32 results about "Density based clustering" patented technology

The invention discloses an analysis method and system based on AI video recognition behavior monitoring, and relates to the technical field of video recognition, and the method comprises the steps: collecting video data, calculating the overall pixel gradient change of adjacent frames, marking preliminary candidate frames, calculating the edge saliency intensity of each frame, and analyzing a salient region point set; clustering and calculating the geometric center point of each cluster as an anchor point value, and forming an anchor point set through the center points of each cluster; according to the method, concentrated extraction of salient points is realized through a density-based clustering algorithm DBSCAN, so that an extraction result has logic uniqueness and local integrity of a salient target, a local feature map is extracted by using a shallow convolutional network, target texture and edge information is effectively captured, bounding box optimization is realized by rotating an IoU loss function, and the robustness of the target is improved. And the fitting precision of the boundary of the salient region is improved.

The application discloses an environmental monitoring data analysis method and related device, relates to the technical field of environmental monitoring, obtains the standard sample data consistent with the attribute of the to-be-analyzed environmental monitoring data, calls a density-based clustering algorithm to process the two kinds of data, obtains a clustering result, and in the case that a turning point in a K-distance graph of the to-be-analyzed environmental monitoring data does not satisfy a preset turning point condition, determines a reference turning point by using a sliding window algorithm, so that the clustering result is more accurate; when the clustering result satisfies a condition, multi-dimensional quality analysis is performed on the to-be-analyzed environmental monitoring data, and the efficiency and accuracy of environmental monitoring data analysis are further improved; when the to-be-analyzed environmental monitoring data is subjected to multi-dimensional quality analysis, at least data stability and correlation of the to-be-analyzed environmental monitoring data are analyzed, so that the accuracy of an analysis result obtained by multi-dimensional analysis is greatly improved, and the accuracy of application of the analysis result to other scenes, such as data quality evaluation, is further improved.

According to an aspect, there is provided a computer-implemented method comprising the following. Initially, information on a plurality of anomaly events relating to operation of a target system is obtained. The information comprises one or more time series of anomaly score data. Segments satisfying one or more pre-defined criteria for anomalous operation are detected from the one or more time series. The one or more pre-defined criteria are defined to exclude fully non-anomalous anomaly score data. Anomaly durations and standardized anomaly scores are determined for the segments. Partition or density based clustering is performed in a two-dimensional space formed by the standardized anomaly scores and the anomaly durations to form n clusters, and m smallest clusters of the n clusters are identified. At least one of the following is performed: outputting information on the m smallest clusters or causing adjusting of operation of the target system based on the m smallest clusters.

This invention discloses a method and system for identifying cross-connection issues in low-voltage distribution areas. It acquires voltage and current time-series data of users in the low-voltage distribution area, cleans the data using a mask interpolation strategy, and adaptively extracts voltage drop and current surge features based on the median absolute deviation algorithm. It constructs a voltage trend similarity model based on local regularization kernels and a mutation co-occurrence similarity model based on event level weights, employing a dual-track strategy combining daily rigorous assessment and feature revival to generate suspected voltage correlation pairs. A density-based clustering algorithm is used to generate an initial population, and a voltage similarity-weighted centroid screening mechanism is introduced to remove loose outliers in the feature space. A pairwise cross-validation model is constructed based on electrical causality, and the population is topologically refined using relative voltage response criteria and a maximum fully connected clique extraction algorithm to determine the final cross-connection population. This invention significantly improves the accuracy and robustness of cross-connection identification.

The application discloses a Tor network hidden malicious organization mining method based on multi-dimensional fusion detection. The method comprises the following steps: crawling the network state consensus published by Tor Metric in the latest half year, sorting the online relay list and mining the relay related information; performing multi-dimensional security evaluation based on the relay list and the characteristics, and screening out suspicious relays with abnormal scores higher than a threshold; comprehensively using the relay server descriptor file and the network public asset mapping platform information to obtain the internal attribute behavior and external intelligence features of the relay node, and modeling the similarity of relay pairs; using a random forest algorithm to calculate the correlation degree of relay pairs; based on the correlation degree of relay pairs, using an OPTICS clustering algorithm to perform density-based clustering, and dividing the suspicious relays into organizations. The application combines the internal attributes and external intelligence features of the relay node, respectively uses the similarity modeling and clustering algorithm to calculate the correlation degree of relay pairs and divide the organizations, and realizes efficient identification of the Tor network hidden malicious organizations.

