Large-model-based intelligent overall planning and distribution method for iron and steel industry

Abstract

The invention provides an intelligent overall distribution method based on a large model and oriented to the iron and steel industry, and relates to the technical field of artificial intelligence. The intelligent overall distribution method based on the large model for the iron and steel industry is realized by a user side, a task demand input module, a large model semantic understanding module, a sub-model scheduling module and an execution and result analysis module. The method comprises the steps that the user side inputs task demands of iron and steel industry production; the task demand input module performs standardization processing on task demands to obtain standardized text data; the large model semantic understanding module processes the text data to obtain a task target, task model requirements and parameters of a task model; dynamic scheduling is carried out through a sub-model scheduling module, and sub-models are determined; executing the task by the sub-model to obtain an execution result of the sub-model; and the execution and result analysis module processes the execution result and outputs a comprehensive analysis result meeting task requirements. According to the invention, the task processing efficiency can be improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent overall allocation method based on a large model for the steel industry. Background Art

[0002] With the rapid development of modern manufacturing and intelligent production, the number of production processes has increased dramatically, especially in the important industrial fields of steel and metallurgy. The traditional production management model usually relies on manual setting of tasks and manual scheduling of different models for corresponding calculations and analysis. However, with the increasing complexity and diversified needs of industrial production, the existing single task processing method obviously cannot meet the real-time, accurate and efficient task execution requirements of industrial sites.

[0003] In the production process of steel rolled products, managers usually need to monitor and analyze mechanical performance indicators in real time, including yield strength and tensile strength, to make multi-dimensional decisions and scheduling. However, traditional mechanical performance prediction models can only predict a single indicator, and every time the task requirements change, the model must be reconfigured and adjusted, which is cumbersome and inefficient. More importantly, the needs of production management are often complex and dynamic. For example, a user may want to query "the proportion of a batch of rolled products that meet the standard yield strength." This requirement is usually not clearly stated, and results need to be obtained from different data sources and multiple models.

[0004] In the existing steel industry production management system, the single-task learning (STL) method is usually used to process production data. The single-task model is often trained and predicted according to specific production goals. For example, the mechanical property prediction model can usually only focus on a single mechanical property indicator, or requires manual selection of different sub-task models and manual scheduling. The above processing methods have the following problems when facing diversified production tasks: (1) The limitations of the single-task processing mode; the traditional system needs to manually split the task when facing complex tasks, resulting in slow system response and low efficiency; (2) Inflexible model scheduling; in the existing technology, model calling relies on pre-set hard-coded rules, and it is impossible to automatically select and call the appropriate sub-model according to the actual needs of the user; (3) Inaccurate task requirement identification; the existing system cannot intelligently understand the complex task requirements proposed by the user, and requires manual intervention and disassembly and processing through the traditional rule engine, and cannot directly input complex task goals through natural language. The above shortcomings limit the flexibility and adaptability of traditional industrial management systems, making it difficult to meet rapidly changing production needs and diversified task goals. Summary of the invention

[0005] In order to solve the technical problems that the traditional steel industry production management system in the prior art has slow response speed and low efficiency; model calling relies on pre-set hard-coded rules, cannot automatically select and call appropriate sub-models according to the actual needs of users, and cannot intelligently understand the complex task requirements proposed by users, requires manual intervention and disassembly and processing through traditional rule engines, and cannot directly input complex task goals through natural language, the embodiment of the present invention provides a large model-based intelligent overall allocation method and system for the steel industry. The technical solution is as follows:

[0006] On the one hand, a large-model-based intelligent overall allocation method for the steel industry is provided, the method is implemented by a large-model-based intelligent overall allocation device for the steel industry, and the method includes:

[0007] S1, the user terminal inputs the task requirements of steel industry production; the task requirement input module receives the task requirements and obtains standardized text data by standardizing the task requirements;

[0008] S2, the task requirement input module uploads the standardized text data to the large model semantic understanding module, and processes the large model semantic understanding module to obtain task objectives, task model requirements and task model parameters;

[0009] S3, the task requirement input module uploads the task goal, task model requirements and task model parameters to the sub-model scheduling module, and dynamically schedules the sub-model through the sub-model scheduling module to determine the sub-model; according to the task goal, the sub-model executes the task and obtains the execution result of the sub-model;

[0010] S4. The sub-model scheduling module uploads the execution result of the sub-model to the execution and result analysis module, processes it through the execution and result analysis module, and outputs a comprehensive analysis result that meets the task requirements of the steel industry production input by the user end.

[0011] Optionally, the task requirements for steel industry production input by the user terminal in S1 are natural language text.

[0012] Optionally, the step S1 of obtaining standardized text data by standardizing the task requirements includes:

[0013] S11, formatting the input task requirements to obtain formatted text data;

[0014] S12. According to the formatted text data, use the word segmenter of the GPT pre-trained model to perform word segmentation on the formatted text data to obtain a text sequence after word segmentation;

[0015] S13, according to the text sequence after word segmentation, using the NLTK stop word list, the text sequence after word segmentation is processed to remove the stop words, and the text after the stop words are removed is obtained;

[0016] S14. Use the BERT-based dependency analysis model to perform dependency analysis on the text after removing stop words to obtain the grammatical relationship between words in the text data;

[0017] S15. According to the text after the stop words are removed and the grammatical relationship between words in the text data, the GPT pre-training model is used to perform feature encoding to obtain an encoding vector.

