416 results about "Stiffness matrix" patented technology

Disclosed in the present invention are a railway vehicle-track-bridge system dynamic response prediction method and system, and a medium. The method comprises: inputting vehicle speed samples and track irregularity samples into a vehicle-track-bridge system coupled random distribution physical model to obtain corresponding bridge dynamic responses and effective loads of the vehicle-track-bridge system, and extracting a global stiffness matrix of the vehicle-track-bridge system; constructing a training sample set; constructing a fitness function considering the effective loads, and then on the basis of the training sample set, using a genetic algorithm to optimize parameters of a BP neural network prediction model and training same to obtain a bridge dynamic response prediction model; and using the bridge dynamic response prediction model to carry out bridge dynamic response prediction. By introducing effective loads in a vehicle-track-bridge system into a fitness function in a genetic algorithm, a neural network model and a vehicle-track-bridge physical model are organically combined, thereby improving the prediction precision.

The invention relates to the field of thin-wall structure displacement field monitoring, and discloses a coupled functional unilateral inverse finite element cylindrical thin-wall structure displacement field monitoring method, which comprises the following steps of: performing discretization by adopting a four-node inverse shell unit, and arranging a sensor at a geometric center position of each unit; based on the basic mathematical model of the inverse finite element, establishing an overall pseudo-stiffness matrix and an overall pseudo-force matrix; and calculating to obtain a displacement field of the structure by solving a linear equation set formed by the overall pseudo-stiffness matrix, the overall pseudo-force matrix and the global displacement vector. According to the method, the structural displacement field is calculated by using the strain information measured at one side, and the global deformation is accurately and reversely deduced under the condition that only the local strain data is obtained by combining the advantages of finite element analysis and inverse problem solution. By optimizing local strain data solution, high cost and low efficiency of global strain measurement in a traditional method are avoided, and the method has remarkable practical advantages.

A solid structure deformation and damage analysis method based on a quasi-bond finite element method comprises: carrying out geometric modeling on a target structure body, and distribution and subdivision to generate a grid of a traditional finite element method; dividing the target structure body into a finite element region and a quasi-bond region, and calculating a finite element region system stiffness matrix and a quasi-bond region system stiffness matrix to obtain an overall stiffness matrix of the target structure body; setting a boundary condition for the target structure body, applying an external load, and calculating a system force matrix under current load and boundary state; and judging a quasi-bond breakage condition according to a node trial displacement, then calculating a node displacement of the structure body and an equivalent damage parameter at each node, and outputting cloud charts of a displacement field and an equivalent damage field.

The invention belongs to the technical field of material structure optimization, and discloses an enhanced structure design method and system for a three-period minimal curved surface, and the method comprises the steps: modeling a geometric model of the three-period minimal curved surface, and exporting the model; expanding a design domain of the geometric model to obtain a corresponding expanded design space of the three-period minimal curved surface; a design space file is converted into an analysis data structure, a general expression mode is constructed for the three-period minimum curved surface, an equal geometric element stiffness matrix is solved by using a Gaussian integral method and the elastic performance of each point in a design domain and assembled into a global stiffness matrix, and the structural response of the model is solved; constructing a local density distribution function in the Bayesian unit, and assembling all LDDFs to obtain a global density distribution function for presenting structural topology; and constructing a stiffness maximization topological optimization mathematical model for the TPMS, calculating an objective function and analyzing sensitivity, and carrying out iterative updating on design variables in the optimization model based on an optimal criterion method to obtain a reinforced structure configuration.

