Aerodynamics task processing method and device based on multiple agents

Abstract

The invention provides an aerodynamic task processing method and device based on multiple agents, and relates to the technical field of aircraft simulation design, and the method comprises the steps: obtaining a task target and working condition parameters inputted by a user; based on the capability of each agent group in the multi-agent system, generating a plurality of sub-tasks based on the task target; distributing the plurality of sub-tasks to corresponding agent groups in the multi-agent system for task processing to obtain a calculation result corresponding to the task target; the multi-agent system comprises a first agent group, a second agent group and a third agent group. According to the multi-agent-based aerodynamic task processing method and device provided by the invention, the role of a person in a traditional CFD normal form is replaced by the agents, automatic execution of the whole process of the CFD task is realized, and the processing efficiency of the aerodynamic task is greatly improved.

Description

Technical Field

[0001] The present application relates to the technical field of aircraft simulation design, and in particular to a multi-agent-based aerodynamic task processing method and device. Background Art

[0002] Aerodynamics is a branch of mechanics that studies the force characteristics, gas flow patterns, and accompanying physical and chemical changes of aircraft or other objects in relative motion with air or other gases.

[0003] Taking Computational Fluid Dynamics (CFD) as an example, CFD technology has been developed since the 1970s and has become one of the core technical means for aerodynamic performance analysis. It has significantly reduced dependence on wind tunnel experiments, effectively shortened the R&D cycle and reduced costs.

[0004] However, computational fluid dynamics (CFD) is a related technology, and its related software is complex to operate. The degree of automation in the entire process is insufficient. Key steps such as grid generation and turbulence model selection are completely manually operated, and the accuracy of its calculation results is highly dependent on the experience of engineers. Summary of the Invention

[0005] The purpose of this application is to provide a multi-agent-based aerodynamic task processing method and device, which is used to use intelligent agents to replace the human role in the traditional CFD paradigm, realize the automatic execution of the entire CFD task process, and greatly improve the processing efficiency of aerodynamic tasks.

[0006] This application provides a multi-agent-based aerodynamic task processing method, including: Obtain the task objectives and working condition parameters input by the user; based on the capabilities of each agent group in the multi-agent system, generate multiple subtasks based on the task objectives; assign the multiple subtasks to the corresponding agent groups in the multi-agent system for task processing, and obtain the calculation results corresponding to the task objectives; wherein, the multi-agent system includes: a first agent group, a second agent group and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are correct; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; the third agent group is used to perform fluid mechanics calculations based on the geometric model generated by the geometric agent, and generate the calculation results of the target task.

[0007] Optionally, the assigning of the multiple subtasks to the corresponding agent groups in the multi-agent system for task processing to obtain the calculation results corresponding to the task objectives includes: inputting the first subtask and the operating condition parameters into the member agents of the first agent group, and the member agents analyzing the operating condition parameters based on a first preset prompt word, judging whether the operating condition parameters are correct, and formatting and unifying the operating condition parameters; wherein, the first subtask is: the subtask corresponding to the first agent group among the multiple subtasks; the member agents of the first agent group include: a parameter detection agent and a parameter conversion agent; the parameter detection agent is used to detect whether the operating condition parameters are correct; the parameter conversion agent is used to format and unify the input operating condition parameters.

[0008] Optionally, the assigning of the multiple subtasks to corresponding agent groups in the multi-agent system for task processing to obtain calculation results corresponding to the task objectives includes: selecting at least one member agent from the second agent group that matches the task requirements of the second subtask; the second subtask is: the subtask corresponding to the second agent group in the multiple subtasks; the second subtask and the operating condition parameters are input into the at least one member agent, and the at least one member agent performs modeling based on a second preset prompt word and relevant knowledge in the knowledge base to generate the required geometric model; wherein, when generating the geometric model, the at least one member agent calls the application interface corresponding to the second agent group; the knowledge base is a database containing knowledge related to aerodynamics.

[0009] Optionally, the assigning of the multiple subtasks to corresponding agent groups in the multi-agent system for task processing to obtain calculation results corresponding to the task objectives includes: selecting at least one member agent from the third agent group that matches the task requirements of the third subtask; the third subtask is: the subtask corresponding to the third agent group among the multiple subtasks; the third subtask, the operating parameters and the geometric model output by the second agent group are input into the at least one member agent, and the at least one member agent performs calculations based on a third preset prompt word to obtain calculation results corresponding to the task objectives; wherein, when performing calculations, the at least one member agent calls the application program interface corresponding to the computational fluid dynamics energy agent group.

[0010] Optionally, after assigning the multiple subtasks to corresponding agent groups in the multi-agent system for task processing and obtaining the calculation results corresponding to the task objectives, the method also includes: inputting the geometric model output by the second agent group, the given optimization task and the given operating parameters into the fourth agent group, and having the member agents in the fourth agent group perform aerodynamic optimization on the geometric model output by the second agent group based on a fourth preset prompt word and calling the corresponding optimization tool to obtain an optimized geometric model; inputting the third subtask, the given operating parameters and the optimized geometric model into at least one member agent of the third agent group, and having the at least one member agent perform calculations based on the third preset prompt word to obtain optimized calculation results; wherein, the at least one member agent is a member agent in the third agent group that matches the task requirements of the third subtask.

