815 results about "Node count" patented technology

The invention belongs to the field of fault monitoring, and particularly relates to a fault self-healing method and system under an intelligent operation and maintenance global P2P architecture, and the method comprises the steps: constructing a global operation and maintenance monitoring double-layer node network which comprises a basic monitoring subnet and a collaborative management and control subnet; the basic monitoring subnet collects platform operation data in real time according to a preset inspection strategy, determines an abnormal root node and an association node through causal association analysis, generates an early warning signal and uploads the early warning signal to the collaborative management and control subnet; after the collaborative management and control subnet receives the early warning, a fault self-recovery strategy is started, resources are dynamically allocated, abnormal node data batch migration and recovery are executed, and a self-recovery result is fed back to the basic monitoring subnet to trigger loop check until abnormity is eliminated; according to the method, real-time monitoring and intelligent analysis of operation and maintenance data are realized through cooperative work of the double-layer node network, and disaster tolerance and load adaptive adjustment capability of the system are effectively improved in combination with an automatic fault self-healing mechanism and data difference migration.

The invention provides a real estate full life cycle intelligent management method and system based on Internet of Things perception, and the method comprises the steps: distributing a unique hardware identifier of equipment, combining a lightweight digital certificate with a pre-shared key to achieve two-factor authentication, and storing a root certificate in a cloud end; an access authentication process is optimized, an identity label and a certificate signature request need to be submitted when a node accesses for the first time, a temporary session key is generated after the gateway verifies, and abnormal access is immediately isolated and reported; data integrity verification is implemented, and after a sensor collects data, an abstract is generated and data encryption transmission is carried out; after decryption, the receiving end recalculates the abstract for comparison, and if not, the abstract is discarded and nodes are marked to be abnormal; a data anomaly detection model is constructed, energy consumption fluctuation is determined based on historical data, a frequency baseline is collected, data reasonability is monitored in real time, secondary authentication is triggered when the data is abnormal, and node data access is suspended if the data fails; a hierarchical key system is established, hardware is solidified by a root key, session keys are alternated for 24 hours, and data encryption keys are dynamically derived and distributed through encryption channels.

The invention discloses an electronic lead seal data management method based on a block chain, relates to the technical field of internet data services, and aims to solve the problems of node data asynchronization, state updating failure and potential safety hazards of an intelligent contract in the prior art. The data consistency is ensured through a distributed clock synchronization algorithm and a version number verification mechanism; the reliability and fault tolerance of state updating are improved by adopting an improved Byzantine fault-tolerant algorithm and an asynchronous two-stage transaction submission mechanism; data integrity and traceability are guaranteed through a state machine detection mechanism and transaction log records; the security and maintainability of the smart contract are improved based on an automatic code auditing algorithm and a proxy contract mode of a graph neural network; real-time data monitoring and exception repair are realized by using an under-chain auditing mechanism and a cross-chain data consistency verification protocol; according to the invention, the data consistency, the system safety and the reliability of electronic lead seal data management are obviously improved.

The invention relates to the technical field of execution traceability, in particular to a block chain-based flood storage and detention area ecological compensation execution traceability processing method and system, and the method comprises the following steps: extracting remote sensing data to generate a delay factor list, recognizing ecological trigger nodes to generate a calibration node data set, constructing a reconciliation path structure, and extracting boundary nodes; and generating a permission mapping and responsibility signature block, and writing the permission mapping and responsibility signature block into a responsibility tracking chain structure area. According to the method, logic reconciliation is carried out on compensation and payment behaviors through timestamps and hash records, node tracking and continuity inspection of fund payment paths are realized, operation authority matching is carried out in combination with a contract authority table after time jump is identified and boundary nodes are calibrated, and a responsibility tracking structure chain is constructed step by step; according to the method, responsibility binding of fund behavior nodes and whole-process signature traceability and chain traceability of a transfer behavior are effectively realized, the authenticity discrimination capability of compensation execution and the operation responsibility traceability capability are enhanced, and the verification precision and process integrity of a fund execution path are improved.

