A post-disaster rescue task cooperative scheduling method considering performance differences of unmanned aerial vehicles

Abstract

The application discloses a post-disaster rescue task cooperative scheduling method considering performance differences of unmanned aerial vehicles, comprising: constructing an unmanned aerial vehicle capability vector and a task demand vector; calculating the matching degree of the unmanned aerial vehicle and the task by using a weighted cosine similarity; establishing a bidirectional preference sequence on the task side and the unmanned aerial vehicle side; designing and introducing an improved Gale-Shapley stable matching algorithm to realize the stable matching between the unmanned aerial vehicle and the task in a many-to-many mode, so that an initial scheduling scheme is obtained; introducing an environment perception factor, when a re-scheduling condition is triggered due to dynamic environment changes, re-scheduling the initial scheduling scheme, and judging immediate switching or delayed switching based on a smooth switching execution strategy; and distributing the unmanned aerial vehicles to execute all rescue tasks according to the final scheduling scheme. The application can realize the automatic matching of the performance differences of the unmanned aerial vehicles and the task demands, greatly improve the accuracy, real-time performance and stability of the rescue task scheduling, and is suitable for emergency rescue scenes with multiple unmanned aerial vehicle cooperation.

Description

Technical Field

[0001] This invention belongs to the field of UAV swarm collaborative scheduling technology, specifically relating to a collaborative scheduling method for disaster relief missions that takes into account the performance differences of UAVs. Background Technology

[0002] With the rapid development of drone technology, multi-drone collaborative operations have been gradually applied to post-disaster emergency scenarios such as flood relief, earthquake rescue, and forest fire monitoring. Compared with traditional ground rescue methods, drones have advantages such as rapid deployment, high maneuverability, and wide field of vision, enabling them to perform various rescue missions in complex environments, including disaster reconnaissance, communication relay, and material delivery. However, in practical applications, the drones involved in rescue operations typically come from different models or different mission groups, exhibiting significant differences in endurance, payload capacity, sensing capabilities, communication capabilities, maneuverability, and environmental adaptability. The parallel execution of multiple tasks and dynamic environmental changes further complicate the drone scheduling problem.

[0003] Existing multi-drone scheduling methods are mostly based on human experience, static rules, or heuristic allocation strategies, making it difficult to accurately quantify and describe the performance differences of drones, and also difficult to comprehensively characterize the differentiated needs of rescue missions across different capability dimensions. In the highly dynamic scenario of post-disaster relief, environmental factors and drone status change constantly over time, and traditional scheduling methods often lack real-time adjustment capabilities, easily leading to problems such as capability mismatch, mission interruption, or resource waste. When the number of drones and the scale of missions increase, mission-drone matching exhibits high-dimensional and multi-constraint coupling characteristics, making it difficult for existing methods to obtain a scheduling scheme that balances stability and executability. Therefore, the field of drone swarm collaborative scheduling technology urgently needs a collaborative scheduling scheme that can improve the accuracy of multi-drone post-disaster relief mission scheduling while also considering stability and overall execution efficiency. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention proposes a collaborative scheduling method for disaster relief missions that considers the performance differences of unmanned aerial vehicles (UAVs). By vectorizing the UAV capabilities and mission requirements, a matching-degree-based scheduling mechanism is constructed to achieve precise adaptation between UAVs and missions. Furthermore, an improved Gale-Shapley stable matching algorithm and a dynamic scheduling mechanism are introduced to update the scheduling scheme in real time under changing environmental and state conditions. Combined with a smooth switching execution strategy, this effectively reduces the risk of mission interruption. This invention aims to improve the accuracy, stability, and overall execution efficiency of multi-UAV disaster relief mission scheduling.

[0005] To achieve the above-mentioned technical objectives, the present invention provides the following technical solution:

[0006] A collaborative scheduling method for disaster relief missions that takes into account the performance differences of unmanned aerial vehicles (UAVs) includes the following steps:

[0007] A multi-dimensional quantitative analysis of the performance differences of drones and the requirements of rescue missions was conducted to construct a drone capability vector and a mission requirement vector.

[0008] Based on the UAV capability vector and mission requirement vector, the matching degree between the UAV and the rescue mission is calculated, and the UAV side preference sequence and mission side preference sequence are generated.

[0009] Based on the UAV-side preference sequence and the mission-side preference sequence, and considering the UAV capability constraints, an improved Gale–Shapley stable matching algorithm is adopted to perform stable many-to-many matching between UAVs and rescue missions, and to obtain an initial scheduling scheme.

[0010] The drone is controlled to perform rescue missions according to the initial scheduling plan. During the rescue mission, the drone status information and environmental status information are monitored in real time. When the real-time detected status information meets the preset rescheduling trigger conditions, the drone capability vector, mission requirement vector, and weight vectors participating in the calculation of matching degree are updated. The matching degree is recalculated and the improved Gale-Shapley stable matching algorithm is executed to generate a new scheduling plan.

[0011] When the scheduling plan changes, the system determines whether to immediately switch or delay the switch based on the smooth switching execution strategy; and allocates drones to perform all rescue missions according to the final scheduling plan.

[0012] Furthermore, the multi-dimensional quantitative analysis of UAV performance and rescue mission requirements, and the construction of UAV capability vector and mission requirement vector, are specifically as follows:

[0013] The six dimensions of capability—endurance, payload capacity, perception capability, communication capability, maneuverability, and environmental adaptability—are selected as the construction factors for the UAV capability vector.