[0018] Optionally, the S2 is processed by the large model semantic understanding module to obtain task objectives, task model requirements and task model parameters, including:

[0019] S21, input the encoding vector into the intent classification network to generate a task type label;

[0020] S22. Generate task tag information according to the task type tag;

[0021] S23, according to the coding vector and the task label information, using the information extraction function to convert the content of the task requirement input by the user end into structured data; decomposing the structured data to generate multiple subtask objectives;

[0022] S24. According to the structured data and multiple subtask objectives, a model selection function is used to determine the task model of each subtask and the parameters of each subtask model.

[0023] Optionally, the step S3 of dynamically scheduling by the sub-model scheduling module, determining the sub-model for executing the task, and obtaining the execution result of the sub-model includes:

[0024] Dynamically schedule according to each subtask goal and the parameters of each subtask model to determine the submodel that executes each subtask;

[0025] According to each sub-task goal, each sub-model executes the corresponding sub-task to obtain the execution result of each sub-model.

[0026] Optionally, the step S4 integrates and analyzes the results through the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user end, including:

[0027] S41, according to the execution result of each sub-model, use multimodal data processing technology to perform multimodal preprocessing on the execution result of each sub-model to obtain the preprocessed execution result;

[0028] S42, according to the execution result after preprocessing, using a long short-term memory network to perform feature extraction to obtain key features of each mode;

[0029] S43. According to the key features of each mode, a weighted average fusion method is used to perform fusion processing, and a comprehensive analysis result that meets the task requirements of steel industry production input by the user end is output.

[0030] Optionally, after the step of integrating and analyzing the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user end in S4, the step further includes:

[0031] The intelligent coordination system based on large models for the steel industry receives user feedback on the output comprehensive analysis results;

[0032] According to user feedback, the large-model-based intelligent coordination system for the steel industry dynamically adjusts and optimizes the model scheduling strategy, task processing flow and prediction accuracy.

[0033] On the other hand, a large-model-based intelligent overall allocation system for the steel industry is provided, which is applied to a large-model-based intelligent overall allocation method for the steel industry, and the system includes:

[0034] The user terminal is used for inputting task requirements of steel industry production by the user terminal;

[0035] The task requirement input module is used for the task requirement input module to receive the task requirement and obtain standardized text data by standardizing the task requirement;

[0036] The large model semantic understanding module is used for the task requirement input module to upload the standardized text data to the large model semantic understanding module, and the large model semantic understanding module processes the data to obtain the task objectives, task model requirements and task model parameters;

[0037] The sub-model scheduling module is used for the task requirement input module to upload the task objectives, task model requirements and task model parameters to the sub-model scheduling module, and dynamically schedule the sub-model through the sub-model scheduling module to determine the sub-model; according to the task objectives, the sub-model executes the task and obtains the execution result of the sub-model;

[0038] The execution and result analysis module is used for the sub-model scheduling module to upload the execution results of the sub-model to the execution and result analysis module, and integrate and analyze the results through the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user end.

[0039] Optionally, the task requirements for steel industry production input by the user end are natural language text.

[0040] Optionally, the step of obtaining standardized text data by standardizing the task requirements includes:

[0041] Format the input task requirements to obtain formatted text data;

[0042] According to the formatted text data, the word segmenter of the GPT pre-trained model is used to perform word segmentation on the formatted text data to obtain a text sequence after word segmentation;

[0043] According to the text sequence after word segmentation, the NLTK stop word list is used to remove the stop words from the text sequence after word segmentation to obtain the text after the stop words are removed;

[0044] The BERT-based dependency analysis model is used to perform dependency analysis on the text after the stop words are removed to obtain the grammatical relationship between words in the text data;

[0045] According to the text after removing stop words and the grammatical relationship between words in the text data, the GPT pre-training model is used to perform feature encoding to obtain the encoding vector.

[0046] Optionally, the processing by the large model semantic understanding module to obtain task objectives, task model requirements and task model parameters includes:

[0047] Input the encoded vector into the intent classification network to generate a task type label;

[0048] Generate task label information based on task type label;

[0049] According to the encoding vector and task label information, an information extraction function is used to convert the task requirements input by the user into structured data; the structured data is decomposed to generate multiple sub-task objectives;

[0050] According to the structured data and multiple subtask objectives, a model selection function is used to determine the task model of each subtask and the parameters of each subtask model.

[0051] Optionally, the performing dynamic scheduling through the sub-model scheduling module, determining the sub-model for executing the task, and obtaining the execution result of the sub-model includes:

[0052] Dynamically schedule according to each subtask goal and the parameters of each subtask model to determine the submodel that executes each subtask;

[0053] According to each sub-task goal, each sub-model executes the corresponding sub-task to obtain the execution result of each sub-model.

[0054] Optionally, the integration and analysis processing by the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user terminal includes:

[0055] According to the execution result of each sub-model, multi-modal preprocessing is performed on the execution result of each sub-model using multi-modal data processing technology to obtain the execution result after preprocessing;

[0056] According to the execution results after preprocessing, the long short-term memory network is used for feature extraction to obtain the key features of each mode;

[0057] According to the key features of each mode, a weighted average fusion method is used for fusion processing to output a comprehensive analysis result that meets the task requirements of steel industry production input by the user end.

[0058] Optionally, after the step of integrating and analyzing the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user terminal, the step further includes:

[0059] The intelligent coordination system based on large models for the steel industry receives user feedback on the output comprehensive analysis results;

[0060] According to user feedback, the large-model-based intelligent coordination system for the steel industry dynamically adjusts and optimizes the model scheduling strategy, task processing flow and prediction accuracy.

