A construction method of a large factory panoramic image automatic collection and management system based on air-ground cooperative data middle station

Abstract

A construction method of a large factory panoramic image automatic acquisition and management system based on air-ground collaborative data platform, the system comprises an unmanned aerial vehicle automatic hangar, an unmanned aerial vehicle, a cloud management platform, an air-ground collaborative data platform and a project management terminal. The unmanned aerial vehicle automatic hangar is deployed at the project site; the cloud management platform integrates task planning, route generation, panoramic map splicing, time axis comparison, BIM comparison and AI identification modules; the air-ground collaborative data platform serves as a data hub and connects the two-way data flow among the terminals. The construction method comprises hangar deployment and route pre-planning, task automatic triggering and take-off constraint check, unmanned automatic flight and panoramic acquisition, data return and cloud splicing and archiving, progress comparison, BIM comparison, AI hidden danger identification and closed-loop treatment analysis and application, and forms deliverable digital assets. The present application solves the technical problems of incomplete coverage of traditional manual inspection, data fragmentation, difficult traceability of hidden engineering and hidden danger response lag.

Description

Technical Field

[0001] This invention belongs to the field of building engineering information management and UAV remote sensing monitoring technology, specifically involving a method and system for automated acquisition, processing and management of aerial and ground collaborative panoramic images for large-scale industrial plant projects of millions of square meters. Background Technology

[0002] As construction projects expand to the million-square-meter and ultra-large scale, the traditional on-site management model, which relies primarily on manual inspections, fixed-point photography, and regular meetings, has revealed significant shortcomings: managers cannot quickly, comprehensively, and in real-time grasp the actual situation of high-risk areas such as deep foundation pits and steel structure roofs; data collection, which depends on manual labor, results in low efficiency, limited coverage, and discontinuous image recording, creating "blind spots in management vision" and a "data real-time deficit." Existing drone applications still require professional pilots for on-site operation, making it impossible to achieve high-frequency, regular, and unmanned automated data collection, and also making it difficult to form a data closed loop with project progress, safety, and quality management. Summary of the Invention

[0003] The purpose of this invention is to provide an automated acquisition and management system for panoramic images of large factories based on an air-ground collaborative data platform, and to address the technical problem to be solved.

[0004] To achieve the above objectives, the present invention adopts the following technical solution.

[0005] A construction method for an automated acquisition and management system of panoramic images of large factory buildings based on an air-ground collaborative data platform, characterized by: The automated acquisition and management system for panoramic images of large factory buildings includes an automated drone hangar, drones, a cloud management platform, an air-ground collaborative data platform, and a project management terminal. The automated hangar for unmanned aerial vehicles (UAVs) is deployed at a fixed site on the project site; the UAVs are connected to the automated hangars via a multimodal communication link, and the UAVs are equipped with a high-resolution panoramic camera, an edge computing module, and an RTK high-precision positioning module. The cloud management platform communicates with the drone automated hangar. The cloud management platform integrates a task planning and distribution module, an automatic flight path generation module, a panoramic image stitching and registration module, a panoramic image timeline comparison module, a BIM model comparison module, a safety and quality AI recognition module, and a hidden danger closed-loop management module. The drones are indirectly connected to the cloud management platform via an automated drone hangar as a gateway, or directly connected to the cloud management platform via a 4G / 5G network; the air-ground collaborative data platform, as the system's data hub, is deployed in the enterprise's internal data center or the cloud, and is used to realize bidirectional data flow and standardized data processing between the automated drone hangar, drones, the cloud management platform, and the project management terminal; the project management terminal is deployed on the user side and communicates with the cloud management platform and / or the air-ground collaborative data platform. The automated hangar for drones receives automated inspection tasks from the cloud management platform, controls the drones to perform panoramic image acquisition along a preset route, and uploads the acquired data to the cloud management platform for automated processing and analysis after standardization by the air-ground collaborative data platform. The processing results are then pushed to the project management terminal. The construction method includes the following steps: Step 1: Deployment of automated drone hangars and pre-planning of flight routes: Select deployment points on the project master plan to set up automated drone hangars and connect them to the network; In the cloud management platform, set up multiple panoramic data collection points over the project site according to project management needs, and generate a set of waypoints and flight routes covering the entire construction area. Step 2: Project managers select the target automated hangar and preset route through the cloud management platform, set the task execution cycle or trigger conditions, and the cloud management platform automatically checks the takeoff constraints. Once the constraints are met, the automated data collection process begins. Step 3, unattended automatic flight and panoramic data acquisition: Upon arrival of the mission time or receipt of the trigger command, the automatic hangar automatically opens the hatch, the UAV takes off autonomously and flies to each acquisition point in sequence according to the planned route, automatically performs multi-angle hovering shooting at each acquisition point, and generates raw image data. Step 4, Automatic Data Backhaul and Cloud Stitching Processing: After completing the data collection, the drone automatically returns to home, lands, and connects to the automated hangar for charging; the raw image data is automatically uploaded to the cloud management platform through the network module of the automated hangar; the cloud management platform automatically performs panoramic image stitching, geographic information registration, and image enhancement processing, and automatically archives the data according to the naming rules. Step 5: The cloud management platform performs one or more of the following analyses on the archived panoramic image: Compare panoramic images of the same collection point at different times on the same screen to generate a visual progress change report; By comparing real-world images with BIM models, early or late processes can be identified and early warnings can be generated. AI identifies potential safety and quality hazards, automatically generates alarm work orders, and pushes them to the responsible person to complete closed-loop handling. Step Six: Data Asset Accumulation Throughout the Construction Process: The data generated in Steps One through Five will be stored and accumulated into deliverable digital assets.

