127 results about "Graph algorithms" patented technology

The invention provides a low-airspace public route construction method, device and equipment, and relates to the technical field of aircraft airspace route planning. The method comprises the following steps: determining a plurality of target candidate nodes corresponding to a city based on ground traffic topological characteristics of the city corresponding to a low airspace and spatial distribution characteristics of low-risk ground features; performing triangulation on the plurality of target candidate nodes to obtain an initial edge set; wherein the initial edge set comprises a plurality of initial edges; performing sparse processing on the plurality of initial edges based on a greedy tensor graph algorithm, and determining a plurality of target edges from the plurality of initial edges; wherein the length of the shortest path between the two nodes of the target edge is greater than the greedy tensor coefficient times of the Euclidean distance between the two nodes; and constructing a low-airspace public route based on the plurality of target edges. By adopting the technical scheme provided by the invention, the utilization efficiency and safety of the low airspace can be effectively improved by constructing the unified low airspace public route.

The invention relates to the technical field of water supply network monitoring, and discloses a water supply network leakage monitoring and positioning system and method based on big data analysis, and the method comprises the steps: obtaining continuous audio stream data and network topological graph data collected by a distributed sensor, and calculating background baseline parameters; performing multi-scale wavelet transform decomposition on the current audio frame, calculating an energy entropy feature and generating a single-frame abnormal score; applying a cumulative sum control chart algorithm to output a cumulative abnormal index; generating an enhanced anomaly score through a graph attention network; and performing leakage judgment and positioning based on the abnormal scores of the multiple sensors. The technical problem that weak leakage signal detection sensitivity is low is solved.

The invention relates to a knowledge graph construction and retrieval optimization method based on water affair data, and the method comprises the steps: taking the unified knowledge expression of a water affair object and an event thereof as a target, determining a water affair business field and business range, and constructing a water affair knowledge graph conceptual model covering a static knowledge ontology and a dynamic event ontology; performing knowledge extraction on the obtained multi-source heterogeneous water affair data, converting the data into a unified triple form, and constructing a data layer of a water affair knowledge graph; and on the basis of the data layer, node, path and region query is performed on the knowledge graph by adopting a graph algorithm, and semantic enhancement retrieval and reasoning are performed in combination with a large language model. By constructing the water affair knowledge graph and introducing knowledge extraction and retrieval enhancement capability of a large language model, dynamic modeling and semantic reasoning of water affair objects, events and relationships thereof are realized, high-risk area identification, emergency path planning and emergency early warning are supported, and the knowledge utilization rate and intelligent decision level in the water affair field are improved.

The invention discloses a multi-dimensional graph series-parallel and visualization method for telecommunication network fraud cases based on a graph database. According to the method, intelligent series-parallel analysis across time-space cases is realized by constructing a knowledge graph containing various entities such as cases, personnel, accounts and equipment and deeply fusing a hard association matching algorithm and a soft similarity graph algorithm. The system has the functions of white list filtering, time sequence behavior analysis, NLP information extraction and the like, and can effectively identify criminal gangs, track fund flow directions and mine crime modes. And finally, dynamic visual display is carried out through a plurality of interactive views such as a force-oriented graph, a time axis and a mulberry-based graph, and clue intelligent pushing and report automatic generation are supported. According to the method, the case association mining depth, the serial-parallel analysis efficiency and the interpretability of a complex network are remarkably improved, and an efficient technical support is provided for the public security organization to fight against telecommunication network fraud.

The invention belongs to the field of robot positioning and mapping algorithms, and provides a visual SLAM method and device for point-line feature fusion and a medium, and the method comprises the following steps: preprocessing an image, and obtaining point features and line features; matching the point features and the line features to obtain matched point features and matched line features; obtaining the initial pose of the camera through the matched point features and the matched line features, and constructing map points; constructing a three-dimensional map straight line through the matched line features; and optimizing the three-dimensional map points and the three-dimensional map straight lines. Complementary advantages are formed through extraction, matching, mapping and optimization processes of tight coupling point features and line features, and the fundamental problem of insufficient features in a weak texture environment is effectively solved.