[0061] On the other hand, a large-scale model-based intelligent coordinated allocation device for the steel industry is provided, and the large-scale model-based intelligent coordinated allocation device for the steel industry comprises: a processor; a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, any one of the above-mentioned large-scale model-based intelligent coordinated allocation methods for the steel industry is implemented.

[0062] On the other hand, a computer-readable storage medium is provided, in which at least one instruction is stored, and the at least one instruction is loaded and executed by a processor to implement any one of the above-mentioned large-model-based intelligent coordinated allocation methods for the steel industry.

[0063] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0064] The embodiment of the present invention first inputs the task requirements of steel industry production through the user end; the task requirement input module receives the task requirements, and obtains standardized text data by standardizing the task requirements; the task requirement input module uploads the standardized text data to the large model semantic understanding module, and processes it through the large model semantic understanding module to obtain the task objectives, task model requirements and task model parameters; secondly, the task requirement input module uploads the task objectives, task model requirements and task model parameters to the sub-model scheduling module, and dynamically schedules the sub-model through the sub-model scheduling module to determine the sub-model; according to the task objective, the sub-model executes the task and obtains the execution result of the sub-model; finally, the sub-model scheduling module uploads the execution result of the sub-model to the execution and result analysis module, and processes it through the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of steel industry production input by the user end.

[0065] The embodiment of the present invention can automatically understand the task objectives when the user proposes a complex task, and intelligently schedule related models to process the task without manual intervention, which greatly improves the efficiency of task processing; through the combination of large models and sub-models, the system can handle multiple task requirements at the same time, and integrate the output results of each task to ensure the accuracy and consistency of the comprehensive analysis results; through natural language processing technology, the system can understand the user's unstructured needs and convert them into structured task instructions, thereby realizing intelligent task allocation and predictive analysis; the feedback and optimization of the system enable the model scheduling strategy and task execution process to be continuously improved, optimizing the system's resource utilization and task response speed. The system architecture designed by the present invention has a high degree of flexibility and scalability, can adapt to various complex steel industry production management tasks, and can be widely used in different industrial fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0067] Figure 1This is a flow chart of an intelligent overall allocation method based on a large model for the steel industry provided by an embodiment of the present invention;

[0068] Figure 2 It is a schematic diagram of the overall process of an intelligent overall allocation method based on a large model for the steel industry provided by an embodiment of the present invention;

[0069] Figure 3 It is a block diagram of an intelligent overall allocation system based on a large model for the steel industry provided by an embodiment of the present invention;

[0070] Figure 4 It is a structural diagram of a large-model-based intelligent overall allocation device for the steel industry provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0071] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0072] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0073] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same. "of", "corresponding, relevant" and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, the meanings they intend to express are the same.

[0074] In the embodiments of the present invention, sometimes a subscript such as W1 may be written as a non-subscript such as W1. When the difference is not emphasized, the meanings to be expressed are the same.

[0075] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0076] The embodiment of the present invention provides a large-model-based intelligent overall allocation method for the steel industry. The method can be implemented by a large-model-based intelligent overall allocation device for the steel industry. The large-model-based intelligent overall allocation device for the steel industry can be a terminal or a server. Figure 1The flowchart of the intelligent overall allocation method based on a large model for the steel industry is shown. The processing flow of the method may include the following steps:

[0077] S1. The user inputs the task requirements of steel industry production; the task requirement input module receives the task requirements and obtains standardized text data by standardizing the task requirements.

[0078] Among them, the task requirement input module is used for users to input task requirements through the system interface; the input form supports complex natural language descriptions without the need for specific instructions or formats, which can improve the flexibility of the system and user experience.

[0079] Optionally, the user terminal of S1 inputs the task requirements of steel industry production as natural language text.

[0080] In a feasible implementation, the user can input the task request through natural language. The system does not need a fixed input format, and the user can flexibly express complex requirements. The input task request includes descriptive requirements, such as "count the number of rolled products whose yield strength meets the standard."

[0081] Optionally, the specific implementation process of S1 may include S11-S15:

[0082] S11, formatting the input task requirements to obtain formatted text data;

[0083] In a feasible implementation, the text of the input task requirement is formatted, including conversion to lowercase, removal of special characters, and deletion of spaces; the consistency of word segmentation is ensured through formatting.

[0084] S12. According to the formatted text data, use the word segmenter of the GPT pre-trained model to perform word segmentation on the formatted text data to obtain a text sequence after word segmentation;

[0085] In a feasible implementation, the process of using the word segmenter of the GPT pre-trained model to segment the formatted text data is expressed by the following formula (1):

[0086] (1)

[0087] Among them, Tokenized_text represents the text sequence after word segmentation; Normalized_text represents the input text.

[0088] Among them, the GPT pre-training model is a model with powerful natural language processing capabilities that can generate text through autoregression.

[0089] In a feasible implementation, a tokenizer of a GPT pre-trained model is used to break down the text into subunits. For example, "user requirement input" is split into ["user", "requirement", "input"]. If there are words not in the dictionary, the tokenizer of the GPT pre-trained model will further break down the subunits into sub-words. The result after tokenization not only retains the original sentence structure but also retains semantic information and is mapped to specific word vectors in the encoding of the GPT pre-trained model for convenient subsequent processing.

[0090] S13. According to the tokenized text sequence, use the NLTK stopword list to perform stopword removal processing on the tokenized text sequence to obtain the text after stopword removal.

[0091] Among them, NLTK (Natural Language Toolkit) is a natural language processing library that can provide stopword lists in multiple languages. Among them, stopwords refer to words that frequently appear in the text but usually have little semantic content, such as stopwords like "de", "shi", and "he".

