Robot team collaborative management method and system for high-risk operations of power plants

Abstract

The application discloses a robot team-oriented high-risk operation of power station collaborative management method and system, relates to the intelligent management technical field, and the method comprises the following steps: segmenting a local model from a power station BIM model, aligning with a monitoring area, combining a heterogeneous team configuration to fit an inspection path, assigning a multi-machine cooperative scanning of a UAV, and acquiring multi-modal monitoring data. Data is analyzed by an edge hub to generate a dynamic risk map, combined with a four-legged robot detection capability matrix for task matching, and an inspection task sequence is output. The task sequence is optimized by compensation to complete risk collaborative investigation. The application solves the technical problems in the prior art that traditional power station inspection relies on manual or single equipment, has a monitoring blind area, and has low inspection efficiency, and cannot assess risks in real time, achieves the technical effects of cooperative operation of heterogeneous robots, comprehensive coverage of high-risk areas, improved inspection efficiency, and dynamic risk assessment and timely response.

Description

Technical Field

[0001] The present invention relates to the field of intelligent management technology, and in particular to a robot team-based collaborative management method and system for high-risk operations in power plants. Background Art

[0002] As power plant equipment continues to expand in scale and complexity, traditional manual inspections and single-use monitoring methods are no longer sufficient to meet the safety monitoring needs of high-risk operating areas. Traditional methods typically rely on fixed cameras and manual operation, resulting in blind spots, low efficiency, and false detections. This is particularly true in high-risk operating areas of power plants, where traditional monitoring equipment has limited viewing angles and coverage, making it impossible to capture all potential risks in real time. Furthermore, manual inspections are slow, unable to respond to sudden risks, and unable to cope with complex operating environments, severely impacting the safety management and operational efficiency of power plants. To address these issues, inspection systems based on robots, drones, and intelligent data analysis are gradually being adopted, but existing technologies still face challenges in path planning, task allocation, and risk assessment. Summary of the Invention

[0003] This application provides a robot team-based collaborative management method and system for high-risk operations in power plants, which is used to solve the technical problems in the existing technology that traditional power plant inspections rely on manual labor or single equipment, have blind spots in monitoring, low inspection efficiency, and cannot assess risks in real time.

[0004] The first aspect of the present application provides a robot team collaborative management method for high-risk operations in power plants, the method comprising: segmenting a first local BIM model from a power plant BIM model, wherein the first local BIM model is aligned with a first monitoring area in horizontal space; performing inspection coverage path fitting based on the heterogeneous team configuration of the first monitoring area and the first local BIM model to obtain a first group of monitoring coverage paths; H drones in the heterogeneous team configuration perform multi-machine collaborative scanning of the first monitoring area along the first group of monitoring coverage paths to obtain multimodal monitoring data; a first edge hub receives and performs risk coupling analysis based on the multimodal monitoring data to generate a dynamic risk map; constructing M detection capability matrices of the M quadruped robots in the heterogeneous team configuration, and performing capability-task matching on the dynamic risk map and the M detection capability matrices to output M inspection task sequences; performing adjacent area inspection compensation optimization on the M inspection task sequences, outputting M compensation task sequences, and performing risk collaborative investigation on the dynamic risk map.

[0005] The second aspect of the present application provides a robot team collaborative management system for high-risk operations in power plants, the system comprising: a model segmentation module for segmenting a first local BIM model from a power plant BIM model, wherein the first local BIM model is aligned with a first monitoring area in horizontal space; an inspection coverage path fitting module for fitting an inspection coverage path according to the heterogeneous team configuration of the first monitoring area and the first local BIM model to obtain a first set of monitoring coverage paths; a multi-machine collaborative scanning module for performing a first monitoring along the first set of monitoring coverage paths according to H drones in the heterogeneous team configuration. Multi-machine collaborative scanning of the area to obtain multimodal monitoring data; a risk coupling analysis module, which is used to receive the first edge center and perform risk coupling analysis based on the multimodal monitoring data to generate a dynamic risk map; a task matching module, which is used to construct M detection capability matrices of the M quadruped robots in the heterogeneous team configuration, and perform capability-task matching on the dynamic risk map and the M detection capability matrices, and output M inspection task sequences; a risk collaborative investigation module, which is used to perform adjacent area inspection compensation optimization on the M inspection task sequences, output M compensation task sequences, and perform risk collaborative investigation on the dynamic risk map.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] The present application provides a collaborative management method and system for robot teams in high-risk power plant operations, which relates to the field of intelligent management technology. The system segments the local model from the power plant BIM model, aligns it with the monitoring area, combines heterogeneous team configurations to fit the inspection path, assigns drones to scan and acquire multimodal monitoring data, and analyzes and generates dynamic risk maps. The system combines the capability matrix of quadruped robots to perform task matching and compensation optimization, ensuring efficient inspection and risk assessment of high-risk operation areas in power plants. This solves the technical problems in the prior art that traditional power plant inspections rely on manual labor or single equipment, have blind spots in monitoring, have low inspection efficiency, and cannot assess risks in real time. It achieves the technical effect of comprehensively covering high-risk areas, improving inspection efficiency, and achieving dynamic risk assessment and timely response through the collaborative operation of heterogeneous robots. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0009] Figure 1A flowchart of a collaborative management method for robot teams in high-risk power plant operations provided in an embodiment of the present application;

[0010] Figure 2 Schematic diagram of the structure of a robot team-based collaborative management system for high-risk operations in power plants provided in an embodiment of the present application.

[0011] Explanation of the accompanying drawings: model segmentation module 11, inspection coverage path fitting module 12, multi-machine collaborative scanning module 13, risk coupling analysis module 14, task matching module 15, risk collaborative investigation module 16. DETAILED DESCRIPTION

[0012] This application provides a robot team-based collaborative management method and system for high-risk operations in power plants, which is used to solve the technical problems in the existing technology that traditional power plant inspections rely on manual labor or single equipment, have blind spots in monitoring, low inspection efficiency, and cannot assess risks in real time.

[0013] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0014] It should be noted that the terms "first", "second", etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices.

[0015] Example 1, as Figure 1 As shown, the present application provides a robot team collaborative management method for high-risk operations in power plants, the method comprising:

[0016] P10: Segment a first local BIM model from the power plant BIM model, wherein the first local BIM model is aligned with the first monitoring area in horizontal space.

[0017] Specifically, the power plant's overall BIM (Building Information Model) must first be processed to segment the first local BIM model. BIM, short for Building Information Model, digitally creates a three-dimensional model of the building and its surroundings, incorporating various data such as the building's geometry, physical properties, and construction information. Through segmentation, a sub-model of a specific area within the power plant is extracted, creating the first local BIM model. This sub-model is used for further monitoring and risk analysis. Extracting the local BIM model can be achieved through a combination of Geographic Information Systems (GIS) and BIM technology, ensuring accurate extraction of data for specific areas within the larger power plant data, facilitating subsequent operations.

