Industrial design automation method based on artificial intelligence large model and intelligent agent

Abstract

The invention discloses an industrial design automation method based on an artificial intelligence large model and an intelligent agent, relates to the technical field of industrial design, and solves the technical problem that industrial design automation is difficult to realize at present. The method comprises the following steps: automatically matching a large model according to input industrial product related information; extracting a design optimization problem through a problem analysis agent based on a large model, and decomposing the design optimization problem into optimization tasks; a target function or a simulation code is generated through the target quantification agent, and an interface is provided; mapping from intelligent body fitting simulation input to output is built through the agent model, and an interface is provided; packaging the interface through an optimization packaging agent and executing an optimization task; according to the embodiment of the design automation, a hierarchical agent architecture for industrial design automation by utilizing an artificial intelligence large model is obtained through generalization. According to the invention, high automation of industrial design can be realized.

Description

Technical Field

[0001] The present invention relates to the field of industrial design technology, and in particular to an industrial design automation method based on an artificial intelligence large model and an intelligent agent. Background Art

[0002] In recent years, large language models (LLMs) have rapidly emerged and demonstrated unprecedented capabilities in natural language processing tasks. These models have been trained on a large corpus and have the ability to understand and generate human language, thus opening up new avenues for applications in various industries. Large language models are also widely used in the field of industrial design.

[0003] The design cycle of industrial products usually includes market research, demand analysis, conceptual design and detailed design. The development of large language models has brought great development opportunities for industrial design. At the market research level, studies have shown that various LLMs have responded to classic behavioral economics experiments in a way that is consistent with intuition and experience. For example, a method currently used is for market researchers to enhance the response of large language models by fine-tuning and combining previous survey data from similar backgrounds; this method improves the consistency of large language models' responses to existing product features and more importantly new product features with human responses. In addition, at the demand analysis level, some people have explored the potential of large language models in promoting the requirements engineering (RE) process, proposed key directions and SWOT analysis for the research and development of requirements engineering using large language models, and improved the efficiency and accuracy of tasks related to requirements. In terms of conceptual design, some people have proposed an artificial intelligence-enhanced multimodal collaborative design framework, using large language models to establish an iterative prompt mechanism to generate precise visual solutions. In addition, some people have proposed a morphological analysis method for conceptual design based on large language model enhancement, which provides targeted guidance and support for designers when applying morphological analysis by refining the design process into three main stages: decomposition, generation and combination. In terms of detailed design, some people have made fine-tuning to make the large language model more conducive to API calls of industrial software, thereby realizing the use of LLM for automated detailed design; in addition, there are methods to achieve the optimal design of shear wall structures by converting the simulation intentions of CAE engineers into code.

[0004] In industrial design, product design optimization is an important way to improve product quality and performance. It is not only related to product functionality, production efficiency and cost, but also directly affects market competitiveness and sustainable development. The industrial design process usually involves multiple interrelated and coupled processes. In order to achieve global optimal operation, it is necessary to comprehensively consider the complexity of problems in industrial design, multi-objective, multi-scale and dynamic optimization. At present, various heuristic algorithms and deep learning algorithms have been applied to industrial design, and the scope of application is constantly expanding, covering fields such as mechanical design and circuit design.

[0005] Specifically, in the paper titled "Large language models as evolutionary optimizers" (Liu S, Chen C, Qu X, et al. Large language models as evolutionary optimizers [C] / / 2024 IEEE Congress on Evolutionary Computation (CEC). IEEE, 2024: 1-8.), a large language model is used as an evolutionary operator for zero-shot learning, showing that a large language model can solve complex NP (nondeterministic polynomial) difficult combinatorial optimization problems (NP-hard problems). However, for small-scale problem execution, the complete optimization process of this method is very expensive and time-consuming (about half a day). In addition, in industrial design, both mechanical design and circuit design require mathematical modeling of physical world products to simulate the physical model of the product; however, the optimizer used in this method lacks mathematical modeling technology for the physical world, making it difficult to quantify the quality of the design, and therefore difficult to apply to industrial problems.