[0092] In a feasible implementation, stopword removal can reduce unnecessary corpus interference and highlight the important content in the task requirements input by the user. Stopwords include conjunctions and articles.

[0093] Among them, using the NLTK stopword list can ensure that the removed words will not affect the overall semantics of the text. Using the NLTK stopword list to traverse the tokenized text sequence and remove the words in the stopword library to obtain the text after stopword removal can be represented by the following formula (2):

[0094] (2)

[0095] Among them, Filtered_Tokens represents the text after stopword removal; Stopwords represents stopwords.

[0096] Among them, the text after stopword processing can greatly reduce the model calculation burden and highlight the key information in the task request.

[0097] S14. Use a BERT-based dependency parsing model to perform dependency analysis on the text after stopword removal to obtain the grammatical relationships between words in the text data.

[0098] Among them, the BERT-based dependency parsing model is a method of using the BERT model for dependency syntactic analysis. Dependency syntactic analysis is an important task in natural language processing that can identify the dependency relationships between words in a sentence and construct a dependency tree to better understand the structure and meaning of the sentence.

[0099] In a feasible implementation, dependency analysis is used to analyze the grammatical structure of the text and identify the subject, predicate, and object components in the sentence.

[0100] In a feasible implementation, a BERT-based dependency analysis model is used to perform dependency analysis on the formatted text data, and the grammatical relationship between words in the text is marked. For example, the BERT-based dependency analysis model is used to analyze "the number of statistical yield strengths that meet the standards", and the dependency structure obtained is: statistics (predicate), quantity (object), strength (modification); according to the obtained dependency structure, the component categories of the identified text include: noun phrases and verb phrases, and corresponding grammatical tags are added to each word.

[0101] Among them, the dependency analysis process can accurately extract task objectives and requirement content.

[0102] S15. According to the text after the stop words are removed and the grammatical relationship between words in the text data, the GPT pre-training model is used to perform feature encoding to obtain an encoding vector.

[0103] In a feasible implementation, each word after segmentation is converted into a corresponding word vector to obtain a multi-dimensional embedding representation; wherein the word vector retains semantic relations and contextual information, which helps the system understand complex requirements.

[0104] The specific process of obtaining the encoding vector can be expressed by the following formula (3) and formula (4):

[0105] (3)

[0106] (4)

[0107] Among them, Encoded_Text represents the encoded vector; h[CLS] represents the output vector of the [CLS] tag generated by the GPT pre-training model; [CLS] Token represents the [CLS] tag at the first position in the sequence preprocessed by GPT input.

[0108] In a feasible implementation, the encoding vector is generated according to the text sequence using formula (3), and the semantic features of the entire input can be represented by the [CLS] bit vector of GPT, or the overall vector can be extracted through a pooling operation.

[0109] S2. The task requirement input module uploads the standardized text data to the large model semantic understanding module, and processes it through the large model semantic understanding module to obtain the task objectives, task model requirements and task model parameters.

[0110] Among them, the large model is a pre-trained model based on the GPT large language; the large model semantic understanding module is used for the system to call the large-scale pre-trained language model for semantic analysis and intent recognition after receiving the user's task requirements. Through natural language processing technology, the system converts the user's natural language task requirements into structured task goals, and identifies the main task goals by extracting key features of the task requirements input by the user; decomposes the sub-tasks in complex tasks to ensure that each task goal is clear and unambiguous. For requirements that include multiple tasks, the system automatically refines the requirement content and outputs the task structure and key parameters of each task model.

[0111] Optionally, the specific implementation process of S2 may include S21-S24:

[0112] S21, input the encoding vector into the intent classification network to generate a task type label;

[0113] Among them, the intent classification network is a conventional technical means of natural language processing; it is used to identify and understand the intentions expressed by users in text, language or other forms of input, and can be applied to chatbots, voice assistants, search engines, and customer service systems. The intent classification network can accurately respond to user needs.

[0114] Among them, the task type labels include classification, regression and data statistics; the task type label is used to indicate the nature of the task, which is expressed by the following formula (5):

[0115] (5)

[0116] Among them, W represents the weight of the intent classifier; b represents the bias parameter of the intent classifier; Intent_Type represents the task type label; softmax() represents the activation function of the output layer.

[0117] S22. Generate task tag information according to the task type tag;

[0118] The task label information includes: operation keywords and constraints; the task label information is expressed by the following formula (6):

[0119] (6)

[0120] Among them, Generated_Tags represents the generated task tag information; Keywords represents the operation keywords; Constraints represents the constraint conditions.

[0121] S23, according to the coding vector and the task label information, using the information extraction function to convert the content of the task requirement input by the user end into structured data; decomposing the structured data to generate multiple subtask objectives;

[0122] In a feasible implementation, the process of obtaining structured data can be expressed by the following formula (7):

[0123] (7)

[0124] Among them, Structured_Data represents the structured data format; f() represents the information extraction function.

[0125] Among them, the structured data is further decomposed into multiple subtasks, and the operations and goals of each subtask are clearly marked.

[0126] S24. According to the structured data and multiple subtask objectives, a model selection function is used to determine the task model of each subtask and the parameters of each subtask model.

[0127] In a feasible implementation, according to the structured data and multiple subtask objectives, a model selection function is used to extract the model type and key parameters of each subtask. Through model selection, the objectives and operation requirements of each subtask are analyzed, and the corresponding model of each subtask is determined. The model mapping can be expressed by the following formula (8):

[0128] (8)

[0129] Among them, Model represents the selected model; g() represents the model selection function.

