38 results about "Sparse grid" patented technology

In various examples, systems and methods are disclosed relating to generating each initial feature map of a plurality of initial feature maps based on a respective input image of an input dataset, each initial feature map, incorporating depth data of the respective input image, corresponds to a plurality of pixels of the respective input image, generating a sparse feature point cloud including a plurality of features determined using the plurality of initial feature maps, transforming the sparse feature point cloud into multi-resolution sparse grids, each of the multi-resolution sparse grids comprising a plurality of voxels, modeling, using a plurality of neural networks according to a hierarchal architecture, the multi-resolution sparse grids to construct a hierarchical volume representation, and providing constructed content based on the hierarchical volume representation.

The invention discloses a pressure vessel lower chamber model order reduction method combining sparse grids and POD, and the method comprises the steps: adding one to the grade of the sparse grids after iteration initialization, and obtaining test grid points according to important grid points; executing the lower chamber full-order computational fluid mechanics model corresponding to the test grid points to obtain a first snapshot; reconstructing a second snapshot by using the mapping relation and the POD modal matrix; calculating a root-mean-square error of the two snapshots, and if the root-mean-square error is smaller than a threshold value, completing construction of a reduced-order model Otherwise, the grid points with the errors meeting the requirements are stored in a non-important grid point set, the grid points which do not meet the requirements are marked as important grid points, and the snapshot matrix is updated. Performing singular value decomposition on the snapshot matrix, updating the mapping relation, judging whether the maximum sparse grid grade is reached or not, and continuing iteration if the maximum sparse grid grade is not reached. According to the method, the field distribution of the thermal hydraulic parameters in the lower chamber can be quickly obtained through matrix algebraic operation, and the method can be used for application scenes such as parameterization analysis, design optimization and digital twinning.

The invention discloses a high-precision adaptive numerical simulation method and system for solving a nonlinear integral-differential equation, and relates to the technical field of computer simulation and numerical calculation. Comprising the steps of setting simulation conditions; constructing a grid space based on the computational domain; constructing a global cubic spline interpolation function based on the solutions on the grid points; if the updated grid point location exists, calculating a right-end integral item by using an adaptive numerical simulation integral algorithm; calculating a heuristic solution and an error estimation value; dynamically adjusting the tentative step length and updating the time according to the estimated value and the ODE tolerance; checking gradient and curvature distribution of solutions, encrypting grids for areas with violent solution changes, and sparse grids for areas with smooth solution changes; using a global cubic spline interpolation function to demap the old grid space to the adjusted new grid space; and iteratively executing the steps, and outputting an integral solution in the computational domain. And finally, collaborative improvement of integral simulation calculation in precision, efficiency and stability is realized.

The application discloses a network-constructing type energy storage multi-scenario operation boundary quantification method and system considering coupling constraints, comprising: acquiring multi-source parameters and constructing a characteristic vector; the characteristic vector comprises a first characteristic vector: a network-constructing working condition vector and a second characteristic vector: a battery state response vector; based on the failure mechanism of each dimension of electricity, heat and life, a multi-dimensional dynamic safety threshold vector is determined; and a joint safety margin function is constructed according to the multi-dimensional dynamic safety threshold vector, the first characteristic vector and the second characteristic vector; combined with the joint safety margin function, a boundary quantification algorithm based on adaptive sparse grid and GPR dynamic reconstruction is used to solve a failure boundary point set; a failure boundary is constructed according to the failure boundary point set; and according to the failure boundary, a double-track quantification evaluation and a state label output mechanism are used for boundary triggering output or safe operation output. The application significantly improves the real-time performance, fineness and engineering landing ability of the safety operation boundary quantification evaluation under the network-constructing energy storage full working condition atlas.

The present invention discloses a high-strength and tough aviation thin plate imitating an armadillo with a shell and a preparation method thereof, relating to the fields of aviation materials and laser additive manufacturing technology. The high-strength and tough aviation thin plate of the present invention includes a plurality of connection layers arranged in parallel, and a sparse grid layer and a dense grid layer located between adjacent connection layers. The sparse grid layer and the dense grid layer include closely arranged hollow block units, and the connection layer, the sparse grid layer and the dense grid layer form a closed cavity. The preparation method of the present invention uses laser melting deposition additive manufacturing technology to form the connection layer, the sparse grid layer and the dense grid layer, and selective laser melting additive manufacturing technology to form the sparse hollow blocks and the dense hollow blocks. The high-strength and tough thin plate of the present invention can be used for load-bearing structural parts and heat dissipation structural parts in fields such as aerospace.

