41 results about "Spread Networks" patented technology

The invention discloses a false information intelligent detection and traceability method based on cross-modal consistency verification. The method specifically comprises the following steps: S1, multi-modal content feature extraction; s2, carrying out cross-modal consistency verification; s3, performing deep counterfeiting detection; s4, propagation anomaly detection; s5, checking knowledge enhancement facts; s6, information tracing and variation tracking; s7, comprehensive authenticity evaluation; s8, a model training strategy; a multi-task learning framework is adopted, and parameters of all modules are jointly optimized. According to the method, a multi-dimensional consistency verification system is constructed by fusing multi-source information such as video content, text semantics, a propagation network and an external knowledge base, and automatic identification, authenticity evaluation and information traceability of various false information are realized.

The invention relates to the technical field of power grid exception handling, and discloses a power grid exception event handling method and system based on a knowledge graph. The method comprises the following steps: collecting time sequence data flow and space coordinate information of a power grid abnormal event in real time through multi-source sensing equipment, and fusing equipment operation parameters to construct a dynamic event data pool; performing graph structure conversion on the dynamic event data pool based on a knowledge graph topological connection relationship, calculating event node propagation path intensity and generating an event propagation network model; key abnormal indexes are separated out from the model by adopting a multi-modal feature extraction technology, risk probability calculation is completed in combination with an adaptive learning algorithm, and an abnormal risk distribution diagram is output; and classifying event levels according to the graph, and automatically generating a processing instruction sequence. According to the method, omnibearing data perception and accurate analysis of the abnormal events of the power grid are realized, an event propagation rule is clearly presented, risk assessment is enabled to be closer to an actual operation state, and accurate guidance is provided for abnormal disposal.

The invention discloses a legal contract risk monitoring and early warning method, and particularly relates to the technical field of risk early warning. The method comprises the following steps: constructing a multi-contract risk propagation network by identifying transmissible terms in a plurality of legal contract texts, and generating a multi-contract propagation path topology network model by adopting a graph theory algorithm; identifying a cascade trigger mechanism of risk propagation terms between contracts by using time sequence causal analysis, and outputting a cascade risk propagation link; performing sensitivity grading on each risk propagation path to obtain propagation path sensitivity grading data; integrating the cascaded risk propagation link and propagation path sensitivity grading data, constructing a contract network risk propagation comprehensive evaluation model, and outputting a comprehensive risk evaluation result; comparing the comprehensive risk assessment result with a preset risk early warning threshold rule to generate corresponding risk prompt information; according to the method, the propagation trend and the diffusion path of the legal contract risk can be effectively predicted, and the timeliness and the accuracy of contract risk management and control are improved.

The invention discloses a bidirectional adaptive video super-resolution method based on frame difficulty index, and belongs to the field of computer vision. The invention provides a frame-level dynamic reconstruction method for solving the problems that simple frame calculation is redundant and difficult frame reconstruction is insufficient due to the fact that an existing model adopts a fixed calculation strategy for video frames with different difficulties. According to the method, a motion detail decoupling propagation network is constructed, motion information is efficiently transmitted by utilizing a shallow forward propagation branch, and a deep backward propagation branch focuses on recovering texture details so as to decouple a time sequence propagation task; meanwhile, a frame reconstruction difficulty evaluation network is introduced to generate a global difficulty index, so that the receptive field weight of the adaptive time sequence fusion network and the refining depth of the dynamic refining network are regulated and controlled. According to the method, through explicit modeling frame-level reconstruction difficulty, adaptive matching of the model capacity and the video frame feature complexity is realized, and the video reconstruction performance is remarkably improved under limited computing power.

The invention relates to a product security event evolution process prediction method and device, and electronic equipment. The method comprises the steps that a data set is built through multi-source heterogeneous data, and the multi-source heterogeneous data is data having an incidence relation with a product safety event; cleaning data and extracting event entities and attributes from the data set; constructing a affair graph by using a causal reasoning method of a time sequence association rule; taking entities in the affair atlas as nodes of a propagation network, and mapping causal relationships in the affair atlas as edges of the propagation network and weights on the edges; creating agent attributes and event state transition rules; and constructing a multi-agent model, and predicting the evolution process of the product security event by using the multi-agent model. According to the invention, the technical problem of lack of research on the evolution law of the product security event in the prior art is solved.

