A multi-level high-low air multi-machine cooperative scheduling decision-making agent optimization modeling method

Abstract

The application discloses a kind of multi-level high-low air multi-machine cooperative scheduling decision-making agent optimization modeling methods: construct multi-level airspace structure model, multi-machine resource capability model and task demand model, high altitude, mid-air, low-altitude airspace, aircraft state capability, task time window, priority and capability demand are uniformly described;Further construct decision-making agent model, form including airspace state, aircraft state, task state, cooperative relationship state and risk state State space, and action space such as task allocation, route selection, height layer switching, waiting avoidance and rescheduling;On this basis, multi-machine cooperative relationship model and cooperative scheduling constraint set are established, considering time, route, height, safety interval, energy consumption, communication, airspace capacity and conflict resolution and other constraints comprehensively, and through scheduling strategy optimization solution and dynamic feedback update, generate high-low air multi-machine cooperative scheduling scheme.

Description

Technical Field

[0001] This invention relates to a decision-making agent optimization modeling method for multi-level high and low altitude multi-aircraft collaborative scheduling, belonging to the technical fields of UAV collaborative control, airspace resource scheduling, intelligent decision modeling and multi-agent system. It is particularly applicable to task allocation, path coordination, conflict resolution and dynamic rescheduling scenarios of various flight nodes such as high-altitude platforms, low-altitude UAVs, general aviation aircraft and eVTOL aircraft in complex airspace environments. Background Technology

[0002] With the rapid development of applications such as low-altitude economy, drone logistics, urban air traffic, emergency inspection, and airspace monitoring, the demand for collaborative operation of various types of high- and low-altitude aircraft in the same area, at different altitudes, and in multiple mission scenarios is constantly increasing. Different aircraft have significant differences in range, speed, payload, communication capabilities, endurance, mission capabilities, and airspace usage rules. Traditional methods relying on fixed rules, manual planning, or a single scheduling algorithm are insufficient to meet the collaborative scheduling requirements under multi-level, multi-objective, multi-constraint, and real-time dynamic changing conditions.

[0003] Existing multi-aircraft scheduling methods typically focus on a single altitude level, a single mission type, or a single type of UAV swarm, paying more attention to a single aspect of path planning, mission allocation, or obstacle avoidance control. They lack unified modeling of multi-level airspace structures (high-altitude, mid-altitude, and low-altitude), cross-altitude mission coordination, differences in multi-aircraft capabilities, communication link status, airspace capacity constraints, and dynamic risk factors. Furthermore, existing methods often lack a clear agent modeling mechanism for the scheduling decision-making body, making it difficult to transform mission requirements, aircraft capabilities, environmental conditions, airspace rules, and scheduling objectives into a computable, optimizable, and iteratively updatable decision model.

[0004] In practical applications, high- and low-altitude multi-aircraft collaborative scheduling needs to simultaneously consider factors such as task priority, route conflicts, area capacity, flight safety intervals, energy consumption constraints, communication coverage, real-time situational changes, and handling of local anomalies. When task requirements change, some aircraft experience abnormal states, local airspace becomes restricted, or communication link quality deteriorates, the system needs to quickly complete task reallocation, path replanning, and scheduling strategy adjustments. Existing technologies often suffer from problems such as static scheduling models, singular decision objectives, insufficient expression of collaborative relationships, weak dynamic reconfiguration capabilities, and difficulty in interpreting optimization results due to the lack of optimization model optimization methods for multi-level airspace collaboration.

[0005] Therefore, it is necessary to propose an optimization modeling method for decision-making agents in multi-level high- and low-altitude multi-aircraft collaborative scheduling. By constructing a unified decision-making agent model, the multi-level airspace structure, multi-aircraft states, multi-task requirements, multi-constraints, and optimization objectives are comprehensively expressed. The collaborative scheduling strategy is generated through an optimization solution mechanism, thereby improving the collaborative scheduling efficiency, security, and adaptability of high- and low-altitude multi-aircraft systems in complex dynamic scenarios. Summary of the Invention

[0006] Purpose of the invention: The technical problem to be solved by the present invention is to provide an optimization modeling method for decision-making intelligent agents for multi-level high and low altitude multi-aircraft collaborative scheduling, which addresses the shortcomings of existing multi-aircraft scheduling methods, such as incomplete airspace hierarchical expression, insufficient modeling of the capabilities of multiple types of flight nodes, fragmentation of task allocation and path coordination, weak rescheduling capability in dynamic environments, and lack of unified intelligent agent model support for scheduling decisions.

[0007] The purpose of this invention is to construct an optimization modeling method with a decision-making intelligent agent as the core, which unifies the modeling of multi-level airspace at high and low altitudes, aircraft resources, mission requirements, operational status, cooperative relationships, constraints, and optimization objectives. Through state perception, capability matching, task decomposition, cooperative relationship construction, scheduling strategy optimization, and dynamic feedback updates, it realizes the collaborative scheduling decision of multi-level high and low altitude multi-aircraft systems.

[0008] To address the aforementioned technical problems, this invention discloses a decision-making agent optimization modeling method for multi-level high- and low-altitude multi-machine collaborative scheduling, comprising the following steps:

[0009] Step 1: Divide the target airspace into multiple altitude layers, and establish an airspace map model for each altitude layer to obtain a multi-level airspace structure model;

[0010] Step 2: Model the aircraft participating in the coordinated scheduling as scheduling resource objects with state, capabilities, and constraints;

[0011] Step 3: Represent the tasks to be scheduled as decomposable and matchable task objects, and establish a task capability matching relationship between the aircraft and the task;

[0012] Step 4: Construct a decision-making agent model that includes state space, action space, constraints, and objective function. The inputs are the current airspace state, aircraft state, mission state, and matching degree results to obtain a scheduling scheme that satisfies the constraints.

[0013] Step 5: Establish a multi-aircraft collaborative relationship diagram based on the mission association, spatial proximity, communication connection, path conflict, and resource complementarity relationships among the aircraft;

[0014] Step 6: Establish a set of collaborative scheduling constraints to determine the feasibility and filter constraints of task allocation results, path selection results, and conflict states during the scheduling process;

[0015] Step 7: Perform a comprehensive cost calculation on task allocation, candidate paths, height layers, waiting time, and collaborative support relationships, and select the scheduling scheme that satisfies the constraints and has the lowest cost;

[0016] Step 8: During task execution, perform dynamic feedback updates to trigger local rescheduling or global re-optimization.

[0017] In step 1, the multi-level spatial structure model is represented as a set of height layers. ,in, Indicates the first Each height level, Indicates the number of height layers; for each height layer Establish a spatial domain map model ,in, Indicates height layer The set of spatial nodes within, This represents the set of walkable edges between spatial nodes; the spatial nodes The attribute is represented as ,in, Indicates the spatial location of the node. Indicates the height level to which the node belongs. Indicates node capacity. Indicates the node risk level. Indicates the node's passage status; the spatial edge The attribute is represented as ,in, Represents a node To node The distance between them Indicates the estimated travel time. Indicates the cost of passage. This indicates the channel risk value.

