26 results about "Cluster group" patented technology

The application discloses a kind of based on risk scene's electric power market parallel clearing method and system, it is characterized by: S1, multi-dimensional risk scene synchronization generation;S2, scene space-time feature extraction;S3, scene clustering group;S4, distributed parallel clearing calculation;S5. Risk index aggregation analysis.The application improves risk coverage by scene coupling generation, solves the problem that scene and clearing calculation are separated and cannot capture multiple risk coupling effects;Through distributed architecture for clearing calculation, improve the efficiency of calculation;Through risk index, the cross-market chain reaction effect can be quantified.The application improves risk coverage by scene coupling generation, solves the problem that scene and clearing calculation are separated and cannot capture multiple risk coupling effects;Through distributed architecture for clearing calculation, improve the efficiency of calculation;Through risk index, the cross-market chain reaction effect can be quantified.

The application discloses an adaptive clustering federated learning modeling method for precision medicine, and relates to the technical field of precision medicine, and comprises the following steps: data collection and preprocessing and model construction and training; the application accurately determines the optimal number of the global model through adaptive clustering, and divides clustering groups according to the similarity of the client model parameters, realizes independent training of the groups, effectively improves the adaptability of the model to different client data characteristics, samples and label distribution differences, avoids the problem that a single model has poor performance in some clients, and significantly enhances the ability of the model to capture complex medical patterns; the model generalization is optimized through grouped training, so that the model can better cope with new data distribution, under the premise of ensuring the privacy and security of medical data, the accuracy and reliability of the model in precision medical scenes such as disease diagnosis and prognosis prediction are greatly improved.

The present application relates to the technical field of cluster communication, and proposes an intelligent fusion cluster group call method, steps of which include: configuring group members and member types thereof, wherein the member types at least include a cluster terminal, a dispatch station, and a non-cluster terminal, wherein the non-cluster terminal includes at least one of a traditional fixed phone, a mobile phone and a smart camera; a cluster core network performs corresponding media adaptation operation and right adaptation operation on different types of group members in a group call establishment process and a group call process according to the pre-configured member types, so that the non-cluster terminal can access and participate in the cluster group call service. The present application enables the public network terminal and the intelligent device originally not supporting the group call and the right operation to be integrated into the cluster communication process, expands the terminal compatibility and the service coverage range of the cluster system, and meanwhile maintains the necessary media control and right management characteristics in the cluster dispatch.

The application is suitable for the technical field of computers, and provides a federated learning method for heterogeneous clients, comprising: distributing an initial global model to all client models for training, and collecting model update parameters of all client models after training; based on the model update parameters, constructing multi-view feature representation; based on the multi-view feature representation, using a spectral clustering method for client clustering grouping to obtain multiple client clusters; using a two-stage aggregation method to aggregate each client cluster to obtain global gradient parameters, iteratively update the initial global model, and obtain a final global model. The application solves the problems of poor single-view gradient clustering accuracy in the first round of existing federated learning and non-discriminative aggregation of old client gradients, realizes high-precision clustering grouping and dynamic fair aggregation under the premise of not leaking original data privacy, and significantly improves the generalization ability of the global model and the training efficiency of large-scale systems in a heterogeneous environment.

Systems and methods for determining how topics evolve in a messaging system extract at least one N-gram from data content (e.g., caption of messages) in the messaging system and detect anomalous behavior in N-gram frequencies over time. The anomalous behavior is used to select candidate N-grams for a determination of whether a topic of a candidate N-gram is evolving or fading. The candidate N-grams are clustered into cluster groups that are used to train at least one time series forecasting model to predict N-gram frequencies in a future time window. A time series of the N-gram frequency is divided into old and recent partitions and pattern recognition is applied to the predicted N-gram frequencies to identify an evolving or fading topic when the difference between a frequency of each anomaly and an average rolling median for each partition is greater for the most recent partition.

This invention discloses a satellite trunking terminal user information publishing and comment / like interaction system. After a satellite terminal forms a cluster social circle through online cluster grouping and adding friends, in addition to supporting one-call / group calling and real-time SMS services, it will also support value-added services such as terminal users publishing information to their group user circle and friend user circle, and receiving comments and likes. The terminal APP adds two communication sections, group activity and friend activity. Clicking on these sections displays information published by associated users, which can be in the form of SMS messages, data, or status emoticons. Users can comment or like these messages. Users can also actively publish information and receive comments and likes from other users. Associated information is exchanged bidirectionally between the satellite terminal and the ground station core network. The satellite trunking air interface data transmission path ensures the integrity and reliability of the interactive data, allowing users to publish information and comment / like simply by clicking to send to the satellite, adding new application functions to trunking communication.