[0018] The core characteristic of the first local BIM model is its ability to represent the spatial information of a specific operational area within a power plant. Compared to the BIM model of the entire power plant, the local model's spatial scope is more specific and narrow, focusing on the specific area requiring monitoring. This enables subsequent monitoring route planning and risk identification to be carried out on a more targeted and efficient basis.

[0019] After the first local BIM model is segmented, it must be precisely aligned horizontally with the actual first monitoring zone. The first monitoring zone is the first monitoring area, and subsequent zones refer specifically to regions. Horizontal spatial alignment means that on a two-dimensional plane, the coordinate system of the local BIM model is completely consistent with the geographic coordinate system of the monitoring zone, ensuring that every point in the model accurately corresponds to a specific location in the actual monitoring area. The alignment process involves coordinate transformation and geographic information system (GIS) technology. By matching the coordinate system of the BIM model with the geographic coordinate system in GIS, coordinate transformation algorithms (such as affine transformation and projection transformation) are used to map the point, line, and surface elements in the BIM model to the geographic coordinate space of the GIS. This process requires the use of high-precision positioning equipment (such as GPS and RTK) to obtain the actual geographic coordinate data of the monitoring zone as a basis for alignment.

[0020] By completing this step, the first local BIM model of the power station can accurately reflect the spatial layout of the monitoring area and provide an accurate spatial basis for subsequent multi-machine collaborative scanning, dynamic risk map generation and other tasks.

[0021] P20: According to the heterogeneous team configuration of the first monitoring area and the first local BIM model, inspection coverage paths are fitted to obtain a first set of monitoring coverage paths.

[0022] Furthermore, step P20 in this embodiment of the present application further includes:

[0023] P21: Locally call the power plant's global control information; P22: Based on the power plant's global control information, perform collaborative perception area positioning in the first local BIM model to obtain a collaborative perception virtual boundary; P23: Extract the building height distribution information of the collaborative perception virtual boundary from the first local BIM model; P24: Use the heterogeneous team configuration as an equipment scheduling constraint, and use the building height distribution information and the collaborative perception virtual boundary as monitoring range constraints to perform flight trajectory fitting, and output the first set of monitoring coverage paths.

[0024] It should be understood that according to the first monitoring area and the first local BIM model of the power station, combined with the heterogeneous team configuration, the inspection coverage path is fitted to generate an optimized inspection path to ensure that the robot can effectively cover the monitoring area.

[0025] First, based on the heterogeneous team configuration of the first monitoring area and the first local BIM model, the inspection coverage route was fitted. Heterogeneous team configuration refers to the combination and configuration of different types of monitoring equipment (such as drones and quadruped robots) within the monitoring area. Each device has unique performance parameters, such as flight altitude, endurance, and sensor type. These parameters serve as equipment scheduling constraints and are used in the subsequent route planning process. The first local BIM model provides detailed building structure and spatial layout information within the monitoring area, including equipment and facility locations and obstacle distribution, providing the basic geographic data support for route planning.

[0026] Furthermore, the local system accesses global power plant management and control information. This information covers a wide range of aspects, including the overall power plant operating status, equipment distribution, and operational plans. This information is crucial for understanding the role and importance of monitoring zones within the entire power plant and provides macro-level guidance for subsequent path planning.

[0027] Next, based on the accessed global power plant control information, the collaborative sensing area is located within the first local BIM model, resulting in a collaborative sensing virtual boundary. A collaborative sensing area refers to the area within which different devices must collaborate during monitoring. By locating these areas, the scope and boundaries of device coordination are clearly defined, ultimately determining the collaborative sensing virtual boundary. This boundary serves as a constraint on the monitoring range and is used in the subsequent flight trajectory fitting process.

[0028] Next, the building height distribution information for the collaboratively perceived virtual boundary is extracted from the first local BIM model. This building height distribution information reflects the height variations of buildings within the monitoring area, which is particularly important for drone flight altitude planning. During monitoring missions, drones need to adjust their flight altitude based on building heights to avoid collisions and ensure accurate monitoring data.

[0029] Finally, flight trajectories are fitted, using heterogeneous team configurations as equipment scheduling constraints and building height distribution information and collaborative sensing virtual boundaries as monitoring range constraints. This process comprehensively considers equipment performance parameters, the spatial layout within the monitoring area, and collaborative work requirements. Using a path planning algorithm, the first set of monitoring coverage paths is output. These paths will guide drones and quadruped robots to efficiently and safely perform monitoring tasks within the monitoring area, ensuring comprehensive coverage of the entire area while meeting the collaborative work requirements between devices.

[0030] Through the above steps, the entire inspection route planning process not only takes into account the actual layout and monitoring requirements of the power plant, but also incorporates factors such as building height and robot configuration to ensure the accuracy and feasibility of route planning.

[0031] Furthermore, step P22 of the embodiment of the present application further includes:

[0032] P22-1: Decompose the power plant's global control information to obtain power plant monitoring deployment information and power plant airspace restriction information. P22-2: Define the monitoring range distribution in the first local BIM model based on the power plant monitoring deployment information, and reversely select the first compensation blind area distribution based on the defined results. P22-3: Filter and define the collaborative sensing virtual boundary in the first compensation blind area distribution based on the power plant airspace restriction information. The power plant monitoring deployment information includes fixed camera deployment data, temperature and humidity sensor deployment data, and gas detector deployment data.

[0033] Optionally, to more accurately determine the collaborative perception virtual boundary, further refinement can be performed. First, by decomposing the power plant's global control information, the power plant's monitoring deployment information and airspace restriction information are extracted. Power plant monitoring deployment information includes the deployment data of static monitoring equipment deployed within the power plant, such as the deployment location and number of fixed cameras, temperature and humidity sensors, and gas detectors. These devices are used to monitor the environment and equipment conditions within the power plant to ensure that key parameter data can be obtained in real time. Airspace restriction information includes no-fly zones within the power plant area, such as the demarcation of no-fly zones, drone flight altitude restrictions, and flight ranges. This information is the restriction that must be followed during drone inspection missions.