[0011] Optionally, before assigning the multiple subtasks to corresponding agent groups in the multi-agent system for task processing and obtaining the calculation results corresponding to the task objectives, the method also includes: obtaining text to be processed related to aerodynamics, and performing text segmentation on the content in the text to be processed according to semantics to obtain multiple knowledge texts; inputting the multiple knowledge texts into a vector representation model for vector conversion to obtain a knowledge vector corresponding to each knowledge text in the multiple knowledge texts; and associating each knowledge text in the multiple knowledge texts with the corresponding knowledge vector and storing them in a database to obtain the knowledge base.

[0012] The present application also provides an aerodynamic task processing device based on a multi-agent, comprising: A task division module is used to generate multiple subtasks based on the task goal based on the capabilities of each agent group in the multi-agent system; a task processing module is used to assign the multiple subtasks to the corresponding agent groups in the multi-agent system for task processing to obtain calculation results corresponding to the task goals; wherein, the multi-agent system includes: a first agent group, a second agent group and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are correct; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; the third agent group is used to perform fluid mechanics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task. Optionally, the task processing module is specifically used to input the first subtask and the operating condition parameters into the member agents of the first agent group, and the member agents analyze the operating condition parameters based on the first preset prompt word, determine whether the operating condition parameters are correct, and format and unify the operating condition parameters; wherein, the first subtask is: the subtask among the multiple subtasks corresponding to the first agent group; the member agents of the first agent group include: a parameter detection agent and a parameter conversion agent; the parameter detection agent is used to detect whether the operating condition parameters are correct; the parameter conversion agent is used to format and unify the input operating condition parameters.

[0013] Optionally, the task processing module is specifically used to select at least one member agent from the second agent group that matches the task requirements of the second subtask; the second subtask is: the subtask corresponding to the second agent group among the multiple subtasks; the task processing module is also specifically used to input the second subtask and the operating condition parameters into the at least one member agent, and the at least one member agent performs modeling based on the second preset prompt word and relevant knowledge in the knowledge base to generate the required geometric model; wherein, when generating the geometric model, the at least one member agent calls the application interface corresponding to the second agent group; the knowledge base is a database containing knowledge related to aerodynamics.

[0014] Optionally, the task processing module is specifically used to select at least one member agent from the third agent group that matches the task requirements of the third subtask; the third subtask is: the subtask corresponding to the third agent group among the multiple subtasks; the task processing module is also specifically used to input the third subtask, the operating parameters and the geometric model output by the second agent group into the at least one member agent, and the at least one member agent performs calculations based on a third preset prompt word to obtain a calculation result corresponding to the task goal; wherein, when performing calculations, the at least one member agent calls the application program interface corresponding to the computational fluid dynamics energy agent group.

[0015] Optionally, the task processing module is also used to input the geometric model output by the second intelligent agent group, the given optimization task and the given operating parameters into the fourth intelligent agent group, and the member intelligent agents in the fourth intelligent agent group perform aerodynamic optimization on the geometric model output by the second intelligent agent group based on the fourth preset prompt word and call the corresponding optimization tool to obtain the optimized geometric model; the task processing module is also used to input the third subtask, the given operating parameters and the optimized geometric model into at least one member intelligent agent of the third intelligent agent group, and the at least one member intelligent agent performs calculation based on the third preset prompt word to obtain the optimized calculation result; wherein, the at least one member intelligent agent is a member intelligent agent in the third intelligent agent group that matches the task requirements of the third subtask.

[0016] Optionally, the device also includes: a knowledge base construction module; the knowledge base construction module is used to obtain text to be processed related to aerodynamics, and perform text segmentation on the content in the text to be processed according to semantics to obtain multiple knowledge texts; the knowledge base construction module is also used to input the multiple knowledge texts into a vector representation model for vector conversion to obtain a knowledge vector corresponding to each knowledge text in the multiple knowledge texts; the knowledge base construction module is also used to associate each knowledge text in the multiple knowledge texts with the corresponding knowledge vector and store it in a database to obtain the knowledge base.

[0017] The present application also provides a computer program product, comprising a computer program / instruction, which, when executed by a processor, implements the steps of any of the above-described multi-agent-based aerodynamic task processing methods.

[0018] The present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of any of the above-described multi-agent-based aerodynamic task processing methods are implemented.

[0019] The present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-described multi-agent-based aerodynamic task processing methods.

