Industrial process modeling guiding method and system based on large language model

Abstract

The invention discloses an industrial process modeling guiding method and system based on a large language model, and the method comprises the steps: firstly obtaining current industrial process state information, including industrial process initialization information, process module topological structure information and process module calculation state information; then preprocessing and data cleaning are carried out on the obtained industrial process state information to obtain structured process state data; and then cue word extraction is carried out on the structured process state data based on a preset cue word template, and cue words are obtained. And finally, taking the obtained cue word as model input, and generating a corresponding industrial process modeling guide result in real time through a pre-constructed large language model and outputting the result.

Description

Technical Field

[0001] The present invention relates to the field of industrial process modeling, and particularly to an industrial process modeling guidance method and system based on a large language model. Background Art

[0002] The process simulation platform is the core R & D and design platform in the petrochemical industry, and its applications have penetrated all life cycle business links such as technology R & D, engineering design, and production optimization, playing a key role in the reliability and efficiency of the production process.

[0003] However, in process modeling, the operation suggestions provided by traditional platforms are usually static and cannot be adjusted and updated according to the real-time changes of the process state. Users often need to solve problems by themselves during the process operation, and the accuracy and efficiency of the operation are limited. Moreover, traditional process simulation platforms lack intelligent reasoning capabilities, and they cannot understand the logical relationships and possible results in the process, restricting the choices and innovations of users during the process operation. Therefore, how to further improve the intelligence and efficiency of the process simulation system during modeling has become an urgent problem for practitioners in the same field. Summary of the Invention

[0004] The following gives a brief overview of one or more aspects to provide a basic understanding of these aspects. This overview is not an exhaustive survey of all contemplated aspects, and is neither intended to identify key or decisive elements of all aspects nor to define the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed description that follows.

[0005] The object of the present invention is to solve the above problems, and provides an industrial process modeling guidance method and system based on a large language model, which can automatically perform requirement quality analysis based on objective criteria, thereby ensuring the consistency of the requirement quality analysis results and improving the analysis efficiency.

[0006] The technical solution of the present invention is as follows:

[0007] The present invention provides an industrial process modeling guidance method based on a large language model, including the following steps:

[0008] Step S1: Obtain the current industrial process state information, including industrial process initialization information, process module topology structure information, and process module calculation state information;

[0009] Step S2: Perform preprocessing and data cleaning on the obtained industrial process state information to obtain structured process state data;

[0010] Step S3: Extract prompt words from the structured process state data based on a preset prompt word template to obtain prompt words;

[0011] Step S4: Use the obtained prompt words as the input of the model, and generate and output the corresponding industrial process modeling guidance results in real time through a pre-constructed large language model.

[0012] According to an embodiment of the industrial process modeling guidance method based on a large language model of the present invention, the industrial process initialization information includes chemical components, petroleum analysis data selection, physical property method selection, and flash evaporation algorithm selection. The process module topological structure information includes the number of inlets and outlets of each module and the connection status of inlets and outlets. The process module calculation status information includes the calculated status and the uncalculated status. Among them, after obtaining the industrial process initialization information, the industrial process modeling guidance method based on the large prediction model performs structured preprocessing on the obtained industrial process initialization information to obtain structured process status data.

[0013] According to an embodiment of the industrial process modeling guidance method based on a large language model of the present invention, the industrial process status information includes industrial process initialization data, process module topological structure data, and process module calculation status data. Among them, the industrial process modeling guidance method based on the large prediction model presets a corresponding JSON structure for the industrial process status information to be structured, and then uses the preset JSON structure to perform structured preprocessing on the obtained industrial process status information to obtain structured process status data corresponding to the JSON structure.

[0014] According to an embodiment of the industrial process modeling guidance method based on a large language model of the present invention, in step S3, the industrial process modeling guidance method based on the large language model extracts prompt words through the following steps:

[0015] Step C1: Extract the corresponding status key values for the obtained structured process status data.

[0016] Step C2: Judge the process status of the corresponding structured process status data according to the extracted status key values to obtain process status information. Among them, the process status information includes the integrity of the process initialization status, the rationality of the component quantity selection, the process module topological structure status, the process module outlet connection status, and the process module calculation status.

[0017] Step C3: Use a preset python script and a prompt word template to convert the extracted process status information into prompt words as the input of the large language model.

[0018] According to an embodiment of the industrial process modeling guidance method based on a large language model of the present invention, the industrial process modeling guidance method based on the large language model pre-constructs a large language model through the following steps:

[0019] Step B1: Obtain sample industrial process status information, including sample industrial process initialization information, sample process module topology information, and sample process module calculation status information;

[0020] Step B2: Perform data cleaning and structured preprocessing on the obtained sample industrial process status information to obtain structured sample process status data;

[0021] Step B3: Use a preset Python script and combine it with manual evaluation to convert the structured sample process status data into natural language to obtain the corresponding sample text description;

[0022] Step B4: Determine the corresponding industrial process modeling orientation label based on the converted sample text description;

[0023] Step B5: Use the sample text description and the corresponding industrial process modeling orientation label as seed sample data to expand the sample data scale to obtain an expanded sample text description;

[0024] Step B6: Use the expanded sample text description as the sample data for model training and the industrial process modeling orientation label corresponding to the expanded sample text description as the sample label to train the initial large language model to obtain a large language model for extracting industrial process modeling orientation results.

[0025] According to an embodiment of the industrial process modeling orientation method based on a large language model of the present invention, in step B5, the industrial process modeling orientation method based on a large language model uses the self-generation instruction framework SELF - INSTRUCT to expand the sample data scale, including:

[0026] The first step is to migrate the self-generation instruction SELF - INSTRUCT to the single-task multi-instance domain to generate the SELF - INSTRUCT single-task instance generation instruction;

[0027] The second step is to expand the sample data scale based on the SELF - INSTRUCT single-task instance generation instruction.