The invention discloses a power system risk assessment device and method based on extended dimension adaptive sparse grid interpolation and a medium, and relates to the field of power system risk assessment, the device comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, a processor executes a computer program to realize a power system risk assessment method based on extended dimension adaptive sparse grid interpolation, and the method comprises the following steps: firstly, extending DASGI by using a relaxation technology, so that the DASGI can effectively approach a complex risk assessment function which takes a load shedding amount as an output, has strict non-negativity, and gathers a large number of results near a zero value; based on the constructed EDASGI proxy model, an analytic relationship between the unit commitment and the system risk is established; the iterative recursion solving process of the RSUC is converted into a mixed integer quadratic programming problem, efficient and unified solving is achieved, and the efficiency and precision of risk decision making in a high-dimensional uncertainty scene are remarkably improved.

This paper presents a sparse grid polynomial-based orbital game situation analysis method, belonging to the field of aerospace orbital adversarial and intelligent decision analysis technology. Addressing the problems of lack of systematic representation of multi-source uncertainties, low computational efficiency of traditional quantification methods, and difficulty in achieving multi-objective equilibrium optimization in existing high-orbit multi-to-multi dynamic adversarial scenarios, this paper constructs a POMDP multi-to-multi dynamic game model incorporating uncertainties in high-orbit dynamics and control execution, strategy behavior, and communication delay. A surrogate model is established using sparse grid sampling and polynomial chaotic expansion, and global sensitivity analysis is performed. Based on this, a multi-objective optimization model is constructed, and the Pareto optimal policy set is solved. This achieves game strategy generation that balances performance, robustness, and economy, and is applicable to spacecraft strategy optimization and space situation analysis in high-orbit multi-to-multi dynamic adversarial missions.

In various examples, systems and methods are disclosed relating to generating each initial feature map from a plurality of feature maps based on a respective input image of an input data set, wherein each initial feature map incorporating depth data of the respective input image corresponds to a plurality of pixels of the respective input image, generating a sparse feature point cloud containing a plurality of features determined using the plurality of initial feature maps, transforming the sparse feature point cloud into multi-resolution sparse grids, wherein each of the multi-resolution sparse grids comprises a plurality of voxels, modeling the multi-resolution sparse grids using a plurality of neural networks according to a hierarchical architecture to construct a hierarchical volume representation,and providing constructed content based on the hierarchical volume representation.,

The present invention discloses an adaptive subdivision grid topology optimization method and system based on a balanced quadtree. The steps of the method include establishing a sparse grid model and defining loads and boundaries; establishing a data structure for storing unit and node information; initializing design variables, calculating the objective function and constraint function, and the sensitivity of the function to the design variables; calculating the subdivision factor of each unit and determining whether it is greater than a threshold; for grids greater than the subdivision factor, using a recursive method to decompose the grid units and update the unit and node data structures; establishing a stiffness matrix, a constraint matrix and a calculation matrix for the updated model, and updating the design variables; repeating the above steps until convergence. Based on the above method, the corresponding modules are constructed and a system is formed. The method of the present invention ensures high computational efficiency while achieving extremely high boundary accuracy, greatly broadening the scope of application of the adaptive topology optimization method.

The invention relates to the technical field of three-dimensional scene reconstruction, in particular to a color fusion processing method and device applied to a three-dimensional space point cloud. The method comprises the steps of obtaining a target space coordinate of a target space point; the target space point comprises at least one space point to be subjected to color fusion in the three-dimensional space point cloud; according to the target space coordinates, analyzing each target space point and a grid identifier corresponding to a preset index grid space to obtain a target index identifier corresponding to each target space point; the index grid space is a grid space of a sparse grid containing a three-dimensional space point cloud; performing color fusion operation on the target space points according to the target space coordinates, the target index identifier corresponding to each target space point and a preset fusion grid space; the fusion grid space is a grid space of dense sub-grids of the index grid space. Fast access is achieved through dense grids, redundant occupation is reduced through sparse grids, and therefore the point cloud color fusion efficiency is improved.