[0006] Preferably, the large factory building in step one is a steel structure factory building, and the panoramic data collection points set up cover at least the bottom area of ​​the deep foundation pit, the steel structure roof area, the material storage area, the component processing area and the main construction transportation roads in the site, and each collection point records its spatial coordinates, hovering height and multi-angle shooting parameters.

[0007] Preferably, in step two, the task execution cycle is set to perform panoramic data acquisition tasks at least twice a day at fixed times; before each task is triggered, the system automatically checks the takeoff constraints, and when all conditions are met, the UAV takes off to perform the task; if any condition is not met, the task is delayed or canceled, and the management personnel are notified. The conditions for triggering takeoff include at least the following: wind speed not exceeding a preset threshold, no rainfall, and drone battery power not lower than a preset minimum power threshold.

[0008] Preferably, in step three, for high-risk locations where it is inconvenient for personnel to enter or exit, namely the bottom area of ​​the deep foundation pit and the steel structure roof area, the drone is used to perform shooting parameters with lower flight altitude, higher overlap rate and shorter collection cycle at the special collection point by setting a flight path.

[0009] Preferably, the naming rule in step four is as follows: automatically archive according to the naming rule of collection time + spatial location code, and create folder structure archives according to collection point number and year and month, supporting multi-dimensional retrieval by time range, location area, and task batch; Step four also includes: automatically marking the spatial coordinates and process information of the corresponding locations in the panoramic view for the layered construction of concealed works, basic structures and steel structure node welding, forming a visual archive that can be used for final acceptance and subsequent operation and maintenance.

[0010] Preferably, the panoramic image timeline comparison in step five includes: supporting simultaneous comparison and display of any two panoramic images from different times on the same screen, and automatically generating a visual progress comparison report that marks the differences in construction status changes; supporting sliding comparison of panoramic images from the same collection point over multiple consecutive days to form a "daily-granular" visual growth record of the entire life cycle of the construction area.

[0011] BIM model comparison includes: importing the project's BIM 3D model and construction schedule; using algorithms to spatially register and visually overlay the panoramic images obtained from real-world data collection with the BIM model; automatically identifying the current actual progress and comparing the actual progress with the planned progress; highlighting advanced or lagging processes with color markings, flashing prompts, or list warnings, and pushing them to the project management terminal for managers to locate abnormal processes.

[0012] Safety and quality AI identification should recognize at least the following two types of potential hazards: Construction safety hazards include: not wearing safety protective equipment, cables dragging on the ground, uncovered bare soil, lack of edge protection, and fire hazards. Structural quality hazards include: cracks and leaks in concrete structures, appearance defects in welds of steel structures, and component misalignment.

[0013] Preferably, the closed-loop handling process in step five is as follows: First, an alarm work order is generated, which must include the type of hazard, location information, on-site images, and discovery time; then, the work order is automatically pushed to the work account of the person in charge of the corresponding area; subsequently, the system receives feedback from the person in charge on the completion of rectification and on-site images after rectification; for work orders that have not been rectified within the time limit, the system automatically generates an escalation alarm. This forms a complete closed-loop management process of automatic discovery, background analysis, task assignment, rectification feedback, and review and archiving.

[0014] Preferably, step five also includes setting up independent monitoring routes for high-risk areas such as the bottom of the deep foundation pit and the steel structure roof, and performing special data collection at a higher frequency than in ordinary areas; panoramic images of high-risk locations are marked as restricted access areas in the cloud management platform, and can only be viewed by authorized personnel; when the edge AI identifies illegal operations in high-risk areas, it immediately triggers a real-time alarm and pushes it to the mobile phone of the safety management personnel.

[0015] Preferably, the digital assets in step six include a panoramic image database organized by timeline, a visual work card for concealed works organized by spatial location, a safety and quality closed-loop record organized by hazard type, and a BIM-real scene comparison report organized by process. After the project is completed, these assets are packaged and delivered to the construction unit for auxiliary review of completion acceptance and subsequent operation and maintenance inquiries.

[0016] Compared with the prior art, the present invention has the following features and beneficial effects.

