111 results about "Delaunay triangulation" patented technology

The present invention relates to the technical field of natural ventilation in buildings, and in particular to a computationally assisted decision-making method and system for climate-adaptive building cavity design. The method comprises: using a Delaunay triangulation method to generate an initial mesh, using a Laplace operator-based mesh refinement method to adaptively refine the initial mesh, so as to obtain an adaptive mesh system, wherein the adaptive mesh system is used for dynamically adjusting the mesh density of conditional PINNs; constructing a multi-task learning framework within the conditional PINNs, wherein the multi-task learning framework is used for jointly predicting a plurality of physical field variables within the conditional PINNs; and iteratively training the conditional PINNs, and using a gradient descent algorithm to minimize a loss function until a predetermined number of training iterations or loss convergence is reached, thereby generating a key physical field variable prediction model in building cavity design. The present invention can implement efficient and accurate prediction of key physical quantities in building cavity design, and can be adapted to different building layouts and functional space characteristics.

The invention is suitable for the technical field of road detection, and relates to a semi-rigid base layer loose disease GPR image recognition and three-dimensional reconstruction method, which comprises the following steps: constructing a semi-rigid base layer loose disease random medium model, carrying out forward modeling on a road structure, generating simulation data, and expanding a data set in combination with a data enhancement technology; a DN module fusing a dynamic deformation convolution module and a RepNCSPELAN4 module is designed, meanwhile, a WIoU loss function is introduced, training is conducted through an obtained data set, and identification of the semi-rigid base layer loose disease is achieved; calculating the area of a disease region through pixel and physical size conversion based on an instance segmentation result obtained through identification; a three-dimensional point cloud is generated by adopting inverse distance weighted interpolation, and the disease volume is calculated in combination with a Delaunay triangulation method. According to the method, accurate disease recognition is achieved, meanwhile, two-dimensional recognition is converted into three-dimensional reconstruction, and a quantitative basis is provided for maintenance priority division and repair scheme formulation.

The invention provides an airborne point cloud building contour point extraction method based on a constraint triangulation network and semantic ring analysis, and belongs to the field of airborne laser radar data processing, and the method comprises the steps: obtaining an airborne laser radar point cloud data set of a to-be-processed region; performing classification processing by using a filtering algorithm to obtain a surface point set and a non-surface point set; building candidate seed points are extracted, and building point clouds are extracted; a Delaunay triangulation network is constructed based on the building point clouds, and constraint is carried out based on the distance between the point clouds; constructing a concurrent triangle set, and extracting contour points and contour connecting edges; and constructing a topological graph, performing traversal search on the topological graph, determining the minimum inclusion rectangle of the closed loop, and judging the contour point type corresponding to the closed loop according to the geometrical shape parameter and the spatial distance parameter of the minimum inclusion rectangle. According to the method, the triangulation network is constructed by using the disordered contour points, and the sequence relationship and semantic information between the points are considered, so that accurate extraction of the semantic contour points is realized.

The invention provides a hydropower station reservoir area three-dimensional terrain modeling method and device based on multi-source data fusion, and relates to the technical field of three-dimensional terrain modeling, and the method comprises the steps: obtaining a multi-source data set containing point cloud data, image data, texture data and actual measurement data of a hydropower station reservoir area; preprocessing the multi-source data set to obtain a preprocessed data set; and according to the preprocessed data set, constructing a three-dimensional terrain model by using an improved Delou inner triangulation algorithm, a non-uniform rational B-spline surface reconstruction technology and a Laplacian smoothing operator. The geometric accuracy and the natural form of the three-dimensional terrain model can be remarkably improved.

The invention discloses a photovoltaic power station wind field evaluation method and system, electronic equipment and a storage medium, and relates to the technical field of photovoltaic power station engineering. The method comprises the following steps: carrying out classification processing on original point clouds, generating DEM data by using Delaunay triangulation in combination with a natural neighborhood interpolation method, and carrying out correction through ground control points and least square adjustment; and constructing a three-dimensional terrain model according to the corrected data and image data, determining a computational domain size and a boundary condition, performing terrain optimization, executing LES solution after applying near-surface turbulence inlet disturbance synthesized by PRFG3, and comparing and evaluating the obtained transient wind field data with field actually measured wind speed data. According to the method, accurate capture of complex terrain wind field features and dynamic adaptive correction of the model are realized, so that the evaluation period is shortened, the surveying and mapping cost is reduced, the precision and reliability of wind field evaluation are remarkably improved, and the demand of wind resource evaluation of a large-scale photovoltaic power station is met.