The application provides a kind of online social platform topic propagation topology fingerprint graph construction method, comprising: constructing topic propagation network graph, determining source node, extracting maximum connected subgraph, calculating subgraph each node forwarding volume, node hop count, adjacent edge forwarding intensity;From each subgraph source node, with forwarding intensity as weight, the shortest weighted path from source node as filter function, forward search out connected component sequence, record each connected component start time step and duration, the same connected component is merged;From each subgraph recorded search end node, start reverse search, get ring sequence from reverse search, record each ring structure start time step and duration, the same ring structure is merged;According to connected component sequence set, ring sequence set constructs topic propagation fingerprint graph.The application can effectively extract the significant topology of connected path and loop formed in the process of online social platform topic propagation, and quantify its duration.

The application relates to a depth completion method, device and equipment based on radiation difference and spatial distance, a bilateral propagation network is constructed based on a multi-scale architecture; wherein, an ultra-network is designed in a preprocessing module, the bilateral propagation network has a preference for the nearest value in the radiation difference and the spatial neighborhood in the depth propagation process through filtering weights; the preprocessing module processes a color image and a sparse depth map to obtain corresponding image encoding and depth features under each scale; the image encoding and the depth features of the same scale are input into a multi-modal fusion module for fusion to obtain a dense depth map; a depth optimization module iteratively updates the depth propagated by the multi-modal fusion module based on the sparse depth map, and the completed dense depth map is obtained by weighted combination of the dense depth maps under each scale. The application solves the problems of ambiguity of sparse depth map representation and spatial invariance of convolution operation in the multi-modal fusion stage.

The application relates to the technical field of intelligent response to comment crisis, and discloses an intelligent response method and device for comment crisis, electronic equipment and a storage medium, wherein the method comprises the following steps: through deep semantic analysis and clustering of multi-source comment data, independent crisis event entities can be accurately and efficiently identified and aggregated from noise information; through construction of a time sequence social communication network model and quantification of user node influence, scientific prediction and evaluation of crisis diffusion trend and intensity are realized, so that a crisis index can more accurately reflect the severity of an event; and in combination with cultural characteristics related to the event and the crisis index, a first-level response language suitable for the event is dynamically generated.The application has the beneficial effect that the response strategy is improved in pertinence, timeliness and cultural suitability, so that a subject can more actively and effectively manage and resolve online comment crisis, and automation and intelligentization of a main process from identification to response are realized.

This disclosure provides a target model training method, a multimodal data processing method and apparatus therefor, an electronic device and a program, which can be applied to artificial intelligence-based content generation scenarios in the field of artificial intelligence technology, particularly in the fields of computer vision, deep learning and large-scale models. [Solution] The specific solution includes inputting sample data into a preset model to obtain initial multimodal features of the sample data, and, if the parameters of the image / text encoder in the preset model are fixed, using the initial multimodal features and preset noise features to train the N-layer spread network in the preset model to obtain a target model.

The invention provides a new energy station operation simulation assessment method and system based on comprehensive early warning, and relates to the field of new energy power generation, and the method comprises the steps: collecting operation and fault data, constructing a parameter propagation network to recognize fluctuation abnormity, and generating a risk propagation path; a turning point is determined according to the fluctuation change rate, and adjustment and disturbance time periods are divided; constructing a staged evolution scene, and injecting a control instruction and an interference signal; and evaluating the adjustment and anti-interference capability based on the response and the deviation between the instability trajectory and the safety boundary, and generating an assessment result. According to the method, early warning and targeted evaluation of the operation risk of the new energy station are realized.

