A three-dimensional point cloud multi-modal intelligent analysis system for power transmission lines

Abstract

The application provides a power transmission line three-dimensional point cloud multi-modal intelligent analysis system, the system comprises a data acquisition module, an intelligent identification and construction module, a simulation analysis module and a report output module; the data acquisition module is used for receiving and cleaning power transmission line multi-source data; the intelligent identification and construction module is used for combining multi-source data to construct and update the three-dimensional space model of the power transmission line; the simulation analysis module is used for combining real-time environmental parameters to perform multi-working-condition physical simulation on the power transmission line equipment in the three-dimensional space model and evaluate potential hidden danger risks; the report output module is used for outputting the risk report of each power transmission line equipment; the application realizes intelligent construction of the three-dimensional power transmission line model under multi-modal data driving, and improves the efficiency, accuracy and risk response capability of power transmission line inspection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power inspection and intelligent analysis, and particularly relates to a three-dimensional point cloud multi-modal intelligent analysis system for a power transmission line. BACKGROUND

[0002] With the continuous expansion of the number and distribution range of power transmission lines, the traditional manual inspection method has been difficult to meet the high-frequency and high-precision line state perception demand; in recent years, the development of multi-source perception technologies such as point cloud, image and infrared provides a basis for comprehensive perception of line state, but how to efficiently fuse multi-modal data, accurately identify complex hidden danger features, and realize systematic and intelligent risk analysis and auxiliary decision-making is still a key challenge in the current intelligent inspection of power transmission lines.

[0003] After consulting relevant disclosed technical solutions, a technical solution with publication number CN118246730A proposes a power transmission line hidden danger analysis method, device, computer equipment, storage medium and computer program product; the method comprises: acquiring original three-dimensional point cloud data for a target power transmission line; in the case that the original three-dimensional point cloud data meets the preset data verification requirements, extracting key point data in the original three-dimensional point cloud data; determining the fitting plane corresponding to the original three-dimensional point cloud data according to the key point data, and determining the distance difference between the fitting plane and the target power transmission line according to the key point data; determining the hidden danger analysis result corresponding to the target power transmission line according to the distance difference and the preset safety distance threshold; this solution can efficiently, comprehensively and reliably analyze the hidden dangers of the power transmission line based on the three-dimensional point cloud data, timely determine the accurate power transmission line hidden danger analysis result, and improve the analysis efficiency of the power transmission line safety hidden danger; but this solution only analyzes hidden dangers based on single point cloud data, lacks multi-modal information fusion, and is difficult to identify non-geometric hidden dangers such as thermal anomalies; at the same time, its hidden danger judgment mainly depends on static geometric fitting, and lacks the modeling ability of dynamic response to environmental changes and working conditions. SUMMARY

[0004] The present application relates to the technical field of power inspection and intelligent analysis, and particularly relates to a three-dimensional point cloud multi-modal intelligent analysis system for a power transmission line.

[0005] The present application adopts the following technical solutions:

[0006] The system comprises a data acquisition module, an intelligent identification and construction module, a simulation analysis module and a report output module; the data acquisition module is used for receiving and cleaning power transmission line multi-source data; the intelligent identification and construction module is used for constructing and updating the three-dimensional space model of the power transmission line in combination with the multi-source data; the simulation analysis module is used for performing multi-working-condition physical simulation on the power transmission line equipment in the three-dimensional space model in combination with real-time environmental parameters and evaluating potential hidden danger risks; and the report output module is used for outputting risk reports of each power transmission line equipment.

[0007] The data acquisition module comprises a multi-source perception acquisition unit and a data cleaning and registration unit; the multi-source perception acquisition unit is used for acquiring power transmission line multi-source data collected by unmanned aerial vehicles through an unmanned aerial vehicle management platform, wherein the power transmission line multi-source data comprises point cloud data, visible light image data and infrared image data; and the data cleaning and registration unit is used for performing denoising, time synchronization and coordinate registration processing on the power transmission line multi-source data.

[0008] The intelligent identification and construction module comprises a three-dimensional model construction unit, a multi-modal target identification unit and a feature parameter extraction unit; the three-dimensional model construction unit is used for constructing the three-dimensional space model of the power transmission line in combination with the power transmission line multi-source data after cleaning processing; the multi-modal target identification unit is used for identifying typical defects of the power transmission line equipment in the three-dimensional space model; and the feature parameter extraction unit is used for extracting structural feature parameters of the power transmission line equipment in the three-dimensional space model.

