Power operation point cloud map construction method and system based on inspection robot

Abstract

The invention relates to the technical field of video survey, in particular to an electric power operation point cloud map construction method and system based on an inspection robot. The method comprises the following steps: acquiring multi-source sensor data of a power operation environment; carrying out synchronous calibration on the obtained multi-source sensor data, and checking the time consistency of the multi-source sensor data; carrying out point cloud registration by using an improved point cloud registration algorithm on the premise of passing the time consistency verification to obtain an optimized weight matrix; performing pose estimation on the inspection robot based on the pose data from the inertial measurement unit, and calculating the pose data of the inspection robot; fusing the optimized weight matrix and the pose data of the inspection robot, and carrying out filtering updating on the fused point cloud matrix by adopting a filtering algorithm; and constructing a power operation point cloud map of the power operation environment based on the filtered and updated point cloud matrix. Accurate three-dimensional visualization support can be provided for electric power operation, and the operation efficiency and safety are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of video measurement, and in particular to a method and system for constructing a point cloud map of an electric power operation based on an inspection robot. Background Art

[0002] As the power system becomes more complex and its scale continues to expand, the stable operation of power equipment has become particularly important. Traditional power inspections mainly rely on manual inspections, which are not only time-consuming and labor-intensive, but also have problems such as high human resource costs, low inspection efficiency, and high risks. In order to improve inspection efficiency and safety, inspection robots have emerged. Through autonomous navigation and intelligent sensing technology, inspection robots can accurately locate, monitor the status, and detect faults of power equipment, greatly reducing the workload and risks of manual inspections.

[0003] In power operations, building an accurate point cloud map is the basis for efficient inspection robot operations. Point cloud maps can provide robots with accurate environmental information to help them complete navigation, obstacle detection and operation planning. Based on point cloud maps, inspection robots can accurately locate the position of power equipment and achieve efficient automatic inspection operations.

[0004] The power operation environment is complex and changeable, including a large number of power equipment, lines and complex terrain. Traditional point cloud construction methods are difficult to meet the needs of these specific environments. The precise positioning and identification of power equipment requires point cloud maps to have high resolution and accuracy, which puts higher requirements on existing point cloud processing algorithms. The real-time and safety of power inspection tasks require that the construction and update speed of point cloud maps be fast enough to support the real-time operation of inspection robots.

[0005] Although the existing technology has made some progress in point cloud map construction, there are still many technical problems that need to be solved in the power operation scene, such as the acquisition and processing efficiency of the point cloud matrix, environmental adaptability, equipment positioning accuracy, etc. Therefore, in view of the specific needs of power operations, the development of a power operation point cloud map construction method and system based on inspection robots has important practical significance and application value.

[0006] The technical routes of existing technologies mainly include: 1. SLAM technology based on LiDAR: Scan the environment with LiDAR sensors to obtain a three-dimensional point cloud matrix, and use the SLAM (Simultaneous Localization and Mapping) algorithm to simultaneously perform robot positioning and environmental map construction. LiDAR SLAM is widely used in unmanned driving and robot navigation.

[0007] Existing problems: For complex power operation environments, the resolution and scanning accuracy of LiDAR may be insufficient, especially in high-altitude power equipment and dense line environments, which are prone to data loss or reduced accuracy. The SLAM algorithm has high real-time requirements in power operations, but in complex environments, the computing efficiency may not be sufficient, affecting the efficiency of real-time navigation and operations.

[0008] 2. 3D reconstruction technology based on depth camera: Use depth camera to capture the depth information of the environment, and generate point cloud map through 3D reconstruction algorithm for robot navigation and power equipment identification. Depth camera is usually combined with RGB camera to provide richer environmental information.

[0009] Problems: Depth cameras are unstable in strong light, low light or bad weather conditions, which may cause the quality of the point cloud matrix to deteriorate and affect the accuracy of map construction. When dealing with complex power equipment, 3D reconstruction algorithms are easily affected by occlusion and reflection, resulting in errors in the reconstruction results, affecting the precise positioning and identification of the equipment.

