Intelligent scheduling method for low-altitude economic activities

Abstract

The present application relates to the technical field of low-altitude economy, and particularly relates to an intelligent scheduling method for low-altitude economic activities, comprising the following steps: step one: obtaining the capability-cost image of a UAV, quantitatively evaluating a task chain through a task chain economy evaluation module, and generating a serialized task chain package; step two: constructing a hierarchical combined auction mechanism for task chain economy optimization, including a strategy layer, a cluster layer and an execution layer, the strategy layer issuing the serialized task chain package, the cluster layer clustering and resource pre-matching the task chain package in combination with the capability-cost image of each UAV, and the execution layer completing the claiming of the serialized task chain package by the UAV through distributed auction; step three: optimizing the real-time flight path and charging decision for the UAVs that have completed claiming; step four: constructing a dual-chain architecture in which a consortium chain and a directed acyclic graph chain coexist; and step five: evaluating the task chain execution contribution and resource allocation fairness based on the Shapley value and the Gini coefficient, and dynamically feeding back and optimizing.

Description

Technical Field

[0001] This invention relates to the field of low-altitude economic technology, and in particular to an intelligent scheduling method for low-altitude economic activities. Background Technology

[0002] The rapid development of the low-altitude economy in recent years, with its diversified applications such as drone logistics, urban air traffic, agricultural plant protection, emergency rescue, and aerial inspection, has led to increasingly strained low-altitude airspace resources and a growing risk of flight conflicts. There is an urgent need to achieve efficient allocation and dynamic optimization of airspace resources through intelligent means to meet the growing demands of low-altitude economic activities and ensure flight safety and operational efficiency. This requires integrating advanced technologies such as artificial intelligence, big data, cloud computing, the Internet of Things, 5G communication, and digital twins to construct an intelligent scheduling system with core functions including real-time airspace perception, intelligent decision analysis, dynamic path planning, conflict avoidance prediction, precise resource allocation, and task collaborative optimization. This system enables autonomous path planning for low-altitude aircraft, collaborative obstacle avoidance among multiple aircraft, spatiotemporal optimization of airspace resources, and real-time monitoring of operational status. The automation and intelligentization of key capabilities such as real-time monitoring and early warning, and rapid response and handling of emergency events will improve the utilization rate and operational efficiency of low-altitude airspace, reduce operating costs and safety risks, enhance the robustness and scalability of the system, and support the efficient collaborative operation and seamless connection of various low-altitude economic activities. Its strategic significance lies not only in cultivating new productive forces and driving technological innovation and economic restructuring, but also in building a three-dimensional operation and support system of "underground-ground-low-altitude", promoting innovative collaboration between the low-altitude economy and 6G integration, smart cities, intelligent transportation and other fields, and providing full-chain support from "technology exploration" to "large-scale implementation" for urban safe development, livable environment construction and the cultivation of new economic growth drivers. Ultimately, it will form a modern low-altitude transportation ecosystem with scientific allocation of low-altitude airspace, precise flight safety assurance and coordinated development of industrial ecology.

[0003] Existing technologies treat drones as homogeneous resources or allocate tasks based on rough classifications, failing to deeply quantify their heterogeneous capabilities and dynamic costs. This results in insufficient accuracy in matching drone capabilities with complex task requirements and low efficiency in providing specialized services in existing scheduling methods. Therefore, an intelligent scheduling method for low-altitude economic activities is proposed. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing an intelligent scheduling method for low-altitude economic activities.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A method for intelligent scheduling of low-altitude economic activities includes the following steps: Step 1: Integrate the city information model platform, integrated sensing base station, meteorological sensor, and blockchain evidence storage system to construct an integrated air-ground-pipeline data acquisition network. Based on the differences in drone payload capacity, cruising speed, and sensor configuration, the drone swarm is subdivided into different types. For each subdivision, historical mission data and real-time status are integrated. A dynamically updated capability-cost profile model is constructed using machine learning to obtain the drone's capability-cost profile. Digital twin technology is used to generate a three-dimensional dynamic mission environment. Serialized mission chains are modeled within this environment. The overall path cost, time cost, and energy cost of the mission chain are quantitatively evaluated using a mission chain economic evaluation module. This generates a serialized mission chain package, providing a basis for precise matching between the mission chain and the drone's capability-cost profile. More specifically... Step 2: Construct a hierarchical combined auction mechanism for optimizing the economics of the task chain. The hierarchical combined auction mechanism includes a strategy layer, a cluster layer, and an execution layer. The strategy layer publishes serialized task chain packages. The cluster layer uses the particle swarm optimization algorithm to cluster and pre-match resources for the task chain packages based on the capability-cost profiles of each UAV. The execution layer completes the UAV's claim to the serialized task chain package through a distributed auction. The auction process is automatically executed by a blockchain smart contract to ensure transparency and trustworthiness. Step 3: Optimize the real-time flight path and charging decisions for the completed UAVs based on the deep deterministic policy gradient algorithm. The reward function based on the deep deterministic policy gradient algorithm integrates the evaluation results (task chain cost parameters) of the task chain economic evaluation module and uses the integrated sensing base station network to realize the UAV status perception. More specifically, the state space of the deep deterministic policy gradient algorithm integrates the UAV's real-time capability-cost profile parameters, the current task chain execution progress, and information on surrounding charging facilities. Its action output must dynamically optimize the flight path and charging plan while meeting the professional requirements of each sub-task in the task chain, so as to substantially realize the collaborative benefits predicted by the task chain economic evaluation module. Step 4: Construct a dual-chain architecture with a consortium blockchain and a directed acyclic graph blockchain. The consortium blockchain stores static contract data (task chain package parameters and drone capability-cost profile), while the directed acyclic graph blockchain stores dynamic execution data (dynamic task execution data generated in Step 3 and charging transaction records). The blockchain smart contract automatically executes the auction settlement and payment verification of the hierarchical combined auction mechanism and supports zero-knowledge proof for efficient batch storage of evidence. Step 5: Evaluate the contribution of the task chain execution and the fairness of resource allocation based on the Shapley value and Gini coefficient. Dynamically optimize the parameters of the particle swarm optimization algorithm, the reward function weights based on the deep deterministic policy gradient algorithm, and the consensus parameters of the blockchain. Pre-configure a resilience behavior library for each drone and activate preset emergency strategies in the event of communication interruption or extreme situations to ensure the basic operation capability of the system.