[0009] Further, the three-dimensional model construction unit comprises a point cloud geometry modeling subunit, a multi-source semantic fusion subunit and a model updating subunit; the model updating subunit is used for completing updating of the three-dimensional space model in combination with continuously updated power transmission line multi-source data; the model updating subunit divides the three-dimensional space model into at least one candidate updating area based on a preset space division strategy, calculates an updating priority of each candidate updating area, and proportionally allocates system video memory resources to perform updating of the three-dimensional space model in combination with the updating priority of each candidate updating area; wherein the allocation mode of the video memory resources comprises:

[0010] S11: grid dividing the point cloud space in the three-dimensional space model according to a space coordinate range to form at least one candidate updating area, each candidate updating area corresponding to a point cloud segment within a certain range;

[0011] S12: calculating an updating priority of each candidate updating area:

[0012] ;

[0013] wherein, is the first Update priority of each candidate update region; For the first The voxel overlap rate of the candidate update region is determined by the voxel overlap rate of the _th candidate update region. The point cloud data before and after the update in each candidate update region are voxelized, and the historical voxel set and the current voxel set of the candidate update region are obtained respectively. The ratio of their intersection and union is calculated. For the first The rate of change of point cloud density for each candidate update region is obtained by calculating the difference between the current point cloud density and the historical point cloud density. For the first The time interval since the last update for each candidate update region. The preset time normalization coefficient; , and To update the priority weight coefficients, they are pre-set according to the experiment and can be dynamically adjusted according to actual operational needs;

[0014] S13: Allocate memory usage for each candidate update region based on its update priority:

[0015] ;

[0016] in, For the first Memory usage of each candidate update region This represents the total available video memory capacity of the system.

[0017] Furthermore, the simulation analysis module includes an environmental parameter acquisition unit, an operating condition simulation calculation unit, and a hidden danger risk assessment unit; the environmental parameter acquisition unit is used to acquire real-time environmental parameters; the operating condition simulation calculation unit is used to combine real-time environmental parameters with the structural characteristic parameters of the transmission line equipment to simulate the operating status and spatial variation behavior of the transmission line equipment under different operating conditions; the hidden danger risk assessment unit is used to analyze potential hidden danger risks based on the simulation results of the operating condition simulation calculation unit.

[0018] Furthermore, the point cloud geometric modeling subunit is used to extract spatial contour information of the transmission line area based on point cloud data, and to perform structural modeling of transmission line equipment and environmental elements within the transmission line area to generate a point cloud structure model; the multi-source semantic fusion subunit is used to extract texture features and thermal features from visible light image data and infrared image data, and to generate semantic annotation labels based on texture features and thermal features, and to map the semantic annotation labels to the corresponding spatial locations in the point cloud structure model, thereby realizing category recognition and state annotation of each region in the point cloud structure model, and thus constructing a three-dimensional spatial model with spatial geometric attributes and semantic recognition information.

[0019] Further, the multi-modal target recognition unit extracts device surface defects from the visible light image data by combining a pre-established defect detection model, and identifies temperature abnormality defects from the infrared image data in combination with a device standard operating temperature account, and registers the device surface defects and the temperature abnormality defects as typical defects into corresponding positions of the three-dimensional space model.

[0020] Further, the report output module generates a standardized risk report of each power transmission line device based on the output results of the multi-modal target recognition unit and the hidden danger risk assessment unit, and provides the report to the user by accessing a deepseek model for semantic fusion and language organization.

[0021] The beneficial effects achieved by the present application are:

[0022] The present application fuses multi-modal data of point clouds, visible light images and infrared images, constructs a three-dimensional space model with semantic recognition capability, and performs working condition simulation and hidden danger assessment in combination with environmental parameters, so as to realize intelligent analysis and dynamic risk identification of power transmission line equipment; by introducing a region priority update mechanism of video memory perception, differential modeling scheduling is realized under limited computing resources, and the resource utilization and inspection response efficiency of the system are improved; at the same time, the intelligent report output mode generated by the large model enhances the expression ability and practical value of the inspection results. BRIEF DESCRIPTION OF DRAWINGS

[0023] The present application can be further understood from the following description in conjunction with the accompanying drawings. The components in the drawings are not necessarily drawn to scale, but the emphasis is placed on showing the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0024] Figure 1 The present application is a whole module schematic diagram.

