112 results about "Weighted network" patented technology

The invention discloses a failure chain quantitative analysis and risk assessment method and system based on a multi-level security model, and aims to solve the defects that accident cause analysis of a complex social technology system is inaccurate, and a risk assessment result is lack of effective verification. According to the method, a multi-level causal model is systematically constructed, a multi-dimensional failure chain (MDFC) is extracted, multi-dimensional risk quantification is performed on the MDFC, a directed weighted failure propagation network is constructed based on the multi-dimensional risk quantification, and structural features of the directed weighted failure propagation network are analyzed to identify key risk factors. The core innovation of the method is that reverse accident reason tracing and forward risk propagation path analysis based on the weighted network are fused, mutual verification and iterative optimization are realized by comparing analysis results of the two paths, so that the understanding of an accident evolution mechanism is deepened, and the reliability of evaluation is improved. The system vulnerability can be revealed more comprehensively, powerful support is provided for formulating accurate risk control measures, and the overall safety level of a complex system is improved.

The invention relates to a construction site environment dynamic regulation and control method and system fused with AIoT. According to the method, a multi-source environment sensor network is deployed to collect original data of a construction site environment in real time, a data set with aligned timestamps is generated, an abnormal event is detected by using an isolated forest algorithm after standardized denoising processing, and a specific parameter type and a time window are marked; a causal intensity matrix between parameters is constructed through Granger causal test on the basis of environment data intercepted in an abnormal time period, a directed weighted network adjacency matrix representing a pollution propagation path is generated accordingly, and an abnormal source is accurately positioned through node influence propagation calculation and timestamp verification; and finally, a regulation and control instruction sequence is dynamically generated according to the shortest influence path of the source node in the propagation network, so that autonomous analysis of implicit association among construction site environment parameters, accurate positioning of a pollution source and dynamic distribution and instant deviation correction of a multi-stage regulation and control strategy are realized.

The invention provides an unmanned aerial vehicle ad hoc network transmission and calculation integrated resource scheduling method based on task driving, and the method comprises the steps: building a multi-dimensional resource pool model which comprises the communication bandwidth, calculation resources and storage resources of an unmanned aerial vehicle, and collecting the resource state vector of each unmanned aerial vehicle node in real time; a dynamic topology sensing network is constructed, link duration is predicted through relative motion speed between unmanned aerial vehicle nodes, and a network structure chart with weights is generated; constructing a decision model based on a fusion architecture of a preset message passing neural network and a deep reinforcement learning network, and inputting the network topology features of the network structure chart and the resource state vector into the decision model; and outputting an optimal scheduling strategy including target node selection and multi-hop path planning through the decision model, and maximizing system benefits while meeting constraints of tasks on communication and computing resource quality. The problems that existing unmanned aerial vehicle networking communication is high in time delay, low in reliability and difficult to calculate and maximize utilization of resources are solved.

The invention provides a key node identification method for a disease marker expression regulation network, and belongs to the technical field of disease markers, and the method comprises the steps: firstly carrying out the preprocessing and quality control of original data, including batch effect removal, abnormal sample identification and the like; then identifying differential expression genes through multiple difference analysis and a pre-training model, and constructing a gene expression correlation network; and integrating multi-source regulation and control data to construct a multi-level weighted network, calculating network node features, and carrying out representation learning and module division. And based on multi-dimensional features such as network topology features, module contribution degree and biological importance, a neural network model is trained to carry out key node identification. And finally, optimizing the model through multi-layer verification such as pathway enrichment, disease gene overlapping, expression stability, time sequence change and network disturbance, and finally obtaining a verified key node set. The problem that in the prior art, the interaction relation between molecules is ignored, and consequently some key regulation and control nodes are possibly missed is solved.

