634 results about "Mesh node" patented technology

The invention relates to a transformer multi-physics field coupling numerical simulation method and device and a storage medium. The method comprises the following steps: performing coupling calculation on a target working condition by using an electromagnetic, heat, flow and force coupling calculation model of a transformer to obtain transformer loss data and temperature rise data; mapping the temperature rise data to a transformer electromagnetic and force coupling calculation model by using a grid node data mapping method to serve as an initial value, and performing coupling calculation on a target working condition to obtain dynamic data of an internal structural member of the transformer; performing fast Fourier transform on the dynamic data in a preset frequency range to obtain dynamic data under each frequency; and mapping the temperature rise data and the dynamic data under each frequency to a transformer sound and force coupling calculation model as initial values, and carrying out coupling calculation on a target working condition to obtain transformer oil tank vibration and external noise data. Compared with the prior art, the method has the advantages that the calculation amount is effectively reduced, and the accuracy of predicting the internal state of the transformer is improved.

The invention discloses an algae biomass prediction method based on a bidirectional coupling hydrodynamic model and a physical information neural network, and relates to the field of algae biomass prediction.The algae biomass prediction method comprises the steps that firstly, water regimen data, water quality data and meteorological data are obtained and preprocessed, and then a hydrodynamic model based on a shallow water equation set is constructed; the model is coupled with a convection-diffusion equation and a wind stress-turbulence model so as to simulate spatial distribution of nutritive salt concentration, water temperature distribution and a vertical and horizontal mixing process driven by wind power; operating the model based on a time sequence of meteorological and water regimen data, and outputting multi-dimensional physical field data; thirdly, performing spatial interpolation on the water quality data to enable the water quality data to be in space-time alignment with grid nodes of the hydrodynamic model; a target loss function is constructed by taking a residual error in the shallow water equation set as a regular term, and a physical information neural network is trained to finally obtain a target prediction model; and the accuracy of algae biomass prediction and the robustness of the target prediction model are obviously improved.

The invention provides a tunnel deformation parameter inversion method and device based on distributed optical fibers, and relates to the technical field of tunnel deformation parameter inversion, and the method specifically comprises the steps: selecting a monitoring point in a tunnel, arranging an optical fiber sensor, and obtaining a Brillouin scattering signal generated by the optical fiber sensor under the stress effect of the corresponding monitoring point; extracting a Brillouin frequency shift corresponding to the monitoring point, and determining an environmental effect frequency shift correction value by using the temperature change data, the humidity change data and the service time of the optical fiber sensor; determining the structural strain of the monitoring point according to the Brillouin frequency shift and the environmental effect frequency shift correction value, generating a three-dimensional strain field cloud picture, performing spatial registration on the three-dimensional strain field cloud picture and the geometric topology of the model, and performing three-dimensional grid division; the structure strain of the monitoring point is mapped to the grid node at the corresponding position, the structure strain of other grid nodes is determined based on the space distance attenuation law, inversion of the whole tunnel deformation parameters is completed, and the tunnel deformation monitoring precision and efficiency are improved.

The invention relates to the technical field of equipment intelligent diagnosis, in particular to a generator digital twinning diagnosis method based on temperature field online simulation, which comprises the following steps: respectively constructing a three-dimensional virtual model, a temperature field simulation model and an intelligent agent diagnosis model; the digital twin is in data connection with the physical entity; data alignment is achieved through a unified space-time coordinate system, a mapping relation is established between a vertex of the three-dimensional virtual model and a grid node in the temperature field simulation model, and a fault probability matrix output by the intelligent agent diagnosis model is matched to a three-dimensional grid of the three-dimensional virtual model through spatial interpolation; according to a fault probability output by the intelligent agent diagnosis model, correcting a boundary condition of the temperature field simulation model, triggering local grid adaptive encryption of the temperature field simulation model, and marking a fault position in real time through a three-dimensional virtual model; and the simulation result of the temperature field simulation model verifies the fault probability output by the intelligent agent diagnosis model. Through the method, the fault identification accuracy can be effectively improved.

