Machine nest site design method and device, electronic equipment and storage medium

Abstract

The application provides a UAV nest site selection design method and device, electronic equipment and a storage medium, and belongs to the technical field of unmanned aerial vehicles. In the method, a double-layer network architecture of "original + extension" is constructed, and an incomplete information game model is introduced, so that fixed (i.e. the nest position in the original network set) and temporary nests (i.e. the nest position in the extended network set) can be flexibly distinguished according to weather and demand changes. The marginal contribution is quantified by using the Shapley value, and high-value strategy combinations are screened, so that multi-objective collaborative optimization of patrol efficiency, deployment cost and network robustness is realized, and the unity of individual rationality and overall benefit is ensured. Further, by using the improved particle swarm algorithm based on game tree mapping, combined with the gravity operator guidance and the inferior particle elimination and replacement mechanism, the convergence speed and optimization accuracy of the algorithm in the complex solution space are significantly improved, and finally an optimized control strategy scheme with scientific site selection and fine resource allocation is output.

Description

Technical Field

[0001] This invention relates to the technical field of unmanned aerial vehicles (UAVs), and in particular to a method, apparatus, electronic device, and storage medium for UAV nesting design. Background Technology

[0002] With the continuous iterative evolution and integrated innovation of next-generation information technologies such as artificial intelligence, cloud computing, big data, the Internet of Things, and mobile internet, drone inspection has become an emerging inspection technology. Its unique advantages, such as flexible deployment, wide detection range, and immunity to ground traffic congestion, make it a promising application in numerous fields including traffic monitoring, urban law enforcement, municipal infrastructure inspection, and emergency management, attracting widespread attention from academia and industry. However, many challenges remain regarding the scientific selection of drone nest locations and network design.

[0003] Current research largely assumes that drone nests used for inspection coexist with other airports, neglecting the influence of geographical factors, urban security factors, infrastructure distribution, and the density of inspection tasks. Furthermore, from an urban perspective, there is a lack of technical support for the comprehensive network design of drone inspections in environments with multiple drones conducting collaborative inspections. Nest location planning fails to fully integrate multiple dimensions such as urban functional layout, inspection needs, and meteorological conditions, resulting in low inspection coverage efficiency. Network robustness design lacks a systematic approach, making it difficult to handle inspection task scheduling in complex environments. Resource allocation models are simplistic, failing to effectively integrate game theory and optimization algorithms, and cannot achieve coordinated optimization of nest locations and inspection strategies in dynamic scenarios. Existing game theory models are mostly limited to scenarios with complete information, making it difficult to adapt to incomplete information environments where strategies are not shared in real-world networks.

[0004] In summary, traditional UAV nesting methods have shortcomings in terms of dynamic adaptability, multi-objective collaborative optimization, and algorithm convergence efficiency. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a nesting design method, apparatus, electronic device and storage medium to alleviate the shortcomings of traditional UAV nesting methods in terms of dynamic adaptability, multi-objective collaborative optimization and algorithm convergence efficiency.

[0006] In a first aspect, the present invention provides a nesting site selection design method, comprising: Collect geographic facility data, historical meteorological data, and inspection requirement standards for the target area, and construct a basic feature dataset after fusion processing; Based on the aforementioned basic feature dataset, candidate nest location sets are selected, and the K-Means fuzzy clustering algorithm is used to divide the candidate nest location sets into an original network set and an extended network set, and the union of the two is established as the set of players in the game. Based on the set of players in the game, the strategy space and multidimensional reward function under incomplete information are defined, and the adjustment set weights and the prior probabilities of the edge-adding set are initialized, thereby constructing an initial structure of the game tree that includes all potential competitive paths and all potential cooperative alliance paths. Based on the initial structure of the game tree, all potential cooperative alliance paths are traversed, the expected return value of each player in each game is calculated using the Shapley value formula, and strategy combinations with expected return values ​​greater than the benefits of independent deployment are selected to form the optimal strategy solution space. The game tree nodes are mapped to a particle swarm. The strategy vector in the optimal strategy solution space is encoded as the initial position of each particle in the particle swarm. The multidimensional reward function is mapped to the fitness function of the particles. At the same time, the velocity of each particle is initialized. Then, the particle state is iteratively optimized using a velocity-position update formula containing a gravity operator. A mechanism for eliminating inferior particles and replacing the dimension of the global optimal strategy is executed until the convergence condition is met. The global optimal strategy vector that maximizes the global reward function and encodes the nest location deployment and resource allocation information is obtained. Then, the global optimal strategy vector is decoded and analyzed to obtain an optimized control strategy scheme that includes the nest location deployment scheme and resource allocation strategy.

[0007] Furthermore, the K-Means fuzzy clustering algorithm is used to divide the candidate nest location set into an original network set and an extended network set, including: Extract the feature vector of each candidate nest location from the candidate nest location set; Based on the feature vector, calculate the first Euclidean distance from each candidate nest location to the preset original network cluster center, and the second Euclidean distance from each candidate nest location to the preset extended network cluster center; The network type to which each candidate nest location belongs is determined based on the first Euclidean distance and the second Euclidean distance, wherein the network type includes: the original network set, or the extended network set, wherein the candidate nest locations in the original network set are fixed nest candidate points, and the candidate nest locations in the extended network set are temporary nest candidate points.

[0008] Furthermore, based on the set of players in the game, a strategy space and a multidimensional reward function under incomplete information are defined, and the adjustment set weights and the prior probabilities of the increasing edge set are initialized. This leads to the construction of an initial structure for the game tree containing all potential competitive paths and all potential cooperative alliance paths, including: The strategy space of each player in the set of players in the game is defined as a binary choice set, and the global strategy space is composed of the combination of strategies of all players. For any strategy combination, through a multidimensional reward function Calculate the theoretical payoff for each player, where, Indicates strategy combination When it happens, the people involved The theoretical profit value, This indicates an indicator of inspection efficiency. This represents the deployment cost coefficient. Indicators representing network robustness , , These are the weighting coefficients; We set adjustment set weights to characterize the degree of belief bias of players regarding the strategy choices of other players, and set incremental edge set prior probability set to characterize the probability that the first player believes the second player is willing to form a cooperative alliance with them. The game tree skeleton is constructed with the empty alliance state as the root node and the process of players making decisions on whether to choose a location or not as the branch paths. At each decision node of the game tree, a cooperative branch pointing to the cooperative alliance state is generated based on the probability value in the prior probability set of the augmenting edge set. The theoretical payoff values ​​of each generated branch node are weighted and corrected using the adjustment set weights to obtain the expected payoff value of each node, thereby forming the initial structure of the game tree that includes all potential competitive paths and all potential cooperative alliance paths.

[0009] Furthermore, by traversing all potential cooperative alliance paths, the expected return value for each player in each game is calculated using the Shapley value formula, and strategy combinations whose expected return value is greater than the benefit of independent deployment are selected, including: Traverse all potential cooperative alliance paths in the initial structure of the game tree. For any player and their potential cooperative alliance, through... Calculate the marginal contribution of the player to the potential cooperative alliance, where, The person in question Regarding the potential cooperative alliance marginal contribution Indicates that the player is included. New Alliance Total alliance revenue, This indicates that no players were included. The original alliance Total alliance revenue; Using the Shapley value formula Calculate the expected return value for the player, where, Indicates the person involved The expected return value among all possible potential cooperative alliances. n Denotes the set of players in the game. NThe total number of people involved in the game. Indicating potential cooperative alliances S The number of players included. N { j} indicates the person outside the game. The set of all other players besides [the other player]. S N { j} indicates a potential cooperative alliance S It does not include Any sub-alliance; Calculate the independent deployment benefit when the player deploys independently; Players whose expected return value is greater than the benefit of independent deployment and their corresponding site selection strategy combinations are selected, and all the selected site selection strategy combinations constitute the optimal strategy solution space.

