45 results about "Physical computation" patented technology

The invention discloses an interdisciplinary PBL teaching plan intelligent generation system based on knowledge graph and multi-Agent cooperation, and the system comprises an interdisciplinary knowledge base construction module which constructs knowledge graphs of mathematics, physics and computer subjects through semantic vectorization and entity alignment; the retrieval enhancement module is used for dynamically supplementing knowledge blind areas based on an RAG mechanism; the teaching demand analysis module is used for analyzing teacher input and generating a teaching target matrix; the structured modeling module is used for decomposing tasks and designing a teaching framework according to a three-dimensional target system (cognition, ability and subject literacy); the teaching plan generation module is used for generating subject contents through multi-Agent collaboration; and the teaching plan optimization module verifies interdisciplinary consistency and outputs a standardized teaching plan. According to the method, the knowledge graph and multi-Agent dynamic weight distribution are creatively fused, deep integration of interdisciplinary knowledge is realized, teaching plan generation efficiency and scientificity are remarkably improved, and intelligent lesson preparation support is provided for teachers.

The invention provides an atom pair distribution function calculation method based on a back propagation algorithm, and the method comprises the steps: building a micromapping relation between a material microstructure and experimental PDF data through constructing a physical calculation layer containing lattice parameters and atom displacement; defining a loss function by taking experimental data as a supervision signal, and synchronously optimizing tens of thousands of atomic positions and lattice parameters by utilizing gradient back propagation of the loss function covering the experimental data and model calculation result deviation; and a special optimization algorithm CrystalAdam and dynamic learning rate scheduling are combined, so that global efficient search of a high-dimensional parameter space is realized. Therefore, compared with a traditional modeling method, the method provided by the invention not only can remarkably improve the optimization efficiency, but also can provide finer local adjustment in a complex structure space, thereby achieving higher precision.

The invention discloses an electrical impedance tomography method, which integrates a physical model and deep learning to construct a deep learning framework with physical constraints so as to improve physical consistency and reconstruction precision. According to the method, a neural network based on an encoder-decoder structure is adopted, and measured boundary voltage is used as input to predict conductivity distribution of a target object; in order to improve the performance of the neural network, a physical Jacobian matrix is introduced, and the sensitivity of boundary voltage to conductivity is described; and constructing a physical regularization loss function, comparing a physically calculated Jacobi matrix with a neural network predicted Jacobi matrix, and combining a final total loss function with data error loss and physical consistency loss. According to the method, data driving model training is guided through physical constraints, combination of physical information and deep learning is achieved, the precision of electrical impedance reconstruction and the generalization ability of the model are remarkably improved, and a new solution thought is provided for the fields of medicine, flexible electronics, environment monitoring and the like.

The invention belongs to the technical field of computational fluid mechanics and artificial intelligence crossing, and discloses an anisotropic turbulence modeling method and system based on scale adaptive physical convolution. Comprising a physical anchor point expansion layer used for improving a physical constraint coverage range, a physical information neural network basic full-connection layer, a convolution layer comprising a feature scale convolution kernel generation mechanism and a loss calculation process. According to the method, a physical anchor point expansion layer is introduced, additional physical calculation points are generated by setting a physical point expansion coefficient on the basis of training data points of a traditional physical information neural network, and the additional physical calculation points are combined with an original data point set to form a physical constraint calculation point set which is wider in coverage range and more reasonable in distribution; a convolution kernel size self-adaptive generation mechanism based on a turbulence characteristic scale is utilized, the turbulence characteristic scale is obtained through vorticity field analysis and energy spectrum calculation, the size of a convolution kernel is determined according to the turbulence characteristic scale, convolution operation can be matched with multi-scale physical characteristics of turbulence in a self-adaptive mode, and pertinence is higher.

