Multi-brand and multi-type unmanned aerial vehicle equipment access configuration method

By performing three-dimensional modeling and multi-dimensional difference calculation on the drone protocol, dynamic adaptation configuration parameters are generated, and the adaptability and automation problems of access configurations of multi-brand and multi-type drone equipment are solved, achieving efficient and accurate access configurations.

CN120416882APending Publication Date: 2025-08-01浙江志诚云信息科技有限公司
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202510632030.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-05-16
Publication Date
2025-08-01

AI Technical Summary

Technical Problem

The existing UAV equipment access configuration methods are insufficient in multi-brand and multi-type environments, making it difficult to achieve efficient and accurate access configuration, especially when facing new or heterogeneous systems, it requires manual intervention or frequent reconfiguration.

Method used

By modeling the three-dimensional protocol feature of the target UAV protocol, the communication layer feature matrix, the data layer association map and the control layer spatiotemporal behavior tensor, combined with the multi-dimensional protocol difference calculation model, the feature comparison and analysis are performed, a dynamic adaptation configuration parameter set is generated, and the access configuration automation and stability is achieved through a closed-loop optimization mechanism.

Benefits of technology

It realizes accurate modeling and dynamic configuration of multi-brand and multi-type drone equipment, improves the degree of automation and adaptability of access configurations, and enhances the accuracy, compatibility and long-term adaptability of configuration details.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure BDA0005405614920000041
    Figure BDA0005405614920000041
  • Figure BDA0005405614920000044
    Figure BDA0005405614920000044
  • Figure BDA0005405614920000047
    Figure BDA0005405614920000047
Patent Text Reader

Abstract

The invention relates to the field of unmanned aerial vehicle communication and protocol adaptation, and discloses a multi-brand and multi-type unmanned aerial vehicle equipment access configuration method, which comprises the following steps of: performing three-dimensional feature modeling for a target protocol by constructing a communication layer feature matrix, a data layer association map and a control layer space-time behavior tensor; based on a multi-dimensional protocol difference degree calculation model, the communication features, the data structure difference and the control behavior difference are fused, and the difference between a target protocol and a historical protocol is measured; the method comprises the following steps: selecting a most relevant basic configuration template through a protocol feature clustering and deviation compensation strategy, introducing a communication feature compensation item of a historical protocol, and generating a personalized access configuration parameter set adaptive to a target protocol; the method is suitable for access configuration generation of multiple types of unmanned aerial vehicle equipment, the adaptability, compatibility and stability of configuration can be improved, and the automatic configuration capability of an unmanned aerial vehicle protocol in a heterogeneous environment is enhanced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of drone communication and protocol adaptation, and specifically provides a method for accessing and configuring multi-brand and multi-type drone devices. Background Art

[0002] With the rapid development of drone technology, a variety of drone devices of different brands and types have emerged in the market, and there are significant differences at the protocol level among these devices. Each drone device usually relies on different communication protocols, data exchange mechanisms, and control strategies, which makes it increasingly complex to access and configure devices in a multi-brand and multi-type drone environment. The existing drone access and configuration methods mainly focus on the optimization of specific protocols or device types, and it is difficult to adapt to diverse and dynamically changing protocol requirements. Especially when facing new drones or heterogeneous systems, it is difficult to achieve efficient and accurate access and configuration.

[0003] Existing solutions usually adopt static configuration templates, which are customized for a specific device or protocol, ignoring the differences between protocols and the dynamic adaptation requirements. These methods often rely on manually defined rules or pre-configured parameters, and there is a lag in supporting new protocols, and it is difficult to flexibly adjust according to device characteristics during operation. In addition, the existing methods lack a comprehensive multi-dimensional protocol feature modeling, and cannot effectively quantify the differences between different protocols at the communication, data, control, etc. levels, resulting in poor accuracy and adaptability of access configuration in a complex and changing usage environment.

[0004] In addition, the existing technology lacks a dynamic learning and optimization mechanism for protocol differences. In practical applications, due to the continuous changes of devices or protocols, traditional methods cannot adaptively adjust the configuration, and often require manual intervention or frequent manual re-configuration, which affects the flexibility and automation level of the system. Especially in the scenario of multi-protocol collaboration, how to dynamically update the protocol configuration to cope with the changing protocol characteristics and network environment is an urgent problem to be solved.

[0005] Therefore, the existing technology has significant defects in the adaptability, intelligence of drone device access configuration, and configuration stability in a multi-protocol environment, and there is an urgent need for an access configuration method that can handle multi-brand, multi-type protocol differences and can achieve real-time dynamic optimization. Summary of the Invention

[0006] Aiming at the deficiencies of the existing technology, the present invention provides a method for accessing and configuring multi-brand and multi-type drone devices, which can accurately model and dynamically configure the target drone protocol in a multi-protocol heterogeneous environment, and improve the automation degree and adaptability of protocol access.

[0007] To achieve the above object, the present invention is implemented through the following technical solutions: A method for accessing and configuring multi-brand and multi-type UAV devices, comprising the following steps:

[0008] Perform three-dimensional protocol feature modeling on the target UAV protocol, and respectively extract the communication layer feature matrix, the data layer association map, and the control layer spatio-temporal behavior tensor;

[0009] Based on the multi-dimensional protocol difference degree calculation model, perform feature comparison and analysis on the target protocol and the historical protocol. The difference degree fuses the communication feature distance, the data graph structure difference, and the control mode difference;

[0010] According to the protocol feature clustering result and the difference compensation strategy, fuse the basic configuration template and the historical deviation compensation item to generate an access configuration parameter set for the target UAV protocol, and realize the generation of dynamic adaptation configuration.

[0011] Preferably, the three-dimensional protocol feature modeling of the target UAV protocol includes:

[0012] Construct the communication layer feature matrix where n represents the number of communication units; m represents the dimension of the preset communication features; The element m ij is generated by the encoding function φ(·), that is where represents the original value of the i-th communication unit in the j-th communication feature dimension, and φ is the feature mapping function from discrete to numerical;

[0013] Construct the data layer association map G p =(V p , E p , W p ), where V p is the vertex set, representing the set of data fields in the protocol p; E p is the edge set, representing the edge pairs where there are semantic or temporal associations between the fields; W p is the edge weight function, defined as where e ij =(v i , v j ) is the directed edge from the slave node v i to v j , d ij represents the structural distance between the fields v i and v j , and σ is the smoothing parameter;

[0014] Construct the control layer spatio-temporal behavior tensor where T represents the number of time steps of the protocol control sequence, S represents the number of state dimensions, C represents the number of control command types, and the tensor Used to capture the change behavior of the protocol control process in three dimensions of time, state, and instructions, as the basis for subsequent modeling and difference analysis.