[0010] Furthermore, the particle state is iteratively optimized using a velocity-position update formula incorporating a gravitational operator, including: In each iteration, using the formula and Update the particle velocity and particle position to obtain the updated particle velocity and particle position, where, Indicates the first Iter In the next iteration, the particle i The velocity vector, Indicates the first Iter Particles in 1st iteration i The velocity vector, oh Represents the gravitational operator, c 1. c 2 represents individual learning factors and social learning factors, respectively. Represents particles i The historical best position, namely the particle i The position vector where the fitness value is optimal in each iteration. Indicates the first Iter Particles in 1st iteration i The position vector, This represents the global optimal position of the particle swarm, that is, the position vector of all particles in the current particle swarm when their fitness values ​​are optimal in each iteration. Indicates the first Iter After the nth iteration, the th i The particle in the first m Positional components in 2D space, Indicates the first Iter After the first iteration, the... i The particle in the first m Positional components in 2D space, Indicates the first Iter During the iteration, the first i The particle in the first m Velocity components in 2D space.

[0011] Furthermore, a mechanism for eliminating inferior particles and replacing the dimension of the globally optimal strategy is implemented, including: Calculate the fitness value of all updated particle positions and sort the updated particles in ascending order of the fitness value; Eliminate the lower-ranked particles according to a predetermined ratio to obtain the remaining particles; Select the global best particle with the highest fitness value from the remaining particles, and extract a preset number of key dimensions from its strategy vector. The extracted key dimensions are randomly replaced with the corresponding dimensions of other particles in the remaining particles to obtain a new generation of particle swarms, which then enters the next iteration.

[0012] Further, the globally optimal policy vector is decoded and parsed, including: Read the dimension representing the location status from the global optimal strategy vector, extract the corresponding geographical coordinates, and form the nest location deployment scheme; The resource allocation strategy is obtained by reading the numerical dimension representing resource allocation in the global optimal strategy vector, and determining the number of resident UAVs at each nest location and the weather trigger threshold for dynamic temporary nests in combination with the network type.

[0013] Secondly, the present invention also provides a nesting location design apparatus, comprising: The data collection and construction unit is used to collect geographic facility data, historical meteorological data and inspection requirement standards of the target area, and construct a basic feature dataset after fusion processing. The filtering and partitioning unit is used to filter the candidate nest location set based on the basic feature dataset, and to use the K-Means fuzzy clustering algorithm to partition the candidate nest location set into an original network set and an extended network set, and to establish the union of the two as the set of players in the game. The construction unit is used to define the strategy space and multidimensional reward function under incomplete information based on the set of players in the game, and initialize the adjustment set weights and the prior probabilities of the edge set, thereby constructing an initial structure of the game tree that includes all potential competitive paths and all potential cooperative alliance paths. The calculation and filtering unit is used to traverse all potential cooperative alliance paths based on the initial structure of the game tree, calculate the expected return value of each player in each game using the Shapley value formula, and filter out the strategy combinations whose expected return value is greater than the independent deployment benefit to form the optimal strategy solution space. The iterative optimization unit maps game tree nodes to a particle swarm, encodes the strategy vector in the preferred strategy solution space as the initial position of each particle in the particle swarm, maps the multidimensional reward function to the fitness function of the particles, initializes the velocity of each particle, and then iteratively optimizes the particle state using a velocity-position update formula containing a gravity operator. It also executes a mechanism for eliminating inferior particles and replacing the dimension of the global optimal strategy until the convergence condition is met, obtaining a globally optimal strategy vector that maximizes the global reward function and encodes nest location deployment and resource allocation information. Finally, the globally optimal strategy vector is decoded and analyzed to obtain an optimized control strategy scheme containing nest location deployment and resource allocation strategies.

[0014] Thirdly, the present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the computer program to implement the method described in the first aspect.

[0015] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the method described in the first aspect.

[0016] This invention provides a nest location design method, comprising: collecting geographical facility data, historical meteorological data, and inspection requirement standards of the target area, and constructing a basic feature dataset after fusion processing; filtering candidate nest location sets based on the basic feature dataset, and using the K-Means fuzzy clustering algorithm to divide the candidate nest location sets into an original network set and an extended network set, and establishing the union of the two as the game player set; defining the strategy space and multidimensional reward function under incomplete information based on the game player set, and initializing the adjustment set weights and the prior probabilities of the edge-adding set, thereby constructing an initial structure of a game tree containing all potential competitive paths and all potential cooperative alliance paths; and traversing all potential cooperative alliance paths based on the initial structure of the game tree, calculating the mid-game value of each game using the Shapley value formula. The expected return value of each player is determined, and strategy combinations with expected returns greater than those of independent deployment are selected to form the optimal strategy solution space. The game tree nodes are mapped to a particle swarm, and the strategy vectors in the optimal strategy solution space are encoded as the initial positions of each particle in the swarm. The multidimensional reward function is mapped to the fitness function of the particles, and the velocity of each particle is initialized. Then, the particle state is iteratively optimized using a velocity-position update formula containing a gravity operator, and a mechanism for eliminating inferior particles and replacing the dimension of the global optimal strategy is executed until the convergence condition is met. The global optimal strategy vector that maximizes the global reward function and encodes the nest location deployment and resource allocation information is obtained. The global optimal strategy vector is then decoded and analyzed to obtain an optimized control strategy scheme that includes nest location deployment and resource allocation strategies. As described above, the UAV nesting design method of this invention effectively solves the problem of poor dynamic adaptability of traditional methods by constructing a "original + extended" two-layer network architecture and introducing an incomplete information game model. It can flexibly distinguish between fixed (i.e., nesting locations in the original network set) and temporary (i.e., nesting locations in the extended network set) deployments based on weather and demand changes. By using Shapley value to quantify marginal contributions and screen high-value strategy combinations, it achieves multi-objective collaborative optimization of inspection efficiency, deployment cost, and network robustness, ensuring the unity of individual rationality and overall benefits. Furthermore, by using an improved particle swarm optimization algorithm based on game tree mapping, combined with gravity operator guidance and inferior particle elimination and replacement mechanism, it significantly improves the convergence speed and optimization accuracy of the algorithm in complex solution space, avoids local optimum traps, and finally outputs an optimized control strategy scheme that combines scientific site selection and fine resource allocation. This greatly improves the overall operational efficiency and environmental adaptability of the UAV inspection system and alleviates the shortcomings of traditional UAV nesting methods in terms of dynamic adaptability, multi-objective collaborative optimization, and algorithm convergence efficiency. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating a nesting site selection design method provided in an embodiment of the present invention; Figure 2 A schematic diagram of a nesting location design game provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a nesting location design device provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0019] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. 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.

[0020] Traditional UAV nesting methods have shortcomings in dynamic adaptability, multi-objective collaborative optimization, and algorithm convergence efficiency.