[0006] In addition, in the paper titled "A large language model-based computing platform for automated building energy modeling" (Jiang G, Ma Z, Zhang L, et al. EPlus-LLM: A large language model-based computing platform for automated building energy modeling [J]. Applied Energy, 2024, 367: 123431.), fine-tuning technology is used to adjust the general large language model to a code model suitable for application programming interface (API) calls, thereby achieving highly automated modeling and simulation. This method enables the large language model to establish an understanding of the feedback of the physical world through the API, supporting the automation of industrial design through API calls. However, the design of this method is in an open-loop state. Given the relatively weak mathematical capabilities of the large language model, this open-loop design method is unreliable and requires the introduction of an optimization process to ensure the stability of the design results. Therefore, in the paper titled "Intelligent design and optimization system for shear wall structures based on large language models and generative artificial intelligence" (Qin S, Guan H, Liao W, et al. Intelligent design and optimization system for shear wall structures based on large language models and generative artificial intelligence [J]. Journal of Building Engineering, 2024, 95: 109996.), the large language model is used as the core controller to interact with engineers, interpret their language descriptions and convert them into executable computer code. Subsequently, the system autonomously completes the intelligent design task using appropriate structure generation and optimization methods. However, in this approach, the interaction between humans and the large language model is continuous, and the design needs to be constantly adjusted, making it difficult to achieve a highly automated system.

[0007] In the process of implementing the present invention, the inventors found that there are at least the following problems in the prior art: Currently, it is difficult to achieve a high degree of automation in industrial design. Summary of the invention

[0008] The purpose of the present invention is to provide an industrial design automation method based on an artificial intelligence large model and an intelligent agent to solve the technical problem that it is difficult to achieve high automation of industrial design in the prior art. The various technical effects that can be produced by the preferred technical solution among the various technical solutions provided by the present invention are described in detail below.

[0009] To achieve the above object, the present invention provides the following technical solutions: The present invention provides an industrial design automation method based on an artificial intelligence large model and an intelligent agent, comprising the following steps: Extract design optimization problems based on input relevant information of industrial products, and automatically match the large language model based on the design optimization problems; Decomposing the design optimization problem into a target quantification task, a proxy model establishment task and an optimization initialization task through the large language model; Generate an objective function according to the target quantization task, and convert the objective function into a target quantization interface; Establishing a task and target quantization interface according to the proxy model, using a neural network to fit a mapping of simulator input to output to establish a proxy model, and converting the proxy model to a proxy model interface; According to the optimization initialization task, the target quantization interface and the proxy model interface, the proxy model is optimized to obtain an optimization result, and the optimal design solution of the industrial product is output according to the optimization result.

[0010] Optionally, the automatically matching a large language model according to the design optimization problem includes: Acquire evaluation information of various different large language models from the historical call records of the large language model; According to the task type of the design optimization problem and the evaluation information of the large language model, the optimal large language model is automatically matched to the design optimization problem.

[0011] Optionally, generating the objective function according to the target quantization task includes: calling simulation software to generate the objective function according to the complexity of the target quantization task, or directly allowing the large language model to generate the objective function.

[0012] Optionally, the calling of simulation software to generate the objective function includes: Using the finite element analysis model in the simulation software to pre-process the target quantification task; According to the pre-processing results, the target function is generated by combining the thinking chain and retrieval enhancement generation and calling the simulator to simulate.

[0013] Optionally, the method further includes: using the large language model as an analyzer to determine the complexity of the target quantization task.

[0014] Optionally, the step of establishing a task and target quantization interface according to the proxy model and using a neural network to fit the mapping of simulator input to output to establish the proxy model comprises: Acquire a solution space of the proxy model establishment task through the large language model; Perform uniform sampling in the solution space, and construct a structural data set of the industrial product through the target quantization interface and the sampled data; The neural network is trained using the structured data set to obtain the proxy model.

[0015] Optionally, after the neural network is trained using the structured data set, the method further includes: evaluating the convergence curve and fitting effect of the neural network through the large language model; if the convergence curve diverges or the fitting effect is unsatisfactory, adjusting the learning rate and number of training rounds of the neural network through the large language model and retraining.

[0016] Optionally, before training the neural network, the method further includes: normalizing the structured data set using the large language model.

[0017] Optionally, performing target optimization on the proxy model according to the optimization initialization task, the target quantization interface and the proxy model interface includes: By searching for enhanced generation, an optimization algorithm matching the initial optimization task is selected from a preset algorithm library; The target quantization interface and the proxy model interface are packaged into the target function of the target function, and the target function is optimized by the selected optimization algorithm.