[0015] Preferably, the feature comparison and analysis of the target protocol and the historical protocol based on the multi-dimensional protocol difference degree calculation model includes:

[0016] For the communication feature matrix of the target protocol p and any historical protocol q and calculate the Euclidean distance to obtain the communication feature difference term where ||·|| F represents the Frobenius norm;

[0017] Calculate the graph edit distance for the data layer association graph G p =(V p , E p , W p ) and G q =(V q , E q , W q ), denoted as GED(G p , G q ), which is used to measure the difference in data field structure and connection relationship;

[0018] Perform Tucker decomposition on the control layer behavior tensors and respectively to obtain the core tensors and Then calculate the control behavior difference where ||·|| * represents the nuclear norm of the core tensor;

[0019] Combine the above three differences to obtain the multi-dimensional protocol difference degree D pq , and the calculation formula is:

[0020]

[0021] where α and β are non-negative weighting coefficients.

[0022] Preferably, the protocol difference degree calculation also includes the introduction of the time constraint matrix C t =[c ij , where c ij represents the co-occurrence frequency of instruction i to instruction j in multiple protocol samples, and the calculation method is:

[0023]

[0024] where, Indicates whether there is a valid execution path from instruction i to instruction j in the k-th protocol instance; Is an indicator function, with a value of 1 indicating existence and 0 indicating non-existence; Z is a normalization constant, K represents the total number of protocol instances for statistics, and C represents the number of types of control instructions.

[0025] Preferably, the method of fusing the basic configuration template and the historical deviation compensation item according to the protocol feature clustering result and the difference compensation strategy to generate the access configuration parameter set for the target UAV protocol includes:

[0026] Based on the protocol feature vectors of multiple historical protocols, use the spectral clustering algorithm to cluster the historical protocols and the target protocol to obtain the set of cluster centers μ1, μ2,..., μ K , where K is the number of clusters, and μ k Represents the center vector of the k-th cluster;

[0027] Calculate the similarity score s of the protocol feature representation vector f of the target protocol p p to the center of each cluster k = sim(f p , μ k ), and select the most relevant cluster k * = arg max k s k , to obtain the basic configuration template C most similar to the target protocol base = C k * ;

[0028] On the basis of the basic configuration template, introduce the deviation compensation items ΔC of several historical protocols j , to form the final access configuration parameter set C of the target protocol target , and its calculation formula is:

[0029]

[0030] Among them, Represents the set of historical protocols whose similarity to the target protocol p is higher than the threshold, and γ j Is the compensation weight coefficient, satisfying

[0031] Preferably, the calculation method of the deviation compensation item ΔC of the historical protocol j includes:

[0032] By minimizing the difference between the target protocol p and the historical protocol j in the communication feature dimension, the optimal compensation vector ΔC j is obtained, and its calculation formula is:

[0033]

[0034] Among them, is the communication feature matrix of the target protocol, is the communication feature matrix of the historical protocol j, Δ is the compensation matrix to be solved, and |·|F represents the Frobenius norm;

[0035] Take the optimal ΔC j as the compensation item of protocol j and participate in the weighted fusion generation of the target protocol access configuration.

[0036] Preferably, the method further includes the step of performing closed-loop optimization on the access configuration, specifically including:

[0037] Enable the forward absorption channel to accept the new feature vector of the target protocol and update the communication layer feature matrix. The update method is:

[0038]

[0039] Among them, represents the communication feature matrix at time t, and ΔM t is the newly observed protocol feature change amount in the current period, represents matrix-level splicing or augmentation operation to generate the updated communication feature matrix

[0040] Enable the feedback correction channel to dynamically adjust the edge weights of the data layer graph. According to the current protocol execution error e t update the graph weight The calculation formula is:

[0041]

[0042] Among them, represents the edge weight of edge e ij in the data layer graph at time t, e t is the protocol execution error measurement index, λ is the error sensitivity adjustment coefficient, and Φ(e t ) is the feedback correction function, and the domain is e t ≥0;

[0043] The closed-loop optimization process is automatically triggered at a fixed time interval Δt or when the protocol error e t exceeds the preset threshold, and continues to iterate until the configuration is stable or the error converges.

[0044] The present invention also provides a protocol adaptation device for UAV communication access, including:

[0045] A feature modeling module for three-dimensional protocol feature modeling of the communication protocol of a target unmanned aerial vehicle (UAV), constructing a communication layer feature matrix, a data layer association graph, and a control layer behavior tensor;

[0046] A difference analysis module for calculating the difference degree index between the target protocol and the historical protocol based on a multi-dimensional protocol difference degree calculation model and combining historical protocol data;

[0047] A configuration generation module for generating an access configuration parameter set for the target protocol by fusing a basic configuration template and historical compensation items according to the protocol feature clustering result and the difference compensation strategy;

[0048] A closed-loop optimization module for dynamically optimizing the access configuration, including: augmenting the communication feature matrix through a forward absorption channel and updating the edge weights of the data layer graph based on the protocol execution error through a feedback correction channel;

[0049] A control unit for coordinating the work processes of each module and triggering the configuration update and optimization iteration process according to the difference degree and the error feedback result.

[0050] The present invention also provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method as described above is implemented.

[0051] The present invention also provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method as described above is implemented.

[0052] The present invention provides a method for accessing and configuring multi-brand and multi-type UAV devices. It has the following

[0053] Beneficial effects:

[0054] 1. By constructing a three-dimensional protocol modeling method including a communication layer feature matrix, a data layer graph structure, and a control layer spatio-temporal behavior tensor, the present invention can comprehensively depict the feature expressions of UAV protocols in multiple dimensions such as structure, semantics, and behavior, thereby realizing fine-grained identification and modeling support for protocol differences and laying a complete data foundation for subsequent configuration generation and difference calculation.

[0055] 2. By introducing a multi-dimensional protocol difference degree calculation model composed of communication feature distance, data graph structure difference, and control behavior tensor difference, the present invention significantly enhances the system's expression ability for complex differences between protocols. Among them, the time sequence response ability of the control layer difference is further improved by introducing a time constraint matrix, which can effectively quantify the synchronous relationship between instruction behaviors.

[0056] 3. The present invention performs clustering modeling on historical protocol features through a spectral clustering algorithm, and selects the basic configuration template most relevant to the target protocol by using a similarity matching mechanism between the clustering cluster center vector and the target protocol feature vector, so that the generated configuration has structural consistency and semantic generality, which is beneficial to improving the matching degree and coverage breadth of the initial access configuration.

[0057] 4. The present invention can perform personalized correction on the basic configuration template by introducing a historical protocol compensation term calculated based on the minimum deviation criterion of communication features and combining a similarity weighting mechanism for fusion, so as to achieve local adaptation to the target protocol features while maintaining the generality of the configuration, and significantly improve the detail accuracy and compatibility of the configuration.

[0058] 5. The present invention realizes the dynamic update ability of the access configuration by establishing a closed-loop optimization mechanism including a forward absorption channel and a feedback correction channel. Among them, the communication feature matrix is structurally extended by splicing, and the edge weight of the data graph is adaptively adjusted through an error feedback function, which can continuously optimize the access configuration during the system operation, and enhance the long-term adaptability and stability of the configuration. Brief Description of the Drawings

[0059] Figure 1 It is a schematic flowchart of the method of the present invention;

[0060] Figure 2 It is a schematic structural diagram of the device of the present invention;

[0061] Figure 3 It is a schematic structural diagram of the computer device of the present invention.

