31 results about "Eulerian lagrangian" patented technology

The invention discloses a cooperative online and offline control method for an unmanned aerial vehicle power transmission line broken strand repairing robot, and aims to realize efficient and safe maintenance of a power transmission line. Based on Newton-Euler and Euler-Lagrange equations, establishing an eight-degree-of-freedom nonlinear dynamic model of the unmanned aerial vehicle-broken strand repair robot; and designing and optimizing an extended state observer and a nonlinear error feedback control law to suppress swinging. An airborne visual servo system is adopted to capture the posture of the robot so as to correct the trajectory of the unmanned aerial vehicle, and a double-closed-loop control strategy based on a spherical swing mechanism is applied to realize stable tracking of an arc trajectory. Nonlinear load trajectory tracking control based on a six-dimensional force sensor is introduced, suspension point tension and the position of the unmanned aerial vehicle are fused to estimate the pose of the robot, self-adaption to different load weights is achieved through a load speed observer, and real-time data exchange between the unmanned aerial vehicle and the robot is supported through a low-delay and high-reliability communication protocol.

This invention discloses a spatial domain impedance learning control method for a five-bar parallel robot, comprising the following steps: Step 1, establishing an Eulerian-Lagrange model of the five-bar parallel robot; Step 2, establishing a human-machine system interaction force model based on the trajectory tracking error of the parallel robot; Step 3, estimating impedance dynamics based on the Eulerian-Lagrange model; Step 4, designing an iterative learning strategy for human-machine interaction force based on the human-machine system interaction force model and impedance dynamics; Step 5, conducting control tests on the five-bar parallel robot using the iterative learning strategy for human-machine interaction force. The learning control method proposed in this invention ensures the convergence of tracking errors, considers the uncertainties in robot modeling, and further guarantees the safety of human-machine interaction.

The application discloses an analysis method for key local vibration of fluid-solid coupling bridges based on a substructure method, and relates to the technical field of bridge vibration analysis, and comprises the following steps: dividing a response key area in a bridge structure into substructures, performing static condensation, and obtaining a condensed substructure model; establishing a whole bridge-wave coupling model, and adopting a dynamic explicit solving strategy based on a coupling Euler-Lagrange method; independently solving the substructures by using a dynamic implicit method; coupling the substructures to the whole model by an interface coordination condition, and jointly calculating the whole explicit model and the local implicit model. The analysis method adopts an explicit method to process a strong nonlinear fluid-solid coupling process at a whole level, and establishes a fine substructure model through Guyan static condensation at a local area, so that the analysis precision is effectively improved.

The application relates to the technical field of underground engineering and geotechnical mechanics, and provides an underground engineering blasting simulation method based on secondary development of ABAQUS, which comprises the following steps: step 1, a geometric model is established, and a calculation region is divided into an Euler domain and a Lagrange domain; step 2, a JH-2 constitutive model is called and realized through a UMAT user subroutine interface of ABAQUS; step 3, a JWL state equation is used for modeling in the Euler domain; step 4, a coupled Euler-Lagrange method is applied; step 5, boundary conditions and blasting loading parameters of the geometric model are set; and step 6, transient dynamics calculation is carried out, and results are output for analyzing blasting effects. The scheme not only can provide a high-precision numerical test platform for underground engineering blasting design and safety control, reduces engineering risks, optimizes blasting parameter configuration, but also has remarkable engineering application value and popularization significance.

The invention relates to the technical field of industrial dust removal equipment design and fluid simulation, and discloses a macroscopic numerical simulation and verification method for dust deposition uniformity of a metal filter bag deduster, which comprises the following steps: constructing a deduster full-size model loaded with a porous jumping medium boundary; establishing an Euler-Lagrange gas-solid coupling model to calculate a flow field and a particle trajectory; designing a diversion structure orthogonal experiment based on the flow field characteristics, and obtaining operation resistance and deposition mass data under different parameter combinations; establishing an evaluation system comprising a flow distribution coefficient and a deposition uniformity index, and screening an optimal structure; a physical experiment platform is constructed for powder coating and segmented scraping weighing, and the reliability of the method is verified by comparing relative errors of experiments and simulation data. According to the method, through macroscopic equivalent modeling and quantitative evaluation, the deposition behavior in the metal filter bag dust collector is accurately predicted, the problem of uneven load of the filter bag is effectively solved, and a reliable basis is provided for structural optimization.