The invention relates to the technical field of gamma radiation dose calculation, and particularly discloses a gamma radiation dose calculation method based on a Unity engine, a program, equipment and a storage medium. In order to solve the problems that ray tracing is complex and low in efficiency in complex irregular geometry by means of an existing point kernel integration method, the physical calculation capacity of a Unity engine is introduced, ray tracing is conducted through Physics. Raycast interface iteration, and gamma radiation dose calculation under the multi-layer shielding condition is optimized. The method comprises the following steps: acquiring radioactive source energy, intensity, shielding body density, mass attenuation coefficient and detection point coordinates; constructing a radioactive source model and generating a point kernel set through parallel array ray discretization; performing ray tracing on the point core set to obtain a list of all intersection points of rays and the shielding body; and calculating the shielding thickness of each layer according to the intersection point list, calculating the dose contribution of each point core in combination with the accumulation factor, and summing to obtain a total dose value.

In an example of a game according to an exemplary embodiment, a movement of a movable dynamic object placed in a virtual space is controlled based on physical calculations, and based on an operation input, a player character is caused to perform an object operation action including a first operation of moving a dynamic object specified based on the object operation action and a second operation of forming an assembly object by linking the specified dynamic object to the other dynamic object, and if the assembly object including the dynamic object specified based on the object operation action includes the dynamic object that the player character is mounting, the dynamic object is prevented from moving based on the object operation action.

The invention discloses a virtual character behavior processing method and system based on semantic decoupling and elastic coupling and a computer readable storage medium. According to the method, unstructured action data is decomposed into physical layer intention descriptors, measurable style descriptors and high-dimensional semantic layer features by utilizing a parallel physical calculation engine and an AI semantic analysis engine through a double-track feature decoupling mechanism. Besides, an elastic coupling mechanism between intentions and styles is introduced, style parameters are dynamically clamped based on a physical priority principle, and physical topology collapse caused by style overload is prevented. According to the method, a physical-semantic dual index system is constructed, database autonomous evolution based on manifold density analysis is supported, and the problems that in the prior art, virtual character action generation is poor in physical controllability, semantic understanding is lacked, and cross-scene generalization ability is weak are effectively solved.

The present application relates to the technical field of short-circuit current calculation, and particularly relates to a flexible DC system short-circuit current calculation method based on a physical information neural network, comprising: setting a model input feature vector, the model input feature vector being used to represent a system operating state before a fault and fault information, and performing data preprocessing on the model input feature vector; constructing a hybrid driving calculation model, the hybrid driving calculation model comprising a physical calculation module and a neural network module, and being coupled based on a preset fusion architecture; performing end-to-end training and optimization on the hybrid driving calculation model by using a preset composite loss function; and performing flexible DC system short-circuit current calculation based on the optimized hybrid driving calculation model, so that the problems of poor precision, speed and convergence, poor interpretability and weak generalization ability in the prior art are solved.

This invention provides an underwater long-range ultra-high-definition polarization imaging system and method, relating to the field of data processing technology. The system comprises, from bottom to top, an optical sensing layer, a physical computing layer, an AI inference enhancement layer, an adaptive control layer, and a decision optimization layer, forming a hierarchical control structure where data flows upward and strategies are passed downward. The optical sensing layer employs a non-uniform polarization sampling array, a spectral polarization joint coding unit, and a dual-modal illumination unit to simultaneously acquire intensity, multi-directional polarization, multispectral images, and depth information. The physical computing layer calculates polarization parameters in real time and inverts water turbidity within the image signal processor hardware pipeline. The AI ​​inference enhancement layer embeds the physical model as a differentiable module into a neural network to achieve scattering suppression and detail enhancement. The adaptive control layer adopts a time-scale separated dual-closed-loop architecture. This invention can adapt to different water turbidity levels, output ultra-high-definition images in real time, and significantly improve the quality of long-range underwater imaging.