[0025] Figure 2 The present application is a specific allocation process schematic diagram of the model update subunit for video memory resources.

[0026] Figure 3 The present application is a function variation schematic diagram of three functions in the update priority calculation under normalized input. , and

[0027] Figure 4 The present application is a dynamic adjustment method flowchart of the update priority weight coefficient. DETAILED DESCRIPTION

[0028] ​In order to make the purpose, technical solutions and advantages of the present application clearer and more apparent, the present application will be further described in detail below in conjunction with its embodiments; it should be understood that the specific embodiments described herein are only for explaining the present application and are not intended to limit the present application; for those skilled in the art, other systems, methods and / or features of the embodiments will become apparent after reading the following detailed description; all such additional systems, methods, features and advantages are intended to be included within the present specification; included within the scope of the present application, and protected by the appended claims; additional features of the disclosed embodiments are described in the following detailed description, and will be apparent from the following detailed description.

[0029] The same or similar reference numerals in the drawings of the embodiments of the present application correspond to the same or similar components; in the description of the present application, it should be understood that if the orientations or positional relationships indicated by the terms "upper", "lower", "left", "right" and the like are based on the orientations or positional relationships shown in the drawings, they are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or components referred to must have a particular orientation, be constructed and operated in a particular orientation, therefore the terms describing the positional relationships in the drawings are only used for exemplary illustration, and cannot be understood as limiting the present patent, for those skilled in the art, the specific meanings of the above terms can be understood according to the specific circumstances.

[0030] Embodiment one:

[0031] As shown in Figure 1 The present embodiment provides a power transmission line three-dimensional point cloud multi-modal intelligent analysis system, the system comprises a data acquisition module, an intelligent identification and construction module, a simulation analysis module and a report output module; the data acquisition module is used for receiving and cleaning power transmission line multi-source data; the intelligent identification and construction module is used for constructing and updating the three-dimensional space model of the power transmission line in combination with the multi-source data; the simulation analysis module is used for performing multi-working-condition physical simulation on the power transmission line equipment in the three-dimensional space model in combination with real-time environmental parameters and evaluating potential hidden danger risks; the report output module is used for outputting risk reports of each power transmission line equipment;

[0032] The data acquisition module comprises a multi-source perception acquisition unit and a data cleaning and registration unit; the multi-source perception acquisition unit is used for obtaining power transmission line multi-source data collected by unmanned aerial vehicles through an unmanned aerial vehicle management platform, the power transmission line multi-source data comprising point cloud data, visible light image data and infrared image data; the data cleaning and registration unit is used for performing denoising, time synchronization and coordinate registration processing on the power transmission line multi-source data;

[0033] The intelligent recognition construction module comprises a three-dimensional model construction unit, a multi-modal target recognition unit and a feature parameter extraction unit; the three-dimensional model construction unit is configured to construct a three-dimensional space model of the power transmission line in combination with the multi-source data of the power transmission line after cleaning processing; the multi-modal target recognition unit is configured to recognize typical defects of the power transmission line equipment in the three-dimensional space model; and the feature parameter extraction unit is configured to extract structural feature parameters of the power transmission line equipment in the three-dimensional space model.

[0034] Further, the simulation analysis module comprises an environment parameter acquisition unit, a working condition simulation calculation unit and a hidden danger risk assessment unit; the environment parameter acquisition unit is configured to acquire real-time environment parameters; the working condition simulation calculation unit is configured to simulate the running state and spatial change behavior of the power transmission line equipment under different working condition conditions in combination with the real-time environment parameters and the structural feature parameters of the power transmission line equipment; and the hidden danger risk assessment unit is configured to analyze potential hidden danger risks based on the simulation results of the working condition simulation calculation unit.

[0035] Further, the three-dimensional model construction unit comprises a point cloud geometry modeling subunit, a multi-source semantic fusion subunit and a model updating subunit; the point cloud geometry modeling subunit is configured to extract spatial contour information of the power transmission line area based on point cloud data, and to perform structural modeling on the power transmission line equipment and environmental elements in the power transmission line area to generate a point cloud structure model; the multi-source semantic fusion subunit is configured to extract texture features and thermal features from visible light image data and infrared image data, and to generate semantic annotation labels according to the texture features and thermal features, and to map the semantic annotation labels to corresponding spatial positions of the point cloud structure model to realize class recognition and state annotation of each region in the point cloud structure model, thereby constructing a three-dimensional space model with spatial geometric attributes and semantic recognition information; and the model updating subunit is configured to complete updating of the three-dimensional space model in combination with continuously updated multi-source data of the power transmission line.