[0028] According to an embodiment of the industrial process modeling orientation method based on a large language model of the present invention, in step B6, the industrial process modeling orientation method based on a large language model performs the model training of the large language model through the following steps:

[0029] The first step is to use the expanded sample text description as the model input and input it into the initial large language model to obtain the corresponding industrial process orientation result and output it;

[0030] In the second step, based on the industrial process modeling orientation tags and industrial process modeling orientation results corresponding to the augmented sample text, the model parameters of the initial large language model are iteratively optimized to obtain a trained large language model.

[0031] The present invention also provides an industrial process modeling orientation system based on a large language model, including a state information acquisition unit, a state information processing unit, a prompt information extraction unit, and a modeling orientation result generation unit; wherein,

[0032] The state information acquisition unit is used to acquire the current industrial process state information, including industrial process initialization information, process module topology structure information, and process module calculation state information;

[0033] The state information processing unit is used to preprocess and clean the acquired industrial process state information to obtain structured process state data;

[0034] The prompt information extraction unit is used to extract prompts from the structured process state data based on a preset prompt template to obtain prompts;

[0035] The modeling orientation result generation unit is used to take the acquired prompts as model inputs, and through a pre-constructed large language model, generate corresponding industrial process modeling orientation results in real time and output them.

[0036] According to an embodiment of the industrial process modeling orientation system based on a large language model of the present invention, the industrial process initialization information includes chemical components, petroleum analysis data selection, physical property method selection, and flash evaporation algorithm selection, the process module topology structure information includes the number of inlets and outlets of each module and the connection state of inlets and outlets, and the process module calculation state information includes a calculated state and an uncalculated state; wherein, after the state information acquisition unit acquires the industrial process state information, the state information processing unit performs structured preprocessing on the acquired industrial process state information to obtain structured structured process state data.

[0037] According to an embodiment of the industrial process modeling orientation system based on a large language model of the present invention, the industrial process state information includes industrial process initialization data, process module topology structure data, and process module calculation state data; wherein, the state information processing unit presets a corresponding JSON structure for the industrial process state information to be structured, and then uses the preset JSON structure to perform structured preprocessing on the acquired industrial process state information to obtain structured process state data corresponding to the JSON structure.

[0038] According to an embodiment of the industrial process modeling orientation system based on a large language model of the present invention, the prompt information extraction unit extracts prompts through the following steps:

[0039] Step C1: Extract the corresponding status key values for the obtained structured process status data;

[0040] Step C2: Determine the process status of the corresponding structured process status data according to the extracted status key values, and obtain process status information; wherein, the process status information includes the completeness of the process initialization status, the reasonableness of the component quantity selection, the status of the process module topological structure, the connection status of the process module outlet, and the calculation status of the process module;

[0041] Step C3: Use a preset Python script and prompt template to convert the extracted process status information into a prompt for input to the large language model.

[0042] According to an embodiment of the industrial process modeling guiding system based on a large language model of the present invention, the industrial process modeling guiding system based on a large language model pre-constructs a large language model through the following steps:

[0043] Step B1: Obtain sample industrial process status information, including sample industrial process initialization information, sample process module topological structure information, and sample process module calculation status information;

[0044] Step B2: Perform data cleaning and structured preprocessing on the obtained sample industrial process status information to obtain structured sample process status data;

[0045] Step B3: Use a preset Python script and combine manual evaluation to convert the structured sample process status data into natural language to obtain the corresponding sample text description;

[0046] Step B4: Determine the corresponding industrial process modeling guiding label based on the converted sample text description;

[0047] Step B5: Use the sample text description and the corresponding industrial process modeling guiding label as seed sample data to expand the scale of the sample data to obtain an expanded sample text description;

[0048] Step B6: Use the expanded sample text description as the sample data for model training and the industrial process modeling guiding label corresponding to the expanded sample text description as the sample label to train the initial large language model to obtain a large language model for extracting the industrial process modeling guiding result.

[0049] 17. According to an embodiment of the industrial process modeling guiding system based on a large language model of the present invention, in step B5, the industrial process modeling guiding method based on a large language model uses the self-generated instruction framework SELF-INSTRUCT to expand the scale of the sample data, including:

[0050] First step, migrate the self-generated instruction SELF-INSTRUCT to the single-task multi-instance domain to generate the SELF-INSTRUCT single-task instance generation instruction;

[0051] Second step, expand the sample data scale based on the SELF-INSTRUCT single-task instance generation instruction.

[0052] According to an embodiment of the industrial process modeling guidance system based on a large language model of the present invention, in step B6, the industrial process modeling guidance system based on a large language model performs model training of the large language model through the following steps:

[0053] First step, use the expanded sample text description as the model input and input it into the initial large language model to obtain the corresponding industrial process guidance result and output it;

[0054] Second step, based on the industrial process modeling guidance label and the industrial process modeling guidance result corresponding to the expanded sample text description, iteratively optimize the model parameters of the initial large language model to obtain the trained large language model.

[0055] The present invention also provides a computer-readable medium storing computer program code, and the computer program code realizes the above method when executed by a processor.

[0056] The present invention also provides an industrial process modeling guidance device based on a large language model, including:

[0057] A memory for storing instructions executable by a processor; and

[0058] A processor for executing the instructions to realize the above method.

[0059] The present invention has the following beneficial effects compared with the prior art: For the automatic generation of industrial process modeling guidance, the present invention obtains the current industrial process state information, preprocesses and cleans the current industrial process state information to obtain structured process state data, and then according to the structured process state data, combines with a preset prompt word template for further processing to obtain a model input prompt word, inputs the prompt word into a pre-constructed large language model, obtains the process modeling guidance result output by the large language model, and generates the corresponding industrial process modeling guidance result in real time through the pre-constructed large language model and outputs it, thereby realizing industrial process modeling guidance. Through the present invention, not only the intelligence of industrial process modeling is enhanced, but also the accuracy and efficiency of users when using the industrial process simulation system for modeling are improved, and the efficiency of industrial process modeling is increased. Description of the Drawings

[0060] After reading the detailed description of the embodiments of the present disclosure in conjunction with the following drawings, the above features and advantages of the present invention can be better understood. In the drawings, the components are not necessarily drawn to scale, and components with similar relevant characteristics or features may have the same or similar reference numerals.

