Gas insulated combined switch multi-light inspection fusion data detection method and device and medium
By using a multi-light inspection fusion data detection method, which leverages the features of visible light, infrared images, and laser point clouds to drive step-by-step registration and adaptive weighted fusion, the problem of registration accuracy and efficiency in GIS inspection is solved, and high-precision equipment status assessment is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- STATE GRID SHANGHAI MUNICIPAL ELECTRIC POWER CO
- Filing Date
- 2026-02-28
- Publication Date
- 2026-07-14
AI Technical Summary
In the current technology for gas-insulated switchgear (GIS) inspection, the infrared image resolution is not high, the visible light image lacks depth information, and the pixel information is not accurately associated with the equipment components, resulting in low detection accuracy and failing to meet the registration accuracy requirements of component-level geometric dimensions.
A multi-light inspection data fusion method using visible light images, infrared images, and laser point clouds is adopted. Through feature-driven step-by-step registration strategy and adaptive weighted fusion algorithm, the registration accuracy and efficiency are improved by utilizing GIS equipment structure and thermal feature information, and the deep fusion of the three types of data is achieved.
It improves the registration accuracy and efficiency of GIS inspection, provides rich and reliable multi-dimensional information, meets the needs of GIS equipment status assessment, and enhances the accuracy and reliability of inspection.
Smart Images
Figure CN122391056A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of gas-insulated switchgear testing, and in particular to a method, apparatus, and medium for detecting multi-optical inspection fusion data of gas-insulated switchgear. Background Technology
[0002] Gas-insulated switchgear (GIS) plays a crucial role in power systems, but due to its complex structure and harsh operating environment, it is prone to faults such as partial discharge. Timely detection and assessment are essential for ensuring power grid stability. Currently, GIS inspection mostly relies on single visible light or infrared images. The former can intuitively present the appearance and provide detailed visual information about the equipment's appearance, such as loose screws and corrosion of components, but it is not sensitive to internal thermal defects. The latter can detect thermal features, but it is greatly affected by environmental interference, and its positioning accuracy and imaging quality are limited, affecting intuitiveness. In response, some existing technologies, such as Chinese patent CN120278890A, provide a solution for the inspection of electrical equipment by fusing visible light and infrared images.
[0003] However, the above method has significant drawbacks when applied to the detection of gas-insulated combination switches. On the one hand, the resolution of infrared images is not high, and visible light images lack depth information. If the pixel information and the components of the device are accurately correlated, the pixel alignment accuracy is not high, resulting in low prediction accuracy.
[0004] While some existing technologies, such as Chinese patent CN116258744A, disclose target tracking methods based on the fusion of visible light, infrared, and lidar data, their technical approach involves: first, registering and fusing infrared and visible light images to obtain an enhanced two-dimensional image; simultaneously, lidar independently performs three-dimensional target detection. Finally, the three-dimensional point cloud target is projected onto a unified two-dimensional coordinate system and correlated with the two-dimensional target on the fused image to complete tracking. This invention uses a two-dimensional image as a reference, leveraging the advantages of visible light, infrared, and lidar, fusing information from several heterogeneous sensors, utilizing the complementarity and redundancy of sensor performance, expanding the information acquisition range of the sensor group, improving target detection decision confidence, reducing ambiguity, enhancing reliability, fault tolerance, and system anti-interference capabilities, and improving target tracking accuracy. However, its focus is on achieving accurate detection and identification of equipment or observed objects and their movement trajectories, and it cannot meet the requirements for registration accuracy at the component-level geometric dimensions. Summary of the Invention
[0005] The purpose of this invention is to address the aforementioned shortcomings by providing a method, apparatus, and medium for detecting multi-light inspection fusion data of gas-insulated combination switches.
[0006] The objective of this invention can be achieved through the following technical solutions: A detection method for multi-light inspection fusion data of gas-insulated switchgear includes: Step S1: Acquire visible light image, infrared image and laser point cloud of the gas-insulated combination switch under test; Step S2: Register the visible light image and the laser point cloud; Step S3: Register the infrared image and the laser point cloud; Step S4: Obtain the visible light pixel value, infrared pixel value, and point cloud attribute value of each target location. Query the weight values of the visible light pixel value, infrared pixel value, and point cloud attribute value according to the task category, and perform weighted fusion to obtain the fused attribute value. Step S5: Construct a feature vector based on the fused attribute values and input it into the trained detection model to obtain the detection result.
[0007] During the acquisition of visible light images, infrared images, and laser point clouds, the target device's position and orientation remain consistent.