The application belongs to the field of robot control, and particularly relates to a mobile robot dynamic event triggering control method based on reinforcement learning. In view of the unknown wheel slip effect of the robot and the communication resource constraint problem, a kinematics model and an Euler-Lagrange dynamics model of a four-wheel Mecanum wheel mobile robot are established, the motion deviation caused by four-wheel slip is characterized as non-matching disturbance, the trajectory tracking problem is converted into an extended dimension system adjustment problem constrained by the worst case disturbance by combining the tracking error and the expected trajectory and introducing the steady state control, a dynamic event triggering mechanism is introduced to reduce the calculation burden and transmission load, a Critic reinforcement learning algorithm is designed to solve the Hamilton-Jacobi-Isaacs equation to obtain an approximate optimal controller under the dynamic event triggering condition, and the consistent ultimate bounded stability of the closed loop system is analyzed based on the Lyapunov theory, so that the high-precision trajectory tracking performance of the robot is ensured on the premise of saving communication and calculation resources.

The invention discloses an analysis method for key local vibration of a fluid-structure interaction bridge based on a substructure method, and relates to the technical field of bridge vibration analysis, and the method comprises the following steps: dividing a response key region in a bridge structure into substructures, and carrying out static polycondensation to obtain a polycondensed substructure model; an integral bridge-wave coupling model is established, and a dynamic explicit solving strategy based on a coupling Euler-Lagrange method is adopted; independently solving the substructures by adopting a dynamic implicit method; and coupling the substructures into the overall model through an interface coordination condition, and carrying out joint calculation of the overall explicit model and the local implicit model. According to the analysis method, an explicit method is adopted to process the strong nonlinear fluid-solid coupling process on the overall level, a fine substructure model is established in a local area through Guyan static polycondensation, and the analysis precision is effectively improved.

The invention belongs to the technical field of robot control, and particularly relates to an anti-disturbance multi-finger cooperative grabbing control method with global preset performance, which comprises the following steps: establishing a kinetic model of a dexterous hand system according to an Euler-Lagrange equation; describing a communication topological structure of the dexterous hand system through a virtual coupling mechanism, and constructing a collaborative grabbing error of the dexterous hand system; a preset time function and a global preset performance function are introduced to convert the collaborative grabbing error, and the overall potential energy of the dexterous hand system is calculated; with the purpose of minimizing the overall potential energy of the dexterous hand system, joint control input torque is designed in combination with a distributed control architecture; designing a self-adaptive updating law for unknown disturbance according to an internal model principle, and obtaining a distributed global preset performance control law in combination with a joint control input torque; according to the method, the robustness of the system under disturbance of joints and external force is remarkably improved.

The invention discloses an image segmentation method of a parametric type active contour model based on Finsler-Laplacian. According to the method, an anisotropic Finsler norm is introduced under a static external force field, and a new energy minimization problem is provided and comprises a data fidelity term and a regularization term based on the Finsler norm. A corresponding Euler-Lagrange equation is derived through a variational method and is converted into a time-evolved parabolic partial differential equation for numerical solution. A comparison principle is established under the framework of the viscous solution, and the existence of the understanding, the continuity about the space Lipschitz and the continuity about the time consistency are proved. The model has the advantages of being large in capture range, insensitive to initial contours, high in noise immunity, capable of effectively retaining image edge details and the like, and shows excellent performance in medical image and natural image segmentation.

The application relates to the technical field of automatic control, in particular to an anti-interference control method and system for an Euler-Lagrange system containing unknown interference, the method comprising the following steps: establishing an Euler-Lagrange dynamics model containing unknown interference and model uncertainty, unifying the unknown interference and the model uncertainty into a bounded and differentiable lumped disturbance term, constructing a preset time disturbance observer to perform specified time estimation on the lumped disturbance term; defining a system tracking error, designing a non-singular preset time synchronous sliding mode surface fused with a time-varying scaling function; and based on the output value of the preset time disturbance observer and the non-singular preset time synchronous sliding mode surface, designing a preset time synchronous controller with disturbance feedforward compensation. Under the premise of considering unknown interference and system uncertain dynamics, the system interference and uncertainty can be effectively reconstructed, the anti-interference capability of the system is improved, and vibration suppression and attitude stabilization are realized.