The invention relates to the technical field of radiation field calculation, and relates to an improved non-uniform radiation field grid division method and system. Comprising the following steps: carrying out initial uniform grid setting on a radiation field calculation area; performing N-time sparse processing on the initially set uniform grid; calculating the distance from the center point of each sparse grid to the ellipsoid surrounded by the source item, and determining the weight multiple of the grid according to the distance; traversing all source items and grids in sequence; carrying out refining processing on all the grids with the weight multiples greater than 1 according to the corresponding multiples; and summarizing and outputting all grid intersection points, namely a final radiation field calculation dot matrix. According to the radiation field change rule of the area near the source item, the judgment basis of whether the grid needs to be refined is improved, and whether the ratio of the distance from the grid center point to the ellipsoid surrounded by the source item to the source linearity is smaller than 1.5 is modified from whether the grid is an extension grid, so that grid division near the source item is more reasonable, and the accuracy of grid division is improved. And the calculation efficiency and the calculation precision of the radiation field are better considered.

This invention discloses a structural topology optimization design method considering the uncertainty of multi-peak loads, belonging to the field of uncertain structural optimization design. It mainly includes three parts: establishing a multi-peak load uncertainty model, solving for the random response, and solving for the optimization formula. This invention describes load uncertainty using a Gaussian mixture model, solves for the Gaussian mixture model coefficients based on the EM algorithm, and establishes a multi-peak load distribution probability model. By decorrelating random variables, sparse grid technology is used to solve for the mean, standard deviation, and sensitivity of the response, thereby solving for the topology optimization model considering the multi-peak load uncertainty. This method addresses the problem of low structural product reliability that may occur in structural designs considering multi-peak load uncertainty by establishing an accurate probabilistic model of multi-peak load uncertainty. This method is simple and easy to implement, readily applicable in engineering, and can improve the design efficiency of structural designers considering complex load conditions.

The application discloses a kind of calculation methods for simulating toxic and harmful gas diffusion concentration in three-dimensional virtual training environment, S1, determine toxic gas diffusion effective influencing factor;S2, the physical model corresponding to training environment is constructed, and the effective influencing factor of gas diffusion is derived;S3, according to gas toxic dose and shelter size, sparse grid is divided, and grid density is distinguished;S4, CFD model is constructed, and the concentration of simulated harmful gas diffusion is calculated;S5, concentration value time sequence situation result offline database of a period of time is established, and gas concentration situation result is stored;S6, when training, in virtual environment, quickly display.The method of the present application can consider terrain, building, meteorological conditions and other factors affecting gas concentration diffusion, can measure the gas diffusion concentration value of any point in space in real time, and its diffusion concentration value is realistic and practical, so that the simulation effect of harmful gas diffusion concentration value in virtual reality chemical hazard simulation training environment is more realistic and credible, and the training effect is improved.

The application discloses a water area adaptive underwater image enhancement method, which comprises the following steps: 1) based on a feature extraction method of multiple groups of convolution and sparse grid attention, an input underwater image is encoded to obtain encoded features of local details and global color contour information; 2) the domain difference of the underwater environment is constructed as a water area classification problem, and the difference information between different water areas is used to generate water area sensitive features to realize image enhancement of different water areas; 3) based on the enhanced image generated by the water area sensitive features, an aggregation loss and a multi-stage training strategy are used to constrain the enhanced result to realize high-quality underwater image enhancement. The water area adaptive method disclosed by the application fully utilizes the domain difference between different water areas and the difference of local and global information, and realizes adaptive and high-quality underwater image enhancement.

A kind of underwater vehicle noise prediction correction method based on bayesian inference, calculation model (1) includes Corcos autospectrum and Corcos normalized cross power spectral density function calculation model;According to calculation model (1), construct the schematic diagram (2) of bayesian network, the key factor identification and state analysis of the key factor of the modeling of the influencing factor of turbulent fluctuation pressure in bayesian network in schematic diagram (2) of bayesian network are built-in bayesian inference;According to the schematic diagram (2) of bayesian network, model in NETICA software, construct directed acyclic bayesian network (3);Reference calculation model (1) and directed acyclic bayesian network (3) are simulated setting (4) and noise signal comparison (5).The present application has the advantages of ensuring the precision of low-frequency noise prediction in sparse grid large size, improves the problem that the grid scale of finite element model and the space related scale of turbulent boundary layer pressure do not match and the calculation amount is too large when fine grid is used.

