Simulation modeling method and system based on natural language and computer equipment

Abstract

The invention relates to a simulation modeling method and system based on a natural language and computer equipment. Comprising the steps that natural language modeling simulation is achieved according to the technology of a large model, an intelligent entity and a protocol module, and the intelligent entity serves as a communication bridge between the protocol module and the large model. After a user question based on a natural language is obtained through an intelligent entity, a response of a large model to the natural language and communication interaction between the large model and a protocol module are triggered; the related target tool interface provided by the protocol module is called by the intelligent entity, so that the tool calling specification is ensured to be accurate, and the protocol module is enabled to determine and execute the target tool after the request is responded for multiple times. Therefore, through interactive use of the natural language of the large model, the use threshold of the user is reduced, the accessibility of the model is improved, codes do not need to be written or complex graphs do not need to be drawn, and the simulation modeling efficiency is greatly improved.

Description

Technical Field

[0001] This application relates to the field of simulation modeling technology, and in particular to a simulation modeling method, system and computer device based on natural language. Background Technology

[0002] Currently, in technical activities such as power system model building, the mainstream approach heavily relies on specialized programming languages ​​(such as Python, MATLAB, and C++) and specific simulation software (such as PSS / E, PowerFactory, and PSCAD). Technicians need to build models by writing code or operating complex graphical interfaces, and the presentation of simulation models is mostly in the form of code or graphical connections. This significantly increases the technical threshold for model building, and presenting models through code or complex graphical connections is not conducive to model reuse, maintenance, and modification. Furthermore, the process of translating design ideas into code or graphics consumes a significant amount of manpower and time, further impacting overall modeling efficiency. Summary of the Invention

[0003] Based on this, the purpose of this application is to realize a natural language-based simulation modeling method that lowers the technical threshold to improve accessibility, increases modeling efficiency and speed, and enhances model readability, so as to solve the technical problems mentioned above.

[0004] Firstly, this application provides a simulation modeling method based on natural language. Applied to intelligent entities, it includes: The system acquires user questions based on natural language and determines a list of target tools associated with the user questions through a protocol module; the protocol module represents the module that interacts based on the model context protocol. The target parameters associated with the target tool list are determined by the large model, and the execution of each target tool is triggered according to the target parameters to obtain the initial simulation model; The initial simulation model is updated using a large model to obtain the target simulation model.

[0005] In one embodiment, the intelligent entity communicates with the protocol module at least via an HTTP interface, message queue, or remote procedure call; the step of determining the target tool list associated with the user's question through the protocol module includes: parsing the user's question through a large model to determine the function or tool identifier to be queried; and initiating a first request to the protocol module based on the tool identifier, so that the protocol module determines the target tool list according to the first request.

[0006] In one embodiment, the message in the first request includes at least a request type, smart entity identity information, a request timestamp, and tool filtering conditions; the protocol module determines the target tool list based on the first request, including: determining an initial tool list based on the request type and verifying identity based on the smart entity identity information; filtering and formatting the initial tool list based on the request timestamp or the tool filtering conditions to obtain the target tool list; and feeding back the target tool list to the smart entity through the protocol module.

[0007] In one embodiment, triggering the execution of each target tool according to the target parameters to obtain an initial simulation model includes: initiating a second request to the protocol module based on the target parameters, so that the protocol module triggers the execution of each target tool to obtain the execution result of the initial simulation model; the execution result includes the execution status; the above method further includes: performing a new triggering operation on the target tool according to the execution status.

[0008] In one embodiment, triggering a new operation on a target tool based on the execution state includes: traversing the execution states corresponding to each target tool; when any execution state is a non-target state, returning to the process of determining the list of target tools associated with the user's question through the protocol module; when each execution state is a target state, associating the corresponding target tools according to the tool information corresponding to the execution state to obtain a target association result.

[0009] In one embodiment, the method further includes: when obtaining the next user question based on natural language, the protocol module determines the tool list corresponding to the next user question based on the target association result.