[0061] Figure 1 is a flowchart showing the steps of an embodiment of the industrial process modeling guidance method based on a large language model of the present invention.

[0062] Figure 2 is a flowchart showing the steps of an embodiment of the prompt extraction of the present invention.

[0063] Figure 3 is a flowchart showing the steps of an embodiment of constructing a large language model of the present invention.

[0064] Figure 4 is a system architecture diagram showing an embodiment of the industrial process modeling guidance system based on a large language model of the present invention. Detailed Embodiments

[0065] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some examples or embodiments of the present application. For those of ordinary skill in the art, without creative efforts, the present application can also be applied to other similar scenarios based on these drawings. Unless obvious from the language context or otherwise stated, the same reference numerals in the figures represent the same structure or operation.

[0066] As shown in the present application and the claims, unless the context clearly indicates an exceptional situation, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0067] Unless otherwise specifically stated, the relative arrangements of components and steps, numerical expressions, and numerical values set forth in these embodiments do not limit the scope of the present application. At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Technologies, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, such technologies, methods, and devices should be regarded as part of the authorization specification. In all the examples shown and discussed herein, any specific values should be construed as merely exemplary and not as a limitation. Therefore, other examples of the exemplary embodiments may have different values. It should be noted that like reference numerals and letters denote like items in the following drawings, and thus, once an item is defined in one drawing, further discussion thereof is not required in subsequent drawings.

[0068] When detailing the embodiments of the present invention, for ease of illustration, the cross-sectional views showing the device structure are enlarged locally not in accordance with the general scale, and the schematic diagrams are only examples and should not limit the scope of protection of the present invention herein. In addition, in actual production, three-dimensional spatial dimensions of length, width, and depth should be included.

[0069] In the description of the present application, it should be understood that the orientation or positional relationships indicated by orientation words such as "front, back, up, down, left, right", "lateral, vertical, perpendicular, horizontal", and "top, bottom" are generally based on the orientation or positional relationships shown in the drawings, and are only for ease of describing the present application and simplifying the description. Without contrary description, these orientation words do not indicate and imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and thus should not be construed as limiting the scope of protection of the present application; the orientation words "inside, outside" refer to the inside and outside relative to the contour of each component itself.

[0070] Disclosed herein is an embodiment of an industrial process modeling guiding method based on a large language model. Figure 1 is a flowchart showing an embodiment of the industrial process modeling guiding method based on a large language model of the present invention. Please refer to Figure 1 The following is a detailed description of each step of the industrial process modeling guiding method based on a large language model.

[0071] Step S1: Obtain the current industrial process status information, including industrial process initialization information, process module topological structure information, and process module calculation status information.

[0072] Step S2: Preprocess and clean the obtained industrial process status information to obtain structured process status data.

[0073] In this embodiment, the industrial process initialization information includes chemical components, petroleum analysis data selection, physical property method selection, and flash algorithm selection. The process module topology structure information includes the number of inlets and outlets of each module, the connection status of inlets and outlets, and the normative mapping of the connection relationships of inlets and outlets of each module. The process module calculation status information includes the calculated status and the uncalculated status. After obtaining the industrial process initialization information, it is first necessary to perform structured preprocessing on the obtained industrial process initialization information to obtain structured process status data for subsequent modeling.

[0074] Specifically, in this implementation, for the process module topology structure information, the normative mapping of the connection relationships of inlets and outlets of each module plays a crucial role in judging the correctness of the process module topology. For different industrial modules, the restrictions on the connection relationships of inlets and outlets are different. For the flash tank module, usually 1 inlet stream module needs to be connected, and 1 - 3 outlet stream modules can be accepted. For the stream module, usually 0 - 1 non-stream inlet modules need to be connected, and 0 - 1 non-stream outlet modules are connected. When defining the connection relationships of inlets and outlets of each module, the normative mapping table of the connection relationships of inlets and outlets as shown below can be referred to:

[0075] O mapping ={A module[1] <a1 front ,a1 back >,

[0076] A module[2] <a2 front ,a2 back >...A module[i] <ai front ,ai back >}

[0077] Among them, A module[i] represents the chemical reaction modules involved in the process simulation platform, such as two categories: stream and non-stream. ai front represents the upper and lower limits of non-self-class inlet modules of the reaction module, and a1 back represents the upper and lower limits of non-self-class outlet modules of the reaction module.

[0078] In this embodiment, the industrial process status information includes industrial process initialization data, process module topology structure data, and process module calculation status data. Among them, the industrial process modeling guidance method based on the large prediction model presets a corresponding JSON structure for the industrial process status information to be structured, and then uses the preset JSON structure to perform structured preprocessing on the obtained industrial process status information, so as to obtain structured process status data in the corresponding JSON structure.

[0079] In one implementation, Python libraries and methods are used to extract industrial process initialization data, process module topology data, and process module calculation status data from the database process table during the process initialization phase, process construction phase, and process calculation completion phase. Then, the extracted data is integrated into a preset JSON structure to remove redundant data (such as initialization configuration detailed parameters, process module calculation results, etc.), obtaining pure industrial process initialization data, process module topology data, and process module calculation status data.

[0080] Taking the process construction phase as an example, when the user selects the trigger-oriented suggestion during the process construction phase, the orientation request will carry the current process ID and be passed to the backend. After passing through the ORM framework, the current industrial process initialization data, process module topology data, and process module calculation status data are obtained based on the process ID. Then, these data are stored in the process table in JSON form. Finally, the obtained data is reconstructed into a unified structured description, and redundant data is removed through data cleaning, thus obtaining the following data structure O root :

[0081] O root ={("prepare":O prepare ),

[0082] ("graphic":O graphic ),

[0083] ("module_var":O module_var )}

[0084] Among them, O prepare represents the current process initialization configuration data after processing, and the structure is expressed as:

[0085] O prepare ={("compounds":A comp ),

[0086] ("oil":A oil ),

[0087] ("methods":A methods ),

[0088] ("algorithm":A algor )}

[0089] Among them, A comp is the list of chemical components selected for the current process after cleaning, A oil is the list of petroleum analysis data selected for the current process after cleaning, A methodsData list of physical property methods selected for the current process after cleaning, A algor Data list of flash algorithms selected for the current process after cleaning. All of the above data objects can be empty, indicating that no configuration is made.