The invention provides a power collection system topology optimization method and device based on a variable weight minimum spanning tree, and relates to the technical field of offshore wind power topology optimization. The method comprises the following steps: grouping a plurality of fans based on the angle difference among the plurality of fans according to the position information of the offshore booster station to obtain a plurality of fan sets; a Delaunay triangulation method is adopted to construct a reference grid connecting line of each fan set; and based on the reference grid connection line of each fan set, generating the optimal topology of the to-be-processed offshore current collection system based on a variable weight minimum spanning tree algorithm. According to the method, the grouping clustering method, the Delaunay triangulation method and the variable weight minimum spanning tree algorithm are combined, and the topology design of the offshore wind plant current collection system is efficiently, economically and rapidly achieved. According to the invention, topological connection and cable specification type selection can be optimized synchronously, and a global optimal solution is sought between the cable path length and the type cost difference caused by the change of the current-carrying capacity, so that the fine control of the overall cost of the current collection system is realized.

The invention, which relates to the technical field of mine geology and urban safety, discloses a refined collaborative prediction method and system for subsidence of a global goaf, comprising the steps of establishing a hierarchical storage system, and correcting a working face calculation boundary; extracting the maximum influence radius of all the working faces and four-dimensional coordinates of the mining boundary of each working face, and calculating to obtain a global prediction range to generate a global surface prediction grid; point set insertion and Delaunay triangulation are adopted to obtain an effective triangular unit, a local influence domain range of the effective triangular unit is calculated, and an influenced sub-grid is screened from the global surface prediction grid; and calculating a predicted point movement deformation value based on a parallel computing architecture, accumulating the predicted point movement deformation value to a global result array through a coordinate index mechanism, synthesizing a deformation component in a specific direction, generating a visual result, and performing statistical analysis. According to the method, the problems of poor adaptability and low efficiency of a traditional scheme are effectively solved, and the prediction precision and efficiency are remarkably improved.

The invention discloses a miniature intelligent terminal intestinal image processing method and device, a storage medium and computer equipment. The method comprises the following steps: preprocessing an original intestinal tract image set acquired by a monocular endoscope, calculating according to the preprocessed intestinal tract image set to obtain a motion calibration scale factor and a texture calibration scale factor, and adaptively fusing to obtain a conversion coefficient from a pixel to an actual size; extracting polyp region key feature points of each image of the preprocessed intestinal image set, constructing a Delaunay triangulation graph according to each polyp region key feature point, and performing non-rigid deformation matching based on a cross-graph convolutional network and an optimal transmission algorithm to obtain a matched feature point set; and sparse 3D reconstruction is carried out according to the matched feature point set to obtain a polyp point cloud, and the polyp size is calculated according to the Euclidean distance between two farthest points in the polyp point cloud and the conversion coefficient. According to the method, the problems of monocular image scale ambiguity and non-rigid deformation can be solved, and the more accurate polyp size can be obtained.

The invention discloses a substation danger area judgment method and system based on electric field gradient perception, and relates to the field of safety monitoring, and the method comprises the steps: arranging an electric field sensor array in a key equipment area of a substation; performing noise reduction processing on the data of each sensor node to obtain an electric field vector set; obtaining the electric field intensity change rate of each monitoring point through a space vector differential algorithm, and generating an electric field gradient tensor; dividing the electric field gradient value, and constructing a dynamic danger level mapping rule; performing spatial interpolation through a Delaunay triangulation algorithm to generate an electric field gradient equipotential surface; constructing an electric field gradient space distribution model, and generating a dynamic danger boundary envelope body in real time; and correcting the boundary of the dangerous area through the pre-trained gradient compensation coefficient matrix and triggering a graded early warning signal. The method has the advantages that the electric field gradient change of the transformer substation is monitored in real time through the electric field sensor array and the wavelet denoising technology, the dangerous area is dynamically evaluated, and the accuracy and real-time performance of dangerous area judgment are effectively improved.