[0021] Based on this, the nest location design method of this invention effectively solves the problem of poor dynamic adaptability of traditional methods by constructing a "original + extended" two-layer network architecture and introducing an incomplete information game model. It can flexibly distinguish between fixed (i.e., nest locations in the original network set) and temporary nest (i.e., nest locations in the extended network set) deployments according to weather and demand changes. By using Shapley value to quantify marginal contributions and screen high-value strategy combinations, it achieves multi-objective collaborative optimization of inspection efficiency, deployment cost and network robustness, ensuring the unity of individual rationality and overall benefits. Furthermore, by using an improved particle swarm algorithm based on game tree mapping, combined with gravity operator guidance and inferior particle elimination and replacement mechanism, it significantly improves the convergence speed and optimization accuracy of the algorithm in complex solution space, avoids local optimum traps, and finally outputs an optimized control strategy scheme that combines scientific location selection and fine resource allocation, greatly improving the overall operational efficiency and environmental adaptability of the UAV inspection system.

[0022] To facilitate understanding of this embodiment, a nesting location design method disclosed in this embodiment of the invention will first be described in detail.

[0023] Example 1: According to an embodiment of the present invention, an embodiment of a nesting location design method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0024] Figure 1 This is a flowchart of a nesting location design method according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps: Step S102: Collect geographic facility data, historical meteorological data and inspection requirement standards for the target area, and construct a basic feature dataset after fusion processing; Specifically, the nesting site selection design method of the present invention can be applied to power facility inspection scenarios. Of course, it can also be applied to other inspection scenarios. The embodiments of the present invention do not impose specific limitations on the application scenarios of the above method.

[0025] When the method of this invention is applied to a power facility inspection scenario, the target area can be the administrative region of a city, and the geographical facility data can include the latitude and longitude distribution of substations and the topology of transmission lines. These data can also be collectively referred to as power facility data, thereby determining key inspection areas. The historical meteorological data can include the city's meteorological data for the past 5 years, specifically including average wind speed (gridized), number of precipitation days, etc., and then a meteorological influence factor matrix is ​​constructed (this matrix is ​​usually a two-dimensional matrix, its dimension defined as meteorological influence factor matrix ∈ Where N (number of rows) represents the number of spatial grids, that is, the total number of discrete grids into which the target city or inspection area is divided, and M (number of columns) represents the meteorological feature dimension, that is, the types of meteorological indicators collected and used for evaluation, such as average wind speed, number of precipitation days, etc. The above inspection requirement standards are combined with the inspection standards of the power sector to determine the inspection frequency requirements of each region, such as the substation in the city center being inspected once every two weeks, and the suburban transmission lines being inspected once a month.

[0026] After obtaining the above data, the geographic coordinates of candidate satellite docking stations (described below) and deployment cost coefficients are determined using geographic infrastructure data. These coefficients are related to land attribute factors (0.1 for self-owned substation land; 0.5 for leased commercial land; 0.9 for newly constructed land acquired through land acquisition), accessibility factors (normalized values ​​of distance to the nearest main road calculated based on road network data; the greater the distance, the larger the value (higher maintenance costs)), and power supply factors (whether there is an existing high-voltage interface; 0.8 for no interface requiring new line construction, 0.2 for an interface), and can be the sum of these three factors. A meteorological influence factor matrix is ​​constructed using historical meteorological data (historical meteorological data (wind speed, precipitation, thunderstorms) are matched with the geographic grid space to convert physical quantities into dimensionless influence factors in the [0,1] interval). Based on this, network robustness indices (robustness refers to the network's ability to survive under adverse conditions. The worse the weather and the greater the distance, the lower the robustness. This can be obtained by designing formulas based on the above principles) and theoretical inspection frequency (the worse the weather, the faster the equipment ages, or the more frequent the inspections need to be (or conversely, the theoretical frequency is reduced because it is difficult to fly into the area). This can also be achieved by designing formulas, even without adhering to this principle). Using the inspection demand standard combined with the theoretical inspection frequency, spatial demand density is generated through normalization (obtained by multiplying the inspection demand standard by the theoretical inspection frequency and then normalizing). Combining the above processing results, a basic feature dataset containing geographic coordinates, deployment cost coefficients, spatial demand density, and network robustness indices is constructed.

[0027] Step S104: Based on the basic feature dataset, filter the candidate nest location set, and use the K-Means fuzzy clustering algorithm to divide the candidate nest location set into the original network set and the extended network set, and establish the union of the two as the set of players in the game. Specifically, combining inspection task priorities and basic feature datasets, candidate drone nesting locations are selected within the city (covering areas around substations, power transmission line hubs, and areas with relatively stable weather conditions). Areas deemed absolutely unsuitable are eliminated using "inspection task priority." If the network robustness index of a grid is below a certain threshold (meaning the location experiences frequent strong winds or thunderstorms), it cannot physically take off or land regardless of the task priority, and is directly eliminated. If the deployment cost coefficient of a grid exceeds the company's budget limit (meaning it's prime real estate in a core business district), it is directly eliminated. If the inspection task priority of a certain area is... "Special grade" points (such as nuclear power plants and core hubs) are forcibly added to a temporary set even if their costs are slightly excessive or their robustness is slightly low. After initial screening, the remaining points need to be queued. During queuing, the values ​​in the basic feature dataset react with the weights of the task priorities, i.e., a weighted calculation is performed. The parameters include task weights (directly converted from the inspection task priorities), demand density (i.e., spatial demand density), robustness (i.e., network robustness), and cost penalty (i.e., deployment cost). The resulting numerical scores are sorted in descending order, and the top n points are selected to form the candidate nest locations, thus obtaining the candidate nest location set. P ={ p 1, p 2,…, p n Then, the K-Means fuzzy clustering algorithm is used to divide the candidate nest location set into an original network set and an extended network set. The candidate nest locations in the original network set correspond to the fixed nests in the core inspection area, and the candidate nest locations in the extended network set correspond to the dynamically supplemented temporary nests. The union of the original network set and the extended network set is established as the set of players in the game.

[0028] Step S106: Based on the set of players in the game, define the strategy space and multidimensional reward function under incomplete information, and initialize the adjustment set weights and the prior probabilities of the edge-adding set, thereby constructing the initial structure of the game tree containing all potential competitive paths and all potential cooperative alliance paths. Step S108: Based on the initial structure of the game tree, traverse all potential cooperative alliance paths, use the Shapley value formula to calculate the expected return value of each player in each game, and select strategy combinations whose expected return value is greater than the benefit of independent deployment to form the optimal strategy solution space. Step S110: Map the game tree nodes to a particle swarm, encode the strategy vector in the optimal strategy solution space as the initial position of each particle in the particle swarm, and map the multidimensional reward function to the fitness function of the particles. At the same time, initialize the velocity of each particle, and then use the velocity-position update formula containing the gravity operator to iteratively optimize the particle state. Then, execute the mechanism of eliminating inferior particles and replacing the dimension of the global optimal strategy until the convergence condition is met, and obtain the global optimal strategy vector that maximizes the global reward function and encodes the nest location deployment and resource allocation information. Then, decode and analyze the global optimal strategy vector to obtain the optimized control strategy scheme containing the nest location deployment scheme and resource allocation strategy.

[0029] The above provides a brief overview of the nesting site selection design method of the present invention. The specific details involved are described in detail below.

[0030] In an optional embodiment of the present invention, the K-Means fuzzy clustering algorithm is used to divide the candidate nest location set into an original network set and an extended network set, specifically including the following steps: (1) Extract the feature vector of each candidate nest location from the candidate nest location set; Specifically, extract the three-dimensional features of each candidate nest location: geographic coordinates ( x , y ), density of surrounding inspection needs d (i.e., spatial demand density), to obtain the feature vector of each candidate nest location ( x , y , d ).

