63 results about "Division algorithm" patented technology

The invention discloses an adaptive task division and decision joint optimization-based cloud edge computing resource collaborative scheduling method for a complex manufacturing system. The method comprises the following steps of 1, constructing a hierarchical cloud-edge collaborative computing network model; constructing a multi-objective optimization model, and defining an objective function and constraint conditions; 2, dynamically predicting and calculating a resource state through a resource sensing module based on an LSTM neural network, and generating a node resource prediction matrix; 3, dividing a calculation task generated by the manufacturing system into fine-granularity, medium-granularity and coarse-granularity subtask sets by adopting a multi-granularity subtask division algorithm (MSPA), and mapping the subtasks to corresponding calculation nodes; 4, constructing a task unloading decision model based on the D3QN, and dynamically selecting unloading nodes and an execution sequence of the subtasks in combination with a multi-objective optimization reward function; and 5, iteratively optimizing parameters of the D3QN model through a target network updating mechanism and a self-adaptive exploration strategy to realize real-time dynamic adjustment of a task scheduling decision.

The invention provides an electronic data evidence storage verification method and system based on hierarchical hash and an intelligent contract, and relates to the technical field of block chains and data security, and the method comprises the steps: obtaining to-be-stored electronic data; to-be-stored electronic data is divided into data blocks according to a preset rule, a unique identifier and a timestamp are added to each data block, and a standardized data unit is generated; performing hierarchical hash calculation on the standardized data unit, generating a leaf node hash value by each data block, aggregating the leaf node hash values layer by layer according to a Merkle tree structure, and generating root hash; calling a block chain smart contract, submitting the tree structure hash, the timestamp and the public key encrypted signature to a block chain network, after verification of a consensus node, writing the evidence information into the block chain, analyzing a hash path submitted by a user by the smart contract to perform data verification, if the hash path is consistent with the data verification, returning verification success, otherwise, triggering an exception handling contract; through the dynamic partitioning algorithm and the layered hash structure, the calculation overhead of large file processing is reduced.

The invention provides an intelligent business resource scheduling method and system based on big data, and relates to the technical field of data processing, and the method comprises the steps: 1, capturing equipment layer sensor data, business layer order data and logistics layer vehicle track data in real time through distributed data collection nodes, and carrying out the real-time acquisition of the data; a dynamic time window division algorithm is adopted to carry out time sequence alignment processing on the heterogeneous data flow, a ternary state vector matrix with a unified timestamp is generated, the row dimension of the matrix represents a collection time sequence, and the column dimension of the matrix corresponds to equipment layer operation parameters, a service layer task queue state and logistics layer transportation node coordinates respectively. According to the invention, through multi-source data fusion, three-dimensional space analysis and an intelligent algorithm, resource gap accurate identification and cross-level scheduling optimization are realized, and the business processing efficiency and the resource utilization rate are improved.

The invention discloses a model cavity detection method and system. The method comprises the following steps: acquiring target model data to be detected; based on a voxel grid division algorithm, dividing the target model data to obtain a plurality of model parts; based on a seed filling algorithm, determining cavity information corresponding to each model part; and according to the cavity information corresponding to each model part and preset area communication information, identifying isolated cavity data corresponding to the target model data. Therefore, accurate recognition and positioning of the isolated cavity in the complex model can be achieved, the accuracy and efficiency of subsequent geometric repair and structural analysis are improved, printing errors in subsequent 3D printing are reduced, and the printing quality is improved.