[0020] The multi-agent-based aerodynamic task processing method and device provided in the present application first obtains the task goal and working condition parameters input by the user; then, based on the capabilities of each agent group in the multi-agent system, multiple subtasks are generated based on the task goal; secondly, the multiple subtasks are assigned to the corresponding agent groups in the multi-agent system for task processing to obtain the calculation results corresponding to the task goal; wherein, the multi-agent system includes: a first agent group, a second agent group, and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are correct; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; the third agent group is used to perform fluid mechanics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task. In this way, the role of humans in the traditional CFD paradigm is replaced by agents, and the automatic execution of the entire CFD task process is realized, which greatly improves the processing efficiency of aerodynamic tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0022] Figure 1 This is a schematic diagram of the structure of the multi-agent system provided by this application; Figure 2 This is a flowchart of the multi-agent-based aerodynamic task processing method provided by this application; Figure 3 This is a schematic diagram of the structure of the multi-agent-based aerodynamic task processing device provided by this application; Figure 4 It is a structural diagram of the electronic device provided in this application. DETAILED DESCRIPTION

[0023] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0024] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of this application can be implemented in an order other than that illustrated or described herein, and that the objects distinguished by "first," "second," and the like are generally of the same type, and do not limit the number of objects; for example, the first object can be one or more. In addition, the term "and / or" in the specification and claims refers to at least one of the connected objects, and the character " / " generally indicates that the objects connected are in an "or" relationship.

[0025] Fluid mechanics provides the fundamental theoretical support for aerodynamics, which is an extension of fluid mechanics to the context of gas flow. The typical process of computational fluid dynamics in related technologies includes: establishing a geometric model, generating structured / unstructured grids, numerically solving equations, and post-processing analysis. Specifically, this process may include the following: 1. Establish a geometric model: The purpose is to define the computational domain of fluid flow and ensure that it accurately reflects the physical problem. The key contents include: Geometric modeling tools: Use CAD software (such as SolidWorks, CATIA, AutoCAD) or dedicated CFD pre-processing tools (such as ANSYS DesignModeler, SpaceClaim) to build a geometric model. Simplification and optimization: Remove small features (such as bolt holes and chamfers) to reduce the amount of calculation while retaining key geometric features (such as curvature and boundary layer areas). Computational domain setting: Determine the scope of the computational domain based on flow characteristics (for example, when simulating pipe flow, ensure that the inlet / outlet is far enough away to avoid boundary interference). Format and compatibility: Export to formats such as STEP, IGES or STL to ensure compatibility with mesh generation tools.

[0026] 2. Generate structured / unstructured meshes: The purpose is to discretize continuous geometry into discrete elements for numerical calculations. Mesh types include: Structured meshes: Nodes are regularly arranged (hexahedrons / quadrilaterals), which are computationally efficient and suitable for simple geometries (such as pipes and airfoils). Unstructured meshes: Composed of triangles / tetrahedrons, suitable for complex geometries (such as car bodies and biological blood vessels). Hybrid meshes: Combining the two (such as prismatic layers for boundary layers and tetrahedrons for the exterior). Key steps include: Meshing tools: ANSYS Meshing, ICEM CFD, Gmsh, Pointwise, etc. Mesh quality check: Evaluate indicators such as aspect ratio, orthogonality, distortion, and Jacobian determinant to avoid negative volume or highly distorted elements.

[0027] 3. Numerical Solution of the Governing Equations: The goal is to discretize and solve the governing equations (such as the Navier-Stokes equations) to obtain flow field data. Key steps include: Equation Selection: Simplify the equations based on the problem (e.g., for incompressible flow, select k-ε, k-ω, SST, or LES for turbulence models). Boundary Condition Setting: Specify conditions such as inlet (velocity, pressure), outlet (static pressure), wall (no-slip / slip), and symmetry planes. Discretization Method: Finite Volume Method (FVM, most commonly used), Finite Element Method (FEM), or Finite Difference Method (FDM).

[0028] 4. Post-processing Analysis and Verification: This aims to extract flow field information, verify the reliability of the results, and support engineering decision-making. Key operations include visualization tools such as ParaView, Tecplot, CFD-Post, and MATLAB. Flow Field Display: Graphs of velocity contours, pressure distribution, streamlines, and vorticity are plotted to identify flow separation and vortex structures. Quantitative Analysis: Calculate force / torque coefficients, heat transfer rates, flow rates, and other parameters, and compare them with experimental data or literature.

[0029] Based on the above, it can be seen that the operation of relevant software for computational fluid dynamics is very complex, and the degree of automation of the entire process is insufficient. In particular, key steps such as grid generation and turbulence model selection are completely manually operated, resulting in the accuracy of the calculation results being highly dependent on the experience of engineers. Although large language model technology has made significant breakthroughs in general fields, it still has significant limitations in the field of aerodynamics: First, the general large language model itself does not have the numerical calculation capabilities to directly handle computational fluid dynamics (CFD) tasks, and has not yet established a deep integration mechanism with mainstream computational fluid dynamics (CFD) solvers, making it impossible to achieve full-chain automation from geometric modeling to post-processing; second, the general large language model lacks a specialized aerodynamic knowledge base and tool support, making it difficult to effectively integrate expert experience and simulation data, and is prone to non-factual outputs in key decision-making links. These technical shortcomings have seriously restricted the practical application value of artificial intelligence technology in the field of aerodynamic design.