[0010] 3. Point cloud map construction based on multi-sensor fusion: Comprehensive use of multiple sensors such as lidar, depth camera, inertial measurement unit (IMU), etc., and improve the accuracy and robustness of point cloud maps through data fusion technology. Sensor fusion technology can complement each other in different environments and enhance the map construction effect.

[0011] Existing problems: The data synchronization and fusion algorithms of multiple sensors are complex, and are prone to timing errors and data inconsistencies, which affect the accuracy of point cloud maps. The diversity of sensors increases the hardware cost and computational complexity of the system, especially in scenarios where power inspection robots need to operate stably for a long time, which makes maintenance and debugging more difficult. Summary of the invention

[0012] To achieve the purpose of the present invention, the present invention is implemented by the following technical solutions:

[0013] A method for constructing a point cloud map of electric power operation based on an inspection robot, wherein the point cloud map is formed in the form of a point cloud matrix, and the method comprises the following steps:

[0014] S1. Acquire multi-source sensor data of an electric power operation environment, wherein the multi-source sensor data includes a basic point cloud matrix from a laser radar, a registration point cloud matrix from a depth camera, and position and posture data from an inertial measurement unit, wherein the position and posture data includes position data and attitude data;

[0015] S2. Synchronize and calibrate the acquired multi-source sensor data to verify the time consistency of the multi-source sensor data;

[0016] S3. On the premise that the temporal consistency check is passed, the basic point cloud matrix from the lidar and the registration point cloud matrix from the depth camera are registered using the improved point cloud registration algorithm to obtain an optimized weight matrix.

[0017] S4, using a Kalman filter algorithm, estimating the posture of the inspection robot based on the posture data from the inertial measurement unit, and calculating the posture data of the inspection robot;

[0018] S5, fusing the optimized weight matrix in step S3 and step S4 with the posture data of the inspection robot, and filtering and updating the fused point cloud matrix using a filtering algorithm;

[0019] S6. Construct a power operation point cloud map of the power operation environment based on the filtered and updated point cloud matrix.

[0020] Furthermore, the improved point cloud registration algorithm includes:

[0021] S31, selecting the nearest points in the basic point cloud matrix and the registration point cloud matrix, and calculating the weight of each point, wherein the weight is calculated by the following formula: ,

[0022] in, represents the weight of the i-th point; is the dynamic adjustment parameter of the i-th point, with an initial value of 1; represents the i-th point in the basic point cloud matrix, represents the i-th point in the registration point cloud matrix with the same sequence number as that in the base point cloud matrix, represents the natural exponential function, point Coordinate parameters and points The coordinate parameters satisfy: ,

[0023] Among them, the basic point cloud matrix and the registration point cloud matrix are both expressed based on the OXYZ orthogonal coordinate system. Indicate point The X-axis coordinate, Indicate point The Y-axis coordinate of Indicate point The Z-axis coordinate of Indicate point The X-axis coordinate, Indicate point The X-axis, Indicate point The Z-axis coordinate of

[0024] Indicate point and Point The Euclidean distance between them satisfies: ,

[0025] S32. According to the weight of each point, the initial weight matrix is ​​iteratively optimized using the weighted least squares method to obtain an optimized weight matrix that satisfies: T * = angry MY T [ f T ] f T = ∑ i =1 m ω i ∙ T∙ p i - q i 2 ,

[0026] in, represents the optimization weight matrix; represents the initial weight matrix; the initial weight matrix and the optimized weight matrix Each element of represents the weight value of the corresponding point; the initial value of all elements of the initial weight matrix is ​​1, and is updated after each iterative optimization;

[0027] Represents the optimized weight matrix The maximum number of points in ; represents an iterative optimization function; angry MY T [] represents the minimum value; angry MY T [ f T ] Represents the iterative optimization function Iteratively calculate the optimal value of the weight that makes it minimum as the optimal weight value , by all the optimal weight values Composition optimization weight matrix .