The invention provides a CNN-oriented batch matrix multiplication parallel optimization method and system on a SW architecture, and belongs to the technical field of artificial intelligence parallel optimization. Comprising the following steps: respectively converting an input feature map and a convolution kernel in a convolution layer into an input matrix and a weight matrix, and processing the input matrix and the weight matrix into a plurality of groups of independent matrix multiplication tasks in batches; the main core encapsulates a matrix multiplication task into a parameter structure array, the parameter structure array is transmitted to the slave core through single DMA, and the slave core divides rows of an input matrix into row block tasks by adopting a dynamic row block division algorithm according to the total number of threads and the height of the matrix; and executing sub-matrix multiplication calculation on the distributed independent row blocks, asynchronously prefetching matrix sub-blocks by adopting a double-buffer DMA (Direct Memory Access), and executing matrix multiply-accumulate calculation. The parallel processing efficiency of batch matrix multiplication between the master core and the slave core of the SW processor can be improved, and the algorithm performance is optimized.

The invention belongs to the field of micro-architecture design of an integrated circuit processor. The invention provides an H2P branch prediction chip circuit architecture based on BrPerceptron, a variable operation track of a program is constructed based on a context-sensitive variable tracking mechanism, a program execution path topology is constructed based on a basic block division algorithm according to the variable operation track, multi-dimensional correlation detection is performed according to the program execution path topology to judge whether an H2P branch exists or not, and if yes, the H2P branch is predicted. And when the H2P branch instruction is judged to be the H2P branch instruction, enabling the BrPerceptron predictor to output, performing multi-stage feature fusion and outputting a prediction result, and realizing high-precision prediction for the H2P branch instruction.

The invention discloses an automatic segmentation method and system based on a deep foundation pit point cloud model in the technical field of deep foundation pit modeling, and aims to solve the optimization problem of an existing deep foundation pit modeling technology. The method comprises the following steps: carrying out model simplification on an obtained deep foundation pit point cloud model by adopting a dynamic threshold voxel reduction algorithm to obtain a simplified point cloud model; dividing the underground diaphragm wall and the foundation pit ground for the simplified point cloud model by using a geometric constraint random optimization algorithm to obtain an underground diaphragm wall set and a foundation pit ground set; filtering the underground diaphragm wall and the foundation pit ground in the simplified point cloud model according to the underground diaphragm wall set and the foundation pit ground set to obtain a preliminarily filtered point cloud model; performing support layer division on the preliminarily filtered point cloud model by using a geometric constraint random optimization algorithm to obtain a support layer set; and carrying out independently supported point cloud division on the support layer set by adopting a dynamic radius domain division algorithm to obtain an independently supported point cloud set.

The invention discloses an intelligent operation and maintenance management and control method and system based on a digital twinning wind power plant, and relates to the technical field of digital twinning, and the method comprises the steps: obtaining a wind power plant fan time sequence signal, constructing a feature vector, carrying out the Hash mapping, and storing the feature vector into a segmented Hash table; judging an operation mode through a Bloom filter, generating a label triple stream and summarizing abnormal hash keys; counting the Hash value frequency in the time window and marking a high-risk value, and generating a Hash frequency distribution vector; calculating mutual information between nodes, and constructing an adjacent forest through a minimum generation forest; carrying out sampling and low-frequency signal inflection point storage on the sample flow; constructing an exception decision-making tree, and judging exceptional nodes; extracting weighted sub-graphs and dividing the weighted sub-graphs into fault mode clusters; and calculating a fault urgency degree score and generating an operation and maintenance work order. Through a multi-source time sequence signal, Hash mapping, a decision tree and a spectrum division algorithm, wind power plant fault detection, mode recognition and operation and maintenance scheduling are optimized, and the operation efficiency and sustainability are improved.

The invention discloses a dynamic prediction-based three-dimensional space-time modeling method for a pipe network geographic information system, which comprises the following steps of: S1, constructing a multi-modal sensor array, and carrying out multi-modal data acquisition and preprocessing; s2, realizing asynchronous data stream space-time alignment by adopting a dynamic time warping algorithm; s3, constructing a dynamic prediction model based on a space-time causal graph convolutional network; s4, applying an adaptive grid division algorithm driven by reinforcement learning; and S5, a three-dimensional pipe network state matrix with a timestamp is generated, a photon mapping real-time rendering technology is supported, and the model refreshing delay is less than 0.5 second. According to the method, real-time simulation and sub-second three-dimensional rendering of the pipe network state are achieved through dynamic space-time modeling and multi-modal data fusion, accident response delay is shortened to 0.5 second from 30 minutes, the prediction error is reduced by more than 40%, reinforcement learning is adopted to optimize grid division and transfer learning is adopted to adapt to the pipe network material, the maintenance resource waste is reduced by 35%, and the maintenance efficiency is improved. And full-life-cycle dynamic optimization is achieved.