[0017] 1. This invention, by deploying automated drone hangars at project sites and coordinating with task planning and automatic scheduling on a cloud management platform, achieves fully automated panoramic image acquisition at least twice daily, and up to four times daily in high-risk areas, without any human intervention. Drones automatically take off, fly along preset routes, capture images from multiple angles, and automatically return to recharge, forming a complete unmanned operation loop, significantly reducing labor costs and increasing inspection frequency and data timeliness. Simultaneously, the air-ground collaborative data platform in this invention serves as the system's data hub, connecting bidirectional data flow between the automated drone hangars, drones, the cloud management platform, and project management terminals. Through various protocols such as MQTT / HTTPS, RESTful API, message queues, and WebSocket, standardized data exchange between heterogeneous systems is achieved, solving the problem of fragmented multi-source data in traditional construction management and providing a unified data foundation for applications such as progress analysis, quality inspection, and safety control.

[0018] 2. The cloud management platform in this invention integrates modules such as panoramic image stitching and registration, timeline comparison, BIM model comparison, and AI-based safety and quality recognition, achieving automated processing from raw images to structured analysis results. Specifically, it allows for simultaneous comparison of panoramic images from the same acquisition point at different times, generating a visualized progress change report and supporting daily-level traceability of the entire construction process; spatial registration and overlay comparison of real-scene images and BIM models automatically identifies and warns of advanced or lagging processes, assisting managers in accurately locating deviations; and AI automatically identifies construction safety hazards such as failure to wear safety protective equipment, cables dragging on the ground, uncovered bare soil, lack of edge protection, and fire hazards, as well as structural quality hazards such as concrete cracks, leaks, defects in steel structure welds, and component misalignment, significantly improving the timeliness and accuracy of hazard detection.

[0019] 3. This invention forms a complete closed-loop management process, from AI automatically identifying potential hazards and generating alarm work orders containing type, location, image, and time, to automatically pushing the responsible parties, receiving rectification feedback, and automatically escalating alarms upon exceeding the deadline. This process includes automatic discovery, background analysis, task assignment, rectification feedback, and review and archiving. This mechanism ensures that potential hazards during construction can be tracked to the end and handled properly, effectively solving the problems of missed hazard reporting, delayed handling, and unclear responsibilities in traditional management.

[0020] 4. This invention targets concealed works such as pipeline pre-embedding, layered foundation construction, and steel structure node welding. The system automatically marks spatial coordinates and process information in the panoramic image, forming visualized archived data that can be used for final acceptance and subsequent operation and maintenance. During final acceptance, the quality of concealed works can be verified without damaging the structure; in the later operation and maintenance phase, it can be quickly located by coordinates or process, significantly reducing the difficulty and cost of maintenance. Simultaneously, this invention sets independent monitoring routes for high-risk areas, such as the bottom of deep foundation pits and steel structure roofs, performing specialized data collection at a higher frequency than in ordinary areas. High-risk images are marked as restricted access areas, accessible only to authorized personnel. When the edge AI identifies illegal operations, it immediately pushes real-time alarms to the mobile devices of safety management personnel, achieving enhanced monitoring and rapid response in high-risk areas. Furthermore, this invention structurally stores the panoramic image database, visualized work cards for concealed works, safety and quality closed-loop records, and BIM-real-scene comparison reports generated throughout the construction process, organized by multiple dimensions such as timeline, spatial location, hazard type, and process organization. After project completion, these can be packaged and delivered to the construction unit. These digital assets can be used to assist in the final acceptance review, reducing the workload of on-site verification; they can also be used for later operation and maintenance inquiries to quickly locate hidden facilities; and they can also serve as a knowledge base for similar projects to enhance the company's digital management capabilities. Detailed Implementation

[0021] This construction method for an automated acquisition and management system for panoramic images of large-scale factory buildings based on an air-ground collaborative data platform is applicable to large steel structure factory building projects, which include deep foundation pit engineering, main steel structure engineering, internal equipment foundations, and pipeline pre-embedding engineering.