The invention relates to the technical field of cold-chain medicine transportation, in particular to an intelligent monitoring and early warning system for a cold-chain medicine transportation environment. Comprising a multi-source heterogeneous data acquisition module, a time-space synchronization data fusion module, a three-dimensional thermodynamic distribution modeling module, a drug property degradation risk assessment module, a dynamic grading early warning module and a path self-optimization execution module, the multi-source heterogeneous data acquisition module is used for generating multi-source heterogeneous data; the space-time synchronization data fusion module generates an environment parameter matrix; the three-dimensional thermodynamic distribution modeling module calculates an equivalent thermal load value of each grid node; the drug property degradation risk assessment module predicts residual effective time; the dynamic grading early warning module generates a dynamic early warning signal; and the path self-optimization execution module generates optimized transportation path parameters. According to the invention, the medicine is prevented from losing efficacy due to environment change, and meanwhile, the response speed of cold-chain transportation and the medicine quality guarantee are improved.

The invention discloses a rapid calculation method based on a non-coordination and reanalysis theory and a computer device. The invention relates to the technical field of structural design optimization, and solves the problem that it is difficult to realize efficient structural response analysis while ensuring the calculation precision in the prior art. The method comprises the following steps: selecting a structural member of a to-be-calculated structural response and discretizing the structural member; preprocessing the data, and determining a coarse and fine grid index set corresponding to interface fine grid nodes; solving a unit shape function, and embedding the shape function into a body stiffness matrix according to a degree-of-freedom position to serve as an interface constraint condition; calculating a variable element stiffness matrix, and completing the construction of a modified model stiffness matrix; carrying out iterative calculation to obtain a basis vector; and calculating and solving the structural response of the modified model. According to the method, the efficient response analysis of the non-coordinated variable structure is realized on the premise of ensuring the calculation precision.

A method for predicting gas explosion. The method includes: obtaining a geometric model with spatial obstacles; determining a simulation dataset of the gas explosion based on the geometric model, where the simulation dataset is a dataset including process data of multiple gas explosion scenes in the geometric model, and a state vector of a spatial grid node in the simulation dataset includes a first moment, coordinate information and a target pressure value; optimizing, using the simulation dataset, a physics-informed GNN to obtain an optimized physics-informed GNN; and predicting the gas explosion in an obstructed gas explosion scenario using the optimized physics-informed GNN. This method is applied to use optimized physics-informed GNN to realize spatiotemporal second-level prediction of overpressure distribution and explosion wave propagations occurring in the obstructed gas explosion scenario, enhance the accuracy, the reliability and the efficiency of gas explosion prediction in the obstructed gas explosion scenario.

The invention provides a grid quality optimization method, computer equipment, a program product and a storage medium. The method comprises the steps that an initial grid is acquired, the initial grid is converted into a built-in data structure, and a grid object is acquired, wherein the grid object comprises a node set, a tetrahedron set and topological adjacency information; traversing each grid node of the grid object, and adding a constraint condition for each grid node; regarding the discretized three-dimensional grid as a hyperelastic body, and calculating the energy of the current grid for each tetrahedron unit in the grid object through a hyperelastic model; performing iterative calculation on the energy minimum value of the current grid according to the constraint condition, and applying random disturbance after each N iterations until a convergence threshold value or the maximum number of iterations is reached; and the grid corresponding to the minimum energy value is output as the optimal grid. According to the scheme, the overall quality of the grid can be effectively improved, and the boundary retention capability and the restoration capability of the degraded unit are enhanced, so that the simulation quality is improved.