[0130] Among them, the methods of model selection function include: cross-validation method, model evaluation index and model selection criteria; among them, the model selection method used in this application is the BIC Bayesian Information Criterion.

[0131] In a feasible implementation, the corresponding model parameters are extracted according to the specific requirements of each subtask, and the key parameter set of the model is expressed by the following formula (9):

[0132] (9)

[0133] Among them, Key_Params represents the key parameter set of the model; h() represents the extraction function of the model parameters.

[0134] S3. The task requirement input module uploads the task objectives, task model requirements and task model parameters to the sub-model scheduling module, and dynamically schedules the sub-model through the sub-model scheduling module to determine the sub-model; according to the task objectives, the sub-model executes the task and obtains the execution result of the sub-model.

[0135] Among them, the sub-model scheduling module is used for the system to automatically select the most suitable model to perform a specific task according to the requirements of each sub-task; the sub-model scheduling module will optimize and allocate models according to conditions such as task complexity, time requirements, and computing resources to ensure efficient completion of tasks; for tasks with complex requirements, the system can call multiple sub-models at the same time for collaborative processing to improve the prediction effect.

[0136] Optionally, the specific implementation process of S3 may include:

[0137] Dynamically schedule according to each subtask goal and the parameters of each subtask model to determine the submodel that executes each subtask;

[0138] According to each sub-task goal, each sub-model executes the corresponding sub-task to obtain the execution result of each sub-model.

[0139] S4. The sub-model scheduling module uploads the execution results of the sub-model to the execution and result analysis module, which processes the sub-model and outputs a comprehensive analysis result that meets the task requirements of the steel industry production input by the user.

[0140] Among them, the execution and result analysis module is used for the system to summarize and analyze the results of all subtasks; through multimodal data fusion technology, the system can integrate the results of different sub-models and conduct in-depth analysis, which can provide more accurate and targeted prediction results; the final output results will be presented as clear reports or visualized data for users to view and understand.

[0141] In a feasible implementation, each sub-model completes its own task according to the task requirements, and each sub-model performs its own task according to the parameters of the task model; after each sub-model completes the task, the execution result obtained is returned to the large model semantic understanding module for preliminary verification, including: checking the format of the data, the validity and rationality of the execution result. Among them, preliminary verification of the execution result can ensure that the execution of all sub-tasks meets expectations.

[0142] For key tasks, the system can verify the accuracy of the execution results based on rule verification or through the main model cross-verification method, which can improve the reliability of the execution results. For example, whether the generated text and annotated image targets match or support specific user needs.

[0143] Optionally, the specific implementation process of S4 may include S41-S43:

[0144] S41, according to the execution result of each sub-model, use multimodal data processing technology to perform multimodal preprocessing on the execution result of each sub-model to obtain the preprocessed execution result;

[0145] In a feasible implementation, the data of each modality is cleaned, a median filter or a Gaussian filter is used to remove noise in the image, a regular expression is used to process special characters and punctuation marks in the text, and missing values ​​in the numerical data are filled by interpolation methods including linear interpolation or K-nearest neighbor interpolation. At the same time, a hash table or set is used to deduplicate, and the preprocessed execution result is obtained.

[0146] In a feasible implementation, multimodal features can be compared and fused in the same dimension by performing multimodal preprocessing on the execution results.

[0147] S42, according to the execution result after preprocessing, using a long short-term memory network to perform feature extraction to obtain key features of each mode;

[0148] Among them, the long short-term memory network is a conventional technical means, and the present invention will not be further elaborated here.

[0149] S43. According to the key features of each mode, the weighted average fusion method is used for fusion processing to output a comprehensive analysis result that meets the task requirements of steel industry production input by the user.

[0150] Among them, the weighted average fusion method is a commonly used image fusion method, which assigns a weight to each image and calculates the weighted average of the corresponding pixel values ​​of all images at each pixel point to obtain the final fused image.

[0151] In a feasible implementation method, in order to ensure the accuracy of the fusion result, the system dynamically adjusts the weight of each feature according to the task requirements and the priority of the modality; uses weighted averaging, attention mechanism or joint embedding method to fuse multimodal data to obtain the final fusion result; and uses the final fusion result as a comprehensive analysis result that meets the task requirements of steel industry production input by the user.

[0152] In a feasible implementation, Figure 2 It is a general flow chart of an intelligent coordinated allocation method based on a big model for the steel industry provided by an embodiment of the present invention; in a feasible implementation mode, flexible task requirements are input, and according to the task requirements, the system performs semantic understanding of the big model, performs big model analysis by calling the interface, and obtains comprehensive analysis results; wherein the calling interface includes: data model, picture model and text model.

[0153] Optionally, after the step of integrating and analyzing the execution with the result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user end in S4, the step further includes:

[0154] The intelligent coordination system based on large models for the steel industry receives user feedback on the output comprehensive analysis results;

[0155] In a feasible implementation, the system receives user feedback on output results, records task execution results and user feedback, and facilitates continuous optimization of system performance.

[0156] According to user feedback, the large-model-based intelligent coordination system for the steel industry dynamically adjusts and optimizes the model scheduling strategy, task processing flow and prediction accuracy.