[0010] In one embodiment, updating the initial simulation model using a large model to obtain a target simulation model includes: obtaining the execution result of the initial simulation model fed back by the intelligent entity and the knowledge base associated with the intelligent entity using the large model; preprocessing the execution result using the large model and then retrieving and updating it based on the knowledge base to obtain the final result of the target simulation model; the final result is presented in natural language.

[0011] In one embodiment, the process of constructing a knowledge base associated with intelligent entities includes: acquiring various modeling rule data corresponding to various modeling platforms; the modeling rule data includes the type of function interface; segmenting the modeling rule data to obtain multiple fragment data; converting the multiple fragment data into vectors through a text embedding model, and storing the multiple vectors in a vector database to form a knowledge base.

[0012] Secondly, this application also provides a simulation modeling system based on natural language. The system includes intelligent entities, protocol modules, and a large model, wherein the protocol modules represent modules that interact based on model context protocols; wherein: Intelligent entities are used to obtain user queries based on natural language and trigger a large model to parse them to obtain the tool identifier to be queried. A smart entity is used to initiate a first request to the protocol module based on the tool identifier; The protocol module is used to determine the target tool list based on the first request and feed the target tool list back to the intelligent entity; A large model is used to determine the target parameters associated with the target tool list when obtaining the target tool list fed back by the intelligent entity, and to feed back the target parameters to the intelligent entity; A smart entity is used to initiate a second request to the protocol module based on the target parameters; The protocol module is used to trigger the execution of each target tool according to the second request, and obtain the execution result of the initial simulation model; A large model is used to update the execution results of the initial simulation model sent by the intelligent entity to obtain the final result of the target simulation model.

[0013] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the above-described natural language-based simulation modeling method.

[0014] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-described natural language-based simulation modeling method.

[0015] The aforementioned natural language-based simulation modeling method, system, computer equipment, and readable storage medium, implemented using large-scale models, intelligent entities, and protocol modules, utilizes intelligent entities as a communication bridge between the protocol modules and the large-scale model. After acquiring user queries in natural language through the intelligent entities, the large-scale model responds to the natural language and engages in communication and interaction with the protocol modules. The protocol modules provide interfaces for relevant target tools that the intelligent entities can invoke, ensuring accurate and standardized tool calls. This allows the protocol modules to determine and execute the target tool after responding to multiple requests. Therefore, the interactive use of natural language within the large-scale model lowers the user barrier and improves model accessibility, eliminating the need for writing code or drawing complex graphics, thus significantly improving the efficiency of simulation modeling.

[0016] In addition, the large model will update and optimize the execution results of the initial simulation model, and finally present them to the user in the form of natural language, which enhances the readability and understandability of the simulation model. Attached Figure Description

[0017] Figure 1 This is a communication principle diagram of a natural language-based simulation modeling system in one embodiment; Figure 2 This is a flowchart illustrating a natural language-based simulation modeling method in one embodiment. Figure 3 This is a schematic diagram of the process of intelligent entity invocation in one embodiment; Figure 4 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] Given the current deep reliance on specialized programming languages ​​and specific simulation software, which requires users to have a strong programming foundation and proficiency in the complex modeling syntax and graphical interface operation of related software, this application designs a simulation modeling system based on natural language, including intelligent entities, protocol modules, and a large model. The protocol modules represent modules that interact based on the model context protocol.

[0020] The system comprises: an intelligent entity, which acquires user queries based on natural language and triggers a large model to parse them to obtain the identifier of the tool to be queried; an intelligent entity, which initiates a first request to the protocol module based on the tool identifier; a protocol module, which determines the target tool list based on the first request and feeds the target tool list back to the intelligent entity; a large model, which determines the target parameters associated with the target tool list when acquiring the target tool list fed back by the intelligent entity and feeds the target parameters back to the intelligent entity; an intelligent entity, which initiates a second request to the protocol module based on the target parameters; a protocol module, which triggers the execution of each target tool based on the second request to obtain the execution result of the initial simulation model; and a large model, which updates the execution result of the initial simulation model sent by the intelligent entity to obtain the final result of the target simulation model.