The invention discloses a contour and filling direction adaptive path planning and SLM support-free forming method, which has the following beneficial effects that a low-angle overhanging region is separated based on a region division algorithm, a scanning vector is controlled by a contour on the basis of contour features of complex structure features and an overhanging direction recognition algorithm, and the accuracy of the contour and filling direction adaptive path planning and SLM support-free forming method is improved. The expansion direction of the lower surface area is a filling direction, and adaptively generating a scanning path and the filling direction; different from a conventional path filling strategy, the method can meet the complex structure characteristics of a curved surface, a round hole, rotation, a special-shaped runner, a curved surface and the like, and the scanning line vector and the filling direction of a lower surface area adapt to the structure contour characteristics, so that the influence of the placing position and the scraper direction on the forming thermal strain accumulation and the lower surface forming quality is reduced; according to the method, the high-quality and stable forming requirements of the low-angle feature of 30 degrees or below of existing mainstream metal materials can be met, and the data processing time cost, the powder material consumption and the post-processing time in the whole forming process are effectively reduced.

The invention discloses a hypergraph segmentation method considering logic replication under a multi-FPGA (Field Programmable Gate Array) system, which provides a staged layout strategy, and firstly segments an original circuit and then reduces cross-FPGA communication delay through logic replication on the premise of ensuring that the limitation of programmable resources of each FPGA is not exceeded. The method specifically comprises a pre-layout stage and a post-layout stage: in the pre-layout stage, a multi-stage division algorithm of a multi-resource weighting strategy is used to obtain a preliminary segmentation result, a multi-dimensional weight superposition strategy is applied to solve various resource constraint problems, and possible resource violations are repaired through a resource fine tuning algorithm; and in the post-layout stage, the logic units are selectively copied based on a simulated annealing algorithm, so that the inter-chip communication overhead is further reduced. By means of the method, the high-quality FPGA segmentation scheme copied through the inter-chip logic can be rapidly generated, and the operation frequency and verification efficiency of a multi-FPGA system are effectively improved.

The invention relates to the technical field of big data processing and data integration, in particular to a Kettle-based database dynamic sub-table processing method and system, which comprises the following steps: receiving sub-table configuration parameters through a Kettle graphical interface, the sub-table configuration parameters comprising a fragmentation key, a sub-table algorithm and a sub-table number; table division rule metadata is loaded, a dynamic table division strategy engine is constructed, and the strategy engine supports Hash fragmentation, range fragmentation and time fragmentation algorithms; the construction of the dynamic table division strategy engine comprises the following steps: automatically recommending a table division algorithm based on data sampling analysis; providing a Grouping script interface to support a user-defined fragmentation rule; maintaining table division rule version history, and supporting rule rollback; the method has the beneficial effects that the large table is split into the plurality of sub-tables according to the specified rule, so that the data volume of a single table is remarkably reduced, query and update operations can focus on data in a smaller range, and the query response speed and the update efficiency are greatly improved.

The application provides a high-resolution reconstruction method for sheet microstructure modeling, and relates to the technical field of metal structure reconstruction. The application obtains sheet thickness and average grain size by performing morphology and EBSD characterization on a metal material, compiles a microstructure reconstruction program based on a space segmentation algorithm and a sheet division algorithm, and establishes a microstructure reconstruction model; the three-dimensional reconstruction model closest to the real microstructure is obtained by adjusting the control parameters in the reconstruction program, namely sheet orientation, grain distribution and iteration number; the application can quickly and accurately reconstruct the sheet microstructure, and the generated microstructure model has high compatibility and can be transplanted to various finite element software for simulation analysis.