The invention relates to the technical field of intelligent response of crisis comment, and discloses an intelligent response method and device of crisis comment, electronic equipment and a storage medium, and the method comprises the steps: carrying out the deep semantic analysis and clustering of multi-source comment data, can precisely and efficiently recognize and aggregate an independent crisis event entity from noise information, and improves the accuracy of the crisis event entity. By constructing a time sequence social communication network model and quantifying user node influence, scientific prediction and evaluation of crisis diffusion trend and intensity are realized, so that a crisis index can more accurately reflect the severity of an event, and an adaptive first-level response skill is dynamically generated in combination with cultural features associated with the event and the crisis index. The method has the beneficial effects that the pertinence, the timeliness and the cultural suitability of the response strategy are improved, so that the subject is helped to more actively and effectively manage and resolve the online comment crisis, and the automation and the intelligentization of the main process from identification to response are realized.

The invention discloses a drill jumbo toughness evaluation method based on personalized customization and environmental disturbance. The drill jumbo toughness evaluation method comprises the following steps: constructing a personalized demand library and an environmental disturbance library; constructing a forward propagation network of demand change, a reverse feedback network of environmental disturbance and a propagation network in which the two networks act at the same time, and identifying a component and a path which are jointly propagated and influenced by the demand change and the environmental disturbance; setting a first scene for environment disturbance assessment, a second scene for personalized customization demand change assessment, and a third scene for assessment when the environment disturbance and the personalized customization demand change act at the same time; establishing a toughness evaluation index system of the drill jumbo; and based on the calculated values of the four indexes, generating a comprehensive toughness evaluation result in the scene to be evaluated by adopting a weighted fusion method. According to the method, the toughness of the drill jumbo under the scene of environment disturbance and personalized customization demand change can be improved.

This invention discloses a convex clustering analysis method and system for assembly error propagation networks. The method includes the following steps: S1, acquiring the assembly error propagation network data to be clustered; S2, constructing a convex optimization problem based on the network data acquired in step S1, and transforming it into an equality-constrained optimization problem; S3, solving the equality-constrained optimization problem using the alternating direction multiplier method, iterating repeatedly until a preset iteration termination condition is met, and obtaining the cluster centers corresponding to each network data; S4, completing the clustering of the network data according to the distance between the cluster centers, and outputting the clustering results. This invention applies the convex clustering framework to the clustering problem of network sets, which can effectively handle graph-structured data and overcome the shortcomings of traditional clustering methods, such as being prone to getting trapped in local optima, being sensitive to initial values, and requiring a preset number of clusters, thus achieving efficient and robust network set clustering.

This invention discloses a method for mitigating social propagation risk and selecting robust groups based on extreme value dependency networks and maximally independent sets. The method includes: acquiring user interaction data from a social network and constructing a graph structure based on the user interaction data; obtaining target node pairs from the graph structure for a preset propagation activity period and calculating the extreme value dependency of the target node pairs; constructing propagation networks with target nodes as vertices and extreme value dependency as the edge construction criterion under different dependency thresholds, evaluating the propagation networks, and determining the extreme value dependency propagation network based on the evaluation results; dividing the extreme value dependency propagation network into communities and determining central nodes from the divided communities; constructing a minimum extreme value dependency group based on the extreme value dependency propagation network; and identifying high-risk node groups in the minimum extreme value dependency group using a propagation loss index to obtain the identification result. This invention can achieve the identification, isolation, and control of tail propagation linkage relationships.

The application discloses an influence maximization method based on pass-through and autoregressive influence estimation, relates to the technical field of information processing, and models propagation dynamics according to propagation trajectory data of nodes in a given propagation network G, obtains a propagation model M(x, G; theta), and is used for modeling a node state evolution process, wherein x represents a node state, and theta is a model parameter; an initial state x of a node is mapped to a propagation final state y, that is, a final infection probability of the node, by using the propagation model M(x, G; theta), the initial state x of the node is mapped to a continuous state vector z, an end-to-end propagation proxy model y=M(z, G; theta) is constructed, the model parameter theta is fixed, the propagation final state y is set as an optimization target, the continuous state vector z corresponding to the initial state x of the node is optimized by using a gradient descent method, and a seed node set is obtained according to an optimization result. 0 0 s 0 The application provides an influence maximization method which can consider diffusion effect, calculation efficiency and application flexibility.​​​