The invention discloses a road structure performance catastrophe early warning method and system. The method comprises the following steps: S1, synchronously acquiring data by using a three-dimensional ground penetrating radar and a deflectometer; s2, introducing a dielectric constant correction signal, quantifying a volume defect, and taking the volume defect as a physical constraint inversion mechanical parameter; s3, fusing the multi-dimensional parameters to construct a damage potential index, and predicting the residual life by using a graph space-time neural network; and S4, on-line mutation detection is carried out based on a Bayesian and cumulative sum control chart algorithm, and the weight of the model is updated through a treatment feedback closed loop. According to the method, physical-data dual-drive modeling is adopted, the problems of one-sided detection means, poor model interpretability and parameter solidification are solved, and accurate evaluation of road performance degradation and continuous evolution of the model are achieved.

The invention discloses a communication network operation and maintenance fault tracking method and system. The method comprises the following steps: constructing a communication network operation and maintenance knowledge graph; optimizing the knowledge graph by using a graph algorithm and a communication domain rule; utilizing a PCST algorithm to generate an optimization sub-graph of which the relevancy with the fault query is greater than a first threshold value, and converting the optimization sub-graph into a first natural language text; performing integrity judgment on the first natural language text by utilizing a large language model; when the first natural language text is judged to be incomplete, entities and relationships which do not appear in the optimized sub-graph are extracted from the first natural language text to serve as supplementary query words; iteratively generating a supplementary sub-graph on the optimized knowledge graph based on the supplementary query word to obtain a complete natural language text; performing fault reasoning on the complete natural language text by using a large language model to obtain a reasoning answer; and constructing a four-layer reverse tracing link according to the reasoning answer, the first natural language text and the second natural language text. And the communication network fault tracking precision is improved.

This invention relates to a multi-agent orchestration and execution method supporting complex cyclic graph structures, comprising: S1: receiving workflow configuration and constructing a directed graph data structure; S2: performing a depth-first search traversal on the directed graph data structure, detecting and outputting strongly connected components in the directed graph data structure; S3: abstracting strongly connected components containing multiple nodes into super nodes, and keeping single-node strongly connected components as ordinary nodes, thereby constructing a directed acyclic graph containing super nodes; S4: performing topological sorting on the directed acyclic graph to generate a hierarchical execution plan containing multiple execution layers; S5: scheduling execution items according to the hierarchical execution plan, and when an execution item is the super node, starting a loop executor for iterative execution. This method, by constructing a multi-layered technical system, integrates graph algorithm innovation, unified context management, abstraction layer design, metadata-driven techniques, etc., to achieve automated orchestration and execution of multi-agent workflows supporting cyclic graph structures.

This invention relates to the field of orchard navigation technology, specifically a 3D method for orchard operations based on a grid map. The method includes: constructing an orchard point cloud map using a mapping algorithm; filtering the acquired 3D point cloud map and constructing a triangular grid map of the orchard, embedding the orchard's GPS information into the triangular grid map; based on the triangular grid map with GPS information, performing ICP registration between the point cloud data scanned by the LiDAR and the points on the triangular grid map using the RMCL 3D positioning algorithm, outputting the 3D pose of the operation robot in the orchard; updating the 3D pose of the operation robot in the orchard using GNSS information; planning the navigation path of the operation robot using grid navigation based on the GPS information of the target point; and controlling the operation robot to dynamically avoid obstacles and navigate to the target point. This invention enables precise positioning and autonomous navigation of the operation robot in hilly orchards, allowing it to complete tasks such as orchard inspection, plant protection, harvesting, and transportation.

The invention provides a casting X-ray image defect detection method based on an SAHI algorithm, and the method comprises the steps: carrying out the defect marking of an obtained casting X-ray image, and obtaining an initial data set; slicing the initial data set based on an overlapping image cutting algorithm, and adjusting and marking to obtain an image cutting data set; constructing a deep learning network model, performing pre-training on the deep learning network model based on the initial data set to obtain a pre-training model, and performing fine tuning on the pre-training model based on the cut graph data set to obtain a target network model; segmenting the X-ray image of the casting to be detected into a plurality of overlapped image blocks based on an SAHI algorithm; and performing defect detection on the plurality of image blocks based on the target network model to obtain a plurality of detection results, mapping the detection results back to an image coordinate system of the X-ray image of the casting to be detected, and fusing the detection results to obtain a defect detection result. According to the method, the calculation cost during model training is reduced, the training efficiency is improved, and the detection precision of the internal defects of the casting is improved.