The application discloses a method for dividing different cleanliness grade areas in a constant-temperature and constant-humidity dust-free workshop in the field of information technology, and comprises the following steps: preliminarily dividing different cleanliness grade areas according to production process requirements and functions of each area, obtaining a region division scheme by using a space division algorithm, including space position coordinates, area and cleanliness grade requirement parameters of each area; setting a buffer zone between different cleanliness grade areas, determining the spatial layout of the buffer zone, including position coordinates, area and shape parameters, by calculating the space position coordinates and area of each area, and determining the cleanliness grade transition requirements of the buffer zone according to the cleanliness grade difference between adjacent areas; analyzing airflow organization and pressure difference control conditions under the optimized region division scheme, obtaining airflow velocity, airflow direction and pressure difference parameters of each area, and obtaining an airflow organization and pressure difference control scheme meeting the cleanliness grade requirements of each area by using a gradient descent iterative optimization algorithm.

The invention discloses a power distribution network multi-point source storage cluster dynamic division and hybrid collaborative voltage treatment method, and belongs to the technical field of power system operation and control, and the method comprises the steps: constructing a cluster dynamic division comprehensive index based on a scale index and a sensitivity consistency index; proposing a cluster dynamic division algorithm based on division indexes, and ensuring that the sensitivity in the same cluster tends to be consistent after division; identifying a key bus according to the bus voltage offset, calculating the required voltage regulation quantity, determining a cluster hybrid collaborative algorithm control target, constructing a priority index, and establishing a reactive power and active power compensation sequence based on the priority index; the voltage violation of the key bus is eliminated with the purpose of minimizing the power compensation requirement, and the adjustment of the large-range bus voltage is realized. According to the method, voltage out-of-limit can be effectively suppressed in the high-photovoltaic-permeability power distribution network, the utilization rate of reactive compensation is remarkably increased, the demand of active compensation is reduced, and the service life of equipment is prolonged while control fairness is considered.

A system is provided robustly protects important information from high-level cyberattacks and physical destruction, including cryptographic analysis using quantum computers and electromagnetic pulse attacks, while enabling restoration without theft by a third party. The system encrypts and partitions file data using predetermined encryption and division algorithms based on a customer specified parameter, allots each file data to multiple sets of distributed file management groups comprising node groups at multiple bases in different regions of the world, distributes and records the file data to be saved in the nodes located at each base that belong to corresponding distributed file management groups, generates and encrypts index information of each distributed and recorded corresponding file data, and records the index information in node groups of a specified base in the consortium chain.

The invention discloses a storage chip aging prediction method and device based on topological sorting and a medium, and relates to the technical field of chip aging tests.The method comprises the steps that an initial static topological graph is constructed based on a predefined internal dependency relationship of a storage chip; collecting performance monitoring data of the storage chip, performing time sequence relevance identification, and generating a dynamic dependency edge; fusing the dynamic dependent edge with the initial static topological graph to generate a mixed topological graph; decomposing the mixed topological graph through a graph community division algorithm to generate a plurality of sub-graph communities; performing topological sorting on each sub-graph community to generate a functional component test sequence of each sub-graph community; and respectively executing an aging test task on each sub-graph community and the corresponding functional component test sequence through a plurality of distributed test nodes to generate a plurality of pieces of performance parameter attenuation data. According to the method, the performance monitoring data and the static topology are fused, the hybrid topological graph of the chip is constructed, and the problem of relevance deficiency caused by data separation is solved.