The invention discloses an ultra-parallel microwave photon Isin machine system based on wavelength division multiplexing and an optimization method, and belongs to the technical field of photon calculation and combination optimization. The Isin machine system comprises a multi-wavelength light source module, a microwave modulation and wavelength division multiplexing coding module, a nonlinear calculation and evolution module and a demultiplexing and dynamic feedback module, all the modules are optically and electrically connected to form a closed-loop system, and different optimization problems can be coded to independent wavelength channels to achieve parallel solving. The optimization method is applied to the Isin machine system, and closed-loop optimization is completed through the steps of coupling matrix mapping, microwave modulation loading, nonlinear evolution, dynamic feedback control and the like. The method supports synchronous processing of 1024 optimization problems, and energy consumption of a single problem is lt; the method is mainly applied to the fields of logistics scheduling, financial modeling, communication network optimization and the like, and an efficient physical calculation solution is provided for a large-scale combination optimization problem.

This application provides a method, system, device, and medium for lead compound design based on deep reinforcement learning. It innovatively achieves de novo design of covalent inhibitors targeting affinity, proposes an innovative method for constructing novel drug molecules based on multiple different bioactive fragments, and optimizes the model's exploration space. The application of the physics computation software Autodock-GPU as a scoring function enhances the model's generalization ability and running speed. The application of Munchausen reinforcement learning techniques to the Soft Actor Critic discrete action space model improves the model's optimization performance. Compared to other models, this application expands and improves upon the model's functionality and the properties of the generated molecules.

An apparatus and method is provided for simulating a physical computer system using virtualization. The disclosed system virtualizes devices of a physical computer system by modeling hardware and software components that are physically present within the physical computer system. The system simulates changing at least one of a piece of hardware or software in the virtualized computer system and assesses an effect of the change in the virtualized computer system to determine a potential effect of the change if the change were implemented on the physical computer system.

The invention discloses a simulation system and method of a flight simulator, relates to the technical field of flight simulation, and can solve the problem that flight simulation is not perfect enough. The system comprises a simulation cabin, an operation loading module and a simulation calculation module, the simulation cabin comprises a simulation control instrument; the simulation calculation module comprises a target physical calculation engine, and a flight calculation model is deployed in the target physical calculation engine; the operation loading module is used for acquiring a manipulation instruction based on the simulation manipulation instrument in real time and transmitting the manipulation instruction to the simulation calculation module; the simulation calculation module inputs the control instruction into a target physical calculation engine, the target physical calculation engine is used for solving a flight simulation result when the aircraft takes off on the ski-jump deck, and the flight simulation result comprises flight attitude information, flight motion information and collision information; the collision information is used for indicating the collision information of the aircraft and the ski-jump deck, and determining the ski-jump take-off effect of the flight simulator according to the flight simulation result in real time.

A server includes physical compute nodes. Each physical compute node includes a host and physical management resources. The physical management resources include a physical management processor. The server includes a distributed hypervisor to provide a distributed application operating environment that is hosted by the physical management resources. The distributed hypervisor to allocate, from the physical management processors, virtual processors for the distributed application operating environment to execute applications to manage the physical compute nodes. The distributed hypervisor includes a plurality of hyper-kernels that are associated with respective physical compute nodes. Each hyper-kernel is hosted on the physical management resources of the associated physical compute node.

The present disclosure provides a pipelining and parallelizing graph execution method for neural network model computation and apparatus, and provides a pipelining and parallelizing graph execution method for neural network model computation and apparatus in a deep learning training system. The method includes the graph execution flow in a neural network model computation process and a process of cooperative work of all functional modules. The pipelining and parallelizing graph execution method for neural network model computation includes creating a graph executive on a native machine according to a physical computation graph compiled and generated by a deep learning framework.

This invention discloses an industrial agent reasoning framework and method based on physical first-principle constraints, belonging to the field of industrial intelligence technology. The framework includes a data access layer, a task parsing layer, a candidate solution generation layer, a physical constraint verification layer, a conflict rollback and replanning layer, an execution decision layer, and a knowledge accumulation layer. The physical constraint verification layer uses at least one physical first-principle model as a pre-constraint hard constraint gating before the candidate decision sequence enters the execution decision layer. It performs parameter mapping, boundary condition solving, consistency calculation, residual determination, and executability determination on the physical computation variables obtained from the structured reasoning task mapping, and outputs structured conflict information when the verification fails. The conflict rollback and replanning layer performs targeted rollback and replanning based on the violation type, violation node level, and source of missing boundary conditions in the structured conflict information. The execution decision layer only outputs execution-type output results for candidate decision sequences that have obtained the pass flag. The knowledge accumulation layer writes the structured conflict information and the corrected pass results into a rule base, template base, or historical case base to serve as filtering conditions, initial parameter values, tool sorting criteria, or threshold setting criteria during subsequent candidate generation. This invention can improve the physical feasibility, reliability, interpretability, and auditability of industrial agent decision-making results.