The invention discloses a jacket platform real-time correction and state prediction method based on a reduced-order model, and relates to the technical field of ocean engineering structure health monitoring and safety early warning. Comprising the steps of establishing a jacket platform finite element model; obtaining a sample matrix; constructing a reduced order model; reducing the order of the reduced-order model until the calculation precision of the reduced-order model meets a preset precision requirement, and obtaining a global reduced-order basis matrix and the reduced-order model meeting the preset precision requirement; reconstructing a stiffness matrix and a mass matrix by using the global reduced-order basis matrix; obtaining a corrected finite element model; inverting the marine environment load based on the on-site platform structure strain response data, and deducing the marine environment load; inputting the marine environment load into a pre-trained LSTM model, and outputting time series data including future load through the LSTM model; and predicting the future operation state trend of the jacket platform. The operation safety and reliability of the jacket platform 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.

The invention discloses a nonlinear beam unit analysis method considering prestressed tendon bonding slippage, which comprises the following steps: establishing an overall structure analysis model according to the geometric dimension of a prestressed concrete beam, dividing calculation units, and assembling unit stiffness matrixes into a structural overall stiffness matrix; applying load and boundary conditions to the overall structure analysis model, and solving the overall analysis model by adopting a Newton iteration method; and extracting calculation results of structural deformation, stress, section internal force, prestressed tendon relative slippage and the like, drawing a load-deformation curve, and calculating structural rigidity and bearing capacity according to the slope and peak value of the curve. The calculation unit provided by the invention is an 8-degree-of-freedom beam unit, not only is the stress characteristic of the beam structure considered, but also the bond slip characteristic of the prestressed tendon is considered, the problem that a traditional beam unit analysis model cannot consider the bond slip effect of the prestressed tendon is solved, and compared with an entity unit analysis model, the convergence is better, and the calculation efficiency is higher.

The invention discloses a prestressed beam simulation method, system and device based on a vector type finite element, and relates to the technical field of finite element simulation, and the method comprises the steps: building a prestressed beam numerical model through a vector type finite element method; dispersing the prestressed beam numerical model into a plurality of nodes and a plurality of prestressed beam units; calculating a stiffness matrix of each prestressed beam unit; determining a control equation of each node based on the stiffness matrix; and solving the control equation of each node by adopting a central difference method, and updating the state of each node according to the solving result to finish the simulation of the prestressed beam. The method can achieve the simulation of the complex behaviors of the prestressed beam, and is high in calculation efficiency and good in calculation stability.

The invention relates to the technical field of force sensors, and discloses a nonlinear decoupling method and system for a three-axis force sensor, and the method comprises the steps: collecting multi-dimensional original data, and carrying out the adaptive noise reduction, dynamic selection of a wavelet basis, and construction of a filtering matrix, and obtaining a noise reduction signal through calculation; constructing a dynamic PINN proxy model, fusing material parameters and environment parameters, outputting a stiffness matrix, establishing an online parameter updating mechanism, calculating the obtained optimization parameters, and constructing an enhanced dynamic compensation matrix; the output stiffness matrix and the enhanced dynamic compensation matrix are fused, and a main singular value is reserved; reconstructing to obtain a reconstructed strain field, establishing a multi-physics field coupling model to compensate quantum-thermal coupling errors, and compensating wire inductive coupling under high frequency; constructing a multi-scale graph structure, strengthening a loss function through physical constraint, performing crystal plasticity hysteresis compensation, fusing multi-scale features, and calculating and outputting a decoupling force value; and an intelligent calibration decision is established, the drift trend is predicted, and the calibration period is dynamically adjusted.

The invention discloses a structure optimization method and device for a collision working condition, a vehicle body part, a vehicle, equipment and a medium, and the method comprises the steps: determining a continuously changing displacement matrix and stiffness matrix of each finite element on an equivalent model in a collision simulation process, and carrying out the optimization of the collision condition according to the displacement matrix and stiffness matrix of the finite element at a target moment; and updating the load of the target load step applied to the finite element to update the collision simulation on the equivalent model, and based on the equivalent model after collision simulation, carrying out topological optimization on the model structure. According to the method, the stiffness matrix change of the finite element under the action of different load steps is analyzed, different stiffness states under the different load steps are determined, and on the basis of updating of the target load step, each load step can truly show the real-time state of collision, so that the nonlinear process in the collision process is restored with high precision, and the collision accuracy is improved. And the effect and accuracy of structure optimization are enhanced, and lightweight design is achieved while the collision performance of vehicle body parts is guaranteed.