This invention belongs to the field of robotic arm control and provides a robust control method and system for a lifting robotic arm. It includes determining five nonlinear motion equations of the robotic arm dynamics using the Euler-Lagrange equations and representing them in matrix form. The robotic arm dynamics matrix form is divided into driven and non-driven forms, and after reduction processing, a reduced-mode dynamic model of the robotic arm is obtained. A robust controller is constructed by building a second-order sliding mode controller and a fast terminal sliding mode controller. The robotic arm dynamics matrix form is transformed into a state-space form to obtain a state observer. The state observer is subtracted from the state-space form to obtain the estimation error dynamic equation. The dynamic state of the robotic arm is inferred using the state observer. The control mode of the robust controller is determined based on the estimation error dynamic equation corresponding to the dynamic state, and the corresponding control signal is generated based on the robust controller. This invention significantly improves the safety and efficiency of robotic arms in practical applications.

The application belongs to the technical field of motor system control, and discloses a motor feasible task design method based on motor performance parameters and controller structure, which comprises the following steps: first, determining two types of motor performance parameters of the maximum torque and the peak power of the selected motor; second, based on Euler-Lagrange equation, establishing motor dynamics differential equation containing unknown rotational inertia, unknown friction resistance, and describing the characteristics of motor generalized speed; third, designing an adaptive controller to realize speed tracking control and online estimation of unknown rotational inertia, unknown friction resistance and unknown external disturbance; fourth, based on motor performance parameters and adaptive controller structure, establishing task feasibility constraint inequality, and determining motor feasible task in combination with the task load condition. The application can realize the analysis and determination of motor feasible task based on the control strategy and the maximum speed and the maximum torque of the motor without experimental test.

This invention discloses a data-driven method for reconstructing the flow field in a water-fish coupled scenario, relating to the intersection of fluid mechanics and artificial intelligence. The method includes the following steps: using a CFD model to simulate the water-fish coupled flow field under various scenarios, constructing a water-fish coupled flow field dataset; based on the water-fish coupled flow field dataset, constructing dynamic meshes for the fish body and background using a convolutional neural network prediction method via a Lagrangian module; constructing convolutional neural network models for both the fish body and background dynamic meshes; and constructing a rapid flow field reconstruction model based on the convolutional neural network model and the Eulerian-Lagrangian method to accurately predict the water-fish coupled flow field. This invention solves the problems of high cost, low efficiency, low accuracy, and unsuitability for widespread application of traditional flow field numerical simulation methods.

This application provides a numerical simulation method, system, and medium for erosion wear of titanium alloy pumps and valves. The method includes: acquiring the internal flow channel geometric parameters of the titanium alloy pump and valve, and constructing a three-dimensional geometric model of the internal flow channel; performing mesh generation to generate a computational mesh model; setting titanium alloy material property parameters and flow field boundary conditions; analyzing the fluid medium and solid particles inside the pump and valve using the Euler-Lagrange coupling method based on the computational mesh model, titanium alloy material property parameters, and flow field boundary conditions; analyzing the solid particle parameters based on the motion trajectory information of the solid particles in the flow field, calculating the erosion rate data at different locations on the pump and valve wall, and generating an erosion wear distribution cloud map of the pump and valve inner wall; by using the Euler-Lagrange coupling method, the motion characteristics of individual particles and their impact behavior on the wall can be accurately captured, the erosion rate data can be accurately analyzed, and thus the erosion wear of the pump and valve can be accurately analyzed.

This invention discloses a NOx control method for a chain grate rotary kiln based on multi-source fuel utilization. This method, targeting different levels of environmental warnings and ultra-low emission requirements, first establishes a solid fuel combustion characteristic database including anthracite, bituminous coal, and their blends based on industrial analysis, elemental analysis, and thermogravimetric experiments. Simultaneously, it employs CFD numerical simulation technology, coupling an Eulerian-Lagrange multiphase flow model and a P-1 radiation model, to analyze the flow field and NOx formation patterns under pulverized coal and natural gas co-firing conditions. Upon receiving control commands, it calculates the inlet concentration threshold based on the real-time efficiency of the denitrification system, intelligently recommends the optimal fuel ratio using the multi-source database, and outputs corresponding kiln tail temperature, combustion air volume, and raw material structure control parameters. This invention, through a dual approach of "experimentation + simulation," achieves the suppression of fuel-type and thermal NOx formation at the source, meeting differentiated environmental indicators while ensuring stable production.