The application relates to a V-belt life optimization design method based on a particle swarm algorithm, which comprises the following steps: constructing a life calculation model of a rubber V-belt; adopting a uniform test design method to take the tensile strength, reference force elongation, hardness and elastic modulus of the V-belt as sample inputs, and dividing a training set, a verification set and a test set; obtaining a life prediction value by using a pre-established life calculation model; taking the maximum value of the life of the V-belt as an objective function, the tensile strength, reference force elongation, hardness and elastic modulus of the V-belt as to-be-optimized parameters, and the slip rate and tension force failure threshold of the dynamic operation of the V-belt as constraint conditions. Compared with traditional optimization of the volume and size of the whole belt transmission system and simple physical calculation optimization of the life, the application has better global search capability and faster convergence speed for multi-parameter optimization, and has obvious practical value and broad application prospect.

The invention relates to the technical field of rock material elastic modulus field and Poisson's ratio field prediction, in particular to a rock elastic modulus and Poisson's ratio field prediction method based on physical constraints, which comprises the following steps: constructing a rock mechanical sample data set; the data set comprises a rock elastic modulus field, a Poisson's ratio field, a strain field, a stress field and space coordinates, a physical information neural network is constructed, an encoder, a decoder and a physical calculation layer are constructed, and the physical calculation layer is used for calculating a mechanical response strain field; the method comprises the steps of constructing a multi-element composite loss function fused with physical constraints on the basis of a physical information neural network, performing iterative training on the physical information neural network on the basis of the multi-element composite loss function, and outputting a high-resolution elastic modulus field and a Poisson's ratio field of a target rock test piece. It is ensured that the predicted elastic modulus and Poisson's ratio parameter field meets the basic law of rock mechanics, and the credibility of the prediction result is improved.

The invention provides a Gaussian-Seidel-based MMPA burnup equation set accelerated solving method, belongs to the field of reactor physical calculation, and aims to solve the technical problems of high solving complexity, long calculation time and low efficiency of an existing MMPA burnup equation set. The method comprises the following steps: firstly, based on current reactor burnup working condition parameters, accurately constructing a burnup matrix under the current burnup condition through nuclear data processing and burnup chain modeling; the method comprises the steps that firstly, an MMPA burnup equation is constructed, an MMPA burnup equation set suitable for Gaussian Seidel iteration solution is obtained through derivation, finally, an iteration convergence criterion is set, and the Gaussian Seidel iteration method is adopted for conducting iteration solution on the MMPA burnup equation set till the convergence condition is met; and finally, outputting a nuclide density distribution result at a corresponding moment. According to the method, the calculation efficiency can be greatly improved on the premise that the calculation precision is guaranteed, and the problems that a traditional MMPA burnup equation is high in solving complexity and long in calculation time consumption are solved.

For at least one floating object disposed in the air, of dynamic objects disposed in a virtual space, movements of the dynamic objects being controlled based on physical calculation, a load applied to the floating object is calculated based on a game process. In a first case in which the load does not exceed a first limit load in a first direction including at least a downward direction in the virtual space, a first force that maintains a position of the floating object in the virtual space, is additionally applied to the floating object. In a second case in which the load exceeds the first limit load in the first direction, the first force is removed. The position and an orientation of the floating object are updated based on the physical calculation.