[0006] The above further includes: Furthermore, based on differences in drone payload capacity, cruising speed, and sensor configuration, drone swarms are further subdivided into types, namely, drones focused on heavy-load logistics delivery, high-speed cruising inspection, and professional monitoring equipped with multispectral or infrared sensors. For each sub-type, a capability-cost profile model reflecting its unique operating costs is independently constructed and updated. The capability-cost profile model outputs the overall path cost (energy consumption cost per unit distance), time cost (operation cost per unit time), and energy cost (data acquisition value gain due to its dedicated sensor configuration) of the corresponding sub-type drones when performing a specific task. The capability-cost profiling model uses mission type, flight distance, payload weight, and ambient wind speed from historical missions as input features, and actual measured energy consumption and time as supervision labels to train a regression prediction model to predict the cost of new missions. The model also uses the drone's real-time battery level and battery health as dynamic parameters to correct the prediction results online.

[0007] Furthermore, digital twin technology is used to generate a three-dimensional dynamic task environment that integrates geographic information, meteorological data, and facility distribution. In this three-dimensional dynamic task environment, a task chain consisting of multiple spatiotemporally or logically related sub-tasks is modeled, and the overall path cost, time cost, and energy cost of the task chain are quantitatively evaluated through a task chain economic evaluation module. Based on the evaluation results, the optimized task chain is packaged into a serialized task chain package.

[0008] Furthermore, based on the global task requirements and the economic evaluation results of the task chain, the strategic layer generates and publishes a serialized task chain package containing at least two related sub-tasks to the cluster layer, and sets the initial rules and constraints for the auction. The specific process of generating the task chain package at the strategic layer includes: based on the geographic information and task location distribution in the three-dimensional dynamic task environment, identifying multiple sub-tasks that are adjacent in time and space and are continuously executed by the same UAV to reduce empty load costs or improve service efficiency (a sub-task represents an indivisible independent job request unit proposed by the task issuer, which is the smallest logical object for task chain combination and resource scheduling), logically binding and serializing the sub-tasks to form a task chain package with overall economic value, and assigning it a unique chain identifier.

[0009] Furthermore, the cluster layer receives task chain packets from the strategy layer and uses the particle swarm optimization algorithm, combined with the capability-cost profile of the UAVs, to perform cluster analysis and resource pre-matching on the task chain packets, generating a preliminary matching scheme between the task chain packets and the UAV cluster. The specific process includes: The cluster layer receives multiple task chain packages issued by the strategy layer, and at the same time obtains the real-time capability-cost profile of each drone in the cluster. Initialize the particle swarm optimization algorithm by setting the particle population size, number of iterations, and algorithm parameters; Each particle is position-encoded, and the position encoding includes the cluster center information of the task chain package and the mapping relationship between each cluster center and a specific UAV or UAV group, so that each particle represents a complete task chain package clustering and UAV pre-matching scheme. Calculate the fitness value for each particle; Based on the fitness value and through the iterative update rules of the particle swarm optimization algorithm, the entire particle population is guided to evolve towards a better matching scheme. After the preset iteration termination condition is met, the particle with the best fitness is selected, the position information of the particle with the best fitness is decoded, and a preliminary matching scheme is generated and output. The preliminary matching scheme includes the preliminary task chain package clustering results and the preferred UAV candidate set that matches each clustered task chain package, which serves as the constraint framework and input for the subsequent distributed auction in the execution layer.

[0010] Furthermore, based on the preliminary matching scheme issued by the cluster layer, the execution layer initiates a distributed auction for the task chain package within the drone cluster. The specific steps are as follows: Auction initialization: The execution layer receives the preliminary matching scheme from the cluster layer; Auction Trigger: For each task chain package, the execution layer broadcasts an auction trigger signal to the corresponding candidate drone set through a local communication network (using a self-organizing peer-to-peer mesh network protocol, through which drones in the candidate drone set exchange auction status synchronization information and relay bid transactions to the blockchain node during the auction process, so as to enhance communication reliability in the edge network environment). The auction trigger signal includes the unique identifier of the task chain package, task details, auction deadline, and auction rules defined by the blockchain smart contract address. Distributed bidding: Each drone in the candidate drone set calculates its private cost and expected revenue for completing the task chain package based on its current real-time status data and its own capability-cost profile, and generates an encrypted bidding bid. The encrypted bid is then submitted to the designated blockchain smart contract through the local communication network. Smart contract adjudication: After the auction deadline is reached, the blockchain smart contract automatically executes the pre-programmed auction logic, including decrypting all received encrypted bids, verifying the validity and compliance of the bids, and determining the winning drone based on the preset optimal evaluation criteria; Results Storage and Notification: The blockchain smart contract generates an immutable transaction record by writing the final bidding result, winning price, and key auction parameters into the blockchain. The auction result is broadcast to all participants through the local communication network or blockchain event, thus completing the claiming and allocation of the task chain package.

[0011] Furthermore, the specific steps for co-optimizing flight path and charging decisions using a deep deterministic policy gradient algorithm after the drones have been claimed are as follows: Constructing an integrated state space: The state space integrates real-time information from three dimensions, including capability-cost profile parameters reflecting the current operational efficiency of the UAV, execution state parameters characterizing the completion progress of the sequential task chain, and resource environment parameters describing the location and status of available charging facilities in the vicinity. The resource environment parameters are sensed and updated in real time through a sensory integrated base station network. Collaborative Decision Generation: A continuous action space is defined, which outputs a continuous action vector for controlling the UAV's flight. This vector includes a three-dimensional heading angle, flight speed, and a decision variable representing whether to initiate a reservation request to a specific charging facility (the charging decision variable is a discrete action used to choose from the following options: continue the mission without charging, fly to the nearest fixed charging station, fly to the reserved mobile charging platform, or call for wireless charging service while performing a specific hovering sub-task; this decision is output together with the continuous adjustment of heading and speed, and is executed collaboratively by the flight control system). This achieves integrated decision-making on flight path and charging timing. The policy network takes the state space as input and outputs specific action values. The action values ​​must ensure that they meet the professional requirements of the UAV type (payload, sensors) for the current sub-task to be executed. Reward Calculation and Strategy Optimization: A reward function is constructed based on the evaluation results of the task chain economic evaluation module integrated with the deep deterministic policy gradient algorithm. This function is used to evaluate the immediate merits of action values. The reward function aims to maximize long-term cumulative rewards. The calculation results of the reward function are input into a critic network based on the deep deterministic policy gradient algorithm to update network parameters and drive the policy network to continuously optimize, thereby generating a charging collaborative optimization strategy that maximizes long-term cumulative rewards. This substantially realizes the overall collaborative benefits of the task chain pre-evaluation. More specifically, the training process of the deep deterministic policy gradient algorithm adopts a combination of offline pre-training and online fine-tuning: First, offline pre-training is performed using historical task data and simulation data generated from a 3D dynamic task environment to initially learn a general strategy. During the actual execution of the task chain by the UAV, online fine-tuning is performed using real-time perception data and reward feedback to adapt to specific dynamic environments and individual performance differences.