The invention discloses a method, a device, equipment and a medium for determining sparse parameters of meteorological observation, relates to the technical field of meteorological numerical forecasting, and aims to solve the problem that an atmospheric state displayed by high-resolution observation data cannot be accurately described by a mode due to the fact that a set sparse parameter value is selected for sparse processing in the conventional meteorological numerical forecasting. The method comprises the steps of determining an upper limit value of a sparsification parameter based on a background field water vapor correlation scale; performing horizontal grid division on the satellite observation data to obtain a plurality of sparse grid units; calculating a local standard deviation corresponding to each sparse grid unit, and determining local standard deviation probability density distribution corresponding to each sparse parameter value; a first target sparsification parameter value is determined based on the local standard deviation probability density distribution. The meteorological observation sparsification parameter determination method provided by the invention is used for determining the sparsification parameter value which enables the regional numerical forecasting assimilation mode to be matched with the resolution of the sparsified observation data, and the meteorological forecasting precision is improved.

The invention discloses a sparse grid-based lower chamber flow field POD coefficient adaptive fitting method. The method comprises the steps of obtaining to-be-fitted POD coefficient discrete data; parameters and sets are initialized, and the initialization parameters and sets comprise one-dimensional sparse grid node positions, all power primary functions, sampling dimensions d corresponding to POD coefficients, d-dimensional sparse grid point positions, all power primary functions, active point sets, good point sets, surplus sets corresponding to the active point sets and primary function sets; sampling and set updating are carried out on the grid points of the preset level, test grid points are obtained, the test process is executed, and then a POD coefficient fitting formula is determined. According to the method, multiple power basis function adaptive selection strategies are provided, the optimal basis function form can be selected for different parameter intervals, and the fitting precision of the POD coefficient is remarkably improved.

This invention discloses an exploration grid optimization method and system. The method includes: reading N original boundary sampling coordinates of a target mining area, constructing a mining area constraint boundary of the target mining area, obtaining the current grid spacing pairs of the mining area constraint boundary, dividing the mining area constraint boundary into grids with the current grid spacing pairs as the step size to obtain J candidate exploration grids, extracting the grid coordinates of the J candidate exploration grids, calculating the parameter variation intensity of the current grid spacing pairs based on the grid coordinates of the J candidate exploration grids, iteratively updating the current grid spacing pairs with a preset step size, and simultaneously calculating the updated parameter variation intensity. If the updated parameter variation intensity is lower than a set threshold, the updated current grid spacing pairs are determined as the optimal grid spacing pairs. This invention can truly reflect the degree of control of the current grid density on the ore body variation, avoiding over-densification or over-sparse grids caused by a lack of quantitative standards.

The invention provides a laser radar point cloud rapid robust ground segmentation method based on an adaptive grid, and the method comprises the steps: dividing an adaptive polar coordinate network for an input point cloud, and obtaining each sector which comprises a basic ring and an adaptive ring; for each sector, performing forward robust first-order filtering on the adaptive ring along the radial direction to obtain the ground height and the radial slope after filtering; according to the ground height and the radial slope obtained after filtering, the filtering estimation height of each point and the deviation between the height of the point and the filtering estimation height are calculated, the deviation is compared with a self-adaptive threshold value for a high-confidence grid unit and a sparse grid unit, and ground point classification is completed; and carrying out ground point classification on the low-confidence grid units by adopting a comprehensive scoring mode again.

The present invention belongs to the field of computer graphics and discloses an efficient error-controllable mesh simplification method based on improved envelope testing. The method is as follows: for the mesh of the original model, first generate error envelope polyhedra for all patches to form a set B, and construct a sparse grid acceleration structure for the set B. Then, use Qslim to select an edge e to be folded, set the set of patches adjacent to e as D, quickly find the envelope bodies in the set B of envelope bodies that intersect the patches in the set D based on the sparse grid, and define their set as B'. Finally, calculate the intersection lines between the cutting planes in B' and the triangular patches in the set D, thereby realizing the reduced-dimensional coverage test. If the envelope bodies in B' completely contain the patches D, fold the edge e, and thus complete one folding. Repeat until no edges can be folded. The present invention effectively controls the error of the simplified model by using error envelope bodies, and realizes higher simplification efficiency based on sparse grid search and reduced-dimensional coverage test.