[0090] O graphic Represents the data dictionary of the topological structure of the current process module after processing, which represents each module in the process and the modules connected to the outlet of each module. This object can be empty, indicating that there are no modules on the current canvas.

[0091] O module_var Represents the calculation status data of each module in the current process after processing. The structure is represented as:

[0092] O module_var ={("block":O block ),("mstr":O mstr )}

[0093] Among them, O block is the calculation status dictionary data of the non-stream module in the current process, and O mstr is the calculation status dictionary data of the module in the current process. In addition, the value of the sub-module in the above dictionary data is A status , and the structure is represented as:

[0094]

[0095] Among them, when the value of A status is 0, it means that the module is normal; when the value is 1, it means that the module is abnormal.

[0096] Step S3: Extract prompt words from the structured process status data based on a preset prompt word template to obtain prompt words.

[0097] In this embodiment, a preset python script and a prompt word template are used to extract prompt words. Figure 2 is a step flowchart showing an embodiment of prompt word extraction of the present invention. Please refer to Figure 2 , and the following is a detailed description of each step of prompt word extraction.

[0098] Step C1: Extract the corresponding status key values for the obtained structured process status data.

[0099] Step C2: Judge the process status of the corresponding structured process status data according to the extracted status key values to obtain process status information; among them, the process status information includes the integrity of the process initialization status, the rationality of the component quantity selection, the topological structure status of the process module, the connection situation of the process module outlet, and the calculation status of the process module.

[0100] In this embodiment, according to the status key values to be extracted and the process status information, a Python script (including Python libraries and scripts) is preset. After obtaining the structured process status data through the above steps, according to the obtained structured process status data, the preset Python script is used, combined with some manual definitions, to extract the corresponding status key values. The process status of the corresponding structured process status data is judged through the extracted status key values, so as to obtain the process status information. Among them, the process status information includes the integrity of the process initialization status, the reasonable selection of the number of components, the topology structure status of the process module, the connection status of the process module outlet, the calculation status of the process module, etc. The obtained process status information is further combined as the prompt words for input to the large language model.

[0101] Step C3: Use the preset Python script and the prompt word template to convert the extracted process status information into the prompt words for input to the large language model.

[0102] In this embodiment, for the content of the prompt words to be converted from the process status information, a Python script and a prompt word template are preset. The obtained process status information is converted into a complete text description through the preset Python script and the prompt word template as the prompt words for input to the pre-constructed large language model. Among them, the preset prompt word template includes the description of the current process initialization configuration, the description of the current process module topology structure, and the description of the current process module calculation status, and specifies the modeling guiding task.

[0103] Step S4: Use the obtained prompt words as the model input, and the corresponding industrial process modeling guiding results are generated and output in real time through the pre-constructed large language model.

[0104] In this embodiment, after obtaining the prompt words through the above steps, the obtained prompt words are input into the pre-constructed large language model, so as to generate and output the corresponding industrial process modeling guiding results in real time. Figure 3 The figure shows the step flow chart of an embodiment of constructing a large language model of the present invention. Please refer to Figure 3 , and the following is a detailed description of each step of constructing the large language model.

[0105] Step B1: Obtain the sample industrial process status information, including the sample industrial process initialization information, the sample process module topology structure information, and the sample process module calculation status information.

[0106] Step B2: Perform data cleaning and structured preprocessing on the obtained sample industrial process status information to obtain the structured sample process status data.

[0107] Step B3: Use the preset Python script and combine it with manual evaluation to convert the structured sample process status data into natural language to obtain the corresponding sample text description.

[0108] In this embodiment, before constructing a large language model, the large language model is first obtained from the historical database for processing, so as to obtain data for training the large language model. Among them, after obtaining the sample industrial process status information, the obtained sample industrial process status information is cleaned, invalid data, such as incomplete data, malformed data, etc., is eliminated, and the data is deduplicated, and then structured preprocessed to obtain structured sample process status data. The specific steps are consistent with step S2 and will not be repeated here. Finally, the cleaned and deduplicated structured data is converted into natural language by combining manual evaluation standards with python scripts to obtain the corresponding sample text description, such as "the current process initialization configuration has been completed."

[0109] Assume that the current process has reaction modules. At this time, they are connected as follows: the outlet of module MSTR1 is connected to COMP1, and the outlet of module COMP1 is connected to MSTR2. The outlet of module MSTR2 is not connected to other modules. Processing steps: Module MSTR1 has not completed calculation, and module MSTR2 has not completed calculation".

[0110] Step B4: Determine the corresponding industrial process modeling-oriented label based on the converted sample text description.

[0111] In this embodiment, a python script is preset for the conversion relationship between the sample text description and the corresponding industrial process modeling guidance label. Through the preset python script, combined with manual evaluation, the structured sample process state data is converted into natural language, thereby obtaining the corresponding sample text description, and then the corresponding industrial process modeling guidance label is determined according to the obtained sample text description. The sample text description is classified by the industrial process modeling guidance label.

[0112] Specifically, in this embodiment, the industrial process modeling - oriented tags include process health, incomplete initialization configuration, calculation anomaly, and process topology structure anomaly. Among them, the process health tag is defined as all configuration parameters of the process being complete and the process module topology structure being correct. The incomplete initialization configuration tag is defined as the lack of process initialization state data. The calculation anomaly tag is defined as all configuration parameters of the process being complete and the process module topology structure being correct, but there is a situation where the calculation of the process module is not completed. The process topology structure anomaly tag is defined as all configuration parameters of the process being complete and the process module topology structure being incorrect. Through these four industrial process modeling - oriented tags, the sample text descriptions are divided into four categories: process health, incomplete initialization configuration, calculation anomaly, and process topology structure anomaly, and then model training is carried out.

[0113] Step B5: Use the sample text description and the corresponding industrial process modeling - oriented tag as seed sample data to expand the scale of the sample data, and obtain the expanded sample text description.

