Intelligent numerical simulation and evaluation system based on homotopy matrix method and RAG enhancement

Abstract

The invention discloses an intelligent numerical simulation and evaluation system based on a homotopy matrix method and RAG enhancement. The intelligent numerical simulation and evaluation system is used for high-precision simulation and evaluation of the water outlet process of a pipe folding unmanned aerial vehicle. According to the method, a multi-body system dynamic model is constructed through a homotopy matrix method, and intelligent construction, parameter optimization and multi-physics field coupling solution of a numerical pool can be realized in combination with a computational fluid mechanics (CFD) solver OpenFOAM and RAG technology. According to the method, the problems of complex modeling, low calculation efficiency and insufficient parameter optimization of a traditional method are solved, and efficient simulation and real-time evaluation of the water outlet process of the pipe folding unmanned aerial vehicle are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical fields of computational fluid dynamics (CFD) and unmanned aerial vehicle design, and specifically relates to an intelligent numerical simulation and evaluation system based on the homotopy matrix method and RAG enhancement. Background Art

[0002] Due to its folding structure and high water resistance, the pipe-fold unmanned aerial vehicle is widely used in fields such as maritime rescue and military reconnaissance. However, its water emergence process involves complex fluid-structure interaction effects (such as wave impact and air-water interface phase change), and the traditional numerical simulation methods have the following problems: 1) High modeling complexity: It is necessary to manually set up multi-body dynamics models and fluid boundary conditions, relying on the experience of engineers; 2) Low computational efficiency: Large-scale multi-body coupling simulations require high-performance computing resources and take a long time; 3) Insufficient parameter optimization: Lack of an adaptive optimization mechanism based on historical data.

[0003] In the prior art, multi-body system dynamics modeling methods (such as Lagrange equations and Newton-Euler methods) have high computational complexity and are difficult to adapt to dynamic coupling requirements. The homotopy matrix method uses a 4×4 matrix to uniformly describe the kinematic and dynamic characteristics of a multi-body system, but it has not been combined with CFD simulation and RAG technology. In addition, the existing solutions have not been adapted to the particularity of the water emergence process of the pipe-fold unmanned aerial vehicle, especially there are gaps in the intelligent generation of wave conditions and multi-body dynamics coupling. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent numerical simulation and evaluation system based on the homotopy matrix method and RAG enhancement, to solve the problems of complex modeling, low computational efficiency, and insufficient parameter optimization of traditional methods, and to achieve efficient simulation and real-time evaluation of the water emergence process of the pipe-fold unmanned aerial vehicle.

[0005] To solve the above technical problems, an intelligent numerical simulation and evaluation system based on the homotopy matrix method and RAG enhancement provided by the present invention is characterized by including the following steps:

[0006] Step 1: Multi-body dynamics modeling using the homotopy matrix method

[0007] Use the homotopy matrix method to establish the dynamics model of the unmanned aerial vehicle multi-body system, and uniformly describe the kinematic and dynamic characteristics of the system through the position matrix, velocity matrix, and acceleration matrix; the kinematic and dynamic characteristics of the unmanned aerial vehicle multi-body system are described by the following matrices:

[0008] 1. Position matrix

[0009]

[0010] Among them, is the rotation matrix, is the translation vector;

[0011] 2. Velocity matrix

[0012]

[0013] wherein, is the skew-symmetric matrix of angular velocity, is the linear velocity;

[0014] 3. Acceleration matrix

[0015]

[0016] wherein, is the skew-symmetric matrix of angular acceleration, a i is the linear acceleration;

[0017] 4. Dynamic equation

[0018]

[0019] wherein, M i is the mass matrix, C i is the Coriolis force matrix, is the external force, is the fluid force;

[0020] Step 2: Construction of RAG-enhanced numerical wave tank

[0021] Retrieve historical wave condition data based on RAG technology, generate initial boundary parameters and optimize the turbulence model coefficients;