[0008] Step S1 further includes: Preprocessing of visible light and infrared light images, including radiometric correction and grayscale transformation, removes noise and interference. Filtering and downsampling of laser point cloud data removes redundant points, improving data quality and fusion efficiency.
[0009] Step S2 includes: Step S2-1: Extract the geometric feature points of the set features on the visible light image. The set features include the flange edge and the outline of the basin insulator. Step S2-2: Establish sparse corresponding points between the visible light image and the laser point cloud data, and apply the point-to-point ICP algorithm based on KD tree acceleration for registration. In the initial registration process, the maximum number of iterations is set to 100, the distance threshold is twice the average density of the point cloud, and the root mean square error change is less than 0.001 mm as the convergence condition.
[0010] Step S3 includes: Step S3-1: Extract the defined significant thermal region feature points based on the temperature gradient from the infrared image; Step S3-2: According to the thermal field constraints, project each thermal feature point onto the laser point cloud; Step S3-3: Construct a joint descriptor using the feature points of the hot region and the three-dimensional geometric features near their projection points on the point cloud, wherein the three-dimensional geometric features include normal direction and curvature; Step S3-4: Match the joint descriptor with the texture feature descriptor of the corresponding region on the visible light image, and use the RANSAC algorithm to eliminate false matches.
[0011] In step S3-2, when registering the infrared image and the laser point cloud, the physical relationship between thermal distribution and geometric structure is introduced as a constraint condition, wherein; At the flange connection, the surface temperature distribution is uniform. For a conductor, its temperature distribution is related to the current load and heat dissipation conditions.
[0012] The point cloud attribute value is reflection intensity.
[0013] When the task is to detect thermal defects caused by partial discharge and the region is a conductor, the weight of the visible light pixel value is 0.6, the weight of the infrared pixel value is 0.3, and the weight of the point cloud attribute value is 0.1. When the task is to locate the discharge position, the weight of the visible light pixel value is 0.7, the weight of the infrared pixel value is 0.2, and the weight of the point cloud attribute value is 0.1.
[0014] A detection device for multi-light inspection fusion data of gas-insulated combination switches includes a memory, a processor, and a program stored in the memory. When the processor executes the program, it implements the method described above.
[0015] A storage medium having a program stored thereon, which, when executed, implements the method described above.
[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. This application proposes a feature-driven step-by-step registration strategy for GIS, which makes full use of the structural and thermal feature information of GIS equipment to improve registration accuracy and efficiency, solves the problem of high-precision geometric registration of visible light, infrared light and laser point cloud data in complex power equipment scenarios, and provides a key guarantee for the accuracy of subsequent data fusion.
[0017] 2. Design an adaptive weighted fusion algorithm based on GIS inspection needs. Dynamically adjust data weights according to different inspection locations and targets to achieve deep fusion of the three types of data. This enables the fused data to achieve an optimized balance in reflecting the appearance, thermal state, and three-dimensional geometric structure of GIS, quickly responding to the multi-dimensional inspection needs of GIS and providing rich and reliable information for GIS equipment status assessment. Attached Figure Description
[0018] Figure 1 This is a schematic diagram of the structure of the method of the present invention. Detailed Implementation
[0019] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.
[0020] A detection method for multi-optical inspection fusion data of gas-insulated switchgear, such as Figure 1 As shown, it includes: Step S1: Acquire visible light image, infrared image and laser point cloud of the gas-insulated combination switch under test; Visible light data is acquired using a visible light camera, infrared data using an infrared thermal imager, and laser point cloud data using a laser scanner. During the acquisition of visible light images, infrared images, and laser point clouds, the target device's position and orientation must remain consistent. In some cases, it is required that the viewpoints of the visible light and infrared images be completely identical. Consistency is achieved through the following methods: The data acquisition module rigidly mounts the visible light camera, infrared thermal imager, and laser scanner onto the device using any fixing device, ensuring that the relative positions and orientations of the three devices relative to the target device remain unchanged during the acquisition process. In addition, in other embodiments, consistency is also achieved by the following method: the data acquisition module is equipped with a spatial positioning device, such as a laser tracker, a total station or a high-precision visual positioning system, which simultaneously records the precise spatial position and attitude parameters of each sensor when acquiring visible light images, infrared images and laser point cloud data, for coordinate system alignment in the subsequent registration stage.
[0021] Step S1 also includes: Preprocessing of visible light and infrared light images, including radiometric correction and grayscale transformation, removes noise and interference. Filtering and downsampling of laser point cloud data removes redundant points, improving data quality and fusion efficiency.