[0018] Step 2 specifically involves the set of aircraft participating in the scheduling being represented as follows: ,in, Indicates the first One aircraft, Indicates the number of aircraft; for aircraft Constructing state vectors ,in, express Location at any given moment Indicates flight altitude. Indicates flight speed. Indicates the heading angle. Indicates remaining energy. Indicates the task execution status. Indicates health status. Indicates the status of the communication link; for aircraft Constructing capability vectors and construct a constraint set The capability vector is used to characterize the aircraft's range, payload, speed, communication, perception, and mission execution capabilities, while the constraint set is used to characterize the aircraft's altitude, speed, energy, payload, safety interval, and communication limitations.

[0019] Step 3 specifically involves: the set of tasks to be executed is represented as follows: ,in, Indicates the first One task, Indicates the number of tasks; each task Represented as ,in, Indicates the starting point or area of ​​the task. Indicates the task target point or target area. Indicates the earliest start time of the task. Indicates the latest time to complete the task. Indicates task priority. Indicates the capabilities required for the task. Indicates task execution constraints; constructs the aircraft. With the task Task capability matching function , represented as ,in, Indicates the degree of ability matching. Indicates the time window matching degree. Indicates energy matching degree, Indicates the degree of spatial reachability matching. Indicates the degree of matching of the communication link. to Represents the weighting coefficient; when At that time, the aircraft With the task The combination was removed from the candidate combination.

[0020] In step 4, the state space of the decision-making agent model is represented as follows: ,in, Indicates the state of the airspace. Indicates the aircraft's status. Indicates the task status. Indicates the state of collaborative relationship. Indicates a risk state; the action space is represented as ,in, Indicates the task assignment action. This indicates a route selection action. Indicates the action of switching between different height levels. Indicates a waiting action. Indicates an evasive action. This represents a rescheduling action; the objective function is expressed as follows: ,in, This indicates the total task completion time. Indicates total energy consumption. Indicates airspace operational risks. This indicates the cost of mission delays. Indicates conflict penalty items, This indicates the benefits of aircraft load balancing. to This represents the target weight coefficient.

[0021] In step 5, the multi-machine collaborative relationship model is a multi-machine collaborative relationship diagram. ,in, Represents the set of aircraft nodes. This represents the set of collaborative edges between aircraft; collaborative edges are established between two aircraft when there is a mission association, spatial proximity, communication connection, path intersection, resource complementarity, or succession execution relationship between them; aircraft With aircraft The collaborative edge weights between them are represented as ,in, Indicates the strength of task association. Indicates spatial proximity. Indicates communication quality. Indicating conflict risk, Indicates the degree of complementarity of abilities. to Represents the weighting coefficient; when At that time, determine the aircraft With aircraft There is a strong collaborative relationship between them, and they are jointly scheduled during optimization.

[0022] In step 6, the set of cooperative scheduling constraints is represented as follows: Among them, task allocation constraints include Time window constraints include Energy constraints include Height constraints include Safety interval constraints include Spatial capacity constraints include Communication constraints include When detected When conflicts occur within the same airspace node or adjacent airway within the same time window, the conflict is resolved through speed adjustment, waiting, detour, or altitude layer switching.

[0023] In step 7, based on the task capability matching degree Filter each task The corresponding set of candidate aircraft According to the multi-level spatial domain model Generate a set of candidate paths And based on the multi-machine collaborative relationship diagram Determine strong cooperative relationships or potential conflict relationships among candidate aircraft, and then construct a set of candidate scheduling schemes. For the candidate scheduling scheme set Each candidate scheduling scheme Calculate the overall cost And determine whether it satisfies the set of coordinated scheduling constraints. If candidate scheduling scheme Not satisfying the set of cooperative scheduling constraints If the candidate scheduling scheme is satisfied, then it is eliminated; if the set of cooperative scheduling constraints is satisfied, then the candidate scheduling scheme is eliminated. If the first option is not found, it is retained as a feasible solution; finally, the feasible solution with the minimum overall cost is selected as the current optimal scheduling solution, denoted as: .

[0024] In step 8, during mission execution, the aircraft status, mission status, airspace status, communication link status, and risk status are continuously collected, and the state space of the decision-making agent is updated. When the change in state Exceeding the preset threshold Or, new tasks, task cancellations, or task priorities may appear. Changes, remaining energy of the aircraft Below the energy threshold Aircraft health status Below the health status threshold Communication link quality Below the link threshold airspace nodes Passage status Changes, number of airspace nodes occupied Exceeding the capacity limit The predicted trajectory contains Potential conflicts, or mission arrival time Beyond the task time window When a dynamic feedback update is triggered, the scope of the state change is determined. If the state change only affects a single aircraft, a single mission, or a local route, a local rescheduling is performed. If the state change affects multiple missions, multiple aircraft, or multiple altitudes, a global re-optimization is performed. The local rescheduling must satisfy the set of cooperative scheduling constraints. And the increment of the objective function Less than the preset threshold When the local rescheduling result is used... Otherwise, the result of global re-optimization is used. .

[0025] The system comprises the following modules: a multi-level airspace modeling module for constructing a multi-level airspace structure model; a multi-aircraft resource capability modeling module for constructing aircraft state attributes, capability attributes, and constraint attributes; a mission requirement modeling module for constructing mission requirement models and mission capability matching relationships; a decision-making agent modeling module for constructing state space, action space, and objective function; a multi-aircraft collaborative relationship modeling module for constructing a multi-aircraft collaborative relationship graph; a collaborative scheduling constraint establishment module for establishing a set of collaborative scheduling constraints; a scheduling strategy optimization solution module for generating multi-level high- and low-altitude multi-aircraft collaborative scheduling schemes; and a dynamic feedback update module for triggering local rescheduling or global re-optimization based on real-time state changes.

[0026] Beneficial effects:

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] First, this invention constructs a unified modeling method for multi-level airspace at high and low altitudes, which can simultaneously express the relationships between different altitude levels, different regions, different routes, and different airspace constraints, overcoming the problem that existing methods are only applicable to a single airspace level or local scheduling scenarios.

[0029] Second, this invention integrates aircraft status, mission requirements, airspace rules, and collaborative relationships into a decision-making intelligent agent model, transforming the multi-aircraft collaborative scheduling process from a traditional rule-driven or single-point optimization approach into a computationally calculable, optimizable, and feedback-updated intelligent decision-making process, thereby improving the integrity and adaptability of the scheduling model.

[0030] Third, by constructing a multi-aircraft collaborative relationship model, this invention can effectively describe the mission association, spatial proximity, communication connection and resource complementarity among aircraft, providing structured support for cross-altitude layer, multi-mission, and multi-aircraft collaborative scheduling, and improving multi-aircraft collaborative efficiency.

[0031] Fourth, this invention comprehensively considers multiple objectives such as task completion efficiency, energy consumption cost, conflict risk, airspace utilization and scheduling stability during the scheduling optimization process, which can generate a safer, more stable and efficient collaborative scheduling scheme and avoid the local optimum problem caused by single objective optimization.