The invention relates to the technical field of three-dimensional modeling, in particular to a terrain modeling method and device based on a domestic three-dimensional graphic platform Railworks, and the method comprises the steps: collecting terrain data along a rail transit engineering project, and generating a structured original data pool; the method comprises the following steps: extracting and processing a discrete three-dimensional point set through a Railworks platform, and generating a triangular mesh by adopting a Gift Wrapping algorithm and a Delaunay triangulation algorithm; reconstructing a three-dimensional terrain surface model, adding an attribute database, and integrating elevation and gradient attributes according to industrial standards; and a downstream bridge and roadbed design module is connected through a data synchronization mechanism, so that design parameters are automatically updated. The device correspondingly realizes functions of each step. According to the method, the problems of high data security risk and low collaboration efficiency caused by dependence on foreign software are solved, an autonomous solution is provided, and the data processing security and the cross-professional design collaboration efficiency are improved.

The invention relates to the field of precision manufacturing of aero-engine blades, and particularly discloses an aero-engine blade grinding and polishing removal amount control method which comprises the steps that an industrial robot carries 3D visual equipment to conduct multi-angle scanning on a blade pasted with mark points, the number of public mark points between adjacent scanning poses is forcibly restrained to be larger than or equal to 3, and the number of the mark points is smaller than or equal to 3; a rigid body transformation matrix is solved based on space coordinate transformation, and high-precision point cloud automatic splicing is achieved; blade point clouds before and after grinding and polishing are reconstructed through Delaunay triangulation with consistent parameters, and a triangular mesh continuous curved surface is generated; calculating the minimum space distance from each vertex of the target curved surface to the source curved surface as a removal amount error epsilon, and positioning an out-of-tolerance region based on epsilon distribution; and generating a compensatory processing track by adopting an equal residual height method. The problem of point cloud registration caused by rare geometric features of the blade is solved, machining decision is driven through curved surface space errors, closed-loop accurate control over the removal amount is achieved, and the consistency of the grinding and polishing quality of the blade is improved.

The invention discloses an industrial complex curved surface point cloud reconstruction method based on adaptive projection, and aims to solve the problem of surface reconstruction of high-noise, non-uniformly-distributed and complex geometric point clouds. The method comprises the following steps of firstly performing down-sampling based on spatial distribution on an original three-dimensional point cloud, and then denoising and homogenizing the point cloud through an adaptive weighted local optimal projection algorithm; fitting a local plane by adopting a local least square method, projecting the point cloud to two dimensions, and then carrying out Delaunay triangularization; mapping the two-dimensional triangulation network topological relation to three dimensions to generate a three-dimensional triangulation network; and finally, abnormal triangles are eliminated through side length constraint, and the grid quality is optimized. According to the method, through whole-process cooperative processing, the point cloud data quality and complex curved surface adaptability are remarkably improved, it is ensured that reconstructed grids are smooth, regular and consistent in precision, and the high-precision requirement of complex curved surface reverse engineering in the fields of aviation, molds and the like can be met.

The invention relates to the technical field of digital twinning and simulation, and discloses a full-stack digital twinning simulation method and system for a power grid dispatching master station, and the method comprises the steps: obtaining a voltage power flow result data set and a temperature field distribution data set, building a unified virtual coordinate system, building an initial two-dimensional thermal distribution diagram according to an initial mapping matrix, and carrying out the simulation of the voltage power flow result data set and the temperature field distribution data set; carrying out Delaunay triangulation on the image to generate a mapping error distribution diagram; constructing a Markov chain state sequence, calculating a state transition probability matrix, and predicting and generating a corrected mapping matrix; the edge node loads the correction matrix to generate a lightweight state vector, the lightweight state vector is uploaded to the cloud through multi-level cache to form a set, and the cloud updates the matrix and issues a new version; according to the method, the voltage-temperature rise mapping relation is captured, predicted and corrected in real time along with working condition drifting, hidden coupling deviation caused by inconsistent data calibers is restrained, and it is guaranteed that the high-precision and high-consistency electrical-thermodynamic combined situation is always maintained in the operation process of full-stack digital twin simulation.

The invention provides a network RTK base station coordinate precision detection method and system, and the method comprises the steps: carrying out the base station networking based on a Delaunay triangulation algorithm, and determining a baseline list; obtaining satellite precision orbit information, selecting a to-be-solved baseline vector from the baseline list, and resolving the to-be-solved baseline vector by adopting a fast static positioning algorithm to obtain a preset high-precision relative coordinate result; the preset high-precision relative coordinate result is converted into coordinate values of all reference stations, the coordinate values of all the reference stations are compared with preset known coordinate values in a system, coordinate difference values of all the reference stations on all the base lines are obtained through calculation, gross error elimination is conducted based on the coordinate difference values, and processed coordinate difference values are obtained; and integrating the processed coordinate difference values of each reference station on the plurality of base lines, performing consistency analysis and statistical judgment, determining the reference station with abnormal coordinates in combination with a preset threshold value, and outputting a diagnosis alarm result. According to the invention, the reliability and precision of the network RTK service are effectively guaranteed.