[0012] Furthermore, a dual-chain architecture is constructed, in which consortium blockchains and directed acyclic graph chains coexist, specifically including: Deploy a consortium blockchain maintained by multiple authorized nodes to store static contract data generated during low-altitude economic scheduling. The static contract data serializes the structured parameters of the task chain package and the capability-cost profile of each UAV. Deploy a directed acyclic graph chain that can be written to at high speed by a wide range of participants to store dynamic execution data, including the real-time trajectory, state changes and charging transaction records of the UAV during mission execution; The rules of the tiered combined auction mechanism are written into a blockchain smart contract and deployed on the consortium blockchain; Once the distributed auction at the execution layer is completed and the drone claims the task chain, the smart contract is automatically executed. The smart contract verifies the identity of the winner and the validity of their bid according to the pre-set auction rules, and automatically locks the task deposit, transfers the task chain execution rights, and settles and pays the task reward based on the verification results. Upon completion of the auction, the smart contract will take the auction matching result, i.e. the binding relationship between the task chain package and the specific drone, as a key contract fact, and together with the timestamp of its generation time, generate a unique transaction hash and store it as an immutable record in the consortium blockchain. During task execution, each newly generated piece of dynamic execution data, when written to the directed acyclic graph chain, contains a hash pointer pointing to the key contract facts stored on the consortium chain corresponding to its task chain, thereby establishing a two-way verifiable association. After the task is completed, the drone that performed the task or the designated aggregation node generates a batch proof of evidence based on the dynamic execution data written into the directed acyclic graph chain using zero-knowledge proof technology. The batch proof of evidence confirms that a set of dynamic execution data does indeed completely and correctly satisfy the preset execution constraints of the corresponding task chain package without disclosing all the original data details. The verification node only needs to verify the validity of the batch proof of evidence to confirm the task completion status and feeds back the verification result as the final settlement basis to the smart contract of the consortium chain.

[0013] Furthermore, the specific steps for evaluating the contribution of the task chain execution and the fairness of resource allocation are as follows: Dynamic evaluation of task chain contribution based on Shapley value: After the drone swarm completes one or more task chains, collect the actual energy consumption data, task completion quality data, and time cost data of each drone during the execution process; For each completed task chain, the Shapley value calculation model is adopted, taking all drones participating in the task chain as cooperative alliance members, taking the overall economic benefits of the task chain as the total value, and calculating and allocating the specific contribution value of each drone in the task chain based on the difficulty of the sub-tasks undertaken by each drone, the actual resource consumption, and the marginal contribution to the final completion quality. Resource allocation fairness assessment based on Gini coefficient: The scheduling system periodically calculates the cumulative contribution gains of all drones in the cluster based on the Shapley value, as well as the total task load they undertake, within a certain time window. Based on the distribution of the ratio of cumulative contribution revenue to task load, the Gini coefficient is calculated to quantitatively assess the fairness of the matching between task revenue and resource input in the entire drone swarm, and to identify whether some drones are in an unfair state of "high load, low revenue" or "low load, high revenue" for a long time.

[0014] The present invention has the following beneficial effects: 0. In this invention, by introducing a subdivision of UAV types based on payload capacity, cruising speed, and sensor configuration, and constructing a dynamic capability-cost profile model for each type, the multidimensional capabilities of UAVs are characterized in a refined and quantitative manner. Combined with task chain economic modeling in a digital twin environment, the costs and benefits of serialized task packages are pre-quantified and evaluated, thereby achieving precise matching of the task chain with the most suitable UAV type at the source.

[0015] 1. In this invention, by designing a combined auction mechanism oriented towards the economics of task chains, multiple sub-tasks that are spatially and temporally adjacent and continuously executed by the same UAV to reduce empty-load costs or improve service efficiency are intelligently bundled into task chain packages for release and bidding. The combined auction mechanism incentivizes UAVs to actively undertake related task packages from an economic model perspective, enabling UAVs to operate continuously within a single task cycle. This significantly reduces the ineffective empty-load mileage generated by task transfers, directly reducing the energy consumption and time costs for operators. Furthermore, by improving the single-slot efficiency of individual UAVs, it optimizes the resource utilization rate of the overall fleet, achieving a leap from local task completion to maximizing global economic benefits. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of an intelligent scheduling method for low-altitude economic activities proposed in this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 As shown, this invention is an intelligent scheduling method for low-altitude economic activities, comprising the following steps: Step 1: Integrate the city information model platform, integrated sensing base station, meteorological sensor, and blockchain storage system to construct an integrated air-ground-pipeline data acquisition network. Based on the differences in drone payload capacity, cruising speed, and sensor configuration, the drone swarm is subdivided into different types. For each subdivided type, historical mission data and real-time status are integrated. A dynamically updated capability-cost profile model is constructed using machine learning to obtain the drone's capability-cost profile. Digital twin technology is used to generate a three-dimensional dynamic mission environment. More specifically, multi-source data from the city information model platform, 5G-A integrated sensing base station, meteorological sensor, and blockchain storage system are integrated. Based on geographic information system and building information modeling technology, a three-dimensional dynamic digital twin environment is generated, integrating the real-time status of urban infrastructure, the full-dimensional status of drones, and airspace control policies. This provides a real-time, integrated virtual mapping space for the unified economic signal decision-making. Serialized task chains are modeled within the mission environment. The overall path cost, time cost, and energy cost of the task chain are quantitatively evaluated through a task chain economic evaluation module, generating serialized task chain packages to achieve a precise matching basis between the task chain and the drone's capability-cost profile. More specifically… Step 2: Construct a hierarchical combined auction mechanism for optimizing the economics of the task chain. The hierarchical combined auction mechanism includes a strategy layer, a cluster layer, and an execution layer. The strategy layer publishes serialized task chain packages. The cluster layer uses the particle swarm optimization algorithm to cluster and pre-match resources for the task chain packages based on the capability-cost profiles of each UAV. The execution layer completes the UAV's claim to the serialized task chain package through a distributed auction. The auction process is automatically executed by a blockchain smart contract to ensure transparency and trustworthiness. Step 3: Optimize the real-time flight path and charging decisions for the completed UAVs based on the deep deterministic policy gradient algorithm. The reward function based on the deep deterministic policy gradient algorithm integrates the evaluation results (task chain cost parameters) of the task chain economic evaluation module and uses the integrated sensing base station network to realize the UAV status perception. More specifically, the state space of the deep deterministic policy gradient algorithm integrates the UAV's real-time capability-cost profile parameters, the current task chain execution progress, and information on surrounding charging facilities. Its action output must dynamically optimize the flight path and charging plan while meeting the professional requirements of each sub-task in the task chain, so as to substantially realize the collaborative benefits predicted by the task chain economic evaluation module. Step 4: Construct a dual-chain architecture with a consortium blockchain and a directed acyclic graph blockchain. The consortium blockchain stores static contract data (task chain package parameters and drone capability-cost profile), while the directed acyclic graph blockchain stores dynamic execution data (dynamic task execution data generated in Step 3 and charging transaction records). The blockchain smart contract automatically executes the auction settlement and payment verification of the hierarchical combined auction mechanism and supports zero-knowledge proof for efficient batch storage of evidence. Step 5: Evaluate the contribution of the task chain execution and the fairness of resource allocation based on the Shapley value and Gini coefficient. Dynamically optimize the parameters of the particle swarm optimization algorithm, the reward function weights based on the deep deterministic policy gradient algorithm, and the consensus parameters of the blockchain. Pre-configure a resilience behavior library for each drone and activate preset emergency strategies in the event of communication interruption or extreme situations to ensure the basic operation capability of the system.