[0032] Fifth, the present invention has dynamic feedback update capability. When the mission changes, the aircraft malfunctions, the local airspace is restricted, or the communication conditions change, the scheduling scheme can be quickly adjusted through the status update and rescheduling mechanism, thereby improving the robustness and continuous operation capability of the high and low altitude multi-aircraft system in complex dynamic environments. Attached Figure Description

[0033] Figure 1 Flowchart of Multi-level High- and Low-Altitude Multi-Aircraft Cooperative Scheduling Decision-Making Intelligent Agent Optimization Modeling Method

[0034] Figure 2 Schematic diagram of multi-level high and low altitude airspace structure model

[0035] Figure 3 Decision-making agent optimization modeling structure diagram

[0036] Figure 4 Multi-machine collaborative relationship model and constraint association diagram Detailed Implementation

[0037] This embodiment provides an optimization modeling method for decision-making agents in multi-level high- and low-altitude multi-aircraft collaborative scheduling. Based on multi-level airspace structure, various types of aircraft resources, mission requirements, collaborative relationships, and scheduling constraints, this method constructs a unified decision-making agent model and generates multi-aircraft collaborative scheduling schemes through optimization. The aircraft may include high-altitude platforms, medium-altitude UAVs, low-altitude UAVs, logistics UAVs, inspection UAVs, emergency communication UAVs, mapping UAVs, or other flight nodes with autonomous flight capabilities.

[0038] The decision-making agent in this invention is an abstract decision-making unit constructed for the high- and low-altitude multi-aircraft collaborative scheduling process. This decision-making agent can receive information such as airspace status, aircraft status, mission status, collaborative relationship status, and risk status, and output scheduling actions such as mission allocation, aircraft selection, route adjustment, altitude layer switching, waiting and avoidance, and rescheduling.

[0039] Step 1: Construct a multi-level spatial structure model

[0040] It is used to divide the continuous high and low altitude operating space into a computable hierarchical airspace structure, providing basic data for aircraft path selection, altitude layer allocation and conflict resolution.

[0041] This embodiment uses the example of multiple UAVs collaboratively performing inspection, delivery, and communication relay tasks within an urban area. The target area includes the low-altitude terminal flight area, the mid-altitude transfer flight area, and the high-altitude relay coverage area. Among them, the low-altitude layer corresponds to the terminal flight space between buildings, above roads, and above parks; the mid-altitude layer corresponds to the cross-regional transfer channels; and the high-altitude layer corresponds to the area where long-endurance platforms or communication relay platforms are located.

[0042] First, obtain the airspace boundaries, altitude range, available airways, mission area, restricted area, communication coverage, obstacle distribution, environmental conditions, and real-time airspace occupancy status of the target area. Based on the altitude range and operational requirements, divide the target airspace into several altitude layers, denoted as... .in, Represents a set of spatial hierarchy levels. Indicates the first Each height level, Indicates the number of height layers.

[0043] For each height layer Establish a spatial domain map model .in, Indicates height layer The set of spatial nodes within, This represents the set of traversable edges between airspace nodes. Airspace nodes are determined based on the actual operational area and can correspond to locations such as low-altitude grids, airway intersections, mission area entrances, waiting points, or altitude layer switching points. For example, in the urban low-altitude layer, a horizontal dimension of [missing information] can be used to define [missing information]. A spatial grid with a height and thickness of 30m is used as an airspace node; in mid-altitude transfer corridors, route intersections or altitude layer switching points can be used as airspace nodes. The aircraft is composed of nodes. Fly to Node , indicating its along the airspace boundary Through adjacent airspace units or airway channels.

[0044] Each airspace node The attribute is represented as .in, Indicates the spatial location of the node. Indicates the height level to which the node belongs. Indicates node capacity. Indicates the node risk level. Indicates the node's access status.

[0045] Node capacity This indicates the nodes that are allowed to enter within the same time window. The maximum number of aircraft. For low-altitude grids with dense buildings and narrow passageways, The value is relatively small; for open areas or hollow transfer nodes, The value is relatively large.

[0046] Node risk level Used to indicate where the aircraft passes through nodes. The overall risk at that time, with a value range of [0,1]. This indicates that the node has the lowest risk. This indicates the node with the highest risk. During the scheduling process, The larger the value, the lower the priority for the node to be selected as a passing node or a waiting node. (Can be set) Low-risk node This is a medium-risk node. A node is designated as high-risk; when a node is located in a restricted area, its access status will be changed. Set as , will not be included in the candidate path set.

[0047] Node risk level Based on a comprehensive calculation of node obstacle risk, environmental risk, congestion risk, communication risk, and airspace restriction risk, it is expressed as: .in, Indicates node obstacle risk, Indicates node environmental risk. Indicates the risk of node congestion. Indicates node communication risks. This indicates node limitation risk. to This represents the weighting coefficient. All risk sub-items are normalized to [0,1].

[0048] In urban areas By node The distance to surrounding buildings, towers, trees, or other obstacles is determined; the closer the distance, the better. The larger; Determined by environmental data such as wind speed, rainfall, and visibility, the worse the environmental conditions, the more likely it is that... The larger; The number of nodes in use in real time determines the capacity; the closer the usage is to the capacity limit, the better. The larger; Determined by the quality of the communication signal at that node; the worse the signal quality, the lower the quality. The larger; The degree of restriction is determined by whether the node is in a restricted area, a temporary control area, or a low-access area; the higher the degree of restriction, the lower the access level. The larger.

[0049] Node congestion risk Represented as .in, express Time-space nodes The number of aircraft inside, This represents the node capacity. When the number of aircraft within a node is... hour, near When the number of aircraft within a node reaches or exceeds hour, This indicates that the node has Capacity limit reached.

[0050] Node access status Used to represent nodes Whether the aircraft is currently allowed to enter can be set to a value of 0. , Or. Among them, This indicates normal passage. This indicates that traffic is restricted. This indicates that the road is impassable. If This node does not participate in candidate path generation; if This node can participate in candidate path generation, but will incur a corresponding passage cost during path selection.

[0051] Each airspace edge The attribute is represented as .in, Represents a node To node The distance between them Indicates the estimated travel time. Indicates the cost of passage. This indicates the channel risk value.

[0052] Channel risk value Used to indicate the aircraft along the node To node The overall risk during flight through the passageway is defined, with a value ranging from [0,1]. This indicates that the channel has the lowest risk. This indicates the highest channel risk. Channel risk value. Based on a comprehensive calculation considering the risks at both ends of the channel, channel occupancy status, communication status, and channel restriction status, it is expressed as follows: .in, and Representing nodes respectively and nodes The node risk level, This indicates the real-time risk of congestion in the passageway. Indicates channel communication risks, This indicates the risk of channel restrictions. to This represents the weighting coefficient. All sub-items are normalized to [0,1].

[0053] When the channel When located in densely populated areas, at the edge of communication coverage, or when many aircraft are already passing through, The value is relatively high; when the channel When located in an open area, with stable communication and minimal aircraft usage, The value is relatively low.