[0021] like Figure 1 As shown, Figure 1This is a communication principle diagram for a natural language-based simulation modeling system. The intelligent entity shown is the Agent, which is an entity capable of perceiving its environment and making autonomous decisions. It can be a software program or a hardware device, possessing autonomy, environmental awareness, and decision-making execution capabilities. Based on different user needs, Agents can be categorized as single Agents, multiple Agents, and Agents that interact with humans. The large model shown is the LLM (Large Language Model), an artificial intelligence model designed to understand and generate human language. The protocol module shown is the MCPserver (Model Context Protocol), an open standard protocol that defines how applications and large models exchange context information. It acts like a "universal plug" or "USB interface" in the field of intelligent interaction, enabling seamless integration between the artificial intelligence model and external resources through standardized specifications. The target tool is the Tools shown.

[0022] In one embodiment, the natural language-based simulation modeling system can be deployed on a server, which can be implemented using a standalone server or a server cluster consisting of multiple servers. Alternatively, it can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0023] In one embodiment, such as Figure 2 As shown, a natural language-based simulation modeling method is provided, applied to intelligent entities, through... Figure 1 This is achieved through a natural language-based simulation modeling system, including the following steps: Step 202: Obtain user questions based on natural language and determine the list of target tools associated with the user questions through the protocol module.

[0024] Among them, the protocol module represents the module that interacts based on the model context protocol.

[0025] Specifically, such as Figure 3 As shown, Figure 3This is a flowchart illustrating the process of an intelligent entity invoking a function. When a user sends a natural language-based user question to the intelligent entity (Agent), such as inquiring about model details ("Which generators are connected to bus B5?") or requesting an explanation of simulation results ("Why did line L1-2 become overloaded after the fault?"), the large model (LLM model) is triggered to respond to the natural language. The LLM model analyzes the user question and determines whether a function or tool identifier needs to be invoked. When a tool invocation is required, communication interaction between the protocol module and the intelligent entity is triggered, allowing the protocol module (MCP) to determine the list of target tools associated with the user question. This list of target tools includes various target tools required for simulation modeling.

[0026] It is easy to understand that describing model concepts and structures in natural language is much faster than writing code or drawing complex graphics. It is especially suitable for early design, proof of concept and rapid iteration of simulation models. Technicians can directly use descriptive language to express system structure, equipment characteristics and control logic.

[0027] In one embodiment, when the large model responds to natural language, it packages user questions into user prompts. A user prompt is content directly entered by the user to ask a question or provide contextual information, guiding the large model to generate a specific answer. It is the primary way for users to interact with the large model, typically involving specific tasks or questions.

[0028] Step 204: Determine the target parameters associated with the target tool list through the large model, and trigger the execution of each target tool according to the target parameters to obtain the initial simulation model. Specifically, refer to Figure 3 As shown, the intelligent entity sends the list of target tools determined by the protocol module to the large model. After responding to the user's question, the large model generates tool invocation instructions to invoke the required target parameters based on the function descriptions in the target tools. The target parameters for function calls are generally built-in and do not need to be generated by the large model. Next, the intelligent entity invokes the application program, executing the actual function calls through the backend target tool interface and target parameters, and obtaining the execution results of the initial simulation model.

[0029] In one embodiment, when the intelligent entity obtains the target parameters determined by the large model, it initiates an execution request for the corresponding target tool to the protocol module in order to obtain the execution result of the initial simulation model.

[0030] Step 206: Update the initial simulation model using the large model to obtain the target simulation model.

[0031] Specifically, the intelligent entity feeds back the execution results of the initial simulation model to the larger model. The larger model then updates the execution results of the initial simulation model based on the knowledge base associated with the intelligent entity, further optimizing its responses to user questions to make them more fluent and conform to human communication habits, and generating the final result of the target simulation model. Finally, the intelligent entity presents the final result fed back by the larger model to the user in natural language. The knowledge base is determined based on various modeling rule data corresponding to different modeling platforms, such as Simulink.