[0022] Specifically, the radiometric correction methods are as follows: white balance correction and lens distortion correction are performed on visible light images; non-uniformity correction (NUC) and atmospheric transmittance correction are performed on infrared images. The specific method of grayscale transformation is to perform linear or nonlinear grayscale stretching transformation on the infrared image based on the temperature range to enhance the visualization effect of thermal features.
[0023] The specific method of point cloud filtering is to use the statistical outlier removal algorithm, namely the SOR algorithm, and / or the radius outlier removal algorithm, namely the ROR algorithm, to filter out noise points in the laser point cloud data; The specific method of point cloud downsampling is as follows: a voxel grid downsampling algorithm is used on the filtered laser point cloud data to reduce the point cloud density while ensuring geometric features.
[0024] Step S2: Register the visible light image and the laser point cloud, including: Step S2-1: On the visible light image, use the Harris corner detector or SIFT feature extraction algorithm to identify geometric feature points such as flange edges and basin insulator outlines; in the laser point cloud data, identify the corresponding three-dimensional feature points through curvature calculation or feature line extraction algorithms. Step S2-2: Establish sparse corresponding points between the visible light image and the laser point cloud data, and apply the point-to-point ICP algorithm based on KD tree acceleration for registration. In the initial registration process, the maximum number of iterations is set to 100, the distance threshold is twice the average density of the point cloud, and the root mean square error change is less than 0.001 mm as the convergence condition.
[0025] The ICP refinement method is as follows: A point-to-point ICP algorithm based on KD-tree acceleration is applied for initial registration, with a maximum iteration count of 100, a distance threshold of twice the average point cloud density, and a convergence condition of: Where: RMSE is the root mean square error, N is the total number of corresponding points, and p i The location of the laser point cloud is calculated or estimated from feature points in a visible light image. i In the laser point cloud, with p i The corresponding location of the laser point cloud.
[0026] In this embodiment, the root mean square error variation is less than 0.001 mm; Step S3: Register the infrared image and the laser point cloud, including: Step S3-1: Extract the defined significant thermal region feature points based on the temperature gradient from the infrared image; Step S3-2: According to the thermal field constraints, project each thermal feature point onto the laser point cloud. When registering the infrared image and the laser point cloud, the physical relationship between the thermal distribution and the geometric structure is introduced as a constraint condition. At the flange connection, the surface temperature distribution is uniform. For a conductor, its temperature distribution is related to the current load and heat dissipation conditions.
[0027] When registering infrared images with preliminary registration data, the physical relationship between thermal distribution and geometric structure is introduced as a constraint. For example, for flange connections, it is assumed that the surface temperature distribution is relatively uniform; for conductors, the temperature distribution is related to the current load and heat dissipation conditions. This prior knowledge is used to construct thermal-geometric consistency constraints, which are then added to the objective function of feature matching to optimize the registration results.
[0028] Step S3-3: Construct a joint descriptor using the feature points of the hot region and the three-dimensional geometric features near their projection points on the point cloud, where the three-dimensional geometric features include normal direction and curvature; Step S3-4: Match the joint descriptor with the texture feature descriptor of the corresponding region on the visible light image, and use the RANSAC algorithm to eliminate false matches.
[0029] Between the infrared image and the initially registered data, based on the correlation between thermal features and geometric structure, an improved feature matching algorithm is used to further refine the registration, ensuring that the three are highly consistent in the spatial coordinate system, thus laying the foundation for fusion. The improved feature matching algorithm is as follows: The system extracts significant thermal feature points based on temperature gradients from infrared images, pre-defined by the administrator. These thermal feature points are then projected onto a pre-registered 3D point cloud. A joint descriptor is constructed using the thermal feature points and their corresponding 3D geometric features near the projection points on the point cloud, such as normal direction and curvature. This descriptor is then matched with the texture feature descriptor of the corresponding region on the visible light image. The RANSAC algorithm is used to eliminate mismatches, thus achieving accurate registration between the infrared image and the visible light point cloud data.
[0030] Register the target and ensure that the final registration result ensures that the spatial coordinate error of the pixels in the visible light image, the pixels in the infrared image, and the three-dimensional points in the laser point cloud for the same physical location is less than a set value, which can generally be 1 mm.