The invention discloses a seismic geological disaster monitoring and early warning device, and the device comprises an intelligent sensing system which carries out the real-time collection of geological parameters, facility states and environmental factors through a multi-mode sensor array, and comprises the steps: capturing the strain of a pipeline through a distributed optical fiber sensor, monitoring the vibration through an MEMS accelerometer, and providing deformation data through an InSAR satellite; the analysis system constructs a complex network modeling engine, fuses the historical transition probability matrix and the real-time deformation rate parameter by using a directed weighted network model, and combines a two-channel causal inference engine which comprises a physical causal channel and a data causal channel; the physical causal channel is embedded into a coulomb fracture criterion to calculate a fault stress accumulation rate, the Darcy law simulates pore pressure propagation, and the data causal channel generates an anti-fact sample through a CaualGAN and extracts a real causal chain in combination with a time causal convolutional network; and the execution and feedback system is used for triggering graded early warning based on an analysis result of the analysis system.

The invention discloses a low-altitude airspace planning method and system based on a micro-terminal area, and relates to the field of low-altitude airspace planning, and the method comprises the steps: generating a corresponding take-off and landing area for a low-altitude take-off and landing point or a take-off and landing field, and generating a take-off and landing hub area for an adjacent take-off and landing point or a take-off and landing field, so as to construct the micro-terminal area of the low-altitude airspace; constructing a risk area, and forming a restricted area of the low-altitude airline network together with the control area and the no-fly area; on the basis of the restricted area, generating alternative intermediate points connected with the take-off and landing points or the take-off and landing fields; based on the data of the alternative intermediate points and the data of the micro-terminal area, constructing an initial low-altitude airline network connected with all the points; calculating a weight corresponding to each side line in the initial low-altitude airline network, and optimizing the initial low-altitude airline network based on a weighted network; and removing redundant intermediate points based on the optimized network, constructing access points of the airline network and each micro-terminal area, and generating a final low-altitude airline network. According to the invention, a more reliable technical basis is provided for low-altitude route network planning.

The invention relates to a hyperspectral de-mixing method based on a double-branch cross-weighted network, and the method comprises the following steps: obtaining a hyperspectral multi-mode data set which comprises hyperspectral image data and laser radar data, and inputting the hyperspectral image data and the laser radar data into the double-branch cross-weighted network; processing the hyperspectral image data through a spectral encoder to obtain primary spectral features; a spatial encoder processes laser radar image data to obtain primary spatial features, the primary spectral features and the primary spatial features are input into an adaptive feature selection module to be processed to obtain precise spectral features and precise spatial features, and the obtained precise spectral features and precise spatial features are added to be fused; and obtaining an abundance coefficient graph and an end member from a fused result through an activation function, and obtaining a reconstructed hyperspectral image by using a decoder. According to the invention, the unmixing effect of different substances of similar materials in a multi-modal scene can be improved.

The invention discloses a multipath channel parameter reconstruction and beam coverage prediction method, and relates to the technical field of wireless communication. Aiming at the problems of low data utilization efficiency, limited environment modeling precision, strong coupling of measured data and antenna beam configuration and the like in the prior art, the method comprises the steps of configuring antenna beams and orientation parameters, collecting reference signal receiving strength, performing deterministic channel modeling simulation, establishing a mapping model, performing sparse recovery, constructing a channel propagation model, evaluating a channel map and the like. In combination with a coherence perception weighting network (CARE-Net) algorithm and ray tracing (RT) physical prior guidance, accurate reconstruction from a low-dimensional reference signal receiving strength (RSRP) measurement value to a high-dimensional multipath parameter is realized. The method has the advantages that the reconstructed multi-path parameters are decoupled from the antenna configuration, the method has the cross-beam / cross-configuration generalization prediction capability, the channel coverage performance under different antenna configurations can be accurately predicted, a reliable basis is provided for wireless network optimization, and the method is low in cost, high in precision and strong in generalization.