[0157] In a feasible implementation, by analyzing the feedback data, the system can dynamically adjust and optimize the model scheduling strategy, task processing flow and prediction accuracy, thereby improving the task execution effect; the system can monitor the performance indicators of the model in real time, including response time and resource utilization, dynamically adjust the computing resource allocation according to the actual load, and use multi-model integration technology to improve the stability and prediction accuracy of the system. At the same time, optimize the task processing process, introduce workflow management and anomaly detection mechanisms, realize the automation of task processing, and respond quickly to abnormal situations. By establishing a continuous learning mechanism, regularly collecting user feedback and new data to retrain the model, continuously optimizing feature engineering and hyperparameter settings, ensuring that the model is always in the best state and improving prediction accuracy; through the above steps, the system can maintain efficient and accurate task execution capabilities in a changing environment.

[0158] The embodiment of the present invention first inputs the task requirements of steel industry production through the user end; the task requirement input module receives the task requirements, and obtains standardized text data by standardizing the task requirements; the task requirement input module uploads the standardized text data to the large model semantic understanding module, and processes it through the large model semantic understanding module to obtain the task objectives, task model requirements and task model parameters; secondly, the task requirement input module uploads the task objectives, task model requirements and task model parameters to the sub-model scheduling module, and dynamically schedules the sub-model through the sub-model scheduling module to determine the sub-model; according to the task objective, the sub-model executes the task and obtains the execution result of the sub-model; finally, the sub-model scheduling module uploads the execution result of the sub-model to the execution and result analysis module, and processes it through the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of steel industry production input by the user end.

[0159] The embodiment of the present invention can automatically understand the task objectives when the user proposes a complex task, and intelligently schedule related models to process the task without manual intervention, which greatly improves the efficiency of task processing; through the combination of large models and sub-models, the system can handle multiple task requirements at the same time, and integrate the output results of each task to ensure the accuracy and consistency of the comprehensive analysis results; through natural language processing technology, the system can understand the user's unstructured needs and convert them into structured task instructions, thereby realizing intelligent task allocation and predictive analysis; the feedback and optimization of the system enable the model scheduling strategy and task execution process to be continuously improved, optimizing the system's resource utilization and task response speed. The system architecture designed by the present invention has a high degree of flexibility and scalability, can adapt to various complex steel industry production management tasks, and can be widely used in different industrial fields.

[0160] Figure 3 The block diagram of a large-model-based intelligent overall allocation system for the steel industry is shown according to an exemplary embodiment. The system is used for a large-model-based intelligent overall allocation method for the steel industry. Figure 3 The system includes a user terminal 310, a task requirement input module 320, a large model semantic understanding module 330, a sub-model scheduling module 340, and an execution and result analysis module 350. Among them:

[0161] The user terminal 310 is used for the user terminal to input the task requirements of steel industry production;

[0162] The task requirement input module 320 is used for the task requirement input module to receive the task requirement and obtain standardized text data by standardizing the task requirement;

[0163] The large model semantic understanding module 330 is used for the task requirement input module to upload the standardized text data to the large model semantic understanding module, and the large model semantic understanding module processes the data to obtain the task objectives, task model requirements and task model parameters;

[0164] The sub-model scheduling module 340 is used for the task requirement input module to upload the task objectives, task model requirements and task model parameters to the sub-model scheduling module, and dynamically schedule through the sub-model scheduling module to determine the sub-model for executing the task and obtain the execution result of the sub-model;

[0165] The execution and result analysis module 350 is used for the sub-model scheduling module to upload the execution results of the sub-model to the execution and result analysis module, and integrate and analyze the results through the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user end.

[0166] Optionally, the task requirements for steel industry production input by the user end are natural language text.

[0167] Optionally, the step of obtaining standardized text data by standardizing the task requirements includes:

[0168] Format the input task requirements to obtain formatted text data;

[0169] According to the formatted text data, the word segmenter of the GPT pre-trained model is used to perform word segmentation on the formatted text data to obtain a text sequence after word segmentation;

[0170] According to the text sequence after word segmentation, the NLTK stop word list is used to remove the stop words from the text sequence after word segmentation to obtain the text after the stop words are removed;

[0171] The BERT-based dependency analysis model is used to perform dependency analysis on the text after the stop words are removed to obtain the grammatical relationship between words in the text data;

[0172] According to the text after removing stop words and the grammatical relationship between words in the text data, the GPT pre-training model is used to perform feature encoding to obtain the encoding vector.

[0173] Optionally, the processing by the large model semantic understanding module to obtain task objectives, task model requirements and task model parameters includes:

[0174] Input the encoded vector into the intent classification network to generate a task type label;

[0175] Generate task label information based on task type label;

[0176] According to the encoding vector and task label information, an information extraction function is used to convert the task requirements input by the user into structured data; the structured data is decomposed to generate multiple sub-task objectives;

[0177] According to the structured data and multiple subtask objectives, a model selection function is used to determine the task model of each subtask and the parameters of each subtask model.

[0178] Optionally, the performing dynamic scheduling through the sub-model scheduling module, determining the sub-model for executing the task, and obtaining the execution result of the sub-model includes:

[0179] Dynamically schedule according to each subtask goal and the parameters of each subtask model to determine the submodel that executes each subtask;

[0180] According to each sub-task goal, each sub-model executes the corresponding sub-task to obtain the execution result of each sub-model.

[0181] Optionally, the integration and analysis processing by the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user terminal includes:

[0182] According to the execution result of each sub-model, multi-modal preprocessing is performed on the execution result of each sub-model using multi-modal data processing technology to obtain the execution result after preprocessing;

[0183] According to the execution results after preprocessing, the long short-term memory network is used for feature extraction to obtain the key features of each mode;

[0184] According to the key features of each mode, a weighted average fusion method is used for fusion processing to output a comprehensive analysis result that meets the task requirements of steel industry production input by the user end.