[0031] (2) Calculate the first Euclidean distance from each candidate nest location to the preset original network cluster center and the second Euclidean distance from each candidate nest location to the preset extended network cluster center based on the feature vector; Specifically, the K-Means fuzzy clustering algorithm is used to calculate the first Euclidean distance from each candidate nest location to the preset original network cluster center, and the second Euclidean distance from each candidate nest location to the preset extended network cluster center, thus obtaining the Euclidean distance matrix. D =[ d ij ] n×n ,in, , x ik For position p i The k Dimensional features (such as geographic coordinates, density of surrounding inspection needs, etc.). The first cluster center of the original network or the cluster center of the extended network k Dimensional features.

[0032] Two virtual ideal centers are artificially set up to represent the optimal characteristics of a "fixed nest" and a "temporary nest," respectively.

[0033] (3) Determine the network type of each candidate nest location based on the first Euclidean distance and the second Euclidean distance. The network type includes: the original network set or the extended network set. The candidate nest locations in the original network set are fixed nest candidate points, and the candidate nest locations in the extended network set are temporary nest candidate points.

[0034] Specifically, for each candidate nest location, the first Euclidean distance and the second Euclidean distance are compared. If the first Euclidean distance is less than the second Euclidean distance, the network type of the candidate nest location is determined to be the network type corresponding to the network cluster center corresponding to the first Euclidean distance; if the first Euclidean distance is greater than the second Euclidean distance, the network type of the candidate nest location is determined to be the network type corresponding to the network cluster center corresponding to the second Euclidean distance. Figure 2 As shown.

[0035] In an optional embodiment of the present invention, based on the set of players in the game, a strategy space and a multidimensional reward function under incomplete information are defined, and the adjustment set weights and the prior probabilities of the increasing edge set are initialized, thereby constructing an initial structure of the game tree containing all potential competitive paths and all potential cooperative alliance paths. Specifically, the following steps are included: (1) Define the strategy space of each player in the set of players in the game as a binary choice set, and the global strategy space is composed of the strategy combinations of all players; Specifically, define the set of players in the competitive-cooperative game algorithm. N (i.e., the set of players in the game), strategy space S and multidimensional reward function U In practice, since positional strategies are not shared within the network, it is necessary to assume that players cannot obtain the states, strategies, and payoffs of other players. The incomplete information competition-cooperative game model is represented as: ,in, N={1,2,3,…,m} ( m (Number of network locations) In the game model, the first... i The strategy space of each participant; This indicates the strategies adopted by other game participants; This represents its reward function; the higher the efficiency of drone inspection, the greater its reward.

[0036] The binary choice set mentioned above can be {addressed, not addressed}.

[0037] (2) For any strategy combination, through a multidimensional reward function Calculate the theoretical payoff for each player, where, Indicates strategy combination When it happens, the people involved The theoretical profit value, Indicates the efficiency of inspection (e.g., the area covered per unit time). This represents the deployment cost coefficient. Indicators representing network robustness , , These are the weighting coefficients. + + =1.

[0038] (3) Set adjustment set weights to characterize the degree of belief bias of players regarding the strategy choices of other players, and set incremental edge set prior probability set to characterize the probability that the first player believes the second player is willing to form a cooperative alliance with him; Specifically, according to the players i Preferences, initial adjustment set weights and the prior probability set of the increasing edge set .

[0039] (4) Using the empty alliance state as the root node, and the process of players making site selection or non-site selection decisions in turn as the branch paths, construct the game tree skeleton; Specifically, the players i The initial strategy is considered as an intermediate node in the game process, and a game tree node is established. This involves traversing all players, initializing the initial strategies of players at each drone hull location, establishing the corresponding game tree nodes, and constructing the initial layer of the game tree. (The root node and first-level nodes have been generated). Location j Whether one can join a given partnership depends on prior probabilities. and marginal contribution .

[0040] (5) At each decision node of the game tree, determine whether to generate a cooperative branch pointing to the cooperative alliance state based on the probability value in the prior probability set of the increasing edge set. Specifically, if the probability value is lower than a preset threshold (the exemplary range of the preset threshold is set to [0.5, 0.7]), the branch is not generated; if the probability value is higher than or equal to the preset threshold, the branch is generated.

[0041] (6) The theoretical payoff values ​​of each generated branch node are adjusted by adjusting the weight of the adjustment set to obtain the expected payoff value of each node, thereby forming the initial structure of the game tree containing all potential competitive paths and all potential cooperative alliance paths.

[0042] The specific calculation logic of the weighted adjustment is as follows: multiply the theoretical profit value by the trust gain factor, which is calculated based on the adjustment set weights among the players involved in the current branch. ,in, This is the corrected expected return value. This is the theoretical profit value. Let be the adjustment set weight for player i against player j. This adjustment amplifies the expected payoff if the adjustment set weight is positive (indicating trust), representing the added value of high reliability in the cooperative path; conversely, it reduces the expected payoff if the adjustment set weight is negative (indicating distrust), representing the cost reduction due to potential default risk. Ultimately, this forms the initial structure of a game tree containing all potential competing paths and all potential cooperative alliance paths, where the node values ​​have already incorporated the trust dimension adjustment.

[0043] The above process will be illustrated with an example below: Root node initialization → Branch generation (competition / cooperation determination) → Node benefit calculation (theoretical value → expected value) → Recursive expansion → Termination and pruning.

[0044] Root Initialization: State definition: Create a root node to represent the initial state of the game, i.e., the empty alliance state.

[0045] Attribute settings: At this time, no player makes any decisions, and the cumulative benefit is 0.

[0046] Hierarchical label: Labeled as level 0 ( Level=0 ).

[0047] Branch Generation Strategy: From the current node (let's assume it's node) Nodek The set of players who have made decisions is... Dk Let's start by considering the next player in the game who needs to make a decision. i ( i∈N Dk Generate child node branches for this node: Basic branch (competitive path): action A (Location selection): Generate a child node representing a player in the game. iSelect "Site Selection".

[0048] action B (No location selection): Generates a child node representing a player. i Select "No location selected".

[0049] Note: This is the standard competitive path for all game trees.

[0050] Probability Branch (Collaboration Path - Key Innovation Point): Query probability: Search the prior probability set of the increasing edge set to obtain the current players. i With existing alliance members (or specific targets) j The probability of cooperation.

[0051] Threshold determination: If the probability of cooperation is greater than or equal to a preset threshold: an additional "cooperation edge" is generated, pointing to a special "cooperative alliance state" child node. This edge represents the probability that the two parties will reach a cooperative consensus under incomplete information.

[0052] If the probability of cooperation is less than the preset threshold: no cooperative edge is generated, or a virtual edge with very low weight is generated (depending on the specific algorithm implementation, usually the branch is pruned directly to simplify the tree structure).

[0053] Significance: This step visualizes the "incomplete information" as the topological structure of a tree (some paths are open, and some are closed).

[0054] Payoff Calculation & Correction: For each newly generated child node (representing a new temporary strategy combination): Calculate the theoretical return (UtheoryUtheory) → Call the multidimensional return function formula:

[0055] Apply the adjustment set weight correction: obtain the current player's belief bias weights towards other participants; perform the correction calculation to obtain the expected payout value.

[0056] Assigning values ​​to node attributes: The expected return value is stored in the attributes of the child node, serving as the basis for subsequent evaluation of the path's merits.