The invention relates to the technical field of mobile communication, and discloses a prediction-based network congestion avoidance system, which comprises a data sensing end, a space-time prediction end, a strategy generation end and an execution feedback end, and is characterized in that the data sensing end, the space-time prediction end, the strategy generation end and the execution feedback end are jointly integrated with a voice alarm module; a data sensing end is arranged, when network state monitoring is carried out, a network topological graph with weights is constructed in real time, a timestamp index feature matrix is generated, the time delay defect of a traditional passive response mechanism is eliminated, and an active sensing mechanism constructed at the end captures link dynamic features before service damage; the real-time performance and the first-onset performance of congestion risk identification are ensured, the problem of service interruption caused by post-detection is overcome, and the core defect of insufficient prediction model precision is solved by setting a space-time prediction terminal and based on a collaborative modeling architecture of a graph convolutional neural network and an attention mechanism during congestion risk analysis.

The invention provides a weighting network key node identification method based on layering. The method comprises the following steps: step 1, establishing a multi-layer network initial model based on a layering network thought; 2, selecting two layers of networks with the maximum similar modularity gain after combination to perform iterative combination, and obtaining a multi-layer network model by taking the similar modularity gains of any two layers of networks are non-positive as a loop termination condition; step 3, considering betweenness indexes of each network node in the multi-layer network model, measuring importance of each network node in the multi-layer network model, and dividing the network nodes into interactive nodes, first-level non-interactive nodes and second-level non-interactive nodes; and 4, evaluating the importance of the interactive nodes, the primary non-interactive nodes and the secondary non-interactive nodes by adopting three betweenness indexes. According to the method, multi-layer network modeling is carried out on the premise that the global attributes of the key nodes of the complex network have limitation, and the recognition accuracy of the key nodes in the network nodes of the multi-layer network is ensured.

A machine learning based data structuring method for automating dimensional data modelling process in data repositories is disclosed. The ML-based data structuring method includes obtaining datasets from databases; classifying the data comprising attributes and measures based on historical data using a ML-based classifier model; determining associations between the datasets using primary and foreign keys, SQL logs, usage of the datasets in creating transformations, and performance of fuzzy string match; assigning weightages to the determined associations between the datasets based on utilization of the determined associations using weighted network graphs; validating the determined associations between the datasets based on recurrent utilization of the determined associations; clustering the datasets based on the validated associations between the datasets by detecting dimensional models in the weighted network graphs; and generating actionable insights on each of the clustered datasets by performing exploratory data analysis, influencer analytics, and forecasting of the data.

The invention relates to the technical field of data processing, and discloses an urban space structure collaboration measurement method and system, and a storage medium. The method comprises the following steps: performing gridding processing on urban space elements through multi-temporal data acquisition to obtain a three-dimensional data matrix; calculating a space-time dynamic coupling coefficient by adopting an improved gravity model according to the three-dimensional data matrix; constructing a space-time bidirectional weighting network through a bidirectional weighting algorithm; a dynamic collaboration degree index DSCI is calculated based on network self-adaption; and determining a collaboration degree threshold value through multi-level nesting identification, and obtaining a measurement report. According to the method, through constructing the space-time bidirectional weighted network and the dynamic coordination degree index DSCI adaptive calculation algorithm, the space-time dynamic coupling relation quantification precision of the urban space structure coordination measurement is improved. Meanwhile, a self-adaptive collaboration degree threshold combination is established through multi-level nesting recognition, and the dynamic adaptability of the collaboration evaluation standard to the urban development stage and policy environment change is improved.