[0031] Step S4: Obtain the visible light pixel value, infrared pixel value, and point cloud attribute value of each target location. Query the weight values of the visible light pixel value, infrared pixel value, and point cloud attribute value according to the task category, and perform weighted fusion to obtain the fused attribute value. To address GIS inspection needs, an improved weighted fusion algorithm is employed for data fusion. This algorithm integrates visible light texture information, infrared thermal features, and the 3D geometric information of laser point clouds to generate fused data containing comprehensive information, thus more accurately reflecting the operational status of the GIS. Within the registered spatial coordinate system, for each 3D location, the fused attribute value is calculated by multiplying the corresponding visible light pixel value, infrared pixel value, and point cloud attribute value by their respective weight values W. v W i W l The summation yields the point cloud attribute values, which can be reflection intensity, color, temperature, or geometric attributes. The specific formula for calculating the fusion attribute value is as follows: Among them W v ,+ W i +W l =1, weights are dynamically allocated based on the detection task; I v T i and A lThese are the corresponding visible light pixel values, infrared pixel values, and point cloud attribute values; Weights are assigned based on the detection priorities of different parts of the GIS. For example, infrared light data has a high weight when detecting thermal defects caused by partial discharge, while laser point cloud data has a high weight when locating the discharge location. The improved weighted fusion algorithm is as follows: Establish a mapping table between GIS component types and inspection tasks. Here, GIS components can be conductors, flanges, pot insulators, and shells, and inspection tasks can be thermal defect detection, discharge location, and visual inspection. Based on the current detection task and the component type of the area to be analyzed, weights are assigned to visible light data, infrared data, and laser point cloud data from a predefined weight table.
[0032] For example, when the task is to detect thermal defects caused by partial discharge and the region is a conductor, the weight of the visible light pixel value is 0.6, the weight of the infrared pixel value is 0.3, and the weight of the point cloud attribute value is 0.1. When the task is to locate the discharge position, the weight of the visible light pixel value is 0.7, the weight of the infrared pixel value is 0.2, and the weight of the point cloud attribute value is 0.1.
[0033] In other embodiments, the improved weighted fusion algorithm may further be: Calculate the confidence index C for the features extracted from each data source. v C i C l For example, the signal-to-noise ratio (SNR) of temperature values in infrared images, the sharpness / curvature saliency of geometric features in point clouds, and the contrast of textures in visible light images. Weight W i W v and W l The calculation is based on the confidence level and is performed dynamically. The formula is as follows: Step S5: Construct a feature vector based on the fused attribute values and input it into the trained detection model to obtain the detection result.
[0034] Furthermore, in some embodiments, the registered fused data also needs to be evaluated, specifically in the following manner: Construct a precision evaluation system specifically for GIS inspection using visible light, infrared light, and laser point cloud data fusion; In terms of geometric accuracy, the three-dimensional positional deviation and dimensional error between the fused data and the real GIS model are calculated; in terms of thermal feature accuracy, the fused data and the standard heat source are compared in terms of temperature measurement values, heat distribution uniformity, and other indicators; in terms of detection reliability, the accuracy and false alarm rate of the fused data in identifying typical GIS faults such as partial discharge and insulation aging are evaluated, and the accuracy and reliability of the fused data are comprehensively quantified.
[0035] The specific method for evaluating geometric accuracy is as follows: A high-precision laser scanner is used to acquire a 3D model of the GIS equipment as the certified GIS model. Select at least X1 feature points from the fused data. The feature points can be the flange center or the bolt vertex. X1 can be 50. Calculate the Euclidean distance between their three-dimensional coordinates and the corresponding points in the identified GIS model. Statistically calculate the average error, root mean square error and maximum error. The Euclidean distance d between feature points is calculated using the following formula: ; in, x f To fuse the x-coordinates of feature points in the data, x r To fuse the y-coordinates of feature points in the data, y f To fuse the z-coordinates of feature points in the data, y r To determine the x-coordinate of the corresponding feature point in the GIS model (reference model), z f To determine the y-coordinate of the corresponding feature point in the GIS model (reference model), z r To determine the z-coordinate of the corresponding feature point in the GIS model (reference model).
[0036] Dimensions of key components in the fused measurement data, such as flange spacing and insulator diameter, are compared with the design dimensions or high-precision measurements in the GIS model to calculate the percentage of dimensional error. Thermal feature accuracy evaluation: At critical locations on the surface of GIS equipment, such as conductor connections or housings, attach or place standard heat sources with known emissivity, such as high-precision blackbody radiation sources or calibrated temperature sensors. In the fused data, the temperature measurement value of the area where the standard heat source is located is read and compared with its actual temperature value. The temperature measurement error ΔT is calculated as follows: Wherein: T fused To integrate the temperature measurements at the location of the standard heat source in the data, T refThis represents the actual temperature value of the standard heat source. Analyze the heat distribution, such as temperature standard deviation, in the standard heat source region of the fused data to evaluate its ability to characterize uniformity.