[0030] In response to the above-mentioned technical problems existing in the relevant technologies, an embodiment of the present application provides an aerodynamic task processing method based on multi-agents. This method uses agents to replace the role of humans in the traditional CFD paradigm to realize the automatic execution of the entire CFD task process. In addition, it combines professional domain knowledge (such as textbooks, technical documents), tools and large language models (LLM) to improve the accuracy of agent decisions and outputs.

[0031] like Figure 1As shown in the figure, it is a flow chart of the aerodynamic task processing method based on multi-agent provided by the embodiment of the present application. After receiving the task objectives and working condition parameters input by the user, the multi-agent system first generates subtasks corresponding to each agent group, and then sends each subtask to the corresponding agent group, and then executes the subtasks according to the corresponding task objectives and working condition parameters input by the user. Figure 1 The order shown is that each agent group performs the corresponding subtask in turn. It should be noted that after receiving the user's aerodynamic optimization task, the aerodynamic optimization agent group optimizes the geometric model and then performs calculations through the CFD agent group.

[0032] The following describes in detail the multi-agent-based aerodynamic task processing method provided in the embodiment of the present application through specific embodiments and their application scenarios in combination with the accompanying drawings.

[0033] like Figure 2 As shown, an embodiment of the present application provides a multi-agent based aerodynamic task processing method, which may include the following steps 201 to 203: Step 201: Obtain task objectives and operating condition parameters input by the user.

[0034] For example, the above-mentioned task objective can be the task objective of any aerodynamic task. The user needs to specify the task objective and the operating parameters required to calculate the task objective. For example, the user can enter "What are the lift and drag coefficients of RAE2822.dat at Mach numbers of 0.3, 0.8, and angles of attack of -2, 0, and 2 degrees?" For example, to enable the agent to autonomously perform task calculations, the overall task needs to be divided into multiple subtasks and assigned to multiple agents using the Autogen framework for step-by-step completion. Each agent writes prompts for the task and uses function calling technology to call various CFD tools to ensure accurate and reliable results.

[0035] It's important to note that the large language model (LLM) that the agent relies on, with its prompt word design, is much more capable of completing split, single tasks than simply completing the entire complex task. Therefore, using multiple large language models to work together can significantly improve performance. While the performance ceiling of a single large language model has slowed, the performance of models with fewer parameters is still rapidly improving. Therefore, using multiple, smaller large language models is more cost-effective. Furthermore, models with fewer parameters facilitate local deployment and fine-tuning.

[0036] Step 202: Based on the capabilities of each agent group in the multi-agent system, generate multiple subtasks based on the task goal.

[0037] Among them, the multi-agent system includes: a first agent group, a second agent group and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are incorrect; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; the third agent group is used to perform fluid mechanics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task.

[0038] Step 203: Allocate the multiple subtasks to corresponding agent groups in the multi-agent system for task processing to obtain calculation results corresponding to the task objectives.

[0039] It should be noted that since the task requirements of different subtasks are different, the corresponding calculation methods are not exactly the same. Therefore, the embodiment of the present application adopts the method of agent grouping to assign one or more member agents to each agent group. After the corresponding subtask is assigned to the agent group, the corresponding member agent in the group will be selected according to the task requirements to perform the calculation task.

[0040] Specifically, in the above step 203, the step of executing the corresponding subtask for the first agent may further include the following step 203a: Step 203a: input the first subtask and the operating condition parameters into the member agents of the first agent group; the member agents analyze the operating condition parameters based on the first preset prompt words, determine whether the operating condition parameters are correct, and perform formatting and unification operations on the operating condition parameters.

[0041] Among them, the first subtask is: the subtask corresponding to the first agent group among the multiple subtasks; the member agents of the first agent group include: a parameter detection agent and a parameter conversion agent; the parameter detection agent is used to detect whether the operating condition parameters are correct; the parameter conversion agent is used to format and unify the input operating condition parameters.

[0042] For example, the first agent group includes two member agents, namely the parameter detection agent and the parameter conversion agent. After receiving the first subtask assigned by the system, the first agent group assigns the first subtask to the two member agents to perform different computational tasks. Each member agent has a corresponding preset prompt word. Specifically, the first preset prompt word also includes a prompt word corresponding to the parameter detection agent and a prompt word corresponding to the parameter conversion agent.

[0043] Specifically, in the above step 203, the step of executing the corresponding subtask for the second agent may further include the following steps 203b1 and 203b2: Step 203b1: Select at least one member agent from the second agent group that matches the task requirements of the second subtask.

[0044] The second subtask is: a subtask among the multiple subtasks corresponding to the second agent group.