[0019] It should be noted that the specific analysis process for the parameters of the dynamic feedback optimization particle swarm optimization algorithm, the reward function weights based on the deep deterministic policy gradient algorithm, and the consensus parameters of the blockchain is as follows: Dynamically adjust particle swarm optimization algorithm parameters: When the Gini coefficient is too high, indicating poor fairness, the feedback mechanism will reduce the individual learning factor that guides particles to learn from their historical best position and increase the social learning factor that guides particles to learn from the global best position. Dynamically adjust the reward function weights of the deep deterministic strategy gradient algorithm: Based on the Shapley value evaluation results, if it is found that the drone has damaged the mission completion quality or time due to excessive pursuit of low energy consumption, the feedback mechanism will reduce the negative penalty weight of the energy consumption term in the reward function and correspondingly increase the positive reward weights of the mission completion quality and time efficiency terms. Dynamically adjust blockchain consensus parameters: Based on the real-time system load reflected by the frequency and scale of task execution data storage, and the fairness of node (drone or edge server) participation indirectly reflected by the Gini coefficient, dynamically adjust the block interval and the upper limit of each block size in the consensus mechanism, and introduce a weighted probability based on historical contribution (Shapley value) when selecting verification nodes to optimize storage efficiency and incentivize honest participation.

[0020] It should be noted that the specific analysis process of the resilience behavior library is as follows: The preset emergency strategies in the resilience behavior library are divided into at least three categories: The first category is communication interruption response strategies, including returning along a preset safe corridor, hovering and waiting at the last known valid GPS point, and attempting to switch to a backup power frequency communication link. The second category is extreme weather response strategies, including immediately taking shelter to the nearest weather shelter, reducing flight altitude to a safe airspace, and executing emergency landing procedures. The third category is energy emergency response strategies, including interrupting current non-critical tasks and flying directly to the nearest backup charging station or landing point; When the trigger condition is communication interruption and the drone's current task chain priority is emergency, the drone will execute a hybrid strategy of attempting to switch communication links and continuing to execute core sub-tasks; When both communication interruption and energy levels fall below the red threshold are triggered simultaneously, regardless of task priority, the strategy of flying directly to the nearest preset safe landing point will be enforced. The events, causes, and results of the activation of the resilience behavior library within each cycle will be recorded and used as data input for adjusting the load weight in the Gini coefficient calculation or the emergency strategy itself.

[0021] In one embodiment, the drone swarm is subdivided into types based on differences in payload capacity, cruising speed, and sensor configuration. This is represented by classifying drones into types focused on heavy-load logistics delivery, high-speed cruising inspection, and professional monitoring equipped with multispectral or infrared sensors. For each sub-type, a capability-cost profile model reflecting its unique operating costs is independently constructed and updated. The capability-cost profile model outputs the overall path cost (energy consumption cost per unit distance), time cost (operation cost per unit time), and energy cost (data acquisition value gain due to its dedicated sensor configuration) of the corresponding sub-type of drones when performing a specific task. The capability-cost profiling model uses mission type, flight distance, payload weight, and ambient wind speed from historical missions as input features, and actual measured energy consumption and time as supervision labels to train a regression prediction model to predict the cost of new missions. The model also uses the drone's real-time battery level and battery health as dynamic parameters to correct the prediction results online.

[0022] It should be noted that the specific analysis process for classifying drones is as follows: The classification criteria for the heavy-duty logistics type is that the load capacity is higher than the first threshold and the standard cargo mounting device is configured. The classification criteria for the high-speed cruise inspection type is that the cruise speed is higher than the second threshold and the wide-angle vision sensor is configured. The classification criteria for the professional monitoring type is that the sensor configuration includes at least one of a multispectral imager or an infrared thermal imager.

[0023] In one embodiment, digital twin technology is used to generate a three-dimensional dynamic task environment that integrates geographic information, meteorological data, and facility distribution. In the three-dimensional dynamic task environment, a task chain consisting of multiple spatiotemporally or logically related sub-tasks is modeled, and the overall path cost, time cost, and energy cost of the task chain are quantitatively evaluated through a task chain economic evaluation module. Based on the evaluation results, the optimized task chain is packaged into a serialized task chain package.

[0024] It should be noted that the specific analysis process for the quantitative evaluation of the task chain economics assessment module is as follows: In a three-dimensional dynamic task environment, simulate the execution of all sub-tasks contained in the task chain; Calculate the total flight path cost from the starting point to the final point, the total time cost under the time window constraints of each sub-task, and the total energy consumption cost considering the impact of load changes and environmental factors; The assessment evaluates the synergistic economic benefits of combining multiple subtasks into a task chain, resulting in reduced intermediate idle or waiting mileage.

[0025] In one embodiment, the strategy layer generates and publishes a serialized task chain package containing at least two related subtasks to the cluster layer based on global task requirements and task chain economic evaluation results, and sets the initial rules and constraints for the auction. The specific process of generating the task chain package at the strategic layer includes: based on the geographic information and task location distribution in the three-dimensional dynamic task environment, identifying multiple sub-tasks that are adjacent in time and space and are continuously executed by the same UAV to reduce empty load costs or improve service efficiency (a sub-task represents an indivisible independent job request unit proposed by the task issuer, which is the smallest logical object for task chain combination and resource scheduling), logically binding and serializing the sub-tasks to form a task chain package with overall economic value, and assigning it a unique chain identifier.