[0028] Furthermore, in step S31, the dynamic adjustment parameter of the i-th point According to the density and dynamic adjustment of the basic point cloud matrix and the registration point cloud matrix, including: for the i-th point, adjust the dynamic adjustment parameters of the i-th point according to the distribution density of the i-th point's neighborhood points ,satisfy: ,

[0029] in, represents the i-th point in the basic point cloud matrix, represents the i-th point in the registration point cloud matrix with the same sequence number as that in the base point cloud matrix, represents the jth point in the basic point cloud matrix, represents the jth point in the registration point cloud matrix that has the same sequence number as the base point cloud matrix. and Respectively represent the Euclidean distance between the jth point and the ith point in the basic point cloud matrix and the registration point cloud matrix, The maximum value of the points in common between the base point cloud matrix and the registered point cloud matrix.

[0030] Furthermore, in step S31, selecting the nearest points in the basic point cloud matrix and the registration point cloud matrix includes:

[0031] S311, aligning the coordinate origins of the basic point cloud matrix and the registration point cloud matrix, and making the basic point cloud matrix and the registration point cloud matrix overlap;

[0032] S312, for the points in the basic point cloud matrix , select the nearest point from the registration point cloud matrix ,satisfy: ,

[0033] in, Indicates that the point cloud matrix is ​​selected so that The point with the smallest value , as the point in the registration point cloud matrix and the base point cloud matrix The nearest point .

[0034] Furthermore, after performing point cloud registration on the basic point cloud matrix from the laser radar and the registration point cloud matrix from the depth camera using the improved point cloud registration algorithm, step S3 further includes:

[0035] S33, performing splicing processing on the registered multi-frame point cloud matrices, and integrating the multi-frame point cloud matrices into one by aligning the transformation matrix;

[0036] S34. During the integration process, based on the calculated weights, the density of the registered multi-frame point cloud matrix is ​​adjusted so that the basic point cloud matrix from the lidar and the registered point cloud matrix from the depth camera maintain a consistent density in different regions;

[0037] S35. Through iterative optimization, the integrated point cloud matrix is ​​globally adjusted based on the optimized weight matrix, so that the overall accuracy and consistency of the point cloud map reaches a preset threshold.

[0038] Further, step S4 includes:

[0039] S41, data preprocessing: filtering the position and posture data of the inertial measurement unit to obtain filtered position and posture data of the inertial measurement unit;

[0040] S42, state prediction: predicting the current posture data of the inspection robot based on the historical posture data of the inertial measurement unit;

[0041] S43, Kalman gain calculation: combining the filtered posture data of the inertial measurement unit with the predicted current posture data of the inspection robot, calculating the Kalman gain to determine the weight relationship between the filtered posture data of the inertial measurement unit and the predicted current posture data of the inspection robot;

[0042] S44, state update: according to the Kalman gain, the predicted current posture data of the inspection robot is corrected to obtain the posture data of the inspection robot.

[0043] Furthermore, step S4 further includes:

[0044] S45, covariance matrix update: after obtaining the posture data of the inspection robot, the covariance matrix is ​​updated as the basis for the next posture estimation.

[0045] The present invention also provides a system for constructing a point cloud map of electric power operations based on an inspection robot, which is used to execute the method for constructing a point cloud map of electric power operations based on an inspection robot. The system comprises: an inspection robot module: equipped with a variety of sensors, including a laser radar, a depth camera, and an inertial measurement unit, which is used to collect the point cloud matrix of the electric power operation environment and the posture data of the inspection robot; a data preprocessing module: connected to the inspection robot module, used to filter and time-synchronize the collected point cloud matrix and the posture data of the inertial inspection robot, and verify the stability and consistency of the point cloud matrix and the posture data; a posture estimation module: based on the Kalman filtering algorithm, the posture of the robot is estimated using the data of the inertial measurement unit. Generate the current posture data of the robot; Point cloud registration module: Based on the posture data generated by the posture estimation module, perform the point cloud matrix registration operation, and align and integrate the multi-frame point cloud matrices into one; Global optimization module: Perform global optimization processing on the registered point cloud matrix, adjust the density of the registered multi-frame point cloud matrix, so that the basic point cloud matrix from the lidar and the registered point cloud matrix from the depth camera maintain consistent density in different areas; Refined processing module: Perform refined processing on the optimized point cloud matrix, remove redundant data and optimize the structure of the point cloud matrix to generate a high-precision point cloud map; Quality verification and calibration module: Perform quality verification and calibration on the generated point cloud map to ensure the accuracy and practicality of the point cloud map.