[0045] Step 203b2: input the second subtask and the working condition parameters into the at least one member agent, and the at least one member agent performs modeling based on the second preset prompt word and relevant knowledge in the knowledge base to generate the required geometric model.

[0046] Wherein, when generating the geometric model, the at least one member agent calls the application program interface corresponding to the second agent group; and the knowledge base is a database containing knowledge related to aerodynamics.

[0047] For example, after receiving the second subtask assigned by the system, the second agent group will also select a matching member agent to process according to the task requirements of the second subtask.

[0048] It should be noted that the operating parameters in step 203b2 may also be the operating parameters processed by the parameter conversion agent in the first agent group. Because the member agents in the second agent group perform calculations involving a wealth of specialized knowledge, they also need to access relevant knowledge from a knowledge base containing aerodynamics-related knowledge.

[0049] Specifically, in the above step 203, the step of executing the corresponding subtask for the third agent may further include the following steps 203c1 and 203c2: Step 203c1: Select at least one member agent from the third agent group that matches the task requirements of the third subtask.

[0050] The third subtask is: a subtask among the multiple subtasks corresponding to the third agent group.

[0051] Step 203c2: input the third subtask, the working condition parameters and the geometric model output by the second agent group into the at least one member agent, and the at least one member agent performs calculations based on the third preset prompt word to obtain the calculation results corresponding to the task target.

[0052] Wherein, when performing calculations, the at least one member agent calls the application program interface corresponding to the computational fluid dynamics agent group.

[0053] For example, combining Figure 1 As shown in Table 1 below, this is an example of a multi-agent system provided in an embodiment of the present application, which includes agent groups, member agents within the groups, member agent responsibilities, and plug-in tools that can be called by the agent groups.

[0054] Table 1 Combined with the above Table 1, Figure 1 As shown, after receiving the task objectives and operating parameters input by the user, the multi-agent system generates a first subtask for the flow parameter agent group (the first agent group mentioned above), a second subtask for the geometry agent group (the second agent group mentioned above), and a third subtask for the CFD agent group (the third agent group mentioned above). The system then assigns the subtasks to the corresponding agent groups, which then select appropriate member agents to handle the subtasks based on their requirements. These member agents use pre-defined prompts and simultaneously invoke corresponding external tools (the aforementioned application programming interface) to perform calculations.

[0055] It should be noted that the content shown in Table 1 is only an example and does not constitute a limitation on the member agents in each group of the multi-agent system. Users can add or delete member agents in the agent group according to actual conditions.

[0056] Optionally, in an embodiment of the present application, when the multi-agent system outputs calculation results, the user can also perform aerodynamic optimization on the calculation results.

[0057] For example, after step 203, the multi-agent-based aerodynamic task processing method provided in the embodiment of the present application may further include the following steps 204 and 205: Step 204: The geometric model output by the second intelligent agent group, the given optimization task, and the given operating parameters are input into the fourth intelligent agent group. The member intelligent agents in the fourth intelligent agent group perform aerodynamic optimization on the geometric model output by the second intelligent agent group based on the fourth preset prompt word and call the corresponding optimization tool to obtain the optimized geometric model.

[0058] Step 205: Input the third subtask, the given operating parameters, and the optimized geometric model into at least one member agent of the third agent group, and have the at least one member agent perform calculations based on a third preset prompt word to obtain an optimized calculation result.

[0059] The at least one member agent is a member agent in the third agent group that matches the task requirements of the third subtask.

[0060] Exemplarily, after receiving the optimization task input by the user, the multi-agent system will hand it over to the fourth agent group to regenerate a new geometric model according to the optimization task, and input the regenerated geometric model into the third agent group for recalculation.

[0061] It should be noted that the given operating parameters can be newly entered by the user or entered for the first time. That is, when entering an optimization task, the user can enter new operating parameters. The fourth agent group will regenerate the geometric model based on the newly entered operating parameters, and the third agent group will also perform calculations based on the newly entered operating parameters.

[0062] Optionally, in an embodiment of the present application, the knowledge in books related to aerodynamics may be converted into a vector knowledge base.

[0063] For example, before step 202, the multi-agent-based aerodynamic task processing method provided in the embodiment of the present application may further include the following steps 206 to 208: Step 206: Acquire text to be processed related to aerodynamics, and perform text segmentation on the content in the text to be processed according to semantics to obtain multiple knowledge texts.

[0064] Step 207: Input the plurality of knowledge texts into a vector representation model for vector conversion to obtain a knowledge vector corresponding to each of the plurality of knowledge texts.

[0065] Step 208: Associate each knowledge text in the plurality of knowledge texts with the corresponding knowledge vector and store them in a database to obtain the knowledge base.