[0022] The automated acquisition and management system for panoramic images of large factory buildings includes an automated drone hangar, drones, a cloud management platform, an air-ground collaborative data platform, and a project management terminal. The automated drone hangar is deployed at a fixed site on the project site, providing an unmanned base for drones to autonomously take off and land, charge / swap batteries, store data, and transmit data. In this embodiment, a location with open terrain, no obstructions, close to the construction area but avoiding main roads is selected in the project's overall plan for the deployment of an automated drone hangar. The hangar connects to a cloud management platform via a 4G / 5G cellular network. Obstacles are cleared within a 5-meter radius of the hangar to ensure safe drone takeoff and landing. The automated drone hangar has autonomous charging, environmental perception, flight status self-checking, and network communication functions. The drones and the automated drone hangar are connected via a multimodal communication link, which includes, but is not limited to: a physical contact interface for autonomous charging and data exchange; a wireless data / image transmission link for real-time flight control and status monitoring; a 4G / 5G cellular network link for beyond-line-of-sight communication; and an RTK differential data link for centimeter-level precise guided landing. The drones are equipped with a high-resolution panoramic camera, an edge computing module, and an RTK high-precision positioning module. The cloud management platform communicates with the automated drone hangar. The cloud management platform is deployed in the data center of a third-party public cloud service provider, such as Alibaba Cloud, Tencent Cloud, or Huawei Cloud. The hangar accesses the cloud platform via a 4G / 5G cellular network or the internet. Alternatively, the cloud management platform can be deployed in a user-built data center or an internal server room, communicating with the hangar via an internal LAN or VPN. Or, the cloud management platform can be deployed on an integrated server at the project site, directly connected to the hangar via a LAN. The cloud management platform integrates modules for task planning and deployment, automatic flight path generation, panoramic image stitching and registration, panoramic image timeline comparison, BIM model comparison, safety and quality AI identification, and hazard closed-loop management. The drones are indirectly connected to the cloud management platform via an automated drone hangar as a gateway, or directly connected to the cloud management platform via a 4G / 5G network. The air-ground collaborative data platform, as the system's data hub, is deployed in the enterprise's internal data center or the cloud to enable bidirectional data flow and standardized data processing between the automated drone hangar, drones, the cloud management platform, and the project management terminal. The data platform is connected to the hangar via MQTT / HTTPS protocol through 4G / 5G or dedicated network, to the cloud management platform via RESTful API and message queue, and to the project management terminal via HTTPS / WebSocket protocol. The project management terminal is a PC web browser, tablet, or smartphone application deployed on the user side. It communicates with the cloud management platform and / or the air-to-ground collaborative data platform via HTTPS / WebSocket protocol through a local area network, the internet, or a VPN network. The project management terminal does not communicate directly with the UAV automated hangar or the UAV itself; its control commands and status queries for the underlying devices are indirectly implemented by calling service interfaces provided by the cloud management platform or data platform. The project management terminal is used to present inspection results, hazard alarms, and panoramic image comparison information to management personnel, and to receive task creation, route editing, and parameter configuration commands from management personnel. The automated hangar for drones receives automated inspection tasks from the cloud management platform, controls the drones to perform panoramic image acquisition along a preset route, and uploads the acquired data to the cloud management platform for automated processing and analysis after standardization by the air-ground collaborative data platform. The processing results are then pushed to the project management terminal. The construction method includes the following steps: Step 1: Deployment of automated drone hangars and pre-planning of flight routes: Select deployment points on the project master plan to set up automated drone hangars, and connect the automated drone hangars to the network via 4G / 5G cellular network or wired broadband; In the cloud management platform, set up multiple panoramic image data collection points over the project site according to project management needs, and generate a set of waypoints and flight routes covering the entire construction area. Step 2: Project managers select the target automated hangar and preset route through the cloud management platform, set the task execution cycle or trigger conditions, and the cloud management platform automatically checks the takeoff constraints. Once the constraints are met, the automated data collection process begins. Step 3, unattended automatic flight and panoramic data acquisition: Upon arrival of the mission time or receipt of the trigger command, the automatic hangar automatically opens the hatch, the UAV takes off autonomously and flies to each acquisition point in sequence according to the planned route, automatically performs multi-angle hovering shooting at each acquisition point, and generates raw image data. Step 4: Automatic Data Transmission and Cloud-Based Stitching Processing: After completing data collection, the drone automatically returns, lands, and connects to the automated hangar for charging. The raw image data is automatically uploaded to the cloud management platform via the hangar's network module. The cloud management platform automatically performs panoramic image stitching, geographic information registration, and image enhancement processing, and automatically archives the images according to naming rules. Specifically, upon receiving the raw images, the cloud management platform automatically performs the following processing: Panoramic image stitching: stitching 16 original images from each collection point into a complete spherical panoramic image; Geographic information registration: Read the coordinates recorded by the UAV RTK module and write them into the panoramic image EXIF ​​information; Image enhancement: Automatically performs color balance, dehazing, and sharpening. Automatic archiving: Name the images according to the naming rules and save them to the folder path.

[0023] Step 5: The cloud management platform performs one or more of the following analyses on the archived panoramic image: Compare panoramic images of the same collection point at different times on the same screen to generate a visual progress change report; By comparing real-world images with BIM models, early or late processes can be identified and early warnings can be generated. AI identifies potential safety and quality hazards, automatically generates alarm work orders, and pushes them to the responsible person to complete closed-loop handling. Step Six: Data Asset Accumulation Throughout the Construction Process: The data generated in Steps One through Five will be stored and accumulated into deliverable digital assets.

