Simulation method, system and application of material world

Abstract

The invention discloses a substance world simulation method and system and application, relates to the technical field of substance world simulation, and solves the technical problems that the realized function is relatively single, only functions under a specific task can be realized, and a powerful universal scheme is lacked, so that the efficiency of the substance world simulation method is relatively low. Analyzing the input data through an LLM model to obtain a model selection instruction and a task execution instruction; generating an optimal model and a model execution plan according to the model selection instruction and the RAG technology; and the final result is output according to the task execution instruction and the optimal model, and the efficiency and accuracy of material world simulation are improved through highly automatic and intelligent model selection, flexible task processing and advanced data processing.

Description

Technical Field

[0001] The present application relates to simulation technology of the material world, and in particular to a material simulation method and system based on multi-scale simulation and high-performance computing, and its application in material science, drug design and chemical engineering. Background Art

[0002] The Material World Simulator is a computational tool used to simulate and study various phenomena and processes in the material world. It is widely used in materials science, chemistry, physics and other fields to help scientists understand the properties of materials, predict the behavior of materials, and design and optimize new materials. Large Language Models (LLMs) have made remarkable achievements in the field of natural language processing and have powerful language understanding and generation capabilities. Retrieval-augmented generation (RAG) technology further improves the performance of LLMs in dealing with open-world problems by combining external search engines.

[0003] With the rapid development of materials science, the demand for the research and development of new materials is becoming increasingly urgent. However, in order to obtain the properties of existing materials, traditional experimental methods are time-consuming, costly, and difficult to fully cover all possible material combinations and conditions; existing material world simulation methods often rely on specific physical or chemical models, which have limited applicability and are difficult to handle complex and variable material systems. At the same time, the functions implemented by existing material world simulation solutions are relatively simple and can only achieve functions under a specific task. There is a lack of powerful general solutions, resulting in low efficiency of material world simulation methods; therefore, further improvements are still needed for material world simulation methods. Summary of the invention

[0004] The present application aims to solve at least one of the technical problems existing in the prior art; to this end, the present application proposes a simulation method, system and application of the material world, which are used to solve the technical problem that the functions implemented by the prior art are relatively single, can only realize functions under a specific task, lack of powerful general solutions, and lead to low efficiency of the material world simulation method.

[0005] To achieve the above-mentioned purpose, the first aspect of the present application provides a simulation system of the material world, comprising: a data acquisition module, a data analysis module, an early warning module and a database;

[0006] The data acquisition module is used to acquire input data through a data acquisition device; the input data is text data;

[0007] The data analysis module: parses the input data through the LLM model to obtain model selection instructions and task execution instructions; generates the most suitable model and model execution plan according to the model selection instructions and RAG technology; outputs the final result according to the task execution instructions and the most suitable model;

[0008] The early warning module: makes prompts according to the alarm signal and contacts relevant personnel.

[0009] Furthermore, the step of parsing the input data through the LLM model to obtain the model selection instruction and the task execution instruction includes:

[0010] Step 1: Get input data;

[0011] Step 2: Parse the input data through the NLP algorithm to obtain word decomposition and part-of-speech tagging; then perform syntactic analysis on it to obtain a dependency syntax tree; perform named entity recognition on the input data to obtain marked entity data; use intent recognition and slot filling technology to obtain task parameters for the marked entity data and dependency syntax tree;

[0012] Step 3: Generate model selection instructions and task execution instructions through task parameters.

[0013] Through the above steps, the present application greatly reduces the possibility of human intervention, so that the present invention can be widely applied to different research fields and task types, and improves the flexibility and versatility of the system.

[0014] Furthermore, the generating of the most suitable model and model execution plan according to the model selection instruction and the RAG technology includes:

[0015] Get the retrieved data in the model selection command;

[0016] Searching the model database according to the search data to obtain a model search result, and obtaining a model selection suggestion by generating a model according to the model search result;

[0017] The selected model in the model selection suggestion is taken as the most suitable model;

[0018] According to the search data, a search is performed in the literature and database to obtain the execution search results, and according to the execution search results, a model execution plan is obtained by generating a model.