[0114] In this embodiment, before training the large - language model, in order to expand the scale of the training data, after obtaining the sample text description and the corresponding industrial process modeling - oriented tag through the above steps, the sample text description and the corresponding industrial process modeling - oriented tag are used as seed sample data to expand the scale of the sample data, and the expanded sample text description is obtained as the data for model training.

[0115] In one implementation, the self - generated instruction framework SELF - INSTRUCT is used to expand the scale of the sample data. Among them, when expanding the scale of the sample data, first, the self - generated instruction SELF - INSTRUCT is migrated to the single - task multi - instance domain to generate the SELF - INSTRUCT single - task instance generation instruction, and then the scale of the sample data is expanded based on the SELF - INSTRUCT single - task instance generation instruction.

[0116] Step B6: Use the expanded sample text description as the sample data for model training and the industrial process modeling - oriented tag corresponding to the expanded sample text description as the sample label to train the initial large - language model, and obtain the large - language model for extracting the industrial process modeling - oriented result.

[0117] In this embodiment, after expanding the scale of the sample data through the above steps, the initial large language model is trained with the expanded sample text description as the sample data for model training and the industrial process modeling-oriented label corresponding to the expanded sample text description as the sample label, so as to obtain a large language model for extracting the industrial process modeling-oriented result. Among them, when performing model training, first, the expanded sample text description is used as the model input and input into the initial large language model to obtain the corresponding industrial process-oriented result and output it. Then, based on the industrial process modeling-oriented label corresponding to the expanded sample text description and the industrial process modeling-oriented result, the model parameters of the initial large language model are iteratively optimized to obtain the finally trained large language model.

[0118] In addition, in this embodiment, when performing the model training of the large language model, the large language model after parameter fine-tuning can also be verified from the integrity of the initial process parameters, the rationality of the process topology structure, and the integrity of the process calculation state to verify whether it can accurately understand and respond to different industrial process states to ensure the effectiveness of the trained large language model. At the same time, the fine-tuned large language model can also be applied to other data sets collected from the actual environment to further verify its effectiveness and evaluate its applicability and intelligence in the real application environment, thereby improving the model training efficiency of the large language model.

[0119] This specification also provides an industrial process modeling-oriented system based on a large language model. Figure 4 It is a system architecture diagram showing an embodiment of the industrial process modeling-oriented system based on the large language model of the present invention. As Figure 4 shown, in this embodiment, the industrial process modeling-oriented system based on the large language model (hereinafter sometimes simply referred to as the industrial process modeling-oriented system) includes a status information acquisition unit, a status information processing unit, a prompt information extraction unit, and a modeling-oriented result generation unit. Among them, the status information acquisition unit is used to acquire the current industrial process status information, including industrial process initialization information, process module topology structure information, and process module calculation state information. The status information processing unit is used to preprocess and clean the acquired industrial process status information to obtain structured process status data. The prompt information extraction unit is used to extract prompts from the structured process status data based on a preset prompt template to obtain prompts. The modeling-oriented result generation unit is used to use the obtained prompts as the model input and generate and output the corresponding industrial process modeling-oriented result in real time through a pre-constructed large language model.

[0120] In this embodiment, the industrial process initialization information includes chemical components, petroleum analysis data selection, physical property method selection, and flash algorithm selection. The process module topological structure information includes the number of inlets and outlets of each module, the connection status of inlets and outlets, and the normative mapping of the connection relationship between the inlets and outlets of each module. The process module calculation status information includes the calculated status and the uncalculated status. After obtaining the industrial process initialization information, it is first necessary to perform structured preprocessing on the obtained industrial process initialization information to obtain structured process status data for subsequent modeling.

[0121] Specifically, in this implementation, for the process module topological structure information, the normative mapping of the connection relationship between the inlets and outlets of each module plays a crucial role in judging the correctness of the process module topological structure. For different industrial modules, the restrictions on the connection relationship between the inlets and outlets are different. For the flash tank module, usually 1 inlet stream module needs to be connected, and 1 - 3 outlet stream modules can be accepted. For the stream module, usually 0 - 1 non-stream inlet modules need to be connected, and 0 - 1 non-stream outlet modules are connected. When defining the connection relationship between the inlets and outlets of each module, the following normative mapping table of the connection relationship between the inlets and outlets can be referred to:

[0122] O mapping ={A module [1]<a1 front ,a1 back ,

[0123] A module[2] <a2 front ,a2 back >...A module[u] <ai front ,ai back >}

[0124] Among them, A module[i] represents the chemical reaction modules involved in the process simulation platform, such as two categories: stream and non-stream. ai front represents the upper and lower limits of the non-self type inlet modules of the reaction module, and a1 back represents the upper and lower limits of the non-self type outlet modules of the reaction module.

[0125] In this embodiment, the industrial process status information includes industrial process initialization data, process module topological structure data, and process module calculation status data. Among them, the industrial process modeling guiding method based on the large prediction model presets a corresponding JSON structure for the industrial process status information to be structured, and then uses the preset JSON structure to perform structured preprocessing on the obtained industrial process status information, so as to obtain structured process status data in the corresponding JSON structure.

[0126] In one embodiment, Python libraries and methods are used to extract industrial process initialization data, process module topology data, and process module calculation status data from the database process table during the process initialization stage, process construction stage, and process calculation completion stage. Then, the extracted data is integrated into a preset JSON structure to remove redundant data (such as initialization configuration detailed parameters, process module calculation results, etc.), obtaining pure industrial process initialization data, process module topology data, and process module calculation status data.

[0127] Taking the process construction stage as an example, when the user selects the trigger guidance suggestion during the process construction stage, the guidance request will carry the current process ID and be passed to the backend. After passing through the ORM framework, the current industrial process initialization data, process module topology data, and process module calculation status data are obtained based on the process ID. Then, these data are stored in the process table in JSON format. Finally, the obtained data is reconstructed into a unified structured description, and redundant data is removed through data cleaning, resulting in the following data structure O root :

[0128] O root ={("prepare":O prepare ),

[0129] ("graphic":O graphic ),

[0130] ("module_var":O module_var )}

[0131] Among them, O prepare represents the current process initialization configuration data after processing, and the structure is represented as:

[0132] O prepare ={("compounds":A comp ),

[0133] ("oil":A oil ),

[0134] ("methods":A methods ),

[0135] ("algorithm":A algor )}

[0136] Among them, A comp is the list of chemical components selected for the current process after cleaning, A oil is the list of petroleum analysis data selected for the current process after cleaning, A methodsList of physical property method data selected for the current process after cleaning, A algor List of flash algorithm data selected for the current process after cleaning. All of the above data objects can be empty, indicating no configuration.