The invention discloses a multi-split cross-platform distributed monitoring system state data acquisition method, which belongs to the field of system operation and maintenance and data monitoring, and comprises the following steps: acquiring node data through a data acquisition module arranged on a distributed monitoring system node, and introducing a self-adaptive acquisition strategy into the data acquisition module, adaptively adjusting the node data acquisition frequency according to the resource utilization condition; a node data monitoring strategy is formulated according to user requirements, and the monitoring strategy is optimized by using a hybrid monitoring algorithm; a Delta compression technology is utilized to compress node data to obtain a compressed data packet, the compressed data packet is transmitted, and a multi-level transmission optimization mechanism is introduced in the transmission process of the compressed data packet. According to the method, the technologies such as message bus publishing and subscribing, a customized data acquisition strategy and data flow optimization are integrated, so that consistent state monitoring can be realized among different platforms and hardware architectures, and the efficiency and reliability of system operation and maintenance are greatly improved.

The invention discloses a communication fault diagnosis method and device based on a power system, equipment and a medium. The method comprises the following steps: S1, collecting and outputting multi-source data; s2, preprocessing the multi-source data; s3, performing topological constraint attention mechanism weighting processing on the data matrix; s4, performing dynamic threshold calculation on the fusion feature matrix; s5, positioning the abnormal node list; s6, calculating to obtain a posterior probability; s7, matching the rule base to obtain a standardized disposal instruction; according to the method, the communication traffic is collected through the compressed sensing collection algorithm, the line impedance parameters are integrated into the observation matrix in combination with the topological parameters, key node data are collected preferentially, and the effective data acquisition rate is increased; the accuracy of switch false alarm identification is improved by outputting a time-space unified fusion feature matrix; the computing resources are focused to the abnormal nodes and neighborhoods thereof through the abnormal node mask matrix, so that the positioning precision of the aged equipment is improved, and the judgment omission rate caused by switch aging is prevented from being improved.

The invention provides a computing power service dynamic resource allocation method and system applied to AI model training. The method comprises the following steps: firstly, collecting real-time computing power resource use data (including computing node load, memory occupation and data transmission delay) and model training state data (including training task stage identification, model parameter updating frequency and training data batch processing progress) in AI model training; generating a computing power resource demand association feature set, constructing a computing power resource dynamic allocation decision model including resource allocation priority judgment, adjustment amplitude calculation and scheduling opportunity selection units based on the set, and outputting a computing power resource allocation scheme (including computing node number, memory capacity and data transmission bandwidth adjustment instructions) through the model. Resources are scheduled according to the scheme, and new data are collected to update the feature set, so that dynamic and accurate allocation of computing power resources is realized, and the resource utilization rate and the training efficiency are improved.

The invention provides a computing power center resource centralized self-adaptive control scheduling system and method, and the method comprises the steps: S1, a user submits a computing power training task through an API, and sends the task to a DAG task composer, and forms a task list to be scheduled; s2, the intelligent monitoring analysis module is used for continuously collecting node data and calculating real-time CS and HS scores of each node; s3, analyzing the dependency relationship type data by a task composer, and calling a dynamic programming algorithm in an adaptive scheduler to make a decision; s4, the adaptive scheduler performs dynamic task distribution according to the task list of the DAG task orchestrator and the intelligent monitoring analysis module, and S5, the distributed runtime and environment management module is used for uploading the prepared Docker mirror image to the distributed storage; and S6, through the unified communication adaptation layer, the mirror image is pulled from the distributed storage on the target node and the container is started, the user does not need to concern the difference of the operating system in the whole process, and efficient and intelligent adaptive scheduling is realized.

The invention belongs to the technical field of cloud side-end collaborative data processing, and discloses a cloud side-end collaborative heterogeneous data fusion processing system, which establishes dynamic association of data features, resource states and service priorities through a scheduling module, and adopts a reinforcement learning algorithm combining DQN and an attention mechanism to perform fusion processing on heterogeneous data. Task allocation weights are dynamically adjusted with the minimum delay, the highest precision and the maximum business value as targets; the sensing module and the preprocessing module complete lightweight processing such as feature extraction and noise reduction redundancy elimination at the edge end, and the cross-node data transmission amount is reduced. The complex fusion task dispatches cloud computing power support, and reserves resources in advance by means of an LSTM-based task urgency predictor; the cross-modal fusion module designs a double-stage fusion algorithm, in the first stage, cross-modal semantic association is enhanced through multi-head self-attention, and in the second stage, the fusion weight is adjusted according to the reliability coefficient of the sensing module.