[0019] Furthermore, the outputting of the final result according to the task execution instruction and the optimal model includes:

[0020] Obtaining task execution instructions, an optimal model, and a model execution plan; the model execution plan includes initialization parameters and operating parameters;

[0021] Initialize the optimal model by initializing parameters;

[0022] Achieve operation simulation of the most suitable model through operation parameters;

[0023] Generate alarm signals based on equipment operating environment data;

[0024] The results after running the simulation are output to get the final result.

[0025] Further, generating an alarm signal according to the equipment operating environment data includes:

[0026] Obtain the operating temperature in the equipment operating environment data at intervals of t;

[0027] When the operating temperature is higher than the temperature threshold for N consecutive times, a high temperature alarm signal of the equipment is generated;

[0028] Otherwise, when the continuous When the operating temperatures are all higher than the temperature threshold, a high temperature warning signal of the equipment is generated; otherwise, no operation is performed; where D is the proportional coefficient, D∈(0,1).

[0029] A second aspect of the present invention provides a method for simulating a material world, comprising:

[0030] S0: Get input data;

[0031] S1: Analyze the input data through the LLM model to obtain model selection instructions and task execution instructions;

[0032] S2: Generate the most suitable model and model execution plan based on model selection instructions and RAG technology;

[0033] S3: Output the final result according to the task execution instructions and the most suitable model;

[0034] S4: Provide prompts based on the alarm signal and contact relevant personnel.

[0035] Another aspect of the present invention provides a simulation application of the physical world for use in materials science, drug design, and chemical engineering.

[0036] Compared with the prior art, the beneficial effects of this application are:

[0037] 1. This application parses the input data through the LLM model to obtain model selection instructions and task execution instructions; generates the most suitable model and model execution plan based on the model selection instructions and RAG technology; outputs the final result based on the task execution instructions and the most suitable model, and improves the efficiency and accuracy of the simulation of the material world through highly automated and intelligent model selection, flexible task processing and advanced data processing.

[0038] 2. This application generates the most suitable model and model execution plan through model selection instructions and RAG technology. Automated model selection significantly reduces the time and error risk of manual operation, and improves the efficiency and reliability of simulation tasks. At the same time, the combination of RAG technology ensures the intelligence and accuracy of model selection, and generates a detailed task execution plan, which improves the accuracy and scientificity of simulation tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0040] Figure 1 A schematic diagram of the principle of a simulation system of the material world of the present application;

[0041] Figure 2 This is a flow chart of a method for simulating the material world of this application. DETAILED DESCRIPTION

[0042] The technical solution of the present application will be described clearly and completely in conjunction with the embodiments below. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.

[0043] See also Figure 1 , the first aspect of the present application provides a simulation system of the material world, including: a data acquisition module, a data analysis module, an early warning module and a database;

[0044] Data acquisition module: obtain input data through data acquisition equipment; the input data is text data; the data acquisition equipment includes various sensors, etc.;

[0045] Data analysis module: Analyze the input data through the LLM model to obtain model selection instructions and task execution instructions. Model selection instructions are instructions for selecting models, and task execution instructions are instructions for executing tasks; Generate the most suitable model and model execution plan based on the model selection instructions and RAG technology. The most suitable model is an optimal model selected by the system based on the model selection instructions, and the model execution plan is a plan for executing the most suitable model; Output the final result based on the task execution instructions and the most suitable model; The final result includes the result content and result format;

[0046] Early warning module: Prompts are given based on alarm signals and relevant personnel are contacted; alarm signals include equipment high temperature alarm signals and equipment high temperature early warning signals, etc.

[0047] Through the above steps, this embodiment achieves highly automated and intelligent model selection, flexible task processing and advanced data processing, thereby improving the efficiency and accuracy of the simulation of the material world.

[0048] In this embodiment, the input data is parsed by the LLM model to obtain the model selection instruction and the task execution instruction, including:

[0049] Step 1: Get input data; the input data is a natural language;

[0050] Step 2: Parse the input data through the NLP algorithm to obtain word decomposition and part-of-speech tagging; then perform syntactic analysis on it to obtain a dependency syntax tree; perform named entity recognition on the input data to obtain marked entity data; obtain task parameters through intent recognition and slot filling technology for the marked entity data and the dependency syntax tree; in this embodiment, word segmentation and part-of-speech tagging, syntactic analysis, named entity recognition, intent recognition and slot filling are implemented through the NLP algorithm; for example, in this embodiment, the input data is set to "Predict the thermal conductivity of a new composite material using molecular dynamics", and the task parameters obtained after being processed by the NLP algorithm are {task: Predict; property: thermal conductivity; material: composite material; model: molecular dynamics};