According to the WIFI mesh channel dynamic allocation method and device provided by the invention, the size and position allocation of the RU are intelligently adjusted by monitoring the network flow, the equipment density and the channel interference in real time, so that the channel resource utilization rate is improved; a four-dimensional candidate channel scoring model and a decision engine are adopted for channel allocation and interference prediction, millisecond-level RU dynamic scheduling is achieved, and the low-delay service requirements of AR / VR, 4K videos and the like are met; a global channel state is determined through signaling interaction and collaborative decision among Mesh nodes, and conflicts and performance bottlenecks caused by local optimization are avoided; and a Wi-Fi 6 / 6E / 7 RU flexible division mechanism is supported, different frequency bands and bandwidth configurations are adapted, and efficient access of heterogeneous equipment is ensured. In a word, according to the scheme disclosed by the invention, the problems of static property, low efficiency and interference in RU allocation in the existing WiFi Mesh network are solved, and high-reliability and low-delay intelligent channel resource management is realized through dynamic priority layering, service type adaptation and interference avoidance technologies.

The invention discloses a boundary grid node-oriented shipborne unmanned aerial vehicle delivery scheduling method and system, and relates to the technical field of shipborne unmanned aerial vehicle path optimization, and the method comprises the steps: carrying out the systematic grid division of a boundary region based on a geographic coordinate system and marine environment constraints, and constructing a boundary grid node set; mapping attribution is carried out based on boundary grid nodes, and path feasibility is evaluated; and according to an evaluation result, screening and scheduling candidate tasks, generating a corresponding time window, constructing an optimal path, and completing dynamic scheduling and path planning of multiple tasks through time, space and platform constraint linkage. By constructing a task space mapping structure, an airspace schedulability scoring mechanism and a path multi-dimensional cost evaluation model, dynamic, efficient and safe scheduling of the shipborne unmanned aerial vehicle in a boundary sea area is realized, the task matching precision, the path feasibility and the system robustness are remarkably improved, and the method has high intelligent cooperation capability and environmental adaptability.

The invention relates to the field of three-dimensional scene rendering, in particular to a three-dimensional scene data generation method, an analysis method, rendering equipment and a computer storage medium. According to the method, LOD (Level of Detail) grading processing is performed on three-dimensional scene data, foreground data in the three-dimensional scene data is divided into a plurality of LOD layers, each LOD layer is divided into a plurality of grid nodes, and each grid node is independently stored, so that the data can be loaded as required. Meanwhile, different compression modes are adopted for different data, the GPU can directly analyze and read the compressed data, decompression in a memory is not needed, occupation of the memory and video memory bandwidth is greatly reduced, and rendering efficiency is improved.

The invention relates to a valve health degree monitoring and maintaining method and system based on an electric actuator, and belongs to the technical field of mechanical control and automation. The method comprises the steps that valve maintenance data of the electric actuator are obtained, a position-pressure bimodal grid is constructed based on the valve maintenance data, pressure testing is conducted on a valve, torque-opening curves in the opening and closing processes are analyzed, and valve health degree data are obtained; an electric execution monitoring model is constructed, a valve torque curve is obtained, maintenance information labels are added to grid nodes according to the valve maintenance information through a labeling binding method, and a valve health maintenance scheme is generated; for the pressure test, identifying switching time delay, hysteresis and leakage data in a valve torque profile and an operation log through a curve feature extraction method to obtain a valve scheme optimization strategy; and executing the maintenance scheme optimized by the valve scheme optimization strategy, and automatically outputting the updated valve health maintenance scheme based on a recursive execution feedback method.

The invention belongs to the field of machine tool geometric accuracy design, and particularly discloses a machine tool geometric accuracy analysis optimization method based on dynamic sensitivity, which comprises the following steps: constructing an error transfer model, and determining a target function; selecting an SOBOL method based on the global influence demand for quantifying each error term; selecting a Morris method based on the non-linear influence demand of the evaluation error term range change; dividing the travel of each movement shaft of the machine tool into grid nodes; selecting an error initial value for calculation based on parameter input and an error range, performing sensitivity calculation by using SOBOL, and taking a result as an optimization constraint condition; performing sensitivity calculation by using Morris, and taking a result as an optimization iteration basis; carrying out loop iteration and updating a sensitivity result to guide precision optimization, and carrying out multi-round iterative optimization until iterative convergence to obtain an optimized error term; and traversing grid nodes and carrying out sensitivity normalization in the transverse direction to obtain an optimized error curve in which error terms change along with coordinates. The geometric accuracy of machine tool machining can be improved.