[0026] It should be noted that the specific analysis process for obtaining global task requirements is as follows: Multi-source task request access and standardization: Through a standardized application programming interface, original task requests from different task publishers are accessed in real time or near real time. The task publishers include at least government regulatory platforms, commercial logistics order systems, and IoT event reporting terminals. The accessed original task requests are cleaned and formatted, and their core attributes are extracted. The core attributes include at least task type, spatial location coordinates, time window requirements, resource requirements, and priority tags. Construct a dynamic global task requirement pool: All independent task requests after standardization are aggregated into a dynamically updated global task requirement pool. This requirement pool serves as a data layer component in the three-dimensional dynamic task environment. It is indexed and managed using timestamps and unique task identifiers, and each task request is labeled with its source and initial state. The integration of the demand pool with the task chain economic evaluation module: The dynamic global task demand pool is used as input and provided to the task chain economic evaluation module in real time. Based on the geographic information, charging facility distribution, and global capability profile snapshot of the UAV swarm in the three-dimensional dynamic task environment, this module performs preliminary feasibility screening and preliminary economic benefit assessment of the tasks in the demand pool. The task chain economic evaluation module first filters out task requests in the demand pool that cannot be met at present based on the overall capability profile of the current UAV swarm. Then, it performs correlation analysis between pairs or multiple tasks on the remaining tasks. Based on the principles of spatiotemporal proximity and task type compatibility, it identifies potential task groups that may be combined into task chain packages and estimates the economic benefit gain of their combined execution compared to individual execution, providing a quantitative basis for the final decision-making at the strategic level. Based on the output of the task chain economic evaluation module, the strategic layer executes the final decision to generate and release task chain packages. The decision logic includes: selecting related task groups with significant combined economic benefits from the dynamic global task demand pool, sorting and packaging them according to preset optimization objectives to form serialized task chain packages, and specifying recommended UAV capability types and initial auction parameters for each package.

[0027] In one embodiment, the cluster layer receives task chain packages from the strategy layer and uses a particle swarm optimization algorithm, combined with the capability-cost profile of the UAVs, to perform cluster analysis and resource pre-matching on the task chain packages, generating a preliminary matching scheme between the task chain packages and the UAV cluster. The specific process includes: The cluster layer receives multiple task chain packages issued by the strategy layer, and at the same time obtains the real-time capability-cost profile of each drone in the cluster. Initialize the particle swarm optimization algorithm by setting the particle population size, number of iterations, and algorithm parameters; Each particle is position-encoded, and the position encoding includes the cluster center information of the task chain package and the mapping relationship between each cluster center and a specific UAV or UAV group. This allows each particle to represent a complete task chain package clustering and UAV pre-matching scheme. More specifically, the particle position encoding adopts a hybrid encoding structure: the first part is a continuous variable used to represent the cluster center coordinates of the task chain package in the multi-dimensional feature space; the second part is a discrete variable or identifier used to represent a candidate execution group composed of multiple UAVs associated with each cluster center. This candidate execution group is the preliminary screening result based on the UAV capability-cost profile and clustering task characteristics. The fitness value is calculated for each particle. More specifically, the fitness function comprehensively evaluates the matching degree between the drone's capability-cost profile and the requirements of the task chain package in the matching scheme represented by the particle, the expected reduction in air mileage of the execution scheme, and the degree of load balancing among drones in the cluster. The fitness function consists of multiple weighted sub-objective functions, specifically including: the first sub-objective function quantifies the estimated efficiency-cost ratio of the drone to complete the assigned task chain package, which is based on the drone's capability-cost profile; the second sub-objective function calculates the expected air mileage saved due to the connection of tasks within the task chain package; and the third sub-objective function evaluates the variance of the total workload of all drones under the current matching scheme to promote load balancing. Based on the fitness value and through the iterative update rules of the particle swarm optimization algorithm, the entire particle population is guided to evolve towards a better matching scheme. After the preset iteration termination condition is met, the particle with the best fitness is selected, the position information of the particle with the best fitness is decoded, and a preliminary matching scheme is generated and output. The preliminary matching scheme includes the preliminary task chain package clustering results and the preferred UAV candidate set that matches each clustered task chain package. This serves as the constraint framework and input for the subsequent distributed auction in the execution layer. More specifically, the output preliminary matching scheme is a mapping list. This list clearly records the unique identifier of each task chain package, the cluster number to which it belongs, and a list of UAV identifiers sorted by matching priority. The UAVs in this list constitute the legal candidate set for participating in the subsequent distributed auction of the task chain package.

[0028] In one embodiment, the execution layer initiates a distributed auction for the task chain package within the drone cluster based on the preliminary matching scheme issued by the cluster layer. The specific steps are as follows: Auction initialization: The execution layer receives the preliminary matching scheme from the cluster layer; Auction Trigger: For each task chain package, the execution layer broadcasts an auction trigger signal to the corresponding candidate drone set through a local communication network (using a self-organizing peer-to-peer mesh network protocol, through which drones in the candidate drone set exchange auction status synchronization information and relay bid transactions to the blockchain node during the auction process, so as to enhance communication reliability in the edge network environment). The auction trigger signal includes the unique identifier of the task chain package, task details, auction deadline, and auction rules defined by the blockchain smart contract address. Distributed bidding: Each drone in the candidate drone set calculates its private cost and expected revenue for completing the task chain package based on its current real-time status data and its own capability-cost profile, and generates an encrypted bidding bid. The encrypted bid is then submitted to the designated blockchain smart contract through the local communication network. It should be noted that the specific analysis process for distributed pricing is as follows: Each drone invokes its local capability-cost profile model, inputting the sub-task sequence of the task chain package, environmental parameters, and the drone's own real-time battery level, current location, and existing task load; The capability-cost profile model outputs an estimated total cost to complete the entire task chain package. Based on this estimated total cost, the drone adds its expected profit margin and strategy adjustment factor to form the final auction price. The drone uses its private key to digitally sign the bid data, which includes the bid amount, mission chain package identifier, timestamp, and its own identity information. The public address of the blockchain smart contract is used for encryption to form an encrypted quote transaction; Broadcast to blockchain nodes via a peer-to-peer network to invoke the quote submission function of the smart contract; Smart Contract Decision: Upon reaching the auction deadline, the blockchain smart contract automatically executes pre-programmed auction logic, including decrypting all received encrypted bids, verifying the validity and compliance of the bids, and determining the winning drone based on a preset optimal evaluation standard. More specifically, verifying the validity and compliance of the bids includes: first, verifying whether the drone submitting the bid belongs to the legitimate candidate drone set; second, verifying whether the bid was submitted within the auction validity period; and finally, verifying the validity of the bid's digital signature to ensure that the bid has not been tampered with and its source is trustworthy. The preset optimal evaluation standard is the principle of lowest total cost. After verifying all valid bids, the blockchain smart contract automatically compares the decrypted bid values ​​and grants the execution right of the task package to the drone that submitted the lowest valid bid. If the blockchain smart contract does not receive any valid bids after the auction deadline, or if all bids fail the validity verification, the smart contract automatically triggers the auction failure processing logic, records the task package identifier and auction failure event on the blockchain, and returns the task package to the cluster layer to await re-matching and scheduling. Results Storage and Notification: The blockchain smart contract generates an immutable transaction record by writing the final bidding results, winning bid price, and key auction parameters into the blockchain. The auction results are broadcast to all participants through the local communication network or blockchain events, completing the assignment and distribution of task chain packages. More specifically, in the case of post-auction dispute arbitration, if any participating drone raises an objection to the fairness of the auction process or the results and submits an arbitration request within a specified time after the auction results are announced, the preset arbitration node retrieves the complete auction process data stored on the blockchain, including the submission time of all encrypted bids, the decrypted content, and the smart contract execution log, for transparent auditing and automatic adjudication.

