307 results about "Cyber Space" patented technology

The invention discloses a visual language navigation method for cross-modal alignment in a dynamic shielding environment, and the method comprises the steps: collecting multi-modal data through a visual sensor, an inertial measurement unit, a laser radar and the like, and carrying out the preprocessing and time synchronization; sensing the dynamic shielding object through a model composed of a convolutional neural network and a long-short-term memory network, and estimating the future change of the dynamic shielding object in combination with a space-time sequence prediction algorithm; a double-branch convolutional neural network and a Transform based on a dynamic attention mechanism are adopted to respectively extract visual and semantic features and fuse the visual and semantic features; on the basis of occlusion prediction, potential occlusion region features are extracted in advance from a time dimension, an occluded image is repaired by using a generative adversarial network and geometric constraints in a space dimension, and cross-modal feature alignment is optimized through an attention mechanism; planning a path by using a hybrid reinforcement learning algorithm based on a deep Q network-space and a fast exploration random tree, and dynamically adjusting according to real-time shielding; according to the method, the accuracy, adaptability and reliability of visual language navigation in a dynamic shielding environment are improved.

The invention provides a data security method and system fusing an MIN and a unified AI theory, and the method comprises the steps: S1, completing the multi-dimensional identification registration and synchronization of a data space in a multi-identification network MIN based on a data space identity identification DSID and an ontology semantic mapping frame; s2, acquiring a data acquisition record from a multi-identifier network MIN, and constructing an access control atlas in combination with a semantic vector of a data subject identifier DSID-Item; s3, analyzing the access request and verifying the authority, identifying the sensitivity level of the access request, and triggering a privacy enhancement mechanism when the sensitivity level is marked as medium-high-level privacy data; and S4, judging whether the data object reaches a value end point in the life cycle of the data object or not by taking a data space identity (DSID) registered in the multi-identifier network (MIN) as a retrieval anchor point, and if so, executing a verifiable multi-stage destruction operation. According to the invention, a full-life-cycle trusted data security technical scheme can be realized, and intelligent, explainable and efficient security management of data streams in a network space is realized.

The invention discloses a network space map surveying and mapping method and system based on multi-source data fusion, and the method comprises the steps: obtaining a multi-source data set in a unified format based on network flow data, equipment information data and geographic position data; obtaining a network asset entity and an incidence relation graph thereof through entity identification and correlation analysis based on the multi-source data set with the uniform format; obtaining a network space three-dimensional map model through three-dimensional space mapping and visual rendering based on the network asset entity and the association relationship map thereof; based on the network space three-dimensional map model, performing dynamic updating according to the accessed real-time data flow to obtain real-time network space situation data; and obtaining a network security risk assessment result through anomaly detection and threat identification based on the real-time network space situation data. According to the invention, visual display and security situation awareness of the network space are realized, and a brand new decision support tool is provided for network security management.

The invention discloses a network space security intelligent monitoring and analysis system, and relates to the technical field of network security monitoring and analysis, and the system comprises a multi-source data collection module which collects multi-dimensional data in a full-link manner, and the collection frequency is dynamically adjusted along with a network load; the data preprocessing module cleans the fused data and generates a standardized analysis data set; the AI intelligent risk identification module identifies various safety risks in real time through a mixed deep learning model; the real-time response processing module starts differential processing according to a three-level mechanism; the threat traceability analysis module traces an attack link and generates a report; the security situation visualization module displays the security state in multiple dimensions; the data encryption storage module encrypts and protects data and performs double backup; and the system self-optimization module dynamically optimizes the strategy through incremental learning. The method is accurate in risk identification, timely in response processing, reliable in traceability and evidence storage, and efficient in cross-domain cooperation; terminal protection and third-party access control are enhanced, and network space security and stable service operation are comprehensively guaranteed.

The invention relates to the technical field of power network space optimization management, and particularly discloses a power network space optimization analysis method and system based on a power grid GIS, and the method comprises the steps: firstly enabling a power grid GIS port to firstly match the space coordinates of a high-voltage side bus of a power transmission substation and a power distribution feeder starting node, building an electrical connection topology after verification, and evaluating a capacity coupling factor; dividing feeder segments according to topology, calculating a loss rate and a risk index, and correcting the risk index in combination with a capacity coupling factor, thereby realizing accurate quantification of the feeder state and improving risk assessment scientificity; secondly, feature vectors are constructed according to the GIS center points, the loss rates and the risk indexes of the feeder line sections for clustering, GIS layers with labels are output, feeder line clusters in the same state can be recognized conveniently, and support is provided for regional analysis; and finally, detecting a loop gap based on a layer, analyzing connectivity, identifying a weak link and optimizing feedback, accurately positioning a network short board, and providing a targeted reference for an optimization scheme.