The invention relates to the technical field of data storage, in particular to a distributed data storage and access processing method. The method comprises the following steps: presetting a guide configuration file in a distributed storage cluster to carry out identity authentication and communication link establishment on each storage node and a cluster management center so as to generate a security data interaction channel; the data sending end equipment sends a data storage request to the cluster management center through the secure data interaction channel and issues data fragments to proper storage nodes; the data receiving end equipment sends a data access request to the cluster management center through the security data interaction channel, generates an access path and feeds back the access path to the data receiving end equipment; the corresponding resource use condition in each storage node is monitored in real time, and when a preset threshold value is reached, the resource is reallocated to other storage nodes according to a preset data migration strategy and a load balancing algorithm based on a resource use insufficiency result. According to the invention, efficient storage and rapid access of data in a distributed environment can be realized.

To provide a node that can maintain the number of nodes and continuity of transactions even when a specification change occurs in a distributed ledger program that causes inconsistencies in consensus formation.SOLUTION: Provided is a node for a first distributed ledger system that manages a first distributed ledger, comprising: a control unit for, together with some other nodes of the first distributed ledger system, generating a second distributed ledger and registering contents of a plurality of blocks included in the first distributed ledger as blocks of the second distributed ledger by performing consensus formation in order from the lowest block height; and a communication unit for, after the contents of all blocks included in the first distributed ledger have been registered in the second distributed ledger, synchronizing the blocks registered in the second distributed ledger with the remaining other nodes of the first distributed ledger system.SELECTED DRAWING: Figure 1

The invention provides a gray-based continuous deployment method, device and equipment, a medium and a program product, which can be applied to the technical fields of artificial intelligence and software development. The method comprises the following steps: acquiring a first deployment strategy corresponding to a first deployment task, wherein the first deployment strategy comprises the number of nodes participating in gray deployment and the proportion of nodes of different risk levels; selecting nodes from the node pools of different risk levels according to a first deployment strategy to obtain a target batch of nodes; executing the first deployment task in the target batch node; wherein the node pools with different risk levels are obtained by continuously updating according to a second trigger logic, and the updating process of the node pools with different risk levels comprises the following steps: dividing the plurality of nodes into the node pools with different risk levels based on the node features of the plurality of nodes by using a trained risk prediction model, wherein the node features of the plurality of nodes are obtained by continuously updating according to the first trigger logic.

The invention discloses an optimized node selection method based on a BGP-iSec protocol partial deployment scene, which comprises the following steps of: in a first stage, calculating three static topological characteristics of nodes according to a key node selection algorithm, weighting the three static topological characteristics and AS node levels in proportion, calculating a mixed weight of the nodes, and selecting the three static topological characteristics according to the mixed weight; preliminarily determining candidate nodes; and in the second stage, a greedy algorithm is adopted, the node with the maximum gain for the current path coverage rate is gradually selected from the candidate nodes according to the priority and added into the to-be-deployed set until the path coverage rate is stable or the number of the nodes reaches the specified deployment cost. According to the method, the key AS node selection strategy is formulated, so that the defense effect of the path manipulation / prefix hijacking attack is optimized under the limited deployment cost.

The invention discloses a metadata management method based on ES cold and hot data read-write separation, and relates to the technical field of metadata management, comprising: marking an SSD disk as a hot node, and marking an SATA disk as a cold node; assigning label parameters in a node starting command to enable the nodes to join according to roles; creating an index template to divide a hot index and a cold index, writing a default route to a hot node, and reading an automatic route of a coordination node; setting an automatic index rolling strategy, writing an active index into a hot node, automatically establishing a new index when a threshold value is reached, and preferentially allocating resources to the hot index; an ILM migration rule is configured, and the overdue index is automatically migrated to a cold node; writing real-time metadata into a hot node, migrating overdue metadata to a cold node regularly or manually, and deleting cold node data according to a strategy; real-time data are read from the hot nodes, historical data are read from the cold nodes, and on-demand access is achieved. The performance, reliability and resource utilization rate of the ES cluster are remarkably improved on the premise that the cost is controllable.

The invention relates to the technical field of anti-interference of intelligent monitoring data acquisition of a new energy power station, in particular to a self-adaptive anti-interference system for data acquisition of an intelligent auxiliary monitoring system. A windowing judgment mechanism is formed based on cross-node stage consistency; thirdly, defining participation paths of the data units with different stage attributes in cross-node association analysis, and aligning and screening multi-node data through data consistency management; when the stage consistency window meets a release condition, controlled stage regression processing is performed on the data unit in the unstable stage, so that the data acquisition node gradually enters a unified stage attribute; according to the method, the instantaneous abnormal processing of the interference influence is converted into staged and structured management, and the continuity, consistency and reliability of data acquisition and analysis in a complex power monitoring scene are greatly improved.