The application belongs to the field of social network information processing, and particularly relates to a key user identification method, device and system in a derivative topic network; the method comprises the following steps: obtaining data information in a topic network; extracting a native topic, a derivative topic user relationship network, a native topic and derivative topic propagation network, user basic attributes and user intimacy from the data information in the topic network; extracting a user feature vector representation through a community label propagation algorithm according to the user basic attributes and user intimacy; extracting a topic vector representation of the native topic and the derivative topic through a joint distribution adaptive algorithm according to the native topic and the derivative topic propagation network; inputting the user feature vector representation and the topic vector representation into a user influence dynamic space-time model based on an attention mechanism, and outputting an opinion leader user or a long tail user. The application can accurately identify key users in a derivative topic network.

A method, apparatus, and device for safety assessment of an energy storage system are disclosed, relating to the field of energy storage system technology. The method includes acquiring first operational data of the energy storage system, the first operational data including fault data describing at least one basic fault type; determining, based on the first operational data, the prior probability of the basic fault type represented by a leaf node in a Bayesian failure propagation network, where the root node in the Bayesian failure propagation network represents the failure mode of the energy storage system, and the directed path from the leaf node to the root node represents the propagation link of the failure mode caused by the basic fault type; inferring the posterior probability of the corresponding failure mode of the energy storage system based on the prior probability and the directed path from the leaf node to the root node; and obtaining the assessment result of the energy storage system based on the posterior probability.

The invention relates to a hot event propagation path prediction method and system based on multi-modal big data. The method comprises the following steps: based on a preset trigger, identifying seed content from a content publishing data stream, and acquiring propagation behavior data by taking the seed content as a center; based on the seed content and the propagation behavior data, constructing a preliminary propagation network structure to obtain an early propagation network; analyzing a propagation chain in the early propagation network to obtain a propagation chain analysis result, and based on the propagation chain analysis result, evaluating a user tendency response behavior to obtain a response tendency prediction matrix; and performing propagation simulation based on the early propagation network, the propagation chain analysis result and the response tendency prediction matrix to obtain propagation probability distribution. By adopting the method, the accuracy of propagation prediction can be improved by utilizing multi-modal data and social network analysis, the combination of content semantics and user behaviors is emphasized, and the method is suitable for the hotspot event propagation analysis of platforms such as social media and the like.

The application discloses a social network influence maximization method based on node reading effect and delay, comprising the following steps: S1, obtaining historical transmission data of multiple articles, and constructing a historical propagation network with time sequence for each article; S2, constructing a propagation network portrait according to the historical propagation network of all articles, and calculating the reading delay of nodes on the historical propagation network; S3, predicting the reading duration of a node after receiving an article, the reading delay, and the probability of propagating the article to other nodes; S4, for a certain node, forming a predicted propagation tree with the node as a root node; and S5, screening a plurality of initial propagation nodes based on the propagation tree. The article realizes influence maximization propagation based on node reading effect and delay.

This invention relates to the field of shipbuilding and repair safety management and risk prevention technology, and provides a three-level prevention and control method for shipbuilding and repair accidents, including: constructing an accident database based on analytical methods and dimensional modeling; performing point-level risk quantification and prediction, using an improved normalized risk index method and a composite incremental risk assessment method to obtain risk quantification values ​​and comprehensive risk coefficient values, and constructing an AutoML model to perform risk prediction; performing line-level cause tracing and chain breaking, using a large language model and LDA topic model to construct a six-level cause classification model to identify the most likely inducing path; performing network-level coupling analysis and propagation prevention and control, constructing a complex network, identifying key nodes and important edges by ranking the importance of nodes and edges, and using Monte Carlo simulation to evaluate the network robustness; and formulating a prevention and control strategy based on the identification results, including multi-dimensional collaboration and targeted blocking of key nodes, thus achieving precise targeted blocking from individual quantification to propagation network prevention and control.