This invention provides a rapid depression detection method based on a weighted degree transfer network, comprising: acquiring and preprocessing the EEG signals of a subject to obtain a one-dimensional EEG time series; mapping the one-dimensional EEG time series into a complex network using a weighted horizontal visualization algorithm; extracting degree and intensity sequences; using the deduplicated set of degree values ​​as nodes of the new network; tracing the degree transfer path between adjacent time points; calculating the product of the degree difference and intensity difference between adjacent time points as the edge weights of the degree transfer path, thereby constructing a weighted degree transfer network; normalizing the network to obtain a transition probability matrix; calculating its Shannon entropy and MPR statistical complexity as a joint feature vector; and using a machine learning classifier to identify the depression state. This invention does not require presetting parameters such as embedding dimension, and by fusing the dynamic difference features of degree and intensity, it can quickly and accurately capture abnormal nonlinear dynamic patterns in the EEG signals of depression.

The invention discloses a refrigerating machine room operation and maintenance method based on a knowledge graph. The method comprises the steps that a directed characteristic graph composed of standard abnormal reference nodes and intermediate nodes is constructed; collecting an on-site abnormal image and extracting a multi-dimensional feature vector of the on-site abnormal image; adding the feature vector as a new node into a graph, and calculating a shortest multi-hop path between the feature vector and the candidate reference node based on a graph algorithm; and judging an abnormal type according to the reference node with the shortest path, and executing reference node feature correction or generating an intermediate node and a new reference node according to the similarity interval and the sample accumulation condition, thereby dynamically evolving the graph structure. The system comprises an atlas management module, a path calculation module, a result management module and a field acquisition robot. According to the method, precise classification and map self-evolution of the on-site variation abnormal images are realized, and the accuracy and adaptability of intelligent operation and maintenance of the refrigerating machine room are improved.

The invention relates to the technical field of networks, in particular to a network data packet analysis system, and provides a system for network data packet analysis, which comprises a protocol decoding engine, a graph optimization engine and a graph traversal engine, in a protocol decoding engine, a plain text concise language specification is adopted to describe basic logic and logic combination of formal languages, description of logic combination and complex nested structures is supported, high expression ability and good readability are achieved, the design is concise enough, the manufacturing difficulty of decoding scripts and the total number of codes can be greatly reduced, and the decoding efficiency is improved. In the graph optimization engine and the graph traversal engine, grammar logic of an original graph structure processed by the protocol decoding engine is converted into graph logic by adopting a graph algorithm to realize performance optimization during operation, and flexible analysis of different types of protocols is supported for a multi-graph engine architecture; and a graph structure designed in a classified manner comprises a character string graph, a regular graph and the like, and a standardized graph traversal mechanism is realized by unifying a multi-graph traversal engine.

Described are examples for creating a graph model for data obtained from multiple data sources and stored in a common data format. A selection of columns of data stored in the common data format that are to be associated with one another in the graph model can be obtained. Based on the selection, a graph model can be created including metadata that defines a list of vertices corresponding to at least a portion of the columns and a list of edges between two or more of the vertices identifying a relationship between corresponding columns. A graph query or graph algorithm can be performed over the graph model of the data as stored in the common data format without requiring copying of the data to a separate graph database.

The application relates to the field of carrier obstacle avoidance scheduling management, and particularly discloses a multi-mobile carrier obstacle avoidance scheduling method and system based on 5G and graph algorithm fusion, which comprises the following steps: acquiring the running state information of each mobile carrier, and performing data preprocessing on the running state information of each mobile carrier at a 5G receiving front end. After the data preprocessing is completed, the 5G receiving front end transmits the data to a 5G transmitting end, the data is transmitted to a warehouse base station belonging to a cloud end through a radio frequency signal based on a 5G transmitting antenna, is converted into a binary digital signal that can be stored by the cloud end, a signal is transmitted to a graph algorithm module by the cloud end to calculate a shortest path planning, a specific obstacle avoidance instruction is generated by sending the path to an obstacle avoidance instruction generation module, each mobile carrier is issued by the cloud end, and obstacle avoidance scheduling control is completed.