[0018] Optionally, the target optimization of the proxy model according to the optimization initialization task, the target quantization interface and the proxy model interface also includes: after the target function is optimized, a convergence check is performed on the optimization result obtained by the target optimization through the large language model; if it is checked that the optimization result has not converged, the large language model adjusts the hyperparameters in the target optimization process for re-optimization.

[0019] A hierarchical agent framework applicable to the field of industrial design, used to execute any of the above-mentioned industrial design automation methods based on artificial intelligence large models and agents to solve industrial design automation problems, including: L0 demand analysis and conceptual design agent, L1 preliminary design agent, and L2 detailed design agent; The L0 demand analysis and concept design agent is used to extract design optimization problems based on relevant information of industrial products, and decompose the design optimization problems into target quantification tasks, agent model establishment tasks and optimization initialization tasks; The L1 preliminary design agent is used to generate an objective function according to the target quantization task, convert the objective function into a target quantization interface, establish a task and a target quantization interface according to the proxy model, use a neural network to fit the mapping of simulator input to output to establish a proxy model, and convert the proxy model into a proxy model interface; The L2 detailed design agent is used to perform target optimization on the proxy model according to the optimization initialization task, target quantization interface and proxy model interface, obtain optimization results, and output the optimal design solution for the industrial product according to the optimization results.

[0020] Implementing one of the above technical solutions of the present invention has the following advantages or beneficial effects: The present invention uses a large language model as a bridge between product simulation and optimization. All process steps in industrial design are automatically completed by the large language model, realizing full-process automation of industrial design from design to optimization. By converting target quantification tasks and proxy model establishment tasks into interfaces and opening them to subsequent operations to receive feedback at the level of the complex physical world, the dependence on the mathematical capabilities of the large language model in the optimization process is eliminated. The proxy model is designed to automatically establish a model, and the proxy model is automatically established through multiple rounds of dialogue before optimization, thereby reducing the repeated simulation time in the optimization process, greatly reducing the simulation time, and ensuring the accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. It is obvious that the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 is a flow chart of an industrial design automation method based on an artificial intelligence big model and an intelligent agent according to a first embodiment of the present invention; Figure 2 is a structural schematic diagram of a ten-bar truss according to an embodiment of the present invention; Figure 3This is a flowchart of step S1 of an industrial design automation method based on an artificial intelligence big model and an intelligent agent in Embodiment 1 of the present invention; Figure 4 This is a flowchart of step S3 of the industrial design automation method based on an artificial intelligence large model and an intelligent agent in Embodiment 1 of the present invention; Figure 5 This is a flowchart of step S4 of the industrial design automation method based on an artificial intelligence large model and an intelligent agent in Embodiment 1 of the present invention; Figure 6 This is a flowchart of step S5 of the industrial design automation method based on an artificial intelligence large model and an intelligent agent in Embodiment 1 of the present invention; Figure 7 is a simulation diagram of a ten-bar truss according to Embodiment 1 of the present invention; Figure 8 is a convergence curve diagram of the ten-bar truss proxy model of the first embodiment of the present invention; Fig. 9 is an optimization result diagram of a ten-bar truss according to Embodiment 1 of the present invention; Fig.10 It is a structural diagram of a hierarchical intelligent agent framework applicable to the field of industrial design according to the second embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solutions and advantages of the present invention clearer, the various exemplary embodiments to be described below will refer to the corresponding drawings, which constitute a part of the exemplary embodiments, wherein various exemplary embodiments that may be used to implement the present invention are described. Unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation methods described in the following exemplary embodiments do not represent all implementation methods consistent with the present disclosure. It should be understood that they are only examples of processes, methods, devices, etc. that are consistent with some aspects of the present disclosure as detailed in the attached claims, and other embodiments may also be used, or the embodiments listed herein may be modified in structure and function without departing from the scope and essence of the present invention.