The invention discloses a GNN-guided hierarchical H-tree low-skew clock tree synthesis method, which comprises the following steps of: firstly, executing fan-out-radius-capacitance three-constraint clustering on a chip register to generate an optimized buffer leaf cluster; spatial distribution and electrical characteristics are extracted, and an optimal sub-tree division scheme is predicted by using a graph neural network GNN; a K-adaptive non-uniform branch H tree is constructed based on the prediction result, a binary chain structure is adopted for a light load area, and a quadtree structure is adopted for a heavy load area; and finally, integrating the global clock network through the sparse grid. According to the method, the capacitance balance constraint and the graph neural network technology are creatively fused, and compared with a traditional H-tree method, the clock skew is reduced by 11.5%, the clock delay is reduced by 13.9%, the number of buffers is reduced by 31.0%, and the comprehensive operation time is remarkably shortened.

The application provides a curved surface image correction method and device based on multi-modal data fusion, comprising the following steps: S1, projecting a sparse grid laser dot array to a to-be-corrected object, collecting an image of the to-be-corrected object with laser marks, and determining whether it is a high-curvature region according to the interval change of adjacent laser dots; S2, projecting a high-density grid laser dot array to the high-curvature region, collecting the image, obtaining fine depth information of each laser dot to form laser point cloud data; S3, constructing and solving an energy minimization model to generate an inverse distortion mapping field of the corresponding curved surface to a two-dimensional flat plane; and S4, inputting the inverse distortion mapping field to a correction model and outputting a corrected image according to the correction model. The application has the beneficial effects that the laser projection is not affected by the surface texture of the to-be-corrected object and environmental light, the precision and strong robustness are improved, the high-curvature region is pre-screened by using the sparse grid laser, and the working efficiency is improved.

The invention discloses a PINN-based gear point cloud statics stress simulation method and system, and relates to the technical field of solid mechanics simulation analysis. The method comprises the following steps: acquiring and importing a target gear geometric model, performing sparse mesh fast finite element analysis on the target gear geometric model, and exporting global mesh and boundary information; establishing a point cloud model based on the global grid and the boundary information, and determining a boundary condition; and constructing a PINN physical information neural network based on the point cloud model and the boundary condition, constructing a loss function to carry out multiple times of staged network training so as to obtain a prediction result of the PINN, and outputting the prediction result as a simulation result. According to the method, the point cloud data can be directly utilized, accurate prediction of the physical field is realized through a staged training strategy, and the convenience, accuracy and engineering practical value of complex structure analysis are remarkably improved.

The invention provides a method for setting an external dynamic noise source positioning scanning switch and a storage medium, and belongs to the technical field of seismic exploration methods. The method comprises the following steps: (1) acquiring a maximum range area of an exploratory area; (2) acquiring a sparse grid energy spectrum and a sparse energy spectrum grid switch array; (3) randomly extracting a part of shots from the seismic data to serve as sparse shots for energy scanning, and setting a grid scanning switch; according to the method, the positioning calculation efficiency is improved, and the positioning effect is ensured, so that the mass data processing efficiency is improved, the calculation resources are saved, and the purposes of reducing the cost and improving the efficiency are achieved.

The invention specifically relates to a hydrogen-doped natural gas pipeline circumferential weld residual life prediction method, and relates to the technical field of oil and gas pipeline safety engineering, and the method comprises the steps: interpolating a physical quantity of a parent unit to a subunit through a shape function, and completing the data transmission of a state variable after conservative correction verification. According to the method, a self-adaptive adjustment mechanism based on a hydrogen concentration gradient module value is constructed, and collaborative optimization of prediction precision and calculation efficiency is realized; aiming at a core area with violent hydrogen concentration change at the circumferential weld crack tip of the hydrogen-doped natural gas pipeline, grids are encrypted through recursive splitting and a 2: 1 balance rule, the coupling effect of stress concentration and hydrogen atom aggregation is accurately captured, and the problem of key physical field distortion caused by a traditional fixed coarse grid is avoided; for a far-field hydrogen concentration uniform distribution region, sparse grids are coarsened through pruning, and the number of invalid calculation units is reduced.