The invention discloses a single-layer latticed shell structure damage identification method and system based on a direct analysis method, and the method comprises the steps: S1, constructing a numerical model according to the numerical value of a latticed shell structure, taking a key member of the latticed shell as a unit, analyzing and determining a distribution position, and preliminarily arranging a monitoring sensor measurement point for each unit; s2, updating a numerical model according to the monitoring data, performing construction analysis on each unit, determining deformation and residual construction stress, performing importance sorting on key components of the latticed shell again, and optimizing a measuring point arrangement scheme; s3, acquiring response data of each unit of the reticulated shell structure by using a monitoring sensor; s4, calculating unit unbalance force under the current reticulated shell configuration according to the response data, and calculating a structural stiffness matrix and a change matrix thereof; and S5, analyzing the change condition of the damage area by using the damage index and the structural stiffness matrix, and obtaining the damage condition of the reticulated shell key component. The structural member damage and the damage degree are accurately identified through the rigidity damage index.

The invention discloses a Fisher information matrix-based deformation monitoring sensor optimization arrangement method. The method comprises the steps of defining observation variables and unknown parameters; establishing a relation matrix; calculating a probability density function and a quality function of the observation variable; calculating a Fisher information matrix according to the deformation reconstruction equation; the traces of the Fisher information matrix are maximized to achieve optimal arrangement of the sensors. According to the method, the relation between the Fisher information matrix and the deformation reconstruction precision is deduced from the basic principle of an inverse finite element, optimal arrangement of the sensor is completed by maximizing the trace of the Fisher information matrix represented by the overall stiffness matrix, and efficient optimal arrangement of the sensor is achieved.

The invention provides a crustal stress prediction method and device, electronic equipment and a storage medium, and relates to the technical field of seismic survey. The method comprises the following steps: obtaining observation seismic data of seismic wavelets in a strong VTI medium, and constructing a to-be-inverted parameter matrix based on a PP wave reflection coefficient corresponding to the strong VTI medium; constructing a posterior probability function obeyed by an inversion parameter matrix corresponding to the to-be-inverted parameter matrix based on a Bayesian inversion theory and observation seismic data, and determining a target functional based on a prior probability function and a likelihood function corresponding to the posterior probability function; determining medium density and each stiffness matrix coefficient based on an inversion parameter matrix solving result of the target functional, and determining a flexibility matrix of the strong VTI medium based on each stiffness matrix coefficient; and predicting the ground stress distribution of the target profile in the strong VTI medium based on the medium density and the positive strain matrix and the flexibility matrix corresponding to the strong VTI. Therefore, the crustal stress prediction accuracy under the strong VTI medium is improved.

The invention provides a real-time reconstruction method and system for an internal temperature field of a key section of a deformed ring forging. The real-time reconstruction method comprises the steps of detecting an edge contour, generating a temperature field reconstruction target section and measuring section edge temperature distribution; performing grid division on the target section according to a heat conduction forward problem to generate a stiffness matrix and a solution equation; a matrix C, a boundary temperature array Ts and a boundary thermal load array Fa are generated, an underdetermined equation set of the internal temperature field reconstruction problem is obtained, and a special solution is obtained; and bringing the special solution of the underdetermined equation set and the boundary thermal load array Fa back to the solving equation to obtain section temperature distribution. According to the method, a heat conduction inverse problem is converted into a heat conduction forward problem, the internal temperature field of the key section can be reconstructed in real time based on the measurable surface temperature for the forge piece in the deformation process, a large number of training samples are not needed, an underdetermined equation set coefficient matrix does not need to be processed, and all information is reserved; the method is simple, efficient, high in controllability, high in fidelity and wider in application range.