This application provides a network channel routing verification method and apparatus. The method constructs a directed graph based on the directed connection relationships of ports in the routing data, and uses graph algorithms to partition connected components and add audit rules based on data dimensions for verification. This allows for rapid and accurate detection of the connectivity and integrity of channel routes. During the verification process, it does not rely on the segment group order attribute in the routing data, avoiding shallow data quality issues that could prevent verification. Furthermore, it can verify the integrity of the channel carrying structure according to the business channel dimension, quickly aggregating all hierarchical channel routing anomaly audit results. For business channels with abnormal aggregated audit results, a secondary verification is performed to ensure the accuracy and reliability of the verification results. This helps vendors rectify data according to the business channel dimension and achieve rapid results.

The application discloses a kind of complex geological conditions under multi-metal ore blending unit automatic division method, comprising the following steps;Step 1: collecting rock powder test data;Step 2: the rock powder test data collected is preprocessed;Step 3: based on space geometry, find out blast pile borehole boundary point by means of-shape method, form peripheral boundary polygon, and-shape method is used to approach blast pile boundary;Step 4: the blast pile borehole boundary point is expanded outward;Step 5: by means of Voronoi diagram algorithm, blast hole is divided into tessellation polygon on the level, and the range including influence area is formed;Step 6: fine ore blending unit solution division and evaluation, the inverse ratio of the square of the distance of ore block and each ore blending center is weighted, the attribution proportion of each ore block to different unit is calculated, and ore blending unit division is realized.The application realizes the whole process automation from rock powder test data acquisition processing, boundary extraction, unit division to resource quantity estimation.

The invention discloses a graph algorithm-based auxiliary identification method and system for a fraud behavior of medicine purchasing at a gang, and relates to the technical field of medical insurance anti-fraud. The method comprises the following steps: acquiring medical insurance medicine purchase settlement data, and preprocessing the medical insurance medicine purchase settlement data; according to the medical insurance medicine purchase settlement data and the corresponding insured persons, constructing a medicine purchase network structure diagram, taking the insured persons and the medicine purchase events as nodes of the medicine purchase network structure diagram, establishing edges between the nodes, and setting a weight for each edge according to analysis of the medical insurance medicine purchase settlement data; utilizing a community discovery algorithm based on modularity optimization to identify a suspected drug-buying gang according to the drug-buying network structure diagram; and analyzing the attributes of the internal nodes of the suspected gang, and grading the suspected degree of the suspected gang. According to the method, accurate identification of abnormal medicine purchasing behaviors of multiple persons can be realized by using a graph algorithm, so that a powerful auxiliary basis is provided for judgment of fraudulent medicine purchasing behaviors.

The invention provides a medical data credible circulation method and system based on multi-level authority dynamic management, and the method comprises the steps: constructing a multi-mode matching index, and carrying out the feature extraction and vectorization of medical text data, structured data and image data; establishing a three-layer index architecture and an intelligent routing layer; constructing a user-permission-data relation graph by applying a graph algorithm, evaluating a permission use mode in real time, and detecting an abnormal behavior; the abnormal judgment threshold is dynamically adjusted in combination with situation factors such as medical emergency degree, abnormal behaviors are subjected to grading processing, and evidence obtaining information is automatically collected. The invention further comprises a data classification method based on security classification and timeliness and a fine-grained permission control mechanism using the minimum permission principle, and dynamic permission granting and recycling are supported. By balancing security and availability, secure sharing and effective utilization of medical data are realized, and diversified data access requirements are met.

The invention discloses an AI intelligent vulnerability noise reduction analysis method and system, and belongs to the technical field of software security. The method comprises the following steps: acquiring core data required by vulnerability analysis, preprocessing the core data, acquiring standardized data, and constructing a knowledge link for the standardized data; the knowledge links are screened through vulnerability conduction path calculation and low-risk path calculation, the conduction probability of vulnerabilities from an existing component to a service core is calculated through a graph algorithm for the screened knowledge links, a threshold value is set, a high-risk path is screened out, and redundant alarms are preliminarily reduced; and accurately identifying false alarms and repeated alarms of the high-risk path based on the conduction probability by adopting a mixed architecture of an AI-based lightweight small model and a large language model. According to the implementation of the method, the manual intervention cost is reduced, the software supply chain vulnerability prevention and control efficiency and reliability are improved, and more targeted analysis support is provided for software full-life-cycle safety.