This invention discloses an industrial network security risk assessment method, belonging to the field of network security assessment technology. The method includes: acquiring standard traffic data and preset threat label data of an industrial wireless network; extracting traffic feature sets; obtaining feature cluster grouping datasets through distributed clustering across multiple working servers; selecting key feature datasets by combining threat labels; obtaining the optimal feature dataset through dimensionality reduction using an autoencoder and inputting it into a traffic detection model to obtain detection results; conducting risk assessment based on the results to obtain quantitative values; and determining the optimal combination of defense strategies accordingly. By using distributed clustering to process traffic features, combining threat labels to filter key information, and optimizing through dimensionality reduction using an autoencoder, the method dynamically quantifies risks and matches the optimal defense strategy based on anomaly detection results, thereby achieving accurate risk assessment and adaptive defense of industrial wireless networks and improving the operational security of industrial wireless networks.

The application provides a multimedia data security measurement method based on multi-dimension correlation. For multimedia data, a plurality of features of a single image in the multimedia data are divided into a plurality of clustering groups based on spatial similarity by a spectral clustering algorithm. Whether the multimedia data is safe is judged by combining content security score, privacy information security score, semantic security score and formal security score, so as to realize security evaluation and optimization of calculation efficiency.

The application belongs to the technical field of cloud computing, and particularly relates to a cloud computing engine resource scheduling method and system based on container technology, which comprises the following steps: capturing container resource consumption trajectories in real time and identifying business feature states; performing load correlation quantitative evaluation by calculating cross-correlation coefficients and maintaining a correlation matrix; performing scheduling decisions based on dual constraints of physical space bearing capacity and load coupling degree, and introducing a dynamic stability buffer threshold to determine a deployment position; and realizing closed-loop rebalancing adjustment by using a feedback compensation link, which can effectively avoid cluster group resonance effects, improve resource utilization, and guarantee system stability.

The application discloses a clustering method and device based on pairwise comparison results, a storage medium and equipment, and belongs to the field of image recognition. According to the comparison score matrix obtained by pairwise comparison of to-be-clustered samples, a weak judgment matrix and a strong judgment matrix are generated. The element value of each position of the weak judgment matrix and the strong judgment matrix indicates whether the two to-be-clustered samples corresponding to the position belong to the same clustering group under the corresponding probability. A preliminary clustering group is obtained according to the weak judgment matrix, and the clustering group is corrected according to the strong judgment matrix to obtain a clustering group result. Compared with hierarchical clustering, the clustering method reduces the complexity of the algorithm and can quickly perform clustering grouping. The clustering process is all matrix calculation, and the calculation is more rapid and efficient. The clustering method does not need to set a merging rule and a stop condition. Multiple comparison features can be selected, the strong and weak judgment matrices can be flexibly adjusted, multiple similarity index problems can be flexibly handled, and the samples can be quickly and accurately clustered.

Embodiments of the present application provide a file processing method and device, electronic equipment and storage medium. The method comprises: determining a connected component to which each newly added sub-file belongs, any two sub-files belonging to the same connected component being able to form an association through direct or indirect relationship; performing file clustering on each sub-file under each connected component to obtain a plurality of file clustering groups, and determining a main file corresponding to each file clustering group, the main file corresponding to each file clustering group being a reference sub-file in each sub-file included in each file clustering group; and determining a file identifier of the reference sub-file as a file identifier associated with the file clustering group to which the reference sub-file belongs. The present scheme realizes local file clustering on each sub-file under each connected component, avoids error accumulation caused by file centroid deviation, and improves file merging efficiency.

The present application relates to the technical field of self-adaptive equipment linkage control of underground coal mine environment, and particularly relates to a coal mine equipment intelligent linkage control method based on personnel positioning.A coal mine equipment intelligent linkage control method based on personnel positioning, comprising the following steps: S1: dividing the edge node area of the underground coal mine according to the explosion-proof type, determining the environment sensitive area and the environment stable area, performing load testing on the edge computing node in the environment sensitive area to determine the corresponding curve type; S2: determining all adjacent areas of each environment sensitive area in the edge node area of the underground coal mine, obtaining the cross-area coordination task success rate, and determining the edge computing capability of the environment sensitive area according to the cross-area coordination task success rate.The present application adjusts the start-stop strategy through area division, sensitive identification and clustering grouping, improves the positioning linkage adaptability, and guarantees the continuous production of the coal mine.