The invention discloses a cyberspace surveying and mapping threat detection method and a cyberspace surveying and mapping threat detection system, and belongs to the technical field of network security. Performing multi-dimensional feature extraction including spatial distribution feature extraction, time distribution feature extraction, protocol interaction feature extraction and load response feature extraction to construct feature vectors, and outputting a prediction feature vector of a next time step according to a plurality of feature vector pairs in a latest time window through a learning model; performing error comparison with the actual feature vector to screen out abnormal features, performing attack intention assessment according to the abnormal features to assess an attack intention comprehensive threat coefficient, and finally assessing a threat risk coefficient in combination with the abnormal features and the attack intention comprehensive threat coefficient. The accuracy of attack intention assessment is improved through the multi-dimensional features, and the assessment accuracy of the threat risk coefficient is improved by combining the abnormal features and the attack intention comprehensive threat coefficient.

The invention provides an efficient task scheduling strategy recommendation method and system for a distributed network surveying and mapping scene, and relates to the crossing field of network space surveying and mapping, network security, distributed scheduling and intelligent optimization recommendation algorithms. Analyzing the task request to extract a task type, a task target network address set, a regional distribution feature, a task execution timeliness requirement and calculation and bandwidth resources required by estimation, and forming a task feature vector; and performing state acquisition and capability evaluation on the distributed surveying and mapping nodes based on task resource requirements defined by the task feature vectors, and performing normalization processing to form node capability vectors. Through multi-source information vectorization and closed-loop optimization, adaptive matching of task scheduling and resource dynamics is realized.

The invention relates to the technical field of network situation awareness, in particular to a network situation intelligent awareness security system based on multi-source data analysis, which accurately acquires a perception threat confidence vector, a service interruption risk vector, an attack chain complete vector and a user behavior deviation vector based on multi-source data fusion and targeted analysis. According to the method, network situation key information can be comprehensively and accurately captured, a reliable basis is provided for subsequent decision making, the comprehensiveness and real-time performance of threat perception are improved, dynamic balance of safety protection, service efficiency and operation and maintenance cost is achieved by relying on a multi-target game mechanism and a network space digital twinning technology, and the safety and reliability of threat perception are improved. A simulation deduction deviation value is introduced as a core index of strategy evaluation, a real network situation is simulated through a digital twin environment, the difference between a strategy execution effect and an expected target is quantified, and in combination with reproduction of an attack scene by the twin environment, deviation value analysis is ensured to be more fit with the real network environment.

Provided in the present invention are a method and system for ensuring the trustworthiness and security of cyberspace. The method comprises: step S1, on the basis of a multi-identifier network, implementing multi-identifier addressing having authentication and packet signing embedded therein; step S2, defining digital customs, a digital passport and a digital visa based on the multi-identifier network; step S3, on the basis of time data, a digital visa key and a hash function, performing calculation to obtain the digital visa based on the multi-identifier network; step S4, on the basis of the hash value of the digital visa and a cross-states passport key, performing calculation to obtain the digital passport based on the multi-identifier network; step S5, implementing the digital customs by means of a multi-identifier network router, and using and maintaining an exit border table by means of the digital customs; and step S6, by means of the digital customs, using an arrival border table to verify the digital passport, and updating the arrival border table. The present invention can effectively guarantee the trustworthiness and security of cyberspace, and enable the traceability of blockchain logs, thereby providing a better foundation for ensuring the order, rule of law and peace of cyberspace.

The invention relates to the technical field of emergency management, in particular to a visual emergency management command system which comprises a network channel judgment module, a broadcast circling adjustment module, a material linkage retrieval module, a bidirectional response matching module and an instruction popup prompt module. And judging the fluctuation condition of the signal-to-noise ratio along with time. According to the method, the signal dynamic change and the communication quality sequence are fused, multi-channel use is optimized in real time, the boundary is dynamically linked according to spatial distribution and equipment motion characteristics in the region circling operation, and multi-dimensional association is realized by combining the spatial distance between a team and a warehouse and a material demand type; the supply-demand relationship is automatically graded according to the actual supply difference and visual identification is realized, and the material gap is displayed through popup window information and geographical highlight linkage, so that multi-link data-driven linkage across network, space and resource dimensions is integrally realized, the fine resource allocation and efficient response capabilities are improved, and the collaborative guarantee in an emergency scene is enhanced.

