in the target state vector, the corrected UAV pose is determined, and based on the target 3D coordinates of multiple map points in the target state vector, the initial 3D point cloud map is determined.

6. The method according to claim 1, characterized in that, Based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information, the device nodes representing the energy-saving equipment and the 3D vector edges representing the conductors are obtained, including: Perform point cloud clustering operations based on spatial distance and geometric features on multiple map points in the global 3D point cloud map to obtain multiple device point cloud clusters, wherein the device point cloud cluster represents a set of map points belonging to the same device instance; Determine the device type corresponding to each of the multiple device point cloud clusters; A parametric model fitting operation is performed on the multiple device point cloud clusters to obtain device pose information of multiple device instances. The device pose information includes three-dimensional spatial coordinates and device orientation. The parametric model fitting operation is to fit the device point cloud clusters into a parametric three-dimensional model corresponding to the device type to determine the three-dimensional spatial coordinates and device orientation of the device instance. The plurality of device instances are respectively identified as nodes representing energy-saving devices, and the device attribute information of the plurality of device instances is assigned to the corresponding nodes, wherein the device attribute information includes the device type and the device pose information.

7. The method according to claim 1, characterized in that, Based on the global 3D point cloud map and the component category and contour mask of the power component instance in the component instance information, the node attribute information representing the energy-saving device and the 3D vector edges representing the conductor are obtained, including: Based on the contour mask of the conductor instance in the component instance information and the global 3D point cloud map, the conductor point set is determined, wherein the conductor point set is the set of map points belonging to the conductor in the global 3D point cloud map. Perform a three-dimensional space curve fitting operation on the set of points of the conductor to obtain a continuous three-dimensional vector line corresponding to the conductor, wherein the continuous three-dimensional vector line represents the continuous trajectory of the conductor in three-dimensional space, and the three-dimensional space curve fitting operation includes at least one of the following: catenary equation, polynomial fitting algorithm; The three-dimensional vector lines corresponding to multiple conductors are defined as the three-dimensional vector edges that characterize the conductors.

8. A device for determining a spatial topology map of a railway line, characterized in that, include: The acquisition module is used to acquire multimodal sensor data, wherein the multimodal sensor data is collected by the sensor system carried by the UAV and includes: visual imaging data, spatial depth point cloud data, attitude inertial sequence data and position coordinate data; The first determining module is used to determine component instance information based on the multimodal sensor data, wherein the component instance information includes the three-dimensional spatial location of the power component instance, the component category, and the contour mask, and the component category includes wires; The second determining module is used to determine a global three-dimensional point cloud map based on the visual imaging data, the spatial depth point cloud data, the attitude inertial sequence data, and the position coordinate data. The global three-dimensional point cloud map is obtained based on the global UAV pose, the global UAV pose is obtained based on the pose optimization function and the initial three-dimensional point cloud map, the initial three-dimensional point cloud map is obtained based on the error correction function, the pose optimization function is used to eliminate the cumulative drift error of the UAV pose, and the error correction function is used to correct the measurement error of the sensor. The third determining module is used to obtain the device node representing the energy-saving device and the three-dimensional vector edge representing the conductor based on the global three-dimensional point cloud map and the component category and contour mask of the power component instance in the component instance information. The fourth determining module is used to perform topology connection operations based on the device nodes and the three-dimensional vector edges to obtain a line space topology map.

9. An electronic device, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the method for determining the line space topology as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, When the instructions in the computer-readable storage medium are executed by the processor of the electronic device, the electronic device is able to perform the method for determining the line space topology as described in any one of claims 1 to 7.

Images