[0062] Among them, 10. Feature modeling module; 20. Difference analysis module; 30. Configuration generation module; 40. Closed-loop optimization module; 50. Control unit. Detailed Embodiment

[0063] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0064] The present invention provides a multi-brand and multi-type unmanned aerial vehicle device access configuration method, which is used to realize dynamic and adaptive access configuration generation and optimization according to the target unmanned aerial vehicle protocol features in a complex heterogeneous communication environment.

[0065] Please refer to the attached Figure 1 , and the multi-brand and multi-type unmanned aerial vehicle device access configuration method may include the following steps:

[0066] S1. Model the three-dimensional protocol features of the target UAV protocol;

[0067] S2. Perform feature comparison analysis on the target protocol and historical protocols based on the multi-dimensional protocol difference calculation model;

[0068] S3. Generate an access configuration parameter set based on the protocol feature clustering results and difference compensation strategy;

[0069] S4. Perform closed-loop optimization on access configuration.

[0070] The following is a detailed description of each step in the method of the present invention, which comprehensively explains the specific implementation principles, technical details and processes of each step.

[0071] In step S1, the communication protocol used by the target UAV is modeled using three-dimensional protocol features to comprehensively model the semantics, control, and data behavior of the protocol at different hierarchical levels. This modeling approach helps support subsequent difference calculation and access parameter generation, and provides a unified protocol representation space for the system.

[0072] In actual implementation, the three-dimensional modeling process includes three parts: communication layer feature extraction, data layer semantic mapping, and control layer behavior modeling. It models the communication format, data association, and control logic respectively, and can more comprehensively represent the protocol behavior characteristics.

[0073] In the communication layer modeling, it is preferred to use vectorized encoding to perform numerical mapping on the message fields in the communication protocol and construct a communication layer feature matrix. Assume that a target protocol contains n communication units, each of which contains m field features of preset dimensions, which are recorded as the original feature matrix:

[0074]

[0075] in, Represents the original discrete value of the jth feature field in the i-th communication unit. To facilitate calculation, the encoding function φ(·) is used to map the discrete value into a numerical vector to form the communication feature matrix:

[0076]

[0077] in, is a feature mapping function from discrete to continuous space. This encoding function can be implemented based on table lookup mapping, word vector mapping, or hash projection to ensure that different protocol fields are consistent on the same semantic dimension.

[0078] Data layer modeling is mainly used to describe the semantic structure and call sequence between data fields in the communication protocol. For this purpose, a directed graph structure G is constructed.p =(V p , E p , W p ), where:

[0079] V p is a set of data fields, and the nodes represent the field entities that appear in the protocol;

[0080] represents that there are reference, transfer, or dependency relationships between fields;

[0081] represents the edge weight function, which is used to quantify the association strength between fields.

[0082] Preferably, the edge weight function is defined as:

[0083]

[0084] where e ij =(v i , v j ) represents the directed edge from field v i to v j , d ij represents the structural distance between fields, and σ is the smoothing control parameter. The structural distance can be comprehensively calculated according to indicators such as the relative offset of fields in the message, the difference in call frequencies, or the semantic label distance. This graph structure can reveal the transfer relationship of the protocol data part and is conducive to the analysis of structural similarity between protocols.

[0085] The control layer modeling aims to perform spatio-temporal modeling on the behavior of the command sequence controlled by the protocol. For this purpose, preferably, a third-order tensor is constructed to record the behavior characteristics of control instructions at different times and in different states.

[0086] where:

[0087] T represents the time step in the protocol control process;

[0088] S represents the number of state variables, such as flight mode, power state, location information, etc.;

[0089] C represents the types of control instructions, such as takeoff, hover, switch channel, etc.

[0090] Each item in the tensor represents the execution state or intensity of the c-th type of control instruction at time step t and in state dimension s. This tensor can be constructed through protocol instruction logs, flight control instruction streams, or embedded call sequences, and can be normalized if necessary to make it have a unified structure when used for Tucker decomposition and behavior comparison in the future.

[0091] The control layer modeling can be further extended to spatio-temporal graph modeling, that is, introducing time tags and state tags to slice and represent tensors, and realizing the modeling of synchronicity and sequentiality between instructions on the premise of unchanged dimensions.

[0092] It should be noted that the three-dimensional modeling results of the communication layer, data layer and control layer will jointly serve as the basis for the protocol feature expression space and be called by subsequent difference analysis steps. Therefore, the three types of modeling need to be unified in terms of semantic consistency, structural completeness and dimension normativity to avoid problems such as vector misalignment and graph incompatibility in subsequent steps.

[0093] Through this three-dimensional protocol feature modeling method, this embodiment can realize the unified representation of complex UAV protocol behaviors at the semantic, structural and dynamic control levels, and constitute the underlying feature basis for protocol difference calculation and access configuration generation.

[0094] In step S2, after completing the three-dimensional protocol feature modeling of the target UAV protocol, it is further necessary to calculate and analyze the difference relationship between the target protocol and multiple historical protocols to support the matching and generation of access configuration parameters. To achieve this function, the present invention conducts feature comparison and analysis based on a multi-dimensional protocol difference calculation model, which is used to quantitatively measure the multi-angle structural differences between the target protocol and historical protocols in three dimensions of the communication layer, data layer and control layer.

[0095] In the specific implementation process, this embodiment preferably adopts a multi-dimensional fusion strategy to jointly model the protocol feature difference results of each layer to form a unified difference metric index.

[0096] In the communication layer difference calculation, the main consideration is how to accurately measure the communication field feature differences between the target protocol and historical protocols. The communication layer is the bottom layer of the protocol, responsible for data transmission and signal encoding, and is a basic module in the protocol architecture. Therefore, accurately comparing the communication layer features can help understand the differences in data transmission and signal structure of the protocol, so as to provide precise configuration guidance for protocol access and collaboration.

[0097] Communication protocols usually consist of multiple data units, and each data unit contains multiple fields. When calculating the communication layer difference, it is first necessary to extract the communication features of each protocol and compare them through appropriate mathematical measurement methods.

[0098] The communication layer features of communication protocols are mainly reflected in the data fields of each protocol. Each data field contains information such as data bits, data types, field lengths, and field values. Therefore, in order to accurately compare communication protocols, it is necessary to first numericalize the features of each data unit in the protocol. Usually, each field in the protocol can be represented as a discrete value, and in order to calculate and compare, these discrete values must be converted into a continuous numerical form.

[0099] In this embodiment, the discrete field values are converted into numerical vectors through a vectorized encoding method to construct a communication feature matrix. Let the communication feature matrix of the target protocol be:

[0100]

[0101] where n represents the number of data units in the protocol, m represents the number of field features included in each data unit, represents the j-th field feature of the i-th data unit in the target protocol. Correspondingly, the communication feature matrix of the historical protocol is:

[0102]

[0103] The discrete field values in the protocol can be mapped to numerical vectors through the function φ(·), so that each communication field has a numerical representation:

[0104]

[0105] where φ is a mapping function from discrete values to numbers, and techniques such as look-up tables, hash mapping, word vectors, or other mapping techniques can be used.