The invention discloses a multi-dimensional network space security intelligent analysis and early warning platform, which relates to the technical field of network security intelligent analysis and early warning, and comprises the following steps: a multi-source network data acquisition module acquires multi-level data and encrypts and gathers an association identifier; the data preprocessing and standardization module filters invalid data to unify a format, and encrypts sensitive information to generate a verification abstract; the multi-dimensional feature extraction module extracts multiple types of features and maps the features to the same vector space; the intelligent security analysis module analyzes threats in combination with a deep learning model and a rule engine; the grading early warning and response module triggers early warning according to grades; the distributed data storage and traceability module adopts a hybrid architecture to ensure that data cannot be tampered and traced; and the visual interaction and management module presents the state in multiple dimensions. According to the platform, multi-source data are integrated to realize multi-dimensional collaborative analysis, and an early warning threshold is dynamically adjusted to avoid false alarm and missing alarm; strategy dynamic optimization, compliance adaptation and full-process tracing are realized, and the network security protection capability is comprehensively enhanced.

A system implements an emergent web space to bind a host computer to one or more client computers. A host computer generates data aspects of characteristically random data to create emergent web spaces in which website resources can be rendered by one or more client computers intersecting the same emergent web spaces. A binding is established between the host computer and one or more client computers when a block of true random data generated by the host computer is provided to the one or more client computers. A unique block of true random data together with the remote storage of data aspects constitutes a zone. A software application executing on the one or more client computers is configured to retrieve data aspects from within the same zone, and further configured to resolve and then render website resources.

The invention relates to the technical field of computer vision and deep learning, and discloses a cascade diffusion model-based virtual fitting method, which comprises the following steps of: inputting a model image and a target clothing image into a multi-condition diffusion model, and outputting a posture control variable and a depth control variable; integrating the posture control variable and the depth control variable into a de-noising U network through a multi-control injection module, and generating a low-resolution try-on image of the model wearing the target garment; inputting the low-resolution fitting image into the super-resolution diffusion model, and performing feature extraction on the low-resolution fitting image; feature codes corresponding to feature extraction are transmitted to a control network space through a full connection layer, then the feature codes are integrated into a super-resolution diffusion model through a cross attention mechanism, the super-resolution diffusion model is guided to reconstruct a low-resolution try-on image, and a high-resolution try-on image is generated. According to the method, high-fidelity and high-resolution clothing try-on image generation is realized through a multi-condition control and super-resolution reconstruction technology.

The invention belongs to the technical field of network space security and social platform governance, and particularly provides an agent-driven active interactive social robot detection method, electronic equipment and a storage medium, and the method comprises the steps: obtaining original data of a to-be-detected account, and forming a structured data packet; constructing a preliminary screening evaluation model, and outputting a detection task state based on the structured data packet; generating an initial detection state; obtaining an interaction record; carrying out evidence extraction on the interaction record to obtain an evidence vector of the round; and performing sequential updating on the initial risk score and the uncertainty based on the evidence vector, and outputting a detection result when a stop condition is met. According to the method, an active interactive evidence obtaining mechanism of a detection agent is introduced, and on the basis of passive preliminary screening, probe interaction is automatically selected and initiated to obtain incremental evidence, so that the recognition capability of a high-simulation social robot is improved, and an auditing report containing key evidence and an interactive link is output.

The invention discloses a network rumor automatic intervention and propagation control method based on an evolutionary game model. The method comprises the following steps: firstly, tagging behavior characteristics and psychological power of network public opinion subjects, and constructing a subject-object-power ternary factor interaction atlas through emotion and activeness analysis of user historical speaking data; on this basis, in combination with an evolutionary multi-agent game, a rumor propagation process is dynamically simulated, and key propagation nodes with strong influence and high mentality plasticity are positioned in real time through a sharp change node detection algorithm. And for the identified key nodes, a scene self-adaptive intervention content generation algorithm is adopted to automatically generate personalized clarification, guidance or pacification type speech, and intervention is implemented through multi-channel accurate pushing. By means of the method, efficient, dynamic and intelligent intervention on network rumor spreading can be achieved, rumor spreading is effectively restrained, and the network space public opinion governance capacity is improved.