[0023] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", etc. indicate the orientation or positional relationship based on the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the elements referred to must have a specific orientation, be constructed and operated in a specific orientation. The terms "first", "second", etc. are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. The term "multiple" means two or more. The terms "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, an integral connection, a mechanical connection, an electrical connection, a communication connection, a direct connection, an indirect connection through an intermediate medium, and can be the internal connection of two elements or the interaction relationship between two elements. The term "and / or" includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0024] In order to illustrate the technical solution of the present invention, a specific embodiment is used below for description, and only the parts related to the embodiment of the present invention are shown.

[0025] Embodiment 1: like Figure 1 As shown, the present invention provides an industrial design automation method based on an artificial intelligence large model and an intelligent agent, comprising the following steps: S1, extracting design optimization problems based on input relevant information of industrial products, and automatically matching a large language model based on the design optimization problems; wherein the industrial products may be household appliances, such as televisions, refrigerators, washing machines, etc.; electronic products, such as laptops, mobile phones, smart watches, etc.; transportation vehicles, such as cars, ships, drones, etc.; the extracted design optimization problems may be optimization of the mechanical structure or circuit structure of these products; S2, decomposing the design optimization problems into target quantization tasks, proxy model establishment tasks and optimization initialization tasks through a large language model; S3, generating an objective function based on the target quantization tasks, and converting the objective function into a target quantization interface; S4, based on the proxy model establishment tasks and the target quantization interface, using a neural network to fit the mapping of the simulator input to the output to establish a proxy model, and converting the proxy model into a proxy model interface; S5, performing target optimization on the proxy model based on the optimization initialization task, the target quantization interface and the proxy model interface, obtaining an optimization result, and outputting the optimal design scheme for the industrial product based on the optimization result.

[0026] The industrial design automation method based on artificial intelligence big model and intelligent agent provided in this embodiment is applied to the field of industrial design, and the big language model is used as a bridge between product simulation and optimization. All process steps in industrial design are automatically completed by the big language model, realizing the full process automation from design to optimization of industrial products; by converting the target quantification task and the agent model establishment task into an interface and opening it to subsequent operations to receive feedback at the level of the complex physical world, the dependence on the mathematical ability of the big language model in the optimization process is eliminated; the agent model is designed to automatically establish the model, and the agent model is automatically established through multiple rounds of dialogue before optimization, thereby reducing the repeated simulation time in the optimization process, greatly reducing the simulation time, and ensuring the model accuracy.

[0027] Next, combine Figures 1 to 9 The specific implementation steps of the industrial design automation method based on the artificial intelligence big model and intelligent agent provided in this embodiment are introduced in detail: First, execute step S1, extract the design optimization problem according to the input relevant information of the industrial product, and automatically match the large language model according to the design optimization problem; Figure 2 As shown in the figure, for example, if you need to design a ten-bar truss structure, which is welded from ten beam units, then its product information can be a cross-sectional view of the structure in CAD software or a PyAnsys modeling code of a 3D model, etc. Based on this information, the design optimization problem that can be extracted is: how to set the horizontal length l of the left half, the width w1 of the left beam, and the width w2 of the right beam to make the structure optimal. Then, according to these optimization problems, a large language model that is most suitable for solving these problems is matched.

[0028] Specifically, Figure 3As shown, the large language model is automatically matched according to the design optimization problem, including: S11, obtaining the evaluation information of various large language models in the historical call records of the large language model; there are many types of large language models, such as: recurrent neural network, convolutional neural network, Transformer model, etc. according to the model architecture; according to the model scale, it can be divided into: small language model, large language model, super large model, etc.; each type is applicable to different task situations, so the processing effect of each model for different task types can be queried through the historical call information. S12, according to the task type of the design optimization problem and the evaluation information of the large language model, the optimal large language model is automatically matched to the design optimization problem. Matching a large language model (i.e., the optimal large language model) that is most suitable for processing this task through historical call records can improve the task processing efficiency and the reliability of the generated results. Intelligent selection and scheduling of calls to different large language models according to different task types can ensure that the most appropriate large language model is selected when processing specific tasks, thereby improving the quality of the generated results. By integrating the advantages of multiple models, in complex industrial design tasks, the major language models can work together to effectively solve different types of problems and ensure the accuracy and stability of the generated results.