[0034] Next, based on the data obtained from the power plant monitoring layout information, the monitoring range distribution is outlined in the first local BIM model. In other words, the coverage of the static monitoring equipment is mapped to the BIM model to form a visual representation of the monitoring area. The purpose of outlining the monitoring range is to clarify which areas are effectively covered by the deployed monitoring equipment and ensure that these areas can be effectively monitored in real time. Based on the results of the monitoring layout, the distribution of the first compensation blind spots is further reversed and screened, that is, those areas not covered by existing monitoring equipment. These blind spots may have monitoring blind spots, and due to the limitations of the equipment layout, they cannot be effectively monitored by existing equipment. Therefore, the screening of compensation blind spots provides an important basis for the planning of subsequent drone and robot missions.

[0035] Next, based on the power plant's airspace restriction information, the collaborative sensing virtual boundary is screened and defined within the first compensation blind spot distribution. This boundary encompasses the monitoring blind spot but not the no-fly zone. This virtual boundary must not only cover the compensation blind spot but also avoid conflict with the power plant's no-fly airspace. No-fly zones are areas that drones cannot enter, potentially due to security, privacy, or other restrictions. Therefore, using airspace restriction information, the collaborative sensing virtual boundary is ensured to encompass the monitoring blind spot while complying with flight restrictions, thereby ensuring the safety and legality of drone missions.

[0036] Through these steps, the system can fully utilize existing monitoring equipment, accurately determine uncovered blind spots, and plan feasible flight paths, making inspection tasks more comprehensive and efficient, while avoiding interference from flight conflict areas, thereby achieving efficient power plant safety monitoring and management.

[0037] Furthermore, step P22-2 of the embodiment of the present application further includes:

[0038] P22-21: Based on the fixed camera deployment data, temperature and humidity sensor deployment data, and gas detector deployment data, coverage range mapping is performed in the first local BIM model to obtain a multimodal monitoring range distribution boundary; P22-22: Solve the intersection of the multimodal monitoring range distribution boundaries and output the effective monitoring range distribution boundary; P22-23: Locate the first monitoring boundary in the first local BIM model; P22-24: Remove the effective monitoring range distribution boundary from the first monitoring boundary to obtain the first compensation blind spot distribution.

[0039] Optionally, in order to accurately identify monitoring blind spots within the power plant's operating area that have not yet been covered by static monitoring equipment, coverage mapping is first performed in the first local BIM model based on the deployment data of fixed cameras, temperature and humidity sensors, and gas detectors. Specifically, the system constructs a perception volume model of each type of equipment in three-dimensional space based on parameters such as the installation coordinates, monitoring angle, effective perception distance, and coverage radius of each type of monitoring equipment, and projects it into the first local BIM model, thereby forming a complete set of multimodal monitoring range distribution boundaries. The boundaries formed by each type of equipment reflect the scope of its real-time monitoring capabilities in space, and are the prerequisite for subsequent compensation blind spot identification.

[0040] After completing the sensing range mapping, the intersection of the spatial boundary areas formed by different sensor types is further solved. The purpose of the intersection operation is to identify areas covered by multiple sensors. These overlapping areas not only have higher sensing stability and data redundancy, but can also be considered as areas with sufficient risk monitoring capabilities. The effective monitoring range distribution boundary output by the intersection operation serves as the sum of the spatial coverage of the current static equipment, laying the spatial foundation for the subsequent isolation of blind spots in monitoring.

[0041] On this basis, the complete boundary required to be covered by this inspection mission is located in the first local BIM model, namely the first monitoring boundary. This boundary is usually pre-set by the task planning system based on the risk level, power plant operation plan, and operation and maintenance requirements, and represents the full coverage target of the heterogeneous robot team for this operation. The system then performs a set difference operation on the first monitoring boundary as the full set, removing the identified effective monitoring range distribution boundaries from it. The resulting area is the first compensation blind spot distribution, which reflects the critical spatial areas that are not currently covered by static equipment but are within the mission scope.

[0042] These compensated blind spots will serve as key target areas for subsequent dynamic inspections and real-time perception by drones or mobile robots. The entire identification process is based on high-precision spatial data and equipment deployment parameters from the BIM model, combined with geometric modeling, spatial overlay, and boundary difference calculations to ensure the accuracy and feasibility of the extracted blind spots, providing rigorous data support for the scientific delineation of collaborative perception virtual boundaries.

[0043] Furthermore, step P24 of the embodiment of the present application further includes:

[0044] P24-1: Fit the flight trajectory based on the building height distribution information and the collaborative perception virtual boundary, and output P benchmark flight trajectories; P24-2: Extract H flight capability matrices of the H drones from the heterogeneous team configuration; P24-3: Allocate the P benchmark flight trajectories based on the H flight capability matrices, and output H groups of benchmark flight trajectories, wherein the flight capability matrix includes flight altitude limit and flight endurance characteristics; P24-4: Smoothly connect the H groups of benchmark flight trajectories, output H monitoring coverage paths, and constitute the first group of monitoring coverage paths.

[0045] Specifically, to more accurately generate the first set of monitoring coverage paths, building height distribution information and a collaborative perception virtual boundary are first extracted from the first local BIM model. The building height distribution information details changes in building heights within the monitoring area, including changes in the highest and lowest points, as well as intermediate heights. The collaborative perception virtual boundary clarifies the areas that the drone needs to focus on monitoring. These areas are typically blind spots and do not fall within no-fly zones. Using this information, a path planning algorithm generates P baseline flight trajectories. The path planning algorithm needs to consider height restrictions to ensure that the drone's flight altitude is above building height to avoid collisions. At the same time, it must ensure that the flight trajectory covers all key areas within the collaborative perception virtual boundary and optimize the flight path to reduce unnecessary detours and improve monitoring efficiency.

[0046] Next, we extract the flight capability matrices of H drones from the heterogeneous fleet configuration. Each flight capability matrix contains key parameters such as altitude limits, flight endurance characteristics, payload capacity, and flight speed. These parameters will serve as an important basis for subsequent task allocation, ensuring that each drone's mission matches its capabilities.

[0047] Then, based on the H flight capability matrices, P benchmark flight trajectories are allocated. This allocation process comprehensively considers factors such as altitude matching, endurance matching, payload matching, and mission priority. Using a suitable allocation algorithm, such as a greedy algorithm, genetic algorithm, or other optimization algorithm, the P benchmark flight trajectories are assigned to the H drones, generating H sets of benchmark flight trajectories. Ensure that the flight trajectory assigned to each drone is within its flight altitude limit, that the flight trajectory length is within the drone's endurance, and that the drone can carry the sensors and other equipment necessary to complete the mission. Prioritize flight missions in critical areas to ensure that high-risk areas receive priority monitoring.