The application discloses a numerical control machine tool positioning precision compensation method based on a position instruction frequency division algorithm. The method can comprise: establishing a starting origin and a detection end point of numerical control machine tool feed shaft positioning precision detection; measuring the length of the detection stroke and the positioning error of the feed shaft between the starting origin and the detection end point through a laser interferometer; according to the length of the detection stroke and the positioning error, iteratively optimizing a position instruction frequency division numerator and a position instruction frequency division denominator; and according to the iteratively optimized position instruction frequency division numerator and the position instruction frequency division denominator, compensating the positioning precision of the numerical control machine tool. The positioning precision of the numerical control machine tool feed shaft is measured through the laser interferometer, the numerical control machine tool positioning precision compensation method based on the position instruction frequency division algorithm is established, and the positioning precision of the numerical control machine tool can be improved.

The present invention discloses a parallelized finite element mesh stitching method for flow channels of reactor unit assemblies, belonging to the field of finite element mesh technology; the present invention classifies cavities according to their characteristics, and then stitches cavities of different categories using different methods. In addition to the original point stitching algorithm, the algorithm also adaptively applies a variety of algorithms such as particle swarm optimization, shortest path problem, and advancing wavefront method, and achieves high-efficiency stitching of large-scale meshes while ensuring the overall mesh quality through parallel processing of the algorithm and tetrahedron quality detection optimization algorithm. The present invention helps to optimize mesh quality, promote the application scenarios of large-scale parallel mesh division algorithms, fill the gap in three-dimensional mesh stitching technology, and further promote the application and development of mesh technology in the field of numerical calculation.

The invention provides a partial discharge intelligent identification method and system based on collaborative reasoning, and belongs to the field of partial discharge detection.The method comprises the steps that a semantic structure of high-dimensional PD data is explicitly deconstructed through a multi-scale feature information graph and a sparse spectrum division algorithm, and feature sub-channels strongly related to a discharge mechanism are separated; semantic interference in field data is effectively inhibited; furthermore, a fuzzy modeler induced by a channel structure is adopted to independently learn a local rule set, and a dynamic fusion mechanism driven by cross-channel prediction consistency is combined, so that the tolerance to voltage phase loss, background noise disturbance and equipment isomerism is remarkably improved. Compared with a traditional neural fuzzy model, the method has the advantages that non-exclusive fuzzy classification of an unknown discharge mode is realized while interpretability is maintained, the problem of model failure caused by data distribution offset in field deployment is solved, the recognition error rate is obviously reduced, and a high-robustness solution is provided for intelligent diagnosis of power equipment.

The invention provides a self-adaptive gridding FPGA (Field Programmable Gate Array) waveform assembly line dynamic reconstruction method and system, which introduces a self-adaptive gridding algorithm, depends on modeling and safe sequence planning based on DAG (Directed Acyclic Graph), and an assembly line parallel scheduling and integrated time sequence calibration mechanism. The problems that in an existing FPGA waveform dynamic reconstruction technology, the resource fragmentation problem is caused by static partition, the system reliability risk is caused by component dependence management deficiency, and the reconstruction delay is large and the real-time performance is poor due to a serial reconstruction mode are solved. According to the method, the utilization rate of FPGA logic resources is remarkably increased, the safety and controllability of the reconstruction process are achieved, the overall reconstruction time is greatly shortened, and meanwhile the processing precision after reconstruction is guaranteed through self-adaptive time sequence calibration. And an efficient hardware dynamic reconstruction solution is provided for application scenes with extremely high requirements on real-time performance and reliability, such as software defined radio and integrated radio frequency systems.

The application discloses a large-scale multi-mode traffic parallel simulation method and system based on CPU and GPU, relates to the technical field of traffic control, and comprises the following steps: acquiring traffic road networks to be simulated and multi-mode traffic demand data, dividing the traffic road networks to be simulated and the multi-mode traffic demand data based on a preset dynamic spectrum division algorithm to obtain a multi-mode traffic network division result; performing dynamic congestion detection and information feedback processing on the multi-mode traffic network division result based on a preset cross-process congestion feedback algorithm and an MPI communication optimization strategy to obtain congestion state information; inputting the congestion state information into a pre-established microscopic traffic intelligent agent model to perform microscopic behavior parallel calculation, and outputting microscopic behavior data based on the limitation of a regionalized Bucket communication mechanism; and performing simulation processing based on the microscopic behavior data to obtain a final traffic parallel simulation result, so that the simulation precision and efficiency are significantly improved.