The application discloses an unmanned ship scheduled time trajectory tracking method based on asymmetric preset performance, which comprises the following steps: S1, a kinematics model and a dynamics model of a three-degree-of-freedom unmanned ship are established based on an Euler-Lagrange method, and the kinematics model and the dynamics model are expressed as a state equation of the unmanned ship system; S2, a system structure with scheduled time convergence characteristics is constructed, and stable convergence of the state of the unmanned ship system within scheduled time is realized; and S3, an asymmetric preset performance function is designed, trajectory tracking errors of the unmanned ship system are constrained, a dynamic evolution process of the errors is limited, and the errors are guaranteed to meet preset performance indexes. The application has the beneficial effects that a novel asymmetric preset performance function is constructed, the dynamic evolution process of system errors can be effectively constrained, the state errors of the system are always kept within preset performance boundaries, and therefore, the stability and control precision of the trajectory tracking process of the unmanned ship are improved.

The application relates to a fixed-time fault-tolerant control method for a space tethered combination under actuator failure, a dynamic motion equation of a tethered robot capturing a target and forming a tethered combination with the target is established according to an Euler-Lagrange method; a learning model of internal uncertainty and actuator failure of the tethered combination system is constructed, an observer with a fixed-time convergence characteristic is designed to accurately estimate the lumped uncertainty composed of the internal uncertainty and the failure; in order to ensure that the system state has a fast convergence characteristic under the lumped uncertainty, a fixed-time non-singular terminal sliding mode surface is designed; finally, in combination with the characteristics of the sliding mode control and the observer, a fixed-time fault-tolerant control scheme based on the observer is proposed, the system state can realize fixed-time stable convergence under the internal uncertainty, and high-performance tracking effect can still be maintained when the actuator has failure.

The invention discloses a rotary kiln temperature field real-time numerical simulation method based on deep learning, and the method comprises the following steps: S1), building a numerical simulation model of a rotary kiln based on an Euler-Lagrange discrete phase model, a coupled radiation model and a chemical reaction model through a computational fluid dynamics method; s2) determining boundary conditions and working condition parameters of the rotary kiln; s3) processing the temperature field sample data set obtained in the step 1) by using an intrinsic orthogonal decomposition method; s4) constructing a deep neural network of the gating circulation unit; s5) in a deployment application stage, obtaining a corresponding intrinsic orthogonal decomposition modal coefficient; s6, the internal temperature field of the rotary kiln is reconstructed, and real-time prediction.The prediction efficiency is greatly improved, computing resource consumption is remarkably reduced, the bottleneck of an existing numerical simulation technology in the aspect of real-time performance is broken through, and rapid and continuous thermal state evaluation and operation optimization decision support can be provided for an industrial site.

This invention discloses a multi-hub motor Euler-Lagrange torque dynamic vector distribution system and its control method. The system treats the four hub motors of a distributed drive electric vehicle as a multi-hub motor electromechanical system. The torque distribution system of the multi-hub motor electromechanical system includes an integrated Euler-Lagrange torque controller and a driving assistance system. The method includes: pre-processing modeling of the four hub motors; pre-setting the Euler-Lagrange torque distribution control law; sampling the state variables of the multi-hub motor electromechanical system; determining the desired equilibrium point of the multi-hub motor electromechanical system; calculating the injection of dynamic interconnection coefficients and injection of friction damping coefficients; outputting the optimal given torque by the Euler-Lagrange torque distribution controller; and coordinating the modulation controller and the multi-half-bridge arm group driver to drive each hub motor. This invention can effectively adapt to various complex road conditions and improve the handling stability of distributed drive electric vehicles.

The invention relates to the technical field of automatic control, and discloses a self-adaptive predefined time tracking control method for an Euler-Lagrange system, which comprises the following steps: establishing a kinetic model of the Euler-Lagrange system of a double-joint mechanical arm, and constructing a reinforcement learning framework comprising an evaluation neural network and an execution neural network; calculating a position error according to a joint position in the kinetic model and a given reference signal; generating a predefined time virtual control law containing predefined convergence time parameters, and calculating a speed error in combination with the joint speed; generating a practical predefined time control law containing predefined convergence time parameters based on the speed error, the position error and the control compensation amount output by the execution neural network; and constructing a weight updating law, applying a practical predefined time control law to the dynamical model, and updating the weight according to the weight updating law. By introducing explicit predefined convergence time parameters, an error convergence time upper limit can be set according to task requirements.