A method of identifying stairs from footfalls includes receiving a plurality of footfall locations of a robot traversing an environment. Each respective footfall location indicates a location where a leg of the robot contacted a support surface. The method also includes determining a plurality of candidate footfall location pairs based on the plurality of footfall locations. The candidate footfall location pair includes a first and a second candidate footfall location. The method further includes clustering the first candidate footfall location into a first cluster group based on a height of the first candidate footfall location and clustering the second candidate footfall location into a second cluster group based on a height of the second candidate footfall location. The method additionally includes generating a stair model by representing each of the cluster groups as a corresponding stair and delineating each stair based on a respective midpoint between each adjacent cluster group.

The application discloses a kind of distribution transformer load prediction method and device, based on self-encoder to distribution transformer historical load data carries out time series feature extraction, obtains load feature vector, carries out clustering to load feature vector, obtains several clustering groups, respectively to the clustering group constructs load time series sample library, by combining the time series convolution network based on attention mechanism and convolution neural network trains the target load prediction model of each clustering group, can make full use of the structural similarity between clustering internal distribution transformer load time series data, effectively capture distribution transformer load multidimension, nonlinear and long short-term dependence, significantly improve the expression ability and prediction accuracy to complex load mode, solve the existing distribution transformer load prediction method difficult to effectively capture the deep-seated characteristics of distribution transformer load, and ignore the time series similarity between different distribution transformers, leading to prediction model difficult to adapt to diversified load mode, the technical problem of low reliability.

The application relates to the technical field of artificial intelligence, can be applied to business system platforms such as medical health and financial technology, and discloses a resource pool management method based on clustering analysis, which comprises the following steps: acquiring multidimensional data, pre-processing the multidimensional data, and outputting numerical characteristics; acquiring industry characteristic data, updating parameters of a clustering model based on a contour coefficient, an elbow rule and the industry characteristic data, inputting customer resources into the updated clustering model, classifying the customer resources based on the numerical characteristics by the clustering model, and outputting clustering grouping results; scoring the customer resources according to the clustering grouping results and customer dynamic indexes, generating customer dimension indexes, scoring business personnel according to business personnel dynamic indexes, generating business dimension indexes, calculating priority evaluation results through the customer dimension indexes and the business dimension indexes; and matching optimal selected business personnel for to-be-matched customers according to the priority evaluation results.

This invention relates to a cutting path optimization method based on dynamic calculation of the starting point of the contour, belonging to the field of CAD / CAM technology. The method includes: acquiring clustered data of parts after layout; calculating the invariant moments of the part contours based on the geometric moments of the contours, and performing similarity clustering; initializing the cutting path; prioritizing the selection of parts within the same cluster group, dynamically calculating the distance from key points on the contour of each part to the current cutting head position, selecting the part with the shortest distance and its optimal starting point, updating the path sequence; and outputting the final path. It also relates to a corresponding system, a computer-readable storage medium, and CNC cutting equipment, wherein the system includes modules for data acquisition, graphic similarity clustering, dynamic calculation, path generation, and output. This invention significantly reduces idle movement paths and improves processing efficiency through graphic similarity clustering and dynamic starting point calculation, and is applicable to CNC equipment such as laser cutting and plasma cutting.

This invention discloses a multi-vehicle accident recognition method based on spatial clustering and static feature analysis. It obtains long-exposure images by accumulating video frames over a given time window. Based on these long-exposure images, it calculates cumulative confidence, pixel change rate features, and trajectory IOU scores. For each feature dimension of the three-dimensional feature vector, K-means clustering is performed based on the feature values ​​of the cluster centers to identify truly stationary vehicles and slow-moving vehicles. The detection boxes corresponding to truly stationary vehicles are expanded into multiple spatial cluster groups. Further analysis of the lane distribution of vehicles within these spatial cluster groups initially identifies multi-vehicle accidents. The bounding rectangles of all vehicle detection boxes within each cluster group are then calculated as cluster regions. On the long-exposure image, the number and location of vehicles within these cluster regions are detected. Finally, the cluster groups corresponding to multi-vehicle accidents, traffic congestion, and single-vehicle abnormal parking labels are determined, solving the problem of single-feature-based judgment failure in complex scenarios.