[0024] In this embodiment, the automated UAV hangar includes a take-off and landing (ADR) system, an energy management system, an intelligent control system, and a communication and scheduling module. The ADR system includes an automatic opening and closing hatch, a high-precision landing platform, and a centering / pushing mechanism; it protects the UAV from wind and rain, provides a platform for precise landing, and guides the UAV to a fixed position after landing. The energy management system includes an intelligent charging module, a mechanical battery swapping device, and a backup power supply; it automatically charges or quickly replaces the UAV's batteries to ensure continuous operation. The intelligent control system includes an industrial control computer, a programmable logic controller (PLC), and a flight control module; acting as the hangar's "brain," it controls hatch opening and closing, robotic arm movements, communicates with the UAV, and executes preset automated task flows. The communication and scheduling module includes 4G / 5G / private network communication, real-time positioning (RTK), and a weather station to receive remote commands, transmit data, and provide centimeter-level precise positioning guidance for the UAV. The weather station can monitor wind speed, rain and snow in real time to determine whether it is suitable for takeoff; environmental and safety protection includes temperature control system, chassis protection, and internal monitoring; to ensure stable operation of the hangar in extreme temperatures ranging from -30℃ to 50℃, and to monitor the internal status of the hangar in real time through built-in cameras.

[0025] In this embodiment, the large factory building in step one is a steel structure factory building; the panoramic data collection points cover key areas at each stage of construction, including the bottom area of ​​the deep foundation pit, the steel structure roof area, the material storage area, the component processing area, and the main construction transportation roads within the site. Each collection point records its spatial coordinates, hovering height, and multi-angle shooting parameters. Parameters include the camera gimbal tilt angle, such as -60°, -30°, 0°, +30°, depending on the gimbal's capabilities. Horizontal rotation angles are typically 0°, 90°, 180°, 270°, or multiple equally spaced angles. The number of shots per angle can be set to one or more bracketed exposures.

[0026] Specifically, in the cloud management platform, based on the project's overall construction plan, a total of 36 panoramic image data collection points were set up, distributed as follows: Bottom area of ​​deep foundation pit: 4 sampling points are located at the four corners of the foundation pit; the hovering height relative to the bottom of the foundation pit is 12 meters; the pitch angle is set to -60°, -30°, 0°, and +30°.

[0027] Steel structure roof area: 8 data collection points, arranged along the ridge line and at the four corners; the hovering height relative to the roof is 18 meters; the pitch angle is set to -30°, 0°, and +30°.

[0028] Material storage area: 6 collection points, hovering height 15 meters.

[0029] Component processing areas: 6 data collection points in the steel structure processing area, with a suspension height of 12 meters; 6 data collection points in the rebar processing area, with a suspension height of 12 meters.

[0030] Main construction and transportation routes within the site: 12 data collection points, one every 50 meters along the route, with a hovering height of 10 meters.

[0031] Each acquisition point records its spatial coordinates (RTK measured), hovering height, and multi-angle shooting parameters.

[0032] In this embodiment, the task execution cycle in step two is set to execute panoramic acquisition tasks at least twice a day at fixed times; before each task is triggered, the system automatically checks the takeoff constraints. When all conditions are met, the UAV takes off to execute the task; if any condition is not met, the task is delayed or canceled, and the administrator is notified. The conditions for triggering takeoff include at least the following: wind speed not exceeding a preset threshold, no rainfall, and drone battery power not lower than a preset minimum power threshold.

[0033] In this embodiment, in step three, for high-risk locations that are inconvenient for personnel to access, namely the bottom area of ​​the deep foundation pit and the steel structure roof area, a preset flight path is used to enable the drone to perform shooting parameters at designated collection points with lower flight altitude, higher overlap rate, and shorter collection cycle. This is used to replace manual labor for real-time safety and quality control and record keeping for future reference. Specifically, the designated collection points are collection points set up for high-risk, difficult-to-access areas that require high-frequency monitoring. The main parameters are adjusted as follows: the flight altitude is reduced to 2-5 meters, compared to the usual 10-15 meters; the distance between adjacent shooting positions is reduced to 2-3 meters, compared to the usual 10-20 meters, to achieve a higher image overlap rate; and the collection cycle is shortened to 4 times per day, compared to the usual 2 times per day.

[0034] In this project, project managers log in to the project management terminal via a PC web browser. After authentication by the data platform, they access the task planning interface of the cloud management platform. They select the deployed automated hangar and set the following tasks: Routine missions throughout the day: executed at 9:00 and 16:00, covering all 36 collection points, with an estimated flight time of 25 minutes; Special mission for high-risk areas: For the bottom area of ​​deep foundation pits and steel structure roof areas, two additional data collections will be conducted daily at 11:00 and 14:00. The flight altitude will be reduced to 3-4 meters, and the distance between adjacent shooting positions will be reduced to 2.5 meters to achieve a higher overlap rate.

[0035] After the task is saved, the system automatically generates the flight path and sends it to the UAV automated hangar.

[0036] An example of automated flight and data acquisition: a mission at 9:00 AM on a certain day. At 9:00, the system automatically checks the takeoff constraints: real-time wind speed 4.2 m / s, meeting the wind speed threshold ≤ 10 m / s; no rain; drone battery level 87%, meeting the minimum battery level threshold ≥ 60%. All conditions are met, and the hangar door automatically opens.