[0066] For example, in the embodiments of this application, building a knowledge base primarily includes the following steps: 1. File loading: Text formats such as PDF / Word / TXT / md. Scanned documents must be converted to text in advance; 2. Text segmentation: Semantic segmentation. For specialized documents, LLM segmentation can be used; 3. Text vectorization: Using an embedding model (such as BERT) to convert text into high-dimensional vectors for similarity retrieval; 4. Vector persistence: Writing the text, its corresponding vectors, and indexes to a database (such as Chroma) for easy storage and retrieval. After the knowledge base is built, text retrieval involves vectorizing user input, performing similarity search, and returning the k closest text blocks.

[0067] The aerodynamic task processing method based on multi-agents provided in the embodiment of the present application first obtains the task goal and working condition parameters input by the user; then, based on the capabilities of each agent group in the multi-agent system, multiple subtasks are generated based on the task goal; secondly, the multiple subtasks are assigned to the corresponding agent groups in the multi-agent system for task processing to obtain the calculation results corresponding to the task goals; wherein, the multi-agent system includes: a first agent group, a second agent group and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are correct; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; the third agent group is used to perform fluid mechanics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task. In this way, the role of humans in the traditional CFD paradigm is replaced by agents to realize the automatic execution of the entire CFD task process, greatly improving the processing efficiency of aerodynamic tasks.

[0068] It should be noted that the multi-agent-based aerodynamic task processing method provided in the embodiments of the present application can be executed by a multi-agent-based aerodynamic task processing device, or a control module within the multi-agent-based aerodynamic task processing device that is used to execute the multi-agent-based aerodynamic task processing method. In the embodiments of the present application, the multi-agent-based aerodynamic task processing device executing the multi-agent-based aerodynamic task processing method is used as an example to illustrate the multi-agent-based aerodynamic task processing device provided in the embodiments of the present application.

[0069] It should be noted that the above-mentioned method figures and multi-agent aerodynamic task processing methods in the embodiments of this application are each described by way of example in conjunction with one of the figures in the embodiments of this application. In specific implementations, the multi-agent aerodynamic task processing methods shown in the above-mentioned method figures may also be implemented in conjunction with any other combinable figures shown in the above-mentioned embodiments, and will not be further described here.

[0070] The following describes the multi-agent-based aerodynamic task processing device provided by the present application. The aerodynamic task processing method based on the multi-agent described below can be referenced to each other.

[0071] Figure 3 A schematic diagram of the structure of the multi-agent-based aerodynamic task processing device provided in the embodiment of the present application is shown in FIG. Figure 3 As shown, specifically including: The task acquisition module 301 is used to obtain the task objectives and working condition parameters input by the user; the task division module 302 is used to generate multiple subtasks based on the task objectives based on the capabilities of each agent group in the multi-agent system; the task processing module 303 is used to assign the multiple subtasks to the corresponding agent groups in the multi-agent system for task processing to obtain the calculation results corresponding to the task objectives; wherein, the multi-agent system includes: a first agent group, a second agent group and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are correct; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; the third agent group is used to perform fluid mechanics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task. Optionally, the task processing module 303 is specifically used to input the first subtask and the operating condition parameters into the member agents of the first agent group, and the member agents analyze the operating condition parameters based on the first preset prompt word, determine whether the operating condition parameters are correct, and format and unify the operating condition parameters; wherein, the first subtask is: the subtask among the multiple subtasks corresponding to the first agent group; the member agents of the first agent group include: a parameter detection agent and a parameter conversion agent; the parameter detection agent is used to detect whether the operating condition parameters are correct; the parameter conversion agent is used to format and unify the input operating condition parameters.

[0072] Optionally, the task processing module 303 is specifically used to select at least one member agent from the second agent group that matches the task requirements of the second subtask; the second subtask is: the subtask corresponding to the second agent group among the multiple subtasks; the task processing module 303 is also specifically used to input the second subtask and the operating condition parameters into the at least one member agent, and the at least one member agent performs modeling based on the second preset prompt word and relevant knowledge in the knowledge base to generate the required geometric model; wherein, when generating the geometric model, the at least one member agent calls the application interface corresponding to the second agent group; the knowledge base is a database containing knowledge related to aerodynamics.

[0073] Optionally, the task processing module 303 is specifically used to select at least one member agent from the third agent group that matches the task requirements of the third subtask; the third subtask is: the subtask corresponding to the third agent group among the multiple subtasks; the task processing module 303 is also specifically used to input the third subtask, the operating parameters and the geometric model output by the second agent group into the at least one member agent, and the at least one member agent performs calculations based on a third preset prompt word to obtain a calculation result corresponding to the task goal; wherein, when performing calculations, the at least one member agent calls the application program interface corresponding to the computational fluid dynamics energy agent group.

[0074] Optionally, the task processing module 303 is also used to input the geometric model output by the second intelligent agent group, the given optimization task and the given operating parameters into the fourth intelligent agent group, and the member intelligent agents in the fourth intelligent agent group perform aerodynamic optimization on the geometric model output by the second intelligent agent group based on the fourth preset prompt word and call the corresponding optimization tool to obtain the optimized geometric model; the task processing module 303 is also used to input the third subtask, the given operating parameters and the optimized geometric model into at least one member intelligent agent of the third intelligent agent group, and the at least one member intelligent agent performs calculation based on the third preset prompt word to obtain the optimized calculation result; wherein, the at least one member intelligent agent is a member intelligent agent in the third intelligent agent group that matches the task requirements of the third subtask.