The application discloses a propagation control method and device, and relates to the technical field of complex networks. The method is used to solve the problem that the existing epidemic propagation control method has too high complexity and low applicability caused by too many removed nodes from the network due to the fact that the propagation characteristics are not considered. The method comprises the following steps: removing each edge included in an initial propagation network from the initial propagation network to obtain a first propagation network; traversing each edge included in a first importance sequence, and setting the edge as a first edge; when the value of the REDL importance of the first edge is less than an initial edge importance threshold value, adding the first edge to the first propagation network and the head of the initial importance sequence; when the first importance sequence is empty, removing A most important edges included in the initial propagation network; and when it is determined that the proportion of infected nodes in a stable state is less than a set value, determining the proportion of edges that need to be removed for complete control of the propagation.

The application provides a detection method and system based on propagation network and content fusion. A monitoring router is determined through the propagation network. Different encoding is performed on three features of user attribute characteristics, structural characteristics and source content characteristics by the monitoring router. Then, fusion detection and classification are performed. Finally, whether an attack is received is analyzed by a control center according to classification information and a forwarding state, so that the problem that the prior art cannot adapt to rich and varied content forms and continuously evolving features of network social media is solved.

This application provides a knowledge graph-driven power grid fault root cause analysis system and method, comprising: processing data of different time granularities using a time-series alignment algorithm based on an initial event sequence to obtain an ordered event list on a unified time axis; if the electrical connection strength of adjacent events in the ordered event list exceeds a preset threshold, it is determined to be direct coupling, and a preliminary causal chain is identified; for the preliminary causal chain, a protection logic rule base is obtained, and matching protection device operation modes are extracted from it to obtain extended causal relationships; through the extended causal relationships, a fault propagation network is constructed using a graph model, and the propagation paths between network nodes are determined to obtain a complete event chain; if there are delayed action events in the complete event chain, the key turning point is determined by comparing the delay duration with the unified time axis; based on the root cause classification results, an evolution path report is generated, the missing links in the report are identified, and supplementary preventive measures suggestions are obtained.

This invention relates to the field of power supply system technology and discloses a smart enclosure safety operation and maintenance system, comprising: a data acquisition module for real-time acquisition of operating data from monitoring units within the distribution box; a data storage and investigation module for storing the operating data and extracting anomalies; an adaptive behavior pattern recognition module for identifying the legality or illegality of anomalies through dynamic behavior profiling and weighted time correlation analysis; an anomaly processing module for generating temporary processing windows based on the anomalies and comparing them with historical data to generate analysis results; a decentralized command propagation network module; and a data processing module for generating execution commands based on the analysis results. This invention's data storage and investigation module stores historical data, extracts anomalies by combining deviation calculations, and the adaptive behavior pattern recognition module determines the legality of anomalies, reducing the false positive rate and improving the accuracy of anomaly identification.

The invention relates to an inventory difference checking method and system based on an inventory dynamic propagation network, and belongs to the field of inventory management. The method comprises the following steps: acquiring a pulse conduction coefficient, and constructing an inventory dynamic propagation network based on the pulse conduction coefficient; acquiring an operation pulse, and injecting the operation pulse into the inventory dynamic propagation network; performing pulse propagation simulation based on the inventory dynamic propagation network, and outputting coherent pulse intensity and dissipated energy of each node; establishing an early warning mechanism based on the dissipated energy hot spots based on the dissipated energy; and after an early warning mechanism is triggered, tracing of a dissipation source is carried out. According to the invention, by constructing the inventory dynamic propagation network and introducing the pulse conduction and energy dissipation model, the interference effect between operations can be effectively captured, the conversion from passive response to active prediction is realized, and the accuracy and robustness of inventory management are significantly improved.