[0046] Furthermore, the system also includes: a partition management module: after the point cloud map is generated, the point cloud matrix is ​​partitioned and managed, and the map is divided into several areas to improve computing efficiency and data management capabilities; a geometric surface and texture generation module: for the point cloud matrix of each area, the corresponding geometric surface and texture information is generated, and texture mapping processing is performed to enhance the visualization effect of the map; a three-dimensional visualization display module: the final generated point cloud map is displayed on a three-dimensional visualization platform, providing a high-precision three-dimensional view of the power operation environment for the maintenance and management of power facilities.

[0047] Beneficial Effects

[0048] The method and system for constructing a point cloud map of power operations based on an inspection robot of the present invention realizes the construction of a high-precision three-dimensional point cloud map of the power operation environment by integrating multiple sensors and advanced data processing technologies. Specific technical effects include: 1. High-precision map generation: Through the comprehensive use of sensors such as lidar, depth camera and IMU, combined with improved point cloud registration algorithm and Kalman filter attitude estimation, high-precision three-dimensional point cloud map of the power operation environment can be generated, effectively improving the spatial resolution and detail performance of the map.

[0049] 2. Data consistency and optimization: The system ensures the consistency and accuracy of the multi-source point cloud matrix by synchronously calibrating and globally optimizing the collected data. Dynamic weight adjustment and weighted least squares optimization further improve the accuracy of point cloud registration and the overall quality of the map.

[0050] 3. Real-time and adaptability: The application of Kalman filter algorithm in attitude estimation enables the system to process IMU data in real time, respond quickly to environmental changes, and adapt to complex situations in dynamic power operation environments.

[0051] 4. Enhanced map visualization: Through refined processing and three-dimensional visualization display, the system not only provides accurate map information, but also improves the visual effect of the power operation environment, enabling maintenance and management personnel to understand and operate more intuitively.

[0052] 5. Efficient data management: The system's partition management and geometric surface generation modules improve data processing efficiency, optimize the map's regional management and visual effects, and facilitate subsequent data analysis and utilization.

[0053] In summary, the method and system of the present invention significantly improve the accuracy, efficiency and practicality of constructing point cloud maps for power operations, and provide effective support for the maintenance and management of power facilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0055] In order to deepen the understanding of the present invention, the present invention will be further described in detail below in conjunction with examples. The examples are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention. Specific embodiment 1

[0057] according to Figure 1 As shown, this embodiment provides a method for constructing a point cloud map of electric power operations based on an inspection robot, wherein the point cloud map is stored in the form of a point cloud matrix, and the method comprises the following steps:

[0058] S1. Acquire multi-source sensor data of an electric power operation environment, wherein the multi-source sensor data includes a basic point cloud matrix from a laser radar, a registration point cloud matrix from a depth camera, and position and posture data from an inertial measurement unit, wherein the position and posture data includes position data and attitude data;

[0059] S2. Synchronize and calibrate the acquired multi-source sensor data to verify the time consistency of the multi-source sensor data;

[0060] S3. On the premise that the temporal consistency check is passed, the basic point cloud matrix from the lidar and the registration point cloud matrix from the depth camera are registered using the improved point cloud registration algorithm to obtain an optimized weight matrix.