The invention discloses a lightweight optical fiber vibration intrusion event identification method and system for perimeter security and protection, and belongs to the technical field of optical fiber sensing technology and mode identification, and the method comprises the steps: collecting an original vibration signal of a perimeter monitoring region through a distributed optical fiber vibration sensing system; performing wavelet threshold de-noising preprocessing on the original vibration signal to obtain a de-noised signal; performing feature extraction on the denoised signal, and constructing a high-dimensional feature vector; performing dimension reduction processing on the high-dimensional feature vector by using a linear discriminant analysis method to obtain a low-dimensional classification feature vector; and inputting the low-dimensional classification feature vector into a pre-trained lightweight convolutional neural network model for classification and identification, and outputting a corresponding intrusion event category. According to the method, through cooperation of front-end LDA dimension reduction and a rear-end lightweight network, the model parameter quantity and calculation overhead are greatly reduced while high recognition precision is guaranteed, efficient real-time deployment on edge equipment is achieved, and the method is suitable for intrusion detection in perimeter security and protection.

The invention provides a product full production cycle carbon footprint tracing evaluation method and system, and relates to the technical field of carbon footprint tracing, and the method comprises the steps: obtaining production activity data of each link of a product full life cycle, constructing a directed weighted network model, and executing path traversal, extracting a traversal path of which the path carbon intensity meets a preset path carbon intensity condition as a first carbon flow path, obtaining the node net carbon emission of each node, and determining the apportionment amount of each node to the third-range indirect emission; and matching the production activity data of the link corresponding to each node with the emission parameters in the carbon emission database to obtain the link carbon emission intensity of each link, and determining the link carbon emission of each link based on the link carbon emission intensity and the node net carbon emission. According to the method and the device, a small-flow high-carbon emission intensity path can be identified, and the range three-emission accounting accuracy is improved to the level of tracing to a specific supply chain node through a double-layer mapping relation of link level microscopic accounting and node level macroscopic tracing.

The invention provides a PageRank-based identification method and system for important nodes in a fused directed weighted network, and relates to the technical field of complex network analysis, the method comprises the following steps: constructing a fused directed weighted network, and fusing two single-layer directed weighted complex networks a and b by multiplexing partial nodes to form a fused network c; the output intensity of the node c of the fusion network is calculated according to the calculation formula that # imgabs0 # and # imgabs1 # are the output intensity of the node vi, and # imgabs2 # and # imgabs3 # are the number of the nodes, belonging to the single-layer directed weighted complex network a and the node number of the single-layer directed weighted complex network b, of neighbor nodes of the node vi; the importance value of a node v to be evaluated is iteratively calculated according to the calculation formula that # imgabs4 # sigma (0 < sigma < 1) is a damping coefficient, n is the total number of nodes of the fusion network, INR (vi) is the importance value of a node source vi pointing to the node v, and wc (vi, v) represents the directed edge weight from the node vi to the node v in the fusion network c; and according to the calculation result of the c node importance value of the fusion network, node importance sorting is carried out, and key nodes are identified.

The invention discloses a change segmentation method and system based on learnable DCT and frequency band division and a medium, and relates to the technical field of image processing, an improved AFBDCTFEM model is taken as a core, a learnable two-dimensional DCT transformation network, an adaptive frequency band division AFB network and a Band Router dynamic weighting network are connected in series to form an integral framework capable of end-to-end back propagation, and the integral framework is used for learning the two-dimensional DCT transformation network, the adaptive frequency band division AFB network and the Band Router dynamic weighting network. While the DCT energy concentration characteristic is maintained, the network is endowed with triple capabilities of tuning a frequency base as required, dynamically selecting a frequency band and automatically estimating a weight; and a substructure of an existing image change region segmentation network is replaced in an end-to-end mode, so that efficient extraction of dual-time-phase image difference features is realized, and the fineness and robustness of image change region segmentation are remarkably improved.

