Industrial large model platform and system thereof

Abstract

The embodiment of the invention provides an industrial large model platform and a system thereof. The industrial large model platform is used for training an industrial large model. The industrial large model platform comprises an infrastructure layer, a base layer, a model adaptation layer, an interaction layer and an application layer. The infrastructure layer is used for constructing basic resources for training an industrial large model; the base layer is used for carrying out industrial multi-modal pre-training so as to improve the processing capability of an industrial large model on multi-modal data; the model adaptation layer is used for carrying out industrial mechanism embedded fine tuning so as to improve the processing capability of an industrial large model on a specific task and the generalization capability of the industrial large model on an unknown field scene; the interaction layer is used for improving the interaction capability of the industrial large model; and the application layer is used for expressing the functions of the industrial large model. According to the industrial large model platform and the system thereof, the problems that an existing industrial large model is poor in generalization ability, low in training efficiency and lack of industrial mechanism understanding are solved.

Description

Technical Field

[0001] The present application relates to the field of computer applications, and in particular to an industrial large model platform and its system. Background Art

[0002] With the advent of the Internet of Things era, industrial big models are constructed by establishing complex mathematical models and computer simulations to simulate, predict and control the operation and behavior of industrial systems.

[0003] However, existing industrial large models have problems such as poor generalization ability, low training efficiency and lack of understanding of industrial mechanisms. Among them, poor generalization ability means that existing industrial large models can only be applied to a single industrial scenario and lack flexibility and adaptability; low training efficiency means that existing industrial large models need to be trained separately, which is time-consuming and resource-intensive; lack of understanding of industrial mechanisms means that existing industrial large models lack in-depth understanding of the specific mechanisms in the industrial field, resulting in inaccurate and unreliable output results of the model.

[0004] Therefore, the existing large industrial models have poor generalization capabilities, low training efficiency and lack of understanding of industrial mechanisms, which are problems that this application urgently needs to solve. Summary of the invention

[0005] The present application provides an industrial large model platform and its system to solve the problems of poor generalization ability, low training efficiency and lack of understanding of industrial mechanisms of existing industrial large models.

[0006] The first aspect of the embodiment of the present application provides an industrial large model platform, which is used to train industrial large models. The industrial large model platform includes: an infrastructure layer, a base layer, a model adaptation layer, an interaction layer and an application layer;

[0007] The infrastructure layer is used to build basic resources for training large industrial models;

[0008] The base layer is used to perform industrial multimodal pre-training on the preset industrial base large model through the preset multimodal data set to improve the processing ability of the industrial large model on multimodal data; the industrial base large model is the base of the industrial large model, and the industrial base large model has the general solution ability of industrial tasks;

[0009] The model adaptation layer is used to embed and fine-tune the industrial mechanism of the pre-trained industrial base big model through a preset specific data set to obtain the industrial big model; wherein the specific data set includes a task-oriented big model and an industry domain big model; the task-oriented big model is used to fine-tune the multi-task instructions of the pre-trained industrial base big model to improve the processing ability of the industrial big model on specific tasks; the industry domain big model is used to embed industry domain knowledge and fine-tune the adapter of the industrial base big model after multi-task instruction fine-tuning to improve the generalization ability of the industrial big model in unknown domain scenarios;

[0010] The interaction layer is used to improve the interaction capabilities of industrial large models through industrial efficient reasoning; the industrial large models are used to interact with technicians and cyber-physical systems respectively;

[0011] The application layer is used to express the functions of large industrial models.

[0012] Optionally, as shown above, the industrial big model platform, the task-oriented big model includes: intelligent question-answering model, scenario recognition model, process decision model, terminal control model, content generation model and scientific discovery model;

[0013] The functions of the industrial big model include: intelligent question and answer, scenario recognition, process decision-making, terminal control, content generation and scientific discovery.

[0014] Optionally, as shown above, the industrial large model platform, the functions of the industrial large model are used to support applications for R&D design, production and manufacturing, testing, business management and operation and maintenance services.