[0061] S4, using a Kalman filter algorithm, estimating the posture of the inspection robot based on the posture data from the inertial measurement unit, and calculating the posture data of the inspection robot;

[0062] S5, fusing the optimized weight matrix in step S3 and step S4 with the posture data of the inspection robot, and filtering and updating the fused point cloud matrix using a filtering algorithm;

[0063] S6. Construct a power operation point cloud map of the power operation environment based on the filtered and updated point cloud matrix.

[0064] Furthermore, the improved point cloud registration algorithm includes:

[0065] S31, selecting the nearest points in the basic point cloud matrix and the registration point cloud matrix, and calculating the weight of each point, wherein the weight is calculated by the following formula: ,

[0066] in, represents the weight of the i-th point; is the dynamic adjustment parameter of the i-th point, with an initial value of 1; represents the i-th point in the basic point cloud matrix, represents the i-th point in the registration point cloud matrix with the same sequence number as that in the base point cloud matrix, represents the natural exponential function, point Coordinate parameters and points The coordinate parameters satisfy: ,

[0067] Among them, the basic point cloud matrix and the registration point cloud matrix are both expressed based on the OXYZ orthogonal coordinate system. Indicate point The X-axis coordinate, Indicate point The Y-axis coordinate of Indicate point The Z-axis coordinate of Indicate point The X-axis coordinate, Indicate point The X-axis, Indicate point The Z-axis coordinate of

[0068] Indicate point and Point The Euclidean distance between them satisfies: ,

[0069] S32. According to the weight of each point, the initial weight matrix is ​​iteratively optimized using the weighted least squares method to obtain an optimized weight matrix that satisfies: T * = angry MY T [ f T ] f T = ∑ i =1 m ω i ∙ T∙ p i - q i 2 ,

[0070] in, represents the optimization weight matrix; represents the initial weight matrix; the initial weight matrix and the optimized weight matrix Each element of represents the weight value of the corresponding point; the initial value of all elements of the initial weight matrix is ​​1, and is updated after each iterative optimization;

[0071] Represents the optimized weight matrix The maximum number of points in ; represents an iterative optimization function; angry MY T [] represents the minimum value; angry MY T [ f T ] Represents the iterative optimization function Iteratively calculate the optimal value of the weight that makes it minimum as the optimal weight value , by all the optimal weight values Composition optimization weight matrix .

[0072] Furthermore, in step S31, the dynamic adjustment parameter of the i-th point According to the density and dynamic adjustment of the basic point cloud matrix and the registration point cloud matrix, including: for the i-th point, adjust the dynamic adjustment parameters of the i-th point according to the distribution density of the i-th point's neighborhood points ,satisfy: ,

[0073] in, represents the i-th point in the basic point cloud matrix, represents the i-th point in the registration point cloud matrix with the same sequence number as that in the base point cloud matrix, represents the jth point in the basic point cloud matrix, represents the jth point in the registration point cloud matrix that has the same sequence number as the base point cloud matrix. and Respectively represent the Euclidean distance between the jth point and the ith point in the basic point cloud matrix and the registration point cloud matrix, The maximum value of the points in common between the base point cloud matrix and the registered point cloud matrix.

[0074] Furthermore, in step S31, selecting the nearest points in the basic point cloud matrix and the registration point cloud matrix includes:

[0075] S311, aligning the coordinate origins of the basic point cloud matrix and the registration point cloud matrix, and making the basic point cloud matrix and the registration point cloud matrix overlap;

[0076] S312, for the points in the basic point cloud matrix , select the nearest point from the registration point cloud matrix ,satisfy: ,

[0077] in, Indicates that the point cloud matrix is ​​selected so that The point with the smallest value , as the point in the registration point cloud matrix and the base point cloud matrix The nearest point .