[0137] O graphic Represents the data dictionary of the topological structure of the current process module after processing, which represents each module in the process and the modules connected to the outlet of each module. This object can be empty, indicating that there are no modules on the current canvas.

[0138] O module_var Represents the calculation status data of each module in the current process after processing. The structure is represented as:

[0139] O module_var ={("block":O block ),("mstr":O mstr )}

[0140] Among them, O block is the calculation status dictionary data of the non-stream module in the current process, and O mstr is the calculation status dictionary data of the module in the current process. In addition, the value of the sub-module in the above dictionary data is A status , and the structure is represented as:

[0141]

[0142] Among them, when the value of A status is 0, it means that the module is normal; when the value is 1, it means that the module is abnormal.

[0143] In this embodiment, when the prompt word information extraction unit extracts prompt words, it uses a preset python script and a prompt word template to extract prompt words. Figure 2 is a flowchart showing the steps of an embodiment of the prompt word extraction of the present invention. Please refer to Figure 2 . The following is a detailed description of each step of the prompt word extraction.

[0144] Step C1: For the obtained structured process status data, extract the corresponding status key values.

[0145] Step C2: Judge the process status of the corresponding structured process status data according to the extracted status key values, and obtain the process status information; among them, the process status information includes the completeness of the process initialization status, the rationality of the component quantity selection, the topological structure status of the process module, the connection status of the process module outlet, and the calculation status of the process module.

[0146] In this embodiment, according to the status key values to be extracted and the process status information, a Python script (including Python libraries and scripts) is preset. After obtaining the structured process status data through the above steps, according to the obtained structured process status data, the preset Python script is used, combined with some manual definitions, to extract the corresponding status key values. The process status of the corresponding structured process status data is judged through the extracted status key values, so as to obtain the process status information. Among them, the process status information includes the integrity of the process initialization status, the reasonable selection of the component quantity, the topology structure status of the process module, the connection status of the process module outlet, the calculation status of the process module, etc. The obtained process status information is further combined as the prompt words for input to the large language model.

[0147] Step C3: Use the preset Python script and the prompt word template to convert the extracted process status information into the prompt words for input to the large language model.

[0148] In this embodiment, for the content of the prompt words to be converted from the process status information, a Python script and a prompt word template are preset. The obtained process status information is converted into a complete text description through the preset Python script and the prompt word template as the prompt words for input to the pre-constructed large language model. Among them, the preset prompt word template includes the description of the current process initialization configuration, the description of the current process module topology structure, and the description of the current process module calculation status, and specifies the modeling guiding task.

[0149] In this embodiment, when generating the modeling guiding result, the modeling guiding result generation unit uses a pre-constructed large language model to generate the corresponding industrial process modeling guiding result in real time and output it. Figure 3 The flowchart shows the steps of an embodiment of constructing a large language model of the present invention. Please refer to Figure 3 , and the following is a detailed description of each step of constructing the large language model.

[0150] Step B1: Obtain the sample industrial process status information, including the sample industrial process initialization information, the sample process module topology structure information, and the sample process module calculation status information.

[0151] Step B2: Perform data cleaning and structured preprocessing on the obtained sample industrial process status information to obtain the structured sample process status data.

[0152] Step B3: Use the preset Python script, combined with manual evaluation, to convert the structured sample process status data into natural language to obtain the corresponding sample text description.

[0153] In this embodiment, before building the large language model, data for training the large language model is first obtained from the historical database for processing. Among them, after obtaining the sample industrial process status information, data cleaning is performed on the obtained sample industrial process status information to remove invalid data, such as incomplete data, data with incorrect formats, etc., and duplicate data is removed. Then, it is subjected to structured preprocessing to obtain structured sample process status data. The specific steps are the same as those in step S2 and will not be elaborated here. Finally, by combining the manual evaluation criteria with a Python script, the cleaned and deduplicated structured data is converted into natural language to obtain the corresponding sample text description, such as "The current process initialization configuration has been completed."

[0154] Assume that the current process has reaction modules, and their connection methods are as follows at this time: The outlet of module MSTR1 is connected to COMP1, and the outlet of module COMP1 is connected to MSTR2. The outlet of module MSTR2 is not connected to other modules. Processing step: Module MSTR1 has not completed the calculation, and module MSTR2 has not completed the calculation."

[0155] Step B4: Determine the corresponding industrial process modeling orientation label based on the converted sample text description.

[0156] In this embodiment, for the conversion relationship between the sample text description and the corresponding industrial process modeling orientation label, a Python script is preset. Through the preset Python script, combined with manual evaluation, the structured sample process status data is converted into natural language to obtain the corresponding sample text description, and then the corresponding industrial process modeling orientation label is determined according to the obtained sample text description. The sample text description is classified through the industrial process modeling orientation label.

[0157] Specifically, in this embodiment, the industrial process modeling orientation labels include process health, incomplete initialization configuration, calculation exception, and process topology structure exception. Among them, the process health label is defined as all configuration parameters of the process being complete and the process module topology structure being correct. The incomplete initialization configuration label is defined as the lack of process initialization status data. The calculation exception label is defined as all configuration parameters of the process being complete and the process module topology structure being correct, but there is a situation where the calculation of the process module has not been completed. The process topology structure exception label is defined as all configuration parameters of the process being complete and the process module topology structure being incorrect. Through these four industrial process modeling orientation labels, the sample text description is divided into four categories: process health, incomplete initialization configuration, calculation exception, and process topology structure exception, and then model training is carried out.

[0158] Step B5: Use the sample text description and the corresponding industrial process modeling-oriented labels as seed sample data to expand the scale of the sample data, and obtain the expanded sample text description.