The embodiment disclosed by the application belongs to the technical field of computing, and particularly relates to a method and device for performing stencil calculation and a computing device. The method comprises the following steps: acquiring a plurality of node data included in a calculation region, and grouping the plurality of node data into a data vector. According to at least one calculation condition included in the stencil calculation, the plurality of node data are subjected to judgment processing, and a judgment result corresponding to each calculation condition is obtained for each node data. For each calculation condition, the judgment result corresponding to the calculation condition is grouped into a judgment result vector, and the judgment result vector is obtained. Based on the data vector and the judgment result vector corresponding to each calculation condition, a calculation operation corresponding to each calculation condition in the stencil calculation is performed, and a calculation result of the stencil calculation is obtained. The application can convert the stencil calculation into vector calculation, and can improve the efficiency of the stencil calculation.

Provided are harmless attack verification and analysis method and system. The method includes: acquiring multi-source security data, determining attack cases in the data to construct an attack knowledge graph; constructing a scenario-based attack flow chart based on the graph, and formulating a harmless attack task based on the chart; determining a number of paths and a number of nodes of attack paths in the harmless attack task, and decomposing the harmless attack task into sub-tasks based on the number of paths and the number of nodes; acquiring path information of each attack path, determining a complexity of the attack path based on the path information, and determining attack priorities of the sub-tasks based on the complexities of the attack paths; and determining an attack sequence of the sub-tasks based on the attack priorities, and performing simulated attack verification on a protection boundary by the sub-tasks in the attack sequence.

The invention relates to a distributed compensation control system and method for a power distribution network, and the system comprises a distributed multi-source module which is used for collecting the node data, equipment operation state data and system global operation data of each key node in real time; the edge decision-making module is used for acquiring local node data of the local key node and generating a plurality of corresponding initial compensation instructions of the distributed compensation equipment according to the local node data; the collaborative optimization module is used for receiving and fusing all the preliminary compensation instructions, the equipment operation state data and the system global operation data, calculating through a unified nonlinear comprehensive evaluation function to obtain a system comprehensive operation index, and generating a global collaborative optimization instruction; and the control execution module is used for distributing the global collaborative optimization instruction to the corresponding edge decision module so as to drive the plurality of distributed compensation devices to execute collaborative compensation operation. Preventive maintenance and fault early warning are achieved, the service life of equipment is effectively prolonged, and the overall reliability of the system is improved.

The invention discloses a pluggable Kubernetes cluster deployment method, a medium and equipment, and solves the problems of rigid deployment process, serious component coupling and low fault recovery efficiency in the prior art. The method comprises the following steps: receiving user module label combination and node resource information through a visual interface, and carrying out structured storage as a deployment context object; after the object is analyzed, a dependent support module is automatically identified and added, a cluster architecture mode is dynamically determined based on the number of Master nodes, and a DAG of a deployment task is constructed; module deployment is executed according to the DAG sequence, and only the fault module and downstream dependency thereof are retrying when failure occurs. According to the method, plug and play of the component is realized through label driving, a deployment process is dynamically arranged based on DAG, intelligent framework cutting is supported, accurate fault isolation and local recovery are realized, time consumption of component switching and resource occupation are greatly reduced, and deployment flexibility and operation and maintenance reliability are remarkably improved.

The invention relates to the technical field of document management, and relates to a legal document key point automatic extraction and association method and system.The new legal document data of a target case is obtained, a preset document fingerprint algorithm is used for automatically comparing the new legal document data with a historical reference document, differential text data fragments are extracted, and the text data fragments are extracted; the method comprises the following steps of: performing deep semantic analysis on different text data fragments to generate candidate triple data containing entities and relationships, intelligently mapping the candidate triple data into a pre-constructed case knowledge graph, quickly positioning graph anchor nodes with data change, and then, performing data change on the basis of change types of the different text data fragments. And executing atomic updating operation on the map anchor node, including updating the dynamic mark of the node state attribute, and finally generating the change source node data with the latest state identifier. The problems that the current legal case data updating depends on manpower, is low in efficiency and is easy to make mistakes are effectively solved.