[0029] In one embodiment, the specific steps for the claimed drone to perform collaborative optimization of flight path and charging decisions using a depth-deterministic policy gradient algorithm are as follows: Constructing an integrated state space: The state space integrates real-time information from three dimensions, including capability-cost profile parameters reflecting the current operational efficiency of the UAV, execution state parameters characterizing the completion progress of the sequential task chain, and resource environment parameters describing the location and status of available charging facilities in the vicinity. The resource environment parameters are sensed and updated in real time through a sensory integrated base station network. Collaborative Decision Generation: A continuous action space is defined, which outputs a continuous action vector for controlling the UAV's flight. This vector includes a three-dimensional heading angle, flight speed, and a decision variable representing whether to initiate a reservation request to a specific charging facility (the charging decision variable is a discrete action used to choose from the following options: continue the mission without charging, fly to the nearest fixed charging station, fly to the reserved mobile charging platform, or call for wireless charging service while performing a specific hovering sub-task; this decision is output together with the continuous adjustment of heading and speed, and is executed collaboratively by the flight control system). This achieves integrated decision-making on flight path and charging timing. The policy network takes the state space as input and outputs specific action values. The action values ​​must ensure that they meet the professional requirements of the UAV type (payload, sensors) for the current sub-task to be executed. Reward Calculation and Strategy Optimization: A reward function is constructed based on the evaluation results of the task chain economic evaluation module, which integrates a deep deterministic policy gradient algorithm. This function evaluates the immediate merits of action values ​​and aims to maximize long-term cumulative rewards. The calculation results of the reward function are input into a critic network based on the deep deterministic policy gradient algorithm to update network parameters and drive continuous optimization of the policy network. This generates a charging collaborative optimization strategy that maximizes long-term cumulative rewards, thereby substantially realizing the overall collaborative benefits of task chain pre-evaluation. More specifically, the training process of the deep deterministic policy gradient algorithm adopts a combination of offline pre-training and online fine-tuning: First, offline pre-training is performed using historical task data and simulation data generated from a 3D dynamic task environment to initially learn a general strategy. During the actual execution of the task chain by the UAV, online fine-tuning is performed using real-time perception data and reward feedback to adapt to specific dynamic environments and individual performance differences. It should be noted that the specific analytical process for constructing the reward function is as follows: The reward function of the evaluation result of the integrated task chain economic evaluation module The specific form is as follows: ,in, The reward for progress in the task chain is proportional to the positive incentive for completing sub-tasks; The energy cost reward is negative and is proportional to the actual energy consumption calculated based on the unit distance energy cost in the drone's capability-cost profile. The time cost reward is negative and is proportional to the product of the time consumed and the unit time operation cost C_t in the capacity-cost profile. A positive value is awarded for successfully completing efficient charging (such as charging on a mobile platform during task breaks), while a negative value is awarded for causing critical task delays due to charging. It is a mission continuity reward, with a positive value, which incentivizes behaviors such as reducing unnecessary hovering or detours and smoothly connecting sub-missions within the mission chain. This is a weighting coefficient, and its initial value comes from the quantitative analysis results of the overall cost structure of the task chain by the task chain economic evaluation module.

[0030] The weighting coefficients in the reward function and The system will be dynamically adjusted based on the deviation between the actual cumulative cost and the expected cost of the drone. If the actual energy consumption or time cost is significantly lower than expected, the system will be adjusted accordingly. and The absolute value is used to reinforce the incentive for this optimization behavior.

[0031] In one embodiment, a dual-chain architecture is constructed that combines a consortium blockchain and a directed acyclic graph chain, specifically including: Deploy a consortium blockchain maintained by multiple authorized nodes to store static contract data generated during low-altitude economic scheduling. The static contract data serializes the structured parameters of the task chain package and the capability-cost profile of each UAV. Deploy a directed acyclic graph chain that can be written to at high speed by a wide range of participants to store dynamic execution data, including the real-time trajectory, state changes and charging transaction records of the UAV during mission execution; It should be noted that the specific analysis process for storing static contract data and dynamic execution data is as follows: The consortium blockchain adopts a consensus mechanism based on proof-of-stake or Byzantine fault tolerance, and produces blocks at a low frequency to ensure strong consistency and eventual determinism of static contract data. The directed acyclic graph chain adopts an asynchronous verification mechanism without the traditional block concept, allowing participating nodes to add new transactions in parallel, achieving high throughput and low latency writing of dynamic execution data; The rules of the tiered combined auction mechanism are written into a blockchain smart contract and deployed on the consortium blockchain; Once the distributed auction at the execution layer is completed and the drone claims the task chain, the smart contract is automatically executed. The smart contract verifies the identity of the winner and the validity of their bid according to the pre-set auction rules, and automatically locks the task deposit, transfers the task chain execution rights, and settles and pays the task reward based on the verification results. It should be noted that the specific analysis process for payment verification automatically executed by smart contracts is as follows: After the contract verifies the validity of the auction results, it automatically locks the corresponding reward from the task issuer's on-chain account. Once feedback confirming the validity of the batch storage certificate by the verification node is received, the locked reward will be automatically transferred to the on-chain account of the drone operator. If verification fails or the task times out, the default logic will be automatically executed, and the deposit will be paid to the task issuer. Upon completion of the auction, the smart contract will take the auction matching result, i.e. the binding relationship between the task chain package and the specific drone, as a key contract fact, and together with the timestamp of its generation time, generate a unique transaction hash and store it as an immutable record in the consortium blockchain. During task execution, each newly generated piece of dynamic execution data, when written to the directed acyclic graph chain, contains a hash pointer pointing to the key contract facts stored on the consortium chain corresponding to its task chain, thereby establishing a two-way verifiable association. After the task is completed, the drone that performed the task or the designated aggregation node generates a batch proof of evidence based on the dynamic execution data written into the directed acyclic graph chain using zero-knowledge proof technology. The batch proof of evidence confirms that a set of dynamic execution data does indeed completely and correctly satisfy the preset execution constraints of the corresponding task chain package without disclosing all the original data details. The verification node only needs to verify the validity of the batch proof of evidence to confirm the task completion status and feeds back the verification result as the final settlement basis to the smart contract of the consortium chain.