The invention relates to the field of network public opinion management and control, in particular to a cross-modal network emergency command system based on a network space model. The method comprises the following steps: acquiring a network public opinion multi-modal data stream, positioning a malicious public opinion propagation source account based on the network public opinion multi-modal data stream, and performing network space cross-layer holographic portraying based on the malicious public opinion propagation source account to generate a source account holographic portraying information set; on the basis, multi-modal feature extraction and network space behavior model construction are carried out on hidden negative guidance behaviors, and an implicit negative information propagation mode model is generated; on the basis, potential malicious account active tracking and risk propagation network mining are carried out, and a potential risk propagation network information set is generated; on the basis, multi-level differential intervention strategies are generated and executed, and a network public opinion intervention strategy instruction set is generated. In the network public opinion emergency command process, pollution of false information to network space is reduced, and construction of healthy and sustainable network ecology is assisted.

The invention belongs to the field of information security, and relates to a multi-agent network security operation method and device, electronic equipment and a storage medium, and the method comprises the steps: constructing a layered architecture from physical hardware to an upper application; receiving a network security global task, and decomposing the network security global task into task subsets in a coarse-grained layer; the atomic tasks are allocated to appropriate agents, a task alliance is formed, and a collaborative strategy is formulated; by improving the credible interference-free information flow control method of the operating system process, interference-free information interaction between intelligent agents is adapted; performing real-time intention synchronization and cooperation adjustment between the intelligent agents; feedback information in the task execution process is collected, the task completion condition is evaluated, and the multi-granularity collaborative decision process is optimized. The requirements of an intelligent network space security protection system are met; the method can be applied to security operation system platform scenes of multi-agent cooperation, multi-robot cooperation, multi-agent software systems, multi-digital person cooperation and the like.

The invention belongs to the technical field of network space security, and discloses a federated learning backdoor attack defense method, system and device, a medium and a program product thereof, the method comprises the following steps: generating an initialized global model through an aggregation server and issuing the initialized global model to a user, and training by the user by using a local data set and the latest global model issued by the aggregation server, generating an updated local model, uploading the updated local model to an aggregation server, then performing anomaly detection on local model gradient information uploaded by a user by the aggregation server, aggregating model gradients of normal users according to a predefined aggregation rule to obtain an encrypted global model, and then issuing the global model to the users in the system; the system, the equipment and the medium are used for implementing the method. A program product comprising a computer program of the method; according to the method, the negative influence of backdoor attack on the federated learning system is avoided, and the robustness of the system is improved.

The invention belongs to the crossing field of network space security, software engineering and artificial intelligence, and discloses a dynamic and static cooperative vulnerability full-link intelligent analysis and verification method based on a large language model and an MCP protocol, and the method comprises the steps: S1, task initialization and context construction; s2, target code deep analysis and static taint preanalysis are carried out; s3, carrying out Fuzzing Harness directional generation and iterative optimization on the basis of the stain paths; s4, performing dynamic and static collaborative sandbox execution and dynamic stain tracking; and S5, vulnerability accessibility judgment and report output of multi-source data fusion are carried out. According to the method, a dynamic and static combination mechanism is adopted, vulnerability analysis and verification can be automatically completed for software target codes, and accurate technical support is provided for vulnerability risk grading and repairing priority ranking.

The invention discloses an urban village identification method based on a graph neural network and multi-source geographic big data, and relates to the technical field of geographic images, and the method comprises the steps: obtaining remote sensing data, social perception data and building footprint data of an urban village; based on road network trajectory data, cutting built-up area boundary data through a hierarchical buffer area strategy to obtain a basic evaluation unit; the streetscape image data, the satellite image data, the interest point data, the building attribute data and the building form data are preprocessed respectively, and visual features, social economic features and building form features are constructed; constructing a graph network structure of the graph neural network model; and according to the instance sample group, in combination with a graph network structure of the graph neural network model, constructing and training a GraphSAGE model so as to improve the recognition probability of the villages in the city. According to the method, the spatial interaction and regional relation between the network spatial structure and the multi-dimensional attributes are effectively understood through the GNN.