[0051] In another embodiment, word segmentation and part-of-speech tagging can be performed by using a rule-based method, such as regular expressions, or other open source NLP toolkits, such as Stanford NLP or SpaCy; syntactic analysis can be performed by using a parsing method based on phrase structure grammar or a transformational grammar; named entity recognition can be performed by using a dictionary matching and rule-based method or other pre-trained models such as XLNet; intent recognition and slot filling can be performed by using a template matching-based method or a traditional machine learning method such as a support vector machine.

[0052] Step 3: Generate model selection instructions and task execution instructions through task parameters.

[0053] This embodiment performs various data processing on the input data through the NLP algorithm, and finally obtains model selection instructions and task execution instructions; through the application of NLP technology, it can automatically and accurately parse complex natural language prompts, generate clear model selection instructions and task execution instructions, greatly reduce human intervention, and improve work efficiency and accuracy.

[0054] In this embodiment, the most suitable model and model execution plan are generated according to the model selection instruction and the RAG technology, including:

[0055] Get the retrieved data in the model selection command;

[0056] Searching the model database according to the search data to obtain a model search result, and obtaining a model selection suggestion by generating a model according to the model search result;

[0057] The selected model in the model selection suggestion is used as the optimal model; in this embodiment, the optimal model obtained through the model selection suggestion is the LAMMPS model;

[0058] According to the search data, the search is performed in the literature and database to obtain the execution search results, and the model execution plan is obtained by generating the model according to the execution search results; this embodiment uses RAG technology to perform information retrieval and generate models; in another embodiment, information retrieval can be performed by TF-IDF method or BM25 algorithm; Seq2Seq model is used to generate specific model selection suggestions and model execution plans;

[0059] In this embodiment, the final result is output according to the task execution instruction and the optimal model, including:

[0060] Obtain task execution instructions, optimal model, and model execution plan; the model execution plan includes initialization parameters and running parameters;

[0061] The optimal model is initialized by initializing the parameters. In this embodiment, the LAMMPS model is initialized with a composite material structure and the simulation temperature is set to 300K.

[0062] The operation simulation of the optimal model is realized through the operation parameters; in this embodiment, the simulation time in the operation parameters is set to 10ns;

[0063] Generate alarm signals based on equipment operating environment data;

[0064] The results after running the simulation are output to get the final result.

[0065] During the task execution phase, this embodiment utilizes high-performance computing resources, parallel computing technology, and integrated GPU acceleration technology to process large-scale computing tasks, which significantly shortens the calculation time and improves simulation efficiency and accuracy.

[0066] In this embodiment, generating an alarm signal according to the device operating environment data includes:

[0067] Obtain the operating temperature in the equipment operating environment data at intervals of t;

[0068] When the operating temperature is higher than the temperature threshold for N consecutive times, a high temperature alarm signal is generated; the value of N is set based on experience;

[0069] Otherwise, when the continuous When the operating temperature of each time is higher than the temperature threshold, a high temperature warning signal of the equipment is generated; otherwise, no operation is performed; wherein D is a proportional coefficient, D∈(0,1), and the specific value is set according to experience. In this embodiment, D is set to 0.7; The symbol for rounding up is used to ensure that the number is an integer.

[0070] In other embodiments, the input data is set to "discover a metal-organic framework material (MOF) with high electronic conductivity"; the final result obtained by the simulation system of the material world is "result content: band structure, state density, and conductivity prediction of MOF materials; result format: band diagram, state density diagram, and conductivity analysis report";

[0071] In another embodiment, the input data is set to "simulate the catalytic reaction path and analyze the transition state of the reaction"; the final result obtained through the simulation system of the material world is "result content: reaction path diagram, energy position and energy barrier of the transition state; result format: visualization diagram of the reaction path, transition state analysis report and reaction energy barrier analysis report".