[0015] Optionally, the industrial large model platform and model adaptation layer shown above are also used to perform scenario knowledge internalization and enhanced self-training on the industrial base large model after industry domain knowledge embedding and adapter fine-tuning through preset limited annotated data, so as to improve the processing capability of the industrial large model in industrial segmentation scenarios.

[0016] Optionally, as shown above, the industrial big model platform, the base layer, is pre-installed with respective tool chains for industrial multimodal pre-training, industrial mechanism embedded fine-tuning and industrial efficient reasoning, as well as multiple industrial template big models; among which, the industrial base big model is one of the multiple industrial template big models.

[0017] Optionally, the industrial large model platform shown above has basic resources including: industrial data, computing resources and industrial knowledge;

[0018] Industrial data is used to provide training data for large industrial models;

[0019] Computing resources are used to provide computing power support for large industrial models;

[0020] Industrial knowledge is used to provide deep logic for large industrial models.

[0021] Optionally, as shown above, the industrial large model platform further includes: a pre-training module supported by the base layer;

[0022] A pre-training module, used to perform industrial multimodal pre-training on the industrial base large model by at least one of a plurality of pre-training methods; wherein the plurality of pre-training methods include: a distributed training method, a transfer learning training method, a self-supervised learning method, and a multi-task training method;

[0023] The pre-training module is also used to set the industrial data called by the industrial base large model and the model parameters of the industrial base large model.

[0024] Optionally, as shown above, the industrial large model platform further includes: an adaptation fine-tuning module supported by the model adaptation layer;

[0025] The adaptive fine-tuning module is used to perform industrial mechanism embedding fine-tuning on the pre-trained industrial base large model through one of a variety of fine-tuning training methods; among them, the various fine-tuning training methods include: industrial knowledge fusion and industrial mechanism embedding method, supervised fine-tuning method, direct preference optimization method and human feedback reinforcement learning method.

[0026] Optionally, as shown above, the industrial large model platform further includes: an interactive reasoning module supported by the interactive layer;

[0027] The interactive reasoning module is used to improve the interactive capabilities of large industrial models through a variety of interactive reasoning optimization methods; among them, the various interactive reasoning optimization methods include: reasoning acceleration, industrial knowledge base retrieval enhancement generation, model orchestration and plug-in mechanism.

[0028] Optionally, as shown above, the industrial large model platform further includes: an evaluation module supported by the application layer;

[0029] The evaluation module is used to evaluate the function of the industrial large model through at least one of multiple evaluation methods; wherein the multiple evaluation methods include: automatic evaluation and manual evaluation.

[0030] A second aspect of the embodiments of the present application provides an industrial large model platform system, the industrial large model platform system comprising: an industrial large model, and a cyber physics system interacting with the industrial large model;

[0031] The industrial large model is a model trained by the industrial large model platform of any one of the first aspects.

[0032] The embodiment of the present application provides an industrial large model platform and its system, including an infrastructure layer, a base layer, a model adaptation layer, an interaction layer and an application layer; the infrastructure layer is used to build basic resources for training industrial large models; the base layer is used to perform industrial multimodal pre-training to improve the processing ability of industrial large models on multimodal data; the model adaptation layer is used to perform industrial mechanism embedded fine-tuning to improve the processing ability of industrial large models on specific tasks and the generalization ability on unknown domain scenes; the interaction layer is used to improve the interaction ability of industrial large models; the application layer is used to express the functions of industrial large models. The following technical effects are achieved: the model adaptation layer improves the generalization ability of industrial large models in unknown domain scenes through industry domain large models, solving the problem of poor generalization ability of existing industrial large models; the model adaptation layer improves the processing ability of industrial large models on specific tasks through industrial base large models, solving the problem of lack of understanding of industrial mechanisms in existing industrial large models; the interaction layer improves the interaction ability of industrial large models through industrial efficient reasoning, solving the problem of low training efficiency of existing industrial large models. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the present application or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0034] Figure 1 An application scenario diagram of the industrial large model platform provided in the embodiment of the present application;

[0035] Figure 2 Structural schematic diagram of the industrial large model platform provided in the embodiment of the present application Figure 1 ;

[0036] Figure 3 Structural schematic diagram of the industrial large model platform provided in the embodiment of the present application Figure 2 .