[0037] The drone took off autonomously and flew sequentially to 36 data collection points along a planned route. At each point, after hovering and stabilizing, the gimbal automatically captured images at preset pitch angle sequences (-60°, -30°, 0°, +30°) and horizontal rotation angles (0°, 90°, 180°, 270°), acquiring one raw image for each angle. The entire mission collected a total of 576 raw images (36 points × 4 pitch angles × 4 horizontal angles) over a flight time of 26 minutes.

[0038] After completing the mission, the drone automatically returns to its home base, achieving centimeter-level precision landing via RTK differential data link, and begins charging by connecting to the hangar's charging port. Raw image data is automatically uploaded to the cloud management platform via the hangar's 5G network module.

[0039] In this embodiment, the naming rule in step four is as follows: the data is automatically archived according to the naming rule of collection time + spatial location code, and a folder structure is established for archiving according to the collection point number and year and month, supporting multi-dimensional retrieval by time range, location area, and task batch; Step four also includes: automatically marking the spatial coordinates and process information of the corresponding locations in the panoramic map for concealed works, such as pipeline pre-embedding, electrical conduits, water supply and drainage pipes, fire protection pipes buried in walls or underground, and the layered construction of the foundation structure, such as foundation cushion layer → rebar binding → concrete pouring → waterproof layer and steel structure node welding, such as the welding nodes of steel columns and steel beams, and high-strength bolt connection points, to form visualized archive data that can be used for final acceptance and subsequent operation and maintenance.

[0040] In this embodiment, the panoramic image timeline comparison in step five includes: supporting simultaneous display of panoramic images from any two different times on the same screen, and automatically generating a visual progress comparison report annotating the differences in construction status changes; supporting sliding comparison of panoramic images from the same collection point over multiple consecutive days, forming a "daily-granular" visual growth record of the entire lifecycle of the construction area. The specific operation of the progress comparison analysis is as follows: For example, when the project reaches the fourth month, the manager selects the collection point of "steel structure roof area" in the project management terminal and retrieves panoramic images from days 90, 95, and 100 for simultaneous sliding comparison. The system automatically generates a progress change report, marking the specific time nodes for steel structure purlin installation and roof color steel plate laying, forming a "daily-granular" growth record of the entire roof construction process.

[0041] BIM model comparison includes: importing the project's BIM 3D model and construction schedule; using algorithms to spatially register and visually overlay the panoramic image captured from real-world footage with the BIM model; automatically identifying the current actual progress and comparing it with the planned progress; highlighting over- or under-progressed processes with color markings, flashing prompts, or list alerts, and pushing these to the project management terminal for managers to locate abnormal processes. For example, if the cloud management platform imports the project's BIM 3D model and construction schedule, and the plan is to complete 80% of the roof corrugated steel panel installation on day 100, the system will compare the panoramic image captured from real-world footage on day 100 with the BIM model, identifying that the actual installation progress is only 65%, lagging by 15 percentage points. The system will automatically mark the lagging area with a flashing red light, generate an alert, and push it to the project management terminal, allowing the project manager to coordinate additional roof work teams.

[0042] Safety and quality AI identification should recognize at least the following two types of potential hazards: Construction safety hazards include: not wearing safety protective equipment, cables dragging on the ground, uncovered bare soil, lack of edge protection, and fire hazards. Structural quality hazards include: cracks and leaks in concrete structures, visible defects in steel welds, and component misalignment. For example, in the panoramic image on day 95, the AI ​​recognition module automatically detected the following hazard: a safety hazard—two workers in the steel structure roof area were not wearing safety helmets. A quality-related hazard was identified: a weld at a steel structure node exhibited a visible porosity defect. The system automatically generated two alarm work orders, each including: hazard type, location coordinates, on-site image screenshot, and discovery time. These work orders were sent to the respective work accounts of the safety officer and quality inspector in their respective areas. The safety officer was required to rectify the issue within 2 hours, ensuring the wearing of a safety helmet and uploading a post-rectification image. The quality inspector was required to complete the weld repair and provide feedback within 4 hours. The system automatically recorded a closed-loop process; work orders not rectified within the specified timeframe were automatically escalated and sent to the project manager.

[0043] In this embodiment, the closed-loop handling process in step five is as follows: First, an alarm work order is generated, which must include the type of hazard, location information, on-site images, and discovery time; then, the work order is automatically pushed to the work account of the person in charge of the corresponding area; subsequently, the system receives feedback from the person in charge on the completion of rectification and on-site images after rectification; for work orders that have not been rectified within the time limit, the system automatically generates an escalation alarm. This forms a complete closed-loop management process of automatic discovery, background analysis, task assignment, rectification feedback, and review and archiving.

[0044] In this embodiment, step five also includes setting up independent monitoring routes for high-risk areas such as the bottom of the deep foundation pit and the steel structure roof, and performing special data collection at a higher frequency than in ordinary areas; the panoramic images of high-risk locations are marked as restricted access areas in the cloud management platform, and can only be viewed by authorized management personnel; when the edge AI identifies illegal operations in high-risk areas, it immediately triggers a real-time alarm and pushes it to the mobile phone of the safety management personnel.