[0029] Then, step S2 is executed to decompose the design optimization problem into target quantification tasks, proxy model building tasks, and optimization initialization tasks through the large language model. The target quantification task is to transform abstract product information into specific and measurable indicators so as to more clearly evaluate the accuracy of the design optimization product. Figure 2 Taking the ten-bar truss structure in the example, its goal can be quantified as solving the maximum volume and maximum deformation of the ten-bar truss structure. The task of establishing the proxy model is to build a proxy model of industrial products and simulate and verify the fitting effect of the proxy model and the actual industrial products. Figure 2 The ten-bar truss structure in the example has a proxy model building task of building a ten-bar truss proxy model 1 with the goal of maximizing the volume and building a ten-bar truss proxy model 2 with the goal of maximizing the deformation. The optimization initialization task is to optimize the model built in the proxy model building task. Figure 2 The optimization initialization task for the ten-bar truss structure in the figure is to optimize the target of agent model 1 and agent model 2. The large language model automatically decomposes the design optimization problem, and the established target quantification tasks and agent model tasks enable the large language model to receive feedback from the complex physical world, eliminating the reliance on the mathematical capabilities of the large language model during the optimization process.

[0030] Next, execute step S3, generate an objective function according to the target quantization task, and convert the objective function into a target quantization interface. After the objective function is generated, use the large language model as a controller and code generator to convert the objective function into the form of code, and then package the code into a target quantization interface, and the interface is always open to subsequent operations (that is, the interface can be called at any time in the subsequent process). It should be noted that the objective function can be converted into Python code, C language code, C++ code or Java code, which can be selected according to actual conditions. In the target quantization task, a large language model is used as an analyzer, and natural language is first used to evaluate the shape and size of the task model, identify optimization variables and solution space, and finally determine the number and complexity of the optimization objectives, and convert the fuzzy optimization requirements into quantified optimization models.

[0031] Further, the objective function is generated according to the target quantification task, including: calling simulation software to generate the objective function according to the complexity of the target quantification task, or directly letting the large language model generate the objective function. Further, the large language model can be used as an analyzer to determine the complexity of the target quantification task. The large language model identifies the complexity of the target quantification task by analyzing whether the optimization target in the target optimization task is easy. For the optimization target that is easy to model, the large language model can directly convert it into a mathematical function (i.e., the objective function), and then convert it into code, and then dynamically verify and package it into a target quantification interface. For the optimization target that is difficult to model, the large language model calls the simulation software to generate the objective function. For industrial products of the circuit type, simulation software such as LTspice, Proteus, PSPICE or MultisimLive, which are specially used for electronic circuits, can be used; for industrial products of the mechanical structure type, simulation software such as Ansys or COMSOLMultiphysics, which are specially used for multi-physics fields, can be used. Calling the simulation software can convert the complex partial differential equations in the physical problems in the task and the engineering problems into the form of global matrix equations to simplify the solution of the equations and save the amount of calculation.

[0032] Specifically, Figure 4As shown, calling the simulation software to generate the target function includes: S31, using the finite element analysis model in the simulation software to pre-process the target quantification task; the pre-processing includes network partitioning, constraint application, load application, etc.; S32, based on the pre-processing results, combining thought chain and retrieval enhancement generation to call the simulator to simulate and generate the target function. Among them, the thought chain (Chain of Thought, CoT) is a technology used in the field of artificial intelligence and machine learning to enhance the model's reasoning ability. It can improve the performance and interpretability of the model on complex problems by explicitly including intermediate reasoning steps in the output of the model; retrieval-augmented generation (RAG) is a model that combines retrieval and generation techniques. It can refer to the information of the external knowledge base to generate answers or content. The proxy model is called through the simulation software's API to obtain a small number of simulation examples and API documents to start the simulation (such as pre-processing, load application and post-processing) to generate the target function, which is converted into code through a large language model, and then the code is passed to the tester to verify the simulation results, and finally packaged into an interface.

[0033] Next, execute step S4, establish the task and target quantization interface according to the proxy model, use the neural network to fit the mapping of the simulator input to the output to establish the proxy model, and convert the proxy model into a proxy model interface. Among them, Computer Aided Engineering (CAE) software can be used as a simulator. Similarly, use the large language model as a controller and code generator to call the neural network to establish the proxy model and convert the proxy model into code, and then package the code into a proxy model interface so that it can be called at any time in the subsequent process. If in step S3, the objective function is directly generated by the large language model, it is not necessary to establish a proxy model and directly enter the optimization initialization task.