[0022] 1. Wave condition retrieval

[0023] According to the target wave spectrum (such as the JONSWAP spectrum), use vector embedding technology to retrieve the boundary condition parameters of similar cases from the database and generate an initial configuration file:

[0024]

[0025] wherein, a i , k i , ω i are the wave amplitude, wave number and frequency respectively, φ i is the random phase;

[0026] 2. Parameter optimization

[0027] Parse historical data through an LLM agent (such as OpenFOAMGPT) to optimize the turbulence model coefficients (such as C μ , σ k ) and grid motion parameters. The formula is:

[0028]

[0029] Step 3: Automated CFD Solving Process

[0030] Integrate the CFD solver with the dynamic mesh method (spring approximation method) to achieve fluid-structure interaction simulation;

[0031] 1. Dynamic Mesh Method

[0032] Update the mesh node displacements using the spring approximation method:

[0033] K spring Δx = F fluid (7)

[0034] Where, is the spring stiffness matrix, is the node displacement vector;

[0035] 2. Turbulence Model Adaptation

[0036] Dynamically switch between the k-ε and k-ω SST models according to the local Reynolds number Re:

[0037]

[0038] Step 4: Integrated Evaluation System

[0039] Evaluate the simulation results in real time through the root mean square error (RMSE) and trigger parameter iterative optimization;

[0040] 1. Compare the simulation results with the experimental data in real time through the RAG agent and calculate the root mean square error (RMSE):

[0041]

[0042] 2. If the RMSE exceeds the threshold (e.g., 0.1 m / s), then trigger parameter iterative optimization until the preset accuracy requirements are met.

[0043] Furthermore, in step 2, the target wave spectrum is the JONSWAP spectrum.

[0044] Compared with the prior art, the beneficial effects of the present invention are:

[0045] 1. Improved modeling efficiency: The homotopy matrix method of the present invention uses matrices to uniformly describe multi-body systems, reducing the modeling time by 60%.

[0046] 2. High calculation accuracy: The optimized parameters of the RAG technology of the present invention make the simulation error less than 4%.

[0047] 3. Dynamic adaptability: The turbulence model and dynamic mesh of the present invention are adaptively adjusted to support efficient solving under complex wave conditions.

[0048] 4. Full-process automation: The present invention completes the integration from modeling to evaluation, reducing manual intervention by 85%. Specific implementation manners

[0049] Next, in combination with the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0050] The present invention provides an intelligent numerical simulation and evaluation system based on the homotopy matrix method and RAG enhancement, including the following steps:

[0051] Step 1: Multi-body dynamics modeling using the homotopy matrix method

[0052] Use the homotopy matrix method to establish the dynamic model of the UAV multi-body system, and uniformly describe the kinematic and dynamic characteristics of the system through the position matrix, velocity matrix, and acceleration matrix; the kinematic and dynamic characteristics of the UAV multi-body system are described by the following matrices:

[0053] 1. Position matrix

[0054]

[0055] Among them, is the rotation matrix, is the translation vector;

[0056] 2. Velocity matrix

[0057]

[0058] Among them, is the skew-symmetric matrix of the angular velocity, is the linear velocity;

[0059] 3. Acceleration matrix

[0060]

[0061] Among them, is the skew-symmetric matrix of the angular acceleration, a i is the linear acceleration;

[0062] 4. Dynamic equation

[0063]

[0064] Among them, M i is the mass matrix, Ci is the Coriolis force matrix, is the external force, is the fluid force;

[0065] Step 2: Construction of RAG-enhanced numerical wave tank

[0066] Retrieve historical wave condition data based on RAG technology, generate initial boundary parameters and optimize the turbulence model coefficients;

[0067] 1. Wave condition retrieval

[0068] According to the target wave spectrum (such as the JONSWAP spectrum), use vector embedding technology to retrieve the boundary condition parameters of similar cases from the database and generate the initial configuration file:

[0069]

[0070] where a i , k i , ω i are the wave amplitude, wave number and frequency respectively, and φ i is the random phase;