Embodiments provide methods and systems for physical uplink shared channel (PUSCH) power control. A user equipment (UE) may associate with a base station (BS), measure reference signal received power (RSRP) data, report the RSRP data to the BS, receive P_0 and P_UE values from the BS, and transmit signals to the BS using P_TX as defined by a PUSCH. A network may initialize a BS with default power values and default cluster groups per cluster, transmit new power settings to the first BS, collect network data from the first BS, validate requirements, update power allocation among UEs, and validate power allocation among the UEs. UEs may be clustered (grouped) based on measurement data, e.g., their reported RSRP values to the first BS and a neighbor BS.

The application provides a kind of identification method and device for commercial aggregation area, comprising obtaining the scene data, performance data, MDT data and work parameter data of the area to be identified after preprocessing, inputting the scene data, performance data, MDT data and work parameter data into target decision tree model to obtain predicted scene data, predicted performance data, predicted MDT data and predicted work parameter data with data label as commercial nature;Based on predicted work parameter data, the predicted scene data is supplemented with scene missing data to obtain target scene data;The comprehensive total score data is obtained by scoring calculation on predicted performance data and predicted MDT data;The density clustering grouping result is obtained by density clustering grouping on target scene data and comprehensive total score data, and the commercial aggregation area is identified from the area to be identified according to the density clustering grouping result, so that the identification result is more convincing, and the identified commercial aggregation area is more accurate through comprehensive scoring combined with scene data.

This invention discloses an unsupervised cross-modal pedestrian re-identification method based on hierarchical clustering and edge-noise learning, comprising: acquiring visible light and infrared images of pedestrians, and extracting visible light and infrared feature vectors; for each visible light feature vector, retrieving the optimal matching infrared feature vector in the infrared feature space, constructing a unified graph and performing feature propagation to obtain enhanced feature vectors; then performing clustering to divide noisy nodes into noisy nodes and non-noisy nodes; constructing boundary collaborative features among noisy nodes of different modalities in the same cluster; filtering noisy nodes based on dynamic thresholds and performing weighted aggregation to form noise collaborative features; constructing a truncated Gaussian distribution to simulate the joining and leaving of pedestrians, treating all pedestrians in the same cluster as a pseudo-cluster group, and obtaining the pseudo-cluster group center; constructing a loss function and training it. This invention focuses on explicitly mining and coordinating boundaries and noise based on existing pseudo-labels to enhance robustness.

This invention relates to the field of power grid control technology, specifically to an intelligent power distribution system and method for monitoring and controlling new energy transformer substations. The system includes: a data acquisition unit for acquiring multi-dimensional operating parameters of each transformer substation unit and status perception parameters of each monitoring node, wherein the multi-dimensional operating parameters include load fluctuation coefficient and reactive power compensation; and a transformer substation clustering unit for clustering all transformer substation units based on the deviation of multi-dimensional operating parameters between different units to determine the transformer substation cluster groups. This invention, through multi-dimensional parameter monitoring and analysis of each transformer substation unit, can classify and cluster transformer substation units according to different operating states and load fluctuation characteristics. This analysis method based on multi-dimensional data and power grid topology can achieve more accurate operating state identification, thereby optimizing the management and operation of the power distribution system and improving the overall reliability and stability of the power system.

The application discloses an unsupervised cross-modal pedestrian re-identification method based on hierarchical clustering and edge noise learning, comprising the following steps: collecting visible light images and infrared images of pedestrians, and extracting visible light feature vectors and infrared feature vectors; for each visible light feature vector, searching for the optimal matching infrared feature vector in the infrared feature space, constructing a unified graph and performing feature propagation to obtain an enhanced feature vector; then, clustering is performed to divide noise nodes and non-noise nodes; boundary collaborative features are constructed between noise nodes of different modalities in the same cluster; noise nodes are screened based on a dynamic threshold, and weighted aggregation is performed to form noise collaborative features; a truncated Gaussian distribution is constructed to simulate the joining and exiting of pedestrians, all pedestrians in the same cluster are regarded as a pseudo-cluster group, and a pseudo-cluster group center is obtained; and a loss function is constructed and trained. The application focuses on explicitly mining and coordinating boundaries and noises on the basis of existing pseudo-labels to enhance robustness.