[0045] In this embodiment, the digital assets in step six include a panoramic image database organized by time axis, a visual work card for concealed works organized by spatial location, a safety and quality closed-loop record organized by hazard type, and a BIM-real scene comparison report organized by process. After the project is completed, these assets are packaged and delivered to the construction unit for auxiliary review of completion acceptance and subsequent operation and maintenance query.

[0046] For example, during the foundation layer construction phase, before the foundation layer is poured, drones perform a special data collection task, taking high-precision photos of the locations of pre-buried electrical conduits and water supply and drainage pipes. The system automatically marks the pipe coordinates and process information in the panoramic image, such as "Area B, axis 3 intersects axis 4, water supply pipe pre-buried, DN100, accepted on May 20, 2026." This marking information forms a visual work card for concealed works, which can be directly retrieved and displayed during final acceptance, allowing verification of the pre-buried quality without damaging the ground. Two years later, during the plant's operation and maintenance phase, if a water supply pipe leak occurs, maintenance personnel can search for "Area B, axis 3 intersects axis 4" in the system to retrieve the panoramic image of that location before concealment, accurately locating the pipeline route and joint position, significantly shortening maintenance time.

[0047] The above embodiments are not exhaustive examples of specific implementation methods, and other embodiments are also possible. The purpose of the above embodiments is to illustrate the present invention, rather than to limit the scope of protection of the present invention. All applications derived from simple variations of the present invention fall within the scope of protection of the present invention.

Claims

1. A construction method for an automated acquisition and management system for panoramic images of large factory buildings based on an air-ground collaborative data platform, characterized in that: The automated acquisition and management system for panoramic images of large factory buildings includes an automated drone hangar, drones, a cloud management platform, an air-ground collaborative data platform, and a project management terminal. The automated hangar for unmanned aerial vehicles (UAVs) is deployed at a fixed site on the project site; the UAVs are connected to the automated hangars via a multimodal communication link, and the UAVs are equipped with a high-resolution panoramic camera, an edge computing module, and an RTK high-precision positioning module. The cloud management platform communicates with the drone automated hangar. The cloud management platform integrates a task planning and distribution module, an automatic flight path generation module, a panoramic image stitching and registration module, a panoramic image timeline comparison module, a BIM model comparison module, a safety and quality AI recognition module, and a hidden danger closed-loop management module. The drones are indirectly connected to the cloud management platform via an automated drone hangar as a gateway, or directly connected to the cloud management platform via a 4G / 5G network; the air-ground collaborative data platform serves as the system's data hub, deployed in the enterprise's internal data center or the cloud, and is used to achieve bidirectional data flow and standardized data processing between the automated drone hangar, drones, the cloud management platform, and the project management terminal. The project management terminal is deployed on the user side and communicates with the cloud management platform and / or the air-ground collaborative data platform. The automated hangar for drones receives automated inspection tasks from the cloud management platform, controls the drones to perform panoramic image acquisition along a preset route, and uploads the acquired data to the cloud management platform for automated processing and analysis after standardization by the air-ground collaborative data platform. The processing results are then pushed to the project management terminal. The construction method includes the following steps: Step 1: Deployment of automated drone hangars and pre-planning of flight routes: Select deployment points on the project master plan to set up automated drone hangars and connect them to the network; In the cloud management platform, set up multiple panoramic data collection points over the project site according to project management needs, and generate a set of waypoints and flight routes covering the entire construction area. Step 2: Project managers select the target automated hangar and preset route through the cloud management platform, set the task execution cycle or trigger conditions, and the cloud management platform automatically checks the takeoff constraints. Once the constraints are met, the automated data collection process begins. Step 3, unattended automatic flight and panoramic data acquisition: Upon arrival of the mission time or receipt of the trigger command, the automatic hangar automatically opens the hatch, the UAV takes off autonomously and flies to each acquisition point in sequence according to the planned route, automatically performs multi-angle hovering shooting at each acquisition point, and generates raw image data. Step 4, Automatic Data Backhaul and Cloud Stitching Processing: After completing the data collection, the drone automatically returns to home, lands, and connects to the automated hangar for charging; the raw image data is automatically uploaded to the cloud management platform through the network module of the automated hangar; the cloud management platform automatically performs panoramic image stitching, geographic information registration, and image enhancement processing, and automatically archives the data according to the naming rules. Step 5: The cloud management platform performs one or more of the following analyses on the archived panoramic image: Compare panoramic images of the same collection point at different times on the same screen to generate a visual progress change report; By comparing real-world images with BIM models, early or late processes can be identified and early warnings can be generated. AI identifies potential safety and quality hazards, automatically generates alarm work orders, and pushes them to the responsible person to complete closed-loop handling. Step Six: Data Asset Accumulation Throughout the Construction Process: The data generated in Steps One through Five will be stored and accumulated into deliverable digital assets.