[0071] 2. Parameter optimization

[0072] Parse historical data through an LLM agent (such as OpenFOAMGPT) to optimize the turbulence model coefficients (such as C μ , σ k ) and grid motion parameters. The formula is:

[0073]

[0074] Step 3: Automated CFD solution process

[0075] Combine the CFD solver with the dynamic grid method (spring approximation method) to achieve fluid-structure interaction simulation;

[0076] 1. Dynamic grid method

[0077] Update the grid node displacement using the spring approximation method:

[0078] K spring Δx = F fluid (7)

[0079] where is the spring stiffness matrix, is the node displacement vector;

[0080] 2. Turbulence model adaptation

[0081] Dynamically switch between the k-ε and k-ω SST models according to the local Reynolds number Re:

[0082]

[0083] Step 4: Integrated Evaluation System

[0084] Evaluate the simulation results in real time through the Root Mean Square Error (RMSE), and trigger the iterative optimization of parameters;

[0085] 1. Compare the simulation results with the experimental data in real time through the RAG agent, and calculate the Root Mean Square Error (RMSE):

[0086]

[0087] 2. If the RMSE exceeds the threshold (such as 0.1 m / s), trigger the iterative optimization of parameters until the preset accuracy requirements are met.

[0088] In the present invention, the target wave spectrum in Step 2 is the JONSWAP spectrum. By introducing the homotopy matrix method and RAG technology, the present invention solves the above problems, simplifies the modeling process, improves the calculation efficiency, and realizes the adaptive optimization of parameters.

[0089] Example 1, Simulation of the Water Ejection Process of a Pipe Folding UAV

[0090] 1. Geometric modeling: Construct a three-dimensional model of the UAV through SolidWorks, export it in STL format and import it into OpenFOAM.

[0091] 2. Homotopy matrix modeling:

[0092] Step 1: Define the position matrix T i of each component of the UAV, i the velocity matrix V i and the acceleration matrix A

[0093] Step 2: Construct the multi-body system dynamics equation according to Equation (4), where the mass matrix M i is calculated through the mass and moment of inertia of the UAV components;

[0094] Step 3: Fluid force is obtained in real time by solving the Navier-Stokes equation through OpenFOAM.

[0095] 3. RAG parameter initialization:

[0096] Step 1: Input the target wave conditions (wave height 2 m, period 8 s);

[0097] Step 2: Retrieve the matching cases in the database through cosine similarity and generate the initial boundary condition file (U, p_rgh);

[0098] Step 3: Call the OpenFOAMGPT agent to optimize C μ parameters.

[0099] 4. Automatic solution:

[0100] Step 1: Generate dynamic mesh code. The example code is as follows:

[0101] dynamicFvMesh dynamicMotionSolverFvMesh;

[0102] motionSolverLibs("libfvMotionSolvers.so");

[0103] solver displacementLaplacian;

[0104] diffusivity quadratic inverseDistance 1.0;

[0105] Step 2: Dynamically switch the turbulence model according to the local Reynolds number;

[0106] Step 3: Allocate resources on the parallel computing cluster and use the interDyMFoam solver for solution.

[0107] Result evaluation:

[0108] Step 1: Extract parameters such as the water outlet velocity and attitude angle from the simulation results;

[0109] Step 2: Calculate the RMSE (Equation (9)). If the error exceeds 0.1 m / s, trigger the optimization of the k-ωSST model parameters;

[0110] Step 3: After iterative optimization, the final RMSE drops to 0.05 m / s, meeting the engineering requirements.

[0111] Example 2, Verification of the accumulation effect of pontoon displacement waves

[0112] 1. Model construction:

[0113] Use the homotopy matrix method to simulate the process of a 50-ton load passing through the pontoon at 9.0 m / s. The pontoon consists of 15 bridge sections with a total length of 100.5 m.

[0114] 2. Parameter setting:

[0115] Added mass coefficient C add = 1.8;

[0116] Roll damping coefficient C roll = 0.36;

[0117] Heaving damping coefficient C heave = 0.26.