[0054] Toll Used to evaluate the passage of an aircraft through a channel The comprehensive cost, with a value ranging from [0,1]. This indicates the lowest passage cost. This indicates the highest travel cost. The travel cost is calculated based on distance, time, energy consumption, risk, and altitude layer switching, and is expressed as follows: .in, Represents the normalized distance cost. Indicates the normalized time cost. This represents the normalized energy cost. Indicates the channel risk value. Indicates the altitude layer switching indicator. to This represents the weighting coefficient. All cost sub-items are normalized to [0,1].

[0055] When the channel When connecting two nodes within the same height layer, When the channel When connecting nodes at different height levels, For example, low-altitude drones fly from the lower atmosphere. Switch to hollow layer When transferring between regions, it is necessary to go through a height layer switching channel. The cost of passage will increase accordingly.

[0056] Estimated travel time Calculated based on the channel distance and the aircraft's reference speed, expressed as: .in, Indicates aircraft Reference flight speed under current airspace conditions.

[0057] After the above modeling, a multi-level airspace structure model covering high, medium, and low altitudes can be obtained. This model can uniformly express the relationships between altitude layers, airway traffic, regional capacity, and traffic risk, and also classify node risk levels. Channel risk value and passage cost Provide a clear range of values ​​and a basis for calculation.

[0058] Step 2: Construct a multi-machine resource capability model

[0059] Step 2 is used to perform unified modeling of the various types of aircraft involved in scheduling, transforming the real-time status, capability attributes, and operational constraints of the aircraft into computable resource objects.

[0060] In this embodiment, the aircraft involved in scheduling include low-altitude inspection UAVs, low-altitude delivery UAVs, medium-altitude transfer UAVs, and high-altitude communication relay platforms. Different aircraft vary in endurance, speed, payload, communication coverage, and mission capabilities; therefore, they are uniformly represented as aircraft resource objects to facilitate mission matching and scheduling optimization.

[0061] Let the set of aircraft participating in the scheduling be . ,in, Indicates the first One aircraft, Indicates the number of aircraft.

[0062] For each aircraft Construct the aircraft state vector , represented as .in, express Time Flight Location, Indicates flight altitude. Indicates flight speed. Indicates the heading angle. Indicates remaining energy. Indicates the task execution status. Indicates health status. Indicates the status of the communication link.

[0063] surplus energy Indicates aircraft exist The amount of energy still available at any given time, expressed in watt-hours, as a percentage, or other equivalent energy units. Remaining energy is calculated based on the aircraft's energy capacity and cumulative energy consumption, and is expressed as... .in, Indicates aircraft Energy capacity Indicates aircraft From the start time to The cumulative energy consumption at any given time. The cumulative energy consumption is expressed as... .in, Indicates flight power. Indicates the power of the mission payload. Indicates communication power. This indicates the power of the sensing device.

[0064] To facilitate comparisons between different aircraft, the remaining energy is further normalized to Its value range is [0,1]. This indicates that energy is plentiful. This indicates that energy has been depleted. During scheduling, settings can be configured. For good energy status, Due to low energy status, Due to insufficient energy, it is necessary to limit the allocation of new tasks or trigger a return to base or take over the execution.

[0065] Task execution status Indicates aircraft The current stage of the task is represented using discrete numerical values. Specifically, Indicates an idle state. This indicates that a task has been assigned and the user has proceeded to the task area. This indicates that a mission is in progress. This indicates that the task is complete. This indicates an abnormal situation or a need for replacement. Task completion status is determined by the completion percentage. Judgment, represented as .in, express The number of task units that have been completed at any given time. This indicates the total number of task units assigned to the task. When At that time, the aircraft The corresponding task has been determined to be in a completed state, that is... When the aircraft experiences insufficient power, link interruption, or abnormal health status, it will... Set as .

[0066] health status Used to indicate aircraft The operational reliability is denoted by a value ranging from [0,1]. This indicates that the aircraft is in good condition. This indicates that the aircraft is unavailable. During scheduling, this can be set... For good health, For those whose health has declined but who are still able to perform low-risk tasks, If the health status is abnormal, no new tasks will be assigned.

[0067] health status Based on a comprehensive calculation of the aircraft's energy status, propulsion status, sensor status, control status, and fault alarm status, it is expressed as follows: .in, Indicates energy health. Indicates the health status of the powertrain system. Indicates the health status of the sensor. Indicates the health status of the control system. This indicates a fault alarm status. to This represents the weighting coefficient. All sub-items are normalized to [0,1], with larger values ​​indicating better conditions. For example, when the low-altitude inspection drone has sufficient battery power, normal motor condition, usable sensors, and no fault alarms, near When a motor malfunction, sensor failure, or fault alarm occurs, the corresponding sub-item is degraded. It then decreases.

[0068] Communication link status Used to indicate aircraft The communication quality with ground platforms, nearby aircraft, or relay nodes, with values ​​ranging from [0,1]. This indicates the best communication quality. This indicates that communication is unavailable. It can be set during scheduling. The communication status is good. The communication status is normal. If the communication status is poor, it is necessary to reduce the allocation of collaborative tasks or strengthen the communication link through relay nodes.

[0069] Communication link status Based on a comprehensive calculation of communication signal-to-noise ratio, packet loss rate, communication latency, and available bandwidth, it is expressed as follows: .in, This represents the normalized signal-to-noise ratio. This represents the normalized packet loss rate. Indicates normalized communication delay. Indicates the normalized available bandwidth. to This represents the weighting coefficient. Higher signal-to-noise ratio and bandwidth, and lower packet loss rate and latency, mean... The closer .

[0070] Simultaneously, construct the aircraft capability vector. , represented as .in, Indicates range capability, Indicates load capacity, Indicates speed capability, Represents communication capabilities, Indicates perceptual ability, Indicates the ability to perform tasks.

[0071] To facilitate a unified comparison across different types of aircraft, the above capability indicators are normalized to [0,1]. The closer the values ​​are to [0,1], the higher the value. The closer the value is to the corresponding ability, the stronger the corresponding ability; the closer the value is to the corresponding ability, the stronger the corresponding ability. The lower the value, the weaker the corresponding ability.

[0072] Range capability Calculated based on remaining energy, reserved return energy, cruising speed, and cruising power, expressed as follows: .in, This indicates that energy reserves have been reserved for return flight. Indicates cruising speed. This represents the cruise power. This value indicates the estimated range the aircraft can still support under current energy conditions. For uniform matching, it is normalized to... ,in, This indicates the aircraft's maximum reference range.

[0073] Load capacity Calculated based on the maximum load and the currently used load, expressed as follows: .in, Indicates aircraft Maximum load capacity express The load has been used at this time. The normalized load capacity is expressed as... .when The larger the value, the more payload missions the aircraft can continue to undertake.

[0074] Speed ​​ability Calculated based on maximum speed, current health status, and environmental impact coefficient, and expressed as follows: .in, Indicates aircraft Maximum speed, Indicates health status. This represents the environmental impact coefficient, with a value range of [0,1]. When environmental conditions are favorable and the aircraft is in good health, its speed capability is high; when there is strong wind, rainfall, or a decline in its health, its speed capability decreases accordingly.