The invention relates to the field of distributed storage, and discloses a distributed memory system based on erasure codes and a dynamic repair mechanism. Comprising a data node cluster, a metadata management cluster, a bandwidth token management module, a dynamic repair composer and a repair agent node. And the data node cluster divides the object into strip generation data fragments and verification fragments and dispersedly stores the data fragments and the verification fragments according to a placement rule. And the metadata management cluster maintains object stripe mapping, a fragment position set, erasure code protection configuration and a version epoch, and records a repair intention log. And the bandwidth token management module grants repair execution permission according to the fault domain budget. The dynamic repair composer generates availability repair and durability repair tasks, the temporary check fragments are written in when the number of the available check fragments is lower than a threshold value in the availability stage, and the configuration fragments are recovered and the temporary fragments are recovered in the durability stage. And the consistency of version epochs is verified before the agent node is repaired and written back, and if not, recalculation is carried out.

The invention discloses an efficient double-link block chain consensus method and device, equipment and a storage medium, and belongs to the field of information security. The method comprises an initialization stage, an input stage, a coding stage, a broadcasting stage and a consensus stage. According to the invention, the parallel design of a coding broadcast link and a consensus confirmation link is adopted, so that the occurrence frequency of transaction congestion is reduced; the integrity required by the asynchronous Byzantine consensus process is realized through three-step broadcasting, the updating problem of the post-quantum component does not need to be considered, and the security is higher; an asynchronous network and a Byzantine opponent model are adopted, so that malicious node attacks which do not exceed the total number of nodes can be tolerated on the premise of not setting a network time delay upper limit; in the encoding stage, the generation of the redundancy share through the RS erasure code can be realized by adopting preprocessing, the data redundancy expansion multiple in the broadcasting stage is far smaller than that of the same type of protocol, and the original transaction can be recovered by collecting part of effective shares in the consensus stage, so that the system time delay is effectively reduced, and the system throughput is improved.

A distributed resource automatic matching method based on credibility grading comprises the steps that multi-dimensional operation performance indexes of edge nodes of a distributed power grid are collected, a node initial credibility evaluation value is generated through quantitative calculation, and a node credibility initial model is constructed; capturing a node operation behavior change trend based on a sliding time window, adjusting each performance index evaluation weight according to a node data fluctuation amplitude, calculating a dynamic trust degree evaluation value by combining a dynamic weight mechanism, dividing the nodes into at least two trust levels according to a preset threshold value, and mapping the nodes of different levels to corresponding physical regions; when a regulation and control task is responded, candidate resources are preferentially matched from high-trust-level nodes, and resource scheduling deployment is completed in combination with task urgency and node electrical characteristic related factors; after the task is completed, the dynamic credibility evaluation value of the node is corrected based on the actual execution effect evaluation result of the node, the corrected evaluation value is used for updating the node credibility model and subsequent resource matching, and a closed-loop feedback mechanism is formed.

A storage system distributes data across multiple nodes, where the data is stored with partial redundancy to meet a predetermined reliability level. The system monitors a number of nodes in the storage system to determine whether a number of nodes changes. Responsive to determining that the number of nodes changes, the system determines whether reliability level of the storage system exceeds the predetermined reliability level by a threshold. Responsive to determining that the reliability level of the storage system exceeds the predetermined reliability level by the threshold, the system adjusts a data storing scheme of the storage system to reduce redundancy and increase available storage capacity in the storage system, while maintaining the predetermined reliability level.

The invention discloses a blockchain-driven separate account data management system and method, and relates to the technical field of blockchains, and the method comprises the steps: achieving data transparency and safety through a hybrid architecture, and achieving multi-node data synchronization and tamper-proofing through a distributed account book technology. The intelligent contract automatically executes the account division rule, and the problems of low fund distribution efficiency, high second-definition risk and the like in e-commerce, supply chain and other scenes are solved. The performance of the system is improved through load balancing and caching technologies, and the data consistency is ensured in combination with a block chain consensus mechanism. And the dynamic account division and cross-border settlement capabilities are optimized. The technical problems that in a cross-e-commerce platform multi-level account division scene, an existing account division system cannot dynamically recognize rule conflicts and process abnormal transaction chains in real time, and is lack of a cross-chain collaborative error correction mechanism are solved, dynamic aggregation and abnormal self-healing of multi-platform account division rules are achieved, and the user experience is improved. The technical effects of improving the timeliness of account division and the recognition accuracy of abnormal transactions are achieved.