[0118] Simulation results:

[0119] The displacement wave accumulation of Node 6 is 0.13 m, and the error from the test data (0.15 m) is 13.3%;

[0120] The distribution of the velocity field and pressure field is more than 90% consistent with the test, verifying the effectiveness of the method.

[0121] The present invention can be used for high-precision simulation and evaluation of the water emergence process of pipe-folded unmanned aerial vehicles. By constructing a multi-body system dynamics model through the homotopy matrix method, combined with the computational fluid dynamics (CFD) solver OpenFOAM and RAG technology, the intelligent construction, parameter optimization and multi-physical field coupled solution of the numerical pool are realized. The core innovations include: 1) Homotopy matrix method: The position, velocity, acceleration and dynamic characteristics of the multi-body system are uniformly described by a 4×4 matrix; 2) Driven by RAG technology: Matching the optimal parameters based on historical wave data to reduce manual intervention; 3) Dynamic coupled solution: Co-optimization of the adaptive turbulence models (k-ε, k-ω SST) and dynamic meshes (spring mesh method); 4) Real-time error correction: Triggering iterative optimization through the root mean square error (RMSE). Experiments show that in complex working conditions with a wave height of 3 m and a flow velocity of 5 m / s, the prediction error of the deployment time of this system is ≤4.2%, the fluid resistance error is ≤5.1%, and the calculation efficiency is increased by 63% compared with the traditional method.

[0122] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as a limitation of the present invention itself. Various changes can be made in its form and details without departing from the spirit and scope of the present invention defined by the appended claims.

Claims

1. An intelligent numerical simulation and evaluation system based on the homotopy matrix method and RAG enhancement, characterized in that: It includes the following steps: Step 1: Multi-body dynamics modeling using the homotopy matrix method. The dynamic model of the UAV multi-body system is established by the homotopy matrix method, and the kinematic and dynamic characteristics of the system are uniformly described by the position matrix, velocity matrix, and acceleration matrix. The kinematic and dynamic characteristics of the UAV multi-body system are described by the following matrices: Position matrix Among them, is a rotation matrix, is a translation vector; Velocity matrix wherein, is the skew-symmetric matrix of the angular velocity, is the linear velocity; Acceleration matrix Among them, is the skew-symmetric matrix of angular acceleration, and a i is the linear acceleration; Dynamic equation Among them, M i is the mass matrix, C i is the Coriolis force matrix, is the external force, is the fluid force; Step 2: Construction of an RAG-enhanced numerical wave tank. Based on RAG technology, historical wave condition data is retrieved to generate initial boundary parameters and optimize the turbulence model coefficients. Wave condition retrieval. According to the target wave spectrum, the boundary condition parameters of similar cases are retrieved from the database using vector embedding technology to generate an initial configuration file: Among them, a i , k i , ω i are the amplitude, wave number, and frequency respectively, and φ i is the random phase; Parameter optimization, through the LLM agent to analyze historical data, optimize the coefficients of the turbulence model (such as C μ , σ k ), and grid motion parameters. The formula is: Step 3: Automated CFD solution process. Combining the CFD solver with the dynamic mesh method to achieve fluid-structure interaction simulation. Dynamic mesh method. The spring approximation method is used to update the displacement of the mesh nodes: K spring Δx = F fluid (7) Among them, is the spring stiffness matrix, is the nodal displacement vector; Turbulence model adaptation. Dynamically switch between the k-ε and k-ω SST models according to the local Reynolds number Re: Step 4: Integrated evaluation system. The simulation results are evaluated in real time through the root mean square error to trigger parameter iterative optimization. The simulation results are compared with the experimental data in real time through the RAG agent to calculate the root mean square error: If the RMSE exceeds the threshold, parameter iterative optimization is triggered until the preset accuracy requirement is met.

2. The intelligent numerical simulation and evaluation system based on the homotopy matrix method and RAG enhancement according to claim 1, characterized in that: The target wave spectrum in Step 2 is the JONSWAP spectrum.

Images