The invention relates to the field of reactor core physical calculation, in particular to a reactor core physical optimization calculation method under a nuclear power unit bias operation condition, and the method comprises the following steps: 1, constructing a geometric model and a reactor core grid model of a reactor in fluid dynamics software; 2, establishing a porous medium model in fluid dynamics software; 3, setting boundary conditions in the hydrodynamic software and carrying out thermal hydraulic calculation; 4, constructing a thermotechnical and physical coupling interface, taking a thermotechnical calculation result of the hydrodynamic software as input, and carrying out physical calculation in physical calculation software; 5, taking the three-dimensional power distribution of the physical calculation software as fluid dynamics software input, and carrying out thermal hydraulic calculation to obtain an optimized fuel assembly upper power peak factor and an optimized fuel assembly volume power peak factor; and 6, repeating the steps 3-5 until the power peak factor converges. According to the method, the physical parameters of the reactor core under the bias ring operation condition are calculated, and the control capability of safe operation of the reactor is further improved.

The invention discloses an urban road traffic state intelligent estimation method based on microphysical representation, and the method comprises the steps: constructing a road section boundary condition based on the arrival and departure accumulated flow information of an intersection and a gate, and forming multi-source traffic perception input through combining with road section observation data; constructing a physical-data hybrid driving model, constructing a deep learning model in a data driving branch, learning a mapping relation between road section boundary cumulative flow and a traffic state in a road section, and converting an indistinguishable Newell traffic flow model into a distinguishable computational graph structure through function approximation and structural conversion in a distinguishable physical branch; and constructing a loss function of the hybrid drive model, so that data drive output and a physical model are kept coordinated, and traffic flow parameters are jointly estimated. Through deep embedding of the microphysical calculation graph, the dependence of a pure physical model on an ideal assumed condition is made up, and meanwhile, the problem that a pure data driving method is insufficient in generalization ability in a sensing blind area is solved.

Movement-related parameters used in virtual physical calculation are changed such that a designated object selected based on an operation input is caused to perform return movement to return to previously recorded positions and orientations, sequentially backward from the time of giving a start instruction based on the operation input. A state in a virtual space including a player character, the designated object, and other objects is updated based on the virtual physical calculation.

This invention relates to the fields of artificial intelligence and computational materials science, and discloses a method and system for autonomous material discovery based on a large language model. The method includes: the system receiving computational task instructions in natural language form; a cognitive layer calling a large language model agent to process the instructions and outputting structured instructions containing action instructions and parameter sets; a control bridging layer merging and pattern-validating the parameter sets based on preset system parameters, generating file entities and writing them to the processing path; an execution layer starting a background daemon process and triggering atomic movement instructions from the operating system; a computation module parsing the file entities, initializing a graph neural network model, and performing iterative computation; during the computation process, collecting and transmitting runtime status data including a loss function, and having the large language model agent dynamically generate the next round of computational task instructions based on the numerical values. This invention establishes a closed loop of instruction parsing, mutual exclusion scheduling, and physical computation, improving the stability of autonomous material computation.

The invention relates to the technical field of earthquake sense prediction, in particular to an earthquake sense prediction method and device based on deep learning and physical knowledge driving and electronic equipment, and the method comprises the steps: collecting multi-dimensional basic parameters which comprise historical earthquake source parameters, historical structure array observation data and geological feature data; extracting physical constraint conditions based on the multi-dimensional basic parameters, wherein the physical constraint conditions comprise a first constraint and a second constraint; by taking the multi-dimensional basic parameters as input and physical constraint conditions as boundaries, constructing a vibration sense prediction model fusing physical calculation and deep learning; and designing a loss function containing a basic error term and a constraint penalty term for the vibration sense prediction model. According to the method, seismic physical laws can be fused, multi-dimensional data are combined to adapt to different geological scenes, and dynamic optimization can be realized through real-time observation.