[0014] The endurance is calculated based on the drone's current remaining battery power, maximum flight time, and unit energy consumption rate.

[0015] The payload capacity is calculated based on the UAV's maximum payload and its current remaining payload capacity;

[0016] The perception capability is calculated by weighting the image resolution, field of view, and number and type of sensors carried by the UAV.

[0017] The communication capability is calculated using a weighted average of communication bandwidth, signal strength, and image transmission stability.

[0018] The maneuverability is calculated based on the UAV's maximum flight speed, rate of climb, and wind resistance level.

[0019] The environmental adaptability is calculated based on the drone's rain and dust protection capabilities, temperature adaptability range, and humidity of the drone's flight environment.

[0020] Six dimensions of requirements—endurance requirements, payload requirements, perception requirements, communication requirements, mobility requirements, and environmental adaptability requirements—are selected as the construction factors for the task requirement vector. The requirements of each dimension are normalized to form the task requirement vector.

[0021] Furthermore, the step of calculating the matching degree between the UAV and the rescue mission based on the UAV capability vector and the mission requirement vector, and generating the UAV-side preference sequence and the mission-side preference sequence, specifically involves:

[0022] The weighted cosine similarity function is used to calculate and weight the similarity between the UAV capability vector and the mission requirement vector in each dimension, so as to obtain the matching degree between the UAV and the rescue mission.

[0023] For a single drone, calculate its matching degree with all rescue missions, sort the rescue missions from high to low according to the matching degree, and obtain the drone side preference sequence of the drone.

[0024] For a single rescue mission, calculate its matching degree with all drones, sort the drones from high to low according to the matching degree, and obtain the mission-side preference sequence for the rescue mission.

[0025] Furthermore, based on the UAV-side preference sequence and the mission-side preference sequence, and considering the UAV capability constraints, an improved Gale-Shapley stable matching algorithm is used to perform stable many-to-many matching between UAVs and rescue missions, resulting in the following initial scheduling scheme:

[0026] Initialize the allocation status of all drones and rescue missions, and introduce the drone-side preference sequence for each drone and the mission-side preference sequence for each rescue mission;

[0027] In the initial request round, each rescue mission sends an execution request to the most preferred drone according to its own mission-side preference sequence; rescue missions requiring multi-drone coordination send execution requests to multiple drones in sequence according to the preference sequence until the coordination quantity requirement is met.

[0028] Each drone sorts all received execution requests according to its own drone-side preference sequence; under the premise that a single drone is allowed to perform multiple rescue missions, the drone's capability constraints are considered to filter and accept the top K execution requests that have the highest preference and meet the drone's capability constraints.

[0029] The rejected mission continues to send execution requests to the next drone in the mission preference sequence; after receiving a new execution request, the drone makes a new decision: that is, it reorders, filters and accepts the rescue missions corresponding to all the received execution requests;

[0030] The process of repeatedly issuing rescue mission requests and making decisions by drones continues until all rescue missions are accepted or the mission-side preference sequence is exhausted. At this point, the repeated operation stops, and an initial scheduling plan is obtained. Rescue missions that have not been accepted even after the mission-side preference sequence has been exhausted are directly placed into the queue of missions that cannot be executed.

[0031] Furthermore, capability constraints include UAV endurance constraints, UAV payload capacity constraints, task acceptance quantity constraints, and execution order constraints, among which:

[0032] The drone's endurance constraint is: the total endurance requirement for a rescue mission undertaken by a drone must not exceed the drone's endurance.

[0033] The drone payload capacity constraint is: the total payload requirement of a rescue mission received by a drone shall not exceed the payload capacity of the drone.

[0034] The task acceptance quantity constraint requires that a drone can only accept a preset number of rescue tasks, and priority is given to rescue tasks that are earlier in the preference sequence.

[0035] The execution order constraint is as follows: when a drone performs multiple rescue missions, the mission execution order is sorted according to the principle of range priority.

[0036] Furthermore, the preset rescheduling triggering conditions specifically include:

[0037] The drone's real-time remaining battery power is below the preset safety threshold;

[0038] The quality of the drone's real-time communication link is lower than the preset stable communication threshold;

[0039] The real-time wind speed of the environment where the drone is located exceeds the maximum permissible wind speed, the real-time precipitation exceeds the maximum permissible precipitation, and the real-time visibility is lower than the minimum permissible visibility.

[0040] The drone malfunctions or its real-time health status measurement falls below the preset minimum health threshold.

[0041] Furthermore, the matching weights are updated as follows:

[0042] A real-time environmental impact factor vector is defined. The original weight vector used to calculate the matching degree between the UAV and the rescue mission is weighted and corrected item by item, and then normalized to obtain the updated weight vector. The environmental impact factors include the impact factors on endurance, payload, visibility on perception capability, electromagnetic interference on communication capability, wind speed on maneuverability, and precipitation on environmental adaptability.

[0043] Furthermore, the step of determining whether to immediately perform a scheduling switch or delay a switch based on the smooth switching execution strategy when the scheduling scheme changes specifically involves:

[0044] The smooth switching execution strategy constructs a switching cost function to comprehensively evaluate the additional range, mission interruption risk, and remaining execution time caused by the UAV continuing to execute the current task versus switching to the new scheduling scheme. When the value of the switching cost function is greater than a preset cost threshold, the scheduling switch is delayed. When the calculation result of the switching cost function is less than or equal to the preset threshold, the scheduling switch is executed immediately.