[0185] Optionally, after the step of integrating and analyzing the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user terminal, the step further includes:

[0186] The intelligent coordination system based on large models for the steel industry receives user feedback on the output comprehensive analysis results;

[0187] According to user feedback, the large-model-based intelligent coordination system for the steel industry dynamically adjusts and optimizes the model scheduling strategy, task processing flow and prediction accuracy.

[0188] The embodiment of the present invention first inputs the task requirements of steel industry production through the user end; the task requirement input module receives the task requirements, and obtains standardized text data by standardizing the task requirements; the task requirement input module uploads the standardized text data to the large model semantic understanding module, and processes it through the large model semantic understanding module to obtain the task objectives, task model requirements and task model parameters; secondly, the task requirement input module uploads the task objectives, task model requirements and task model parameters to the sub-model scheduling module, and dynamically schedules the sub-model through the sub-model scheduling module to determine the sub-model; according to the task objective, the sub-model executes the task and obtains the execution result of the sub-model; finally, the sub-model scheduling module uploads the execution result of the sub-model to the execution and result analysis module, and processes it through the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of steel industry production input by the user end.

[0189] The embodiment of the present invention can automatically understand the task objectives when the user proposes a complex task, and intelligently schedule related models to process the task without manual intervention, which greatly improves the efficiency of task processing; through the combination of large models and sub-models, the system can handle multiple task requirements at the same time, and integrate the output results of each task to ensure the accuracy and consistency of the comprehensive analysis results; through natural language processing technology, the system can understand the user's unstructured needs and convert them into structured task instructions, thereby realizing intelligent task allocation and predictive analysis; the feedback and optimization of the system enable the model scheduling strategy and task execution process to be continuously improved, optimizing the system's resource utilization and task response speed. The system architecture designed by the present invention has a high degree of flexibility and scalability, can adapt to various complex steel industry production management tasks, and can be widely used in different industrial fields.

[0190] Figure 4 is a schematic diagram of a structure of a large-model-based intelligent overall allocation device for the steel industry provided by an embodiment of the present invention, such as Figure 4 As shown, the intelligent overall allocation equipment based on large models for the steel industry can include the above Figure 3 The shown intelligent coordinated allocation system based on a large model for the steel industry. Optionally, the intelligent coordinated allocation device 410 based on a large model for the steel industry may include a first processor 2001 .

[0191] Optionally, the large model-based intelligent coordinated allocation device 410 for the steel industry may also include a memory 2002 and a transceiver 2003 .

[0192] The first processor 2001, the memory 2002 and the transceiver 2003 may be connected via a communication bus.

[0193] Combine the following Figure 4 The components of the large-scale model-based intelligent overall allocation device 410 for the steel industry are introduced in detail:

[0194] The first processor 2001 is the control center of the large-scale model-based intelligent overall allocation device 410 for the steel industry, and can be a processor or a general term for multiple processing elements. For example, the first processor 2001 is one or more central processing units (CPUs), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention, such as one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (field programmable gate arrays, FPGAs).

[0195] Optionally, the first processor 2001 can perform various functions of the large model-based intelligent coordinated allocation device 410 for the steel industry by running or executing a software program stored in the memory 2002 and calling data stored in the memory 2002.

[0196] In a specific implementation, as an embodiment, the first processor 2001 may include one or more CPUs, such as Figure 4 CPU0 and CPU1 are shown in FIG.

[0197] In a specific implementation, as an embodiment, the large model-based intelligent overall allocation device 410 for the steel industry may also include multiple processors, such as Figure 4 The first processor 2001 and the second processor 2004 are shown in FIG. Each of these processors may be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0198] The memory 2002 is used to store the software program for executing the solution of the present invention, and is controlled to be executed by the first processor 2001. The specific implementation method can refer to the above method embodiment, which will not be repeated here.

[0199] Optionally, the memory 2002 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, an optical disc storage (including a compressed optical disc, a laser disc, an optical disc, a digital versatile disc, a Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of an instruction or data structure and can be accessed by a computer, but is not limited thereto. The memory 2002 may be integrated with the first processor 2001, or may exist independently, and may be connected to the first processor 2001 through the interface circuit ( Figure 4 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0200] The transceiver 2003 is used to communicate with a network device or a terminal device.

[0201] Optionally, the transceiver 2003 may include a receiver and a transmitter ( Figure 4 The receiver is used to implement a receiving function, and the transmitter is used to implement a sending function.

[0202] Optionally, the transceiver 2003 may be integrated with the first processor 2001, or may exist independently, and may be connected to the first processor 2001 through the interface circuit ( Figure 4 (not shown) is coupled to the first processor 2001, which is not specifically limited in this embodiment of the present invention.

[0203] It should be noted that Figure 4 The structure of the large model-based intelligent coordinated allocation device 410 for the steel industry shown in the figure does not constitute a limitation on the router. The actual knowledge structure recognition device may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0204] In addition, the technical effects of the large-model-based intelligent coordinated allocation device 410 for the steel industry can refer to the technical effects of the large-model-based intelligent coordinated allocation method for the steel industry described in the above method embodiment, and will not be repeated here.

[0205] It should be understood that the first processor 2001 in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0206] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0207] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0208] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0209] In the present invention, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0210] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0211] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0212] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0213] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of the system or unit, which can be electrical, mechanical or other forms.