[0037] Reference numerals:

[0038] 100-Technical staff;

[0039] 200-industrial large model platform; 210-infrastructure layer; 220-base layer; 221-pre-training module; 230-model adaptation layer; 231-adaptation fine-tuning module; 240-interaction layer; 241-interactive reasoning module; 250-application layer; 251-evaluation module;

[0040] 310-cyber system; 320-physical system. DETAILED DESCRIPTION

[0041] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0042] The technical solution of the present application is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0043] In the embodiments of the present application, words such as "first" and "second" are used to distinguish between identical or similar items with substantially the same functions and effects. Those skilled in the art will understand that words such as "first" and "second" do not limit the quantity and execution order, and words such as "first" and "second" do not necessarily limit the difference. It should be noted that in the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary" or "for example" in this application should not be interpreted as being more preferred or more advantageous than other embodiments or design. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way. In the embodiments of the present application, "at least one" refers to one or more, and "more" refers to two or more.

[0044] It should be noted that the "at..." in the embodiment of the present application can be the instant when a certain situation occurs, or it can be a period of time after a certain situation occurs, and the embodiment of the present application does not specifically limit this. In addition, the industrial large model platform provided in the embodiment of the present application is only used as an example, and the industrial large model platform can also include more or less content.

[0045] In order to clearly describe the technical solutions of the embodiments of the present application, some terms and technologies involved in the embodiments of the present application are briefly introduced below:

[0046] Internet of Things (IoT): refers to a network that connects any object to the Internet through information sensing devices, such as those based on Radio Frequency Identification (RFID), Global Navigation Satellite System (GNSS), infrared sensing or laser scanning, in accordance with agreed protocols, to exchange information and communicate, so as to achieve intelligent identification, positioning, tracking, monitoring and management. The application fields of IoT are wide, including aerospace, shipping, automobile, chemical industry, metallurgy and steel, etc.

[0047] The generalization ability of industrial models refers to the applicability and accuracy of industrial models in unknown domain scenarios. Industrial models with good generalization ability can not only accurately simulate, predict and control industrial scenarios that have been seen in the training data set, but also make reasonable predictions and judgments about unknown domain scenarios. This ability helps industrial models maintain high efficiency and reliability in complex and changing industrial environments.

[0048] Industrial Mechanism: refers to the professional knowledge such as principles, theorems and laws in the production process of the industrial field. Industrial mechanism includes thermodynamic equations and chemical reaction kinetic equations, etc.

[0049] Understanding of the industrial mechanism of the industrial model: refers to the understanding and grasp of the interaction and influence mechanism between the various components within the industrial system during the construction and operation of the industrial model. Through in-depth understanding, the behavior of the industrial system can be simulated and predicted more accurately, and the industrial mechanism of the model can be improved. The industrial model can be more accurate and reliable. At the same time, this understanding also helps to discover problems and bottlenecks in the industrial system, providing strong support for optimizing and improving the industrial system.

[0050] In order to clearly understand the technical solution of the present application, the solution of the prior art is first introduced in detail.

[0051] Industrial big models refer to models that can simulate, predict and control the operation of industrial systems by using complex mathematical models and computer simulation. Industrial big models usually involve knowledge from multiple disciplines and aim to accurately grasp the overall performance and behavior of industrial systems through highly integrated and comprehensive analysis methods.

[0052] However, existing industrial large models have problems such as poor generalization ability, low training efficiency and lack of understanding of industrial mechanisms. Among them, poor generalization ability means that existing industrial large models can only be applied to a single industrial scenario and lack flexibility and adaptability; low training efficiency means that existing industrial large models need to be trained separately, which is time-consuming and resource-intensive; lack of understanding of industrial mechanisms means that existing industrial large models lack in-depth understanding of the specific mechanisms in the industrial field, resulting in inaccurate and unreliable output results of the model.