[0034] In actual optimization cases, the finite element analysis in the simulation software is the most time-consuming part of the whole process, so it is necessary to build a proxy model to approximate the distribution of the simulator. This embodiment builds the proxy model through a neural network model. Specifically, Figure 5 As shown, according to the proxy model establishment task and the target quantization interface, the proxy model is established by using the neural network fitting simulator input to output mapping, including: S41, obtaining the solution space of the proxy model establishment task through the large language model; the solution space is the set of all possible answers generated when processing the target quantization task. S42, uniform sampling is performed in the solution space, and the structural data set of industrial products is constructed through the target quantization interface and the sampling data; the structural data set includes the parameters necessary to generate each industrial product structure. Figure 2Take the ten-bar truss structure in as an example, assuming that the volume and deformation of the structure need to be as large as possible by balancing the horizontal length l of the left half, the width w1 of the left beam and the width w2 of the right beam, then its solution space is 0.05≤l≤0.15, 0.01≤w1, w2≤0.02, and its structural data set is all the ten-bar truss structures that can be constructed in the solution space with parameters l, w1 and w2. S43, use the structural data set to train the neural network to obtain the proxy model. The mapping from simulator input to output is to input the structural parameters of the industrial product into the simulator, and the simulator simulates the output industrial product. The established proxy model is to continuously simulate the process of the simulator simulating the industrial product. Before optimization, the proxy model is automatically established using the neural network through multiple rounds of dialogue, which can reduce the repeated simulation time in the optimization process, greatly reduce the simulation time, and at the same time ensure the model accuracy. Since the simulation results of the simulator may sometimes be very small, the structured dataset is normalized by a large language model before training the neural network to mitigate the impact of the magnitude difference between the targets on the accuracy of the proxy model and ensure the accuracy of the proxy model.

[0035] In order to further ensure the accuracy of the proxy model, a task and target quantization interface is established according to the proxy model. After the neural network is trained, the method also includes: evaluating the convergence curve and fitting effect of the neural network through a large language model; if the convergence curve diverges or the fitting effect is unsatisfactory, adjusting the learning rate and number of training rounds of the neural network to retrain. The quality of the trained neural network is evaluated through a large language model. If the convergence curve is divergent or the fitting effect is not good, it means that the accuracy of the neural network is very poor, and the learning rate and number of training rounds need to be adjusted for retraining until a high-precision proxy model is obtained. Among them, the residual sum of squares method, R² value, residual graph or correlation coefficient r can be used to evaluate whether the fitting effect of the neural network is qualified.

[0036] Finally, execute step S5, according to the optimization initialization task, the target quantization interface and the proxy model interface, optimize the proxy model, obtain the optimization result, and output the optimal design solution of the industrial product according to the optimization result. Only one input is needed to the large language model, and the large language model can automatically complete the design and optimization process of the industrial product, realizing a high degree of automation of the entire industrial design process.

[0037] Further, if Figure 6As shown, according to the optimization initialization task, the target quantization interface and the proxy model interface, the proxy model is optimized, including: S51, by retrieving the enhancement algorithm, selecting the optimization algorithm that matches the optimization initial task in the preset algorithm library; the preset algorithm library can be an algorithm library built by itself, or an existing algorithm library such as Python library, C / C++ open source algorithm library, The Algorithms, etc. S52, the target quantization interface and the proxy model interface are packaged into the objective function of the objective function, and the objective function is optimized by the selected optimization algorithm; wherein, the optimization algorithms include NSGA-II (non-dominated sorting genetic algorithm II), NSWOA (non-dominated sorting whale optimization algorithm), MOPSO (multi-objective particle swarm optimization), and SA (simulated annealing algorithm), etc., and the most suitable algorithm can be selected for optimization according to the actual optimization problem.

[0038] Finally, the optimal solution of target optimization (the optimal solution of a single objective) or the Pareto frontier (a collection of solutions that achieve the best balance between multiple objectives) is obtained. Based on the optimization results, the optimal design solution for industrial products can be automatically obtained. Figure 2 The ten-bar truss structure in FIG. 1 is a multi-objective optimization problem, and the horizontal length l of the left half, the width w1 of the left beam, and the width w2 of the right beam are the optimization variables. The relevant information of the structure is input into the method of this embodiment for automatic design, such as Figure 7 Shown is the automatic simulation result of the structure; Figure 8 is the convergence of the proxy model of the structure; Fig. 9 The multi-objective optimization result of the structure is the Pareto frontier; Figures 7 to 9 The process is fully automated and no human involvement is required.