[0075] communication capability Calculated based on communication coverage radius, communication link status, and available bandwidth, and expressed as follows: .in, Indicates the normalized communication coverage radius. Indicates the status of the communication link. Indicates the normalized available bandwidth. to This represents the weighting coefficient. High-altitude communication relay platforms typically have a large communication coverage radius. The value is relatively high; the communication capability of low-altitude terminal UAVs is mainly affected by link status, obstruction, and available bandwidth.

[0076] Perception Calculated based on sensing radius, recognition accuracy, sampling frequency, and sensor health status, and expressed as follows: .in, Indicates the normalized sensing radius. Indicates the normalized recognition accuracy. Indicates the normalized sampling frequency. Indicates the health status of the sensor. to This represents the weighting coefficient. When the inspection drone is equipped with a high-definition camera, LiDAR, or multispectral sensor, and the sensors are functioning normally... The value is relatively high.

[0077] Task execution capability Based on a comprehensive calculation considering the aircraft's mission type suitability, historical mission completion quality, health status, and payload capacity, it is expressed as follows: .in, Indicates aircraft The degree of adaptability to the target task type Indicates the quality of completion of the normalization historical task. Indicates health status. Indicates normalized load capacity. to This represents the weighting coefficient. The inspection drone's role in the inspection task... A higher value indicates that the delivery drone is more suitable for payload delivery missions. The value is relatively high, indicating that the high-altitude relay platform is responsible for communication relay tasks. The value is relatively high.

[0078] Furthermore, construct the set of constraints for the aircraft. , represented as .in, Indicates a high degree of constraint. Indicates speed constraint, Indicates energy constraints. Indicates load constraints. Indicates safety interval constraints. Indicates communication constraints.

[0079] After the above modeling, each aircraft All are represented as scheduling resource objects with state, capabilities, and constraints. Remaining energy Task execution status Health status Communication link status All capability indicators have clear data sources, value ranges, and practical meanings, which can be used for subsequent capability matching, task allocation, path selection, and scheduling optimization.

[0080] Step 3: Construct a task requirement model

[0081] Step 3 is used to transform the scheduling task into a computable and matchable task object, providing task input for the decision-making agent in step 4.

[0082] Let the set of tasks to be executed be... ,in, Indicates the first One task, This indicates the number of tasks. Each task... Represented as .in, Indicates the starting point or location where the task was issued. Indicates the task target point or target area. Indicates the earliest start time of the task. Indicates the latest time to complete the task. Indicates task priority. Indicates the capabilities required for the task. This indicates task execution constraints.

[0083] This embodiment takes an urban building inspection task as an example. The aircraft is required to reach the target building area within a specified time window, complete the facade inspection and image acquisition, and meet the requirements for range, sensing capabilities, communication links, and return-to-base energy. If the mission involves multiple stages, it can be broken down into... Each task unit corresponds to a specific stage, including takeoff, transfer, arrival at the mission area, inspection and data collection, data transmission, and return.

[0084] In order to determine the aircraft Is it suitable for performing the task? Construct a task capability matching function , represented as .in, Indicates aircraft With the task The overall matching degree between them, with a value range of [0,1], the larger the value, the higher the matching degree; to This represents the weighting coefficient.

[0085] Among them, capability matching degree Used to represent the degree to which the aircraft's capabilities meet mission requirements, calculated as .in, Indicates aircraft In the Values ​​that can be taken in class capabilities Indicates task For the first The required value of class capabilities, Indicates the first The weight of a capability. This item represents the maximum value for a certain capability of the aircraft when the mission requirements are met, and the value is taken according to the proportion of the requirements met when the requirements are not met.

[0086] Time window matching degree This indicates whether the aircraft can complete the mission within the mission time window. Let the estimated completion time of the aircraft be... ,in, This indicates the time required for the aircraft to reach the mission area. Indicates the task execution time. If... ,but ;like ,but .

[0087] Energy matching degree Used to indicate whether the aircraft's energy meets the requirements for mission execution and return, calculated as... .in, Indicates the current remaining energy. This indicates that energy reserves are reserved for safety. Indicates the execution of a task Total energy required. When hour, .

[0088] Spatial reachability matching degree Used to indicate whether the aircraft can reach the mission area and the magnitude of the path risk, calculated as follows: .in, This indicates that a feasible path exists. This indicates that there is no feasible path. This represents the average risk value of the candidate path.

[0089] Communication link matching degree Used to indicate whether the communication quality of the aircraft within the mission area meets the mission requirements, calculated as follows: .in, Indicates aircraft The expected quality of communication links near the mission area. Indicates task Minimum required communication quality.

[0090] when At that time, determine the aircraft Unsuitable for performing tasks It will not enter the candidate aircraft set; when At that time, the aircraft As a task Candidate execution objects.

[0091] Building inspection tasks For example, low-altitude inspection drones If the sensing capability is strong, the flight range meets the requirements, the arrival time is on time, and the communication quality is good, then... High altitude; low-altitude delivery drones Although the range meets the requirements, the perception capability is insufficient. The accuracy is relatively low. Therefore, the mission requirement model can provide candidate aircraft and their matching degrees for subsequent scheduling.

[0092] Step 4: Construct a decision-making agent model

[0093] In this embodiment, the decision-making agent is a decision computation model in the scheduling platform, consisting of a state space, action space, constraints, and an objective function. Its inputs are the current airspace state, aircraft state, mission state, and matching degree result; its output is a scheduling scheme that satisfies the constraints and has a low overall cost. This decision-making agent is a scheduling model independently constructed by this invention and does not rely on a specific large model name.

[0094] The state space of the decision-making agent is represented as .in, It indicates the airspace status, including node capacity, node access status, node risk level, channel risk value, and altitude layer occupancy. It indicates the aircraft's status, including position, altitude, speed, remaining energy, health status, and communication link status; It indicates the task status, including task time window, priority, capability requirements, and completion status; It indicates the status of collaborative relationships, including communication connections, spatial distances, and mission associations between aircraft; This indicates a risk status, including risks of route conflicts, regional congestion, energy shortages, and communication disruptions.

[0095] The action space of the decision-making agent is represented as .in, Indicates the task assignment action. This indicates a route selection action. Indicates the action of switching between different height levels. Indicates a waiting action. Indicates an evasive action. This indicates a rescheduling action.

[0096] For aircraft Its scheduling action is represented as .in, This indicates the assignment of tasks. Indicates the planned path, Indicates the target height layer. Indicates the speed adjustment amount. Indicates the height adjustment amount. This indicates the scheduling mode. The scheduling modes include normal execution, waiting, avoidance, return to base, takeover execution, and rescheduling.

[0097] The objective function of the decision-making agent is expressed as:

[0098] .

[0099] in, , to These represent the target weight coefficients for mission completion time, energy consumption, airspace operation risk, mission delay cost, conflict penalty, and aircraft load balancing benefit, respectively. Each weight coefficient can be configured according to different scheduling scenarios. When the mission has high real-time requirements, it can be appropriately increased. and When the task has high requirements for operational safety, it can be improved. and When the mission focuses more on the balanced utilization of spacecraft resources, it can improve... .