The embodiment of the invention provides a network security information consistency representation method and device based on ontology modeling, and the method comprises the steps: classifying various entities related to network space security, and the relationships and attributes of the entities, and constructing an ontology model of a hierarchical structure; concept conflicts and ambiguity of multi-source data in the ontology model are eliminated through context binding, multi-view reconstruction and fuzzy reasoning; data is injected in real time through a heterogeneous data source to trigger ontology updating, and an ontology model is updated through a collaborative reasoning engine and an atomization updating operation set; and performing cross-level consistency verification on the static structure and the dynamic behavior in the ontology model, and verifying the robustness of the ontology model through a antagonism test. According to the scheme, the heterogeneity, conflict and dynamic variability of multi-source data in the field of network security are effectively solved, so that more accurate and robust security management and decision are realized.

The invention relates to the technical field of mobile communication and Internet of Things, and discloses a heterogeneous network space multi-level mobile Internet of Things cloud system based on artificial intelligence, which comprises a data acquisition module for acquiring real-time state data of a mobile Internet of Things terminal and generating a standardized network-service joint situation vector; the endogenous interference sensing module is used for generating a network interference association diagram based on the vector; the service-network dynamic decoupling module is used for reversely reconstructing terminal service characteristics according to the association diagram and decoupling a key network state sequence; the business intention abstraction module constructs a heterogeneous network cooperation state target according to the sequence; the collaborative scheduling strategy generation module is used for generating a multi-dimensional collaborative scheduling instruction by using a collaborative decision model; the network resource self-growing module is used for performing state sequence evolution after the rehearsal instruction and constructing a compensation network plane; and the instruction distribution and verification module is used for executing the instruction of the fusion compensation logic and updating the association diagram. According to the method, the problems of heterogeneous network interference and scheduling lag of the mobile Internet of Things are effectively solved.

The invention provides an adaptive network attack prediction and traceability system based on large model behavior deviation, and relates to the technical field of network space security. Comprising a double-domain behavior acquisition module, a consistency baseline construction module, a deviation evolution analysis module, a deviation field construction module, a traceability path inference module, an evidence chain generation module, an adaptive strategy response module and a backtracking verification module. The double-domain behavior acquisition module is configured to perform data interaction with a reasoning interface and an internal diagnosis interface of a monitored large model through a data bus or a message queue, and acquire internal domain data and external domain data in real time; wherein the internal domain data comprises a hidden layer state vector sequence, an activation gradient curve, a parameter sensitivity index and a model confidence sequence, the system significantly improves the detection precision and comprehensiveness of abnormal behaviors through double-domain data acquisition and deviation evolution analysis, and can capture the internal state change and external request characteristics of a large model at the same time.

To provide an information processing device, an information processing system, an information processing method, and a program which can contribute to human capital based management.SOLUTION: An information processing device includes: a first acquisition unit which acquires log information of a person on a cyber space and a physical space; a second acquisition unit which acquires human resource information of the person; a storage unit which stores integrated data of log information of the person and human resource information in association with each other; and a first generation unit which generates a hypothesis and a basis for a problem regarding human capital based management on the basis of the integrated data, and outputs the generated hypothesis and the basis.SELECTED DRAWING: Figure 1

The invention relates to the technical field of network space security, and provides a cross-domain honey array collaborative deployment method, device and equipment based on federated learning, and the method comprises the steps: dividing a plurality of network domains into a plurality of autonomous domains, and deploying a local honey array agent module in each autonomous domain; collecting local honey point interaction information in the autonomous domain according to each local honey array agent module; training a local honey spot decoy effectiveness evaluation model according to the local honey spot interaction information, and outputting local model parameters corresponding to each autonomous domain; according to the local model parameters corresponding to each autonomous domain, generating a global honey point deployment strategy through weighted aggregation of a federated learning aggregator; based on the global honey point deployment strategy, a local honey point deployment strategy of each autonomous domain is generated through a cross-domain scheduling mechanism module; according to the local honey point deployment strategy of each autonomous domain, deployment configuration of honey points in each autonomous domain is dynamically adjusted, and joint modeling and dynamic optimization of a multi-organization honey array deployment strategy are realized on the premise of guaranteeing data privacy.