[0048] Finally, the baseline flight trajectory assigned to each drone is smoothed. The purpose of smoothing is to eliminate sudden changes and discontinuities between trajectories, ensuring that the drone can transition smoothly when performing its mission and avoiding flight instability or discontinuous monitoring data caused by sudden changes in the trajectory. The spliced ​​trajectories are smoothed using mathematical smoothing algorithms, such as Bezier curves and spline curves, to ensure smoothness and continuity. At the same time, the smoothed trajectories are safety checked to ensure that they meet flight safety requirements, such as avoiding buildings and obstacles. Ultimately, H smoothed monitoring coverage paths are generated. These paths not only cover all key risk points in the monitoring area, but also ensure the safety and stability of the drone during flight. The H monitoring coverage paths together constitute the first set of monitoring coverage paths, providing a scientific and reasonable flight path for subsequent monitoring tasks.

[0049] Through the above steps, through reasonable flight trajectory fitting, flight capability matrix extraction and task allocation, and smooth trajectory connection, the first set of monitoring coverage paths finally obtained can efficiently and safely guide the UAV to complete the inspection tasks in the high-risk operation areas of the power station.

[0050] P30: The H drones in the heterogeneous team configuration perform multi-machine collaborative scanning of the first monitoring area along the first set of monitoring coverage paths to obtain multimodal monitoring data.

[0051] Furthermore, step P30 in the embodiment of the present application further includes:

[0052] P31: The H drones perform multi-machine collaborative scanning of the first monitoring area along the H monitoring coverage paths to obtain H groups of multimodal data sequences, wherein each group of multimodal data sequences includes thermodynamic profile data, spatial point cloud sequence and gas spatiotemporal distribution matrix; P32: According to the acquisition window of the H groups of multimodal data sequences, the local multimodal monitoring storage is divided to obtain multimodal distribution data; P33: The H groups of multimodal data sequences and multimodal distribution data are spatiotemporally aligned to obtain the multimodal monitoring data.

[0053] It should be understood that the H drones in the heterogeneous team configuration collaboratively scan along the pre-planned first set of monitoring coverage paths, ensuring coverage of the entire first monitoring area and acquiring multimodal monitoring data through collaborative work. Each drone, based on its flight path, collects relevant multimodal monitoring data covering multiple dimensions, including thermodynamic profile data, spatial point cloud sequences, and gas spatiotemporal distribution matrices. Thermodynamic profile data is used to monitor temperature distribution and abnormal heat sources, spatial point cloud sequences help create a three-dimensional model of the area, and the gas spatiotemporal distribution matrix provides information on changes in gas concentration, particularly for monitoring hazardous gas leaks.

[0054] After acquiring monitoring data from each drone, local multimodal monitoring data is segmented for storage based on each data set's acquisition window, specifically the time period and spatial range of data collection. An acquisition window refers to the time period within a specific timeframe during which a drone collects data. By segmenting the locally stored monitoring data according to these acquisition windows, we ensure temporal consistency between acquired data and drone-collected data, allowing different types of monitoring data to be efficiently stored and organized based on their temporal and spatial relevance.

[0055] Next, all H sets of multimodal data sequences are spatiotemporally aligned with the already organized multimodal distribution data. Spatiotemporal alignment involves calibrating data collected from different sources and at different times in time and space so that they can be analyzed and processed within a unified spatiotemporal framework. This process involves precisely matching and adjusting the data's timestamps and spatial coordinates to eliminate data inconsistencies caused by factors such as differences in acquisition equipment and deviations in flight trajectories. After spatiotemporal alignment, the resulting multimodal monitoring data will be highly accurate and consistent, providing reliable data support for subsequent risk coupling analysis and the generation of dynamic risk maps.

[0056] Through these steps, not only can the coordinated scanning of drones along the monitoring coverage path be achieved, and rich multimodal monitoring data be obtained, but also the quality and availability of the data can be ensured by segmenting the data acquisition window and aligning it in time and space, providing a solid data foundation for the effective implementation of the entire monitoring management method.

[0057] P40: The first edge hub receives and performs risk coupling analysis based on the multimodal monitoring data to generate a dynamic risk map.

[0058] Furthermore, step P40 in this embodiment of the present application further includes:

[0059] P41: Decompose the multimodal monitoring data to obtain thermodynamic field data, three-dimensional geometric field data and gas concentration field data; P42: Predefine multimodal risk characteristics, wherein the multimodal risk characteristics are composed of temperature anomaly threshold, structural displacement safety threshold and gas concentration anomaly threshold; P43: Use the temperature anomaly threshold and gas concentration anomaly threshold to map traverse the thermodynamic field data and gas concentration field data to obtain temperature anomaly node distribution and gas anomaly node distribution; P44: After aligning the three-dimensional geometric field data space to the first local BIM model, use the structural displacement safety threshold to perform structural deformation judgment to obtain structural anomaly node distribution; P45: Perform risk coupling analysis on the temperature anomaly node distribution, gas anomaly node distribution and structural anomaly node distribution to generate the dynamic risk map.

[0060] Optionally, the first edge hub receives multimodal monitoring data from multiple drones and performs risk coupling analysis based on this data, ultimately generating a dynamic risk map. This map provides real-time risk assessments for high-risk operating areas within the power plant, facilitating timely response measures.

[0061] To perform this task, the multimodal monitoring data is first decomposed. For example, based on the different physical properties of the monitored data, complex datasets are split into subsets that are easier to process and analyze. This results in different types of data, including thermodynamic field data, 3D geometric field data, and gas concentration field data. Thermodynamic field data primarily reflects temperature variations within the region, 3D geometric field data provides information on the spatial layout and morphology of the power plant area, and gas concentration field data reflects the distribution of gases, particularly the concentration of hazardous gases.

[0062] Next, multimodal risk signatures are predefined based on monitoring requirements and safety standards. These signatures include temperature anomaly thresholds, structural displacement safety thresholds, and gas concentration anomaly thresholds. These thresholds are set based on the safety requirements of the power plant facility and determine the degree of change or anomaly that constitutes a potential safety risk. For example, when the temperature exceeds a preset threshold, it may indicate equipment overheating or failure, abnormal gas concentration may indicate a leak risk, and structural displacement exceeding safety limits may indicate damage to the building or equipment.

[0063] Furthermore, using temperature and gas concentration anomaly thresholds, the system traverses the thermodynamic field data and gas concentration field data, mapping and identifying the distribution of temperature and gas anomaly nodes. By comparing these with pre-set thresholds, it identifies areas where temperatures exceed safety standards and gas concentrations exceed safety standards, thereby marking potential risk areas and providing a basis for further risk assessment.

[0064] Simultaneously, the 3D geometric field data is spatially aligned with the primary local BIM model to ensure that the data matches the primary local BIM model of the power plant. This aligned data accurately reflects the actual structure of the power plant, facilitating subsequent detailed assessment of structural deformation. Using the structural displacement safety threshold, deformation of power plant equipment and buildings can be examined, revealing the distribution of abnormal structural nodes, which may represent safety hazards due to structural damage or movement.