[0106] After obtaining the communication feature matrices of the target protocol and the historical protocol, the next step is to calculate the communication layer difference. The calculation of the communication layer difference is based on matrix distance metrics. Common distance metric methods include Euclidean distance, Manhattan distance, and Frobenius norm, etc.

[0107] In this embodiment, the Frobenius norm is preferably used to measure the difference between the two communication feature matrices. The Frobenius norm is a measure of the overall difference of a matrix, which can reflect the differences between all elements of the matrix and is suitable for calculating the overall deviation between protocol feature matrices. Let the communication feature matrix of the target protocol be The communication feature matrix of the historical protocol is Then the difference metric between them is:

[0108]

[0109] The definition of the Frobenius norm is:

[0110]

[0111] where is the difference matrix between the communication feature matrices of the target protocol and the historical protocol, and ||A|| Fis its Frobenius norm, representing the Euclidean distance between two matrices. This distance quantifies the overall difference of the protocol at the communication layer and can be used for subsequent protocol comparison and selection.

[0112] The advantage of the Frobenius norm is that it can comprehensively consider the differences between all elements in the matrix, rather than just the differences of individual fields. Therefore, it can effectively reflect the similarities and differences of the entire protocol communication layer. If the difference of the communication layer between two protocols is small, the value of the Frobenius norm will be small; otherwise, it will be large.

[0113] To make the calculation of the communication layer difference more accurate, weighted processing can also be considered for specific fields in the communication layer. The influence degrees of communication fields in different protocols on protocol performance, reliability, and efficiency may vary. Therefore, when calculating the communication layer difference, different weights may need to be assigned to different fields according to the importance of the fields. Let the field weight be w j , then the communication layer difference degree can be rewritten as:

[0114]

[0115] By assigning different weights w j to different fields, the influence of some key fields on the protocol difference can be better reflected, thus improving the accuracy and pertinence of the calculation.

[0116] In addition, in some specific application scenarios, the structural association between protocol fields may need to be considered. For example, there are dependency relationships between some fields, and the values of some fields may affect the values of other fields. In this case, a more complex graph structure model can be introduced, and the graph edit distance (GED) can be used to measure the structural differences between protocols, thereby improving the accuracy of the communication layer difference calculation.

[0117] In the calculation of the data layer difference, the purpose is to quantify the difference in the association structure between data fields of the target protocol and the historical protocol. The main characteristics of the data layer are reflected in the semantic relationships between protocol fields and their dependency structures in the protocol. The data layer difference calculation not only considers the numerical differences between individual fields but also pays more attention to the structural differences and dependency changes between fields.

[0118] The core idea of the data layer difference calculation is to construct a data field association graph in the protocol through graph theory methods and use the graph edit distance (GED) to measure the difference between the target protocol and the historical protocol. Through this method, the interdependency relationships between fields and the differences in data flow patterns during data transmission between protocols can be measured more precisely.

[0119] In the process of data layer modeling, it is first necessary to convert the data fields in the protocol and the relationships between them into a graph structure. The data fields in the protocol can be regarded as nodes in the graph, while the dependencies between fields (such as data transmission order, reference, passing, etc.) are regarded as edges in the graph. Each edge can also be attached with a weight, indicating the association strength between two fields. For this purpose, first construct the data graph G p and G q , which respectively represent the data fields and their dependencies in the target protocol and the historical protocol.

[0120] Let the data graph of the target protocol be G p =(V p , E p , W p ), where:

[0121] V p is the set of data fields, representing the data fields in the protocol;

[0122] is the dependency relationship (edge set) between data fields;

[0123] is the weight function of the edge, used to represent the association strength between fields.

[0124] Similarly, the data graph of the historical protocol is G q =(V q , E q , W q ).

[0125] To construct the data graph, the following aspects need to be considered:

[0126] 1. Construction of data field nodes: The fields of each protocol can be represented as a node in the graph. Characteristics such as the type, length, and data structure of the field can be described as attributes of the node.

[0127] 2. Relationships between fields: There are usually dependencies between fields in the protocol. For example, the value of some fields may affect the value of other fields, or data is transmitted between fields in a certain order. These relationships can be represented by the edges of the graph.

[0128] 3. Weights of edges: The weights of edges represent the strength of the dependency relationships between fields. The weights of edges can be designed based on factors such as the dependency frequency of fields, the reference pattern of fields, and the transmission direction of fields.

[0129] After constructing the data graph, the next step is to calculate the data layer differences between the target protocol and the historical protocols. To measure the structural differences between data fields, a commonly used metric is the graph edit distance (GED). The graph edit distance is a metric for measuring the similarity between two graph structures, which calculates the minimum editing cost required to transform one graph into another.

[0130] In this embodiment, the data layer difference metric uses the graph edit distance to quantify the difference between the data graphs G p and G q . The graph edit distance includes the following basic graph editing operations:

[0131] Node insertion / deletion: Adding or deleting data field nodes to / from the graph.

[0132] Node replacement: Replacing data field nodes in the graph.

[0133] Edge insertion / deletion: Inserting or deleting the dependency edges between fields in the graph.

[0134] Edge weight modification: Modifying the association strength between fields.

[0135] The data graph of the target protocol is G p =(V p , E p , W p ), and the data graph of the historical protocol is G p =(V q , E q , W q ). The calculation of the graph edit distance can be expressed as:

[0136] d g =GED(G p , G q )

[0137] where GED(G p , G q ) represents the minimum editing cost required to transform the data graph G p into the data graph G q . This cost is usually obtained by weighted calculation of the editing operations on nodes and edges.

[0138] Specifically, the graph edit distance can be calculated as follows:

[0139] 1. Node editing cost: For the editing operations (insertion, deletion, replacement) between nodes, define the node editing cost function cost node (v i , v j ), where v i ∈Vp and v j ∈V q are nodes in the graph. The node replacement cost is usually defined according to the attribute differences of the nodes (such as field type, field length, etc.).

[0140] 2. Edge editing cost: For the editing operations (insertion, deletion, replacement) between edges, define the edge editing cost function cost edge (e i , e j ), where e i ∈E p and e j ∈E q are edges in the graph. The edge editing cost is usually defined according to the weight differences of the edges or the presence / absence of the edges.

[0141] 3. Total editing cost: The total cost of the graph edit distance is the sum of the costs of all node and edge editing operations:

[0142]

[0143] where k and l represent the number of node and edge editing operations respectively.

[0144] By calculating the graph edit distance GED(G p , G q ), the structural difference degree between the target protocol and the historical protocol at the data layer can be obtained. This difference metric reflects the differences in the data field relevance, dependence, and transmission path of the protocol. A smaller graph edit distance indicates that the data layer structures of the two protocols are more similar, while a larger graph edit distance indicates a greater structural difference between the protocols.

[0145] In practical applications, the field relationships at the data layer usually change dynamically. Therefore, it is also possible to consider dynamically modeling the data graph and performing incremental calculation of the graph edit distance according to the protocol version evolution, so as to further improve the efficiency and accuracy of the difference calculation.