[0159] In this embodiment, before training the large language model, in order to expand the scale of the training data, after obtaining the sample text description and the corresponding industrial process modeling-oriented labels through the above steps, the sample text description and the corresponding industrial process modeling-oriented labels are used as seed sample data to expand the scale of the sample data, and the expanded sample text description is obtained as the data for model training.

[0160] In one implementation, the self-generated instruction framework SELF-INSTRUCT is used to expand the scale of the sample data. Among them, when expanding the scale of the sample data, first, the self-generated instruction SELF-INSTRUCT is migrated to the single-task multi-instance field to generate the SELF-INSTRUCT single-task instance generation instruction, and then the scale of the sample data is expanded based on the SELF-INSTRUCT single-task instance generation instruction.

[0161] Step B6: Use the expanded sample text description as the sample data for model training and the industrial process modeling-oriented labels corresponding to the expanded sample text description as the sample labels to train the initial large language model, and obtain the large language model for extracting the industrial process modeling-oriented results.

[0162] In this embodiment, after expanding the scale of the sample data through the above steps, the expanded sample text description is used as the sample data for model training, and the industrial process modeling-oriented labels corresponding to the expanded sample text description are used as the sample labels to train the initial large language model, so as to obtain the large language model for extracting the industrial process modeling-oriented results. Among them, when performing model training, first, the expanded sample text description is used as the model input and input into the initial large language model, and the corresponding industrial process-oriented result is obtained and output. Then, based on the industrial process modeling-oriented labels corresponding to the expanded sample text description and the industrial process modeling-oriented results, the model parameters of the initial large language model are iteratively optimized, so as to obtain the finally trained large language model.

[0163] In addition, in this embodiment, when training the large language model, the fine-tuned large language model can also be verified from the integrity of the initial process parameters, the rationality of the process topology, and the integrity of the process calculation status to verify whether it can accurately understand and respond to different industrial process states, so as to ensure the effectiveness of the trained large language model. At the same time, the fine-tuned large language model can also be applied to other data sets collected from the actual environment to further verify its effectiveness and evaluate its applicability and intelligence in the real application environment, thereby improving the model training efficiency of the large language model.

[0164] This specification also provides a computer-readable medium storing computer program code, which, when executed by a processor, implements the industrial process modeling guidance method based on a large language model as described above.

[0165] This specification also provides an industrial process modeling guidance device based on a large language model, including a memory storing instructions executable by a processor, and a processor for executing the instructions in the memory to implement the industrial process modeling guidance method based on a large language model as described above.

[0166] The foregoing description of the disclosure has been provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0167] Those skilled in the art will further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, the various illustrative components, blocks, modules, circuits, and steps are described above in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and the design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the invention.

[0168] The various illustrative logical blocks, modules, and circuits described in connection with the embodiments disclosed herein can be implemented or performed with a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration.

[0169] The steps of a method or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such that the processor can read from, and write to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In the alternative, the processor and the storage medium may reside as discrete components in a user terminal.

[0170] In one or more exemplary embodiments, the described functionality may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software as a computer program product, the functions may be stored on or transmitted via a computer-readable medium as one or more instructions or code. The computer-readable medium includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. The storage media may be any available media that can be accessed by a computer. By way of example and not limitation, such computer-readable media can include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. Any connection is properly termed a computer-readable medium. For example, if the software is transmitted from a web site, server, or other remote source using a coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of medium. As used herein, the terms "disk" and "disc" include compact discs (CDs), laser discs, optical discs, digital versatile discs (DVDs), floppy disks, and Blu-ray discs, where disks typically reproduce data magnetically, while discs reproduce data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.

Claims

1. An industrial process modeling-oriented method based on large language models, characterized in that, Including the following steps: Step S1: Obtain the current industrial process status information, including industrial process initialization information, process module topology structure information, and process module calculation status information; Step S2: Preprocess and clean the obtained industrial process status information to obtain structured process status data; Step S3: Extract prompt words from the structured process status data based on a preset prompt word template to obtain prompt words; Step S4: Use the obtained prompt words as model inputs, and generate and output corresponding industrial process modeling guidance results in real time through a pre-constructed large language model.

2. The industrial process modeling-oriented method based on a large language model according to claim 1, wherein The industrial process initialization information includes chemical components, petroleum analysis data selection, physical property method selection, and flash evaporation algorithm selection. The process module topology structure information includes the number of inlets and outlets and the connection status of inlets and outlets of each module. The process module calculation status information includes the calculated status and the uncalculated status. Among them, after the industrial process modeling guidance method based on the large prediction model obtains the industrial process initialization information, it performs structured preprocessing on the obtained industrial process initialization information to obtain structured process status data.

3. The industrial process modeling guidance method based on a large language model according to claim 1, wherein The industrial process status information includes industrial process initialization data, process module topology structure data, and process module calculation status data. Among them, the industrial process modeling guidance method based on the large prediction model presets a corresponding JSON structure for the industrial process status information to be structured, and then uses the preset JSON structure to perform structured preprocessing on the obtained industrial process status information to obtain structured process status data with the corresponding JSON structure.

4. The industrial process modeling guiding method based on the large language model according to claim 1, wherein In step S3, the industrial process modeling guidance method based on the large language model performs prompt word extraction through the following steps: Step C1: Extract the corresponding status key values for the obtained structured process status data; Step C2: Judge the process status of the corresponding structured process status data according to the extracted status key values to obtain process status information. The process status information includes the integrity of the process initialization status, the rationality of the component quantity selection, the process module topology structure status, the process module outlet connection status, and the process module calculation status; Step C3: Use a preset python script and prompt word template to convert the extracted process status information into prompt words as the input of the large language model.