The application relates to the technical field of rehabilitation image processing, and provides a rehabilitation treatment dynamic digitalization auxiliary method based on a graphics algorithm and deep learning. The application aligns real-time collected local images with standard anatomical structures in a rehabilitation anatomical atlas library in high precision by constructing and training an image algorithm comparison deep learning model, matches the deep local anatomical tissues to a display terminal through a local image capture post-processing algorithm and a comparison image matching algorithm, so that a therapist can observe the deep local anatomical tissues in real time during operation, and the accuracy of anatomical positioning and the safety of treatment are improved. By constructing and training a rehabilitation clinical treatment algorithm matching learning model, combining collected clinical data and a rehabilitation clinical standard path knowledge base, rehabilitation diagnosis, rehabilitation treatment list preparation and associated anatomical image output are completed, and are synchronously output to a display terminal, so as to adapt to the operation needs of clinical dynamic precise rehabilitation.

The invention discloses an electric energy meter health state assessment method and system based on a knowledge graph and a medium, and relates to the technical field of intelligent operation and maintaining.According to the scheme, multi-source heterogeneous data are integrated through the knowledge graph technology, a panoramic knowledge graph with an electric energy meter as a core is constructed, and on the panoramic relational graph, a knowledge graph is established; and performing embedded representation learning and comprehensive scoring of the health degree by using the constructed health degree assessment model, performing deep risk analysis by using a graph algorithm based on a scoring result, and enabling the result to operation and maintenance personnel in a visual and explainable manner. And the deep insight of the health state of the electric energy meter and the accurate and foresight early warning of the fault risk are realized.

The application discloses an industrial intelligent agent system and method fusing digital twin and a four-element core, belongs to the field of artificial intelligence, and comprises a digital twin perception layer, a four-element fusion core layer, a hybrid reasoning engine and an interpretable report generator.The digital twin perception layer is used for acquiring real-time state data of a physical entity; the four-element fusion core layer integrates OWL ontology modeling, graph database storage, large language model interaction and multi-physical field agent model, and generates a four-element collaborative knowledge processing result; the hybrid reasoning engine performs parallel rule reasoning, graph algorithm mining, large language model reasoning and agent model reasoning and fuses to generate a comprehensive reasoning conclusion; and the interpretable report generator generates a multi-modal report containing a reasoning path and physical evidence.The application realizes interpretable tracing of a decision-making process through four-element collaboration, realizes millisecond-level physical quantity prediction through an agent model, realizes continuous evolution of knowledge through a dynamic evolution module, and significantly improves the reliability, real-time performance and adaptability of an industrial intelligent system.

The invention discloses an agricultural product intelligent inventory management system and method based on big data, and belongs to the technical field of inventory management. The method comprises the following steps: in each link of an agricultural product supply chain, acquiring agricultural product inventory data, preprocessing, extracting features, and dividing the extracted features into a training set and a test set; the loss rate prediction model is trained and evaluated, and the agricultural product inventory loss rate is predicted in real time; when the predicted agricultural product inventory loss rate does not meet the market demand, constructing an agricultural product supply chain graph model, and screening agricultural product nodes needing replenishment based on the agricultural product inventory loss rate; layering the agricultural products to be replenished based on the refreshing time in the agricultural product information, and calculating the replenishment quantity and replenishment path by using a graph algorithm; based on the replenishment quantity and the replenishment path, calculating replenishment time plans of the agricultural products in different refreshing periods; and constructing a deployment task list, selecting a deployment mode, and making an agricultural product deployment plan.

The invention discloses a pose data adjustment method and device, a storage medium and an electronic device, and relates to the technical field of laser point cloud. The method comprises the steps that target data are acquired, the target data comprise first point cloud data and first pose data, and the first pose data correspond to the first point cloud data; based on the first pose data and a preset window, performing block processing on the first point cloud data to obtain target block data, the target block data including multiple layers of block data, and any layer of block data including multiple blocks; determining a point cloud fitting form of the preset-layer block data, the preset-layer block data being the block data of the bottommost layer in the multi-layer block data, and the point cloud fitting form being used for representing a distribution state of the point cloud data of any block in the preset-layer block data; and adjusting the first pose data based on the point cloud fitting form to obtain second pose data. According to the invention, the technical problems of large error and low mapping precision of a laser mapping algorithm in the prior art are solved.