[0146] In the control layer difference calculation, the purpose is to evaluate the differences in control behaviors between the target protocol and the historical protocol, especially the execution order of protocol control instructions, the timing response characteristics, and the instruction execution relationships under different control modes. The difference calculation in the control layer mainly focuses on the behavioral characteristics of the protocol in operation control, involving the transmission of control signals, the triggering of instructions, and the change of states.

[0147] The difference calculation in the control layer is not just a simple comparison of control instructions in the protocol. Instead, through the tensor modeling and analysis of protocol control behaviors, it quantifies the similarities and differences among different protocols in terms of timing, state changes, and instruction execution. For this purpose, this embodiment adopts a three-dimensional tensor representation method, combines spatio-temporal feature modeling with Tucker decomposition technology, and calculates the difference between the target protocol and the historical protocol through tensor nuclear norm measurement.

[0148] In the control layer modeling, first, a three-dimensional control behavior tensor needs to be constructed to represent the spatio-temporal dynamic behavior of control instructions in the protocol. The control behavior tensor can be constructed based on the timing log of the protocol, the instruction execution order, and the system state information. Let the control behavior tensors of the target protocol and the historical protocol be and respectively, whose dimensions are the time step T, the state space dimension S, and the number of control instruction types C, representing the execution intensity of control instructions at different time steps and under different state conditions.

[0149]

[0150] Among them:

[0151] T represents the time step, usually determined by the control period of the protocol or the time interval between instructions;

[0152] S represents the state space of the protocol, usually including information such as the flight mode of the system, the battery state, and the flight trajectory;

[0153] C represents the types of control instructions, such as takeoff, landing, hovering, channel switching, etc.

[0154] During the construction process of the control layer tensor, each item and represents the intensity or execution state of the c-th type of control instruction in the target protocol and the historical protocol at time step t and state s. Usually, the intensity of the control instruction can be represented as a real number in the range of [0, 1], where 0 means the instruction is not executed, and 1 means the instruction is fully executed.

[0155] For the convenience of comparison, the tensor nuclear norm (Tucker Decomposition) method is used to perform low-rank approximation on the control behavior tensor. This method can effectively extract the dominant features of control instructions in different state and time dimensions, thereby reducing the computational complexity and reducing redundant information.

[0156] Tucker decomposition is a commonly used tensor dimensionality reduction technique that can decompose the original tensor into the product of a core tensor and several factor matrices. In this embodiment, the control behavior tensors of the target protocol and the historical protocol are Tucker decomposed to obtain a low-rank core tensor representation. Specifically, the control behavior tensors of the target protocol and the historical protocol can be respectively represented as:

[0157]

[0158] Among them, and are the core tensors of the target protocol and the historical protocol, representing the low-rank representation after dimensionality reduction; U (1) , U (2) , U (3) and V (1) , V (2) , V (3) are the corresponding factor matrices, representing the principal components of time, state, and control instructions respectively.

[0159] The core idea of Tucker decomposition is to convert the high-dimensional structure of the tensor into a low-dimensional structure, so as to retain the main information of the control behavior with lower computational complexity. Through Tucker decomposition, the spatio-temporal features of the control behavior can be efficiently extracted, and the behavior patterns of the protocol under different states and time conditions can be captured.

[0160] After obtaining the low-rank core tensors of the target protocol and the historical protocol, the difference degree between them can be calculated. The calculation of the difference degree is based on the tensor nuclear norm, which is defined as:

[0161]

[0162] Among them, ||1|| * represents the tensor nuclear norm, which is the sum of the singular values of the tensor and can effectively measure the structural difference between two tensors. The smaller the tensor nuclear norm, the more similar the behavior characteristics of the target protocol and the historical protocol at the control layer; on the contrary, the larger the tensor nuclear norm, the greater the difference in the control behavior between the two.

[0163] After the above three types of difference indicators are modeled, a weighted fusion method will be adopted to generate a unified protocol difference degree indicator:

[0164] D pq = d c + α·d g + β·d t

[0165] Among them, α and β are the fusion weight parameters of the difference dimension, which can be set according to the training set protocol analysis experience or based on the importance of the protocol layer, and are used to balance the influence of the three types of difference items on the overall difference degree.

[0166] Furthermore, to improve the timing response ability of the control layer behavior comparison, a time constraint matrix C is preferably introduced in this embodiment t =[c ij , to capture the synchronicity and timing dependence characteristics between control instructions, and enhance the time sensitivity of the control layer behavior in the difference calculation.

[0167] In the protocol control process, there are usually explicit or implicit timing relationships between control instructions. For example, some instructions often trigger in a specific order or have a synchronous execution mode. To more realistically restore the behavior logic of the control layer, this embodiment constructs a time constraint matrix by statistically analyzing the co-occurrence relationship between instructions in multiple protocol instances to reflect the typical timing paths in the control process.

[0168] Time constraint matrix is a symmetric matrix on the control instruction dimension, where:

[0169] C is the total number of types of control instructions;

[0170] c ij represents the strength of the synchronicity relationship between instruction i and instruction j in multiple protocol instances, that is, in what proportion of protocol instances, there is an effective execution path from i to j.

[0171] The matrix element c ij is calculated as follows:

[0172]

[0173] Among them:

[0174] represents whether there is an effective execution path from instruction i to instruction j in the kth protocol instance;

[0175] is the Indicator Function, defined as:

[0176]

[0177] K is the total number of protocol samples used for statistics;

[0178] Z is the normalization constant, usually taken as K, that is, the number of all protocol instances, which is used to convert the frequency into the synchronicity probability or proportion value;

[0179] Therefore, c ij∈[0,1], the larger its value, the stronger the simultaneity of instruction i to j, that is, the higher the frequency of their co-occurrence, co-execution or logical sequence in the protocol.

[0180] Effective execution path The determination method can be based on protocol logs, state transition diagrams or control flow diagrams. In practical applications, the following method is usually used for extraction:

[0181] Construct a control behavior graph G for protocol instance k k , where the nodes in the graph are instructions and the edges represent the execution relationships between instructions;

[0182] If there is a reachable path from instruction i to instruction j, it is regarded as

[0183] Depth-first search (DFS) or breadth-first search (BFS) can be used to determine whether a path exists in the graph.

[0184] Time constraint matrix C t After construction, the time response ability of the control layer difference calculation can be enhanced in the following two aspects:

[0185] 1. Enhance the weight modulation of the behavior tensor: When constructing the control behavior tensor , perform weight modulation on the control instruction dimension, that is, introduce a time constraint term in the behavior tensor:

[0186]

[0187] Make those instruction pairs with stronger temporal dependence contribute more to the tensor structure and improve the model's sensitivity to temporal patterns.

[0188] 2. As an adjustment factor for the difference distance: The control layer difference distance d t can be adjusted by introducing a time weighting term, for example:

[0189]

[0190] where represents the (i,j)-th element in the control kernel tensor, representing the interaction strength between instruction pairs, and the weight c ij represents its importance.

[0191] Through the above difference analysis process, the structural and behavioral differences between the target protocol and the historical protocol can be quantified in multiple dimensions, providing interpretable basic index support for subsequent configuration parameter generation and fusion strategies.