[0072] See also Figure 2 The second aspect of the present application provides a method for simulating a material world, including:

[0073] S0: Get input data;

[0074] S1: Analyze the input data through the LLM model to obtain model selection instructions and task execution instructions;

[0075] S2: Generate the most suitable model and model execution plan based on model selection instructions and RAG technology;

[0076] S3: Output the final result according to the task execution instructions and the most suitable model;

[0077] S4: Provide prompts based on the alarm signal and contact relevant personnel.

[0078] Another aspect of the present application is to provide a simulation application of the material world, which is applied to fields such as material science, drug design, and chemical engineering.

[0079] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0080] The working principle of this application is as follows: by obtaining input data; parsing the input data through the LLM model to obtain model selection instructions and task execution instructions; generating the most suitable model and model execution plan according to the model selection instructions and RAG technology; outputting the final result according to the task execution instructions and the most suitable model; making prompts according to the alarm signal and contacting relevant personnel. Through highly automated and intelligent model selection, flexible task processing and advanced data processing, the efficiency and accuracy of the simulation of the material world are improved, and the problem of low efficiency of the simulation method of the material world is avoided, that is, the functions realized by the existing technology are relatively single and can only realize the functions under one specific task, and there is a lack of powerful general solutions.

[0081] The above embodiments are only used to illustrate the technical method of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical method of the present application may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present application.

Claims

1. A simulation system of the material world, characterized in that: include: Data collection module, data analysis module, early warning module and database; The data acquisition module is used to acquire input data through a data acquisition device; the input data is text data; The data analysis module: analyzes the input data through the LLM model to obtain model selection instructions and task execution instructions; generates the most suitable model and model execution plan according to the model selection instructions and RAG technology; Output the final result according to the task execution instructions and the most suitable model; The early warning module: makes prompts according to the alarm signal and contacts relevant personnel.

2. A simulation system of the material world according to claim 1, characterized in that: The step of parsing the input data through the LLM model to obtain the model selection instruction and the task execution instruction includes: Step 1: Get input data; Step 2: Parse the input data through the NLP algorithm to obtain word decomposition and part-of-speech tagging; then perform syntactic analysis on it to obtain a dependency syntax tree; perform named entity recognition on the input data to obtain marked entity data; use intent recognition and slot filling technology to obtain task parameters for the marked entity data and dependency syntax tree; Step 3: Generate model selection instructions and task execution instructions through task parameters.

3. A simulation system of the material world according to claim 1, characterized in that: The generating of the most suitable model and model execution plan according to the model selection instruction and the RAG technology includes: Get the retrieved data in the model selection command; Searching the model database according to the search data to obtain a model search result, and obtaining a model selection suggestion by generating a model according to the model search result; The selected model in the model selection suggestion is taken as the most suitable model; According to the search data, a search is performed in the literature and database to obtain the execution search results, and according to the execution search results, a model execution plan is obtained by generating a model.

4. A simulation system of the material world according to claim 1, characterized in that: The output of the final result according to the task execution instruction and the optimal model includes: Obtaining task execution instructions, an optimal model, and a model execution plan; the model execution plan includes initialization parameters and operating parameters; Initialize the optimal model by initializing parameters; Achieve operation simulation of the most suitable model through operation parameters; Generate alarm signals based on equipment operating environment data; The results after running the simulation are output to get the final result.

5. A simulation system of the material world according to claim 4, characterized in that: Generating an alarm signal according to the equipment operating environment data includes: Obtain the operating temperature in the equipment operating environment data at intervals of t; When the operating temperature is higher than the temperature threshold for N consecutive times, a high temperature alarm signal of the equipment is generated; Otherwise, when the continuous When the operating temperatures are all higher than the temperature threshold, a high temperature warning signal of the equipment is generated; otherwise, no operation is performed; where D is the proportional coefficient, D∈(0,1).

6. A method for simulating the material world, applied to a simulation system for the material world as claimed in any one of claims 1 to 5, characterized in that: include: S0: Get input data; S1: Analyze the input data through the LLM model to obtain model selection instructions and task execution instructions; S2: Generate the most suitable model and model execution plan based on model selection instructions and RAG technology; S3: Output the final result according to the task execution instructions and the most suitable model; S4: Provide prompts based on the alarm signal and contact relevant personnel.

7. A simulation application of the material world, applied to a simulation system of the material world as claimed in any one of claims 1 to 5, characterized in that: Includes applications in materials science, drug design, and chemical engineering.

Images