[0075] Optionally, the device also includes: a knowledge base construction module; the knowledge base construction module is used to obtain text to be processed related to aerodynamics, and perform text segmentation on the content in the text to be processed according to semantics to obtain multiple knowledge texts; the knowledge base construction module is also used to input the multiple knowledge texts into a vector representation model for vector conversion to obtain a knowledge vector corresponding to each knowledge text in the multiple knowledge texts; the knowledge base construction module is also used to associate each knowledge text in the multiple knowledge texts with the corresponding knowledge vector and store it in a database to obtain the knowledge base.

[0076] The multi-agent-based aerodynamic task processing device provided by the present application first obtains the task goal and working condition parameters input by the user; then, based on the capabilities of each agent group in the multi-agent system, multiple subtasks are generated based on the task goal; secondly, the multiple subtasks are assigned to the corresponding agent groups in the multi-agent system for task processing to obtain the calculation results corresponding to the task goal; wherein, the multi-agent system includes: a first agent group, a second agent group, and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are correct; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; the third agent group is used to perform fluid mechanics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task. In this way, the use of agents to replace the role of humans in the traditional CFD paradigm realizes the automatic execution of the entire CFD task process, greatly improving the processing efficiency of aerodynamic tasks.

[0077] Figure 4 An example of a physical structure diagram of an electronic device is shown below. Figure 4 As shown, the electronic device may include: a processor (processor) 410 , a communication interface (Communications Interface) 420 , a memory (memory) 430 and a communication bus 440 , wherein the processor 410 , the communication interface 420 , and the memory 430 communicate with each other via the communication bus 440 . The processor 410 can call logic instructions in the memory 430 to execute a multi-agent-based aerodynamic task processing method, which includes: first, obtaining a task objective and operating condition parameters input by a user; then, based on the capabilities of each agent group in the multi-agent system, generating multiple subtasks based on the task objective; and second, assigning the multiple subtasks to corresponding agent groups in the multi-agent system for task processing to obtain calculation results corresponding to the task objective. The multi-agent system includes: a first agent group, a second agent group, and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are incorrect; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; and the third agent group is used to perform fluid dynamics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task. In this way, the use of agents replaces the human role in the traditional CFD paradigm, realizing the automatic execution of the entire CFD task process, greatly improving the processing efficiency of aerodynamic tasks.

[0078] In addition, the logical instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present application, or the part that contributes to the existing technology, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program code.

[0079] On the other hand, the present application also provides a computer program product, which includes a computer program stored on a computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the multi-agent aerodynamic task processing method provided by the above methods, the method including: first, obtaining the task objectives and working condition parameters input by the user; then, based on the capabilities of each agent group in the multi-agent system, generating multiple subtasks based on the task objectives; secondly, assigning the multiple subtasks to the corresponding agent groups in the multi-agent system for task processing to obtain the calculation results corresponding to the task objectives; wherein the multi-agent system includes: a first agent group, a second agent group and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are incorrect; the second agent group is used to generate the geometric model required for the target task based on the condition parameters given in the target prompt word; the third agent group is used to perform fluid mechanics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task. In this way, intelligent agents are used to replace the human role in the traditional CFD paradigm, realizing the automatic execution of the entire CFD task process, greatly improving the processing efficiency of aerodynamic tasks.

[0080] In another aspect, the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned multi-agent-based aerodynamic task processing method, comprising: first, obtaining a task objective and operating condition parameters input by a user; then, based on the capabilities of each agent group in the multi-agent system, generating multiple subtasks based on the task objective; second, assigning the multiple subtasks to corresponding agent groups in the multi-agent system for task processing to obtain calculation results corresponding to the task objective; wherein the multi-agent system includes: a first agent group, a second agent group, and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are incorrect; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; and the third agent group is used to perform fluid dynamics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task. In this way, the use of agents replaces the human role in the traditional CFD paradigm, realizing the automatic execution of the entire CFD task process, greatly improving the processing efficiency of aerodynamic tasks.

[0081] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0082] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0083] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A multi-agent-based aerodynamic task processing method, characterized in that: include: Obtain task objectives and working condition parameters input by the user; generating a plurality of subtasks based on the task goal based on the capabilities of each agent group in the multi-agent system; Assigning the plurality of subtasks to corresponding agent groups in the multi-agent system for task processing to obtain calculation results corresponding to the task objectives; Among them, the multi-agent system includes: a first agent group, a second agent group and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are incorrect; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; the third agent group is used to perform fluid mechanics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task.