[0100] Total task completion time Calculated based on the estimated completion time of all assigned tasks, and expressed as:

[0101] in, Indicates task Estimated completion time; Indicates the preset reference completion time; Indicates the number of tasks.

[0102] After normalization, The smaller the value, the higher the overall completion efficiency.

[0103] Total energy consumption is expressed as

[0104] in, The cumulative energy consumption calculated in step 2 is used. This indicates the aircraft's energy capacity.

[0105] Airspace operation risks are represented as

[0106] in, This represents the set of channels included in all flight paths of the current scheduling scheme; This represents the channel risk value calculated in step 1; and These respectively represent the risk levels of the nodes at both ends of the channel; This represents the risk fusion coefficient.

[0107] The cost of task delay is expressed as

[0108] in, , These represent task times, both derived from the task requirement model established in step 3. If the task is completed on time, the corresponding delay cost is 0.

[0109] Conflict Penalty This term describes the degree of spatial or temporal conflict between aircraft in a candidate scheduling scheme. It is calculated based on the predicted number of conflicting aircraft pairs in the candidate scheduling scheme and is expressed as follows:

[0110] in, This indicates the predicted number of aircraft pairs with insufficient safety intervals. This indicates the total number of all aircraft combinations. If there are no conflicts, then... When the number of conflict aircraft increases It increases accordingly.

[0111] The aircraft load balancing benefit is calculated based on the number of tasks undertaken by all aircraft, and is expressed as follows:

[0112] in, Indicates aircraft Number of tasks undertaken; The standard deviation of the number of missions for all aircraft; This represents the average number of tasks. A smaller standard deviation indicates a more balanced distribution of tasks. The closer it is to 1.

[0113] Step 5: Construct a multi-machine collaborative relationship model

[0114] Step 5 describes the mission associations, spatial proximity, communication connections, path conflicts, and resource complementarity among aircraft, making multi-aircraft scheduling no longer a simple superposition of multiple single-aircraft scheduling results, but a collaborative optimization process that reflects the mutual influence between aircraft.

[0115] In this embodiment, a multi-machine collaborative relationship graph is constructed. .in, Represents the set of aircraft nodes. This represents the set of cooperative edges between aircraft.

[0116] If any two aircraft , If there are task associations, spatial proximity, communication connections, path intersections, resource complementarity, or succession execution relationships between the two, then a collaborative edge is established between them. Specifically, the strength of the task association... Calculations are performed based on the data dependencies, execution order, and task handover relationships between the two aircraft:

[0117] in, , The degree of data dependency between tasks is calculated as the proportion of shared data between the two tasks to the total data volume. This represents the degree of correlation in the task execution order; it is set to 1 when there is a sequential execution relationship between the tasks, and 0 otherwise. The degree of task handover is determined by a value of 1 when the aircraft needs to complete the task handover, and 0 otherwise.

[0118] Spatial proximity Calculated based on the Euclidean distance between the current positions of the two aircraft:

[0119] in, Indicates the spatial distance between two aircraft. This is the distance normalization parameter. The closer the distance, the greater the spatial proximity.

[0120] Communication quality Calculated based on communication link quality:

[0121] in, The strength of the communication signal; This refers to the available bandwidth of the link. For communication stability; all parameters are normalized to Interval. .

[0122] Conflict risk The closest approach distance was calculated based on the predicted flight trajectory. and estimated time to reach the nearest point Defined as

[0123] in, For a safe distance, This is a time-scale parameter. The closer the aircraft are and the sooner an encounter is expected, the greater the risk value.

[0124] Complementarity of abilities Based on the capability vectors of the two aircraft Calculate the cosine complement coefficient:

[0125] The more complementary the abilities, The larger.

[0126] Computational aircraft With aircraft The weight of the collaborative edge between them is :

[0127] .

[0128] in, , to Each weighting coefficient is preset based on system design experience or historical operational statistics. At that time, determine the aircraft With aircraft There is a strong synergistic relationship between them, among which, This represents the threshold for cooperative relationships. For aircraft with strong cooperative relationships, joint scheduling is performed during the optimization process; for aircraft with weak cooperative relationships, local independent scheduling can be performed to reduce the complexity of the solution.

[0129] After the calculation is completed, the resulting collaborative weights will be... Write the corresponding collaborative edge In the edge attributes, a weighted collaborative relationship graph is formed.

[0130] in, This represents the set of collaborative weights corresponding to each collaborative edge. This multi-machine collaborative relationship model allows for the simultaneous consideration of factors such as task allocation, path avoidance, communication maintenance, resource complementarity, and task succession during the scheduling process.

[0131] Step 6: Establish a set of coordinated scheduling constraints

[0132] Step 6 verifies the feasibility of candidate task allocation results, candidate paths, and flight plans, ensuring that the scheduling scheme meets constraints such as task allocation, time window, route accessibility, altitude range, energy margin, airspace capacity, safety interval, and communication link. Candidate schemes that do not meet the constraints are eliminated, while those that do proceed to Step 7 for optimization.

[0133] The set of coordinated scheduling constraints is represented as .in, This indicates task allocation constraints. Indicates time window constraints. Indicates airway traffic constraints. Indicates a high degree of constraint. Indicates energy constraints. Indicates spatial capacity constraints. Indicates safety interval constraints. Indicates communication link constraints. This indicates a conflict resolution constraint.

[0134] In this embodiment, the tasks to be scheduled include building inspection tasks. Park delivery tasks and communication relay tasks The corresponding candidate aircraft are low-altitude inspection drones. Low-altitude delivery drones and high-altitude communication relay platform After matching in step 3, the candidate assignment results are as follows: , , ,the remaining Therefore, for single-machine task execution, it satisfies Task allocation constraints.

[0135] Regarding time window constraints, the aircraft Execute the task At that time, it should meet the following requirements. and ,in In this embodiment, the building inspection task... Time window Corresponding aircraft Flight time Task execution time ,therefore ,satisfy Park delivery mission The time window is [0s, 800s], corresponding to ,satisfy Communication relay task The time window is [0s, 1200s], corresponding to ,satisfy .

[0136] Regarding air route constraints, aircraft Execute the task Candidate paths In the process, each node should satisfy the following: There should be a walkable edge between adjacent nodes. In this embodiment, implement path , implement path , implement path The passage status of each node in the above path satisfies... Furthermore, there are passable edges between adjacent nodes, thus satisfying the route passage constraint.

[0137] Regarding altitude constraints, the aircraft The flight altitude should meet And meets the allowed height range of the task. In this embodiment, and The permissible height range is [60m, 120m], and the planned height is 90m, which meets the requirements. ; The permissible height range is [180m, 250m], and the planned height is 200m, which meets the requirements. .

[0138] Regarding energy constraints, aircraft Execute the task Should meet ,in In this embodiment, implement hour, , , ,therefore This satisfies energy constraints. implement hour, , , ,therefore This satisfies energy constraints. implement hour, , , ,therefore This satisfies energy constraints.