The invention provides a safety coefficient prediction method based on slope three-dimensional finite element analysis and application, and relates to the field of hydraulic geotechnical engineering. Comprising the following steps: inputting a finite element grid, and calculating a geometric topological relation of the three-dimensional finite element grid; the method comprises the following steps: inputting position information and mechanical parameters of a bottom sliding surface and a side sliding surface of a possible sliding block, calculating to obtain a description equation of the possible sliding block, performing cutting calculation by utilizing a sliding surface equation and geometry of a finite element grid to obtain polygons of a boundary surface of the sliding block, decomposing each polygon into triangular grids, and analyzing a stress result based on a slope finite element to obtain a stress result of the boundary surface of the possible sliding block. Calculating stress values of triangular mesh nodes of the boundary surface through line interpolation; integration is carried out on the sliding force and the anti-sliding force of the boundary surface, and the safety coefficient of the possible sliding block body is calculated; according to the method, efficient intersection of the sliding surface and the finite element grid is achieved, geometric processing efficiency is improved, calculation efficiency and calculation accuracy are improved, and the method is suitable for three-dimensional irregular sliding block analysis in the fields of water conservancy, mines and the like.

The invention belongs to the technical field of engineering simulation data processing, and particularly discloses a finite element grid graph structure construction method and system, a terminal and a medium. Comprising the following steps: analyzing original full-amount grid data exported by finite element simulation, segmenting unstructured grid data into a grid vertex coordinate set and a grid unit mark number set, and constructing a graph edge topological structure corresponding to a grid based on a unit mark number relationship; on the basis, loading physical field data and adopting a tolerance-based coordinate matching algorithm to realize accurate mapping of physical field labels and material attributes with grid nodes; and generating standardized finite element grid graph data which can be directly used for graph neural network processing. By means of the method, high-consistency and high-physical-reliability conversion from the finite element simulation data to the graph structure data is achieved, and the physical field modeling and simulation acceleration capacity based on graph learning is improved.

The invention discloses a petroleum pipeline service life prediction system based on a 3D technology, and relates to the technical field of pipeline service life detection, and the system comprises a contour reconstruction module S11 which is used for obtaining the surface three-dimensional point cloud image data of a petroleum pipeline housing in an actual operation environment, and carrying out the surface contour reconstruction processing based on the three-dimensional point cloud image data, extracting space geometric structure data containing a local deformation defect area; the force-heat coupling analysis module S12 is used for constructing a finite element analysis grid according to the space geometric structure data and loading operation condition data into corresponding grid nodes to generate a force-heat coupling stress field, and the operation condition data comprises a periodic internal working pressure value, along-line temperature field data and an axial disturbance stress tensor; the feature extraction module S13 is used for calculating an equivalent stress distribution value of a defect area node based on a force-heat coupling stress field; according to the invention, accurate quantification of fatigue damage evolution of the defect area is realized.

The invention discloses a data detection method and system based on gravitational-magnetic-electric-seismic signal joint gridding, and relates to the technical field of geological prospecting, and the method comprises the steps: determining a research region according to an exploration target and geological conditions, and dividing the research region into grids with the same size; on each grid node, gravity, magnetic, electric and seismic signals are collected; the collected gravity, magnetic and electric shock signals are preprocessed; performing feature extraction on the preprocessed gravity-magnetic-electric-seismic signals by using an independent component analysis algorithm, and performing data fusion by using a convolutional neural network in deep learning; establishing an initial value estimation incidence relation model, and obtaining an initial value prediction result based on the fused data; according to the initial value prediction result, geological parameters are obtained through solving by using a joint inversion algorithm. According to the invention, efficient combined acquisition of gravity, magnetic, electric and seismic signals is realized, the precision and quality of acquired data are improved, and unified management and fusion analysis of the data are realized.