[0065] Finally, a risk coupling analysis is performed on the distribution of temperature, gas, and structural anomaly nodes. This integration of different risk data types evaluates the interrelationships between anomaly nodes and generates a comprehensive dynamic risk map. This map reflects the risk status of the power plant's operating area in real time and dynamically adjusts based on different risk sources, providing a basis for decision-making for power plant managers and ensuring timely and effective safety measures are taken in high-risk operating environments.

[0066] Through these steps, it is possible to combine multiple monitoring data to accurately assess the risk status of the power plant, and through the generated dynamic risk map, real-time monitoring and management of high-risk areas can be achieved, greatly improving the ability to ensure safe operation of the power plant.

[0067] Furthermore, step P45 of the embodiment of the present application further includes:

[0068] P45-1: Predefine the abnormal node fusion distance; P45-2: After spatially aligning the temperature abnormal node distribution, gas abnormal node distribution, and structural abnormal node distribution based on the IPC algorithm, use the abnormal node fusion distance to cluster the abnormal nodes to obtain the coupled risk node distribution; P45-3: Align the coupled risk node distribution to the first local BIM model and locate O power plant equipment; P45-4: Based on the risk conduction characteristics of the O power plant equipment, construct the risk conduction topology of the O coupled risk nodes in the coupled risk node distribution, and output the dynamic risk map. The O coupled risk nodes in the coupled risk node distribution have O abnormal feature identifiers, each of which includes an abnormal device ID, a temperature abnormality, and / or a concentration abnormality, and / or a deformation abnormality.

[0069] In a possible embodiment of the present application, the risk coupling analysis process can be further refined to generate a more accurate and practical dynamic risk map.

[0070] When conducting risk coupling analysis, the anomaly node fusion distance is predefined. This fusion distance is a key parameter, set based on the layout of power plant equipment and the monitoring accuracy requirements. It is used to measure the spatial relationship between different anomaly nodes, especially the interrelationships between multiple risk types (such as temperature anomalies, gas anomalies, and structural anomalies). The anomaly node fusion distance is used to determine which nodes are spatially close and should be considered as the same group of risk sources. In power plant risk assessment, a reasonable fusion distance can help the system aggregate multiple interrelated risk points into a larger risk area, thereby improving monitoring and response efficiency.

[0071] Next, the IPC algorithm was used for spatial alignment, aligning the distribution of temperature anomaly nodes, gas anomaly nodes, and structural anomaly nodes to ensure that these data can be compared and analyzed in the same spatial coordinate system. The IPC algorithm (Iterative Closest Point) is an algorithm commonly used for point cloud data registration and spatial alignment. It aligns data from different sources to a common reference frame through iterative optimization, ensuring accurate matching between data. The aligned data is then clustered according to a predefined anomaly node fusion distance, grouping nodes that are spatially close and have similar anomaly characteristics to obtain a coupled risk node distribution. These coupled risk nodes represent the intersection of multiple risk factors in the power plant, potentially representing complex areas of excessive temperature, excessive gas concentration, or structural damage.

[0072] Next, the resulting coupled risk node distribution is aligned to the first local BIM model, ensuring that the spatial coordinates of all risk nodes are consistent with the actual facility layout of the power plant. This alignment allows the system to accurately locate each of the 0 power plant devices within the plant and associate them with the corresponding risk nodes, ensuring that the status and location of each device are reflected in the risk map.

[0073] Finally, based on the risk transmission characteristics of the O located power plant equipment, a risk transmission topology of O coupled risk nodes was constructed. Risk transmission characteristics refer to the likelihood and path of risk propagation between devices. These characteristics are predefined based on information such as the physical connections, functional relationships, and historical fault data of the devices. By analyzing the operating principles and risk transmission paths of the devices, a topological structure reflecting the spread of risk is constructed, demonstrating the likelihood and path of risk transmission from one node to other nodes. This can intuitively demonstrate the risk transmission path and impact range between power plant equipment, providing a scientific basis for risk management and decision-making.

[0074] In the output dynamic risk map, each of the eight coupled risk nodes is labeled with a corresponding abnormal feature identifier. Each identifier includes the abnormal device ID, abnormal temperature, abnormal gas concentration, and / or abnormal deformation. These feature identifiers provide specific abnormality information, helping operations and maintenance personnel quickly locate and address issues. Through this series of steps, the system not only identifies risk hotspots within the power plant but also clarifies the relationships between individual risk nodes and the risk transmission paths, providing comprehensive risk monitoring and analysis capabilities for the power plant.

[0075] P50: Construct M detection capability matrices of the M quadruped robots in the heterogeneous team configuration, perform capability-task matching on the dynamic risk map and the M detection capability matrices, and output M inspection task sequences.

[0076] Furthermore, step P50 in the embodiment of the present application further includes:

[0077] P51: Calculate the similarity between the O abnormal feature identifiers and the M detection capability matrices, and based on the calculation results, perform capability-task matching on the O coupled risk nodes and the M quadruped robots, and output M groups of abnormal detection tasks, wherein the detection capability matrix includes equipment coverage type and defect coverage type; P52: Serialize the M groups of abnormal detection tasks according to the dynamic risk map, and output the M inspection task sequences.

[0078] Optionally, the task allocation process of the quadruped robots can be further refined to ensure efficient and accurate risk detection. First, for the M quadruped robots in the heterogeneous team configuration, a detection capability matrix is ​​constructed for each of them. The detection capability matrix is ​​an important quantitative description of the performance of the quadruped robot, which includes two key dimensions: equipment coverage type and defect coverage type. Equipment coverage type refers to the types of power station equipment that the robot can detect, such as transformers, transmission lines, etc.; defect coverage type refers to the types of defects that the robot can identify, such as temperature anomalies, gas leaks, structural deformations, etc. By constructing the detection capability matrix, we can clearly understand the advantages and limitations of each quadruped robot in risk detection.

[0079] Next, similarities are calculated between the O anomaly feature identifiers and the M detection capability matrices. These anomaly feature identifiers include potential risk nodes within the power plant, such as equipment failures, overtemperatures, and gas leaks. By calculating the similarity between these identifiers and the detection capability matrix, the system can determine whether each robot is suitable for performing a specific task. Similarity can be calculated using a variety of algorithms, such as cosine similarity based on feature vectors or Euclidean distance, depending on the representation of the anomaly feature identifiers and the detection capability matrix. For example, some robots may be more sensitive to detecting high temperatures or hazardous gases, while others may be more suitable for visual inspection of equipment. Based on the calculation results, the system matches the most appropriate robot to each risk node, ensuring the rationality of task allocation. The output M sets of anomaly detection tasks will assign specific tasks to each robot, ensuring that the robot performs within its capabilities while maximizing overall operational efficiency.