This invention discloses a programmable data plane high-intensity traffic response method and system based on feature distribution, belonging to the field of network traffic classification technology. The programmable data plane high-intensity traffic response method based on feature distribution includes: extracting features from each flow in the data plane and determining whether the features are matched by the current rule; if matched, processing according to the corresponding rule; if not matched, processing as an outlier and recording it; periodically sampling processed flows, triggering rule updates when outliers reach a preset threshold; the control plane uses a density-based clustering algorithm to cluster sampled points to obtain the current traffic feature distribution shape, extracting the boundaries of each cluster to form a high-dimensional rectangle as a new rule, and formulating corresponding processing measures based on spatial features and meta-features, and issuing them to the data plane. This invention achieves line-rate processing and dynamic adaptation on resource-constrained programmable hardware, effectively addressing feature drift under high-intensity traffic while ensuring interpretability.

The present disclosure provides a road generation method, device, equipment and storage medium, relates to the field of data processing, in particular to the field of electronic map, cloud computing and big data technology. The specific implementation scheme is: obtaining position data to be processed; performing clustering analysis on the position data by using a density-based clustering algorithm to obtain data clusters; obtaining longitude and latitude coordinates of boundary points in the data clusters to obtain a boundary point set; connecting the longitude and latitude coordinates in the boundary point set in sequence by using a boundary point discovery algorithm, and drawing a road graph according to the connected boundary points. Through the above scheme, the development and operation cost is saved, and the flexibility and applicability of road generation are ensured.

The invention discloses a crack three-dimensional accurate quantification method and system based on semantic point clouds, which are used for solving the problems of skeleton extraction distortion, tail end missing, noise sensitivity and insufficient multi-scale fusion in the existing crack quantification method, and the method comprises the steps: obtaining three-dimensional point cloud data with crack semantic tags, and carrying out the denoising and smoothing preprocessing; segmenting independent crack individuals through a density-based clustering algorithm; dynamically extracting skeleton points by adopting a curvature weighted L1 median algorithm, and adaptively adjusting the point density according to the curvature; skeleton end points are complemented from the original point cloud through main direction projection; calculating crack length, width, direction and type parameters based on the complete skeleton point set; the system correspondingly comprises a point cloud preprocessing module, a segmentation module, a skeleton extraction optimization module, a parameter calculation module and the like. According to the method, high-precision and automatic three-dimensional quantification of a crack structure with complex bending and strong noise interference is realized, and the accuracy, the integrity and the anti-noise capability of skeleton extraction are remarkably improved.

This invention discloses a programming pattern mining method and system based on a large language model, belonging to the fields of software engineering and code analysis technology. The method includes: scanning and filtering target code repositories to select valid source files; constructing an abstract syntax tree using a static parser and extracting candidate code fragments at three granularities: function level, statement level, and interval level; vectorizing the code fragments using a code embedding model, assembling the AST structure and semantic vectors into meta-information; identifying high-frequency code patterns with semantic similarity using a density-based clustering algorithm; abstracting and generalizing variable elements through sliding window consistency analysis to generate generalized code templates; collaboratively determining the programming patterns of candidates using a multi-agent system, ultimately outputting a code template library; and constructing the code template library into a RAG retrieval knowledge base for use in software engineering tasks such as unit test generation, programming standard recognition, and code completion.