The invention provides a tracking control method for a flexible mechanical arm under preset time. The tracking control method comprises the steps that a coupling effect including connecting rod dynamics and motor actuator dynamics is considered, and a high-order nonlinear dynamic model of a single-connecting-rod flexible joint mechanical arm is established based on an Euler-Lagrange equation; a tracking error and an adaptive filter of the flexible mechanical arm are defined, so that error convergence within finite time is achieved; a preset time function and an obstacle Lyapunov function are constructed, an error conversion function is established, and tracking error constraint is converted into unconstraint; and an actual controller of the flexible mechanical arm is constructed and updated according to the self-adaptive filter, the error conversion function, a preset time function and a barrier Lyapunov function. According to the method and the device, the flexible mechanical arm can be ensured to realize high-precision tracking of the target trajectory within the preset time randomly specified by a user, and all signals in a closed-loop system are kept bounded.

The present application relates to the field of refractory material preparation and industrial automation control, and particularly to a refractory material production optimization system based on big data analysis, comprising a manifold space initialization step: constructing a thermodynamic potential energy manifold data model using raw material characteristic data; a state mapping and positioning step: constructing an observation equation and mapping real-time sintering state data to current state particle coordinates using a damped least squares method; a geodesic solution and control step: solving a second-order Euler-Lagrange differential equation set based on a target energy state position to obtain an optimal geodesic path; and a deviation correction and execution step: generating adaptive control instructions according to the vector deviation of the particle evolution trend and the optimal path to adjust equipment operating parameters; the present application effectively solves the problem of unstable sintering quality caused by raw material fluctuations in traditional production.

This invention discloses a method for visualizing structural mode shapes based on an Euler-Lagrange hybrid frame, comprising: 1. acquiring motion video data of the structure to be measured, and extracting vibration signals of discrete regions of the structure using a sub-pixel precision image matching algorithm; 2. decoupling the vibration signals using a blind source separation algorithm to obtain modal response signals of each order of the structure; 3. calculating the spatial weights of each mode of the structure based on the acquired modal response signals; 4. calculating the image grayscale values ​​associated with a single mode of the structure based on the modal response signals and spatial weights; 5. extracting the dense motion field between frames associated with a single mode of the structure using an optimized Demons algorithm; 6. visualizing the structural mode shapes using a motion compensation algorithm based on the acquired motion field. This invention can better achieve global spatial motion decoupling and improve the accuracy of the motion field, thereby reducing the complexity of the algorithm and improving the quality of visualized structural mode shapes.

The application discloses a method for determining the thickness of a material layer of a grate-type garbage incinerator and related devices, and relates to the technical field of incineration treatment. The method comprises the following steps: through a garbage layer pressure difference prediction model, two accurate parameters, i.e., a measured current primary air preheater outlet pressure and a current combustion 1 section damper opening degree, are converted into a predicted combustion 1 section garbage layer pressure difference. Based on the current combustion 1 section garbage layer pressure difference, the predicted combustion 1 section garbage layer pressure difference and a preset ratio, a combustion 1 section material layer thickness of the grate-type garbage incinerator is calculated. Based on a grate real-time running speed and a total length of the grate of the grate-type garbage incinerator and the predicted combustion 1 section material layer thickness of the grate-type garbage incinerator, a material transmission algorithm based on an Euler-Lagrange hybrid method is used to dynamically simulate and calculate the material layer thickness distribution of the whole grate, so that the material layer thickness is accurately determined, and the problem that the material layer thickness is difficult to accurately measure in a high-temperature and corrosive environment is avoided.

The invention discloses a double-pendulum crown block adaptive control method based on online reinforcement learning, and relates to the technical field of crown block motion control, and the method comprises the following steps: S1, employing an Euler-Lagrange method to establish a dynamic model of a crown block according to the specific parameters of the crown block; s2, converting the kinetic model into a linearized system model; s3, designing a self-adaptive control system framework based on RL; s4, solving the weight of the AN through the iteration of the RL; s5, the weight of the AN is output in real time through the updated AN to serve as the optimal gain parameter generated at the next moment, online reinforcement learning is directly applied to the physical double-pendulum crown block system, an accurate model and offline training are not needed, the complex coupling effect between a lifting hook and goods is effectively overcome, and the reliability of the physical double-pendulum crown block system is improved. And meanwhile, the problem of performance degradation of a traditional reinforcement learning method from simulation to practical application is avoided, so that the control system can keep stable and reliable performance under various working conditions, and the control precision is improved.