[0080] Finally, M groups of anomaly detection tasks are sequenced according to the dynamic risk map, outputting M inspection task sequences. The dynamic risk map not only displays the distribution of coupled risk nodes but also includes risk transmission relationships and priority information. Through sequencing, anomaly detection tasks are arranged in a specific order to form an inspection task sequence. Sequencing can be based on factors such as risk severity, task urgency, and the robot's location, ensuring that the quadruped robot efficiently covers all key risk points when performing its tasks.

[0081] Through the above steps, the system can achieve efficient inspection task allocation based on the matching of robot capabilities and tasks, ensuring that each quadruped robot can work in the area it is best at, maximizing resource utilization and improving overall inspection efficiency.

[0082] P60: Optimize adjacent area inspection compensation for the M inspection task sequences, output M compensation task sequences, and conduct collaborative risk investigation on the dynamic risk map.

[0083] Specifically, after completing the initial planning of M inspection task sequences, these task sequences need to be optimized for adjacent zone inspection compensation. This optimization process aims to address the potential duplication or omission of tasks between adjacent monitoring zones, ensuring that the quadruped robot can seamlessly execute tasks and avoiding risk-based blind spots caused by zone boundaries.

[0084] First, we analyze the distribution of tasks within the M inspection task sequences and identify the task boundaries between adjacent monitoring zones. Since monitoring zones may overlap or have gaps, we need to pay special attention to these boundary areas. By analyzing the location information and task content within the task sequences, we can determine which tasks are located at the intersection of adjacent zones.

[0085] Next, compensation optimization is performed on tasks located at the intersection of adjacent zones. This process involves adjusting the order and content of tasks in the task sequence to ensure that the quadruped robot can cover all key risk points when performing tasks, while avoiding repeated execution of the same tasks. Specific methods of compensation optimization can include task splitting, task merging, and task reallocation. For example, if a task involves two adjacent zones, it can be split into two subtasks, each performed by a quadruped robot in the two zones; if there is duplication in the tasks of two adjacent zones, they can be merged into one task and completed by a single quadruped robot.

[0086] After the compensation optimization is complete, M compensation task sequences are output. These optimized compensation task sequences better adapt to the task requirements between adjacent monitoring zones, ensuring that the quadruped robot can efficiently and accurately complete risk screening during its mission. The compensation task sequences not only consider the task coverage but also optimize the execution order and resource allocation, thereby improving the overall efficiency of task execution.

[0087] Finally, a dynamic risk map is used to collaboratively investigate risks based on the M compensation task sequences. This dynamic risk map provides real-time information on the distribution and evolution of risks within the monitoring area. As the quadruped robots execute tasks according to the compensation task sequences, they can collaboratively investigate risks based on the risk information in the map. During this collaborative investigation, the quadruped robots can share monitoring data in real time, enabling them to promptly identify and address emerging risks, ensuring comprehensive and timely risk control within the monitoring area.

[0088] In summary, the embodiments of the present application have at least the following technical effects:

[0089] This application separates the first local BIM model from the power plant BIM model, aligns it with the monitoring area, combines the heterogeneous team configuration to fit the inspection coverage path, assigns drones to perform multi-machine collaborative scanning, and obtains multimodal monitoring data. The first edge center performs risk coupling analysis based on these data and generates a dynamic risk map. Subsequently, a quadruped robot detection capability matrix is ​​constructed to perform capability-task matching and output an inspection task sequence. Through adjacent area inspection compensation optimization, the compensation task sequence is finally output to complete the risk collaborative investigation of the dynamic risk map, ensuring efficient and comprehensive power plant inspection and risk assessment.

[0090] The technical effect of comprehensively covering high-risk areas, improving inspection efficiency, and realizing dynamic risk assessment and timely response has been achieved through the collaborative operation of heterogeneous robots.

[0091] Example 2, based on the same inventive concept as the robot team collaborative management method for high-risk operations in power plants in the previous embodiment, Figure 2 As shown, this application provides a robot team collaborative management system for high-risk operations in power plants. The system and method embodiments in the embodiments of this application are based on the same inventive concept. The system includes:

[0092] The model segmentation module 11 is configured to segment a first local BIM model from the power plant BIM model, wherein the first local BIM model is aligned with the first monitoring area in horizontal space.

[0093] The inspection coverage path fitting module 12 is used to perform inspection coverage path fitting based on the heterogeneous team configuration of the first monitoring area and the first local BIM model to obtain a first set of monitoring coverage paths.

[0094] The multi-machine collaborative scanning module 13 is used to perform multi-machine collaborative scanning of the first monitoring area along the first set of monitoring coverage paths according to the H drones in the heterogeneous team configuration to obtain multimodal monitoring data.

[0095] The risk coupling analysis module 14 is used to perform risk coupling analysis based on the multimodal monitoring data received by the first edge hub and to generate a dynamic risk map.

[0096] The task matching module 15 is used to construct M detection capability matrices of the M quadruped robots in the heterogeneous team configuration, perform capability-task matching on the dynamic risk map and the M detection capability matrices, and output M inspection task sequences.

[0097] The risk collaborative investigation module 16 is used to perform adjacent area inspection compensation optimization on the M inspection task sequences, output M compensation task sequences, and perform risk collaborative investigation on the dynamic risk map.

[0098] Furthermore, the inspection coverage path fitting module 12 is further configured to perform the following steps:

[0099] The global control information of the power plant is locally called; the collaborative perception area is positioned in the first local BIM model based on the global control information of the power plant to obtain a collaborative perception virtual boundary; the building height distribution information of the collaborative perception virtual boundary is extracted from the first local BIM model; the flight trajectory is fitted using the heterogeneous team configuration as an equipment scheduling constraint, and the building height distribution information and the collaborative perception virtual boundary as monitoring range constraints, and the first set of monitoring coverage paths is output.

[0100] Furthermore, the inspection coverage path fitting module 12 is further configured to perform the following steps:

[0101] By decomposing the power plant's global control information, power plant monitoring deployment information and power plant airspace restriction information are obtained; based on the power plant monitoring deployment information, the monitoring range distribution is framed in the first local BIM model, and the first compensation blind area distribution is reversely selected based on the framed results; and based on the power plant airspace restriction information, the collaborative sensing virtual boundary is framed and selected in the first compensation blind area distribution. The power plant monitoring deployment information includes fixed camera deployment data, temperature and humidity sensor deployment data, and gas detector deployment data.