The invention relates to the field of power grid operation and maintenance, and discloses a power grid abnormity positioning method and device and a medium, and the method comprises the steps: obtaining the new energy generating capacity according to the obtained meteorological data, and obtaining the parameters of a generator set; obtaining power grid simulation time sequence data according to the new energy generating capacity and the generator set parameters; clustering the power grid simulation time series data by adopting a density-based clustering algorithm; adopting an isolated forest algorithm for the clustered power grid simulation time series data to obtain an isolated forest model; and according to the isolated forest model and the acquired power grid time sequence real-time data, positioning an anomaly in the power grid time sequence real-time data. According to the method, the abnormal region can be quickly and accurately positioned in a large amount of power grid operation data, and the complexity of the operation state of a novel power system is effectively dealt with.

To provide a machine-learning program, a machine-learning device, and an information-processing system that enable stable AI operation.SOLUTION: An inference result obtained by inputting operation data to a learned machine learning model trained based on learning data is collected, density-based clustering is executed on the collected inference result to generate a cluster, an estimated label corresponding to the cluster is estimated for each cluster from correct labels corresponding to all correct answers that can be the inference result, and the learned machine learning model is fine-tuned based on the operation data belonging to the cluster and the estimated label corresponding to the cluster.SELECTED DRAWING: Figure 2

The application provides a panel defect point aggregation detection method, system, device and storage medium, and relates to the technical field of defect detection. The method comprises the following steps: acquiring characteristic data of a panel, wherein the characteristic data comprises coordinate information of all defect points on the surface of the panel; performing clustering and division processing on the characteristic data of the panel by using a density-based clustering algorithm to obtain a plurality of clustering regions; performing region merging processing on the plurality of clustering regions to obtain at least one merged region; and comprehensively detecting the defect points of the panel based on the clustering regions and the merged region to obtain a defect point aggregation detection result. The density-based clustering algorithm is applied to the industrial panel defect detection, and the defect point aggregation condition can be more intuitively considered by introducing the defect point merging adjustment between clusters and the total number threshold judgment of the defect points in the cluster, so that the accuracy and reliability of the defect point aggregation detection are improved.

The application discloses a multi-person pose estimation method based on a millimeter wave radar, and specifically comprises the following steps: acquiring original multi-person point cloud data and real-time joint data, calibrating the two, so that they are in the same three-dimensional space coordinate system, obtaining three-dimensional coordinates of the real-time joint and taking the three-dimensional coordinates as label data; processing the original multi-person point cloud data by using a density-based clustering algorithm, removing noise points and separating the point cloud, obtaining separated multi-person point cloud data for constructing a data set; after preprocessing independent point cloud data to be processed, inputting the preprocessed independent point cloud data into a trained point cloud-based pose estimation neural network to obtain a three-dimensional coordinate prediction value of the real-time joint; the pose estimation neural network is sequentially cascaded by a space embedding layer, a first feature extraction layer, a second feature extraction layer, a global feature extraction layer and a multi-branch full connection layer, and the pose estimation neural network is trained on the data set.

The invention discloses a power grid database security protection method and device, equipment and a medium, and relates to the technical field of data security access, the method comprises the following steps: identifying a plurality of groups of historical access logs of a power grid database to obtain a plurality of historical access behaviors, and analyzing and extracting to obtain corresponding access behavior characteristics; performing density-based clustering analysis on all the access behavior characteristics to obtain a plurality of behavior modes; predicting a risk coefficient of each behavior pattern based on a long short-term memory network model, and classifying the plurality of behavior patterns into a plurality of risk levels; matching a first behavior mode corresponding to the plurality of behavior modes according to the current access behavior, determining a protection strategy according to a first risk level determined by the first behavior mode in combination with the behavior intention of the current access behavior, and performing security protection; wherein the behavior intention is obtained based on analysis of the current access behavior. According to the method, the access behavior can be accurately identified, so that the power grid database is reasonably and safely protected.