[0192] In step S3, based on the completed multi-dimensional protocol feature modeling and difference degree calculation, a protocol feature clustering and deviation compensation mechanism is further introduced to integrate historical information and template configuration, and finally generate an access configuration parameter set adapted to the target protocol.

[0193] This process includes: protocol feature clustering, basic configuration template selection, historical deviation compensation item construction, and target configuration fusion generation. The overall solution is based on the clustering structure of the protocol feature vector space, fully considering the structural similarity between the target protocol and existing protocols, and fine-tuning the protocol differences through local compensation to achieve the purpose of automatic adaptation configuration.

[0194] First, on the premise of sufficient accumulation of historical protocols, a protocol feature vector is constructed for each known protocol. This vector is obtained through the aforementioned three-dimensional modeling method, integrating the communication feature matrix, data graph structure information, and control layer behavior tensor, and forming a fixed-dimensional vector representation after dimensionality reduction or feature compression.

[0195] Based on these protocol feature representations, in this embodiment, the spectral clustering algorithm is preferably used to perform clustering analysis on the historical protocol set and the target protocol together. Let the number of clusters be K, and the obtained set of cluster center vectors is represented as:

[0196] μ1, μ2,..., μ K

[0197] Among them, represents the center of the k-th cluster, and d is the dimension of the feature representation.

[0198] For the target protocol p to be accessed, based on the completed feature modeling and feature clustering, it is necessary to determine the most relevant cluster in the clustering space for this protocol, and based on this, select the corresponding basic configuration template as the structural benchmark for generating the access configuration parameters of the target protocol.

[0199] To achieve the above purpose, first, a corresponding protocol feature representation vector needs to be constructed for the target protocol p. The acquisition of this vector depends on the aforementioned protocol modeling results, comprehensively considering the feature matrix representation of the communication layer, the structure graph of the data layer, and the behavior tensor of the control layer, and obtaining a fixed-length protocol feature representation after feature compression or transformation (such as PCA, pooling, etc.).

[0200] This vector semantically represents the structural position of the target protocol in the multi-dimensional communication behavior space and can be used for subsequent similarity matching with the cluster center. At this time, the system has obtained multiple cluster centers by executing the clustering algorithm on multiple historical protocols.

[0201] To determine the degree of association between the target protocol p and each cluster center, the similarity score s between it and each cluster center is further calculated k . The calculation method of this similarity score is preferably based on the kernel similarity calculation method in the high-dimensional vector space, such as cosine similarity or the weighted kernel function based on the inner product. One expression of cosine similarity is:

[0202]

[0203] where · represents the vector dot product, and ||·|| represents the two-norm of the vector. By calculating the similarity values between the target protocol and all cluster centers, a set of score sequences can be obtained:

[0204] {s1, s2, …, s K}

[0205] Subsequently, select the cluster with the highest similarity score as the most relevant cluster to which the target protocol belongs, and its cluster number is denoted as:

[0206] k * = arg max k s k

[0207] Thus, the system can determine the category identifier closest to the target protocol p in the clustering space. This classification result can be regarded as the semantic attribution of the target protocol in the structural feature space.

[0208] After obtaining the cluster number k * , select the corresponding basic configuration template from the set of in-cluster representative configuration templates predefined or statistically learned in the system:

[0209] C base = C k *

[0210] This basic configuration template C base represents the typical access configuration obtained after sample induction and clustering feature abstraction in the cluster k * , and can represent the general configuration structure of devices of this type of protocol. The source of this template can be:

[0211] The weighted average of the configuration templates of multiple protocols within the cluster;

[0212] The most representative protocol configuration template in frequency statistics or structural analysis;

[0213] The standard configuration mode determined by expert rules or automatic annotation.

[0214] The basic configuration template includes, but is not limited to, general access parameters such as communication parameter items, port configurations, transmission rate control policies, handshake response modes, etc. Its structure is normalized according to the device model, protocol family attributes, and the stability of the usage scenario.

[0215] It should be noted that in this embodiment, the selection of the basic configuration template is not based on a single protocol similarity judgment, but by leveraging the clustering expression ability of protocol features in a high-dimensional space, and enhancing the overall adaptation ability to the target protocol behavior features through the semantic aggregation ability of the clustering center.

[0216] In addition, the template selection process does not depend on the access results or online responses of any target protocol, but is completely based on the protocol knowledge space constructed offline for judgment. Therefore, it has the ability of fast response and independent system operation, which is convenient for large-scale deployment and parallel operation of multiple devices.

[0217] On the basis of obtaining the basic configuration template, in order to improve the pertinence and fineness of the configuration, this embodiment introduces a weighted fusion mechanism for historical deviation compensation items. Specifically, based on the similarity relationship between the target protocol and historical protocols, a number of similar protocol sets N(p) are determined, which is defined as:

[0218] N(p) = {j ∈ historical protocol set | sim(f p , f j ) > τ}

[0219] where τ is the similarity threshold, and f is the feature vector of historical protocol j.

[0220] For each j ∈ N(p), calculate its compensation item ΔC j relative to the target protocol p, which is used to characterize the deviation between it and the target protocol in the communication dimension. The solution goal of this item is to minimize the communication feature matching, and the formula is as follows:

[0221]

[0222] where:

[0223] represents the communication feature matrix of the target protocol;

[0224] is the communication feature matrix of historical protocol j;

[0225] is the deviation compensation matrix to be optimized;

[0226] ||·|| F represents the Frobenius norm, which is used to measure the overall difference degree between matrices.

[0227] By solving the above minimization problem, the compensation term ΔC is obtained j , which is used to reflect the differences between the historical protocol and the target protocol in the communication dimension.

[0228] Finally, by combining all the historical protocol compensation terms that meet the conditions, a complete access configuration parameter set C for the target protocol is formed by using a weighted fusion method target . The fusion formula is as follows:

[0229]

[0230] where:

[0231] γ j ∈[0,1] is the compensation weight coefficient for each historical protocol;

[0232] The weight coefficients satisfy the normalization constraint:

[0233]

[0234] The weight coefficients can be set according to similarity scores, the representativeness or stability indicators of historical protocols. By performing a weighted combination of the deviation compensation terms of multiple historical protocols, the basic template can be calibrated more accurately, improving the pertinence of the configuration.

[0235] In summary, in this embodiment, step S3 generates a personalized access configuration parameter set for the target protocol by fusing the clustering modeling and compensation mechanism of protocol features, making full use of the clustering structure and local difference information of historical protocols. This parameter set not only includes the global basic configuration template but also incorporates the local corrections of historical instances to the target protocol. The overall solution has strong adaptability and promotion ability.

[0236] The above processes are all realized without restricting specific hardware platforms, applicable to the access scenarios of multi-brand and multi-type UAV devices, and have good system compatibility and scalability.

[0237] In step S4, to achieve the continuous effectiveness and dynamic adaptability of the access configuration, a closed-loop optimization mechanism is introduced to perform online updates and structural corrections on the generated access configuration for the target protocol, thereby enhancing the response ability of the configuration system to protocol behavior changes and continuously iterating towards a stable optimal configuration state during operation.