[0057] Recursive Expansion: Use the newly generated child node as the current node, repeat the above steps, and process the decision of the next player.

[0058] The depth of the tree gradually increases until all... NAll players in the game have completed their decisions (i.e., reached the leaf node).

[0059] Termination and Pruning: Termination condition: When the path length equals the total number of players. N The path ends and forms a leaf node when all players have made a decision on "choosing a location / not choosing a location / cooperating".

[0060] Dynamic pruning (optional optimization): During the construction process, if the expected reward value of an intermediate node of a certain path has fallen below a certain extremely low threshold (obviously a poor solution), the expansion of that branch can be stopped in advance to reduce the amount of computation.

[0061] In an optional embodiment of the present invention, all potential cooperative alliance paths are traversed, the expected return value of each player in each game is calculated using the Shapley value formula, and strategy combinations with expected return values ​​greater than the benefits of independent deployment are selected. Specifically, this includes the following steps: (1) Traverse all potential cooperative alliance paths in the initial structure of the game tree. For any player and its potential cooperative alliance, through... Calculate the marginal contribution of individuals within the calculation center to potential cooperative alliances, among which, Indicates the person involved Potential cooperative alliances marginal contribution Indicates that the player is included. New Alliance Total alliance revenue, This indicates that no players were included. The original alliance Total alliance revenue; Specifically, the total revenue of the alliance is obtained by summing or weighting the individual revenues (i.e., theoretical revenue values) of all players in the alliance based on a multidimensional return function.

[0062] (2) Using the Shapley value formula Calculate the expected return for each player in the game, where, Indicates the person involved The expected return value among all possible potential cooperative alliances. n Represents the set of players in the game. N The total number of people involved in the game. Indicating potential cooperative alliances S The number of players included. N { j} indicates the person outside the game. The set of all other players besides [the other player]. S N { j} indicates a potential cooperative alliance S It does not include Any sub-alliance; Specifically, in the allocation model of cooperative game problems, the Shapley value unifies individual rationality and collective rationality, allocating additional payoffs based on the marginal contributions of alliance members to each potential alliance. According to the definition of the Shapley value, the calculation of position... j Expected return function: , Indicates that the player is included. New Alliance Total alliance revenue, This indicates that no players were included. The original alliance Total alliance revenue.

[0063] (3) Calculate the independent deployment benefits when players deploy independently; Specifically, the independent deployment revenue of players is calculated based on the multidimensional reward function.

[0064] (4) Select players whose expected return is greater than the benefit of independent deployment and their corresponding site selection strategy combinations, and form the optimal strategy solution space by selecting all the site selection strategy combinations.

[0065] In an optional embodiment of the present invention, the particle state is iteratively optimized using a velocity-position update formula containing a gravitational operator, specifically including the following steps: In each iteration, using the formula and Update the particle velocity and particle position to obtain the updated particle velocity and particle position, where, Indicates the first Iter In the next iteration, the particle i The velocity vector, Indicates the first Iter Particles in 1st iteration i The velocity vector, oh Represents the gravitational operator, c 1. c 2 represents individual learning factors and social learning factors, respectively. Represents particles i The historical best position, namely the particle i The position vector where the fitness value is optimal in each iteration. Indicates the first Iter Particles in 1st iteration i The position vector, This represents the global optimal position of the particle swarm, that is, the position vector of all particles in the current particle swarm when their fitness values ​​are optimal in each iteration. Indicates the first Iter After the nth iteration, the th i The particle in the first m Positional components in 2D space, Indicates the first Iter After the first iteration, the... i The particle in the first m Positional components in 2D space, Indicates the first Iter During the iteration, the first i The particle in the first m Velocity components in 2D space.

[0066] Specifically, the iterative process of particle swarm optimization is as follows: Game tree nodes are mapped to a particle swarm, where nodes of the same generation correspond to the particle swarm, individual nodes correspond to individual particles, the strategy set corresponds to the gravity operator, and the payoff function corresponds to the reward function. The payoff function matrix is ​​established, represented as: ,in, This indicates that particles are only affected by a strategy. i The gains obtained It is a gravity operator strategy. It's a collaborative framework.

[0067] Record the particle position and velocity, denoted as: , .

[0068] Each player represents a particle Position and velocity at m The search space is iteratively updated, and the position and velocity of each particle in the swarm are updated according to the following formula based on the gravitational operator: , ,in, It is a particle i The speed of the previous generation It is a particle i The historical best strategy, namely, the interaction of particles in the game tree. i The strategy is to backtrack. It is the best strategy for particle swarm optimization. For gravity operators, the value range is set to 0~1. Iter Let be the number of iterations, and be the learning factor. Repeat the iterations until Iter reaches its maximum value.

[0069] In an optional embodiment of the present invention, the execution of the inferior particle elimination and global optimal strategy dimension replacement mechanism specifically includes the following steps: (1) Calculate the fitness values ​​of all updated particle positions and sort the updated particles in ascending order of fitness values; (2) Eliminate the lower-ranked particles by a predetermined proportion (10% ~ 30%, based on the trade-off between "population diversity maintenance" and "convergence speed") to obtain the remaining particles; (3) Select the global best particle with the highest fitness value from the remaining particles and extract the preset number of key dimensions in its strategy vector (the "preset number" is usually set to 1 / 3 to 1 / 2 of the total dimension of the strategy vector, based on the game between "inheritance of advantageous genes" and "avoiding premature convergence"). (4) Randomly replace the extracted key dimensions with the corresponding dimensions of other particles in the remaining particles to obtain a new generation of particle swarms and enter the next iteration.

[0070] Specifically, once the collaborative iteration process begins, the worst performer will be eliminated in each iteration. or The proportion of particles is determined, and the particle with the optimal strategy at this point is selected as a high-quality cooperative partner. m Any dimension in the particle swarm is replaced with other particles in the particle swarm, and the particle reward function is re-evaluated. When the number of iterations reaches a preset maximum value or the reward function converges, the optimized control strategy for the particle swarm is output, including the hive location deployment scheme and resource allocation strategy.

[0071] In an optional embodiment of the present invention, decoding and parsing the globally optimal policy vector includes: (1) Read the dimension representing the location status in the global optimal strategy vector, extract the corresponding geographic coordinates, and form a nest location deployment plan (distinguish between fixed nests and temporary nests). (2) Read the numerical dimension representing resource allocation in the global optimal strategy vector, and determine the configuration of the number of resident UAVs at each nest location and the weather trigger threshold of the dynamic temporary nest in combination with the network type to obtain the resource allocation strategy.

[0072] When the nest location belongs to the original network (fixed nest): The system identifies this node as a backbone node and adopts a "high redundancy configuration strategy." It reads the spatial demand density at this location and calculates the number of resident drones based on this density (using a preset formula). For fixed drone nests, weather trigger thresholds are typically not set, or they are set as the extreme physical parameters of the drones to ensure all-weather operational capability.

[0073] When the nest location belongs to an extended network (temporary nest): Quantity configuration: Read the space demand density and calculate the number of resident drones based on the space demand density (preset formula). Usually, only the number of drones that meet the minimum operating unit is configured.

[0074] Weather threshold setting: Read the network robustness index at this location. Due to the weak risk resistance of temporary nests, the system dynamically sets the weather trigger threshold based on the network robustness index (weighted formula for the network robustness index).