[0102] Furthermore, the inspection coverage path fitting module 12 is further configured to perform the following steps:

[0103] Perform flight trajectory fitting based on the building height distribution information and the collaborative sensing virtual boundary, and output P benchmark flight trajectories; extract H flight capability matrices of the H UAVs from the heterogeneous team configuration; allocate the P benchmark flight trajectories based on the H flight capability matrices, and output H groups of benchmark flight trajectories, wherein the flight capability matrices include flight altitude limits and flight endurance characteristics; smoothly connect the H groups of benchmark flight trajectories, and output H monitoring coverage paths, forming the first group of monitoring coverage paths. Perform flight trajectory fitting based on the building height distribution information and the collaborative sensing virtual boundary, and output P benchmark flight trajectories; extract H flight capability matrices of the H UAVs from the heterogeneous team configuration; allocate the P benchmark flight trajectories based on the H flight capability matrices, and output H groups of benchmark flight trajectories, wherein the flight capability matrices include flight altitude limits and flight endurance characteristics; smoothly connect the H groups of benchmark flight trajectories, and output H monitoring coverage paths, forming the first group of monitoring coverage paths.

[0104] Furthermore, the inspection coverage path fitting module 12 is further configured to perform the following steps:

[0105] Based on the fixed camera deployment data, the temperature and humidity sensor deployment data, and the gas detector deployment data, coverage range mapping is performed in the first local BIM model to obtain a multimodal monitoring range distribution boundary; the intersection of the multimodal monitoring range distribution boundaries is solved to output an effective monitoring range distribution boundary; the first monitoring boundary is located in the first local BIM model; the effective monitoring range distribution boundary is removed from the first monitoring boundary to obtain the first compensation blind spot distribution.

[0106] Furthermore, the multi-machine collaborative scanning module 13 is further configured to perform the following steps:

[0107] The H drones perform multi-machine collaborative scanning of the first monitoring area along the H monitoring coverage paths to obtain H groups of multimodal data sequences, wherein each group of multimodal data sequences includes thermodynamic profile data, a spatial point cloud sequence, and a gas spatiotemporal distribution matrix; according to the acquisition windows of the H groups of multimodal data sequences, the local multimodal monitoring storage is divided to obtain multimodal distribution data; and the H groups of multimodal data sequences and the multimodal distribution data are spatiotemporally aligned to obtain the multimodal monitoring data.

[0108] Furthermore, the risk coupling analysis module 14 is further configured to perform the following steps:

[0109] Decomposing the multimodal monitoring data to obtain thermodynamic field data, three-dimensional geometric field data, and gas concentration field data; predefining a multimodal risk feature, wherein the multimodal risk feature is composed of a temperature anomaly threshold, a structural displacement safety threshold, and a gas concentration anomaly threshold; using the temperature anomaly threshold and the gas concentration anomaly threshold to map and traverse the thermodynamic field data and the gas concentration field data to obtain a temperature anomaly node distribution and a gas anomaly node distribution; after spatially aligning the three-dimensional geometric field data to the first local BIM model, using the structural displacement safety threshold to perform structural deformation judgment to obtain a structural anomaly node distribution; performing risk coupling analysis on the temperature anomaly node distribution, the gas anomaly node distribution, and the structural anomaly node distribution to generate the dynamic risk map.

[0110] Furthermore, the risk coupling analysis module 14 is further configured to perform the following steps:

[0111] Predefine an abnormal node fusion distance; spatially align the temperature abnormal node distribution, gas abnormal node distribution, and structural abnormal node distribution based on the IPC algorithm, and then cluster the abnormal nodes using the abnormal node fusion distance to obtain a coupled risk node distribution; align the coupled risk node distribution to the first local BIM model to locate O power plant equipment; and construct a risk conduction topology for the O coupled risk nodes in the coupled risk node distribution based on the risk conduction characteristics of the O power plant equipment, and output the dynamic risk map. The O coupled risk nodes in the coupled risk node distribution have O abnormal feature identifiers, each of which includes an abnormal device ID, a temperature abnormality, / or a concentration abnormality, and / or a deformation abnormality.

[0112] Furthermore, the task matching module 15 is further configured to perform the following steps:

[0113] Calculate the similarity between the O abnormal feature identifiers and the M detection capability matrices, and based on the calculation results, perform capability-task matching on the O coupled risk nodes and the M quadruped robots, and output M groups of abnormal detection tasks, wherein the detection capability matrix includes equipment coverage type and defect coverage type; serialize the M groups of abnormal detection tasks according to the dynamic risk map, and output the M inspection task sequences.

[0114] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0115] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0116] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A robot team-based collaborative management method for high-risk operations in power plants, characterized by: The method comprises: Segmenting a first local BIM model from the power station BIM model, wherein the first local BIM model is aligned with the first monitoring area in horizontal space; Fitting inspection coverage paths according to the heterogeneous team configuration of the first monitoring area and the first local BIM model to obtain a first set of monitoring coverage paths; H drones in the heterogeneous team configuration perform multi-drone collaborative scanning of the first monitoring area along the first set of monitoring coverage paths to obtain multimodal monitoring data; The first edge hub receives and performs risk coupling analysis based on the multimodal monitoring data to generate a dynamic risk map; Constructing M detection capability matrices of the M quadruped robots in the heterogeneous team configuration, performing capability-task matching on the dynamic risk map and the M detection capability matrices, and outputting M inspection task sequences; Performing adjacent zone inspection compensation optimization on the M inspection task sequences, outputting M compensation task sequences, and conducting risk collaborative investigation on the dynamic risk map; The first edge hub receives and performs risk coupling analysis based on the multimodal monitoring data to generate a dynamic risk map. The method includes: Decomposing the multimodal monitoring data to obtain thermodynamic field data, three-dimensional geometric field data, and gas concentration field data; Predefine a multimodal risk feature, wherein the multimodal risk feature is composed of a temperature anomaly threshold, a structural displacement safety threshold, and a gas concentration anomaly threshold; The temperature anomaly threshold and the gas concentration anomaly threshold are used to map and traverse the thermodynamic field data and the gas concentration field data to obtain the temperature anomaly node distribution and the gas anomaly node distribution; After aligning the three-dimensional geometric field data space to the first local BIM model, the structural displacement safety threshold is used to perform structural deformation judgment to obtain the distribution of structural abnormal nodes; Performing risk coupling analysis on the temperature abnormality node distribution, gas abnormality node distribution, and structure abnormality node distribution to generate the dynamic risk map; Performing risk coupling analysis on the temperature abnormality node distribution, gas abnormality node distribution, and structure abnormality node distribution to generate the dynamic risk map, the method comprising: Predefine abnormal node fusion distance; After spatially aligning the temperature anomaly node distribution, gas anomaly node distribution, and structure anomaly node distribution based on the IPC algorithm, the anomaly node fusion distance is used to cluster the anomaly nodes to obtain the coupling risk node distribution; Align the coupling risk node distribution to the first local BIM model and locate O power station equipment; According to the risk conduction characteristics of the O power station equipment, a risk conduction topology of the O coupling risk nodes in the coupling risk node distribution is constructed, and the dynamic risk map is output.