2. The construction method of the automated acquisition and management system for panoramic images of large factories based on an air-ground collaborative data platform according to claim 1, characterized in that: In Step 1, the large factory building is a steel structure factory building; the panoramic data collection points set up cover the key areas of each stage of construction, including the bottom area of ​​the deep foundation pit, the steel structure roof area, the material storage area, the component processing area and the main construction transportation roads in the site, and each collection point records its spatial coordinates, hovering height and multi-angle shooting parameters.

3. The construction method of the automated acquisition and management system for panoramic images of large factories based on an air-ground collaborative data platform according to claim 1, characterized in that: In step two, the task execution cycle is set to perform panoramic data acquisition tasks at least twice a day at fixed times. Before each task is triggered, the system automatically checks the takeoff constraints. When all conditions are met, the UAV takes off to perform the task. If any condition is not met, the task will be delayed or canceled, and the administrator will be notified. The conditions for triggering takeoff include at least the following: wind speed not exceeding a preset threshold, no rainfall, and drone battery power not lower than a preset minimum power threshold.

4. The construction method of the automated acquisition and management system for panoramic images of large factories based on an air-ground collaborative data platform according to claim 1, characterized in that: In step three, for high-risk locations where it is inconvenient for personnel to enter and exit, namely the bottom area of ​​the deep foundation pit and the steel structure roof area, the drone is used to perform shooting parameters with lower flight altitude, higher overlap rate and shorter collection cycle at the special collection point by setting a flight path.

5. The construction method of the automated acquisition and management system for panoramic images of large factories based on an air-ground collaborative data platform according to claim 1, characterized in that: The naming rules in step four are as follows: archive automatically according to the naming rule of collection time + spatial location code, and create folder structure archives according to collection point number and year and month, supporting multi-dimensional retrieval by time range, location area, and task batch; Step four also includes: automatically marking the spatial coordinates and process information of the corresponding locations of concealed works and steel structure node welding in the panoramic view, forming a visual archive that can be used for final acceptance and subsequent operation and maintenance.

6. The construction method of the automated acquisition and management system for panoramic images of large factories based on an air-ground collaborative data platform according to claim 1, characterized in that: Step 5, the panoramic image timeline comparison, includes: supporting the simultaneous display of any two panoramic images from different times on the same screen, and automatically generating a visual progress comparison report that marks the differences in construction status changes; supporting the sliding comparison of panoramic images from the same collection point over multiple consecutive days, forming a "daily-granular" visual growth record of the entire life cycle of the construction area. BIM model comparison includes: importing the project's BIM 3D model and construction schedule; using algorithms to spatially register and visually overlay the panoramic images obtained from real-world data collection with the BIM model; automatically identifying the current actual progress and comparing the actual progress with the planned progress; highlighting advanced or lagging processes with color markings, flashing prompts, or list warnings, and pushing them to the project management terminal for managers to locate abnormal processes. Safety and quality AI identification should recognize at least the following two types of potential hazards: Construction safety hazards include: not wearing safety protective equipment, cables dragging on the ground, uncovered bare soil, lack of edge protection, and fire hazards. Structural quality hazards include: cracks and leaks in concrete structures, appearance defects in welds of steel structures, and component misalignment.

7. The construction method of the automated acquisition and management system for panoramic images of large factories based on an air-ground collaborative data platform according to claim 6, characterized in that: The closed-loop handling process in step five is as follows: First, an alarm work order is generated, which must include the type of hazard, location information, on-site images, and discovery time. Then, the work order is automatically pushed to the work account of the person in charge of the corresponding area. Subsequently, the system receives feedback from the person in charge on the completion of rectification and on-site images after rectification. For work orders that have not been rectified within the time limit, the system automatically generates an escalation alarm. This forms a complete closed-loop management process of automatic discovery, background analysis, task assignment, rectification feedback, and review and archiving.

8. The construction method of the automated acquisition and management system for panoramic images of large factories based on an air-ground collaborative data platform according to claim 6, characterized in that: Step five also includes setting up independent monitoring routes for high-risk areas such as the bottom of deep foundation pits and steel structure roofs, and performing special data collection at a higher frequency than in ordinary areas; panoramic images of high-risk locations are marked as restricted access areas in the cloud management platform, and can only be viewed by authorized personnel; when the edge AI identifies illegal operations in high-risk areas, it immediately triggers a real-time alarm and pushes it to the mobile phone of safety management personnel.

9. The construction method of the automated acquisition and management system for panoramic images of large factories based on an air-ground collaborative data platform according to claim 6, characterized in that: The digital assets in step six include a panoramic image database organized by timeline, a visual work card for concealed works organized by spatial location, a safety and quality closed-loop record organized by hazard type, and a BIM-real scene comparison report organized by process. After the project is completed, these assets are packaged and delivered to the construction unit for auxiliary review of completion acceptance and subsequent operation and maintenance inquiries.