[0078] Furthermore, after performing point cloud registration on the basic point cloud matrix from the laser radar and the registration point cloud matrix from the depth camera using the improved point cloud registration algorithm, step S3 further includes:

[0079] S33, performing splicing processing on the registered multi-frame point cloud matrices, and integrating the multi-frame point cloud matrices into one by aligning the transformation matrix;

[0080] S34. During the integration process, based on the calculated weights, the density of the registered multi-frame point cloud matrix is ​​adjusted so that the basic point cloud matrix from the lidar and the registered point cloud matrix from the depth camera maintain a consistent density in different regions;

[0081] S35. Through iterative optimization, the integrated point cloud matrix is ​​globally adjusted based on the optimized weight matrix, so that the overall accuracy and consistency of the point cloud map reaches a preset threshold.

[0082] Further, step S4 includes:

[0083] S41, data preprocessing: filtering the position and posture data of the inertial measurement unit to obtain filtered position and posture data of the inertial measurement unit;

[0084] S42, state prediction: predicting the current posture data of the inspection robot based on the historical posture data of the inertial measurement unit;

[0085] S43, Kalman gain calculation: combining the filtered posture data of the inertial measurement unit with the predicted current posture data of the inspection robot, calculating the Kalman gain to determine the weight relationship between the filtered posture data of the inertial measurement unit and the predicted current posture data of the inspection robot;

[0086] S44, state update: according to the Kalman gain, the predicted current posture data of the inspection robot is corrected to obtain the posture data of the inspection robot.

[0087] Furthermore, step S4 further includes:

[0088] S45, covariance matrix update: after obtaining the posture data of the inspection robot, the covariance matrix is ​​updated as the basis for the next posture estimation. Specific embodiment 2

[0090] The present embodiment provides a system for constructing a point cloud map of electric power operation based on an inspection robot, which is used to execute the method for constructing a point cloud map of electric power operation based on an inspection robot. The system comprises: an inspection robot module: equipped with a variety of sensors, including a laser radar, a depth camera, and an inertial measurement unit, which is used to collect the point cloud matrix of the electric power operation environment and the posture data of the inspection robot; a data preprocessing module: connected to the inspection robot module, used to filter and time-synchronize the collected point cloud matrix and the posture data of the inertial inspection robot, and verify the stability and consistency of the point cloud matrix and the posture data; a posture estimation module: based on the Kalman filtering algorithm, using the inertial measurement unit to obtain ... The robot's posture is estimated based on the data of the unit, and the robot's current posture data is generated; point cloud registration module: based on the posture data generated by the posture estimation module, it performs the point cloud matrix registration operation, and aligns and integrates the multi-frame point cloud matrix into one; global optimization module: performs global optimization processing on the registered point cloud matrix to improve the consistency and accuracy of the point cloud matrix and the inspection and lover's posture data; fine processing module: performs fine processing on the optimized point cloud matrix, removes redundant data and optimizes the structure of the point cloud matrix to generate a high-precision point cloud map; quality verification and calibration module: performs quality verification and calibration on the generated point cloud map to ensure the accuracy and practicality of the point cloud map.

[0091] Furthermore, the system also includes:

[0092] Partition management module: After the point cloud map is generated, the point cloud matrix is ​​partitioned and managed, and the map is divided into several areas to improve computing efficiency and data management capabilities; geometric surface and texture generation module: for the point cloud matrix of each area, the corresponding geometric surface and texture information is generated, and texture mapping processing is performed to enhance the visualization effect of the map; three-dimensional visualization display module: the final point cloud map is displayed on a three-dimensional visualization platform, providing a high-precision three-dimensional view of the power operation environment for the maintenance and management of power facilities.