The invention discloses a bay hydrodynamic prediction method based on a hybrid Fourier neural architecture, and belongs to the technical field of hydrodynamic prediction, the method comprises the steps of data conversion, data preprocessing, topological relation construction, feature extraction and fusion, model construction, training and prediction, and the like, the data format is converted through a modified trim2nc data packet, and the data format of the trim2nc data packet is converted into the data format of the trim2nc data packet; performing preprocessing to obtain static and dynamic feature data sets; constructing a grid topological relation by using a Delaunay triangulation algorithm; space and time features are extracted through a graph neural network and a time convolutional network respectively and fused; frequency domain calculation is carried out in combination with a Fourier neural operator network with the irregular geometric boundary processing capability, physical knowledge is coupled to define a loss function training model, and finally rapid and accurate prediction of the gulf hydrodynamic force is achieved. According to the method, the problems of high calculation cost and low efficiency of a traditional hydrodynamic model are solved, the prediction precision and efficiency are considered, and a reliable basis can be provided for bay water environment management and planning.

The invention discloses a face gear on-machine measurement comprehensive error compensation method based on sequential decoupling. Specifically, a sequential decoupling compensation strategy is provided for solving the problem that the measurement precision is limited due to mutual coupling of a measuring head pre-stroke error and a workpiece installation coordinate system error in the prior art. The method comprises the following steps: firstly, establishing a measuring head pre-stroke error model through standard ball calibration, and performing anisotropic compensation on a tooth surface measuring point by utilizing Delou inner triangulation and a weighted interpolation method; secondly, based on the compensated data, the outer cylindrical surface and the tooth top plane of the face gear are accurately fitted through an iterative optimization algorithm, an accurate measurement coordinate system is established, and the initial phase of the C axis is optimized; and finally, reconstructing an actual tooth surface by adopting a two-stage NURBS curved surface fitting method, and calculating a normal deviation. According to the method, an error propagation chain is effectively cut off through a strict execution sequence, the overall precision and reliability of on-machine measurement are remarkably improved, and a key technical support is provided for face gear closed-loop manufacturing.

The invention discloses a method for converting an elevation system contour line of a stock place to a 85 elevation datum, which comprises the following basic steps of: discretizing the contour line based on an adaptive sampling method of contour line curvature, and extracting node data on the contour line; elevation points are properly encrypted in a place with sparse contour lines by using a method for improving Kriging interpolation by using a gradient boosting tree, and elevation point data obtained by arrangement is converted into 85 elevation; according to a weighted soft constraint Delaunay triangulation algorithm, a triangulated irregular network is constructed by using 85-elevation point data, and contour lines are generated. The method provided by the invention has the advantages of being scientific, reasonable, easy to implement, high in production efficiency, wide in application range and the like, the curvature and characteristics of the contour lines are analyzed one by one by fully utilizing the contour lines of the elevation system of the stock place, and the variation characteristics of the complex space are captured more accurately; and the accuracy of conversion from the elevation line contour line of the stock place to the 85 elevation datum can be further improved.

The invention discloses a high-precision three-dimensional modeling method for a tunnel portal and an adjacent region thereof. The method comprises the steps of data acquisition, preprocessing, multi-source data fusion, grid model construction, optimization and detail addition. A laser radar and photogrammetry equipment are used for carrying out data collection on a tunnel portal and adjacent areas, and point cloud and image data are generated. Then, denoising and filtering are carried out on the collected point cloud data, and distortion correction is carried out on the image data; and fusion of the point cloud data and the image data is realized through an SIFT feature matching algorithm. An initial grid model is constructed by adopting a Delaunay triangulation algorithm, and the surface of the model is optimized through a Laplacian smoothing algorithm, so that sharp flaws are reduced. And finally, mapping texture information of the image data to the optimized grid model, and enhancing details and visual effects of the model. In addition, the method combines finite element analysis to pre-evaluate the structural strength of the model to ensure the reliability and precision of the model.