[0032] In the aforementioned natural language-based simulation modeling method, natural language modeling and simulation are implemented using a large model, intelligent entities, and protocol modules. The intelligent entity acts as a communication bridge between the protocol module and the large model. After the intelligent entity receives user queries in natural language, it triggers the large model's response and communication interaction with the protocol module. The protocol module provides relevant target tool interfaces for the intelligent entity to call, ensuring accurate and standardized tool calls. This allows the protocol module to determine and execute the target tool after responding to multiple requests. Therefore, the interactive use of natural language within the large model lowers the user barrier and improves model accessibility. Furthermore, it eliminates the need for writing code or drawing complex graphics, significantly improving the efficiency of simulation modeling. In addition, the large model updates and optimizes the execution results of the initial simulation model, finally presenting them to the user in natural language, enhancing the readability and understandability of the simulation model.

[0033] In one embodiment, a natural language-based simulation modeling method is provided, referencing... Figure 1 As shown, the process includes the following steps: Step 1: The user sends a user question in natural language to the intelligent entity. Step 2: The intelligent entity initiates a first request to the protocol module to determine the target tool list corresponding to the user question, and the protocol module feeds back the target tool list to the intelligent entity. Step 3: The intelligent entity sends the target tool list to the large model. Step 4: The large model determines the target parameters associated with the target tool list and feeds back the target parameters to the intelligent entity. Step 5: The intelligent entity initiates a second request to the protocol module based on the target parameters, so that the protocol module triggers the execution of each target tool and the intelligent entity receives the execution results of the initial simulation model. Step 6: The intelligent entity feeds back the execution results of the initial simulation model to the large model. Step 7: The large model updates the execution results of the initial simulation model to obtain the final result of the target simulation model and feeds back the final result to the intelligent entity. Step 8: The intelligent entity presents the final result to the user in natural language.

[0034] In one embodiment, determining the list of target tools associated with a user's question through a protocol module includes: parsing the user's question using a large model to determine the function or tool identifier to be queried; and initiating a first request to the protocol module based on the tool identifier, so that the protocol module determines the list of target tools according to the first request.

[0035] The intelligent entity and the protocol module communicate at least through HTTP, REST, message queues, or remote calls. The first request represents a request to obtain the list of target tools.

[0036] Specifically, when the large model parses the function or tool identifier requested by the user, the intelligent entity generates a structured request message and initiates a first request to the protocol module to obtain the list of target tools (tools list) based on the tool identifier. The protocol module performs authentication, queries the tool registry, filtering, and formatting operations on the first request, and then returns the list of matching target tools to the intelligent entity. Each tool in the target tool list may include the tool name, a description of the tool's function, the parameters required to invoke the tool, and other metadata.

[0037] In this embodiment, the function or tool identifier is determined by parsing user questions through a large model, and the target tool list is accurately matched by the protocol module, eliminating the tedious manual tool selection and improving the accuracy and efficiency of tool invocation.

[0038] In one embodiment, the protocol module determines the target tool list based on the first request, including: determining an initial tool list based on the request type and verifying identity based on the smart entity's identity information; filtering and formatting the initial tool list based on the request timestamp or tool filtering conditions to obtain the target tool list; and feeding back the target tool list to the smart entity through the protocol module.

[0039] The messages in the first request include, but are not limited to, request type, smart entity identity information, request timestamp, and tool filtering conditions. For example, the request type is "Get Tool List" (GET_TOOLS), and the smart entity identity information could be an authentication token (API Key).

[0040] Specifically, the protocol module first queries the tool registry based on the request type to determine an initial tool list, and then performs identity verification based on the smart entity's identity information. Once verification is successful, the initial tool list is filtered and formatted based on information such as the request timestamp or tool filtering conditions to obtain the target tool list, which is then fed back to the smart entity. Therefore, by having the protocol module determine the initial tool list based on the request type, and then format the target tool list through identity verification, timestamps, or conditional filtering, the security and accuracy of tool invocation are ensured.