[0045] Furthermore, this invention also discloses an application system for a collaborative scheduling method for disaster relief missions that considers the performance differences of unmanned aerial vehicles (UAVs), specifically including:

[0046] The UAV capability modeling module is used to quantify the endurance, payload capacity, perception capability, communication capability, maneuverability and environmental adaptability of each UAV into a multi-dimensional UAV capability vector.

[0047] The rescue mission requirement modeling module is used to quantify the endurance requirements, payload requirements, perception requirements, communication requirements, mobility requirements, and environmental adaptation requirements of each rescue mission into a multi-dimensional mission requirement vector.

[0048] The matching degree calculation module is used to calculate the matching degree between the drone and the rescue mission based on the weighted similarity calculation method;

[0049] The preference sequence construction module is used to construct task-side preference sequences and UAV-side preference sequences based on the matching degree, respectively.

[0050] The stable matching scheduling module is used to generate an initial scheduling scheme between drones and rescue missions based on the mission-side preference sequence and the drone-side preference sequence, using an improved stable matching algorithm while taking into account the drone's capability constraints.

[0051] The dynamic scheduling module is used to monitor the drone status and environmental status information in real time during the execution of rescue missions, and to dynamically update the initial scheduling plan when preset trigger conditions are met.

[0052] The switching control module is used to control the UAV to execute the current task or switch to a new scheduling task based on a smooth switching execution strategy when the scheduling scheme is updated.

[0053] Based on the above technical solution, the present invention has at least the following beneficial effects:

[0054] This invention establishes a scheduling mechanism based on weighted matching degree by constructing a vectorized model of UAV capabilities and a vectorized model of mission requirements. This achieves precise matching between UAV performance differences and rescue mission requirements, effectively avoiding problems of insufficient capabilities or waste of resources, and significantly improving the scheduling accuracy and execution success rate of multi-UAV rescue missions.

[0055] An improved Gale–Shapley stable matching algorithm is designed and introduced to ensure the stability and executability of scheduling results in multi-task, multi-UAV scenarios. Addressing the dynamic nature of the rescue environment, this invention further proposes a real-time scheduling and replanning mechanism. This mechanism can rapidly update the scheduling scheme based on UAV status and environmental changes, and reduce the risk of mission interruption and flight resource waste through smooth switching of execution strategies, thereby improving the overall safety, continuity, and reliability of the system.

[0056] The method of this invention is highly versatile and applicable to various drone collaborative operations and emergency rescue application scenarios, and has good engineering application value. Attached Figure Description

[0057] Figure 1 This is a schematic diagram of the UAV capability vectorization model in the embodiment;

[0058] Figure 2 This is a schematic diagram of the task requirement vectorization model in the embodiment;

[0059] Figure 3 This is a schematic diagram of the task-UAV matching degree calculation structure in the embodiment;

[0060] Figure 4 This is a schematic diagram illustrating the construction of preference sequences on the task side and the UAV side in the embodiment;

[0061] Figure 5 The flowchart of the improved Gale–Shapley stable matching algorithm introduced in this invention is shown below.

[0062] Figure 6 This is a diagram illustrating the overall framework of the dynamic scheduling mechanism.

[0063] Figure 7 Flowchart for real-time weight updates and recalculation of matching degree;

[0064] Figure 8 A schematic diagram illustrating the process of smoothly switching execution strategies;

[0065] Figure 9 This is an overall flowchart of a collaborative scheduling method for disaster relief missions that takes into account the performance differences of unmanned aerial vehicles (UAVs) proposed in this invention. Detailed Implementation

[0066] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the following description is provided in conjunction with the accompanying drawings. Figures 1-9 The present invention will be further described in detail with reference to the embodiments, so as to ensure that those skilled in the art can fully understand the complete process of solving technical problems and achieving technical effects by applying technical means in this application, and can implement it according to the process.

[0067] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. Therefore, this request can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, this request can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] This implementation involves specific examples of flood relief application scenarios as follows: After a flood disaster, affected by factors such as continuous heavy rainfall, rising river levels, and dam breaches, disaster areas often experience large-scale water accumulation, road interruptions, and damaged communication facilities, posing significant challenges to emergency rescue work. Traditional rescue methods relying on ground personnel and vehicles suffer from poor mobility, slow response speed, and high personnel safety risks in such environments. Drones, due to their advantages of rapid deployment, aerial mobility, and wide field of view, have gradually become an important technological means in flood relief. However, in actual rescue operations, the drones participating in the missions typically come from different models, manufacturers, or mission groups, exhibiting significant differences in endurance, payload capacity, sensing capabilities, communication capabilities, maneuverability, and environmental adaptability. Furthermore, flood relief is often not a single task but rather a combination of multiple types of tasks operating in parallel, resulting in significantly different requirements for drone capabilities. Therefore, this embodiment presents a collaborative scheduling method for post-disaster rescue missions that considers the performance differences of drones, such as... Figure 9 As shown, it specifically includes:

[0069] S1. Conduct multi-dimensional quantitative analysis of the performance differences of UAVs and the requirements of rescue missions, and construct UAV capability vectors and mission requirement vectors.