2. The robot team collaborative management method for high-risk operations in power plants according to claim 1, characterized in that: According to the heterogeneous team configuration of the first monitoring area and the first local BIM model, inspection coverage path fitting is performed to obtain a first set of monitoring coverage paths, the method comprising: Locally call the power station's global control information; Perform collaborative sensing area positioning in the first local BIM model based on the power plant global control information to obtain a collaborative sensing virtual boundary; Extracting building height distribution information of the collaborative perception virtual boundary from the first local BIM model; The heterogeneous team configuration is used as an equipment scheduling constraint, and the building height distribution information and the collaborative perception virtual boundary are used as monitoring range constraints to perform flight trajectory fitting, and the first set of monitoring coverage paths is output.

3. The robot team collaborative management method for high-risk operations in power plants according to claim 2, characterized in that: Performing collaborative sensing area positioning in the first local BIM model based on the power plant global control information to obtain a collaborative sensing virtual boundary, the method comprising: By decomposing the power plant global control information, power plant monitoring layout information and power plant airspace restriction information are obtained; Defining the monitoring range distribution in the first local BIM model according to the power station monitoring layout information, and reversely screening the first compensation blind area distribution based on the framing result; The collaborative sensing virtual boundary is defined by screening and framing the first compensation blind area distribution according to the power station airspace restriction information.

4. The robot team collaborative management method for high-risk operations in power plants according to claim 2, characterized in that: The method includes: using the heterogeneous team configuration as an equipment scheduling constraint, and using the building height distribution information and the collaborative sensing virtual boundary as monitoring range constraints to perform flight trajectory fitting, and outputting the first set of monitoring coverage paths. Fitting flight trajectories based on the building height distribution information and the collaborative sensing virtual boundary, and outputting P reference flight trajectories; Extracting H flight capability matrices of the H UAVs from the heterogeneous squad configuration; Allocating the P reference flight trajectories according to the H flight capability matrices, and outputting H groups of reference flight trajectories, wherein the flight capability matrices include flight altitude limits and flight endurance characteristics; The H groups of reference flight trajectories are smoothly connected to output H monitoring coverage paths, which constitute the first group of monitoring coverage paths.

5. The robot team collaborative management method for high-risk operations in power plants according to claim 4, characterized in that: H drones in the heterogeneous team configuration perform multi-drone collaborative scanning of a first monitoring area along the first set of monitoring coverage paths to obtain multimodal monitoring data, the method comprising: The H UAVs perform multi-machine collaborative scanning of the first monitoring area along the H monitoring coverage paths to obtain H groups of multimodal data sequences, wherein each group of multimodal data sequences includes thermodynamic profile data, a spatial point cloud sequence, and a gas spatiotemporal distribution matrix; Segmenting local multimodal monitoring storage according to the acquisition windows of the H groups of multimodal data sequences to obtain multimodal distribution data; The H groups of multimodal data sequences and multimodal distribution data are spatiotemporally aligned to obtain the multimodal monitoring data.

6. The robot team collaborative management method for high-risk operations in power plants according to claim 1, characterized in that: The O coupling risk nodes in the coupling risk node distribution have O abnormal feature identifiers, and each abnormal feature identifier includes an abnormal device ID, a temperature abnormality amount and / or a concentration abnormality amount and / or a deformation abnormality amount.

7. The robot team collaborative management method for high-risk operations in power plants according to claim 6, characterized in that: Constructing M detection capability matrices of the M quadruped robots in the heterogeneous team configuration, performing capability-task matching on the dynamic risk map and the M detection capability matrices, and outputting M inspection task sequences, the method includes: Calculate the similarity between the O abnormal feature identifiers and the M detection capability matrices, and perform capability-task matching on the O coupling risk nodes and the M quadruped robots based on the calculation results, and output M groups of abnormality detection tasks, where the detection capability matrix includes equipment coverage type and defect coverage type; The M groups of anomaly detection tasks are serialized according to the dynamic risk map, and the M inspection task sequences are output.

8. The robot team collaborative management method for high-risk operations in power plants according to claim 3, characterized in that: The power station monitoring deployment information includes fixed camera deployment data, temperature and humidity sensor deployment data, and gas detector deployment data.

9. The robot team collaborative management method for high-risk operations in power plants according to claim 8, characterized in that: Defining the monitoring range distribution in the first local BIM model according to the power station monitoring layout information, and reversely screening the first compensation blind area distribution based on the framing result, the method comprising: Perform coverage mapping in the first local BIM model based on the fixed camera deployment data, the temperature and humidity sensor deployment data, and the gas detector deployment data to obtain a multimodal monitoring range distribution boundary; Solving the intersection of the multimodal monitoring range distribution boundaries and outputting the effective monitoring range distribution boundary; Locating a first monitoring boundary in the first local BIM model; The effective monitoring range distribution boundary is removed from the first monitoring boundary to obtain the first compensation blind area distribution.

10. A robot team-based collaborative management system for high-risk operations in power plants, characterized by: The system is used to execute the robot team collaborative management method for high-risk power plant operations according to any one of claims 1 to 9, comprising: A model segmentation module is configured to segment a first local BIM model from the power plant BIM model, wherein the first local BIM model is aligned with the first monitoring area in horizontal space; An inspection coverage path fitting module is used to fit the inspection coverage path according to the heterogeneous team configuration of the first monitoring area and the first local BIM model to obtain a first set of monitoring coverage paths; a multi-drone collaborative scanning module, configured to perform multi-drone collaborative scanning of a first monitoring area along the first set of monitoring coverage paths according to the H drones in the heterogeneous team configuration, to obtain multimodal monitoring data; a risk coupling analysis module, configured to perform risk coupling analysis based on the multimodal monitoring data received by the first edge hub and generate a dynamic risk map; A task matching module is used to construct M detection capability matrices of the M quadruped robots in the heterogeneous team configuration, perform capability-task matching on the dynamic risk map and the M detection capability matrices, and output M inspection task sequences; The risk collaborative investigation module is used to optimize the adjacent area inspection compensation of the M inspection task sequences, output M compensation task sequences, and conduct risk collaborative investigation of the dynamic risk map.

Images