[0041] In one embodiment, triggering the execution of each target tool based on target parameters to obtain an initial simulation model includes: initiating a second request to the protocol module based on the target parameters, so that the protocol module triggers the execution of each target tool to obtain the execution result of the initial simulation model.

[0042] The execution result includes at least the request ID, execution duration, execution data, and execution status. The second request represents the execution request of the target tool.

[0043] Specifically, when the intelligent entity obtains the target parameters determined by the large model, it initiates an execution request for the corresponding target tool to the protocol module. At this point, the core task of the protocol module is to safely and efficiently schedule and execute the target tool. The execution process of each target tool is also the running process of the initial simulation model, and the final execution result will be fed back to the intelligent entity. It's easy to understand that when considering the initial simulation model without taking into account factors such as simulation duration and data, and only focusing on whether the model was successfully built, the execution status in the execution result becomes the key factor. Therefore, it is also necessary to dynamically update the acquisition and execution status of different target tools based on the execution status, that is, to initially adjust the construction process of the initial simulation model.

[0044] In this embodiment, the target tool is securely and efficiently scheduled and executed through the protocol module, and the results are fed back to the intelligent entity. Combined with the dynamic adjustment of the execution status tool acquisition and execution, the initial simulation model is successfully constructed, and the reliability and efficiency of subsequent modeling are improved.

[0045] In one embodiment, the method further includes: triggering a new operation on the target tool based on the execution status.

[0046] In one embodiment, triggering a new operation on a target tool based on its execution state includes: traversing the execution states corresponding to each target tool; when any execution state is a non-target state, returning to the process of determining the list of target tools associated with the user's question through the protocol module; and when each execution state is a target state, associating the corresponding target tools based on the tool information corresponding to the execution state to obtain the target association result.

[0047] The target status includes executed, while the non-target status includes modified and paused, not executed, etc.

[0048] Specifically, traversing the execution state of each target tool can be understood as the process of all target tools responding to the "execution" trigger operation in parallel, and then determining whether the target tool list needs to be updated. When the intelligent entity determines that the execution state is in the non-target state of "not executed," there may be a target tool execution error caused by an abnormal request from the intelligent entity. This can be understood as the target tool being executed in parallel. At this time, it is necessary to re-determine a new target tool list associated with the user's query. When the execution state is determined to be in the non-target state of "modification paused," it indicates that the user's requirement is temporary maintenance or updating of the model description, such as changing parameters, adding devices, or adjusting control logic. For example, "In the existing circuit system, increase the resistance of all lines by 10% to simulate aging." When there are target tools in the existing circuit system's target tool list with the execution state of "modification paused," the large model automatically finds all electrical components containing resistance and generates relevant functions or new target tools. Finally, the original target tool list and the new target tools are executed together until the execution result of the new initial simulation model is obtained. When each execution state is a target state, the intelligent entity will associate its corresponding target tool with the tool information corresponding to the execution state to obtain the target association result.

[0049] In this embodiment, by traversing the execution state of the target tool, tools in non-target states are backtracked and adjusted in a timely manner to ensure the effectiveness and accuracy of tool calls; after all tools have reached the target state, they are associated to form a reliable target association result, which not only ensures the quality of tool collaboration, but also provides a stable foundation for subsequent interactions.

[0050] In one embodiment, the method further includes: when obtaining the next user question based on natural language, the protocol module determines the tool list corresponding to the next user question based on the target association result.

[0051] When the next natural language question arrives, the protocol module determines the corresponding tool list based on the target association results. This allows tool calls to fit the context logic more closely, improves the consistency and accuracy of tool matching, reduces redundant operations, and thus efficiently supports the construction of subsequent simulation models and optimizes the user interaction experience.