The application discloses a steel modular building structure performance prediction method and system based on a graph neural network, and belongs to the technical field of physical calculation. In view of the problems that multi-specialty data is split, index calculation relies on manual work, and numerical simulation is time-consuming and difficult to support rapid iteration in the prior art, integrated digital basic JSON data is first constructed based on DXF drawings; then building material cost is automatically calculated, structure performance indexes are extracted, and building cold and heat loads are calculated; finally, a graph convolution network structure proxy model and a heterogeneous graph convolution network energy consumption proxy model are trained by taking the calculated indexes as a supervision signal, the structure topology and the energy consumption heterogeneous relationship are accurately captured through the graph neural network, and rapid and accurate prediction of structure performance and building energy consumption is realized. The application realizes homologous integration of multi-specialty data and automatic calculation of indexes, and provides technical support for efficient design of steel modular building structures.

The invention relates to the technical field of short-circuit current calculation, in particular to a flexible direct current system short-circuit current calculation method based on a physical information neural network, and the method comprises the steps: setting a model input feature vector which is used for representing a system operation state and fault information before a fault, carrying out data preprocessing on the model input feature vector; a hybrid drive calculation model is constructed, the hybrid drive calculation model comprises a physical calculation module and a neural network module, and coupling is carried out based on a preset fusion architecture; adopting a preset composite loss function to perform end-to-end training and optimization on the hybrid drive calculation model; and performing flexible DC system short-circuit current calculation based on the optimized hybrid drive calculation model. The method solves the problems of poor precision, speed and convergence, poor interpretability and weak generalization ability in the prior art.

The invention provides a miscible multiphase fluid state prediction method and device, equipment and a medium, and the method comprises the steps: obtaining the state of a fluid system at the current moment, and obtaining at least one physical parameter used for controlling the dynamic evolution behavior of the fluid; according to the Newton's law of motion, calculating initial predicted values of the position and the speed of the fluid particles at the next moment; utilizing the trained multiphase fluid prediction network to predict the position correction of fluid particles and the volume fraction of each phase at the next moment; and obtaining the position and speed of the fluid particle at the next moment according to the position of the fluid particle at the current moment, the initial predicted value of the position and the position correction amount. Physical calculation and multiphase fluid prediction network correction are combined, and the network only needs to learn the deviation from physical basic motion, so that the learning difficulty is reduced, and the simulation process is more stable; and physical parameters are independently used as an input, so that the universality and the flexibility of the model are greatly enhanced.

The invention discloses a variational nodal method for directly and quantitatively depicting a first-order anisotropic scattering process, which comprises the following steps of: after an input nodal section and geometric information are obtained, solving a P1 equation set for directly and quantitatively depicting the first-order anisotropic scattering process, deducing to obtain a second-order form of the P1 equation set, and constructing a variational nodal model for solving the P1 equation set in the second-order form; comprising functional construction, spatial discretization and variational derivation, a P1 equation set after numerical discretization is obtained, a power method is adopted for iterative solution, and accurate description of first-order anisotropic scattering is completed. According to the method, diffusion approximation can be eliminated, high algorithm calculation efficiency and iteration stability are ensured, transverse integration is eliminated, the first-order anisotropic scattering effect of light water and part of materials in the pressurized water reactor core is accurately described, and the physical calculation precision of the reactor core is improved.

The invention discloses a mathematical model-based wave visual simulation method and simulation system. The mathematical model-based wave visual simulation method comprises the following steps of: constructing a consistent wave spectrum model; initializing a wave spectrum model, and calculating and generating a wave amplitude array; acquiring the position, speed and course of the ship, and calculating encounter frequency; a wave height field is calculated and generated, and the height of encountered waves is calculated according to the position of the ship; wave force is calculated, the wave height field is accelerated through a GPU to generate a wave height map, and a wave scene is generated; calculating the position, speed, course and attitude of the ship, and updating the three-dimensional model display of the ship; according to the method, the problem that visual simulation and a dynamic response model are disjointed is effectively solved, the ship motion visual scene is highly consistent with a physical calculation result, immersive ship motion visual scene experience can be provided for a user, real interaction between a ship and waves can be accurately simulated, and the ship motion visual scene experience is improved. The practical value of visual simulation in the fields of ship control system design, seakeeping analysis and sailor safety training is improved.