The invention provides a multi-language code specification and security vulnerability synchronous scanning method, which comprises the following steps of: 1, performing cross-language grammar adaptation and grammar tree feature entropy extraction on a multi-language code set to generate a multi-language code grammar tree feature set; step 2, constructing a vulnerability feature belief propagation network based on known vulnerability samples and a code specification library; 3, inputting the multi-language code syntax tree feature set into a vulnerability feature belief propagation network to synchronously execute code specification conformity verification and security vulnerability detection so as to generate a scanning result; step 4, calculating a specification-vulnerability collaborative verification factor, and performing missed scanning and misjudgment correction on the scanning result to obtain a correction result; and step 5, iteratively updating the vulnerability feature belief propagation network based on the correction result to perform calibration and deviation elimination processing on the scanning result to generate vulnerability risk classification and standard violation classification output. According to the scheme, the overall scanning period is relatively short, and the efficient requirement of agile development of a core system can be better met.

The invention discloses an international public opinion focus trend prediction and deduction method. The method comprises the following steps: acquiring multi-platform, multi-language and multi-mode public opinion data through a data acquisition and fusion step; text preprocessing, topic focus recognition and cross-culture sentiment analysis are carried out through a data processing and feature extraction step; a dynamic propagation network is constructed through the propagation network modeling and analysis step, key nodes and propagation paths are identified, and emotion flow is tracked; through prediction and deduction steps, an issue, emotion and network features are coupled, after time sequence construction, state vectorization and network feature extraction, trend prediction is performed by using a time sequence model and a focus migration model, and multi-scenario evolution deduction is performed by adopting multi-subject modeling and a dynamic feedback mechanism; and finally, carrying out visual display and strategy generation through a result generation and output step. And full-process automation and intelligentization from perception and understanding to prediction and deduction of international public opinion focuses are realized.

The invention belongs to the technical field of fault detection, and particularly relates to a cable real-time production fault detection method based on data processing. The method comprises the following steps: firstly, screening key process parameter positions strongly related to a fault, and carrying out normalization and IQR abnormal value suppression processing on a parameter sequence; constructing a directed causal topological graph based on a non-parametric time sequence causal test algorithm of transfer entropy; a prediction value is generated through a causal perception weighted regression model, a standardized residual error is calculated, and a residual error propagation network is constructed to iteratively calculate a node accumulated abnormal energy value; and finally, determining a fault according to the proportion of the abnormal nodes and the continuous abnormal duration. According to the method, the problems of response lag and missed judgment and misjudgment of traditional detection are solved, the real-time performance and accuracy of cable production fault detection are improved, and high-precision production requirements are met.

The application relates to a verification method and device of a power supply contract, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: preprocessing collected data, constructing a multi-dimensional heterogeneous knowledge graph based on the preprocessed data, dynamically allocating the influence of neighbor nodes by using an attention weighting mechanism, constructing and training a weighted heterogeneous relationship propagation network, outputting a business entity risk probability, constructing a high-dimensional feature vector, inputting the high-dimensional feature vector into a trained GBDT model, outputting a GBDT risk probability, outputting an unsupervised anomaly score through a trained unsupervised anomaly detector, counting the total number of times that each business entity triggers an inspection rule in the multi-dimensional heterogeneous knowledge graph as the rule triggering number, and calculating a comprehensive risk index based on the business entity risk probability, the GBDT risk probability, the unsupervised anomaly score and the rule triggering number and performing corresponding early warning. The method can improve the inspection efficiency and accuracy.

This invention discloses an adaptive router and its routing method based on a three-dimensional on-chip network with irregular inter-layer interconnections. It utilizes the vertically connected topology of the three-dimensional on-chip network, a healthy TSV (Through Silicon Via) location mechanism, and a congestion propagation network to adaptively reconstruct routing paths based on congestion level, injection rate, and TSV availability. By designing healthy TSV information tables, healthy TSV distance information tables, healthy TSV congestion information tables, a routing decision module, and healthy TSV propagation and congestion awareness mechanisms, it achieves collaborative operation between distance-based and congestion-based routing algorithms to reduce performance losses caused by congestion and failures. This invention uses relatively low hardware overhead, enhances the reliability and adaptability of three-dimensional on-chip network routing, and effectively reduces the average packet latency in the presence of irregular vertical interconnections and dynamic network traffic changes, significantly improving the overall performance and robustness of the system under fault and congestion environments.