[0075] This invention incorporates multi-dimensional features such as urban functional layout, infrastructure inspection needs, and historical meteorological data into nest location planning, breaking through the limitations of traditional single-geographic coordinate site selection and constructing a candidate set that better fits actual inspection scenarios. A fuzzy membership matrix (i.e., Euclidean distance) is introduced to cluster candidate nest locations, dividing them into an "original network" (core fixed nests) and an "extended network" (dynamic temporary nests), enabling hierarchical deployment of inspection resources and improving the network's adaptability to dynamic demands (such as sudden weather events). The nest location design problem is transformed into a hierarchical optimization of "competitive game" and "cooperative game": Competition phase: Search for Nash equilibrium based on strategy preferences to resolve the strategy choice conflict regarding nest location; Cooperation phase: Optimize alliance resource allocation based on Shapley value to ensure the unity of individual rationality and overall efficiency.

[0076] By utilizing the Shapley value to calculate the "marginal contribution" of joining an alliance based on nested positions, this method serves as the basis for distributing revenue in cooperative game theory, addressing the challenge of balancing fairness and efficiency in traditional resource allocation and ensuring alliance stability. The game tree nodes are mapped to a particle swarm, and through iterative updates of the gravity operator, historical best strategy, and global best strategy, combined with particle elimination and strategy replacement mechanisms, the search for Nash equilibrium and strategy optimization are accelerated, improving computational efficiency.

[0077] The inventive points of this invention are as follows: 1. A "fixed + dynamic" hierarchical network architecture based on fuzzy clustering: Breaking away from the traditional single geographic coordinate site selection, a fuzzy membership matrix is ​​introduced to scientifically divide the nest network into "original network" (fixed nests for routine inspections) and "extended network" (dynamic temporary nests for sudden weather events or high loads), significantly improving the network's adaptability to dynamic environments.

[0078] 2. A competitive-cooperative two-layer game mechanism incorporating Shapley values: This innovative approach transforms the site selection problem into a hierarchical optimization process involving "competition" and "cooperation." The competition layer resolves strategy conflicts (Nash equilibrium), while the cooperation layer utilizes Shapley values ​​to calculate marginal contributions and allocate benefits. This solves the problem of balancing fairness and efficiency in traditional resource allocation, making it particularly suitable for real-world scenarios with incomplete information (lack of strategy sharing).

[0079] 3. An improved particle swarm optimization algorithm based on game tree mapping: By mapping complex game tree nodes to particle swarms for search, a unique "particle elimination and policy dimension replacement" mechanism was designed. By eliminating inferior particles and injecting the optimal policy dimension, the algorithm effectively avoids getting trapped in local optima, significantly improving convergence speed and optimization accuracy under multidimensional constraints.

[0080] Example 2: This invention also provides a nesting location design device, which is mainly used to execute the nesting location design method provided in Embodiment 1 of this invention. The nesting location design device provided in this invention will be described in detail below.

[0081] Figure 3 This is a schematic diagram of a nesting location design device according to an embodiment of the present invention, as shown below. Figure 3 As shown, the device mainly includes: a data acquisition and construction unit 10, a filtering and segmentation unit 20, a construction unit 30, a calculation and filtering unit 40, and an iterative optimization unit 50, wherein: The data collection and construction unit 10 is used to collect geographic facility data, historical meteorological data and inspection requirement standards of the target area, and construct a basic feature dataset after fusion processing. The filtering and partitioning unit 20 is used to filter the candidate nest location set based on the basic feature dataset, and to use the K-Means fuzzy clustering algorithm to partition the candidate nest location set into the original network set and the extended network set, and to establish the union of the two as the set of players in the game. Building unit 30 is used to define the strategy space and multidimensional reward function under incomplete information based on the set of players in the game, and initialize the adjustment set weights and the prior probabilities of the edge-adding set, thereby constructing the initial structure of the game tree containing all potential competitive paths and all potential cooperative alliance paths. The calculation and screening unit 40 is used to traverse all potential cooperative alliance paths based on the initial structure of the game tree, calculate the expected return value of each player in each game using the Shapley value formula, and screen out the strategy combinations whose expected return value is greater than the independent deployment benefit, thus forming the optimal strategy solution space. The iterative optimization unit 50 is used to map game tree nodes to particle swarms, encode the strategy vector in the optimal strategy solution space as the initial position of each particle in the particle swarm, and map the multidimensional reward function to the fitness function of the particles. At the same time, it initializes the velocity of each particle, and then uses the velocity-position update formula containing the gravity operator to iteratively optimize the particle state. It also executes the elimination of inferior particles and the replacement mechanism of the global optimal strategy dimension until the convergence condition is met, and obtains the global optimal strategy vector that maximizes the global reward function and encodes the nest location deployment and resource allocation information. Then, it decodes and analyzes the global optimal strategy vector to obtain the optimized control strategy scheme containing the nest location deployment scheme and resource allocation strategy.

[0082] This invention provides a nest location design device, comprising: collecting geographical facility data, historical meteorological data, and inspection requirement standards of a target area, and constructing a basic feature dataset after fusion processing; filtering candidate nest location sets based on the basic feature dataset, and using the K-Means fuzzy clustering algorithm to divide the candidate nest location sets into an original network set and an extended network set, and establishing the union of the two as the game player set; defining the strategy space and multidimensional reward function under incomplete information based on the game player set, and initializing the adjustment set weights and the prior probabilities of the edge-adding set, thereby constructing an initial structure of a game tree containing all potential competitive paths and all potential cooperative alliance paths; and traversing all potential cooperative alliance paths based on the initial structure of the game tree, calculating the mid-game value of each game using the Shapley value formula. The expected return value of each player is determined, and strategy combinations with expected returns greater than those of independent deployment are selected to form the optimal strategy solution space. The game tree nodes are mapped to a particle swarm, and the strategy vectors in the optimal strategy solution space are encoded as the initial positions of each particle in the swarm. The multidimensional reward function is mapped to the fitness function of the particles, and the velocity of each particle is initialized. Then, the particle state is iteratively optimized using a velocity-position update formula containing a gravity operator, and a mechanism for eliminating inferior particles and replacing the dimension of the global optimal strategy is executed until the convergence condition is met. The global optimal strategy vector that maximizes the global reward function and encodes the nest location deployment and resource allocation information is obtained. The global optimal strategy vector is then decoded and analyzed to obtain an optimized control strategy scheme that includes nest location deployment and resource allocation strategies. As described above, the UAV nesting design device of this invention effectively solves the problem of poor dynamic adaptability of traditional methods by constructing a "original + extended" two-layer network architecture and introducing an incomplete information game model. It can flexibly distinguish between fixed (i.e., nesting locations in the original network set) and temporary nesting (i.e., nesting locations in the extended network set) deployments based on weather and demand changes. By using Shapley value to quantify marginal contributions and screen high-value strategy combinations, it achieves multi-objective collaborative optimization of inspection efficiency, deployment cost, and network robustness, ensuring the unity of individual rationality and overall benefits. Furthermore, by using an improved particle swarm optimization algorithm based on game tree mapping, combined with gravity operator guidance and inferior particle elimination and replacement mechanism, it significantly improves the convergence speed and optimization accuracy of the algorithm in complex solution space, avoids local optimum traps, and finally outputs an optimized control strategy scheme that combines scientific site selection and fine resource allocation. This greatly improves the overall operational efficiency and environmental adaptability of the UAV inspection system and alleviates the shortcomings of traditional UAV nesting methods in dynamic adaptability, multi-objective collaborative optimization, and algorithm convergence efficiency.