The invention discloses an identification method and system for power line tree obstacle cleaning, and the method comprises the steps: obtaining natural tree node information of a target region of a proposed power line and a spatial path of the proposed power line, carrying out the spatial superposition analysis based on the natural tree node information and the spatial path, screening out a potential tree obstacle object, and carrying out the recognition of the potential tree obstacle object. Selecting a geometric volume model of the potential tree obstacle object according to the natural tree node information, calculating a predicted compensation amount of the potential tree obstacle object by using a density-based clustering algorithm in combination with the geometric volume model, and performing cleaning judgment on the potential tree obstacle object based on the natural tree node information to obtain a tree obstacle cleaning strategy; according to the method, the potential tree barrier objects can be screened out without omission, clustering and volume calculation are carried out on the potential tree barrier objects based on natural tree node information in combination with a density-based clustering algorithm and a geometric volume model, the compensation amount of tree barrier cleaning can be accurately and efficiently estimated, and therefore the tree barrier needing to be cleaned and the compensation amount can be comprehensively and accurately recognized.

The invention provides an intelligent desk sorting method and device, equipment and a storage medium, and the method comprises the steps: receiving a desk distribution image in a classroom, and extracting the coordinate of each desk; coordinates of all desks are regarded as a two-dimensional plane discrete point set, a convex hull capable of surrounding the point set is calculated, and vertex coordinates of the convex hull are output; determining an angular point desk from the vertex coordinates of the convex hull; based on the determined angular point desks, performing perspective transformation on the desk distribution image to obtain corrected coordinates of all desks; performing column division on the corrected coordinates of all desks by adopting a density-based clustering algorithm and taking the abscissa of the corrected coordinates of the desks as a clustering feature to obtain a plurality of column clusters; and carrying out row and column coordinate coding on the position of each desk in the desk distribution image based on the plurality of column clusters. The method has higher environmental adaptability and sorting accuracy, and complex scenes such as irregular layout and local shielding can be effectively processed.

The application discloses a mechanical arm polishing track generation method and related devices, and relates to the technical field of machine vision, the method comprises the following steps: based on the polishing demonstration video, the motion track and the posture information of the polishing demonstration object are analyzed by using a posture estimation model; based on the motion track and the posture information, a polishing contact point track is generated, and filtering processing is performed on the polishing contact point track; a density-based clustering algorithm is used to extract the start and end points of the polishing contact point track, and the polishing contact point track is down-sampled based on the start and end points; based on a hand-eye calibration matrix, the point cloud data of a workpiece to be polished and the down-sampled polishing contact point track are used to generate a polishing track in a mechanical arm base coordinate system, and the polishing track is coarsely optimized by using an optimization function; the coarsely optimized polishing track is finely optimized, and a target polishing track of the mechanical arm is obtained. The application guarantees high precision and high efficiency of the generated robot polishing track, and reduces the professional requirements for generating the polishing track.

The application relates to a photovoltaic module abnormality monitoring and fault diagnosis method based on a clustering algorithm; multi-dimensional running time series data of a target photovoltaic module in a preset historical time period is collected, and original data collected is preprocessed to form a regular time series data set; based on the preprocessed time series data set, a feature vector for clustering analysis is constructed, the feature vector is input into a clustering algorithm, the historical normal running state of the photovoltaic module is clustered, and one or more clustering models corresponding to different normal working conditions are established; firstly, the application uses a Gaussian mixture model to accurately depict the complex normal behavior mode of the photovoltaic module under multiple working conditions, and establishes a flexible and robust normal state benchmark; and then, an abnormal data exploratory analysis is conducted through a noise-based density clustering method, unknown fault modes are automatically found and distinguished.

The invention relates to a local density sensing buffer insertion method, and belongs to the field of super-large-scale integrated circuits. The method comprises the following steps: uniformly dividing a physical design area of a chip into a plurality of rectangular grids, traversing and calculating local layout density values of all the grids, and forming a digital density map; based on the digital density map, high-density areas are identified by adopting a density-based clustering algorithm, and each identified high-density area is marked as a density obstacle; aiming at a target line network, generating an interconnection tree capable of avoiding all density obstacles, and generating a series of buffer insertion candidate points on a path of the interconnection tree; an improved dynamic programming algorithm is adopted, searching is carried out in a candidate point set formed by buffer insertion candidate points, and a final buffer insertion position and type are determined by taking a signal time sequence and local density cost caused by buffer insertion as optimization targets together. According to the method, the whole-process density perception optimization from global planning to local decision is realized.