The invention discloses a method and a system for acquiring deformation stress increment of extremely-thin lamellar weak surrounding rock. Relates to the technical field of tunnel engineering. Obtaining a strain increment matrix of the extremely-thin laminar weak surrounding rock under the overall coordinate system, calculating a stress tensor matrix of deformation of the extremely-thin laminar weak surrounding rock under the overall coordinate system by combining the rotation matrix and the element stiffness matrix under an elastic assumption theory, and calculating a deformation stress increment according to the stress tensor matrix after characteristic correction; according to the scheme, the method is improved on the basis of a traditional extremely-thin lamellar surrounding rock mechanical property simulation technology, and a stress increment matrix of deformation of the extremely-thin lamellar weak surrounding rock under an overall coordinate system is calculated on the basis of a strain increment matrix, a rotation matrix and an element stiffness matrix under an elastic assumption theory; after the characteristics of the stress increment matrix are corrected, the deformation stress increment is calculated, the stress deformation characteristics of the extremely-thin lamellar weak surrounding rock are accurately and reasonably described, and the engineering analysis efficiency of the extremely-thin lamellar weak surrounding rock is improved.

The invention provides a shape function KL expansion random field discretization method considering a slope soil body irregular domain, and belongs to the technical field of soil body parameter random field analysis. Comprising the following steps: establishing a slope numerical model and dividing a discrete grid and a form function grid; establishing a probability distribution model of a soil parameter random field; calculating an element stiffness submatrix ke and an element mass matrix Ee; calculating an autocorrelation matrix rho and a global stiffness matrix B; assembling a global matrix; calculating a characteristic value lambda k, a characteristic function phi k and an expansion term number M; and random field discretization of the slope numerical model is realized. The random field discrete technology provided by the invention is used for efficiently and accurately analyzing the random distribution characteristics of the soil parameters in the rock-soil structure. The discrete random field model can be efficiently constructed in a complex geological structure, the calculation cost is remarkably reduced, the method can be conveniently combined with numerical simulation tools such as a finite element method, and the requirement of actual engineering for efficiency is met while the precision is improved.

The invention provides a thick-wall austenitic stainless steel full-focus correction imaging method based on a TRL probe, and the method comprises the steps: collecting a thick-wall austenitic stainless steel welding seam based on a TRL area array probe, obtaining full-matrix data, measuring an elastic stiffness matrix of a welding seam material, and solving a wave equation of ultrasonic waves based on the elastic stiffness matrix, the method comprises the following steps: acquiring a phase velocity and a group velocity, acquiring the magnitude and the direction of the phase velocity and the group velocity, calculating a propagation path of ultrasonic waves in an anisotropic medium based on a ray tracing technology and in combination with a Fermat principle, extracting an amplitude value of a corresponding moment from full matrix data according to propagation time, performing delay superposition on each imaging point in an imaging area, and completing correction of full-focus imaging. According to the thick-wall austenitic stainless steel welding seam correction imaging method based on the TRL area array probe, the detection precision of an imaging result and the detection signal-to-noise ratio are improved, and a more reliable and accurate technical scheme is provided for thick-wall austenitic stainless steel welding seam detection.

The invention relates to the field of geophysics, and discloses a conductivity block continuous change magnetotelluric three-dimensional inversion method, and the method comprises the steps: adding a model constraint term based on the conductivity of a non-structural tetrahedral mesh node into a target function, and obtaining an inversion target function expression; enabling the norm of the conductivity variation of the unit and the surrounding unit to be minimum, and solving a model constraint term to obtain the model constraint term based on the tetrahedral mesh node conductivity; carrying out gradient solution on the inversion objective function expression; specific expressions of elements in stiffness matrix gradients under four different conditions are solved, gradient components of stiffness matrixes corresponding to all nodes are solved, and then the objective function gradient is solved in magnetotelluric inversion; according to the method, updating of the node conductivity model during three-dimensional underground medium inversion is achieved, unknown numbers generated in the inversion process are greatly reduced, the uniqueness of a solution is improved, meanwhile, the time needed by the inversion process is greatly shortened, and a better inversion effect is obtained.