The invention relates to a multi-beam sounding data quality real-time evaluation method based on a Delaunay triangulation network model, and the method comprises the steps: carrying out the thinning of an original discrete data set, generating a thinned data set, and constructing a Delaunay triangulation network; performing fragmentation processing on the original data set according to the Delaunay triangulation network, and extracting an accurate data range of the fragmented data set; and respectively carrying out gridding processing on each fragmented data set to generate fragmented grids, and carrying out multi-beam sounding data quality real-time evaluation. According to the method, the gridding processing of mass and large-area discrete point data of the multi-beam sounding data can be quickly and accurately realized, the problems of real-time quality evaluation of the mass multi-beam sounding data and the like are solved, and the method can be widely applied to the field of multi-beam sounding data processing.

The application discloses a kind of unmanned vehicle auxiliary MEC's task unloading performance evaluation method, including the following steps: constructing the dynamic collaborative unloading model of unmanned vehicle based on Delaunay triangulation network structure;Through the dynamic collaborative unloading model of unmanned vehicle, establish the collaborative switching mechanism based on Delaunay triangulation network structure;According to the dynamic collaborative unloading model of unmanned vehicle, the successful computation probability of user equipment unloading task is calculated, and the successful edge computing probability of typical user equipment is obtained based on the successful uplink communication probability and successful computation probability under the unmanned vehicle switching.The application solves the technical problem of how to establish the service cooperation node of unmanned vehicle cooperative switching mechanism for real-time switching of unmanned vehicle based on the mobile characteristics of unmanned vehicle, and carries out switching performance evaluation.

Apparatus and methods for detecting non-linear data dependencies in machine learning are provided. The methods may include receiving a data set comprising a plurality of data features, the data set for processing by a machine learning model operating on a machine learning system. The methods may also include plotting the data set in multi-dimensional space and linearizing the plotted, multi-dimensional space by feeding the plotted data to an algorithm programmed to execute the Delaunay triangulation method. The methods may further include receiving from the algorithm the data set linearized in a faceted space and detecting linear correlations in the linearized data set. The methods may additional include identifying a first data feature that has a correlation coefficient with a second data feature that is greater than an accuracy metric and removing the first data feature from the data set to create a reduced data set.

The application provides a multi-modal image cooperative head lesion multi-level automatic positioning method, which fuses rigid initial registration and non-rigid accurate registration algorithm based on thin plate spline algorithm, optimizes the thin plate spline algorithm from the local and global levels through the Delaunay triangulation method, balances the registration accuracy, the calculation efficiency and the robustness by adopting a multi-target constraint processing mechanism, simultaneously replaces the traditional dense scheme by light-weight 7 key position points to reduce interference, realizes accurate positioning by combining head posture geometric analysis and initial alignment, and takes into account the theoretical reliability and engineering practicability, thereby providing strong technical support for multi-modal image navigation; the application effectively improves the accuracy of the lesion position cutting point, reduces the dependence on artificial experience in the clinic, and realizes the positioning function of the high-precision and low-cost image navigation system by MRI image reconstruction and dimension reduction, face key position point selection, head posture preliminary alignment and fine registration between virtual and real images.

The application relates to the technical field of image processing, and discloses a code jetting uniformity detection method and system for a code jetting machine, which comprises the following steps: acquiring a digital image of a code character to be detected; calculating the center coordinates, equivalent diameter and edge definition of ink drops; calculating a global definition reference; constructing a Delaunay triangulation network by using the center coordinates of all the ink drops and generating a dual Voronoi diagram; calculating the area and shape factor of the Voronoi unit corresponding to each ink drop as implicit features; calculating the local unevenness index of each ink drop; calculating the spatial variation coefficient and information entropy according to the local unevenness index of all the ink drops; and obtaining an overall uniformity score representing the entire code character by a preset regression model based on the spatial variation coefficient and information entropy. The explicit features representing the morphology of the ink drops and the implicit features representing the spatial relationship between the ink drops by using the Voronoi diagram are combined, so that comprehensive, objective and accurate detection and evaluation of the uniformity of the code jetting can be achieved.

A method for GNN-based automatic bay zoning includes the following steps: S1, using a modified data packet to convert an output file of the hydrodynamic model into a .nc file that can be read by Python software; S2, reading the converted .nc file and performing data preprocessing; S3, using a tsfresh data packet to extract time series feature values of different features; S4, using a Delaunay triangulation algorithm and a four-way matrix or an eight-way matrix to determine spatial connectivity, and constructing an adjacency matrix; S5, constructing a convolutional GNN to learn bay characteristics and spatial connectivity; S6, using a Louvain algorithm or a Spectral Clustering algorithm to perform unsupervised classification of bay; S7, post-processing broken edges of bay zoning; and S8, outputting zoning results in the form of a .shp file to realize the automatic bay zoning.