[0052] In one embodiment, when a target tool responds to the "execute" trigger operation serially, if any currently executing target tool is in a non-target state, the process of determining the list of target tools associated with the user's question through the protocol module is restarted.

[0053] In one embodiment, updating the initial simulation model using a large model to obtain the target simulation model includes: obtaining the execution result of the initial simulation model fed back by the intelligent entity and the knowledge base associated with the intelligent entity through the large model; and after preprocessing the execution result through the large model, retrieving and updating it based on the knowledge base to obtain the final result of the target simulation model.

[0054] The final results are presented in natural language.

[0055] Specifically, large-scale models possess excellent capabilities in processing natural language, but the key to enabling them to proactively complete relevant tasks through natural language lies in "how to make natural language understand the modeling rules and how to call tools to build the model." This is addressed by combining a pre-built knowledge base of intelligent entities with the large-scale model. After the large-scale model receives the execution results of the initial simulation model from the intelligent entities, the results can be pre-processed, including cleaning and denoising. Then, the knowledge base retrieval module searches for relevant knowledge in the knowledge base, and the retrieved knowledge is passed to the large-scale model for natural language processing, making it more fluent and understandable, ultimately yielding the final result of the target simulation model.

[0056] Furthermore, since large models cannot directly perceive external tools or interfaces, they need to be implemented through intelligent entities, including task understanding, tool selection, parameter passing, and result processing. This solves the problem of how to call tools to build models. That is, the large model generates modeling code in natural language (such as MATLAB, Python, etc.), and then the intelligent entity starts the simulation modeling software and executes the code generated by the large model to complete the model building and simulation.

[0057] In this embodiment, by using a large model to process the initial simulation results and combining them with the retrieval and updating of the associated knowledge base, the accuracy and adaptability of the target simulation model results can be improved, the output effect can be optimized, and the practicality of the model can be enhanced.

[0058] In one embodiment, the process of constructing a knowledge base associated with intelligent entities includes: acquiring various modeling rule data corresponding to various modeling platforms; segmenting the modeling rule data to obtain multiple fragment data; converting the multiple fragment data into vectors through a text embedding model, and storing the multiple vectors in a vector database to form a knowledge base.

[0059] The modeling rule data includes the type of function interface.

[0060] Specifically, the knowledge base can be built on other intelligent entities such as Dify, Coze, and FastGPT, collecting relevant modeling rule data from Simulink or other various modeling platforms. The collected data is then cleaned to remove noise, duplicates, and irrelevant information, ensuring data quality and accuracy. The cleaned data is then used to construct the knowledge base on the intelligent entity. This process involves segmenting the text of the modeling rule data into smaller chunks, using a text embedding model (such as GLM) to convert these chunks into vectors, and storing these vectors in a vector database (such as FAISS or Milvus) to ultimately form the knowledge base.

[0061] In this embodiment, by integrating multi-platform modeling rules, the segmented rules are converted into vectors by a text embedding model and stored in a vector database to form a knowledge base. This facilitates quick and accurate retrieval of rules, aids cross-platform modeling, and improves rule reusability and modeling standardization.

[0062] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0063] Based on the same inventive concept, this application also provides a natural language-based simulation modeling apparatus for implementing the above-mentioned natural language-based simulation modeling method. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations of one or more embodiments of the natural language-based simulation modeling apparatus provided below can be found in the limitations of the natural language-based simulation modeling method described above, and will not be repeated here.

[0064] In one embodiment, a natural language-based simulation modeling apparatus is provided, comprising: a target tool determination module, a target parameter determination module, and a model update module, wherein: The target tool determination module is used to acquire user questions based on natural language and determine the list of target tools associated with the user questions through the protocol module.

[0065] The target parameter determination module is used to determine the target parameters associated with the target tool list through the large model, and to trigger the execution of each target tool according to the target parameters to obtain the initial simulation model.

[0066] The model update module is used to update the initial simulation model using a larger model to obtain the target simulation model.

[0067] The modules in the above-mentioned natural language-based simulation modeling can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0068] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 4 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database is used for target parameters. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a natural language-based simulation modeling method.