[0070] In a preferred embodiment, step S1 specifically comprises:

[0071] Six dimensions—endurance, payload capacity, perception capability, communication capability, maneuverability, and environmental adaptability—are selected as the construction factors for the UAV capability vector; for example... Figure 1 As shown, for drones Its six dimensions of capability are denoted as follows: , , , , , Then drone The capability vector is represented as ;

[0072] In this embodiment, battery life Based on the current remaining battery power of drone i Maximum flight time and unit energy consumption rate The calculation is performed, and the formula is expressed as follows:

[0073] ;

[0074] Load capacity Based on the maximum payload of drone i With current remaining load capacity The calculation is performed, and the formula is expressed as follows:

[0075] ;

[0076] Perception According to drones Equipped with image resolution Field of view and the weighted value of the number and type of sensors. The weighted calculation is expressed by the formula:

[0077] ;

[0078] ;

[0079] in, , , , , , , These are the weighting coefficients for each item in perception ability. (Infrared sensor) (LiDAR) (RGB camera) (Gas sensor); For a single drone, The higher the value, the better its sensing ability, and the more sensors there are. The larger the value;

[0080] communication capability According to communication bandwidth Signal strength and image transmission stability The weighted calculation is expressed by the formula:

[0081] ;

[0082] in, Indicates maximum communication bandwidth, Indicates the maximum signal strength;

[0083] Mobility According to drones Maximum flight speed Climb rate and wind resistance level The calculation, expressed by the formula, is as follows:

[0084] ;

[0085] in, These are the weighting coefficients for each item in the mobility capability;

[0086] Environmental adaptability Based on the drone's rain and dust protection capabilities Temperature adaptability range and the humidity of the drone's flight environment The calculation, expressed by the formula, is as follows:

[0087] ;

[0088] in, These are the weighting coefficients for each item in environmental adaptability.

[0089] The aforementioned weighting coefficients are dynamically adjusted according to the needs of different mission scenarios. In addition, during the actual calculation of the drone's capabilities in each dimension, the capability values ​​of each dimension must be normalized to ensure that each dimension is comparable.

[0090] Six dimensions of requirements—range requirements, payload requirements, perception requirements, communication requirements, mobility requirements, and environmental adaptability requirements—are selected as the constructing factors for the task requirement vector, such as... Figure 2 As shown, for a rescue mission The requirements of each dimension are denoted as follows: , , , , , Then the task requirement vector for this rescue mission is represented as: Similarly, the requirements of each dimension also need to be normalized to form a task requirement vector. In this embodiment, the values ​​of each dimension of the task requirement vector are determined by experts through discussion, and appropriate values ​​are set according to different tasks and different requirement tendencies.

[0091] S2. Based on the UAV capability vector and mission requirement vector, calculate the matching degree between the UAV and the rescue mission, and generate the UAV side preference sequence and mission side preference sequence.

[0092] As a preferred embodiment, such as Figure 3 As shown, step S2 in this embodiment specifically includes:

[0093] The weighted cosine similarity function is used to analyze the UAV capability vector. With task requirement vector The similarity scores across various dimensions are calculated and weighted to obtain the matching degree between the drone and the rescue mission. The specific formula for calculating the matching degree is:

[0094] ;

[0095] in, The larger the value, the more likely it is to be a drone. The more suitable for the task ; Indicates drone In the The capability values ​​across each dimension have been normalized and represent the actual executable capabilities of the drone. Indicates task In the The intensity of demand in each dimension is used to express the characteristics of the task; Indicates the first The weight coefficients for each dimension take values ​​between 0 and 1, and satisfy the following conditions: ; at the same time through and These two normalization denominators can perform weighted length normalization on the UAV capability vector and mission requirement vector, ensure that the matching degree is kept in the range of 0-1, prevent excessively large values ​​in the same dimension from causing bias, and solve the problems of inconsistent dimensions and inability to directly compare different scales.

[0096] For a single drone, calculate its matching degree with all rescue missions, and sort the rescue missions from high to low according to the matching degree to obtain the drone's drone preference sequence. ; This means drones i, That is to say, the task j For a single rescue mission, calculate its matching degree with all drones, and sort the drones from highest to lowest matching degree to obtain the mission-side preference sequence for that rescue mission. ;

[0097] like Figure 4 As shown, a bidirectional preference sequence is constructed. For each task... Based on the preference sequence, all drones are sorted from highest to lowest matching degree. The sorted sequence represents the preference for which drone is most likely to perform the task. For example:

[0098] Task 1 (Get) The preference sequence is This means that task 1 is most preferably performed by drone 3, drone 1, drone 4, and drone 2 in that order. Similarly, the drone preference sequence means that for each drone... The tasks are sorted according to their adaptability, and this sorting indicates which tasks the drone is more willing to perform; for example: drones The preference sequence is as follows: This indicates that the tasks that drone 3 most wants to perform are, in order, task 2, task 1, task 4, and task 3.

[0099] S3. Based on the UAV-side preference sequence and the mission-side preference sequence, and considering the UAV capability constraints, an improved Gale–Shapley stable matching algorithm is adopted to perform a many-to-many stable matching between UAVs and rescue missions, and to obtain an initial scheduling scheme.

[0100] As a preferred embodiment, such as Figure 5 As shown, step S3 in this embodiment specifically includes:

[0101] Initialize the allocation status of all drones and rescue missions, and introduce the drone-side preference sequence for each drone and the mission-side preference sequence for each rescue mission;

[0102] In the initial request round, each rescue mission sends an execution request to the most preferred drone according to its own mission-side preference sequence; rescue missions requiring multi-drone coordination send execution requests to multiple drones in sequence according to the preference sequence until the coordination quantity requirement is met.

[0103] Each drone sorts all received execution requests according to its own drone-side preference sequence; under the premise that a single drone is allowed to perform multiple rescue missions, the drone's capability constraints are considered to filter and accept the top K execution requests that have the highest preference and meet the drone's capability constraints.