The invention discloses an oral cavity finite element simulation calculation method based on collision detection. The method comprises the following steps: 1) acquiring oral scanning data of teeth, generating surface grid data of the invisible appliance according to the oral scanning data, and then generating a tetrahedral grid from the surface grid of the invisible appliance for subsequent simulation calculation; 2) on the basis of collision detection, calculating a pushing vector of teeth to the invisible appliance, and taking the pushing vector as a displacement boundary condition of subsequent calculation; 3) constructing a three-dimensional finite element model taking voxels as units, calculating an element stiffness matrix, assembling an overall stiffness matrix, loading displacement boundary conditions, forming a finite element solution equation set, and obtaining a model node displacement result; and 4) calculating stress and strain of the finite element model according to a displacement result. Aiming at the orthodontic mechanical analysis of the invisible orthodontic appliance, the invention provides a modeling method for the interaction between the teeth and the invisible orthodontic appliance based on a collision detection technology, completes the subsequent simulation solution of a finite element model, and gives consideration to the solution precision and speed.

The invention discloses a bridge anti-fatigue performance analysis and optimization method and system based on finite element analysis. The method comprises the following steps: constructing a bridge finite element basic model and a random fleet time-varying load spectrum, and executing damage-rigidity dynamic coupling iterative calculation; a fatigue sensitive area is automatically identified through the strain energy density gradient, a high-precision sub-model is generated, a basic model stiffness matrix is updated in real time through a sub-model calculation result, and a fatigue evolution cloud picture containing a stiffness degradation track is generated; and finally, identifying a damage hotspot based on the cloud picture, and generating an optimization strategy containing a traffic current limiting threshold and a structure reinforcement opportunity by using a genetic algorithm. According to the method, the problem of prediction deviation of traditional static analysis is solved by simulating the reverse influence of damage on rigidity and internal force redistribution, and accurate evaluation and scientific decision-making of the whole life cycle of the bridge are achieved.

The invention discloses a robot scene simulation method applied to a digital robot industry chain, and relates to the technical field of scene simulation. Three-dimensional shape data of a clamp in an actual operation environment is obtained, data fusion correction is carried out in combination with geometric structure information, and a simulation result is obtained; the form deviation caused by manufacturing tolerance, abrasion and mechanical errors is effectively compensated; in the robot grabbing modeling process, part posture changes and grabbing errors in the clamping process are calculated through simulation analysis, the deviation condition of parts in the clamping process is analyzed through a rotation transformation matrix and a rigidity matrix, grabbing points are dynamically adjusted, and assembly deviation caused by the clamping errors is effectively reduced; in the assembly path optimization stage, the path of the robot executing the assembly task is adjusted through the correction parameters in the clamp database, the insertion force and the assembly track of the robot are optimized by integrating the grabbing error and the clamp form deviation, the robot can adjust the assembly strategy in a self-adaptive mode, and the assembly precision is improved.

The invention discloses a vibration solving method for an elliptical perforated plate in any shape based on an elliptical energy element, and belongs to the technical field of thin-wall structures, and the method comprises the steps: carrying out the overall mapping of the elliptical perforated plate in any shape, building an elliptical equation in a dimensionless coordinate system, and carrying out the Boolean operation modeling of geometry; constructing a boundary condition function simulation constraint; simulating board deformation according to a first-order shear deformation theory, and constructing a global trial function in combination with a Legendre polynomial and a boundary condition function; establishing a total energy functional based on a minimum potential energy principle and a Ritz method, solving a total energy functional stationary value, and obtaining element calculation formulas in the stiffness matrix and the mass matrix; and constructing an elliptical energy element based on local mapping, and performing numerical integration on each element of the matrix based on the elliptical energy element and completing vibration problem solving. According to the method, an energy functional and a numerical integration formula are both constructed on a square solution domain in an overall mapping stage, so that the calculation solution process of the elliptical perforated plate in any shape is completely standard.