[0093] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for constructing a point cloud map of electric power operations based on an inspection robot, characterized in that: The electric power operation point cloud map is formed in the form of a point cloud matrix, and the method comprises the following steps: S1. Acquire multi-source sensor data of an electric power operation environment, wherein the multi-source sensor data includes a basic point cloud matrix from a laser radar, a registration point cloud matrix from a depth camera, and position and posture data from an inertial measurement unit, wherein the position and posture data includes position data and attitude data; S2. Synchronize and calibrate the acquired multi-source sensor data to verify the time consistency of the multi-source sensor data; S3. On the premise that the temporal consistency check is passed, the basic point cloud matrix from the lidar and the registration point cloud matrix from the depth camera are registered using the improved point cloud registration algorithm to obtain an optimized weight matrix. S4, using a Kalman filter algorithm, estimating the posture of the inspection robot based on the posture data from the inertial measurement unit, and calculating the posture data of the inspection robot; S5, fusing the optimized weight matrix in step S3 and step S4 with the posture data of the inspection robot, and filtering and updating the fused point cloud matrix using a filtering algorithm; S6. Construct a power operation point cloud map of the power operation environment based on the filtered and updated point cloud matrix.

2. The method for constructing a point cloud map of electric power operations based on an inspection robot according to claim 1, characterized in that: In step S3, the improved point cloud registration algorithm includes: S31, selecting the nearest points in the basic point cloud matrix and the registration point cloud matrix, and calculating the weight of each point, wherein the weight is calculated by the following formula: ， in, represents the weight of the i-th point; is the dynamic adjustment parameter of the i-th point, with an initial value of 1; represents the i-th point in the basic point cloud matrix, represents the i-th point in the registration point cloud matrix with the same sequence number as that in the base point cloud matrix, represents the natural exponential function, point Coordinate parameters and points The coordinate parameters satisfy: ， Among them, the basic point cloud matrix and the registration point cloud matrix are both expressed based on the OXYZ orthogonal coordinate system. Indicate point The X-axis coordinate, Indicate point The Y-axis coordinate of Indicate point The Z-axis coordinate of Indicate point The X-axis coordinate, Indicate point The X-axis, Indicate point The Z-axis coordinate of Indicate point and Point The Euclidean distance between them satisfies: ， S32. According to the weight of each point, the initial weight matrix is ​​iteratively optimized using the weighted least squares method to obtain an optimized weight matrix that satisfies: ， in, represents the optimization weight matrix, represents the initial weight matrix, the initial weight matrix and the optimized weight matrix Each element of represents the weight value of the corresponding point, and the initial value of all elements of the initial weight matrix is ​​1, which is updated after each iterative optimization; Represents the optimized weight matrix The maximum number of points in represents an iterative optimization function; represents the minimum value, Represents the iterative optimization function Iteratively calculate the optimal value of the weight that makes it minimum as the optimal weight value , by all the optimal weight values Composition optimization weight matrix .

3. The method for constructing a point cloud map of electric power operations based on an inspection robot according to claim 2, characterized in that: In step S31, the dynamic adjustment parameter of the i-th point According to the density and dynamic adjustment of the basic point cloud matrix and the registration point cloud matrix, including: for the i-th point, adjust the dynamic adjustment parameters of the i-th point according to the distribution density of the i-th point's neighborhood points ,satisfy: ， in, represents the i-th point in the basic point cloud matrix, represents the i-th point in the registration point cloud matrix with the same sequence number as that in the base point cloud matrix, represents the jth point in the basic point cloud matrix, represents the jth point in the registration point cloud matrix that has the same sequence number as the base point cloud matrix. and Respectively represent the Euclidean distance between the jth point and the ith point in the basic point cloud matrix and the registration point cloud matrix, The maximum value of the points in common between the base point cloud matrix and the registered point cloud matrix.

4. The method for constructing a point cloud map of electric power operations based on an inspection robot according to claim 2, characterized in that: In step S31, the nearest points in the basic point cloud matrix and the registration point cloud matrix are selected, including: S311, aligning the coordinate origins of the basic point cloud matrix and the registration point cloud matrix, and making the basic point cloud matrix and the registration point cloud matrix overlap; S312, for the points in the basic point cloud matrix , select the nearest point from the registration point cloud matrix ,satisfy: ， in, Indicates that the point cloud matrix is ​​selected so that The point with the smallest value , as the point in the registration point cloud matrix and the base point cloud matrix The nearest point .