[0104] In this embodiment, capability constraints are used to limit tasks that the UAV cannot perform, and tasks cannot be executed in any order. Specifically, these constraints include UAV endurance constraints, UAV payload capacity constraints, task acceptance quantity constraints, and execution order constraints.

[0105] The drone's endurance constraint is: the total endurance requirement for a rescue mission undertaken by a drone must not exceed the drone's endurance.

[0106] The drone payload capacity constraint is: the total payload requirement of a rescue mission received by a drone shall not exceed the payload capacity of the drone.

[0107] The task acceptance quantity constraint requires that a drone can only accept a preset number of rescue tasks, and priority is given to rescue tasks that are earlier in the preference sequence.

[0108] The execution order constraint is as follows: when a drone performs multiple rescue missions, the mission execution order is sorted according to the principle of range priority;

[0109] The rejected mission continues to send execution requests to the next drone in the mission preference sequence; after receiving a new execution request, the drone makes a new decision: that is, it reorders, filters and accepts the rescue missions corresponding to all the received execution requests;

[0110] The process of repeatedly issuing rescue mission requests and making decisions by drones continues until all rescue missions are accepted or the mission-side preference sequence is exhausted. At this point, the repeated operation stops, and an initial scheduling plan is obtained. Rescue missions that have not been accepted even after the mission-side preference sequence has been exhausted are directly placed into the queue of missions that cannot be executed.

[0111] S4. Control the UAV to execute the rescue mission according to the initial scheduling plan. During the rescue mission, monitor the UAV status information and environmental status information in real time. When the real-time detected status information meets the preset rescheduling trigger conditions, update the UAV capability vector, mission requirement vector, and weight vectors participating in the matching degree calculation. Recalculate the matching degree and execute the improved Gale-Shapley stable matching algorithm to generate a new scheduling plan. The dynamic rescheduling mechanism can ensure the continuous execution of the mission and improve the overall rescue efficiency. The specific rescheduling framework diagram is shown below. Figure 6 As shown.

[0112] In this embodiment, as a preferred implementation, step S4 specifically includes:

[0113] Define a real-time state vector To reflect the real-time monitored state information, the meanings of each symbol in this vector are shown in Table 1 below:

[0114] Table 1. Meaning of symbols in real-time state vector

[0115]

[0116] The real-time status information comes from the drone's own sensors or the environmental monitoring system of the ground station.

[0117] Next, define a set of system thresholds: security thresholds. Stable communication threshold Maximum permissible wind speed Maximum allowable precipitation Minimum allowable visibility Minimum health threshold Rescheduling will occur when any of the following preset rescheduling trigger conditions are met:

[0118] The drone's real-time remaining battery power is below the preset safety threshold. This indicates that the drone's battery level is below the minimum safe flight requirement and it must exit the mission.

[0119] The real-time communication link quality of the drone is lower than the preset stable communication threshold. This indicates a deterioration in the quality of the communication link; link instability may lead to data interruption and remote control failure.

[0120] The real-time wind speed in the environment where the drone is located exceeds the maximum permissible wind speed. Real-time precipitation exceeds the maximum allowable precipitation Real-time visibility is below the minimum permissible visibility. This indicates that environmental degradation, excessive wind speed or precipitation necessitates increased consideration of drone maneuverability, and decreased visibility requires increased consideration of drone perception capabilities.

[0121] The drone malfunctions or its real-time health status metric falls below a preset minimum health threshold. The drone is unable to perform its mission due to a decrease in its health status caused by itself or an accident.

[0122] When a rescheduling is triggered, the UAV capability vector and mission requirement vector are obtained first from step S1, and the original weight vector used in step S2 to calculate the matching degree is also obtained. ,in Weighted by battery life; As the load capacity weight; Weights for perceptual abilities; Weights for communication capabilities; Assigning weights to maneuverability; Weighting for environmental adaptability;

[0123] Next, as Figure 7As shown, a real-time environmental impact factor vector is defined. The specific meanings of each environmental impact factor are shown in Table 2 below:

[0124] Table 2. Specific meanings of each environmental impact factor

[0125]

[0126] The specific values ​​for each environmental impact factor can be determined by experts through discussion based on the actual situation.

[0127] In this embodiment, the original weight vector used in calculating the matching degree between the UAV and the rescue mission is weighted and corrected item by item using a real-time environmental impact factor vector, and then normalized to obtain an updated weight vector. The formula is expressed as:

[0128] ;

[0129] Reuse The formula for updating the match between the drone and the mission is as follows:

[0130] ;

[0131] in, This represents the updated matching degree. After updating the matching degree, the bidirectional preference sequence between the drone and the rescue mission is updated. A new scheduling scheme is generated by executing the improved Gale–Shapley stable matching algorithm.

[0132] S5. When the scheduling plan changes, determine whether to immediately perform the scheduling switch or delay the switch based on the smooth switching execution strategy. Establishing a smooth switching execution strategy can avoid the risk of mission interruption and waste of flight resources caused by rescheduling.