The invention relates to a BIM (Building Information Modeling)-based building structure modeling method, device, equipment and medium, and the method comprises the steps: aiming at a curvature-mechanics semantic fault problem existing in a process of converting a special-shaped curved surface structure from a BIM geometric model to finite element analysis, generating a curvature tensor field with parameter coordinates by analyzing an NURBS curved surface; integrating along the maximum curvature change direction to construct an internal force flow function, and forming a guide field for controlling the grid density; shell / beam units are divided based on the guide field gradient, space coordinates are marked, and a curvature-stiffness matrix and an internal force flow driving variable cross-section are combined for calculation to generate a structural analysis model; the gradient distribution of the guide field is dynamically corrected through displacement error feedback, and lossless transmission of geometric topology and mechanical properties is achieved. According to the method, the distortion risk of special-shaped structure analysis is remarkably reduced, artificial experience correction is avoided, and high-precision automatic modeling support is provided for a complex building structure.

This invention discloses a method and system for numerically defining a rebar-concrete interface element under both monotonic and cyclic loading, comprising: first, establishing a finite element model of reinforced concrete component, and generating a solver input file within the finite element software; then, adding a series of user-defined interface elements in the solver input file; then, inputting parameters of characteristic points of the axial bond-slip curve between rebar and concrete, the diameter of rebar, and the characteristic parameters of fibers associated with the user-defined element (UEL) in the solver input file; then, setting the element stiffness matrix, the coordinate transformation matrix and residual in the UEL subroutine; finally, submitting the solver input file in the finite element software, and calling the UEL subroutine for calculation.

The invention relates to the technical field of intersection of computational mechanics and deep learning, in particular to a physical information neural network and finite element combined deep learning method. The partial differential equation is embedded into a neural network, and the method comprises the following steps: S1, carrying out grid division on a solution domain, and calculating a stiffness matrix and a node load array; s2, collecting observation data and boundary conditions; s3, taking the fitting error of the energy represented by the stiffness matrix and the observation data as a loss function to train a neural network; s4, executing an optimization algorithm to optimize the neural network; s5, repeating the training and optimization steps until convergence; according to the deep finite element method provided by the invention, a physical information neural network is combined with a finite element method; the method shows remarkable advantages and beneficial effects in the aspects of physical knowledge and observation data fusion, modeling process and loss function design, model training stability and simplicity, calculation efficiency and precision, pre-processing and post-processing technology, transfer learning ability, applicability and the like.

The invention discloses a GIL electromagnetic property detection method and system based on field circuit-mechanical coupling, and the method comprises the steps: building an electromagnetic field-circuit-mechanical coupling mathematical model based on a circuit mathematical model through employing a stiffness matrix of a magnetic vector position, a stiffness matrix of a conductor eddy current, and a load matrix of a conductor conduction current; engineering sections are divided, a finite element two-dimensional simulation model is built, an equivalent external circuit is built, coupling connection is conducted through an external circuit port voltage matrix, and an electromagnetic property detection model of the GIL is obtained; performing iterative solution to obtain transient electromagnetic response of each engineering section, wherein the transient electromagnetic response comprises magnetic flux density, current density and node transient electromagnetic force; detecting whether electromagnetic response abnormity exists in the actual operation process of the GIL or not; and simulating the electromagnetic stress of the GIL component in the abnormal engineering section in the three-dimensional structure field by adopting harmonic response analysis, and checking the electromagnetic and mechanical characteristics of the GIL. According to the method, the electromagnetic characteristics of the GIL can be simulated more accurately, the applicability is wide, the operation speed is high, and guarantee is provided for GIL short-circuit fault monitoring and safe and stable operation.