[0032] In one embodiment, the specific steps for evaluating the contribution of task chain execution and the fairness of resource allocation are as follows: Dynamic evaluation of task chain contribution based on Shapley value: After the drone swarm completes one or more task chains, collect the actual energy consumption data, task completion quality data, and time cost data of each drone during the execution process; For each completed task chain, the Shapley value calculation model is adopted, taking all drones participating in the task chain as cooperative alliance members, taking the overall economic benefits of the task chain as the total value, and calculating and allocating the specific contribution value of each drone in the task chain based on the difficulty of the sub-tasks undertaken by each drone, the actual resource consumption, and the marginal contribution to the final completion quality. It should be noted that the dimensions for calculating marginal contribution specifically include: The added value generated by drones performing specialized sub-tasks that other drones cannot replace due to their specific payload capacity, speed, or sensor configuration; the system benefits brought about by drones saving time occupied by public charging facilities through advance planning in the mission chain; and the total swarm cost saved by drones reducing empty flight mileage due to performing related tasks. Resource allocation fairness assessment based on Gini coefficient: The scheduling system periodically calculates the cumulative contribution gains of all drones in the cluster based on the Shapley value, as well as the total task load they undertake, within a certain time window. Based on the distribution of the ratio of cumulative contribution revenue to task load, the Gini coefficient is calculated to quantitatively assess the fairness of the matching between task revenue and resource input in the entire drone swarm, and to identify whether some drones are in an unfair state of "high load, low revenue" or "low load, high revenue" for a long time.

[0033] It should be noted that the revenue-load ratio distribution on which the Gini coefficient is calculated refers to the virtual or real monetary reward converted from the contribution allocated by the Shapley value calculation. The load is calculated by weighting the actual flight time of the UAV in performing the mission, the payload power consumption, and the continuous working time of its dedicated sensor configuration, in order to comprehensively measure the resource input intensity of the UAV.

[0034] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent scheduling of low-altitude economic activities, characterized in that, Includes the following steps: Step 1: Construct a data acquisition network. Based on the differences in drone payload capacity, cruising speed, and sensor configuration, the drone swarm is subdivided into different types. For each subdivision, historical mission data and real-time status of the type are integrated to construct a capability-cost profile model. Capture the capability-cost profile of the drones and generate a three-dimensional dynamic mission environment. In the three-dimensional dynamic mission environment, the serialized mission chain is modeled. The overall path cost, time cost, and energy cost of the mission chain are quantitatively evaluated through the mission chain economic evaluation module to generate a serialized mission chain package. Step 2: Construct a hierarchical combined auction mechanism for optimizing the economics of the task chain. The hierarchical combined auction mechanism includes a strategy layer, a cluster layer, and an execution layer. The strategy layer publishes serialized task chain packages. The cluster layer uses the particle swarm optimization algorithm to cluster and pre-match resources for the task chain packages based on the capability-cost profiles of each UAV. The execution layer completes the UAV's claim to the serialized task chain package through a distributed auction. The auction process is executed by a blockchain smart contract. Step 3: Optimize the real-time flight path and charging decision for the claimed UAVs based on the deep deterministic policy gradient algorithm, wherein the reward function based on the deep deterministic policy gradient algorithm is integrated with the evaluation results of the task chain economic evaluation module; Step 4: Construct a dual-chain architecture with a consortium blockchain and a directed acyclic graph blockchain. The consortium blockchain stores static contract data, and the directed acyclic graph blockchain stores dynamic execution data. The auction settlement and payment verification of the hierarchical combined auction mechanism are automatically executed through the blockchain smart contract. Step 5: Evaluate the contribution of the task chain execution and the fairness of resource allocation based on the Shapley value and Gini coefficient, dynamically optimize the parameters of the particle swarm optimization algorithm, the reward function weights based on the deep deterministic policy gradient algorithm, and the consensus parameters of the blockchain, and pre-configure a resilience behavior library for each drone to activate preset emergency strategies in the event of communication interruption or extreme situations.

2. The intelligent scheduling method for low-altitude economic activities according to claim 1, characterized in that: Based on differences in payload capacity, cruising speed, and sensor configuration, drone swarms are subdivided into types, namely, drones focused on heavy-load logistics delivery, high-speed cruising inspection, and professional monitoring equipped with multispectral or infrared sensors. For each sub-type, a capability-cost profile model reflecting unique operating costs is independently constructed and updated. The capability-cost profile model outputs the overall path cost, time cost, and energy cost of drones of the corresponding sub-type when performing a specific task. The capability-cost profiling model uses mission type, flight distance, payload weight, and ambient wind speed from historical missions as input features, and actual measured energy consumption and time as supervision labels to train a regression prediction model to predict the cost of new missions.

3. The intelligent scheduling method for low-altitude economic activities according to claim 1, characterized in that: Digital twin technology is used to generate a three-dimensional dynamic task environment that integrates geographic information, meteorological data and facility distribution. In the three-dimensional dynamic task environment, a task chain consisting of multiple spatiotemporally or logically related sub-tasks is modeled, and the overall path cost, time cost and energy cost of the task chain are quantitatively evaluated through a task chain economic evaluation module. Based on the evaluation results, the optimized task chain is packaged into a serialized task chain package.

4. The intelligent scheduling method for low-altitude economic activities according to claim 1, characterized in that: Based on global task requirements and the economic evaluation results of the task chain, the strategy layer generates and publishes a serialized task chain package containing at least two related sub-tasks to the cluster layer, and sets the initial rules and constraints for the auction. The specific process of generating the task chain package at the strategic layer includes: Based on the geographic information and task location distribution in the three-dimensional dynamic task environment, multiple sub-tasks that are adjacent in time and space and are continuously executed by the same UAV to reduce empty load costs or improve service efficiency are identified. The subtasks are logically bundled and serialized to form a task chain package, and each package is assigned a unique chain identifier.