[0036] Specifically, the three-dimensional model construction unit is deployed on a GPU configured with 24G graphics memory to complete the function execution in the unit;

[0037] Specifically, the point cloud geometry modeling subunit extracts target regions such as towers, power conductors, vegetation and ground in the power transmission line area by performing spatial coordinate analysis and density clustering processing on point cloud data, in combination with a point cloud classification network based on deep learning such as PointNet++, and constructs spatial boundaries and contour models for each type of target region to generate a point cloud structure model with structural levels;

[0038] Furthermore, the model update subunit divides the three-dimensional spatial model into at least one candidate update region based on a preset spatial partitioning strategy, calculates the update priority of each candidate update region, and allocates system memory resources proportionally based on the update priority of each candidate update region to perform the update of the three-dimensional spatial model.

[0039] Furthermore, such as Figure 2 , Figure 3 As shown, the specific allocation method of the model update subunit for video memory resources is as follows:

[0040] S11: Divide the point cloud space in the 3D spatial model into a grid according to the spatial coordinate range to form at least one candidate update region. Each candidate update region corresponds to a point cloud segment within a certain range.

[0041] S12: Calculate the update priority for each candidate update region:

[0042] ;

[0043] in, For the first Update priority of each candidate update region; For the first The voxel overlap rate of the candidate update region is determined by the voxel overlap rate of the _th candidate update region. The point cloud data before and after the update in each candidate update region are voxelized, and the historical voxel set and the current voxel set of the candidate update region are obtained respectively. The ratio of their intersection and union is calculated. For the first The rate of change of point cloud density for each candidate update region is obtained by calculating the difference between the current point cloud density and the historical point cloud density. For the first The time interval since the last update for each candidate update region. The preset time normalization coefficient; , and To update the priority weight coefficients, they are pre-set according to the experiment and can be dynamically adjusted according to actual operational needs;

[0044] S13: Allocate memory usage for each candidate update region based on its update priority:

[0045] ;

[0046] in, For the first Memory usage of each candidate update region This represents the total available video memory capacity of the system.

[0047] By comprehensively considering the spatial structure changes, point cloud density fluctuations, and update intervals of each candidate update region in the transmission line, and dynamically allocating memory resources according to update priority, differentiated model update scheduling can be achieved under limited hardware conditions. This effectively avoids memory resources being occupied by low-value regions and prioritizes critical regions that have undergone significant changes, large data fluctuations, or have not been updated for a long time, thereby improving the overall memory utilization, update efficiency, and modeling accuracy of the system.

[0048] Meanwhile, the memory allocation mechanism based on update priority enables the system to prioritize loading data in areas of significant change and complete corresponding model updates, defect identification, and potential risk assessment. This allows users to obtain the most valuable inspection information with limited computing resources, thereby improving the timeliness and response efficiency of defect identification.

[0049] Furthermore, the multimodal target recognition unit extracts equipment surface defects from visible light image data by combining a pre-established defect detection model, and identifies temperature anomaly defects by combining infrared image data with the equipment's standard operating temperature log. The equipment surface defects and temperature anomaly defects are then registered as typical defects to their corresponding positions in the three-dimensional spatial model.

[0050] Furthermore, the feature parameter extraction unit, based on the constructed three-dimensional spatial model, identifies and extracts structural feature parameters of transmission line equipment and environmental elements through spatial measurement. These structural feature parameters include, but are not limited to: tower height, conductor sag, minimum clearance between conductor and ground / vegetation / building, spacing between adjacent towers, conductor direction and arc fitting radius, insulator hanging point position and length, hardware installation height, and shortest distance from tree obstruction to conductor.

[0051] Furthermore, the operating condition simulation calculation unit, based on the physical and mechanical model and combined with the structural characteristic parameters of the transmission line equipment and environmental factors, simulates the operating state and spatial variation behavior of the transmission line equipment under different operating conditions; the environmental parameters include, but are not limited to, external meteorological information such as temperature, wind speed, wind direction, ice thickness, and lightning intensity; during the simulation, the system performs numerical extrapolation on key operating states and spatial variation behaviors such as conductor sag change, tower stress deformation, conductor offset angle, and thermal expansion coefficient based on the physical and mechanical model;

[0052] Furthermore, the hidden danger risk assessment unit identifies potential hidden danger risks based on the key operating states and spatial change behaviors generated by the operating condition simulation unit, combined with equipment operating specifications and safety limit standards.