[0139] For spatial capacity constraints, nodes exist The number of aircraft at any given time should meet the following requirements. In this embodiment, and All passed through nodes The node's capacity The number of aircraft entering this node within the same time window Therefore, it satisfies Capacity constraints.

[0140] For safety separation constraints, any two aircraft and exist The time should meet In this embodiment, the minimum safe distance In the initial candidate solutions, Expected in Passing through the node , Expected in Passing through the node The distance between the two ,because It does not meet the safety interval constraint.

[0141] To address this conflict, conflict resolution constraints are applied. If the park is tasked with delivering goods... Its priority is lower than building inspection tasks. Then for Increase waiting time Make it pass through the node Time from Adjusted to After adjustment and recalculation, the result is... This satisfies the safety interval constraint; at the same time implement The estimated completion time was changed from 295s to 315s, which still meets the requirements. Therefore, the conflict resolution result is valid.

[0142] Regarding communication link constraints, the aircraft Execute the task The time should meet In this embodiment, the building inspection task... Minimum communication quality required Due to obstructions from the target building, The quality of the direct communication link is 0.62, which does not meet the requirements. Combining the collaborative relationships from step 5, As a high-altitude communication relay platform After providing link reinforcement, Communication quality in the mission area has been improved to ,satisfy Park delivery mission Require , correspond ,satisfy Communication relay task Require , correspond ,satisfy .

[0143] Step 7: Perform scheduling strategy optimization and solution.

[0144] Step 7 calculates the comprehensive cost of task allocation, candidate paths, altitude layers, waiting time, and collaborative support relationships, and selects the scheduling scheme that satisfies the constraints and has the lowest cost. In this embodiment, the scheduling strategy optimization solution includes four processes: candidate path generation, candidate scheme construction, constraint screening, and objective function evaluation.

[0145] First, a set of candidate paths is generated. For the aircraft... Execute the task The situation will determine the current position of the aircraft. Mapping to multi-level spatial map The starting node in The target area of ​​the mission Mapped to target node When the node's passage status is satisfied. Height range and the task allows for height layers Under the premise of airspace edge passage cost Generate from arrive Candidate path set .

[0146] Candidate paths can be adopted based on The graph search method generates the graph. Specifically, it starts with the starting node. As the starting point of the search, with the target node As the search endpoint, for each node to be expanded Calculate the evaluation value .in, Indicates starting from the node To the current node The cumulative toll cost is calculated as follows: ; Indicates the current node To the target node The estimated cost can be obtained by normalizing the spatial distance between the two nodes. During the search process, priority is given to expanding... Smaller nodes, and skip them. Impassable nodes and nodes that do not meet the height constraint. The search yields the top... The path with the lowest cost and that meets the passage conditions is selected as the candidate path set. .

[0147] For example, in building inspection tasks Low-altitude inspection drones Current position corresponds to the starting node The target area corresponds to the target node. Based on the above The search yields two candidate paths: and .like Short distance but passes through high-risk nodes The cumulative path cost is 0.48; If the distance is slightly longer but the risk is lower, and the cumulative path cost is 0.36, then both can be considered as candidate paths for subsequent scheme evaluation.

[0148] Secondly, candidate scheduling schemes are constructed. It consists of task assignment variables, executing aircraft, candidate paths, altitude layers, estimated arrival time, estimated completion time, energy consumption, waiting time, and collaborative support relationships, and is represented as follows: .in, Represents task allocation variables. Indicates aircraft Execute the task The path, Indicates the target height layer. Indicates the waiting time. Indicates aircraft With aircraft The collaborative support relationship between them.

[0149] Then, based on the set of cooperative scheduling constraints established in step 6 Candidate scheduling schemes are screened. If a candidate scheme has issues such as unmet time window requirements, insufficient energy, impassable paths, mismatched altitude layers, exceeded airspace capacity limits, insufficient security intervals, or inadequate communication link quality, it is eliminated. If a candidate scheme passes the constraint verification, its objective function value is calculated.

[0150] For all feasible candidate solutions Calculate the objective function value The scheme with the minimum overall cost is selected as the current optimal scheduling scheme, denoted as: .

[0151] In the current embodiment, after the constraint verification in step 6, the following feasible candidate solutions are formed: Building inspection task Low-altitude inspection drones Execution, delivery tasks within the park Low-altitude delivery drones Execution, communication relay task High-altitude communication relay platform Execution. Among them, implement There are two candidate paths at this time. and ; implement Use path And because of At the node There is a safety interval conflict; increase the waiting time by 20 seconds. implement Use path and for Provides communication relay support.

[0152] Perform a comprehensive cost calculation on the candidate solutions. If use Because the path passes through high-risk nodes Corresponding candidate solutions The overall cost is ;like use Corresponding candidate solutions The overall cost is Since both schemes satisfy the constraints of time window, energy, altitude, capacity, security interval, and communication link in step 6, and Therefore, choose This is the current optimal scheduling scheme.

[0153] Finally, the output scheduling result is: Task By aircraft Execution, path is The height level is The estimated completion time is 480 seconds, and the communication link is... Provide relay support; mission By aircraft Execution, path is The height level is The waiting time is The estimated completion time is 315 seconds; Task By aircraft Execution, path is The height level is The estimated completion time is 700 seconds. This scheduling result is output as a multi-level high- and low-altitude multi-aircraft collaborative scheduling scheme and proceeds to step 8 for dynamic feedback updates.

[0154] Step 8: Perform dynamic feedback updates. This is used to update the decision-making agent model based on real-time state changes during task execution, and to trigger local rescheduling or global re-optimization when necessary, enabling the collaborative scheduling scheme to have dynamic adaptability.

[0155] During mission execution, the system continuously collects data on aircraft status, mission status, airspace status, communication link status, and risk status, and updates the decision-making agent's state space accordingly. When detected Change in state at time Exceeding the preset threshold At that time, dynamic feedback updates are triggered.

[0156] In one implementation, the state change is expressed as: .in, express The state of the decision-making agent at any given moment. This represents the state of the decision-making agent at the previous moment.

[0157] Dynamic feedback updates are triggered when any of the following conditions occur:

[0158] First, add tasks, cancel tasks, or prioritize tasks. Changes have occurred;

[0159] Second, the remaining energy of the aircraft Below the energy threshold ;

[0160] Third, the health status of the aircraft Below the health status threshold ;

[0161] Fourth, communication link quality Below the link threshold ;

[0162] Fifth, local airspace nodes Traffic status Changes have occurred;

[0163] Sixth, number of airspace nodes occupied Exceeding the capacity limit ;

[0164] Seventh, the predicted trajectory contains Potential conflicts;

[0165] Eighth, mission arrival time Beyond the task time window .

[0166] After a dynamic feedback update is triggered, the scope of the state change is first determined. If the change only affects a single aircraft, a single mission, or a local route, a local rescheduling is performed. Local rescheduling mainly adjusts the path, altitude, speed, waiting time, or mission succession relationships of the affected aircraft.

[0167] If the changes affect multiple tasks, multiple aircraft, or multiple altitude levels, a global re-optimization is performed. The global re-optimization recalculates the task allocation results, path selection results, altitude level allocation results, and multi-aircraft coordination relationships, and regenerates the scheduling plan. .