The invention relates to the field of perovskite photovoltaic technology, and discloses a perovskite solar cell interface layer optimization and stability enhancement method. The method comprises the following steps: firstly, acquiring real-time operation data of a battery, and screening a to-be-optimized interface layer set; based on an interface material attribute and a target stability threshold value, taking a battery overall structure as a center node, taking a to-be-optimized interface layer as a starting node and taking the target stability threshold value as a termination node, establishing an interface performance weighting network, and exporting a weight of an edge by a matching degree of the material attribute and the threshold value; in the network, aiming at the path of each starting node from a central node to a termination node, applying a self-adaptive optimization strategy to search an optimal solution, and identifying a plurality of optimization tracks; and generating an interface layer optimization scheme according to the optimal solution, and after a selection instruction is received, presenting an interface layer detailed material parameter record corresponding to the selected scheme. According to the method, interface layer optimization and stability enhancement are realized through network construction and optimization trajectory analysis.

The embodiment of the invention provides an external risk factor cause quantitative analysis method and system, and belongs to the technical field of risk assessment. The method comprises the steps that historical accident information is collected, accident cause analysis is performed on the historical accident information, and an external risk factor set is constructed and obtained; obtaining a corresponding relation matrix; constructing a directed weighted network based on the external risk factor set and the relation matrix, and constructing an evaluation system corresponding to the directed weighted network; constructing a corresponding external factor cause quantitative evaluation model; and evaluating the current operation state information of the target chemical enterprise based on the external factor cause quantitative evaluation model to obtain a safety risk evaluation result. According to the scheme, the problem that the external risk factor causes cannot be quantified in the safety risk assessment process of the chemical production process at present is solved, and the accuracy of the safety risk assessment result of the chemical production process is improved.

This application discloses a low-altitude airspace planning method and system based on micro-terminal areas, relating to the field of low-altitude airspace planning. The method includes: generating corresponding take-off and landing areas for low-altitude take-off and landing points or fields, and generating take-off and landing hub areas for adjacent take-off and landing points or fields to construct micro-terminal areas of the low-altitude airspace; constructing risk areas, which, together with control areas and no-fly zones, constitute the restricted areas of the low-altitude flight path network; generating alternative intermediate points connecting each take-off and landing point or field based on the restricted areas; constructing an initial low-altitude flight path network connecting all points based on the data of the alternative intermediate points and the data of the micro-terminal areas; calculating the corresponding weights of each edge line in the initial low-altitude flight path network, and optimizing the initial low-altitude flight path network based on the weighted network; and removing redundant intermediate points based on the optimized network, constructing the flight path network and the entry and exit points of each micro-terminal area to generate the final low-altitude flight path network. This application provides a more reliable technical basis for low-altitude flight path network planning.

The invention discloses a material creep behavior prediction method and device based on a physical mechanism weighted network, and aims to solve the problems of low prediction precision and poor physical interpretability of an existing data-driven model under a small sample condition. The method comprises the following steps: constructing a physical mechanism weighting network which comprises a plurality of weighting mechanism units; each weighting mechanism unit is formed by coupling a weighting layer and a physical mechanism layer, the physical mechanism layer is packaged with a preset creep physical mechanism model, and the weighting layer is configured to represent a competition and conversion relationship of different physical mechanisms under an external load condition through a learnable weighting function; training the physical mechanism weighting network by using creep experiment data of a material so as to determine parameters of the weighting function; and predicting the creep behavior of the target material under a given stress-temperature condition by using the trained physical mechanism weighting network. According to the method, high-precision and physically interpretable creep prediction under small samples and a wide load domain is realized.

The application discloses a face emotion recognition method based on a high-order recursive attention weighting network, which is a HRAM-CNN network model, and learning and training of face emotion recognition are performed, and then an image to be recognized is input into the trained HRAM-CNN network model to perform emotion judgment on each image; the HRAM-CNN network model is a high-order recursive attention mechanism HRAM which is put into a convolution layer of a convolution network model CNN. The application effectively combines the advantages of VisionTransformer and CNN, provides a high-order recursive attention mechanism combining the self-attention mechanism and the channel attention mechanism, and proposes that a large 7*7 kernel is used to construct the attention mechanism, so that adjacent information is effectively captured, and the feature extraction capability of the convolution process is further enhanced; the trained network can perform emotion judgment on each verification image.

