184results about How to "Reduce build time" patented technology

The invention discloses an intelligent legal instrument generation method and system based on a large model, and relates to the technical field of natural language processing. According to the method, law related texts are collected through multiple channels and preprocessed, a law domain knowledge graph is constructed, and after pre-training is conducted through a BERT model, the graph and a document sample are fused for fine tuning training; a user selects a document type and an input parameter by means of the interactive interface and transmits the document type and the input parameter to the model to generate a first draft of the legal document, and then multi-dimensional evaluation and optimization circulation are carried out until the legal document reaching the standard is output; the method covers the whole process of data processing, knowledge graph construction, model training and document generation evaluation, enriches knowledge reserve by using multi-source data, improves model pertinence by combining pre-training and fine tuning, and guarantees document quality through multi-dimensional evaluation. The problems of low legal instrument writing efficiency, unstable quality and the like are effectively solved, the legal instrument generation efficiency and quality are improved, and intelligent development of the legal field is promoted.

The invention discloses a knowledge graph-based nursing corpus automatic creation method, and relates to the field of artificial intelligence technology and medical information processing, and the method comprises a multi-dimensional data acquisition stage, a knowledge graph construction stage, a corpus automatic generation and updating stage, and a corpus quality control and evaluation stage. According to the automatic corpus generation method driven by the knowledge graph, dynamic generation and real-time updating of corpora are achieved by constructing an ontology structure, a corpus generation template and an updating algorithm in the nursing field; through a deep learning model and a transfer learning technology, new nursing knowledge and a diversified information structure are continuously introduced into the corpus, and the diversity of corpus content is increased.

The invention belongs to the technical field of power distribution network fault recovery, and provides a power distribution network fault recovery method and system based on multi-agent reinforcement learning, and the method comprises the steps: taking a recovered load as a target function based on the load priority of a power distribution network, and taking the operation limitation of each node line in the power distribution network as a constraint condition; constructing a power distribution network fault recovery problem model; based on the power distribution network fault recovery problem model, constructing a multi-agent reinforcement learning agent set, and converting the power distribution network fault recovery problem into a Markov decision process; and a multi-agent reinforcement learning framework is constructed based on the agent set of multi-agent reinforcement learning, multi-agent training is carried out, each agent generates a decision according to respective observation information, and a fault recovery strategy is obtained. The objective function and the constraint condition constructed based on the load priority of the power distribution network and the node line operation limitation can ensure the accuracy and the feasibility of the recovery strategy. And the strategy generation time is shortened through the multi-agent parallel decision-making process.

The invention provides a business intelligent report generation method and system based on a large model and a storage medium, and the method comprises the steps: obtaining a user report demand, inputting the user report demand into a pre-trained large model, and carrying out the intention recognition, and obtaining a user intention; determining an SQL query statement according to the user intention, and performing data acquisition according to the SQL query statement to obtain target data; and obtaining a report intention in the user intention, determining a target report according to the report intention, and performing data filling on the target report according to the target data to obtain a business intelligent report. According to the embodiment of the invention, the automatic SQL query and commercial intelligent report generation process effectively reduces the generation time of the commercial intelligent report and improves the generation efficiency of the commercial intelligent report.

The invention relates to the technical field of water affair management, and discloses a modularization-based intelligent water affair light application rapid building system, which comprises a template library management module, which is used for storing and managing a plurality of preset function templates, and each function template is used for executing a specific function in water affair management; the parameter configuration module communicates with the template library management module and is used for performing parameterization configuration on the selected function template according to user requirements, and the parameterization configuration comprises data source configuration, operation parameter setting and output option selection; and the system combination and deployment module is used for combining the function templates subjected to parameterization configuration into a complete application system, and deploying the application system into a running environment through a containerization technology and a Kubernetes arrangement tool. Through modular design and template library management, the system can be rapidly configured and deployed according to different water affair management requirements, the construction time of the system is greatly shortened, and the system adapts to rapidly changing business requirements.