[0069] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0070] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0071] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0072] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps in the above method embodiments.

[0073] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0074] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0075] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A simulation modeling method based on natural language, characterized in that, Applied to smart entities, the method includes: The system acquires user questions based on natural language and determines a list of target tools associated with the user questions through a protocol module; the protocol module represents the module that interacts based on the model context protocol. The target parameters associated with the target tool list are determined by the large model, and the execution of each target tool is triggered according to the target parameters to obtain the initial simulation model; The initial simulation model is updated using a large model to obtain the target simulation model.

2. The method according to claim 1, characterized in that, The intelligent entity and the protocol module communicate at least through an HTTP interface, message queue, or remote call. The step of determining the list of target tools associated with the user's question through the protocol module includes: The user's query is analyzed using a large model to determine the function or tool identifier to be queried; Based on the tool identifier, a first request is sent to the protocol module so that the protocol module can determine the target tool list according to the first request.

3. The method according to claim 2, characterized in that, The message in the first request includes at least the request type, smart entity identity information, request timestamp, and tool filtering conditions; The protocol module determines the target tool list based on the first request, including: The protocol module determines an initial tool list based on the request type and performs identity verification based on the smart entity's identity information. The protocol module filters and formats the initial tool list based on the request timestamp or the tool filtering conditions to obtain the target tool list. The list of target tools is fed back to the intelligent entity through the protocol module.

4. The method according to claim 1, characterized in that, The step of triggering the execution of each target tool according to the target parameters to obtain the initial simulation model includes: Based on the target parameters, a second request is sent to the protocol module to trigger the execution of each target tool, thereby obtaining the execution result of the initial simulation model; the execution result includes the execution status; The method further includes: performing a new triggering operation on the target tool based on the execution state.

5. The method according to claim 4, characterized in that, Based on the execution status, perform new triggering operations on the target tool, including: The execution state of each target tool is traversed. When any execution state is a non-target state, the process returns to the process of determining the list of target tools associated with the user's question through the protocol module. When each execution state is a target state, the corresponding target tools are associated according to the tool information corresponding to the execution state to obtain the target association result; The method further includes: when obtaining the next user question based on natural language, the protocol module determines the tool list corresponding to the next user question based on the target association result.

6. The method according to claim 1, characterized in that, The process of updating the initial simulation model using a large model to obtain the target simulation model includes: The large model is used to obtain the execution results of the initial simulation model fed back by the intelligent entity, as well as the knowledge base associated with the intelligent entity. After preprocessing the execution results using a large model, the results are retrieved and updated based on a knowledge base to obtain the final result of the target simulation model; the final result is presented in natural language.

7. The method according to claim 6, characterized in that, The process of constructing the knowledge base associated with the intelligent entities includes: Acquire various modeling rule data corresponding to different modeling platforms; the modeling rule data includes the type of function interface. The modeling rule data is segmented to obtain multiple data fragments; Multiple data fragments are converted into vectors using a text embedding model, and these vectors are stored in a vector database to form a knowledge base.

8. A simulation modeling system based on natural language, characterized in that, The system includes intelligent entities, protocol modules, and a large model. The protocol modules represent modules that interact based on model context protocols; wherein: Intelligent entities are used to obtain user queries based on natural language and trigger a large model to parse them to obtain the tool identifier to be queried. A smart entity is used to initiate a first request to the protocol module based on the tool identifier; The protocol module is used to determine the target tool list based on the first request and feed the target tool list back to the intelligent entity; A large model is used to determine the target parameters associated with the target tool list when obtaining the target tool list fed back by the intelligent entity, and to feed back the target parameters to the intelligent entity; A smart entity is used to initiate a second request to the protocol module based on the target parameters; The protocol module is used to trigger the execution of each target tool according to the second request, and obtain the execution result of the initial simulation model; A large model is used to update the execution results of the initial simulation model sent by the intelligent entity to obtain the final result of the target simulation model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images