The application discloses a kind of based on hybrid fourier neural architecture's gulf hydrodynamic prediction method, belong to hydrodynamic prediction technical field, the method includes data conversion, data preprocessing, topological relationship construction, feature extraction and fusion, model construction and training and prediction etc. steps, by modified trim2nc data packet conversion data format, static and dynamic feature dataset are obtained after preprocessing;Grid topological relationship is constructed using Delaunay triangulation algorithm;Space and time characteristics are extracted and fused respectively by graph neural network, time convolution network;Combining with the ability of Fourier neural operator network with processing irregular geometric boundary is calculated in frequency domain, coupled physical knowledge defines loss function training model, finally realizes gulf hydrodynamic fast and accurate prediction;The application solves the problems of high cost and low efficiency of traditional hydrodynamic model calculation, and considers the prediction accuracy and efficiency, which can provide reliable basis for gulf water environment management and planning.

The application discloses a simplified visual graph modeling and planning method based on Delaunay triangulation, and first acquires polygonal obstacles in an environment, constructs a limited Delaunay triangulation network by using a centroid point set of the polygonal obstacles; obtains neighbor obstacles of each obstacle from the triangulation network to construct neighbor pairs; then constructs a global environment simplified visual graph by using a visibility relationship between adjacent obstacle vertices; maintains a local region where a starting point and an ending point are located into the global visual graph, and plans a global path by using a Dijkstra algorithm with fusion pruning; generates an elliptical window associated with the path, divides boundary points of obstacles in the elliptical window and elliptical boundary points into two types of samples by using the global path, and finally trains a support vector machine based on a radial basis function to obtain a safe and smooth path, so that the path planning method is more advanced and faster.

The invention discloses a three-dimensional protection envelope generation method for an ancient building lightning receiving system, and belongs to the technical field of building lightning protection. According to the method, a finite element discretization thought and a constrained Delaunay triangulation technology are adopted, and a lightning receiving system is decomposed into three types of basic units including a lightning receiving needle, a linear segment lightning receiving belt and an effective lightning receiving surface; a three-dimensional protection range of each unit is accurately calculated, a continuous integral protection range without splicing gaps is synthesized through Boolean union operation, then the integral protection range is compared with a key part of a roof, and when a blind area exists, the layout of the lightning arrester is automatically adjusted, and iterative optimization is carried out. According to the method, high-precision and automatic calculation of the three-dimensional space protection range of the complex lightning receiving system of the ancient building is achieved, adaptability is high, efficiency is high, it can be ensured that no protection blind area exists in the whole roof area, and reliable technical support is provided for the lightning protection optimization design of the ancient building.

The invention provides a method for reversely obtaining a binary image based on a bituminous pavement surface texture three-dimensional model. An image acquisition platform is built according to a close-range photogrammetry technology, an asphalt pavement surface image is obtained, and the influence caused by an asphalt pavement surface shadow is eliminated; according to the method, asphalt pavement texture image 3D point cloud data is generated through Agisoft Metashape software by adopting multi-view image feature matching, the data is imported into Geomagic Design X for post-processing, and then reverse reconstruction of asphalt pavement surface textures is completed based on a Delaunay triangulation method; and reversely obtaining a binary image according to the reconstructed three-dimensional model.

The invention discloses a webpage classification method and system based on a Delaunay triangulation double-graph neural network, and belongs to the technical field of webpage classification. The method comprises the steps that original graphs and node features of webpage classification are acquired; the original graph comprises a plurality of nodes and directed edges among the nodes, the nodes are webpages, the directed edges are hyperlinks among the webpages, and the node features are webpage contents; constructing a shadow graph by adopting a Delaunay triangulation method; respectively carrying out graph convolution operation on the original graph and the shadow graph to obtain local structure information and global relation information of nodes; and the local structure information and the global relation information are fused to generate final node representation, and all nodes are classified by adopting a neural network classifier based on the final node representation. According to the method, the graph expression capability and the information spreading efficiency are effectively enhanced and the webpage classification effect is improved on the premise of not destroying the original graph structure semantics.