[0214] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0215] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0216] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0217] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An intelligent overall allocation method based on a large model for the steel industry, characterized in that: The intelligent overall planning allocation method based on a large model for the steel industry is implemented by an intelligent overall planning system based on a large model for the steel industry. The intelligent overall planning system based on a large model for the steel industry includes: a user end, a task requirement input module, a large model semantic understanding module, a sub-model scheduling module, and an execution and result analysis module; the method includes: S1, the user terminal inputs the task requirements of steel industry production; the task requirement input module receives the task requirements and obtains standardized text data by standardizing the task requirements; S2, the task requirement input module uploads the standardized text data to the large model semantic understanding module, and processes the large model semantic understanding module to obtain task objectives, task model requirements and task model parameters; S3, the task requirement input module uploads the task goal, task model requirements and task model parameters to the sub-model scheduling module, and dynamically schedules the sub-model through the sub-model scheduling module to determine the sub-model; according to the task goal, the sub-model executes the task and obtains the execution result of the sub-model; S4. The sub-model scheduling module uploads the execution result of the sub-model to the execution and result analysis module, processes it through the execution and result analysis module, and outputs a comprehensive analysis result that meets the task requirements of the steel industry production input by the user end.

2. The intelligent overall allocation method based on a large model for the steel industry according to claim 1 is characterized in that: The user terminal in S1 inputs the task requirements for steel industry production in the form of natural language text.

3. The intelligent overall allocation method based on a large model for the steel industry according to claim 2 is characterized in that: The step S1 obtains standardized text data by standardizing the task requirements, including: S11, formatting the input task requirements to obtain formatted text data; S12. According to the formatted text data, use the word segmenter of the GPT pre-trained model to perform word segmentation on the formatted text data to obtain a text sequence after word segmentation; S13, according to the text sequence after word segmentation, using the NLTK stop word list, the text sequence after word segmentation is processed to remove the stop words, and the text after the stop words are removed is obtained; S14. Use the BERT-based dependency analysis model to perform dependency analysis on the text after removing stop words to obtain the grammatical relationship between words in the text data; S15. According to the text after the stop words are removed and the grammatical relationship between words in the text data, the GPT pre-training model is used to perform feature encoding to obtain an encoding vector.

4. The intelligent overall allocation method based on a large model for the steel industry according to claim 3 is characterized in that: The S2 is processed by the large model semantic understanding module to obtain the task objectives, task model requirements and task model parameters, including: S21, input the encoding vector into the intent classification network to generate a task type label; S22. Generate task tag information according to the task type tag; S23, according to the coding vector and the task label information, using the information extraction function to convert the content of the task requirement input by the user end into structured data; decomposing the structured data to generate multiple subtask objectives; S24. According to the structured data and multiple subtask objectives, a model selection function is used to determine the task model of each subtask and the parameters of each subtask model.

5. The intelligent overall allocation method based on a large model for the steel industry according to claim 4 is characterized in that: The S3 performs dynamic scheduling through the sub-model scheduling module, determines the sub-model for executing the task, and obtains the execution result of the sub-model, including: Dynamically schedule according to each subtask goal and the parameters of each subtask model to determine the submodel that executes each subtask; According to each sub-task goal, each sub-model executes the corresponding sub-task to obtain the execution result of each sub-model.

6. The intelligent overall allocation method based on a large model for the steel industry according to claim 5 is characterized in that: The S4 integrates and analyzes the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user end, including: S41, according to the execution result of each sub-model, use multimodal data processing technology to perform multimodal preprocessing on the execution result of each sub-model to obtain the preprocessed execution result; S42, according to the execution result after preprocessing, using a long short-term memory network to perform feature extraction to obtain key features of each mode; S43. According to the key features of each mode, a weighted average fusion method is used to perform fusion processing, and a comprehensive analysis result that meets the task requirements of steel industry production input by the user end is output.

7. The intelligent overall allocation method based on a large model for the steel industry according to claim 1 is characterized in that: After the step of integrating and analyzing the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user end in S4, the step further includes: The intelligent coordination system based on large models for the steel industry receives user feedback on the output comprehensive analysis results; According to user feedback, the large-model-based intelligent coordination system for the steel industry dynamically adjusts and optimizes the model scheduling strategy, task processing flow and prediction accuracy.

8. An intelligent overall allocation system based on a large model for the steel industry, the intelligent overall allocation system based on a large model for the steel industry is used to implement the intelligent overall allocation method based on a large model for the steel industry as claimed in any one of claims 1 to 7, characterized in that: The system comprises: The user terminal is used for inputting task requirements of steel industry production by the user terminal; The task requirement input module is used for the task requirement input module to receive the task requirement and obtain standardized text data by standardizing the task requirement; The large model semantic understanding module is used for the task requirement input module to upload the standardized text data to the large model semantic understanding module, and the large model semantic understanding module processes the data to obtain the task objectives, task model requirements and task model parameters; The sub-model scheduling module is used for the task requirement input module to upload the task objectives, task model requirements and task model parameters to the sub-model scheduling module, and dynamically schedule the sub-model through the sub-model scheduling module to determine the sub-model; according to the task objectives, the sub-model executes the task and obtains the execution result of the sub-model; The execution and result analysis module is used for the sub-model scheduling module to upload the execution results of the sub-model to the execution and result analysis module, and integrate and analyze the results through the execution and result analysis module to output a comprehensive analysis result that meets the task requirements of the steel industry production input by the user end.

9. An intelligent overall allocation device based on a large model for the steel industry, characterized in that: The large-model-based intelligent overall allocation equipment for the steel industry includes: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program codes, which can be called by a processor to execute the method according to any one of claims 1 to 7.

Images