[0053] The research found that, in order to address the problem of poor generalization ability, the generalization ability of industrial large models can be improved through pre-set large models and diversified training methods, so that they can better adapt to different industrial application scenarios; in order to address the problem of low training efficiency, efficient model training methods can be used to achieve fast training of industrial large models on the platform, improve training efficiency, and reduce time and costs; in order to address the problem of lack of understanding of industrial mechanisms, the mechanism embedded fine-tuning method of the industrial knowledge base can be introduced to enhance the industrial large model's understanding of industrial mechanisms and improve the accuracy and effectiveness of the model in practical applications.

[0054] Based on the above creative discoveries, the industrial large model platform and its system of the present application are proposed. The industrial large model platform is a one-stop large model development and service operation platform, covering multiple key links such as model training, model fine-tuning, model reasoning and model evaluation. In the model training stage, it not only provides data set management and model fine-tuning functions, but also pre-sets multiple large models for users to choose from, supporting a variety of training methods to meet different needs; in the model reasoning stage, the industrial large model platform improves the reasoning speed and the quality of model output through a variety of interactive reasoning optimization methods; in the model evaluation stage, the industrial large model platform provides two methods: automatic evaluation and manual evaluation, to ensure a comprehensive and accurate understanding of the model's performance.

[0055] The following introduces the application scenarios of the industrial large model platform provided in the embodiments of the present application. Figure 1 This is an application scenario diagram of the industrial large model platform provided in the embodiment of this application. It should be noted that: Figure 1 What is shown are merely examples of scenarios to which the embodiments of the present application can be applied, to help those skilled in the art understand the technical content of the present application, but it does not mean that the embodiments of the present application cannot be used in other devices, systems, environments or scenarios.

[0056] like Figure 1 As shown, the application scenario includes three interactive objects: technicians 100, industrial large model platforms 200 and cyber-physical systems (CPS). The three form an organic whole and interactively assist the industrial production process.

[0057] The technicians 100 can be various specialized personnel, including R&D personnel, engineering personnel, production operators, and business managers, etc. The technicians 100 are used to train the industrial big model through the industrial big model platform 200 and interact with the industrial big model according to the different goals and plans of specific industrial tasks.

[0058] The cyber-physical system is a complex system that includes computing, networking, and physical entities. It realizes the interaction between cyberspace and the physical world through the organic integration and deep collaboration of 3C (Computing, Communication, Control) technologies. In the cyber-physical system, the cyber system 310 (Cyber ​​System) and the physical system 320 (Physical System) are two important components, which are the objects of perception, action, and control of industrial large models.

[0059] Cybersystem 310 refers to technologies and systems related to computing, communication, and control, covering multiple aspects such as computer networks, information systems, and control algorithms. It is the core part of the cyberphysical system responsible for data processing, information transmission, and decision-making. Cybersystem 310 may include memory, computing units, databases, industrial software, and industrial networks.

[0060] The physical system 320 refers to a system that exists in the real world and has physical properties and dynamic behaviors. The physical system 320 may include a gear shifter, an engine speed governor, a robot, a machine tool, an automatic guided vehicle, and the like.

[0061] The embodiments of the present application are introduced below in conjunction with the drawings in the specification. Figure 2 Structural schematic diagram of the industrial large model platform provided in the embodiment of the present application Figure 1 The industrial large model platform provided in the embodiment of the present application is used to train industrial large models, such as Figure 2 As shown, the industrial large model platform 200 includes: an infrastructure layer 210, a base layer 220, a model adaptation layer 230, an interaction layer 240 and an application layer 250.

[0062] The infrastructure layer 210 is used to build basic resources for training large industrial models.

[0063] Specifically, the infrastructure layer 210 is a key layer when training industrial large models. In industrial scenarios, especially in scenarios that require processing massive amounts of data and performing complex computing tasks, such as aerospace, shipping, automobiles, chemicals, metallurgy, and steel, building industrial large models has extremely high requirements for storage resources, computing resources, and data processing capabilities. The infrastructure layer 210 is designed and deployed to meet these requirements. The infrastructure layer 210 ensures the smooth progress of the training process and provides a solid foundation for the final application of industrial large models.

