9 results about "Multi mesh" patented technology

The invention discloses a ship bollard identification method and system based on adaptive multi-scale multi-grid division, and belongs to the technical field of computer vision and image processing. The system divides an image foreground region and a background region through a visual saliency calculation module, carries out dense sampling in the foreground region and sparse sampling in the background region by adopting a non-uniform grid generation module, screens candidate grids in combination with gradient direction consistency and texture features, fuses candidate frames of different scales through a multi-scale image pyramid, and finally obtains a multi-scale image. And accurate positioning of the bollard is realized through the fine grid accurate positioning module. The method solves the problem that the prior art is insufficient in adaptability to ship size, shooting distance and resolution change, improves the precision and generalization ability of bollard positioning in a complex scene, and is suitable for bollard detection scenes of various ship images.

The invention relates to the technical field of meteorological data assimilation, discloses an assimilation method suitable for a nonlinear problem, and aims to solve the problem that an incremental variation assimilation method has no trust region constraint and may not converge in practical application. A scheme for solving convergence and calculation efficiency of non-linear cloud radiance assimilation in theoretical and practical application is discussed. According to the method, the comparison of the total variation assimilation method and the increment variation assimilation method is realized under the same framework, the comparison of the multi-grid method and the multi-external circulation method is realized under the same framework, the problem that direct comparison is difficult to realize in different platforms due to complex parameters in different methods is solved, and the system operation complexity is greatly simplified through unified model configuration. Experiments show that by adopting the linear search total variation assimilation method under the multi-grid framework, the accuracy of an analysis field is stably improved, and the minimization process of a nonlinear total variation objective function is accelerated.

Embodiments of the present disclosure disclose a geological modeling method based on adaptive geometric multi-grid and related devices. A specific embodiment of the method includes: according to a division mode indicated by a geological modeling parameter, performing grid division on a calculation region indicated by geological source data according to a first resolution, to obtain a first resolution grid set; then, based on first position information of a fault marker in the calculation region and second position information of a control point, dividing a target grid in the first resolution grid set to obtain grids of multiple different resolutions; thereby, the resolution of the grid in the fault region and the region where the control point is located can be ensured to be high, while the resolution of the grid in other regions can be the first resolution, so that dynamic division of multiple levels of grids is achieved, which is helpful for more efficient and accurate modeling.

The application discloses a kind of multiple grid revision analysis method and device with physical constraint, through the original meteorological observation data of the region to be analyzed is converted to coarse grid background field;Select the first observation site on coarse grid, and obtain all second observation sites in its influence radius, based on wind pressure constraint relationship respectively calculate the first potential height observation value between each second observation site and the first observation site, distance and weight coefficient, by fitting distance and weight coefficient, obtain interpolation coefficient, for the interpolation processing of coarse grid background field, obtain the first analysis field of coarse grid, and it is as fine grid background field;By calculating the deviation of meteorological observation data and fine grid background field, the deviation is interpolated to fine grid background field, to revise fine grid background field, and according to the fine grid background field after revision, generate the second analysis field of fine grid, compared with prior art, the technical scheme of the application can improve the accuracy of meteorological data analysis.

This invention discloses a large-scale power grid simulation method, program product, and storage medium based on algebraic multigrid optimization. The method includes: parsing the netlist of a large-scale power grid to establish a power grid topology matrix; extracting global feature vectors based on the netlist and the power grid topology matrix; compressing the power grid topology matrix to obtain a compressed feature matrix; constructing and training a regression model, which is used to predict the solution time based on the compressed feature matrix, global feature vectors, and coarsening parameters; using the minimum output of the regression model as the objective function, and employing simulated annealing to find the optimal coarsening parameters; establishing a power grid simulation system Ax = b; and solving the power grid simulation system using the algebraic multigrid method with the optimal coarsening parameters to obtain the power grid unknown vectors, thus realizing large-scale power grid simulation. This invention improves solution efficiency and reduces memory consumption while ensuring accuracy.

The invention provides a mixed aggregation algebraic multi-grid solving method for a time sequence linear system, and relates to the field of grid data, which comprises the following steps of: aiming at a first linear system of a time sequence, setting a hierarchical boundary parameter, constructing a coarse layer matrix through Galerkin projection, caching related topology and communication information, and constructing a second linear system of the time sequence; solving the first system by combining multiple grids with a Krylov iteration method, and recording a reference convergence state; judging whether the sparse structure of the matrix is consistent with the cache or not, if yes, multiplexing the cache and only updating the numerical value, then judging whether the current convergence state reaches a degradation threshold or not, if not, re-executing the first construction process to update the cache and then solving, and outputting a complete solution sequence after all systems process. According to the hybrid aggregation algebraic multi-grid solving method for the time sequence linear system, the problem that convergence robustness and low construction overhead are difficult to consider when smooth aggregation and non-smooth aggregation are singly adopted is solved, and convergence robustness and low construction overhead can be considered.

The invention provides a parallel computing method, device and equipment, a medium and a program product, relates to the field of data processing, and is used for improving parallel computing efficiency in a multi-grid algorithm. The method comprises the following steps: acquiring a two-dimensional temporary array about each height layer according to an initial three-dimensional array corresponding to a three-dimensional area in a multi-grid space, filtering the plurality of two-dimensional temporary arrays based on processing units corresponding to the plurality of two-dimensional temporary arrays to obtain two-dimensional arrays corresponding to the plurality of two-dimensional temporary arrays, according to the method, the initial three-dimensional array is updated on the basis of the two-dimensional arrays corresponding to the multiple two-dimensional temporary arrays, the updated initial three-dimensional array is obtained, parallel accumulation is conducted on the updated initial three-dimensional array and the preset three-dimensional array, the target three-dimensional array indicating the three-dimensional area scanning result is obtained, and therefore parallel calculation efficiency is improved.

The invention discloses an efficient parallel method and system based on a space-time four-dimensional multi-grid method, and a medium. The method comprises the following steps: carrying out sea area partitioning according to terrain distribution and mode resolution of a sea area to be researched, and dividing the sea area into N partitioned blocks; performing local encryption on the N partition blocks to generate N groups of locally encrypted analysis grids; respectively carrying out data assimilation in parallel based on the N groups of locally encrypted analysis grids to form N groups of global analysis results; and from the N groups of global analysis results, respectively extracting an analysis field of each group of local encryption areas, and reconstructing the analysis fields into a complete analysis field. According to the method, the calculation speed and the calculation efficiency of the space-time four-dimensional multi-grid variational data assimilation method can be improved, the problem that the memory usage amount is too large in a high-resolution mode application scene is solved, and the business application capability is improved.

A provincial domain power grid ultra-short-term load prediction method based on multi-grid modeling relates to the provincial domain power grid ultra-short-term load prediction field, and comprises the following steps: obtaining distributed photovoltaic load data, gridding numerical weather forecast data and provincial domain power grid topological structure data of each prefecture-level city of a target provincial domain; constructing a load grid chart relationship; constructing a meteorological grid chart relationship; constructing a power grid topology grid chart relationship; constructing a directed graph from the load grid chart to the meteorological grid chart, from the load grid chart to the power grid topology grid chart, from the meteorological grid chart to the load grid chart and from the power grid topology grid chart to the load grid chart; and constructing a graph neural network prediction model to obtain an ultra-short-term power prediction result of each prefecture-level city and an overall load prediction result of the target provincial domain. A graph structure modeling method is adopted, effective utilization of multi-scale spatio-temporal correlation information of multi-source heterogeneous data is achieved, prediction results of all prefecture-level cities in a region can be provided, and prediction precision can be effectively improved.