The invention discloses a new method for constructing an oil reservoir simulation agent model based on transfer learning, and relates to the technical field of petroleum engineering, and the method comprises the steps: building a fine-scale oil reservoir model; constructing a high-fidelity data set; carrying out scale upgrading and coarsening on the oil reservoir model; constructing a low-fidelity data set; constructing a deep learning model; training a pre-training model by using the low-fidelity data; finely tuning the pre-training model by using the high-fidelity data; and predicting output parameters. According to the method, it can be guaranteed that under the condition that a small amount of high-fidelity training data is used, oil reservoir multi-term flow under the variable well control condition can be accurately predicted, dependence on the high-fidelity training data in the agent model building process is effectively reduced, and the agent model building time is shortened on the premise that prediction precision is not affected. According to the method, the embedded module is introduced to process the time-varying well control parameters, the problem that matrix element parameters corresponding to well positions are difficult to effectively diffuse to the whole output result through finite times of convolution is solved, and the global influence of the well control parameters on the output result of the proxy model is improved.

The embodiment of the invention is suitable for the technical field of block chains and data processing, and provides a block chain data processing method and device and computer device.The method comprises the steps that when data stored in a block chain is read according to an account address, state index storage information corresponding to the data is obtained, the state index storage information is constructed in the form of KV key value pairs and used for storing physical K values corresponding to all nodes in the state tree; determining the path depth of a leaf node where the account address is located, wherein the path depth represents the number of target nodes contained on a storage path of the data; respectively determining a current path field of each target node; determining a physical K value corresponding to each target node in the state tree based on the current path field; and reading the physical V value corresponding to the physical K value corresponding to each target node in parallel to obtain a complete storage path, and reading data through the storage path. By adopting the method, the read-write efficiency of the block chain data can be improved.

The invention discloses a mobile terminal path planning method and system based on a distance field, and the method comprises the steps: firstly building an ESDF map, and calculating a gradient value of each grid point in the ESDF map; establishing a state vector and a discretization state transition equation of the mobile terminal based on the vehicle model; and finally, calculating the next state of the mobile terminal moving from the current node by using an improved # imgabs0 # algorithm until an optimal path from the starting node to the terminal node is formed. Wherein the ESDF cost is calculated by using the ESDF value and the gradient value of the node, and the cost function of the improved # imgabs1 # algorithm is the sum of the actual cost, the heuristic estimation cost and the ESDF cost. According to the distance field and gradient information of the ESDF map, collision detection and expansion cost calculation in the # imgabs2 algorithm are combined, and the path planning efficiency is improved while the path safety is guaranteed.

The invention relates to the technical field of tire simulation design, in particular to a tire force transmissibility finite element simulation method and system and a program product. According to the method, an axial symmetry finite element model comprising a tire, a rim, internal air and an additional deformable object is established, material attributes and contact pairs are defined, and a three-dimensional tire model is generated in combination with inflation simulation. An air unit with a displacement freedom degree and a self-adaptive grid division technology are adopted. The method comprises the following steps: simulating the embedding, inflating and loading processes of a tire under different working conditions by dynamically switching a contact pair relationship, performing steady-state dynamics analysis on the force transmissibility of the tire, and calculating an acceleration frequency response function and a force transmissibility curve of a tread and a rim center. According to the method, the analysis efficiency is greatly improved, the simulation time is shortened by 80%, meanwhile, a high-precision simulation result is kept, and the method is suitable for force transmissibility evaluation of tires with multiple air pressures and multiple specifications and has wide application value.

The invention discloses a thin-wall three-period minimal curved surface lattice structure laser additive manufacturing method, which relates to the field of additive manufacturing, and comprises the following steps: (1) according to a three-period minimal curved surface implicit function and additive manufacturing layer thickness, obtaining a single-channel contour path of each layer through direct slicing; (2) generating parallel filling lines in each layer, and intersecting the parallel filling lines with the combination of the contour paths of the layer and the higher layer; (3) combining every two intersection points to form a line segment, and taking the line segment as a filling path if sampling points of the line segment meet the conditions of a functional expression, an inclination angle and a length; (4) taking the filling path of the layer as a supporting layer, and moving the filling path upwards by one layer to form a filling layer; and (5) all the paths are input into additive manufacturing equipment, and different machining parameters are set for manufacturing. The laser path of the thin-wall three-period minimal curved surface lattice structure can be efficiently generated, only one melting channel is contained in the thickness direction of the structure, the wall thickness is smaller than 200 micrometers, and the cavity defect generated in the slicing process can be overcome.