[0083] Optionally, the filtering and segmentation unit is further configured to: extract the feature vector of each candidate nest location from the candidate nest location set; calculate the first Euclidean distance from each candidate nest location to the preset original network cluster center and the second Euclidean distance from each candidate nest location to the preset extended network cluster center based on the feature vector; determine the network type to which each candidate nest location belongs based on the first Euclidean distance and the second Euclidean distance, wherein the network type includes: the original network set or the extended network set, where the candidate nest locations in the original network set are fixed nest candidate points and the candidate nest locations in the extended network set are temporary nest candidate points.

[0084] Optionally, the building unit is also used to: define the strategy space of each player in the set of game players as a binary choice set, and the global strategy space is composed of the strategy combinations of all players; for any strategy combination, a multidimensional reward function is used to... Calculate the theoretical payoff for each player, where, Indicates strategy combination When it happens, the people involved The theoretical profit value, This indicates an indicator of inspection efficiency. This represents the deployment cost coefficient. Indicators representing network robustness , , The weights are set as follows: Adjustment set weights are set to characterize the degree of belief bias of players regarding the strategy choices of other participants; an incremental edge set prior probability set is set to characterize the probability that the first player believes the second player is willing to form a cooperative alliance with them; the empty alliance state is used as the root node, and the process of players making site selection or non-site selection decisions in sequence is used as the branch paths to construct the game tree skeleton; at each decision node of the game tree, the probability value in the incremental edge set prior probability set is used to determine whether to generate a cooperative branch pointing to the cooperative alliance state; the theoretical payoff values ​​of each generated branch node are weighted and corrected using the adjustment set weights to obtain the expected payoff value of each node, thus forming the initial structure of the game tree containing all potential competitive paths and all potential cooperative alliance paths.

[0085] Optionally, the calculation and filtering unit is also used to: traverse all potential cooperative alliance paths in the initial structure of the game tree, and for any player and their potential cooperative alliance, through... Calculate the marginal contribution of individuals within the calculation center to potential cooperative alliances, among which, Indicates the person involved Potential cooperative alliances marginal contribution Indicates that the player is included. New Alliance Total alliance revenue, This indicates that no players were included. The original alliance Total alliance revenue; using the Shapley value formula Calculate the expected return for each player in the game, where, Indicates the person involved The expected return value among all possible potential cooperative alliances. n Represents the set of players in the game. N The total number of people involved in the game. Indicating potential cooperative alliances S The number of players included. N { j} indicates the person outside the game. The set of all other players besides [the other player]. S N { j} indicates a potential cooperative alliance S It does not include Any sub-alliance; calculate the independent deployment benefit when players deploy independently; select players whose expected return value is greater than the independent deployment benefit and their corresponding location strategy combinations, and construct the optimal strategy solution space from all selected location strategy combinations.

[0086] Optionally, the iterative optimization unit is also used to: in each iteration, utilize the formula and Update the particle velocity and particle position to obtain the updated particle velocity and particle position, where, Indicates the first Iter In the next iteration, the particle i The velocity vector, Indicates the first Iter Particles in 1st iteration i The velocity vector, oh Represents the gravitational operator, c 1. c 2 represents individual learning factors and social learning factors, respectively. Represents particles i The historical best position, namely the particle i The position vector where the fitness value is optimal in each iteration. Indicates the first Iter Particles in 1st iteration i The position vector, This represents the global optimal position of the particle swarm, that is, the position vector of all particles in the current particle swarm when their fitness values ​​are optimal in each iteration. Indicates the first Iter After the nth iteration, the th i The particle in the first m Positional components in 2D space, Indicates the first Iter After the first iteration, the... i The particle in the first m Positional components in 2D space, Indicates the first Iter During the iteration, the first i The particle in the first m Velocity components in 2D space.

[0087] Optionally, the iterative optimization unit is also used to: calculate the fitness values ​​of all updated particle positions and sort the updated particles in ascending order of fitness values; eliminate a preset proportion of the updated particles with lower rankings to obtain the remaining particles; select the global best particle with the highest fitness value from the remaining particles and extract a preset number of key dimensions from its policy vector; randomly replace the extracted key dimensions with the corresponding dimensions of other particles in the remaining particles to obtain a new generation of particle swarms and enter the next iteration.

[0088] Optionally, the iterative optimization unit is also used to: read the dimension representing the location status in the global optimal strategy vector, extract the corresponding geographical coordinates, and form a nest location deployment scheme; read the numerical dimension representing resource configuration in the global optimal strategy vector, and determine the configuration of the number of resident drones at each nest location and the weather trigger threshold of dynamic temporary nests in combination with the network type, so as to obtain the resource allocation strategy.

[0089] The device provided in this embodiment of the invention has the same implementation principle and technical effect as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the aforementioned method embodiment.

[0090] like Figure 4 As shown in the embodiment of this application, an electronic device 600 includes a processor 601, a memory 602, and a bus. The memory 602 stores machine-readable instructions that can be executed by the processor 601. When the electronic device is running, the processor 601 communicates with the memory 602 via the bus. The processor 601 executes the machine-readable instructions to perform the steps of the above-described nesting addressing design method.

[0091] Specifically, the memory 602 and processor 601 mentioned above can be general-purpose memory and processor, without any specific limitations. When the processor 601 runs the computer program stored in the memory 602, it can execute the above-mentioned nesting addressing design method.

[0092] The processor 601 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 601 or by instructions in software form. The processor 601 may be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 602, and processor 601 reads the information in memory 602 and, in conjunction with its hardware, completes the steps of the above method.

[0093] Corresponding to the above-described nesting design method, this application also provides a computer-readable storage medium storing machine-executable instructions. When the machine-executable instructions are invoked and executed by a processor, the machine-executable instructions cause the processor to perform the steps of the above-described nesting design method.

[0094] The nesting design apparatus provided in this application embodiment can be specific hardware on the device or software or firmware installed on the device. The implementation principle and technical effects of the apparatus provided in this application embodiment are the same as those in the foregoing method embodiments. For the sake of brevity, any parts not mentioned in the apparatus embodiment can be referred to the corresponding content in the foregoing method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, apparatuses, and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.

[0095] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0096] For example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0097] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0098] In addition, the functional units in the embodiments provided in this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0099] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the nesting location design method described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0100] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0101] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A nesting site selection design method, characterized in that, include: Collect geographic facility data, historical meteorological data, and inspection requirement standards for the target area, and construct a basic feature dataset after fusion processing; Based on the aforementioned basic feature dataset, candidate nest location sets are selected, and the K-Means fuzzy clustering algorithm is used to divide the candidate nest location sets into an original network set and an extended network set, and the union of the two is established as the set of players in the game. Based on the set of players in the game, the strategy space and multidimensional reward function under incomplete information are defined, and the adjustment set weights and the prior probabilities of the edge-adding set are initialized, thereby constructing an initial structure of the game tree that includes all potential competitive paths and all potential cooperative alliance paths. Based on the initial structure of the game tree, all potential cooperative alliance paths are traversed, the expected return value of each player in each game is calculated using the Shapley value formula, and strategy combinations with expected return values ​​greater than the benefits of independent deployment are selected to form the optimal strategy solution space. The game tree nodes are mapped to a particle swarm. The strategy vector in the optimal strategy solution space is encoded as the initial position of each particle in the particle swarm. The multidimensional reward function is mapped to the fitness function of the particles. At the same time, the velocity of each particle is initialized. Then, the particle state is iteratively optimized using a velocity-position update formula containing a gravity operator. A mechanism for eliminating inferior particles and replacing the dimension of the global optimal strategy is executed until the convergence condition is met. The global optimal strategy vector that maximizes the global reward function and encodes the nest location deployment and resource allocation information is obtained. Then, the global optimal strategy vector is decoded and analyzed to obtain an optimized control strategy scheme that includes the nest location deployment scheme and resource allocation strategy.