[0039] In order to ensure the accuracy of target optimization, target optimization is performed on the proxy model according to the optimization initialization task, target quantization interface and proxy model interface. It also includes: after the target function is optimized, the convergence of the optimization result obtained by the target optimization is checked through the large language model; if it is checked that the optimization result has not converged, the hyperparameters in the target optimization process are adjusted and re-optimized to ensure the best optimization result.

[0040] Embodiment 2: like Fig.10As shown, the present invention also provides a hierarchical agent framework suitable for the field of industrial design, which is used to execute an industrial design automation method based on an artificial intelligence large model and an agent as described in Example 1 to solve industrial design automation problems, including: an L0 demand analysis and conceptual design agent, an L1 preliminary design agent, and an L2 detailed design agent; the L0 demand analysis and conceptual design agent is used to extract design optimization problems based on relevant information of industrial products, and decompose the design optimization problems into target quantification tasks, proxy model establishment tasks, and optimization initialization tasks; the L1 preliminary design agent is used to generate an objective function based on the target quantification task, and convert the objective function into a target quantification interface, and according to the proxy model establishment task and the target quantification interface, use a neural network to fit the mapping of the simulator input to the output to establish a proxy model, and convert the proxy model into a proxy model interface; the L2 detailed design agent is used to perform target optimization on the proxy model according to the optimization initialization task, the target quantification interface, and the proxy model interface to obtain an optimization result, and output the optimal design solution of the industrial product according to the optimization result.

[0041] like Fig.10 As shown, the L0-level agent can perform demand analysis and conceptual design. The agent based on the black box multimodal large model communicates with the user to obtain clear requirements. This layer obtains preliminary requirements based on the multimodal capabilities of the large model itself and shallow domain knowledge, performs preliminary conceptual design of the product, and hands it over to the L1-level agent. The L0-level agent uses multimodal understanding of the real world and accurate recognition of long sequences, and uses evolutionary computing to build efficient models, and adopts a variety of talent scales to correspond to different scenarios. The L0 demand analysis and conceptual design agent of this embodiment is the L0-level agent.

[0042] The L1-level agent can perform preliminary design on the product, and impose constraints based on first principles on the results of the conceptual design by integrating the knowledge base and knowledge graph containing professional domain knowledge. By integrating professional domain knowledge, the preliminary design can obtain a product that roughly suits the current needs, reducing the search space for subsequent simulation optimization. At the same time, the L1-level agent also contains the operational knowledge of industrial software, which is the basis for the L2-level agent to perform detailed design. The L1-level agent is a large world model, which is a preliminary design based on the gray box world model. The L1 preliminary design agent of this embodiment is the L1-level agent.

[0043] The L2-level agent can perform detailed design of products based on the white-box knowledge of industrial software, and call industrial software (such as CAD and CAE, etc.) through the code generation capability of the large language model. For example, the conceptual design is first converted into a CAD model, and then the CAE software API is used to perform finite element analysis on the CAD model and the product is optimized at the performance level using the simulation results, and finally the optimally designed industrial product is obtained. The L2-level agent is based on the white-box design iteration of existing tools, which can improve the design and implementation of the agent. The L2 detailed design agent of this embodiment is the L2-level agent.

[0044] The embodiment is only a specific example and does not represent only one implementation mode of the present invention.

[0045] The above description is only the preferred embodiment of the present invention. It is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application belong to the protection scope of the present invention.

Claims

1. An industrial design automation method based on artificial intelligence large model and intelligent agent, characterized in that: The following steps are involved: Extract design optimization problems based on input relevant information of industrial products, and automatically match the large language model based on the design optimization problems; Decomposing the design optimization problem into a target quantification task, a proxy model establishment task and an optimization initialization task through the large language model; Generate an objective function according to the target quantization task, and convert the objective function into a target quantization interface; Establishing a task and target quantization interface according to the proxy model, using a neural network to fit a mapping of simulator input to output to establish a proxy model, and converting the proxy model to a proxy model interface; According to the optimization initialization task, the target quantization interface and the proxy model interface, the proxy model is optimized to obtain an optimization result, and the optimal design solution of the industrial product is output according to the optimization result.