This invention discloses a fault detection method and system for Mecanum wheel automated guided vehicles (AGVs), relating to the field of industrial automation fault diagnosis technology. Based on differential inclusion theory and the Euler-Lagrange equations, this invention establishes a Lur'e differential inclusion-type state-space equation characterizing system nonlinearity, parameter uncertainty, and external disturbances. An adaptive event-triggered mechanism with dynamically adjustable parameters and exponentially decaying terms is designed to optimize data transmission, reduce resource consumption, and strictly exclude Zeno behavior. An adaptive event-triggered interval observer is constructed to achieve robust interval estimation of the system state, and actuator fault detection is achieved through the zero-value inclusion property of the residual interval. This invention solves the problem of traditional methods struggling to balance fault detection accuracy and resource utilization, and is suitable for state monitoring and fault diagnosis of Mecanum wheel AGVs in industrial scenarios.

The invention discloses a PINN-based robot mechanical arm kinetic model correction and parameter identification method, and relates to the technical field of robot dynamics, and the method comprises the following steps: collecting the joint position q, speed, acceleration and execution torque data of a mechanical arm; a double-layer PINN with analysis of Euler-Lagrange dynamics as physical prior is constructed; and simultaneously constraining an observation fitting item and a kinetic equation residual error by joint loss, and introducing a self-adaptive weighting mechanism and energy and symplectic structure constraints based on prediction uncertainty. According to the method, through hierarchical structured PINN and SPD parameterization, the problems that physical parameters are poor in interpretability and physically inconsistent with a data driving model are solved at the same time, the network is prevented from directly damaging a key structure, an identification result has physical significance, a controller is more stable when a correction model is used, physically unreasonable parameter values are not prone to occurring during long-term operation, and the method is suitable for large-scale popularization and application. And the generalization ability and the verifiability are improved.

The invention provides a particle positioning method and device for Euler-Lagrange two-phase flow simulation. The method comprises the following steps: acquiring an initial particle position and a unit where the initial particle position is located; calculating the shortest time for the current particle position to reach each surface in the unit, and recording a corresponding surface sequence number; comparing the shortest time for the particles to reach each surface in the unit with the calculated time step length, and judging whether the particles are still in the unit or not; according to the technical scheme provided by the invention, rapid particle positioning and tracking of Euler-Lagrange two-phase flow simulation can be realized.

The invention belongs to the field of security control of information physical systems, and particularly relates to a design method of a Lagrange system adaptive security tracking controller under spoofing attacks, which comprises the following steps: constructing a network attack model and a nonlinear Euler-Lagrange system model; introducing an error variable according to the model and designing a virtual controller by using a backstepping method; designing an adaptive controller and a parameter adaptive law according to the virtual controller; the distributed adaptive control algorithm based on the projection operator solves the influence of preventing signal transmission in the system caused by network attacks, and overcomes the difficulty in controller design caused by mutual coupling between time-varying characteristics of attacks, uncertainty of a non-linear multi-agent system and consistent tracking targets. The method ensures that a plurality of uncertain Euler-Lagrange systems are stable and consistent tracking errors are bounded under the influence of network attacks.

The application provides a method and system for predicting particle settling, collision and flow characteristics in a pump, comprising: constructing a multi-phase flow training sample library of real-time collected environment and virtual-real fusion; after pre-processing the sample library data, using a double-flow deep neural network architecture to extract multiple source feature vectors in parallel; cross-fusing the multiple source feature vectors to obtain a multi-modal fusion feature vector, and establishing an Euler-Lagrange two-way coupled partial differential control equation; inputting the multi-modal fusion feature vector into a physical information neural network based on a gradient-aware adaptive physical loss function, and outputting particle instantaneous settling velocity, particle collision contact force, local entropy production rate and global reconstructed three-dimensional flow field; based on the output calculation index, hierarchical regulation is executed accordingly. The application breaks the feature mismatch problem, replaces the complex grid iteration with millisecond-level forward reasoning conforming to the conservation law of fluid mechanics, and realizes high-precision intelligent online prediction of complex flow field characteristics.