The invention relates to a non-intrusive and clustering algorithm-based low-speed electric vehicle load calculation method. The method comprises the following steps of: obtaining a time-sharing power utilization sequence and a time-sharing electricity price sequence of residential users in a statistical region; a load sudden increase point is detected; determining the charging start time and duration of the residential users according to the load sudden increase point; meanwhile, low-speed electric vehicle loads of residential users are distinguished; extracting a low-speed electric vehicle charging load sequence; and finally, according to the low-speed electric vehicle charging load sequence, automatically grouping and determining charging behavior modes of different types of users based on a density clustering algorithm. According to the method, through the design of the power level and the charging duration threshold value, the low-speed electric vehicle load is distinguished from the power consumption fluctuation, the special charging load sequence is extracted, the operability is high, and the influence of other types of power consumption loads can be reduced as much as possible.

The application discloses a kind of sand earthquake liquefaction discrimination method and system based on SSA-CNN-SVM model coupling, and the present application relates to the technical field of seismic safety evaluation, comprising the following steps: obtaining historical sand liquefaction sample data set, including key influence index and liquefaction state label, while collecting geographic location information, extract statistical characteristic parameters to calculate anti-liquefaction intensity index and vibration intensity index, and according to the weighted similarity measurement of geographic characteristic parameter, the sample area is divided into multiple geological regions by condensation hierarchical clustering algorithm;In each region, the Mahalanobis distance between sample regions is calculated, and a secondary clustering is carried out using a density-based clustering algorithm to identify similar liquefaction mechanism sample clusters, and a prediction model coupled with a deep learning model and an optimization algorithm is established for each cluster to assess liquefaction risk, significantly improving the accuracy and robustness of sand earthquake liquefaction discrimination.

Systems and methods are provided for implementing large-scale density-based clustering functionalities. In examples, a system selects, for a dataset (which may be sampled at 100% or less), an upper bound value and a lower bound value of a neighborhood radius parameter of a density-based clustering algorithm. The system identifies, using a modified ternary search algorithm, an optimal neighborhood radius parameter value, based on the upper and lower bound values, outputs the optimal neighborhood radius parameter value and / or a corresponding optimal number of clusters within the dataset. The modified ternary search algorithm leverages the near-unimodality of the neighborhood radius parameter, while selection of the upper bound value leverages a characteristic in which the neighborhood radius parameter value increases as the sampling rate decreases, and selection of the lower bound value uses ternary search that takes the number of clusters as a parameter instead of the neighborhood radius parameter.

The invention discloses a multi-beam sonar point cloud-based underwater target detection method. The method comprises the steps of preprocessing an original multi-beam point cloud; segmenting the ground point cloud by adopting a cloth simulation filtering algorithm to obtain candidate target point clouds; a DBSCAN clustering algorithm based on density is adopted to extract a target point cloud cluster; and performing three-dimensional bounding box fitting on the target point cloud cluster by adopting a directional minimum bounding box method, and outputting a final underwater target three-dimensional positioning result. The method has the remarkable advantages of no need of a prior model, high adaptability, high detection precision, automatic processing flow and the like. Statistical filtering is introduced to realize high-efficiency noise elimination, a distribution simulation algorithm is adopted to accurately separate the ground from a target, a density clustering algorithm is utilized to adaptively extract a target area, and a directional minimum bounding box is combined to realize high-precision target positioning, so that the robustness and the intelligent level of point cloud processing are remarkably improved; a solution which is low in cost, high in performance and high in practicability is provided for the fields of underwater detection, seabed structure recognition and the like.

A method and apparatus for clustering point cloud data of a light detection and ranging (LiDAR) are provided. The method includes obtaining point cloud data of a LiDAR, assigning a grid-based index to each of points included in the point cloud data, and performing density-based clustering on the point cloud data based on the grid-based index.