The embodiment of the invention discloses a website layout optimization method and device, electronic equipment and a storage medium, and relates to the artificial intelligence field and the financial science and technology field, and the method comprises the steps: obtaining a plurality of to-be-evaluated websites in a preset region, and the subjective appeal of a user for website layout optimization in the preset region; determining a predefined unsupervised melon score algorithm or supervised melon score algorithm as a target melon score algorithm according to the subjective appeal of the user for network layout optimization in the preset area; according to the method, the target melon score algorithm is adopted to predict the final resource melon score rate of each to-be-evaluated website, and website layout optimization is performed in the preset area according to the final resource melon score rate of each to-be-evaluated website, so that the resource melon score rate of each to-be-evaluated website can be accurately determined, and the website layout in the preset area is effectively optimized.

The embodiment of the application provides a grid division method for additive manufacturing slice type multilayer structure, which can solve the technical problem of poor grid division capability in the related art. The method comprises the following steps: dividing a surface grid on a first surface, and converting a second surface into a point cloud result. According to process constraints, matching grid nodes and surface nodes to obtain surface node pairs. In this way, not only the geometric shape is considered, but also the additive manufacturing layering process. According to the surface node pairs, based on a quadrilateral grid, layered hexahedron-based finite element units are divided, that is, a volume grid is generated through the surface grid. While achieving rapid grid division, the stability of various spatial surfaces is ensured, and a grid division algorithm with high automation degree, good stability and good grid quality is realized. The grid division is performed on the additive manufacturing slice type multilayer structure, the material direction and the layering direction are accurately specified, and the efficiency and accuracy of simulation calculation of the additive manufacturing slice type multilayer structure are improved.

A dynamic resource scheduling method for a space-time big data stream processing engine is proposed in the present application. Current stream processing engines usually use clusters or high-performance hardware to improve the efficiency of processing data, only support processing real-time stream data, cannot process unbounded real-time data and bounded storage data at the same time, and the random allocation of processing tasks cannot improve the data processing rate according to the upper demand. The present application faces space-time big data, improves the data division algorithm, divides the stream data into two types of bounded and unbounded data, and designs a dynamic resource scheduling strategy for data processing tasks from the global and local resource range respectively, so that the algorithm can ensure the real-time processing of unbounded tasks while expanding the total amount of processing bounded tasks, thus enabling the stream processing engine to process unbounded and bounded data simultaneously, and realizing efficient allocation of resources in a cluster environment to improve the throughput of the stream processing engine and reduce the processing delay.

The invention discloses a post-grouting monitoring method and system for a cast-in-place column, and the method comprises the steps: carrying out the partitioning of a three-dimensional distribution data set of a grouting material through employing a grid division algorithm, and determining a coverage range, aiming at a business scene problem that the distribution of the grouting material in an underground space is not uniform, and the deformation of a soil body is difficult to control in real time; and soil deformation and stress change are monitored in real time by using a distributed optical fiber sensing technology. And when deformation exceeds a threshold value, constructing a spatial correlation model of the grouting point and soil deformation through a finite element analysis algorithm, calculating a correlation coefficient of offset and deformation, determining a dynamic adjustment direction of grouting parameters, optimizing grouting flow and pressure by adopting a gradient descent algorithm, and updating grouting material distribution data. According to the method, three-dimensional accurate positioning of the grouting material is achieved in the post-grouting process of the cast-in-place column, real-time space correlation is conducted on positioning data and surrounding soil deformation data, and the grouting monitoring effect is improved.