The invention is suitable for the technical field of data processing, and provides an education evaluation report generation method and system, and the system comprises a data management module which is used for obtaining the data of evaluation indexes in each evaluation item in an evaluation report, and obtaining the index data; the index management module is used for adjusting the evaluation index and the weight thereof, and performing statistical processing on the index data according to the evaluation index and the weight thereof to obtain statistical data; the template management module is used for filling the index data into a Prompt template and inputting the filled Prompt template into a large language model, and the large language model analyzes facts reflected by the index data based on cue words in the Prompt template to obtain analysis content; and the report generation module is used for integrating the analysis content and the statistical data into an evaluation report template to obtain a target evaluation report. The large language model is used for deep data analysis, the evaluation report is generated, the report generation time is remarkably shortened, and manual errors and subjective deviations are reduced.

The invention relates to a non-polar thermoplastic plate with a polar surface coating as well as a preparation method and application of the non-polar thermoplastic plate, and belongs to the technical field of automobile coatings. The problems that in the prior art, in the process of attaching a paint film to the surface of a non-polar thermoplastic plate, a base material is prone to thermal stress or deformation, the paint film is uneven in attachment and prone to falling off or peeling off, cost is high, process time is long, and industrialization is difficult to achieve are solved. According to the method, low-cost polyphenol and amine substances are used for spraying codeposition, and a particle coating with a certain thickness and roughness is formed on the surface of the non-polar thermoplastic plate by regulating and controlling the condition parameters of the spraying process, so that the adhesion capacity of the surface of a matrix is enhanced; the introduced active groups increase the surface polarity of the matrix and enhance the interface bonding between the matrix and the paint film; the particle coating is prepared at normal temperature, so that thermal stress or deformation caused by heating of a matrix is avoided; and by adopting the processes of adding a strong oxidant and spraying and codepositing, the modification time is greatly shortened, the working efficiency is improved, and the industrial production of the modified non-polar thermoplastic plate is promoted.

The invention provides a video generation method and device, electronic equipment and a storage medium, and relates to the technical field of computers, in particular to the technical field of artificial intelligence such as deep learning and large models. According to the specific scheme, the method comprises the steps of firstly receiving a video generation request, then determining a target type to which a target product belongs according to first description information, then obtaining first interaction information of a product video associated with the target type, and finally generating a target video corresponding to the target product according to the first description information and the first interaction information. Therefore, on the basis of shortening the video generation duration and reducing the video generation cost, the video generation efficiency and accuracy are improved.

The invention discloses a laser optical fiber polarizer system suitable for compact ring plasma, which comprises a laser, a first optical fiber coupler, a circularly polarized light generation module, a first collimator, a second collimator, a second optical fiber coupler, a second polarizer, a photoelectric converter and a data processing module, according to the laser fiber polarimeter system, non-contact elements for diagnosing and testing the compact ring plasma are reasonably arranged, so that high-precision testing of density and magnetic field information of the compact ring plasma is realized. According to the invention, the reference signal is compared with the measurement signal passing through the plasma, the phase difference between the signals can be extracted through phase demodulation, and then the density and magnetic field information can be obtained according to the Faraday effect; in addition, the use of open optical devices is reduced, the cost is greatly reduced, the platform building time is shortened, and meanwhile, the interference of the external environment in the experiment process is also reduced.

The invention relates to a concrete mesoscopic model construction method and device, equipment and a storage medium. The method comprises the steps that mesoscopic model parameters and aggregate parameters are obtained; building model walls based on the model size parameters so as to enable the model walls to form a closed model area; determining a plurality of aggregate particle sizes of each graded aggregate based on the aggregate grading proportion and the grading particle size range, and generating a plurality of aggregates based on the aggregate particle sizes; randomly putting a plurality of aggregates in the model area, simulating the motion state of each aggregate based on the wall performance parameters and the aggregate performance parameters until each aggregate is in a stable state, and obtaining the target position of each aggregate; and constructing a concrete mesoscopic model based on the target position of each aggregate and the aggregate particle size. According to the method, all the aggregates are generated at a time and put into the model area, the target positions of all the aggregates in the stable state are calculated, then the concrete mesoscopic model is constructed, the model construction time can be shortened, and it is ensured that model grading meets the expectation.