5. The method for constructing a point cloud map of electric power operations based on an inspection robot according to claim 2, characterized in that: After performing point cloud registration on the basic point cloud matrix from the laser radar and the registration point cloud matrix from the depth camera using the improved point cloud registration algorithm, step S3 further includes: S33, performing splicing processing on the registered multi-frame point cloud matrices, and integrating the multi-frame point cloud matrices into one by aligning the transformation matrix; S34. During the integration process, based on the calculated weights, the density of the registered multi-frame point cloud matrix is ​​adjusted so that the basic point cloud matrix from the lidar and the registered point cloud matrix from the depth camera maintain a consistent density in different regions; S35. Through iterative optimization, the integrated point cloud matrix is ​​globally adjusted based on the optimized weight matrix, so that the overall accuracy and consistency of the power operation point cloud map reaches a preset threshold.

6. The method for constructing a point cloud map of electric power operations based on an inspection robot according to claim 1, characterized in that: Step S4 includes: S41, data preprocessing: filtering the position and posture data of the inertial measurement unit to obtain filtered position and posture data of the inertial measurement unit; S42, state prediction: predicting the current posture data of the inspection robot based on the historical posture data of the inertial measurement unit; S43, Kalman gain calculation: combining the filtered posture data of the inertial measurement unit with the predicted current posture data of the inspection robot, calculating the Kalman gain to determine the weight relationship between the filtered posture data of the inertial measurement unit and the predicted current posture data of the inspection robot; S44, state update: according to the Kalman gain, the predicted current posture data of the inspection robot is corrected to obtain the posture data of the inspection robot.

7. The method for constructing a point cloud map of electric power operations based on an inspection robot according to claim 6, characterized in that: Step S4 also includes: S45, covariance matrix update: after obtaining the posture data of the inspection robot, the covariance matrix is ​​updated as the basis for the next posture estimation.

8. A system for constructing a point cloud map of electric power operations based on an inspection robot, used to execute a method for constructing a point cloud map of electric power operations based on an inspection robot as claimed in any one of claims 1 to 7, characterized in that: The system comprises: Inspection robot module: equipped with laser radar, depth camera, and inertial measurement unit, used to collect multi-source sensor data of the power operation environment. The multi-source sensor data includes the basic point cloud matrix from the laser radar, the registration point cloud matrix from the depth camera, and the posture data from the inertial measurement unit. The posture data includes position data and attitude data Data preprocessing module: connected to the inspection robot module, used to filter and time-synchronize the collected basic point cloud matrix, registration point cloud matrix and posture data from the inertial measurement unit, and verify the time consistency of multi-source sensor data; Posture estimation module: Use the Kalman filter algorithm to estimate the posture of the inspection robot based on the posture data from the inertial measurement unit and calculate the posture data of the inspection robot; Point cloud registration module: On the premise that the temporal consistency check is passed, the basic point cloud matrix from the lidar and the registration point cloud matrix from the depth camera are registered using the improved point cloud registration algorithm to obtain the optimized weight matrix; Global optimization module: performs global optimization processing on the registered point cloud matrix, adjusts the density of the registered multi-frame point cloud matrix, and makes the basic point cloud matrix from the lidar and the registered point cloud matrix from the depth camera maintain the same density in different areas; Refined processing module: Constructs the power operation point cloud map of the power operation environment based on the filtered and updated point cloud matrix.

9. The power operation point cloud map construction system based on the inspection robot according to claim 8 is characterized in that: The system further comprises: Partition management module: After the power operation point cloud map is generated, the point cloud matrix is ​​partitioned and divided into several areas to improve computing efficiency and data management capabilities; Geometric surface and texture generation module: Generates the corresponding geometric surface and texture information for the point cloud matrix of each area, and performs texture mapping processing to enhance the visualization effect of the point cloud map of the power operation; 3D visualization display module: The final generated power operation point cloud map is displayed on a 3D visualization platform, providing a high-precision 3D view of the power operation environment for maintenance and management of power facilities.

Images