2. The method according to claim 1, characterized in that, The K-Means fuzzy clustering algorithm is used to divide the candidate nest location set into an original network set and an extended network set, including: Extract the feature vector of each candidate nest location from the candidate nest location set; Based on the feature vector, calculate the first Euclidean distance from each candidate nest location to the preset original network cluster center, and the second Euclidean distance from each candidate nest location to the preset extended network cluster center; The network type to which each candidate nest location belongs is determined based on the first Euclidean distance and the second Euclidean distance, wherein the network type includes: the original network set, or the extended network set, wherein the candidate nest locations in the original network set are fixed nest candidate points, and the candidate nest locations in the extended network set are temporary nest candidate points.

3. The method according to claim 1, characterized in that, Based on the set of players in the game, a strategy space and a multidimensional reward function under incomplete information are defined, and the adjustment set weights and the prior probabilities of the increasing edge set are initialized. This leads to the construction of an initial game tree structure containing all potential competitive paths and all potential cooperative alliance paths, including: The strategy space of each player in the set of players in the game is defined as a binary choice set, and the global strategy space is composed of the combination of strategies of all players. For any strategy combination, through a multidimensional reward function Calculate the theoretical payoff for each player, where, Indicates strategy combination When it happens, the people involved The theoretical profit value, This indicates an indicator of inspection efficiency. This represents the deployment cost coefficient. Indicators representing network robustness , , These are the weighting coefficients; We set adjustment set weights to characterize the degree of belief bias of players regarding the strategy choices of other players, and set incremental edge set prior probability set to characterize the probability that the first player believes the second player is willing to form a cooperative alliance with them. The game tree skeleton is constructed with the empty alliance state as the root node and the process of players making decisions on whether to choose a location or not as the branch paths. At each decision node of the game tree, a cooperative branch pointing to the cooperative alliance state is generated based on the probability value in the prior probability set of the augmenting edge set. The theoretical payoff values ​​of each generated branch node are weighted and corrected using the adjustment set weights to obtain the expected payoff value of each node, thereby forming the initial structure of the game tree that includes all potential competitive paths and all potential cooperative alliance paths.

4. The method according to claim 1, characterized in that, Traverse all potential cooperative alliance paths, calculate the expected return for each player using the Shapley value formula, and select strategy combinations whose expected return is greater than the benefit of independent deployment, including: Traverse all potential cooperative alliance paths in the initial structure of the game tree. For any player and their potential cooperative alliance, through... Calculate the marginal contribution of the player to the potential cooperative alliance, where, The person in question Regarding the potential cooperative alliance marginal contribution Indicates that the player is included. New Alliance Total alliance revenue, This indicates that no players were included. The original alliance Total alliance revenue; Using the Shapley value formula Calculate the expected return value for the player, where, Indicates the person involved The expected return value among all possible potential cooperative alliances. n Denotes the set of players in the game. N The total number of people involved in the game. Indicating potential cooperative alliances S The number of players included. N { j } indicates the person outside the game. The set of all other players besides [the other player]. S N { j } indicates a potential cooperative alliance S It does not include Any sub-alliance; Calculate the independent deployment benefit when the player deploys independently; Players whose expected return value is greater than the benefit of independent deployment and their corresponding site selection strategy combinations are selected, and all the selected site selection strategy combinations constitute the optimal strategy solution space.

5. The method according to claim 1, characterized in that, The particle state is iteratively optimized using a velocity-position update formula incorporating a gravitational operator, including: In each iteration, using the formula and Update the particle velocity and particle position to obtain the updated particle velocity and particle position, where, Indicates the first Iter Particles in the next iteration i The velocity vector, Indicates the first Iter Particles in 1st iteration i The velocity vector, ω Represents the gravitational operator, c 1. c 2 represents individual learning factors and social learning factors, respectively. Represents particles i The historical best position, namely the particle i The position vector where the fitness value is optimal in each iteration. Indicates the first Iter Particles in 1st iteration i The position vector, This represents the global optimal position of the particle swarm, that is, the position vector of all particles in the current particle swarm when their fitness values ​​are optimal in each iteration. Indicates the first Iter After the nth iteration, the th i The particle in the first m Positional components in 2D space, Indicates the first Iter After the first iteration, the... i The particle in the first m Positional components in 2D space, Indicates the first Iter During the iteration, the first i The particle in the first m Velocity components in 2D space.

6. The method according to claim 5, characterized in that, The mechanism for eliminating inferior particles and replacing dimensions based on the globally optimal strategy includes: Calculate the fitness value of all updated particle positions and sort the updated particles in ascending order of the fitness value; Eliminate the lower-ranked particles according to a predetermined ratio to obtain the remaining particles; Select the global best particle with the highest fitness value from the remaining particles, and extract a preset number of key dimensions from its strategy vector. The extracted key dimensions are randomly replaced with the corresponding dimensions of other particles in the remaining particles to obtain a new generation of particle swarms, which then enters the next iteration.

7. The method according to claim 1, characterized in that, Decoding and parsing the globally optimal policy vector includes: Read the dimension representing the location status from the global optimal strategy vector, extract the corresponding geographical coordinates, and form the nest location deployment scheme; The resource allocation strategy is obtained by reading the numerical dimension representing resource allocation in the global optimal strategy vector, and determining the number of resident UAVs at each nest location and the weather trigger threshold for dynamic temporary nests in combination with the network type.

8. A nesting location design device, characterized in that, include: The data collection and construction unit is used to collect geographic facility data, historical meteorological data and inspection requirement standards of the target area, and construct a basic feature dataset after fusion processing. The filtering and partitioning unit is used to filter the candidate nest location set based on the basic feature dataset, and to use the K-Means fuzzy clustering algorithm to partition the candidate nest location set into an original network set and an extended network set, and to establish the union of the two as the set of players in the game. The construction unit is used to define the strategy space and multidimensional reward function under incomplete information based on the set of players in the game, and initialize the adjustment set weights and the prior probabilities of the edge-adding set, thereby constructing an initial structure of the game tree containing all potential competitive paths and all potential cooperative alliance paths. The calculation and filtering unit is used to traverse all potential cooperative alliance paths based on the initial structure of the game tree, calculate the expected return value of each player in each game using the Shapley value formula, and filter out the strategy combinations whose expected return value is greater than the independent deployment benefit to form the optimal strategy solution space. The iterative optimization unit maps game tree nodes to a particle swarm, encodes the strategy vector in the preferred strategy solution space as the initial position of each particle in the particle swarm, maps the multidimensional reward function to the fitness function of the particles, initializes the velocity of each particle, and then iteratively optimizes the particle state using a velocity-position update formula containing a gravity operator. It also executes a mechanism for eliminating inferior particles and replacing the dimension of the global optimal strategy until the convergence condition is met, obtaining a globally optimal strategy vector that maximizes the global reward function and encodes nest location deployment and resource allocation information. Finally, the globally optimal strategy vector is decoded and analyzed to obtain an optimized control strategy scheme containing nest location deployment and resource allocation strategies.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program thereon, characterized in that, The computer program is executed by the processor to perform the method of any one of claims 1 to 7.

Images