The invention relates to the technical field of geospatial information systems, and particularly discloses a TBM (tunnel boring machine) construction whole-process intelligent decision expert knowledge base construction method, which comprises the following steps of: realizing full-automatic structuring of multi-source heterogeneous data aiming at the characteristics of dispersion, sub-platform and non-structuring of TBM cluster construction information of tunnels along a super-long and super-large railway; constructing a TBM construction whole-process intelligent decision-making knowledge graph mode layer, and establishing a knowledge graph with space-time relevance; and according to user query content, knowledge graph query and knowledge recommendation based on semantic similarity are realized, and dynamic update of the knowledge base is realized. According to the method, association expression is carried out on TBM safety risk information from coverage investigation design to a construction stage by utilizing a knowledge graph, a construction risk disposal expert knowledge base is constructed, and the expert knowledge base is mined in combination with semantic matching and a graph algorithm, so that risk prevention and control information intelligent recommendation for different construction scenes is realized, and the construction risk management efficiency is improved. And intelligent decision-making in the whole process of TBM construction is effectively assisted.

The invention discloses a power distribution network operation order generation method and system based on a low-code platform, and relates to the technical field of power system automation. Through the low-code platform, an operation task text of a power distribution network is obtained, the operation task text is adopted to retrieve a preset atomic chain database, a corresponding atomic chain is matched, and based on a preset text matching model, the power distribution network operation order is generated. Screening each atomic chain according to the operation task text to obtain a plurality of target atomic chains, sorting each target atomic chain by adopting a graph algorithm based on a preset dependency relationship list to generate a corresponding operation sequence, and performing security check on the operation sequence by adopting a low-code platform to obtain a security check result; and generating a corresponding power distribution network operation ticket and displaying the power distribution network operation ticket on the low-code platform. The problems that a traditional power distribution network operation ticket filling mode mainly adopts a manual invoicing mode, completely depends on personal service experience of a dispatcher, is easy to cause writing errors, consumes long time and reduces the operation reliability of a power distribution network are solved.

This invention relates to the field of wireless communication technology, specifically to an intelligent routing planning method and system for ad hoc networks. The method includes: constructing an ad hoc network strategy with a dual-link mechanism, achieving central node election and network construction through a first bidirectional link and a second unidirectional link mechanism; introducing signal encryption and authentication mechanisms to ensure communication security; constructing a network maintenance strategy to support automatic access of new nodes, switching when a central node fails, and periodic network structure updates; constructing a graph model based on node connection relationships and signal strength, and introducing artificial intelligence graph algorithms and reinforcement learning strategies to achieve optimal routing planning under multi-hop paths. The method provided in this application can improve the security, stability, and routing intelligence of ad hoc networks, achieving efficient and reliable wireless communication network planning.

The invention relates to the technical field of software updating, and provides a software upgrading risk prediction method, which comprises the following steps of: obtaining high-dimensional feature vectors corresponding to a to-be-evaluated software version and a plurality of historical software versions respectively, determining a sample covariance matrix according to the high-dimensional feature vectors corresponding to the plurality of historical software versions, and determining the risk of software upgrading according to the sample covariance matrix; according to the high-dimensional feature vector corresponding to the to-be-evaluated software version, the sample covariance matrix and the high-dimensional feature vector corresponding to the current to-be-upgraded software version, determining the mahalanobis distance between the to-be-evaluated software version and the to-be-upgraded software version, and for the multiple high-dimensional feature vectors, determining the to-be-upgraded software version according to the mahalanobis distance through an accumulation and control chart algorithm. And determining global mutation indexes corresponding to the to-be-evaluated software version and the plurality of historical software versions, and determining a risk evaluation index between the to-be-evaluated software version and the to-be-upgraded software version according to the mahalanobis distance and the plurality of global mutation indexes. The accuracy of risk assessment indexes in the software upgrading process is improved, and the accuracy of risk prediction is effectively improved.

The present application belongs to the technical field of data analysis, and specifically relates to a method for detecting an adversarial social robot based on a graph and structure perception. The method comprises the following steps: constructing a user relationship network by taking social platform users as nodes and the forwarding relationship between the users as edges; performing representation learning on different structures in the network by using a structure feature extractor; performing structure similarity calculation between nodes by using a structure self-attention mechanism; training user representation vectors by using a graph Transformer algorithm; and inputting the user representation vectors into a category generative adversarial network for training to obtain a discrimination result of whether the user is a social robot. The method has strong generalization and high robustness by using adversarial thinking and unsupervised training, and can be used for detecting social robots in a large social network platform, thereby providing commercial value for user management, advertisement placement, public opinion safety control and the like of the social platform.