The invention relates to the technical field of coal unloading machine intelligent control, in particular to an intelligent coal unloading machine coal unloading control system and method.An environment fusion sensing unit fuses coal pile point cloud data of a three-dimensional laser scanner and RGB-D camera texture information, a real-time digital twinborn model is constructed through hierarchical semantic segmentation, and a real-time digital twinborn model is constructed; and the autonomous decision planning unit adopts a layered optimization strategy, divides coal unloading grids by a global layer heuristic space division algorithm, plans a grab bucket, generates a spiral progressive cabin cleaning track of a self-adaptive entry point based on the curvature gradient of the coal pile by a local layer, and plans the spiral progressive cabin cleaning track of the self-adaptive entry point based on the curvature gradient of the coal pile. The dynamic posture adjusting unit is used for dynamically correcting the cut-in angle, the excavation depth and the swing amplitude of the grab bucket by collecting excavation resistance moment data and combining with the coal pile thickness to build a coal hardness grading model, and the closed-loop vibration suppression control unit is used for generating a vibration suppression signal by adopting an active damping controller of a Lyapunov stability theory, so that the optimal energy consumption of the grab bucket is realized.

The invention relates to the technical field of building engineering construction, and discloses an engineering deployment method based on a construction stage, and the method comprises the following steps: obtaining basic engineering data and real-time construction dynamic data of a construction project through a multi-source collection device, and generating a standardized construction data set; based on the standardized construction data set, performing construction stage division processing by adopting an intelligent division algorithm, and outputting construction stage division data; a standardized construction data set is formed by integrating multi-source data, construction stage division data is generated based on an intelligent division algorithm, the objectivity and consistency of division of different construction stages are guaranteed, meanwhile, the whole construction process is dynamically divided through the clustering analysis and time sequence pattern recognition technology, the construction progress change is adapted in real time, and the construction efficiency is improved. The accuracy of stage division is ensured, and the rationality and adaptability of project deployment are improved.

The present application belongs to the technical field of coal mine safety, and particularly relates to a dynamic threshold division and multi-factor early warning algorithm model for gas emission of a working face, a dynamic threshold division algorithm model, which comprises an FC-Kmeans model and a GA genetic algorithm model, the FC-Kmeans model comprises two stages: initialization of a clustering center and updating of the clustering center, the GA genetic algorithm model is a GA model based on a fitness function, and is used for clustering quality evaluation of a clustering result output by the FC-Kmeans model; a multi-factor algorithm model, which comprises a Transformer model and a MOA model, the Transformer model is composed of three parts of input, encoder structure and output, and embedding of the MOA model enables information entering the input part of the Transformer model to first enter a multi-head attention layer to extract information, and the multi-head attention layer is composed of multiple self-attention layers. The two models can dynamically divide a threshold range corresponding to a gas early warning level, and comprehensively early warn a real-time state, a fluctuation state and a trend state of gas emission of the working face.

The embodiment of the invention provides an STL (Standard Template Library) grid region division and re-subdivision method and device based on feature preservation. The method comprises the following steps: acquiring an initial grid container; subdividing triangular units in the initial grid container based on a longest edge bisection algorithm; the subdivided triangular units are stored in a subdivision grid container; feature boundary extraction is carried out on triangular units in the subdivided grid container in a dihedral angle threshold value judgment mode; on the basis of the feature edges of the triangular units in the subdivided grid container, performing feature curved surface division on the triangular units in the subdivided grid container through a curved surface division algorithm based on depth-first search to obtain a plurality of feature curved surfaces; and through harmonic mapping and an AFT-Delaunay method, parameterization is carried out on the characteristic curved surface, and mesh generation of the characteristic curved surface is completed. In this way, key geometric features such as sharp edges and ridges of the model can be accurately recognized, the dependence on the initial grid quality is reduced, and the generated grid quality is greatly improved.