The application discloses a physical information guided SAR airplane target detection method, comprising the following steps: constructing an airplane target slice data set and extracting scattering key points; based on a Gaussian mixture model, generating a probability density image corresponding to the airplane target slice as a true value heat map by using the scattering key points; constructing a target scattering key point prediction network and training the same, wherein the true value heat map corresponding to the airplane target slice and a predicted heat map predicted by the prediction network are used to calculate a network loss during the training process; constructing a feature reweighting network; combining the feature reweighting network and the trained target scattering key point prediction network to form a joint model, and connecting the joint model to a deep learning backbone network and a target detection network, so as to construct a physical information guided SAR airplane target detection network; and using a SAR airplane detection image data set to train the SAR airplane target detection network, and saving the trained network model for use in detecting and identifying unknown target categories of SAR airplane detection images.

The application discloses a traditional Chinese medicine compound analysis method based on a multi-dimensional data and information weighted network, and belongs to the field of traditional Chinese medicine compound analysis. The method comprises the following steps: constructing a multi-dimensional disease gene set; acquiring a traditional Chinese medicine compound set and a target set; determining disease gene nodes, compound nodes and target nodes respectively, and constructing a heterogeneous correlation network; performing propagation node screening on the heterogeneous correlation network, and weighting the edge set of the heterogeneous correlation network; running three algorithms on the updated heterogeneous correlation network respectively, and performing equal-weight average fusion on the propagation results of the three algorithms to obtain a final weighted network correlation degree score; sorting the final weighted network correlation degree score from high to low to obtain a key target candidate list; and selecting the first N candidate targets in the key target candidate list for enrichment analysis. The application solves the problems of single data source, lack of functional semantic support in network construction, static node evaluation index, and disconnection between enrichment analysis and core algorithm in the prior art.

The invention provides an intelligent grading and judging method for wooden furniture plates, which comprises the following steps of: constructing a high-quality image database which is acquired by an industrial camera at multiple angles and is bound with material labels, performing multi-stage denoising, enhancement and normalization preprocessing, extracting a key texture feature channel through a lightweight convolutional neural network and a channel attention mechanism, and performing classification and judgment on the key texture feature channel; according to the method, weight distribution and online optimization are carried out on feature channels in combination with a dynamic feature reweighting network, intelligent compensation of feature expression ability after lightweight pruning and quantification of the model is realized, in addition, a weight drift monitoring and local recalibration mechanism is arranged, the discrimination robustness of the model in detection of different batches of plates is guaranteed, and the detection accuracy is improved. According to the method, the edge deployment efficiency is improved, and meanwhile, the accuracy and adaptability of automatic judgment of the plate grade are enhanced.

The invention relates to a near-surface wind field downscaling method based on deep learning, and the method comprises the following steps: S1, obtaining site data, wind field data and topographic data, and outputting wind field and topographic total high-dimensional features by a feature extraction module; s2, inputting the total high-dimensional features of the wind field and the terrain into a multi-scale feature fusion module, and outputting multi-scale fusion features; s3, outputting a first near-surface wind field grid prediction value; s4, a second full-connection prediction module outputs a second near-surface wind field grid prediction value; and S5, obtaining adaptive weighted analysis data, inputting the adaptive weighted analysis data into the adaptive weighted network, and obtaining a final surface wind field grid prediction value based on the first weight and the second weight. Compared with the prior art, the near-surface wind field downscaling method has the advantages of improving the accuracy and efficiency of near-surface wind field downscaling and the like.