The invention discloses a working side slope landslide prevention and control method, system and device and a storage medium, and relates to the field of geological disaster prevention and control, and the method comprises the steps: extracting geological parameters of a soft layer from a three-dimensional geological model constructed based on remote sensing surveying and mapping data, and obtaining a shear strength parameter range of the soft layer according to the geological parameters, generating a two-dimensional slope displacement vector cloud picture and a two-dimensional slope strain rate cloud picture according to the three-dimensional geological model and the shear strength parameter range, generating a coverage range after the landslide occurs in the predicted landslide area according to the geological parameters, the shear strength parameter range, the two-dimensional slope displacement vector cloud picture and the two-dimensional slope strain rate cloud picture, and calculating the coverage range according to the coverage range. And displaying the preset landslide prevention and control method matched with the maximum safety distance. Based on the characteristics of high detection depth and precision of remote sensing surveying and mapping data, the influence of deep geological parameters on the landslide is fully considered in the obtained coverage range, and the landslide dangerous area delimiting precision and the landslide prevention and control measure configuration precision are improved.

The invention discloses a text-based motion video generation method and device, a storage medium and equipment, and belongs to the technical field of computers. Obtaining a positive prompt word and a negative prompt word; performing text coding on the positive prompt words and the negative prompt words to obtain text features; the Animatediff model and the LCM Lora model are inserted into a Stable-Diffusion model, and a video generation model is obtained; processing the text features by using a video generation model to obtain a motion video; and performing video enhancement processing on the motion video by using the video enhancement processing model to obtain a high-definition motion video. According to the method, the LCM Lora model can be used for reasoning acceleration, the ControlNet model is used for enabling the video action to be consistent with the action in the action reference video, and the video enhancement processing model is used for generating the high-definition video.

The invention provides a test data generation method. The method comprises the following steps: acquiring field types of business fields in a plurality of business data tables of a specified business database; obtaining business data associated with each business field according to a data desensitization strategy matched with the field type of each business field; generating a structured file according to each business field and business data; obtaining a primary and foreign key relationship among business fields in the plurality of business data tables; and according to the structured file and the main and foreign key relationship, generating test data for testing each service field, thereby realizing accurate identification and processing of sensitive fields, avoiding information leakage, and ensuring that the generated test data not only conforms to the field type and service scene of the actual service field, but also conforms to the service scene of the actual service field. The logic correctness of the generated test data is ensured, and the effectiveness and accuracy of the test data are improved; meanwhile, manual operation is not needed in the test data generation process, and the labor cost and the test data generation time are reduced.

The invention relates to a cross-platform compiling method and device of a Linux module, computer equipment and a medium. The method comprises the following steps: receiving a compiling file package sent by a development platform, and decompressing the compiling file package to a target compiling directory; adjusting the compiled file package according to an adaptation result of the compiled file package and the target platform; receiving a Linux module code sent by the development platform; and compiling the Linux module code through the adjusted compiling file package. By adopting the method, the compiled file package can be conveniently managed; through transmission of a single compiling file package, a stable compiling environment can be established on an application platform, the situation that a technician needs to regenerate a compiling file package every time the compiling environment is switched is avoided, the number of compiling files needing to be managed by the technician is reduced, and the compiling efficiency is improved. The compiling environment building time needed in the Linux module code compiling process is shortened, and then the development efficiency of the Linux module is improved.

The invention discloses a method for automatically generating a structured emergency plan template in the power industry, and the method comprises the steps: formulating a power scenario plan text writing standard, normalizing a non-structured scenario plan text format, and obtaining a normalized scenario plan text; inputting the standardized scene plan text into a generative large language model, extracting scene information, and generating scene description; establishing a structured emergency plan basic information base; performing semantic matching by using the bidirectional coding pre-training language model, and extracting behavior feature information; the generative large language model generates behavior description information; and constructing a structured plan template generation module, and generating a structured emergency plan template in a specific scene. According to the method, the template generation time is greatly shortened, the labor intensity of workers is reduced, the extraction method based on the large language model is higher in precision, errors caused by manual operation are greatly reduced, rapid generation of the emergency plan template is achieved, and the generation efficiency of the emergency plan template is improved.