2. The method according to claim 1, characterized in that The assigning of the plurality of subtasks to corresponding agent groups in the multi-agent system for task processing to obtain calculation results corresponding to the task objectives includes: Inputting the first subtask and the operating condition parameters into member agents of the first agent group, the member agents analyzing the operating condition parameters based on a first preset prompt word, determining whether the operating condition parameters are correct, and performing a formatting and unification operation on the operating condition parameters; Among them, the first subtask is: the subtask corresponding to the first agent group among the multiple subtasks; the member agents of the first agent group include: a parameter detection agent and a parameter conversion agent; the parameter detection agent is used to detect whether the operating condition parameters are correct; the parameter conversion agent is used to format and unify the input operating condition parameters.

3. The method according to claim 1, characterized in that The assigning of the plurality of subtasks to corresponding agent groups in the multi-agent system for task processing to obtain calculation results corresponding to the task objectives includes: Selecting at least one member agent from the second agent group that matches the task requirements of the second subtask; the second subtask is: a subtask corresponding to the second agent group among the multiple subtasks; Inputting the second subtask and the working condition parameters into the at least one member agent, and having the at least one member agent perform modeling based on the second preset prompt word and relevant knowledge in the knowledge base to generate a required geometric model; Wherein, when generating the geometric model, the at least one member agent calls the application program interface corresponding to the second agent group; and the knowledge base is a database containing knowledge related to aerodynamics.

4. The method according to claim 1 or 3, characterized in that The assigning of the plurality of subtasks to corresponding agent groups in the multi-agent system for task processing to obtain calculation results corresponding to the task objectives includes: Selecting at least one member agent from the third agent group that matches the task requirements of a third subtask; the third subtask being: a subtask from the plurality of subtasks corresponding to the third agent group; Inputting the third subtask, the operating condition parameters, and the geometric model output by the second agent group into the at least one member agent, and having the at least one member agent perform calculations based on a third preset prompt word to obtain a calculation result corresponding to the task objective; Wherein, when performing calculations, the at least one member intelligent agent calls the application program interface corresponding to the computational fluid dynamics energy body group.

5. The method according to claim 1, characterized in that After allocating the plurality of subtasks to corresponding agent groups in the multi-agent system for task processing and obtaining calculation results corresponding to the task objectives, the method further includes: The geometric model output by the second agent group, the given optimization task, and the given operating condition parameters are input into a fourth agent group, and the member agents of the fourth agent group perform aerodynamic optimization on the geometric model output by the second agent group based on a fourth preset prompt word and calling a corresponding optimization tool to obtain an optimized geometric model; Inputting the third subtask, the given operating parameters, and the optimized geometric model into at least one member agent of the third agent group, and having the at least one member agent perform calculations based on a third preset prompt word to obtain an optimized calculation result; The at least one member agent is a member agent in the third agent group that matches the task requirements of the third subtask.

6. The method according to claim 3 or 5, characterized in that Before allocating the plurality of subtasks to corresponding agent groups in the multi-agent system for task processing and obtaining calculation results corresponding to the task objectives, the method further comprises: Acquire a text to be processed related to aerodynamics, and perform text segmentation on the content of the text to be processed according to semantics to obtain multiple knowledge texts; Inputting the plurality of knowledge texts into a vector representation model for vector conversion to obtain a knowledge vector corresponding to each knowledge text in the plurality of knowledge texts; Each knowledge text in the plurality of knowledge texts is associated with a corresponding knowledge vector and stored in a database to obtain the knowledge base.

7. A multi-agent-based aerodynamic task processing device, characterized in that: The device comprises: The task acquisition module is used to obtain the task objectives and working condition parameters input by the user; a task division module for generating a plurality of subtasks based on the task goal based on the capabilities of each agent group in the multi-agent system; A task processing module is used to assign the multiple subtasks to corresponding agent groups in the multi-agent system for task processing, and obtain calculation results corresponding to the task objectives; Among them, the multi-agent system includes: a first agent group, a second agent group and a third agent group; the first agent group is used to check whether the calculation conditions given in the target prompt word are incorrect; the second agent group is used to generate the geometric model required for the target task based on the conditional parameters given in the target prompt word; the third agent group is used to perform fluid mechanics calculations based on the geometric model generated by the geometric agent to generate the calculation results of the target task.

8. The device according to claim 7, characterized in that The task processing module is specifically configured to input the first subtask and the operating condition parameters into the member agents of the first agent group, and the member agents analyze the operating condition parameters based on the first preset prompt word, determine whether the operating condition parameters are correct, and perform a unified formatting operation on the operating condition parameters; Among them, the first subtask is: the subtask corresponding to the first agent group among the multiple subtasks; the member agents of the first agent group include: a parameter detection agent and a parameter conversion agent; the parameter detection agent is used to detect whether the operating condition parameters are correct; the parameter conversion agent is used to format and unify the input operating condition parameters.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the multi-agent-based aerodynamic task processing method as claimed in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon, and when the computer program is executed by a processor, the steps of the multi-agent-based aerodynamic task processing method according to any one of claims 1 to 6 are implemented.

Images