[0037] Example 2 The electronic device of this invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in read-only memory (ROM) or loaded from a storage unit into random access memory (RAM). The RAM may also store various programs and data required for device operation. The CPU, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.
[0038] Multiple components in the device are connected to the I / O interface, including: input units such as keyboards and mice; output units such as various types of displays and speakers; storage units such as disks and optical discs; and communication units such as network interface cards (NICs), modems, and wireless transceivers. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0039] The processing unit executes the various methods and processes described above, such as methods S1 to S5. For example, in some embodiments, methods S1 to S5 may be implemented as computer software programs tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by the CPU, one or more steps of methods S1 to S5 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S5 by any other suitable means (e.g., by means of firmware).
[0040] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.
[0041] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.
[0042] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0043] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a 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 this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
Claims
1. A method for detecting multi-optical inspection fusion data of gas-insulated combination switches, characterized in that, include: Step S1: Acquire visible light image, infrared image and laser point cloud of the gas-insulated combination switch under test; Step S2: Register the visible light image and the laser point cloud; Step S3: Register the infrared image and the laser point cloud; Step S4: Obtain the visible light pixel value, infrared pixel value, and point cloud attribute value of each target location. Query the weight values of the visible light pixel value, infrared pixel value, and point cloud attribute value according to the task category, and perform weighted fusion to obtain the fused attribute value. Step S5: Construct a feature vector based on the fused attribute values and input it into the trained detection model to obtain the detection result.
2. The detection method for multi-light inspection fusion data of a gas-insulated combination switch according to claim 1, characterized in that, During the acquisition of visible light images, infrared images, and laser point clouds, the target device's position and orientation remain consistent.
3. The detection method for multi-optical inspection fusion data of a gas-insulated combination switch according to claim 1, characterized in that, Step S1 further includes: Preprocessing of visible light and infrared light images, including radiometric correction and grayscale transformation, removes noise and interference. Filtering and downsampling of laser point cloud data removes redundant points, improving data quality and fusion efficiency.
4. The detection method for multi-light inspection fusion data of a gas-insulated combination switch according to claim 1, characterized in that, Step S2 includes: Step S2-1: Extract the geometric feature points of the set features on the visible light image. The set features include the flange edge and the outline of the basin insulator. Step S2-2: Establish sparse corresponding points between the visible light image and the laser point cloud data, and apply the point-to-point ICP algorithm based on KD tree acceleration for registration. In the initial registration process, the maximum number of iterations is set to 100, the distance threshold is twice the average density of the point cloud, and the root mean square error change is less than 0.001 mm as the convergence condition.
5. The detection method for multi-light inspection fusion data of a gas-insulated combination switch according to claim 4, characterized in that, Step S3 includes: Step S3-1: Extract the defined significant thermal region feature points based on the temperature gradient from the infrared image; Step S3-2: According to the thermal field constraints, project each thermal feature point onto the laser point cloud; Step S3-3: Construct a joint descriptor using the feature points of the hot region and the three-dimensional geometric features near their projection points on the point cloud, wherein the three-dimensional geometric features include normal direction and curvature; Step S3-4: Match the joint descriptor with the texture feature descriptor of the corresponding region on the visible light image, and use the RANSAC algorithm to eliminate false matches.
6. The detection method for multi-light inspection fusion data of a gas-insulated combination switch according to claim 5, characterized in that, In step S3-2, when registering the infrared image and the laser point cloud, the physical relationship between thermal distribution and geometric structure is introduced as a constraint condition, wherein; At the flange connection, the surface temperature distribution is uniform. For a conductor, its temperature distribution is related to the current load and heat dissipation conditions.
7. The detection method for multi-light inspection fusion data of a gas-insulated combination switch according to claim 1, characterized in that, The point cloud attribute value is reflection intensity.
8. The detection method for multi-light inspection fusion data of a gas-insulated combination switch according to claim 1, characterized in that, When the task is to detect thermal defects caused by partial discharge and the region is a conductor, the weight of the visible light pixel value is 0.6, the weight of the infrared pixel value is 0.3, and the weight of the point cloud attribute value is 0.
1. When the task is to locate the discharge position, the weight of the visible light pixel value is 0.7, the weight of the infrared pixel value is 0.2, and the weight of the point cloud attribute value is 0.
1.
9. A detection device for multi-light inspection fusion data of a gas-insulated combination switch, comprising a memory, a processor, and a program stored in the memory, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1-8.
10. A storage medium having a program stored thereon, characterized in that, When the program is executed, it implements the method as described in any one of claims 1-8.