This invention discloses a method and system for identifying seismic liquefaction of sandy soil based on a coupled SSA-CNN-SVM model. This invention relates to the field of seismic safety assessment technology and includes the following steps: acquiring a historical sandy soil liquefaction sample dataset, including key influencing indicators and liquefaction state labels; simultaneously collecting geographical location information; extracting statistical feature parameters to calculate the liquefaction resistance index and seismic intensity index; performing weighted similarity measurement based on geographical feature parameters; dividing the sample area into multiple geological regions using a hierarchical clustering algorithm; within each region, calculating the Mahalanobis distance between sample areas; performing secondary clustering using a density-based clustering algorithm to identify clusters of samples with similar liquefaction mechanisms; and establishing a prediction model coupled with a deep learning model and optimization algorithm for each cluster to achieve liquefaction hazard assessment, significantly improving the accuracy and robustness of sandy soil seismic liquefaction identification.

The present application belongs to the technical field of industrial energy digital management, and particularly relates to a kind of industrial equipment low carbonization reconstruction efficiency intelligent evaluation method and system, its method includes: collecting the multi-source heterogeneous operation data before and after industrial equipment reconstruction;Dynamic response hysteresis index representing system dynamic instability and hysteresis degree is constructed, and the search radius of density-based clustering algorithm is adaptively adjusted using dynamic response hysteresis index;Carbon sensitive weighted distance cost function is constructed, and the optimal warping path between the measured sequence after reconstruction and the ideal benchmark sequence is found using dynamic time warping algorithm based on cost function;A low-carbon reconstruction efficiency comprehensive scoring model is constructed, and the low-carbon reconstruction efficiency comprehensive score is calculated to evaluate the low-carbon reconstruction efficiency of industrial equipment. The present application can eliminate thermal inertia misjudgment and focus on core emission reduction conditions to achieve scientific and comprehensive performance evaluation.

The invention relates to the technical field of elevator tumble detection systems, and discloses a villa elevator tumble detection system based on millimeter-wave radar, and the system comprises a millimeter-wave radar module which is fixedly installed at the center of the top of an elevator car; the signal processing module is electrically connected to the millimeter wave radar module and is used for processing the original radar data; and the tumble judgment module is electrically connected to the signal processing module. By adopting millimeter-wave radar non-contact detection and combining multi-level threshold judgment logic of'primary triggering-static confirmation-track assistance ', the system can effectively distinguish real falling and similar actions such as normal squatting and stooping, and the problem that a traditional single sensor or threshold method is high in false alarm rate is solved. And a Kalman filtering algorithm and a density-based clustering algorithm are adopted in the signal processing module, so that the precision of target tracking and point cloud segmentation is further ensured, and the analysis accuracy is ensured from a data source.

The invention provides a signaling-based user travel residence point determination method and device, equipment and a medium. The method comprises the following steps: sorting track points in travel track data of a user according to a time sequence to obtain a track point sequence, segmenting the track point sequence according to a preset first time interval to obtain multiple track point subsequences, and clustering the track point subsequences by adopting a density-based clustering algorithm to obtain multiple track point subsequences; and clustering the multiple sections of track point subsequences by adopting a time sliding window method to obtain a clustering set cluster, and finally determining the resident point of the user according to the clustering set cluster. According to the method provided by the invention, the track points moving back and forth in the same path are prevented from being clustered into the resident points, so that the identification states of the track points in the same path are correctly identified, and the identification accuracy of the resident points is improved.

The invention belongs to the technical field of power transmission line three-dimensional vector element extraction, and particularly relates to a laser point cloud-based power transmission line three-dimensional vector element extraction method, which comprises the following steps of: performing height threshold filtering on an original laser point cloud, removing ground points, low vegetation and low-elevation noise points, and retaining related point clouds of an elevated structure; projecting the filtered point cloud to an XY plane, identifying tower point clusters by adopting a density-based clustering method, calculating centroid coordinates of each tower point cluster in the XY plane, and obtaining a point cloud subset only containing a lead after removing the tower point clusters; a minimum spanning tree is constructed by taking a pole tower mass center as a node, tree edges are regarded as transmission line spans, conductor point clouds are classified into the corresponding spans according to the nearest tree edges, and a left side conductor point set and a right side conductor point set are divided based on orthogonal projection in the main direction of the spans.