2. The industrial design automation method based on artificial intelligence big model and intelligent agent according to claim 1, characterized in that: The automatically matching the large language model according to the design optimization problem includes: Acquire evaluation information of various different large language models from the historical call records of the large language model; According to the task type of the design optimization problem and the evaluation information of the large language model, the optimal large language model is automatically matched to the design optimization problem.

3. The industrial design automation method based on artificial intelligence big model and intelligent agent according to claim 1, characterized in that: The generating the target function according to the target quantization task includes: calling simulation software to generate the target function according to the complexity of the target quantization task, or directly using the large language model to generate the target function.

4. The industrial design automation method based on artificial intelligence big model and intelligent agent according to claim 3 is characterized in that: The calling simulation software to generate the target function comprises: Using the finite element analysis model in the simulation software to pre-process the target quantification task; According to the pre-processing results, the target function is generated by combining the thinking chain and retrieval enhancement generation and calling the simulator to simulate.

5. The industrial design automation method based on artificial intelligence big model and intelligent agent according to claim 3 is characterized in that: The method further includes: using the large language model as an analyzer to determine the complexity of the target quantization task.

6. The industrial design automation method based on artificial intelligence big model and intelligent agent according to claim 1, characterized in that: The step of establishing a task and target quantization interface according to the proxy model and using a neural network to fit the mapping of simulator input to output to establish the proxy model includes: Acquire a solution space of the proxy model establishment task through the large language model; Perform uniform sampling in the solution space, and construct a structural data set of the industrial product through the target quantization interface and the sampled data; The neural network is trained using the structured data set to obtain the proxy model.

7. The industrial design automation method based on artificial intelligence big model and intelligent agent according to claim 6, characterized in that: After the neural network is trained using the structured data set, the method further includes: evaluating the convergence curve and fitting effect of the neural network through the large language model; if the convergence curve diverges or the fitting effect is unsatisfactory, adjusting the learning rate and the number of training rounds of the neural network through the large language model to retrain.

8. The industrial design automation method based on artificial intelligence big model and intelligent agent according to claim 6, characterized in that: Before training the neural network, the method further includes: normalizing the structured data set using the large language model.

9. The industrial design automation method based on artificial intelligence big model and intelligent agent according to claim 1, characterized in that: The step of performing target optimization on the proxy model according to the optimization initialization task, the target quantization interface and the proxy model interface includes: By searching for enhanced generation, an optimization algorithm matching the initial optimization task is selected from a preset algorithm library; The target quantization interface and the proxy model interface are packaged into the target function of the target function, and the target function is optimized by the selected optimization algorithm.

10. The industrial design automation method based on artificial intelligence big model and intelligent agent according to claim 9, characterized in that: The target optimization of the proxy model according to the optimization initialization task, the target quantization interface and the proxy model interface also includes: after the target function is optimized, the optimization result obtained by the target optimization is checked for convergence by the large language model; if it is checked that the optimization result has not converged, the large language model adjusts the hyperparameters in the target optimization process for re-optimization.

11. A hierarchical agent framework suitable for the field of industrial design, characterized in that: An industrial design automation method based on an artificial intelligence large model and an agent for executing any one of claims 1 to 10 to solve industrial design automation problems, comprising: an L0 demand analysis and conceptual design agent, an L1 preliminary design agent, and an L2 detailed design agent; The L0 demand analysis and concept design agent is used to extract design optimization problems based on relevant information of industrial products, and decompose the design optimization problems into target quantification tasks, agent model establishment tasks and optimization initialization tasks; The L1 preliminary design agent is used to generate an objective function according to the target quantization task, convert the objective function into a target quantization interface, establish a task and a target quantization interface according to the proxy model, use a neural network to fit the mapping of simulator input to output to establish a proxy model, and convert the proxy model into a proxy model interface; The L2 detailed design agent is used to perform target optimization on the proxy model according to the optimization initialization task, target quantization interface and proxy model interface, obtain optimization results, and output the optimal design solution for the industrial product according to the optimization results.

Images