[0053] Further, the report output module generates a standardized risk report of each power transmission line equipment based on the output results of the multi-modal target recognition unit and the hidden danger risk assessment unit, and provides the report to the user through semantic fusion and language organization by accessing the deepseek model.

[0054] Embodiment Two

[0055] It should be understood that the present embodiment includes all the features of any one of the preceding embodiments and is further improved based on them.

[0056] The present embodiment provides a three-dimensional point cloud multi-modal intelligent analysis system for power transmission lines, which includes a data acquisition module, an intelligent recognition and construction module, a simulation analysis module, and a report output module. The data acquisition module is used to receive and clean multi-source data of power transmission lines. The intelligent recognition and construction module is used to construct and update a three-dimensional space model of power transmission lines in combination with multi-source data. The simulation analysis module is used to perform multi-working-condition physical simulation on power transmission line equipment in the three-dimensional space model in combination with real-time environmental parameters and assess potential hidden danger risks. The report output module is used to output risk reports of each power transmission line equipment.

[0057] Further, as shown in Figure 4 , the present scheme can dynamically adjust the update priority weight coefficients 、 and according to actual operation requirements to realize adaptive matching of model update strategy and actual risk perception ability; the dynamic adjustment of the update priority weight coefficients 、 and is as follows:

[0058] S21: set the feedback period of weight adjustment to monitoring periods, and calculate the update target function corresponding to each update priority weight coefficient after every periods:

[0059] ;

[0060] ;

[0061] ;

[0062] wherein, is the update target function of , is the update target function of , is the update target function of , is the update target function of the i-th The amount of change of the surface defects of the equipment in the power transmission line in the monitoring period is obtained by calculating the difference between the surface defects of the equipment in the monitoring period and the surface defects of the equipment in the previous monitoring period; The amount of change of the typical defects in the power transmission line in the first monitoring period is obtained by calculating the difference between the amount of change of the typical defects in the monitoring period and the amount of change of the typical defects in the previous monitoring period; The amount of change of the typical defects in the power transmission line in the first monitoring period is obtained by calculating the difference between the amount of change of the typical defects in the monitoring period and the amount of change of the typical defects in the previous monitoring period; The amount of change of the typical defects in the power transmission line in the first monitoring period is obtained by calculating the difference between the amount of change of the typical defects in the monitoring period and the amount of change of the typical defects in the previous monitoring period; The amount of change of the typical defects in the power transmission line in the first monitoring period is obtained by calculating the difference between the amount of change of the typical defects in the monitoring period and the amount of change of the typical defects in the previous monitoring period;

[0063] S22: updating the priority weight coefficient according to the updated target function obtained in the previous step by the following way;

[0064] ;

[0065] ;

[0066] ;

[0067] wherein, , and are the updated priority weight coefficients, , and are the priority weight coefficients before updating, is a learning rate for controlling the updating range, which can be preset by a user;

[0068] The weight dynamic adjustment mechanism provided in the scheme enables the system updating strategy to constantly optimize itself according to the running data, automatically enhances the perception ability of the most prominent risk type at present, and improves the inspection efficiency and the accuracy of risk prevention and control.

[0069] The above disclosed content is only the preferred feasible embodiment of the present application, and does not limit the protection scope of the present application, so that any equivalent technical change made by applying the content of the present application and the drawings is included in the protection scope of the present application, and furthermore, the elements can be updated as the technology develops.