[0238] The closed-loop optimization mechanism mainly includes two complementary channels: the forward absorption channel and the feedback correction channel, corresponding to the structural expansion update of protocol communication characteristics and the weight adaptive adjustment of the data graph structure respectively. This optimization mechanism continues to run after the configuration is deployed and can perform fine-grained adjustments to the configuration strategy during the system life cycle.

[0239] Specifically, in the forward absorption channel, to adapt to the newly added or dynamically changing characteristic behaviors gradually exhibited by the target protocol during operation, the system constructs a characteristic change matrix ΔM based on the new characteristics observed in the current cycle of the protocol t , and fuses it with the existing communication characteristic matrix . The fusion is achieved by matrix splicing or augmentation, and the update formula is as follows:

[0240]

[0241] Where:

[0242] represents the communication layer characteristic matrix at time t, and the row-column structure of the matrix corresponds to the channels, parameter items, or behavior patterns in the protocol;

[0243] represents the protocol behavior characteristic change information newly received in the current time cycle;

[0244] represents the splicing operation at the matrix level, preferably the extended splicing operation in the row or column dimension to ensure the dimensional consistency of the feature space;

[0245] The updated communication characteristic matrix will be used as the basic input for the next stage of configuration generation and feature analysis.

[0246] This forward absorption process can be automatically triggered when it is detected that there are situations beyond the original feature coverage range in the protocol behavior, or can be activated through system event driving when new devices are connected and the protocol content changes.

[0247] On the other hand, to achieve dynamic adaptive adjustment of the data layer structure, this embodiment further introduces a feedback correction channel for dynamically updating the edge weights in the data layer association graph according to the protocol execution error during the actual operation process. The update of the graph weights depends on the protocol error metric e t of the current cycle, and the edge weight coefficients are compressed or amplified through a feedback function.

[0248] The edge weight of the data layer graph at each moment is denoted as where e ij represents the edge connection relationship between two data nodes in the graph. The edge weight update formula of the graph is as follows:

[0249]

[0250] where the feedback correction function Φ(e t ) is defined as:

[0251] Φ(e t)=1-tanh(λe t )

[0252] Where:

[0253] Represents the edge e of the graph at time t ij The weight value of

[0254] e t The execution error indicator of the protocol configuration in the current period can be calculated by the deviation between the protocol output and the expected output;

[0255] λ is the error adjustment coefficient, which is used to control the sensitivity of error changes to the weight correction amplitude;

[0256] tanh(·) is a hyperbolic tangent function with a value range of (0,1), which ensures that the feedback function Φ(e t ) is always positive and decreases as the error grows.

[0257] It can be seen that when the protocol execution error is small, Φ(e t ) value is close to 1, the spectrum structure remains basically stable; when the error increases, Φ(e t ) is significantly less than 1, thereby attenuating the corresponding edge weights and prompting the system to relearn the structural importance of edge relationships.

[0258] In order to ensure the efficiency and stability of the closed-loop optimization process, this embodiment further provides a triggering and termination mechanism for the optimization process. The system can trigger the optimization process based on the following conditions:

[0259] Automatic execution after reaching the preset fixed time interval Δt;

[0260] Or, when the current protocol execution error e is detected t The optimization process is initiated immediately after exceeding the set threshold.

[0261] After the optimization is executed, the system continues to iterate until any of the following termination conditions is met:

[0262] The generated communication feature matrix has no significant changes compared with the previous cycle;

[0263] The change in edge weights of the data layer graph is lower than the preset convergence threshold;

[0264] Alternatively, the protocol execution error e t reduced to within the permissible range.

[0265] In summary, step S4 in this embodiment realizes a closed-loop update mechanism for the access configuration of the target protocol by constructing a dual-channel of forward absorption and feedback correction. This mechanism can not only continuously absorb new feature information generated during the operation of the target protocol, but also automatically correct the data graph structure according to the actual execution deviation, ensuring that the access configuration has long-term adaptability and runtime dynamic adjustability.

[0266] The above update mechanism can be deployed in a centralized control system or distributed agent nodes. Preferred software frameworks include a graph computing engine, a protocol behavior analysis module, an edge weight learning unit, etc., which can collaboratively support the realization of the closed-loop optimization function in a multi-type device operating environment.

[0267] Generally speaking, the present invention performs three-dimensional feature modeling on the target protocol by constructing a communication layer feature matrix, a data layer association graph, and a control layer spatio-temporal behavior tensor. On this basis, using a multi-dimensional difference calculation model, it fuses the communication feature distance, the data graph structure difference, and the control behavior difference to accurately measure the multi-level structure difference between the target protocol and historical protocols. Subsequently, based on the historical protocol clustering results, combining protocol similarity analysis and deviation compensation strategies, it selects the most relevant basic configuration template and introduces communication feature compensation terms of multiple historical protocols to generate a personalized access configuration parameter set adapted to the target protocol. Finally, a closed-loop optimization mechanism is constructed to realize the online dynamic update and stability control of the access configuration by forward absorbing new observed features and feedback correcting the graph structure, thereby completing the whole process of efficient access and automatic configuration of heterogeneous UAV protocols.

[0268] The protocol adaptation device for UAV communication access described below can be correspondingly referred to the multi-brand and multi-type UAV device access configuration method described above.

[0269] Please refer to the attached Figure 2 , the present invention also provides a protocol adaptation device for UAV communication access, including:

[0270] A feature modeling module 10 for performing three-dimensional protocol feature modeling on the communication protocol of the target UAV, constructing a communication layer feature matrix, a data layer association graph, and a control layer behavior tensor;

[0271] A difference analysis module 20 for calculating the difference index between the target protocol and historical protocols based on a multi-dimensional protocol difference calculation model in combination with historical protocol data;

[0272] A configuration generation module 30 for generating an access configuration parameter set for the target protocol by fusing a basic configuration template and historical compensation terms according to the protocol feature clustering results and deviation compensation strategies;

[0273] The closed-loop optimization module 40 is used to dynamically optimize the access configuration, including: augmenting the communication feature matrix through the forward absorption channel and updating the edge weights of the data layer graph based on the protocol execution error through the feedback correction channel;

[0274] The control unit 50 is used to coordinate the work processes of each module and trigger the configuration update and optimization iteration process according to the difference degree and the error feedback result.

[0275] The device of this embodiment can be used to execute the above method embodiment, and its principle and technical effects are similar, which will not be elaborated here.

[0276] Please refer to the appendix Figure 3 The present invention also provides a computer device 40, including: a processor 41 and a memory 42. The memory 42 stores a computer program executable by the processor. When the computer program is executed by the processor, it executes the above method.

[0277] The present invention also provides a storage medium 43. A computer program is stored on the storage medium 43. When the computer program is run by the processor 41, it executes the above method.

[0278] Among them, the storage medium 43 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (abbreviated as SRAM), electrically erasable programmable read-only memory (abbreviated as EEPROM), erasable programmable read-only memory (abbreviated as EPROM), programmable read-only memory (abbreviated as PROM), read-only memory (abbreviated as ROM), magnetic memory, flash memory, magnetic disk or optical disc.