The invention relates to the technical field of path planning, in particular to a robot path planning method based on topology awareness, which comprises the following steps of: constructing grids and counting four-way shielding frequency, marking sparse transition density levels according to frequency threshold values, identifying sparse grid clusters and calculating boundary ratios, and marking passage priority or constraint segments. Analyzing path section direction stability, screening continuous priority edges, calculating edge connection fluctuation identification structure sensitive sections, and counting execution failure proportions to mark potential unaccessible path sections. According to the method, the grid shielding frequency in the obstacle distribution diagram is quantitatively calculated, the consistency of the planned path in the global navigation direction is improved, the critical structure area is accurately recognized in combination with path section boundary topological connection density analysis, the adaptive capacity of the path in the structure change area is effectively improved, and the navigation accuracy is improved. The potential inaccessible state of the path segment is dynamically labeled, and the response efficiency and execution stability of path planning to complex environment changes are improved.

This application discloses a two-dimensional DC electric field forward modeling method, apparatus, electronic device, dielectric, and product. The method includes: taking the variational problem of a two-dimensional wavenumber domain point source DC electric field as the research object; dividing the model region of the dielectric model into several triangular elements and expanding the model region with a sparse mesh; reading the conductivity and potential of the principal nodes in each triangular element, and representing the conductivity and potential with quadratic interpolation shape functions to obtain a discrete model; constructing a system of linear equations based on the discrete model and the coefficient matrix of each triangular element; solving the system of linear equations to obtain the wavenumber domain forward modeling result; and performing an inverse Fourier transform on the forward modeling result to obtain the spatial domain electric field potential. The method increases the number of nodes in the mesh during model region meshing, reduces computation time by expanding the model, and utilizes more node property information through biquadratic interpolation of conductivity and potential within the mesh, thereby improving the accuracy of the two-dimensional DC electric field forward modeling fitting.

This invention discloses a sparse mesh radiation field representation method based on point cloud initialization and depth supervision. This radiation field can be used for rendering from new perspectives. The method includes obtaining at least two camera color images, obtaining two camera depth images, generating a 3D point cloud from the depth images, and generating rays from each pixel using camera parameters (including intrinsic and extrinsic parameters) to obtain a set of rays for all pixels. Since real-world scenes are often largely blank, this invention uses a sparse mesh to represent the scene, reducing memory costs. It utilizes the completed 3D reconstruction as a geometric constraint to enhance geometric learning capabilities and uses the completed 3D reconstruction as a guide for initialization, reducing the number of parameters during the mesh generation stage, improving optimization efficiency, and reducing storage costs.

The invention provides a complex structure rock acoustic emission source positioning method and device based on a sparse grid path tracking algorithm, equipment and a medium, and relates to the technical field of rock mechanics. The method comprises the following steps: carrying out three-dimensional space division on a rock with a complex structure by utilizing a self-adaptive octree structure, and dividing the rock into non-cubic sparse grids; based on an arbitrary angle path tracking algorithm and sight visibility detection, calculating a shortest P-wave propagation path between a sensor and a grid node, and calculating minimum propagation time; determining the arrival time difference between the theoretical arrival time and the actual arrival time of the potential acoustic emission source signal to any specific position in the grid through a trilinear interpolation algorithm; and constructing a target function based on the time difference of arrival, and determining the acoustic emission source position of the complex structure rock according to the minimum value of the target function. According to the embodiment of the invention, the calculation cost in the path tracking process is obviously reduced, and the determined position of the acoustic emission source of the rock with the complex structure is more accurate.

The invention discloses a hierarchical sparse voxel representation for generating a synthetic scene. In various examples, systems and methods are disclosed that involve generating each of a plurality of initial feature maps based on a respective input image of an input data set, each initial feature map including depth data for the respective input image, a plurality of pixels corresponding to the respective input image, and a plurality of feature maps corresponding to the plurality of pixels. The method includes generating a sparse feature point cloud including a plurality of features determined using a plurality of initial feature maps, transforming the sparse feature point cloud into multi-resolution sparse grids, each multi-resolution sparse grid including a plurality of voxels, modeling the multi-resolution sparse grids using a plurality of neural networks according to a hierarchical architecture to construct a hierarchical volumetric representation, and providing the constructed content based on the hierarchical volumetric representation.