The invention relates to a tunnel ventilation system parameter optimization method based on neural network model driving, and the method comprises the steps: dividing a tunnel into a plurality of irregular grid nodes according to the three-dimensional geometrical characteristics, aerodynamic characteristics and pollutant distribution characteristics of the tunnel; constructing a nonlinear edge weight calculation function according to the spatial distribution data of the plurality of irregular grid nodes in combination with a pollutant diffusion coefficient, a real-time wind speed field and a spatial position relationship among the grid nodes, and generating an edge weight matrix representing interaction strength among the grid nodes; inputting the grid node information of the plurality of irregular grid nodes and the edge weight matrix into a space-time diagram neural network model to obtain multivariable space-time dynamic characteristics of the pollutants; according to the multivariable space-time dynamic characteristics, pollutant concentration overrun probabilities of the irregular grid nodes are determined; and according to the pollutant concentration overrun probability of the irregular grid nodes, ventilation regulation and control parameters of the tunnel ventilation system are optimized. The design of ventilation system parameters can be optimized.

The invention belongs to the technical field of control detection, and particularly relates to a laser cutting machine cutting control method and system.The laser cutting machine cutting control method comprises the following steps that S1, frequency sweeping excitation is applied to a plurality of preset grid node positions of a laser cutting machine workbench, and vibration response data of a cutting head is collected; a dominant pole and a zero point of a vibration transfer function of each grid node are identified and extracted, spatial distribution is fitted by using a non-uniform rational B-spline surface, and a spatial model of continuous mapping between a cutting head position and a vibration transfer function parameter is established; s2, decomposing a to-be-processed path into a plurality of path infinitesimal segments, and calculating a preliminary feed-forward compensation instruction of each path infinitesimal segment based on a space model and a zero-phase error tracking control algorithm; according to the invention, the contour error caused by structural vibration and non-linear factors is reduced, the processing precision and the cutting surface quality of the laser cutting machine in a full working range are improved, and the forming effect of a complex contour at a high speed is improved.

The invention discloses a bridge structure monitoring method based on multi-source data and digital twinning and a related device, and relates to the technical field of bridge engineering.The method comprises the steps that an initial finite element model of a target bridge is constructed based on a design drawing and structural parameters of the target bridge, and the initial finite element model of the target bridge is constructed based on the initial finite element model and preset load conditions; calculating to obtain simulation mechanical response data of each grid node of the target bridge; based on the simulation mechanical response data and the obtained actually measured mechanical response data of each grid node of the target bridge, obtaining a corrected finite element model; inputting the obtained traffic load data passing through the target bridge into a digital twinborn calculation model constructed based on the corrected influence line data for numerical calculation to obtain an overall structure response cloud picture and a structure response-time curve of the key monitoring point, and fusing the overall structure response cloud picture and the structure response-time curve with the obtained three-dimensional visualization model to obtain a three-dimensional visualization model; and obtaining the structural state of the target bridge. According to the invention, accurate monitoring and real-time state evaluation of the bridge structure are realized.

The invention discloses a curved surface porous structure modeling simulation method and system for finite element analysis, and belongs to the technical field of mechanical engineering simulation, and the method mainly comprises the steps: carrying out the initial grid division of a target curved surface part, and obtaining a grid node set covering a curved surface; according to a preset hole center coordinate, three closest grid nodes are automatically searched in the grid node set, a local reference plane is established, and the hole axis direction is determined; generating an offset sketch plane in the normal direction of the reference plane, projecting the hole center to the sketch plane, drawing a circular sketch, and executing a stretching and cutting command to form a hole; and circularly executing the operation to realize automatic generation of multiple holes on the curved surface. And finally, the porous structure is subjected to grid division again, so that subsequent operation is facilitated. According to the method, the modeling process is combined with parameterization analysis and an automatic process, the efficiency and stability of porous modeling of the complex curved surface structure are remarkably improved, and a basis is provided for finite element pretreatment of the complex curved surface structure.