[0168] In one implementation, the local rescheduling result is represented as The global re-optimization result is expressed as When local rescheduling can satisfy the constraint set And the increment of the objective function Less than the preset threshold When the local rescheduling result is used... Otherwise, the result of global re-optimization is adopted. .

[0169] Through the aforementioned dynamic feedback update mechanism, this invention can quickly modify the original scheduling scheme based on changes in mission, aircraft anomalies, local airspace restrictions, communication quality degradation, and changes in conflict risk, thus preventing the scheduling scheme from becoming ineffective due to environmental changes.

[0170] After processing through steps 1 to 8 above, this embodiment completes the unified scheduling of building inspection, park delivery, and communication relay tasks within the urban park. The scheduling platform determines the building inspection tasks based on the multi-level airspace structure, aircraft resource capabilities, task requirements, collaborative relationships, and constraints. Low-altitude inspection drones Execution, delivery tasks within the park Low-altitude delivery drones Execution, communication relay task High-altitude communication relay platform The system executes the process and resolves conflicts and coordinates scheduling through path selection, waiting adjustments, and link reinforcement. Therefore, this invention can achieve unified modeling, optimization decision-making, and dynamic updates for multiple types of aircraft, multiple mission requirements, and multiple constraints across high and low altitude multi-level airspace.

[0171] This invention provides an approach and method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling. Many methods and approaches exist for implementing this technical solution; the above description is merely a preferred embodiment of the invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications should also be considered within the scope of protection of this invention. All components not explicitly stated in this embodiment can be implemented using existing technologies.

Claims

1. A method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling, characterized in that, Includes the following steps: Step 1: Divide the target airspace into multiple altitude layers, and establish an airspace map model for each altitude layer to obtain a multi-level airspace structure model; Step 2: Model the aircraft participating in the coordinated scheduling as scheduling resource objects with state, capabilities, and constraints; Step 3: Represent the tasks to be scheduled as decomposable and matchable task objects, and establish a task capability matching relationship between the aircraft and the task; Step 4: Construct a decision-making agent model that includes state space, action space, constraints, and objective function. The inputs are the current airspace state, aircraft state, mission state, and matching degree results to obtain a scheduling scheme that satisfies the constraints. Step 5: Establish a multi-aircraft collaborative relationship diagram based on the mission association, spatial proximity, communication connection, path conflict, and resource complementarity relationships among the aircraft; Step 6: Establish a set of collaborative scheduling constraints to determine the feasibility and filter constraints of task allocation results, path selection results, and conflict states during the scheduling process; Step 7: Perform a comprehensive cost calculation on task allocation, candidate paths, height layers, waiting time, and collaborative support relationships, and select the scheduling scheme that satisfies the constraints and has the lowest cost; Step 8: During task execution, perform dynamic feedback updates to trigger local rescheduling or global re-optimization.

2. The method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling according to claim 1, characterized in that, In step 1, the multi-level spatial structure model is represented as a set of height layers. ,in, Indicates the first Each height level, Indicates the number of height layers; for each height layer Establish a spatial domain map model ,in, Indicates height layer The set of spatial nodes within, This represents the set of walkable edges between spatial nodes; the spatial nodes The attribute is represented as ,in, Indicates the spatial location of the node. Indicates the height level to which the node belongs. Indicates node capacity. Indicates the node risk level. Indicates the node's passage status; the spatial edge The attribute is represented as ,in, Represents a node To node The distance between them Indicates the estimated travel time. Indicates the cost of passage. This indicates the channel risk value.

3. The method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling according to claim 1, characterized in that, Step 2 specifically involves the set of aircraft participating in the scheduling being represented as follows: ,in, Indicates the first One aircraft, Indicates the number of aircraft; for aircraft Constructing state vectors ,in, express Location at any given moment Indicates flight altitude. Indicates flight speed. Indicates the heading angle. Indicates remaining energy. Indicates the task execution status. Indicates health status. Representing the communication link status; and constructing the aircraft capability vector. , represented as ,in, Indicates range capability, Indicates load capacity, Indicates speed capability, Represents communication capabilities, Indicates perceptual ability, Indicates the ability to perform tasks.

4. The method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling according to claim 3, characterized in that, Step 2 includes: constructing the set of aircraft constraints. , represented as ,in, Indicates a high degree of constraint. Indicates speed constraint, Indicates energy constraints. Indicates load constraints. Indicates safety interval constraints. Indicates communication constraints.

5. The method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling according to claim 4, characterized in that, Step 3 specifically involves: the set of tasks to be executed is represented as follows: ,in, Indicates the first One task, Indicates the number of tasks; each task Represented as ,in, Indicates the starting point or area of ​​the task. Indicates the task target point or target area. Indicates the earliest start time of the task. Indicates the latest time to complete the task. Indicates task priority. Indicates the capabilities required for the task. Indicates task execution constraints; constructs the aircraft. With the task Task capability matching function , represented as ,in, Indicates the degree of ability matching. Indicates the time window matching degree. Indicates energy matching degree, Indicates the degree of spatial reachability matching. Indicates the degree of matching of the communication link. to Represents the weighting coefficient; when At that time, the aircraft With the task The combination was removed from the candidate combination.

6. The method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling according to claim 1, characterized in that, In step 4, the state space of the decision-making agent model is represented as follows: ,in, Indicates the state of the airspace. Indicates the aircraft's status. Indicates the task status. Indicates the state of collaborative relationship. Indicates a risk state; the action space is represented as ,in, Indicates the task assignment action. This indicates a route selection action. Indicates the action of switching between different height levels. Indicates a waiting action. Indicates an evasive action. This indicates a rescheduling action.

7. The method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling according to claim 1, characterized in that, The objective function of the decision-making agent is expressed as: in, , to These represent the target weight coefficients corresponding to mission completion time, energy consumption, airspace operation risk, mission delay cost, conflict penalty, and aircraft load balancing benefit, respectively. Each weight coefficient is configured according to different scheduling scenarios.

8. The method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling according to claim 7, characterized in that, In step 5, the multi-machine collaborative relationship model is a multi-machine collaborative relationship diagram. ,in, Represents the set of aircraft nodes. Represents the set of cooperative edges between aircraft; when any two aircraft and When there are mission associations, spatial proximity, communication connections, path intersections, resource complementarity, or succession execution relationships between two aircraft, a collaborative edge is established between them.

9. The method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling according to claim 1, characterized in that, In step 6, the set of cooperative scheduling constraints is represented as follows: ;in, This indicates task allocation constraints. Indicates time window constraints. Indicates airway traffic constraints. Indicates a high degree of constraint. Indicates energy constraints. Indicates spatial capacity constraints. Indicates safety interval constraints. Indicates communication link constraints. This indicates a conflict resolution constraint.

10. The method for optimizing the modeling of decision-making agents for multi-level high- and low-altitude multi-machine collaborative scheduling according to claim 1, characterized in that, Step 7 includes four processes: candidate path generation, candidate solution construction, constraint screening, and objective function evaluation.

Images