The invention provides a network topology geospatialization expression method integrating force-oriented layout and layered mapping and related equipment, and is applied to the field of data processing. According to the method, network space element data including nodes, edges and hierarchy attributes and geographic space data serve as input, repulsive force, gravitational force and resultant force formulas are defined by combining algorithm parameters and based on a force-oriented layout algorithm, node positions are iteratively updated to converge, and layout constraints and a target function are applied to generate a two-dimensional network logic topological graph; classifying and processing the network elements to generate a visual result; dividing a network space in a multi-scale manner according to geographic space levels, and constructing a three-dimensional topological structure in combination with a hierarchical mapping strategy; and finally, fusing building three-dimensional model data, simulating facility distribution and machine room topological association, applying a symbolic constraint and flow association target, generating an interactive interface and a multi-dimensional result, and outputting network topology geospatialization information.

The invention relates to a cyberspace security protection capability index measurement method, apparatus and device, and a medium. The method comprises the steps of calculating a static protection strength reference based on historical data and an index system; then collecting and analyzing real-time threat intelligence, and quantizing to generate an attacker intention vector; inputting the static protection intensity reference and the attacker intention vector into a non-cooperative game model for deduction, simulating attack-defense interaction and solving an equilibrium strategy, thereby identifying a protection vulnerability set and obtaining an optimal defense strategy suggestion; dynamically correcting the membership degree of each safety index according to the vulnerability point and the intention of an attacker; and fusing the dynamic membership degree and the optimal defense strategy, and performing weighted synthesis to obtain a final safety protection capability index. According to the method, the prospective and dynamic measurement of the network security protection capability is realized by introducing attacker intention analysis and game deduction.

The invention discloses a hierarchical Mama and multi-modal fusion unknown network threat detection method based on physical boundary perception, and relates to the technical field of network space security and artificial intelligence crossing, and the method comprises the steps: carrying out physical boundary perception preprocessing on original network traffic, and extracting byte modal features and statistical modal features; constructing a layered Mama encoder, extracting a load semantic feature of each data packet, and extracting a time sequence interaction feature between the data packets; inputting the byte modal features and the statistical modal features into a multi-modal gating fusion module to generate an anti-confusion stream representation vector; and based on the mask reconstruction pre-training model, calculating a reconstruction error of the input flow. According to the method, the physical level of a network protocol is strictly aligned, stable detection performance can still be kept, extremely high reasoning speed and low video memory occupation are kept, the limitation that a traditional closed set classifier can only recognize known attacks is broken through, and the blank of an efficient flow detection model in the field of open set detection is filled up.

The invention relates to the technical field of network space security, in particular to a block chain consensus protocol privacy protection method based on a right proof. Through a node capability construction mechanism, hardware acceleration data cache data, environment electromagnetic data and load data are acquired, a comprehensive influence value is calculated through an exponential attenuation correction model and the like, a real-time capability level is analyzed and output, and then a complete set of lightweight multi-level privacy policies and other multi-level privacy policies are matched, so that low-end nodes are prevented from being overloaded due to insufficient computing power, and the security of the low-end nodes is improved. High-end nodes are enabled to exert the computing power advantage, and the adaptability of privacy protection and node computing power is improved; according to the method, through an adaptive scheduling mechanism, attack protection monitoring, intermediate sensitive data and key data monitoring, comprehensive risk value calculation, risk level determination, low-risk maintenance generation, intermediate-risk degradation generation and other adjustment schemes, privacy execution is dynamically optimized, and the overall efficiency is balanced while privacy security is guaranteed.

The application discloses a Chinese error correction method and system based on multi-agent cooperation and retrieval enhancement, and belongs to the technical field of network space security. The method comprises the following steps: 1, based on a large language model and an agent prompt engineering, a plurality of Chinese error correction expert agents and one Chinese error correction judge agent are constructed; 2, an error correction text knowledge base is built, corpus in the knowledge base is converted into a vector database, the Chinese error correction expert agent and the judge agent are accessed, and are used for information retrieval of the Chinese error correction expert agent and the judge agent, so that the context understanding capability is enhanced, and a retrieval enhancement generation function is realized; 3, detection prompts, discrimination prompts and feedback prompts are designed, a multi-agent closed-loop feedback module is constructed, information cooperation of each agent, step-by-step reasoning, result verification and iterative optimization are realized, and a Chinese error correction task is completed. The method can realize generalization error correction under the condition that the training data is limited, and improve the adaptability of the model to unknown errors.