Claims

1. A power transmission line three-dimensional point cloud multi-modal intelligent analysis system, characterized in that, The system comprises a data acquisition module, an intelligent identification construction module, a simulation analysis module and a report output module; the data acquisition module is used for receiving and cleaning power transmission line multi-source data; the intelligent identification construction module is used for constructing and updating a three-dimensional space model of the power transmission line in combination with the multi-source data; the simulation analysis module is used for performing multi-working condition physical simulation on power transmission line equipment in the three-dimensional space model in combination with real-time environmental parameters and evaluating potential hidden danger risks; and the report output module is used for outputting risk reports of each power transmission line equipment. The data acquisition module comprises a multi-source perception acquisition unit and a data cleaning and registration unit. The multi-source perception acquisition unit is used for obtaining power transmission line multi-source data collected by a UAV through a UAV management platform, wherein the power transmission line multi-source data comprises point cloud data, visible light image data and infrared image data; and the data cleaning and registration unit is used for performing denoising, time synchronization and coordinate registration processing on the power transmission line multi-source data. The intelligent identification construction module comprises a three-dimensional model construction unit, a multi-modal target identification unit and a feature parameter extraction unit. The three-dimensional model construction unit is used for constructing a three-dimensional space model of the power transmission line in combination with the cleaned power transmission line multi-source data; the multi-modal target identification unit is used for identifying typical defects of power transmission line equipment in the three-dimensional space model; and the feature parameter extraction unit is used for extracting structural feature parameters of the power transmission line equipment in the three-dimensional space model. The three-dimensional model construction unit comprises a point cloud geometry modeling subunit, a multi-source semantic fusion subunit and a model updating subunit; the model updating subunit is used for updating the three-dimensional space model in combination with continuously updated power transmission line multi-source data; the model updating subunit divides the three-dimensional space model into at least one candidate update area based on a preset space division strategy, calculates an update priority of each candidate update area, and proportionally allocates system video memory resources to each candidate update area to perform update on the three-dimensional space model; wherein the allocation mode of the video memory resources comprises: S11: dividing point cloud space in the three-dimensional space model into at least one candidate update area according to space coordinate ranges to form point cloud segments within a certain range for each candidate update area; S12: calculating an update priority of each candidate update area: ； in, For the first Update priority of each candidate update region; For the first The voxel overlap rate of the candidate update region is determined by the voxel overlap rate of the _th candidate update region. The point cloud data before and after the update in each candidate update region are voxelized, and the historical voxel set and the current voxel set of the candidate update region are obtained respectively. The ratio of their intersection and union is calculated. For the first The rate of change of point cloud density for each candidate update region is obtained by calculating the difference between the current point cloud density and the historical point cloud density. For the first The time interval between each candidate update region and the last update. The preset time normalization coefficient; , and To update the priority weight coefficients, they are pre-set according to the experiment and can be dynamically adjusted according to actual operational needs; S13: allocating video memory occupancy to each candidate update area according to the update priority of each candidate update area. ； wherein, is the total memory usage of the first candidate update region, is the total memory usage of the second candidate update region, is the total available system memory capacity.

2. The power transmission line three-dimensional point cloud multi-modal intelligent analysis system according to claim 1, characterized in that, The simulation analysis module comprises an environmental parameter acquisition unit, a working condition simulation calculation unit and a hidden danger risk evaluation unit; the environmental parameter acquisition unit is used for acquiring real-time environmental parameters; the working condition simulation calculation unit is used for simulating running states and spatial change behaviors of power transmission line equipment under different working condition conditions in combination with real-time environmental parameters and structural feature parameters of the power transmission line equipment; The hidden danger risk evaluation unit is used for analyzing potential hidden danger risks based on simulation results of the working condition simulation calculation unit. 3.The power transmission line three-dimensional point cloud multi-modal intelligent analysis system according to claim 1, characterized in that, The point cloud geometry modeling subunit is configured to extract spatial contour information of the power transmission line region based on the point cloud data, and perform structural modeling on power transmission line equipment and environmental elements in the power transmission line region to generate a point cloud structural model; the multi-source semantic fusion subunit is configured to extract texture features and thermal features from the visible light image data and the infrared image data, and generate semantic labeling labels according to the texture features and the thermal features, and map the semantic labeling labels to corresponding spatial positions of the point cloud structural model to realize class recognition and state labeling of each region in the point cloud structural model, thereby constructing a three-dimensional spatial model with spatial geometric attributes and semantic recognition information.

4. The three-dimensional point cloud multi-modal intelligent analysis system for power transmission lines according to claim 1, characterized in that, The multi-modal target recognition unit extracts equipment surface defects from the visible light image data by combining a pre-established defect detection model, and identifies temperature abnormal defects from the infrared image data by combining an equipment standard operating temperature account, and registers the equipment surface defects and the temperature abnormal defects as typical defects to corresponding positions in the three-dimensional spatial model.

5. The three-dimensional point cloud multi-modal intelligent analysis system for power transmission lines according to claim 1, characterized in that, The report output module generates a standardized risk report of each power transmission line equipment based on the output results of the multi-modal target recognition unit and the hidden danger risk assessment unit, and provides the report to a user by accessing a deepseek model for semantic fusion and language organization.

Images