[0279] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for accessing and configuring a multi-brand and multi-type unmanned aerial vehicle device, characterized in that It includes the following steps: Perform three-dimensional protocol feature modeling on the target UAV protocol, and extract the communication layer feature matrix, data layer association graph, and control layer spatio-temporal behavior tensor respectively; Based on the multi-dimensional protocol difference degree calculation model, conduct feature comparison and analysis between the target protocol and historical protocols. The difference degree integrates communication feature distance, data graph structure difference, and control mode difference; According to the protocol feature clustering results and difference compensation strategy, fuse the basic configuration template and historical deviation compensation items to generate an access configuration parameter set for the target UAV protocol, realizing dynamic adaptation configuration generation.

2. The method for accessing and configuring a multi-brand and multi-type unmanned aerial vehicle device according to claim 1, wherein The performing three-dimensional protocol feature modeling on the target UAV protocol includes: Construct the communication layer feature matrix where n represents the number of communication units; m represents the dimension of the preset communication features; element m of ij is generated by the encoding function φ(·), that is where represents the original value of the i-th communication unit in the j-th communication feature dimension, and φ is a feature mapping function from discrete to numerical values; Construct the data layer association graph G p =(V p , E p , W p ), where V p is the vertex set, representing the set of data fields in protocol p; E p is the edge set, representing the edge pairs with semantic or temporal associations between fields; W p is the edge weight function, defined as where e ij =(v i , v j ) is the directed edge from node v i to v j , d ij represents the structural distance between fields v i and v j , and σ is the smoothing parameter; Construct the spatio-temporal behavior tensor of the control layer where T represents the number of time steps of the protocol control sequence, S represents the number of state dimensions, C represents the number of control command types, and the tensor is used to capture the changing behavior of the protocol control process in the three dimensions of time, state, and instruction, and serves as the basis for subsequent modeling and difference analysis.

3. The method for configuring access of multi-brand and multi-type UAV devices according to claim 1, wherein The conducting feature comparison and analysis between the target protocol and historical protocols based on the multi-dimensional protocol difference degree calculation model includes: The communication feature matrix of the target protocol p and any historical protocol q and perform Euclidean distance calculation to obtain the communication feature difference term where ||·|| F represents the Frobenius norm; For the data layer association graph G p =(V p , E p , W p ) and G q =(V q , E q , W q ), calculate the graph edit distance, denoted as GED(G p , G q ), which is used to measure the differences in the data field structure and connection relationships; For the control layer behavior tensor and perform Tucker decomposition to obtain the core tensors and Then calculate the control behavior difference where ||·|| * represents the nuclear norm of the core tensor; Combining the above three differences gives the multi-dimensional protocol difference degree D pq , and the calculation formula is as follows: Where α and β are non-negative weighting coefficients.

4. The multi-brand and multi-type UAV device access configuration method according to claim 3, wherein, The calculation of the protocol difference degree also includes the introduction of a time constraint matrix C t = [c ij , where c ij represents the co-occurrence frequency of instruction i to instruction j in multiple protocol samples, and the calculation method is as follows: Among them, indicates whether there is a valid execution path from instruction i to instruction j in the k-th protocol instance; is an indicator function, with a value of 1 indicating existence and 0 indicating non-existence; Z is a normalization constant, K represents the total number of protocol instances for statistics, and C represents the number of types of control instructions.

5. The method for accessing and configuring a multi-brand and multi-type drone device according to claim 1, wherein The fusing the basic configuration template and historical deviation compensation items according to the protocol feature clustering results and difference compensation strategy to generate an access configuration parameter set for the target UAV protocol includes: Based on the protocol feature vectors of multiple historical protocols, the spectral clustering algorithm is used to cluster the historical protocols and the target protocol to obtain a set of cluster centers μ1, μ2,..., μ K , where K is the number of clusters, and μ k represents the center vector of the k-th cluster; Calculate the protocol feature representation vector f of the target protocol p p Similarity scores s to each cluster center k = sim(f p , μ k ), and select the most relevant cluster k * = arg max k s k , to obtain the basic configuration template most similar to the target protocol Based on the basic configuration template, introduce the deviation compensation terms ΔC of several historical protocols j to form the final access configuration parameter set C of the target protocol target and its calculation formula is: Among them, represents the set of historical protocols whose similarity to the target protocol p is higher than the threshold, and γ j is the compensation weight coefficient, satisfying 6. The method for accessing and configuring multi-brand and multi-type drone devices according to claim 5, wherein The deviation compensation term ΔC of the historical protocol j The calculation method includes: The optimal compensation vector ΔC is obtained by minimizing the difference between the target protocol p and the historical protocol j in the dimension of communication characteristics j , and its calculation formula is as follows: Among them, is the communication feature matrix of the target protocol, is the communication feature matrix of the historical protocol j, Δ is the compensation matrix to be solved, and |·| F represents the Frobenius norm; Take the optimal ΔC j as the compensation item for protocol j and participate in the weighted fusion generation of the target protocol access configuration.

7. A method for configuring access to multi-brand and multi-type drone devices according to claim 1, characterized in that, The method further includes the step of performing closed-loop optimization on the access configuration, specifically including: Enable the forward absorption channel to accept new feature vectors of the target protocol and update the communication layer feature matrix. The update method is: Among them, represents the communication feature matrix at time t, ΔM t is the change amount of the protocol features newly observed in the current period, and ⊕ represents the matrix-level splicing or augmentation operation, generating an updated communication feature matrix Enable the feedback correction channel to dynamically adjust the edge weights of the data layer graph according to the current protocol execution error e t For the graph weights Update, and the calculation formula is: Among them, represents the edge weight of edge e in the data layer graph at time t ij of which, e t is the protocol execution error metric, λ is the error sensitivity adjustment coefficient, and Φ(e t ) is the feedback correction function, with the domain e t ≥0; The closed-loop optimization process is automatically triggered at fixed time intervals Δt or when the protocol error e t exceeds a preset threshold, and continues to iterate until the configuration is stable or the error converges.

8. A protocol adaptation device for UAV communication access, which is used to execute the method according to any one of claims 1-7, characterized in that, It includes: A feature modeling module, used to perform three-dimensional protocol feature modeling on the communication protocol of the target UAV, and construct a communication layer feature matrix, a data layer association graph, and a control layer behavior tensor; A difference analysis module, used to calculate the difference degree index between the target protocol and historical protocols based on the multi-dimensional protocol difference degree calculation model and combined with historical protocol data; A configuration generation module, used to generate an access configuration parameter set for the target protocol according to the protocol feature clustering results and difference compensation strategy, and fuse the basic configuration template and historical compensation items; A closed-loop optimization module, used to perform dynamic optimization on the access configuration, including: augmenting the communication feature matrix through the forward absorption channel, and updating the edge weights of the data layer graph based on the protocol execution error through the feedback correction channel; A control unit, used to coordinate the work processes of each module, and trigger the configuration update and optimization iteration process according to the difference degree and error feedback results.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1-7.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method according to any one of claims 1-7.

Citation Information

Cited By

  • Drawing life cycle evidence storage method and system based on alliance chain

    CN121682798A