The invention relates to a display method and device of a display interface, computer equipment and a storage medium. The method comprises the steps of obtaining first interface configuration data; from the first interface configuration data, obtaining size information of a preset screen, first style pixel data of each plug-in and a category identifier of each plug-in when the preset screen generates a preset interface; according to the size information and the category identifier of each plug-in, performing pixel conversion on the first pattern pixel data of each plug-in to obtain second pattern pixel data of each plug-in; and in the first interface configuration data, replacing the corresponding first pattern pixel data with the second pattern pixel data to obtain second interface configuration data, and performing interface generation on the to-be-matched display screen according to the second interface configuration data. According to the embodiment of the invention, the requirement of displaying multiple types of plug-ins can be met, the display effect of the interface generated on the to-be-matched display screen is consistent with that of the interface generated on the preset screen, and more accurate global replacement is realized.

The invention belongs to the technical field of time sequence anomaly detection, and particularly relates to a streaming time sequence sub-sequence anomaly detection method fusing global and local features, which comprises the following steps: clustering all sub-sequences in an initial batch, constructing an inverted index, and performing approximate neighbor search on each sub-sequence; the used calculation formula is used for measuring deviation between the subsequences and neighbors of the subsequences and between the subsequences and cluster centers of clusters containing the neighbors of the subsequences and dispersity of neighbor distribution, and global and local feature fusion is achieved; taking each current cluster as a historical cluster, calculating the maximum distance between each sub-sequence in each historical cluster and the cluster center of the cluster, and taking the maximum distance as a density parameter of the historical cluster; and clustering all the subsequences in the latest batch, calculating the Euclidean distance between the cluster center of each new cluster and the cluster center of each historical cluster, carrying out cluster combination based on the comparison of the Euclidean distance and the density parameter of each historical cluster, and repeatedly executing neighbor search and abnormal score calculation. According to the invention, accurate detection of large-scale streaming data can be realized.

The utility model discloses a methanol rectification system capable of improving methanol yield, which is characterized by comprising a crude methanol preheater, a pre-rectification tower, a pre-prognosis methanol pump, a pressurizing tower inlet preheater, a pressurizing tower, an atmospheric tower, an atmospheric tower condenser, an atmospheric tower reflux tank, an atmospheric tower product cooler and a methanol storage tank, and the cold medium outlet of the pressurizing tower inlet preheater is also communicated with the feeding hole of the atmospheric tower through a pipeline. The method has the advantages that in the start-up process, tower bottoms in the pre-rectifying tower are heated by the tower inlet preheater of the pressurizing tower and are supplied to the pressurizing tower and the atmospheric tower respectively, so that the time required for establishing the liquid level of the atmospheric tower can be shortened, the start-up time of the system is shortened, the output time of refined methanol is prolonged, and finally the yield of methanol is increased; the heat carried by the steam discharged from the top of the pressurizing tower can be recycled, and the methanol carried by the steam can be recovered, so that the yield of the methanol is increased.

The invention relates to the technical field of organoid construction, and discloses a heart organoid construction method which comprises the following steps: preparing a culture medium I containing CD lipid, ITS-X, L-AA2P, alpha-MTG, insulin, FGF2 and a GSK-3beta inhibitor; preparing a culture medium II containing CD lipid, ITS-X, L-AA2P, alpha-MTG, insulin, FGF2 (fibroblast growth factor 2) and VEGFA (vascular endothelial growth factor A); the method comprises the following steps: digesting a human induced pluripotent stem cell mass, suspending the human induced pluripotent stem cell mass in a culture medium I, inoculating the human induced pluripotent stem cell mass into an ultra-low adsorption cell culture plate with a U-shaped hole bottom, culturing for 36-40 hours, removing the culture medium I, adding a culture medium II, and culturing until a heart organ is formed. In the method provided by the invention, two culture media are sequentially used for performing two-stage differentiation culture, namely, the heart organoid can be constructed from the human induced pluripotent stem cells dispersed after digestion, so that the construction process of the heart organoid is simplified, the construction time is shortened, and the operation difficulty is reduced.

The invention relates to the technical field of information retrieval, in particular to a large-scale semantic retrieval optimization method based on deep learning, which comprises the following steps: acquiring to-be-retrieved text data, and converting the to-be-retrieved text data into semantic vector representation through a deep learning model; carrying out topological structure analysis on the semantic vector representation to obtain topological characteristic information of a semantic space; on the basis of the topological characteristic information, a topological perception hierarchical navigable small-world graph index with a multi-layer structure is constructed, and each layer comprises a node set and a connection relation thereof; dynamically optimizing and adjusting the topology-aware hierarchical navigable small-world graph index according to an actual query mode and system performance feedback; and receiving a query request, converting the query request into a query vector, executing multi-layer navigation search in the topology-aware hierarchical navigable small-world graph index, and returning a retrieval result related to the semantics of the query vector.