[0133] As a preferred implementation, the smooth switching execution strategy constructs a switching cost function to comprehensively evaluate the additional flight distance, mission interruption risk, and remaining execution time incurred by the UAV continuing to execute the current task versus switching to the new scheduling scheme. In this embodiment, the specific calculation formula for the cost function is as follows:

[0134] ;

[0135] in, Let be the weight coefficients of each term in the cost function, satisfying ; The normalized distance value reflects the additional travel distance required after switching strategies. The calculation process is as follows: Definition For drones Continue current task Until the estimated remaining voyage is completed, For drones The required flight distance after switching to the new scheduling plan. The original distance difference is... ;when When switching will cause excessive flight, a penalty is required; when The time interval indicates that switching saves bandwidth and requires no penalty. Therefore, the effective extra distance for calculating the cost function is defined as follows: , The reference scale parameter for distance normalization represents an acceptable order of magnitude of additional range and is configured by the system; furthermore, this embodiment sets... ,when hour, No additional cost; when hour The distance will not be infinitely magnified in terms of its value;

[0136] This represents the risk of task interruption, indicating the extent of risk to the current task if the task is switched immediately. A higher value indicates a greater risk. In this embodiment, it is also normalized to 0–1. It has almost no impact; 1. Extremely dangerous;

[0137] , representing the normalized remaining execution time, characterizing the current drone Continue to carry out the mission How much longer until it's finished?

[0138] like Figure 8 As shown, when the switching cost function value is greater than the preset cost threshold... When the scheduling switch is delayed, the UAV can hand over the task after completing the current sub-stage or the current task, thereby achieving a smooth transition from the old scheduling scheme to the new scheduling scheme and avoiding frequent task interruptions and excessive resource consumption; when the calculation result of the switching cost function is less than or equal to the preset threshold, the scheduling switch is executed immediately.

[0139] The drones were assigned to perform all rescue missions according to the final scheduling plan.

[0140] The above is a collaborative scheduling method for disaster relief missions that considers the performance differences of UAVs proposed in this invention. This embodiment also provides an application system for this method, which specifically includes:

[0141] The UAV capability modeling module is used to quantify the endurance, payload capacity, perception capability, communication capability, maneuverability and environmental adaptability of each UAV into a multi-dimensional UAV capability vector.

[0142] The rescue mission requirement modeling module is used to quantify the endurance requirements, payload requirements, perception requirements, communication requirements, mobility requirements, and environmental adaptation requirements of each rescue mission into a multi-dimensional mission requirement vector.

[0143] The matching degree calculation module is used to calculate the matching degree between the drone and the rescue mission based on the weighted similarity calculation method;

[0144] The preference sequence construction module is used to construct task-side preference sequences and UAV-side preference sequences based on the matching degree, respectively.

[0145] The stable matching scheduling module is used to generate an initial scheduling scheme between drones and rescue missions based on the mission-side preference sequence and the drone-side preference sequence, using an improved stable matching algorithm while taking into account the drone's capability constraints.

[0146] The dynamic scheduling module is used to monitor the drone status and environmental status information in real time during the execution of rescue missions, and to dynamically update the initial scheduling plan when preset trigger conditions are met.

[0147] The switching control module is used to control the UAV to execute the current task or switch to a new scheduling task based on a smooth switching execution strategy when the scheduling scheme is updated.

[0148] Furthermore, this embodiment also provides specific application effects of the method proposed in this invention in specific examples of flood control and disaster relief application scenarios:

[0149] Assuming that after a flood, the command center simultaneously issues four rescue missions and can deploy five drones. The system utilizes the 6-dimensional drone capability-mission requirement vector of this invention, where all components have been normalized to... That is, the existing collection of drones: Task set: The meaning of the task is defined as follows:

[0150] T1 dam inspection requires drones to have high perception, high maneuverability, and a certain amount of communication data transmission.

[0151] T2 communication relay requires drones to have high communication capabilities, long endurance, and strong environmental adaptability.

[0152] T3 material delivery requires drones to have high payload capacity and a certain range of flight time.

[0153] T4 transfer guidance requires high level of awareness and high level of communication from the UAV;

[0154] According to the method proposed in this invention, the drone capability vectors of the five drones at the initial time zero are calculated:

[0155] ;

[0156] ;

[0157] ;

[0158] ;

[0159] ;

[0160] Task requirement vectors for the 4 rescue missions:

[0161] ;

[0162] ;

[0163] ;

[0164] .

[0165] Given the original weight vector:

[0166] The matching degree between the drone and the rescue mission at the initial time zero is shown in Table 3 below:

[0167] Table 3. Matching Degree Between UAVs and Rescue Missions

[0168]

[0169] Before performing drone-task matching, the following task constraints are added: the T2 communication relay task requires redundancy and necessitates the collaboration of two drones; each of the other tasks requires one drone; each drone executes only one task at a time; thus, the initial scheduling scheme is as follows:

[0170] T1 → U3, T2 → U4 + U1, T3 → U2, T4 → U5;

[0171] The specific matching process is as follows:

[0172] First round of requests: T1 → U3, T2 → U5, T3 → U2, T4 → U5;

[0173] Drone processing requests: U5 receives T2 and T4 simultaneously, sorts them according to its own preference, and accepts T4 first and rejects T2; U3 accepts T1; U2 accepts T3;

[0174] Second round of requests: T2 is rejected by U5, so the preference sequence shifts to U1, the second-ranked player in the preference list; U1 accepts T2;

[0175] Third round of requests: T2 still needs a second drone, so T2 submits its request to the next preferred U3; U3 has a preference for T1 and therefore rejects T2.

[0176] Fourth round of requests: T2 requests U44 again, and then U4 accepts the request;

[0177] Finally, there is no situation where neither drone prefers a particular task pair, thus a stable match is achieved, and the initial scheduling scheme is obtained.

[0178] Next, we introduce environmental and state changes, assuming the following changes: increased rainfall, decreased visibility, increased wind speed, and increased communication interference; U3 experiences a significant decrease in battery power and a decline in link quality due to continuous patrols.