5. The industrial process modeling guiding method based on a large language model according to claim 4, wherein The industrial process modeling guidance method based on the large language model pre-constructs the large language model through the following steps: Step B1: Obtain sample industrial process status information, including sample industrial process initialization information, sample process module topology structure information, and sample process module calculation status information; Step B2: Clean and perform structured preprocessing on the obtained sample industrial process status information to obtain structured sample process status data; Step B3: Use a preset python script and combine manual evaluation to convert the structured sample process status data into natural language to obtain the corresponding sample text description; Step B4: Determine the corresponding industrial process modeling guidance label based on the converted sample text description; Step B5: Use the sample text description and the corresponding industrial process modeling-oriented labels as seed sample data to expand the scale of the sample data, and obtain the expanded sample text description; Step B6: Use the expanded sample text description as the sample data for model training, and use the industrial process modeling-oriented labels corresponding to the expanded sample text description as sample labels to train the initial large language model, and obtain a large language model for extracting industrial process modeling-oriented results.

6. The industrial process modeling guidance method based on a large language model according to claim 5, wherein In Step B5, the industrial process modeling-oriented method based on the large language model uses the self-generated instruction framework SELF-INSTRUCT to expand the scale of the sample data, including: The first step is to migrate the self-generated instruction SELF-INSTRUCT to the single-task multi-instance domain to generate the SELF-INSTRUCT single-task instance generation instruction; The second step is to expand the scale of the sample data based on the SELF-INSTRUCT single-task instance generation instruction.

7. The industrial process modeling guiding method based on a large language model according to claim 5, characterized in that In Step B6, the industrial process modeling-oriented method based on the large language model performs the model training of the large language model through the following steps: The first step is to use the expanded sample text description as the model input, input it into the initial large language model, and obtain and output the corresponding industrial process-oriented result; The second step is to iteratively optimize the model parameters of the initial large language model based on the industrial process modeling-oriented labels and industrial process modeling-oriented results corresponding to the expanded sample text description, and obtain the trained large language model.

8. An industrial process modeling-oriented system based on a large language model, characterized in that, It includes a status information acquisition unit, a status information processing unit, a prompt information extraction unit, and a modeling-oriented result generation unit; among them, The status information acquisition unit is used to acquire the current industrial process status information, including industrial process initialization information, process module topology structure information, and process module calculation status information; The status information processing unit is used to preprocess and clean the acquired industrial process status information to obtain structured process status data; The prompt information extraction unit is used to extract prompts from the structured process status data based on a preset prompt template to obtain prompts; The modeling-oriented result generation unit is used to use the obtained prompts as model inputs, and use a pre-constructed large language model to generate and output the corresponding industrial process modeling-oriented results in real time.

9. The industrial process modeling-oriented system based on a large language model according to claim 8, characterized in that, The industrial process initialization information includes chemical components, petroleum analysis data selection, physical property method selection, and flash evaporation algorithm selection. The process module topology structure information includes the number of inlets and outlets and the connection status of inlets and outlets of each module. The process module calculation status information includes the calculated status and the uncalculated status; among them, after the status information acquisition unit acquires the industrial process status information, the status information processing unit performs structured preprocessing on the acquired industrial process status information to obtain structured structured process status data.

10. The industrial process modeling guiding system based on a large language model according to claim 8, wherein Industrial process status information includes industrial process initialization data, process module topology structure data, and process module calculation status data; among them, the status information processing unit presets a corresponding JSON structure for the industrial process status information to be structured, and then uses the preset JSON structure to perform structured preprocessing on the obtained industrial process status information, so as to obtain structured process status data in the corresponding JSON structure.

11. The industrial process modeling guidance system based on a large language model according to claim 8, wherein The prompt information extraction unit extracts prompts through the following steps: Step C1: Extract the corresponding status key values for the obtained structured process status data; Step C2: Judge the process status of the corresponding structured process status data according to the extracted status key values, and obtain process status information; among them, the process status information includes the completeness of the process initialization status, the rationality of the component quantity selection, the process module topology structure status, the process module outlet connection status, and the process module calculation status; Step C3: Use a preset python script and prompt template to convert the extracted process status information into a prompt for input to the large language model.

12. The industrial process modeling-oriented system based on a large language model according to claim 8, wherein, The industrial process modeling guidance system based on the large language model pre-constructs the large language model through the following steps: Step B1: Obtain sample industrial process status information, including sample industrial process initialization information, sample process module topology structure information, and sample process module calculation status information; Step B2: Perform data cleaning and structured preprocessing on the obtained sample industrial process status information to obtain structured sample process status data; Step B3: Use a preset python script and combine manual evaluation to convert the structured sample process status data into natural language to obtain the corresponding sample text description; Step B4: Determine the corresponding industrial process modeling guidance label based on the converted sample text description; Step B5: Use the sample text description and the corresponding industrial process modeling guidance label as seed sample data to expand the sample data scale to obtain an expanded sample text description; Step B6: Use the expanded sample text description as the sample data for model training and the industrial process modeling guidance label corresponding to the expanded sample text description as the sample label to train the initial large language model to obtain a large language model for extracting the industrial process modeling guidance result.

13. The industrial process modeling guidance system based on a large language model according to claim 12, characterized in that, In step B5, the industrial process modeling guidance method based on the large language model uses the self-generated instruction framework SELF-INSTRUCT to expand the sample data scale, including: The first step is to migrate the self-generated instruction SELF-INSTRUCT to the single-task multi-instance field to generate the SELF-INSTRUCT single-task instance generation instruction; The second step is to expand the sample data scale based on the SELF-INSTRUCT single-task instance generation instruction.

14. The industrial process modeling guidance method based on a large language model according to claim 12, characterized in that In step B6, the industrial process modeling guidance system based on the large language model performs model training of the large language model through the following steps: The first step is to use the expanded sample text description as the model input and input it into the initial large language model to obtain the corresponding industrial process guidance result and output it; Second, based on the industrial process modeling-oriented tags and industrial process modeling-oriented results corresponding to the augmented sample text, iteratively optimize the model parameters of the initial large language model to obtain a trained large language model.

15. A computer-readable medium storing computer program code, characterized in that, The computer program code, when executed by a processor, implements the method according to any one of claims 1-7.

16. An industrial process modeling guiding device based on a large language model, characterized in that, Comprising: a memory for storing instructions executable by the processor; and a processor for executing the instructions to implement the method according to any one of claims 1-7.

Images