The invention discloses a method for directly inducing generation of organoids from IPS and application, and belongs to the field of cell engineering. The invention provides a method for directly and rapidly generating a target organ based on human IPS, and the lung organ, the placenta organ and the blood vessel organ are obtained. According to the method, the organoid represented by three germ layers is induced, so that the method is suitable for constructing various organoid. Compared with the prior art, the method has the advantages that step-by-step induction is not needed, one-step construction is successful, the operation is simple, the construction time is greatly shortened, the process is repeatable, long-term stable culture and passage are realized under in-vitro conditions, and the freezing survival rate is high; the method has the advantages that the method is simple in operation, the cultured organoid is good in homogeneity, RNA-seq results show that high expression of the blood vessel related genes appears in the organoid, and the maximum length of the organoid can reach 400 micrometers, so that the culture method can spontaneously induce expression of part of the blood vessel related genes to support growth of the organoid. The invention provides an important tool for researching pathogenesis of diseases and subsequent drug development.

The invention provides a method and a device for realizing service process recovery by accessing a shared memory, and belongs to the field of storage configuration. Comprising the following steps: dividing a system into a business layer, a memory pool layer, a shared memory connection layer and an operating system layer; initializing all objects needing to be recovered, and constructing the objects on a shared memory area; replacing all global new / deelete operators in the system with a shared memory version; executing a fork operation in the system, and managing the shared memory area for the generated child process inheriting the parent process through a pointer type object of the child process; and in a multi-process model in which a parent process or child process relationship is not shared, calculating the mapping offset during the current call and the last call, and correcting all the pointer objects according to the offset. According to the method, the stability, reliability and efficiency of the system are effectively improved, quick recovery of the shared memory is realized in a multi-process environment, and powerful support is provided for high-performance operation of the system.

The invention relates to the technical field of ray tracing, and provides a fast deterministic channel modeling method based on improved bounding box ray tracing, which comprises the following steps: acquiring a scene containing a plurality of buildings, and determining a surface area heuristic value of a potential segmentation point in the scene; the potential segmentation point is a point used for dividing the space represented by the current node into two subspaces in the process of constructing the hierarchical bounding box tree; the node is a unit in the hierarchical bounding box tree structure; acquiring a set of position surface area heuristic values of all potential segmentation points, and finding out a minimum surface area heuristic value in the set as an optimization problem for modeling; and obtaining a preset threshold value, and when the number of patches in the space represented by the nodes is greater than the threshold value, solving an optimization problem, determining a segmentation plane, and constructing a hierarchical bounding box tree structure by a particle swarm optimization-based traversal segmentation point optimization method.

The invention provides an sql language conversion method, electronic equipment and a storage medium, and the method comprises the steps: carrying out the splitting processing of a to-be-converted natural statement, so as to obtain a plurality of to-be-converted natural segmented words; performing sql language conversion on each to-be-converted natural segmented word to obtain an initial sql language segmented word corresponding to each to-be-converted natural segmented word; if the initial sql language segmented word is any target sql language conversion segmented word, determining the initial sql language segmented word as a target sql language segmented word; and deleting each target sql language segmented word in the plurality of initial sql language segmented words, and splicing the rest initial sql language segmented words after deletion in sequence to obtain a target sql statement. According to the method, the converted initial sql language segmented words are simplified, so that the generation time of the obtained target sql statement is shortened, and the conversion efficiency of the target sql statement is improved.

The application relates to the technical field of network resource management and the technical field of reinforcement learning, and discloses a spectrum decision method, device and equipment based on a double-graph structure and reconstruction cost, which comprises the following steps: inputting original state data of low-altitude intelligent networking into an actor network of a reinforcement learning architecture, generating a predicted spectrum decision scheme of the low-altitude intelligent networking through the actor network, and obtaining an expected reward value output by a critic network based on the original state data and the predicted spectrum decision scheme; determining a trained actor network based on the expected reward value and an actual reward value; and generating a target spectrum decision scheme of the low-altitude intelligent networking based on a current double-graph structure and the trained actor network. The application can improve the generation efficiency of the target spectrum decision scheme of the low-altitude intelligent networking.