The invention belongs to the technical field of aircraft strength design, and particularly relates to an aircraft structure strain field reconstruction method and device based on a deep learning neural network, and the method comprises the steps: S1, carrying out the dimension reduction mapping of irregularly distributed three-dimensional grid nodes of a finite element model, and obtaining regular two-dimensional grid nodes; s2, obtaining calculation results of each two-dimensional grid node under a plurality of working conditions based on finite element simulation, and constructing a training set; s3, training the U-NET deep convolutional network through the training set to obtain a strain field proxy model; and S4, based on the strain field agent model, processing an input image with a small number of measurement point position measurement results to obtain an output image containing an aircraft structure strain field. According to the invention, the strain field of the whole aircraft can be reconstructed through displacement data of a small number of points measured by the static test of the whole aircraft.

The present invention relates to a calculation method for the general scour pattern of a cross-sea bridge considering the turbulence effect. The research area including the cross-sea bridge project is determined and meshed, the elevations of each grid node are obtained, typical computational hydrological boundary conditions are selected, a non-hydrostatic three-dimensional mathematical model including the entire cross-sea bridge project is established, the hydrodynamic fields before and after the bridge construction are calculated, the tidal level and flow velocity are verified, the pulsating velocity at each grid node, the sediment settlement velocity in still water and the settlement velocity under the influence of turbulent kinetic energy are calculated, the sediment carrying capacity is calculated according to the estuarine fine sediment model, and finally the general scour amplitude of the seabed at each grid node position is calculated, thereby obtaining the characteristics of the general scour amplitude in the upstream and downstream areas of the bridge. Combining the plane three-dimensional numerical model and the sediment carrying capacity calculation method considering the turbulence effect, through the dynamic equilibrium theory, the accurate prediction of the general scour pattern of the cross-sea bridge including the depth and range is realized, and the calculation accuracy of the general scour pattern of the cross-sea bridge is improved.

The invention discloses a meshless EFGM and style migration-based multi-material structure topological optimization method, which comprises the following steps of: (1) inputting material attributes such as Young modulus, shear modulus and Poisson's ratio of a structure, and discretizing a design domain by using meshless nodes; (2) transforming the designed meshless density field to obtain a physical density field, and performing length scale filtering; (3) an alternating active phase algorithm is executed, and the displacement field, the system flexibility and the related sensitivity of the structure are solved based on the meshless EFGM; (4) executing a neural style migration program to calculate the difference between the current topological structure and the reference style image, and calculating the derivative of the physical density field of the topological structure to the network parameter; and (5) updating a design variable by using a moving asymptote algorithm, and judging whether an optimization loop is converged or not. According to the method, based on meshless EFGM and style migration multi-material structure topological optimization, topological structures with different artistic styles can be designed, and the method has engineering application value.

The invention relates to an ocean platform pipeline wiring method and system based on an ant colony optimization algorithm, and a medium, and belongs to the technical field of intelligent optimization algorithms. Comprising the following steps: S11, environment modeling: discretizing a three-dimensional space of an ocean platform into a grid model, marking an obstacle region and a feasible region, and calculating a grid node energy value; s12, parameter initialization: constructing a heuristic information matrix based on pipeline inlet and outlet coordinates and equipment parameters, and determining a node transition probability by the matrix through a direction probability allocation strategy and a distance sensitive function; s13, path optimization: generating candidate paths through dynamic smooth search, local pheromone adaptive adjustment and collaborative ecological niche search, and updating pheromones; and S14, constraint verification and result output: verifying whether the path meets the engineering constraint or not, and outputting the three-dimensional coordinate, length and bend number of the optimal pipeline path after iterative optimization. According to the method, efficient, safe and economical wiring of the offshore platform pipeline is achieved.