[0179] The command center has raised higher requirements for T2 relay; T1 inspections are conducted to ensure transmission quality and improve communication and environmental adaptability.

[0180] In this example, the environmental impact factor is set as follows:

[0181] The updated real-time weights are ;

[0182] Due to the change in drone status, the updated U3 experienced a significant decrease in battery life and communication capabilities, at which point its drone capability vector... The U4's range is reduced due to continuous standby, thus affecting its drone capability vector. The capability vectors of the remaining drones remain unchanged for the time being;

[0183] The updated task requirement vector is: T1 dam patrol has increased communication / mobility / environmental requirements, meaning that at this time... T2 communication relay requires higher battery life, that is, at this time ;

[0184] After the matching score was updated, the matching score between Task T2 and each drone changed significantly, including:

[0185] The matching degree between T2 and U5: 0.963→0.959 (still the best fit), the matching degree between T2 and U1: 0.906→0.892, the matching degree between T2 and U3: 0.878 → 0.782 (significantly decreased, due to decreased battery life and communication + greater emphasis on communication / environment), the matching degree between T2 and U4: 0.849→0.824, the matching degree between T2 and U2: 0.829→0.835;

[0186] Although the compatibility between mission T1 and the various drones changed only slightly, the following changes also occurred:

[0187] The matching degree between T1 and U5 increased from 0.967 to 0.971 (more favorable), the matching degree between T1 and U1 increased from 0.987 to 0.985, and the matching degree between T1 and U3 increased from 0.993 to 0.984. Although the matching degree between U3 and T1 is still relatively high, rescheduling is still required because the battery life threshold has been triggered.

[0188] The rescheduled solution is as follows: T2→U3+U4, with U3 serving as the primary relay; although U4's battery life decreases, its load / platform stability allows it to remain a backup relay; T1→U1, U11 still has a high match for patrols and does not trigger the risk of insufficient battery life; T3→U2, U2 is still the most suitable for task deployment; T4→U5. U3 is removed from the execution target of the long-duration patrol task T1 and replaced as the executor of the short-duration / near-distance T2 guiding task; that is, changing the executor of task T1 from U3 to U1 reduces the risk of insufficient battery life, while changing the executor of task T2 from U1 to U3. Although the high battery life requirement of T2 cannot correspond to the significant decrease in U3's battery power and link quality, based on the handshake principle, there is no better choice. Finally, the switching cost function is calculated to determine whether to delay the switch (U3 completes the current patrol segment and then hands it over to U1) or switch immediately (U1 immediately takes over the patrol).

[0189] In summary, the scheduling method proposed in this invention can meet the application requirements of the complex scenario of flood control and disaster relief, which is characterized by multiple tasks, high dynamism, and high risk, and can effectively improve rescue efficiency, scheduling rationality, and system security.

[0190] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

[0191] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A collaborative scheduling method for disaster relief missions that considers the performance differences of unmanned aerial vehicles (UAVs), characterized in that, Specifically, the following steps are included: A multi-dimensional quantitative analysis of the performance differences of drones and the requirements of rescue missions was conducted to construct a drone capability vector and a mission requirement vector. Based on the UAV capability vector and mission requirement vector, the matching degree between the UAV and the rescue mission is calculated, and the UAV side preference sequence and mission side preference sequence are generated. Based on the UAV-side preference sequence and the mission-side preference sequence, and considering the UAV capability constraints, an improved Gale–Shapley stable matching algorithm is adopted to perform stable many-to-many matching between UAVs and rescue missions, resulting in an initial scheduling scheme; specifically: Initialize the allocation status of all drones and rescue missions, and introduce the drone-side preference sequence for each drone and the mission-side preference sequence for each rescue mission; In the initial request round, each rescue mission sends an execution request to the most preferred drone according to its own mission-side preference sequence; rescue missions requiring multi-drone coordination send execution requests to multiple drones in sequence according to the preference sequence until the coordination quantity requirement is met. Each drone sorts all received execution requests according to its own drone-side preference sequence; under the premise that a single drone is allowed to perform multiple rescue missions, the drone's capability constraints are considered to filter and accept the top K execution requests that have the highest preference and meet the drone's capability constraints. The rejected mission continues to send execution requests to the next drone in the mission preference sequence; after receiving a new execution request, the drone makes a new decision: that is, it reorders, filters and accepts the rescue missions corresponding to all the received execution requests; The process of repeatedly issuing rescue mission requests and making decisions by drones continues until all rescue missions are accepted or the mission-side preference sequence is exhausted. At this point, the repeated operation stops and an initial scheduling plan is obtained. Rescue missions that have not been accepted even after the mission-side preference sequence is exhausted are directly placed into the queue of missions that cannot be executed. The drone is controlled to perform rescue missions according to the initial scheduling plan. During the rescue mission, the drone's status information and environmental status information are monitored in real time. When the real-time detected status information meets the preset rescheduling trigger conditions, the drone's capability vector, mission requirement vector, and weight vectors involved in calculating the matching degree are updated. The matching degree is recalculated, and the improved Gale-Shapley stable matching algorithm is executed to generate a new scheduling plan. The specific update of the matching degree weight vector is as follows: A real-time environmental impact factor vector is defined. The original weight vector used to calculate the matching degree between the UAV and the rescue mission is weighted and corrected item by item, and then normalized to obtain the updated weight vector. The environmental impact factors include the impact factors on endurance, the impact factors on payload, the impact factors on perception capability, the impact factors on communication capability, the impact factors on maneuverability, and the impact factors on environmental adaptability. When the scheduling plan changes, determine whether to immediately switch the schedule or delay the switch based on the smooth switching execution strategy; allocate drones to perform all rescue missions according to the final scheduling plan, and repeat the above process until all missions are completed.