The invention relates to a train-rail coupling vibration analysis method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: dividing a plurality of track units, constructing a motion differential equation of a steel rail in the track units in a frequency domain, and solving to obtain a relation between longitudinal displacement and frequency on nodes at two ends of the steel rail and a relation between stress and frequency on the nodes at the two ends of the steel rail; constructing a steel rail dynamic stiffness matrix for the steel rails in the rail units, and constructing a rail dynamic stiffness matrix for the rail units according to the steel rail dynamic stiffness matrix; assembling to obtain an integral dynamic stiffness matrix of the train-rail system, and constructing a transfer matrix of a rail; and determining a track irregularity excitation power spectrum, and solving to obtain a dynamic response index of the train-track system on each degree of freedom under the track irregularity excitation power spectrum according to the transfer matrix and the overall dynamic stiffness matrix. By adopting the method, the calculation efficiency of the dynamic response index can be improved.

The present invention relates to the technical field of the intersection of computational mechanics and deep learning, and specifically to a deep learning method that combines a physical information neural network with a finite element method; specifically, partial differential equations are embedded into a neural network, comprising the following steps: S1, meshing the solution domain, and calculating the stiffness matrix and the node load array; S2, collecting observation data and boundary conditions; S3, using the energy represented by the stiffness matrix and the fitting error of the observation data as a loss function to train the neural network; S4, executing an optimization algorithm to optimize the neural network; S5, and repeating the training and optimization steps until convergence; the deep finite element method proposed in the present invention combines a physical information neural network with a finite element method, and exhibits significant advantages and beneficial effects in terms of the fusion of physical knowledge and observation data, modeling process and loss function design, model training stability and simplicity, computational efficiency and accuracy, pre- and post-processing technology and transfer learning capability, and applicability.

The invention discloses a cluster-CPU-GPU heterogeneous parallel method for solving response of a large-scale anisotropic material composition structure, and belongs to the technical field of high-performance calculation simulation. The method comprises the following steps: firstly, establishing a macroscopic structure and a microstructure, dividing a finite element mesh for a macroscopic model, dividing a calculation region by using a region decomposition algorithm, and realizing macroscopic model unit stiffness matrix and displacement calculation through OpenMP acceleration; secondly, reconstructing the model at the Gaussian point of the macroscopic model, partitioning by means of the MPI technology, calculating the displacement of the microscopic model by using multiple threads in each process, and accelerating the calculation of a stiffness matrix of a microscopic model unit by using OpenMP to realize cluster MPI parallel calculation; during post-processing, visualization is realized by utilizing GPU multi-core advantage calculation macroscopic model stress and strain, and cross-scale heterogeneous parallel simulation is achieved. The method is combined with related algorithms, can efficiently utilize resources to solve the analog simulation problem, and is wide in application prospect.

The invention belongs to the technical field of robot cooperative control, and particularly relates to a distributed self-adaptive control method and system for cooperation of multiple mechanical arms, and the method comprises the steps: decoupling a pure internal force vector from the data of a six-dimensional force sensor at the tail end through pre-identified kinetic parameters; calculating a cooperative confrontation index according to the pure internal force vector and the relative position vector, calculating a self-adaptive internal force penalty weight through nonlinear mapping based on the cooperative confrontation index, establishing a state-space equation of the system based on a double-integral model, introducing an equivalent environment stiffness matrix, mapping a prediction error vector in a prediction time domain into a prediction internal force vector, and obtaining a prediction internal force vector; and a cost function containing the product of the self-adaptive internal force penalty weight and the predicted internal force vector is constructed, an optimal control increment sequence is obtained by minimizing the cost function, and a speed instruction is updated to drive the mechanical arm to move. According to the method, the robustness and the execution stability of the system are effectively improved, and the cooperation safety is guaranteed.