The disclosure provides a traditional Chinese medicine efficacy evaluation method and device based on a node-weighted network, equipment and a storage medium. The method determines target disease protein targets and corresponding target weights based on the comprehensive scoring results of candidate disease protein targets by a public database and a large language model. When determining the protein targets of medicinal materials and the corresponding protein target weights of medicinal materials, the prescription ratio, chemical component information, and the interaction probability between the chemical components and the protein targets of each medicinal material in the traditional Chinese medicine prescription to be evaluated are comprehensively considered. The target disease protein targets and the corresponding disease protein target weights, as well as the protein targets of medicinal materials and the corresponding protein target weights of medicinal materials, are added to a pre-constructed protein interaction network. The obtained node-weighted network focuses on the real pharmacological basis, effectively improves the network biological interpretation, and the multi-dimensional network index determined accordingly is used to evaluate the regulation effect of the traditional Chinese medicine prescription to be evaluated on the target disease, which is more accurate.

The application is suitable for the technical field of charging station site selection, and provides a method for electric vehicle charging station site selection and capacity optimization based on structural entropy, comprising: constructing an initial feature vector of a candidate parking lot based on business information and geographic location information of the candidate parking lot; constructing a weighted network graph and initializing the weighted network graph as a coding tree; calculating the structural entropy of the candidate parking lot based on the coding tree, generating an enhanced feature vector based on the structural entropy and the initial feature vector; clustering the candidate parking lot based on the enhanced feature vector to obtain multiple clusters, and defining each cluster as a demand area; constructing an energy utilization rate function based on the demand area, taking the energy utilization rate function as a fitness function of a genetic algorithm; and optimizing the electric vehicle charging station construction scheme by using the genetic algorithm to obtain an optimal electric vehicle charging station construction scheme which maximizes the fitness function and meets the construction constraint condition. The application can improve the energy utilization rate so that the energy can be fully utilized.

The application discloses a subway network resilience evaluation method and device, and relates to the technical field of traffic resilience evaluation. The method comprises the following steps: constructing a directed weighted network model of a subway network according to subway stations and running routes; determining subway network variable indexes according to the directed weighted network model, and performing initial load distribution of nodes of the directed weighted network model according to the subway network variable indexes; performing iterative attacks on the directed weighted network model with the initial load distributed, re-distributing node loads after each attack, and calculating subway network efficiency after each attack according to re-distribution results; and evaluating subway network resilience according to the subway network efficiency after each attack. The application can more accurately identify key nodes in the network, the evaluation result of the network resilience is more comprehensive and reliable, and the operation efficiency and management level of the subway network are improved.

The invention relates to the field of behavior detection, in particular to a chronic pain protection behavior detection method based on a hierarchical weighted graph convolution model. The method comprises the following steps: acquiring a training set, wherein the training set comprises joint movement data of a plurality of healthy individuals and a plurality of chronic lower back pain individuals in the process of completing daily life tasks; preprocessing the data in the training set, and converting the data into three-dimensional time sequence skeleton data; training a detection model by using the preprocessed training set; the detection model converts the human skeleton image into a layered image structure, and features are extracted by using an image convolutional network; introducing a category-based weighting mechanism, and optimizing a classifier and a weighting network through iterative adversarial training; and after the joint movement data of an individual to be detected in the process of completing the daily life task is preprocessed, the trained detection model is input for detection. According to the method, the receptive field is expanded through a layered structure, the category imbalance problem is optimized through a reweighting mechanism, and the robustness and accuracy of detection are improved.

This invention discloses a microblog group identification method based on community discovery, comprising the following steps: S1, data collection and cleaning; S2, feature extraction and representation; S3, establishing a classification model; S4, community tagging and influence analysis. In this invention, an optimized Dynamic Topic Model (DTM) is used to mine specific groups within the microblog community. Microblog posts from the past year are selected as the research object, and the similarity of topics in posts from different authors is used as the weight of links between authors, mapping the microblog network into a directed weighted network. Community discovery is performed using the Label Propagation Algorithm (LPA), identifying the inherent community structure within the social relationship network. This invention conducts in-depth analysis of user relationships within the microblog network, and based on identification methods for user-generated content characteristics, user association characteristics, and environmental characteristics, it mines potential topics to identify users with similar interests and active user groups in specific fields.