The invention relates to a flow field prediction model training method and device based on Vision Transform, computer equipment and a storage medium. A flow field prediction model is constructed based on Vision Transform and comprises a feature coding layer, a linear projection layer, a coding layer and a decoding layer, and the method comprises the following steps: constructing grid data of a target cascade, and generating simulation flow field data corresponding to the target cascade; coding the coordinate of each grid node to obtain a coordinate code; inputting the coordinate codes and the working conditions of the grid nodes into the flow field prediction model, performing feature extraction processing on the one-dimensional feature sequence through a coding layer and a decoding layer in sequence to obtain predicted flow field data, and adjusting the flow field prediction model according to the difference between the predicted flow field data and the simulated flow field data to obtain the flow field prediction model. And obtaining a trained flow field prediction model. By adopting the method, the flow field prediction efficiency can be improved.

The invention belongs to the technical field of image processing and three-dimensional modeling, and particularly relates to an SLAM three-dimensional modeling method suitable for a complex dynamic scene. By constructing a multi-modal feature hierarchical processing architecture, the system realizes technical breakthrough in three dimensions: firstly, in combination with a pixel mask pattern and a multi-view geometric method, static feature points are reserved while dynamic feature points are removed as far as possible; secondly, according to the optimal flow field mapping relation of the grid nodes, static background restoration of the shielded area is achieved by establishing the geometric constraint relation of adjacent key frames; and finally, the depth perception data and the optimized feature set are fused to construct a tightly coupled three-dimensional reconstruction model, and a closed-loop optimization framework with motion robustness is formed.

The invention discloses a load out-of-cabin process numerical simulation method based on dynamic nested grid motion. Self-adaptive control is carried out on grid node motion by defining a UDF program of grid motion. The method has the advantages that the flow characteristics, especially the flow field characteristics of a boundary layer, in the load cabin-out process can be simulated more accurately by adopting a dynamic nested grid motion control technology; through self-adaptive control of grid node movement, the method effectively solves the problem of isolated units caused by hole digging processing, the improvement remarkably improves the grid quality and the numerical solution precision, it is ensured that the grid quality is kept in numerical simulation of the load cabin-out process, meanwhile, the accuracy and reliability of a simulation result are improved, and the method is suitable for large-scale popularization and application. The problem of solution distortion caused by isolated units at a boundary layer in a traditional nested grid technology is solved.

The invention discloses an image training data optimization method and system based on dynamic distillation, and belongs to the technical field of image processing, and the method specifically comprises the steps: dynamically dividing an input image into non-uniform grids according to an image region feature entropy value; constructing a teacher model and a student model, inputting a non-uniform grid image, extracting a feature vector of each grid node, and calculating a feature difference degree; dynamically generating a distillation temperature coefficient based on the characteristic difference degree, and reconstructing a distillation loss function to obtain an optimized loss value; calculating a loss curvature by using the loss value, screening difficult samples and easy samples, and respectively implementing dual-stage enhancement and feature disturbance enhancement; the feature migration rate is calculated based on the enhanced sample set and the feature vector, the teacher model intermediate layer feature weight is dynamically adjusted and fed back to the model training link, and the knowledge distillation effect is improved.

The embodiment of the invention discloses a method and device for generating a dam surface settlement cloud picture based on a self-adaptive interpolation model. A specific embodiment of the method comprises the steps of performing feature extraction on multi-dimensional monitoring point data of a preset dam structure to generate a dam body scene portrait; according to the dam body scene portrait, generating an optimal interpolation algorithm; optimizing the optimal interpolation algorithm to generate an optimized interpolation algorithm; generating a non-oscillation cloud picture according to dam body physical attributes of a preset dam structure and an optimization interpolation algorithm; performing preset-size grid block division on the non-oscillation cloud picture to generate a cloud picture grid node set; performing parallel processing on each cloud picture grid node in the cloud picture grid node set to generate an accelerated interpolation result; and performing incremental updating on the accelerated interpolation result to generate an updated interpolation result. According to the implementation mode, the accuracy, efficiency and engineering applicability of dam settlement interpolation are improved.