The invention discloses a service agent evaluation method and device, equipment, a storage medium and a program product, and relates to the technical field of agents, and the method comprises the steps: obtaining node data generated by each function node of a service agent when the service agent executes a task; extracting industry features from the node data, and determining an industry field to which the business agent belongs when executing the task based on a matching result of the industry features and a preset industry feature library; screening from a preset evaluation index library to obtain an evaluation index set matched with the industry field; and evaluating the business agent according to the evaluation index set to obtain an evaluation result of the business agent in the industry field. According to the invention, the problems of lack of industry pertinence and inaccurate evaluation in the prior art are effectively solved.

The invention provides a federal incremental modeling method based on dynamic probability optimization in a heterogeneous edge environment. The method comprises the following steps that equipment in each factory counts class density vectors after noise adding and uploads the class density vectors to a server; the server selects a plurality of candidate sets, calculates the distribution offset coefficient of each candidate device set, and selects the minimum set to participate in federal training; equipment training is selected, and optimal parameters are uploaded; the server aggregates and distributes global model parameters; the device calculates a newly-added hidden layer output according to the global parameters and uploads the newly-added hidden layer output to the server; the server performs multi-index evaluation according to the performance of the equipment, optimizes the selected probability of the equipment through an Exp3 algorithm based on multi-index evaluation and fairness constraint, and selects the equipment to participate in training based on the probability; and when the number of hidden layer nodes exceeds the limit or the residual error satisfies the expected tolerance, finishing the training, and distributing to the equipment as a soft measurement model.

The invention relates to the technical field of database synchronization, in particular to a multi-node database synchronization method based on data partitioning, which comprises the following steps of: 1, performing data partitioning and node role allocation; 2, synchronization mechanism configuration, wherein the synchronization mechanism configuration comprises master and slave node copy setting and data initialization and import; 3, incremental data synchronization, asynchronous / semi-synchronous replication and conflict detection; 4, using a protocol to optimize large file and mass data transmission; step 5, performing parallel processing on the fragmented data at a plurality of nodes; and step 6, carrying out redundant backup on the data fragments at a plurality of nodes. According to the invention, data division and node role allocation are carried out; and data distribution: routing the change event to a corresponding node according to a fragmentation rule, or broadcasting a write set through a group communication protocol, carrying out redundant backup on the data fragmentation at a plurality of nodes, and finally checking the consistency of the fragmentation data regularly, so that the method has the advantages of being capable of realizing efficient synchronization of the multi-node database, giving consideration to the performance and the consistency and the like.

The invention discloses a chemical industry park safety risk monitoring system and method based on the Internet of Things, and the system comprises a node data obtaining unit which is used for receiving node output data correspondingly monitored by a monitor in the range of a chemical industry park, and the installation address information corresponding to a node; the node link monitoring unit is used for forming a plurality of data links according to the nodes and the Internet of Things terminal, and dividing the data links into a plurality of node link sub-regions according to directions; the risk searching unit is used for searching the highest risk data chain in each node link sub-region, comparing the highest risk data chains obtained in all the node link sub-regions, and preferentially processing the highest risk data chain of the first priority; and the early warning processing unit is used for triggering early warning information according to the current situation of the highest risk data chain of the first priority and formulating an optimization strategy, and is used for solving the technical problem that an accident source cannot be quickly and accurately determined when an existing platform obtains workshop or equipment accident occurrence information.

The invention provides a mixed retrieval method, system and device based on multi-algorithm fusion and a storage medium, and the method comprises the steps: obtaining various types of document data, preprocessing the document data, and obtaining plain text data; segmenting the plain text data into a plurality of pieces of block node data, and generating a plurality of derivative problems for each piece of block node data by using a large language model; carrying out vectorization coding on the block node data and the derivation problem, and storing the block node data and the derivation problem in a vector database; receiving a user query question, and generating a plurality of related preset questions for the query question by using the large language model; converting the query question and the preset question into a vector form, executing keyword retrieval and vector retrieval, and obtaining corresponding candidate results in a vector database; and carrying out weighted fusion on the candidate results through a weighted reciprocal rearrangement algorithm to obtain most relevant retrieval data. According to the method, the efficiency of information retrieval and the correlation of results can be improved, and the retrieval requirement of a user in a complex semantic scene is met.