5. The intelligent scheduling method for low-altitude economic activities according to claim 4, characterized in that: The cluster layer receives task chain packets from the strategy layer and uses the particle swarm optimization algorithm, combined with the capability-cost profile of the UAVs, to perform cluster analysis and resource pre-matching on the task chain packets, generating a preliminary matching scheme between the task chain packets and the UAV cluster. The specific process includes: The cluster layer receives multiple task chain packages issued by the strategy layer, and at the same time obtains the real-time capability-cost profile of each drone in the cluster. Initialize the particle swarm optimization algorithm by setting the particle population size, number of iterations, and algorithm parameters; Each particle is position-encoded, and the position encoding includes the cluster center information of the task chain package and the mapping relationship between each cluster center and a specific UAV or UAV group. Calculate the fitness value for each particle; Based on the fitness value and through the iterative update rules of the particle swarm optimization algorithm, the entire particle population is guided to evolve towards a better matching scheme. After the preset iteration termination condition is reached, the particle with the best fitness is selected, the position information of the particle with the best fitness is decoded, and a preliminary matching scheme is generated and output. The preliminary matching scheme includes the preliminary task chain package clustering results and the preferred UAV candidate set that matches each clustered task chain package.

6. The intelligent scheduling method for low-altitude economic activities according to claim 5, characterized in that: Based on the preliminary matching scheme issued by the cluster layer, the execution layer initiates a distributed auction for the task chain package within the drone cluster. The specific steps are as follows: Auction initialization: The execution layer receives the preliminary matching scheme from the cluster layer; Auction Trigger: For each task chain package, the execution layer broadcasts an auction trigger signal to the corresponding set of candidate drones through a local communication network. The auction trigger signal includes the unique identifier of the task chain package, task details, auction deadline, and auction rules defined by the blockchain smart contract address. Distributed bidding: Each drone in the candidate drone set calculates its private cost and expected revenue based on its current real-time status data and its own capability-cost profile, and generates an encrypted bidding bid. The encrypted bid is then submitted to the designated blockchain smart contract through the local communication network. Smart contract adjudication: After the auction deadline is reached, the blockchain smart contract automatically executes the pre-programmed auction logic, including decrypting all received encrypted bids, verifying the validity and compliance of the bids, and determining the winning drone based on the preset optimal evaluation criteria; Results Storage and Notification: The blockchain smart contract generates an immutable transaction record by writing the final bidding result, winning price, and key auction parameters into the blockchain. The auction result is broadcast to all participants through the local communication network or blockchain event, thus completing the claiming and allocation of the task chain package.

7. The intelligent scheduling method for low-altitude economic activities according to claim 1, characterized in that: The specific steps for co-optimizing flight path and charging decisions for the claimed drones using a deep deterministic policy gradient algorithm are as follows: Constructing an integrated state space: The state space integrates real-time information from three dimensions, including capability-cost profile parameters reflecting the current operational efficiency of the UAV, execution state parameters characterizing the completion progress of the sequential task chain, and resource environment parameters describing the location and status of available charging facilities in the vicinity. The resource environment parameters are sensed and updated in real time through a sensory integrated base station network. Collaborative decision generation: Define a continuous action space, which outputs a continuous action vector for controlling the flight of the UAV, including a three-dimensional heading angle, flight speed, and a decision variable representing whether to initiate a reservation request to a specific charging facility. The policy network takes the state space as input and outputs specific action values. Reward Calculation and Strategy Optimization: A reward function is constructed based on the evaluation results of the task chain economic evaluation module using a deep deterministic policy gradient algorithm. This function is used to evaluate the immediate merits of action values. The reward function aims to maximize long-term cumulative rewards. The calculation results of the reward function are input into a critic network based on the deep deterministic policy gradient algorithm to update network parameters and drive the policy network to continuously optimize, thereby generating a charging collaborative optimization strategy that maximizes long-term cumulative rewards.

8. The intelligent scheduling method for low-altitude economic activities according to claim 1, characterized in that: Constructing a dual-chain architecture that combines consortium blockchains and directed acyclic graph chains, specifically including: Deploy a consortium blockchain maintained by multiple authorized nodes to store static contract data generated during low-altitude economic scheduling. The static contract data serializes the structured parameters of the task chain package and the capability-cost profile of each UAV. Deploy a directed acyclic graph chain to store dynamic execution data, which includes the real-time trajectory, state changes, and charging transaction records of the UAV during mission execution. The rules of the tiered combined auction mechanism are written into a blockchain smart contract and deployed on the consortium blockchain; Once the distributed auction at the execution layer is completed and the drone claims the task chain, the smart contract is automatically executed. The smart contract verifies the identity of the winner and the validity of their bid according to the pre-set auction rules, and automatically locks the task deposit, transfers the task chain execution rights, and settles and pays the task reward based on the verification results. Upon completion of the auction, the smart contract will take the auction matching result, i.e. the binding relationship between the task chain package and the specific drone, as a key contract fact, and together with the timestamp of its generation time, generate a unique transaction hash and store it as an immutable record in the consortium blockchain. During task execution, each newly generated piece of dynamic execution data, when written to the directed acyclic graph chain, contains a hash pointer pointing to the key contract facts stored on the consortium chain corresponding to its task chain. After the task is completed, the drone that performed the task or the designated aggregation node generates a batch proof of evidence based on the dynamic execution data written into the directed acyclic graph chain using zero-knowledge proof technology. The batch proof of evidence confirms that a set of dynamic execution data does indeed completely and correctly satisfy the preset execution constraints of the corresponding task chain package.

9. The intelligent scheduling method for low-altitude economic activities according to claim 1, characterized in that: The specific steps for assessing the contribution of task chain execution and the fairness of resource allocation are as follows: Dynamic evaluation of task chain contribution based on Shapley value: After the drone swarm completes one or more task chains, collect the actual energy consumption data, task completion quality data, and time cost data of each drone during the execution process; For each completed task chain, the Shapley value calculation model is adopted, taking all drones participating in the task chain as cooperative alliance members, taking the overall economic benefits of the task chain as the total value, and calculating and allocating the specific contribution value of each drone in the task chain based on the difficulty of the sub-tasks undertaken by each drone, the actual resource consumption, and the marginal contribution to the final completion quality. Resource allocation fairness assessment based on Gini coefficient: The scheduling system periodically calculates the cumulative contribution gains of all drones in the cluster based on the Shapley value, as well as the total task load they undertake, within a certain time window. Based on the distribution of the ratio of cumulative contribution benefits to task load, the Gini coefficient is calculated to quantitatively assess the fairness of the matching between task benefits and resource input in the entire drone swarm, and to identify whether some drones are in an unfair state for a long time.

Images