2. The method for collaborative scheduling of disaster relief missions considering the performance differences of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The multi-dimensional quantitative analysis of UAV performance and rescue mission requirements, and the construction of UAV capability vector and mission requirement vector, are specifically as follows: The six dimensions of capability—endurance, payload capacity, perception capability, communication capability, maneuverability, and environmental adaptability—are selected as the construction factors for the UAV capability vector. The endurance is calculated based on the drone's current remaining battery power, maximum flight time, and unit energy consumption rate. The payload capacity is calculated based on the UAV's maximum payload and its current remaining payload capacity; The perception capability is calculated by weighting the image resolution, field of view, and number and type of sensors carried by the UAV. The communication capability is calculated using a weighted average of communication bandwidth, signal strength, and image transmission stability. The maneuverability is calculated based on the UAV's maximum flight speed, rate of climb, and wind resistance level. The environmental adaptability is calculated based on the drone's rain and dust protection capabilities, temperature adaptability range, and humidity of the drone's flight environment. Six dimensions of requirements—endurance requirements, payload requirements, perception requirements, communication requirements, mobility requirements, and environmental adaptability requirements—are selected as the construction factors for the task requirement vector. The requirements of each dimension are normalized to form the task requirement vector.

3. The method for collaborative scheduling of disaster relief missions considering the performance differences of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The process of calculating the matching degree between the UAV and the rescue mission based on the UAV capability vector and the mission requirement vector, and generating the UAV side preference sequence and the mission side preference sequence, specifically involves: The weighted cosine similarity function is used to calculate and weight the similarity between the UAV capability vector and the mission requirement vector in each dimension, so as to obtain the matching degree between the UAV and the rescue mission. For a single drone, calculate its matching degree with all rescue missions, sort the rescue missions from high to low according to the matching degree, and obtain the drone side preference sequence of the drone. For a single rescue mission, calculate its matching degree with all drones, sort the drones from high to low according to the matching degree, and obtain the mission-side preference sequence for the rescue mission.

4. The method for collaborative scheduling of disaster relief missions considering the performance differences of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The UAV capability constraints include UAV endurance constraints, UAV payload capacity constraints, task acceptance quantity constraints, and execution order constraints, wherein: The drone's endurance constraint is: the total endurance requirement for a rescue mission undertaken by a drone must not exceed the drone's endurance. The drone payload capacity constraint is: the total payload requirement of a rescue mission received by a drone shall not exceed the payload capacity of the drone. The task acceptance quantity constraint is: a drone can only accept a preset number of rescue tasks, and priority is given to rescue tasks that are earlier in the preference sequence; The execution order constraint is as follows: when a drone performs multiple rescue missions, the mission execution order is sorted according to the principle of range priority.

5. The method for collaborative scheduling of disaster relief missions considering the performance differences of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The preset rescheduling trigger conditions specifically include: The drone's real-time remaining battery power is below the preset safety threshold; The quality of the drone's real-time communication link is lower than the preset stable communication threshold; The real-time wind speed of the environment where the drone is located exceeds the maximum permissible wind speed, the real-time precipitation exceeds the maximum permissible precipitation, and the real-time visibility is lower than the minimum permissible visibility. The drone malfunctions or its real-time health status measurement falls below the preset minimum health threshold.

6. The method for collaborative scheduling of disaster relief missions considering the performance differences of unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, When the scheduling scheme changes, determining whether to perform an immediate scheduling switch or a delayed switch based on the smooth switch execution strategy specifically involves: The smooth switching execution strategy comprehensively evaluates the additional range, mission interruption risk, and remaining execution time caused by the UAV continuing to execute the current mission versus switching to the new scheduling scheme by constructing a switching cost function. When the switching cost function value is greater than the preset cost threshold, the scheduling switch is delayed; when the calculation result of the switching cost function is less than or equal to the preset cost threshold, the scheduling switch is executed immediately.

7. The application system of the disaster relief mission collaborative scheduling method considering the performance differences of UAVs according to claim 1, characterized in that, include: The UAV capability modeling module is used to quantify the endurance, payload capacity, perception capability, communication capability, maneuverability and environmental adaptability of each UAV into a multi-dimensional UAV capability vector. The rescue mission requirement modeling module is used to quantify the endurance requirements, payload requirements, perception requirements, communication requirements, mobility requirements, and environmental adaptation requirements of each rescue mission into a multi-dimensional mission requirement vector. The matching degree calculation module is used to calculate the matching degree between the drone and the rescue mission based on the weighted similarity calculation method; The preference sequence construction module is used to construct task-side preference sequences and UAV-side preference sequences based on the matching degree, respectively. The stable matching scheduling module is used to generate an initial scheduling scheme between drones and rescue missions based on the mission-side preference sequence and the drone-side preference sequence, using an improved stable matching algorithm while taking into account the drone's capability constraints. The dynamic scheduling module is used to monitor the drone status and environmental status information in real time during the execution of rescue missions, and to dynamically update the initial scheduling plan when preset trigger conditions are met. The switching control module is used to control the UAV to execute the current task or switch to a new scheduling task based on a smooth switching execution strategy when the scheduling scheme is updated.

Images