[0064] The base layer 220 is used to perform industrial multimodal pre-training on the preset industrial base large model through the preset multimodal data set to improve the processing capability of the industrial large model on multimodal data; wherein the industrial base large model is the base of the industrial large model, and the industrial base large model has the general solution capability of industrial tasks.

[0065] Specifically, the base layer 220 is the core support for training industrial large models. The pre-training of industrial base large models involves the use of large-scale, multimodal data sets that are not related to downstream tasks to perform initial training of industrial large models, namely industrial multimodal pre-training, so that industrial large models have the general processing capabilities to understand and process multimodal data in industrial scenarios, and improve them. Among them, the multimodal data set can be preset when building the industrial large model platform, or it can be preset by the user when training the industrial large model according to the industrial large model platform.

[0066] At least one large model may be preset in the base layer 220. The industrial base large model may be selected by the user from these large models, or by the user through a selection algorithm. The industrial base large model has the general solution capability for industrial tasks, providing a basis for more sophisticated model adaptation and scenario application of the industrial large model.

[0067] Furthermore, the base layer 220 is used to perform industrial multimodal pre-training on the preset industrial base large model or the user-built large model through the preset multimodal data set and / or the user-built data set, so as to improve the processing capability of the industrial large model on multimodal data.

[0068] Specifically, the base layer 220 can perform industrial multimodal pre-training through the above-mentioned multimodal data set, or a data set constructed by the user, that is, a user-built data set, or a data set including all data of the multimodal data set and the user-built data set. It should be noted that when performing industrial multimodal pre-training through a user-built data set, the degree of improvement in the processing capability of the industrial large model on multimodal data is affected by the data value of the user-built data set.

[0069] The training object of industrial multimodal pre-training can be the above-mentioned industrial base large model or a user-built large model; wherein the user-built large model can be a large model built by the user himself or a large model built by the user based on the industrial base large model. It should be noted that when the user-built large model is used as the training object, the degree of improvement in the processing ability of the pre-trained user-built large model on multimodal data is affected by the modeling degree of the user-built large model.

[0070] The model adaptation layer 230 is used to embed industrial mechanisms and fine-tune the pre-trained industrial base big model through a preset specific data set to obtain an industrial big model; wherein the specific data set includes a task-oriented big model and an industry domain big model; the task-oriented big model is used to fine-tune the multi-task instructions of the pre-trained industrial base big model to improve the processing capability of the industrial big model on specific tasks; the industry domain big model is used to embed industry domain knowledge and fine-tune the adapter of the industrial base big model after multi-task instruction fine-tuning to improve the generalization capability of the industrial big model in unknown domain scenarios.

[0071] Specifically, after pre-training, the industrial large model can obtain a certain degree of universal generalization ability to cope with industrial tasks. However, in industrial multi-scenario tasks, the industrial large model usually lacks professional knowledge and mechanisms in the industrial field, making it difficult to accurately understand and handle industrial problems, and the credibility of the output results is poor. Therefore, multi-task instruction fine-tuning is required.

[0072] The model adaptation layer 230 is the backbone of training the industrial large model. The secondary training of the industrial base large model, i.e., the embedded fine-tuning of the industrial mechanism, involves using a small-scale specific data set to adapt the pre-trained industrial base large model to industrial tasks and industry fields, so that the industrial large model has the processing ability for specific tasks and the generalization ability in unknown field scenarios, and improves it.

[0073] Industrial task adaptation refers to fine-tuning multi-task instructions based on the pre-trained industrial base large model through the task-oriented large model, so as to enable the industrial large model to have the processing capability for specific tasks and improve it while retaining the general solution capability of the industrial task.

[0074] Industry domain adaptation refers to embedding industry domain knowledge and fine-tuning the adapter based on the industrial base model after fine-tuning the multi-task instructions, so that the industrial model has the ability to generalize in unknown domain scenarios and improve it. Among them, industry fields include discrete manufacturing fields such as aerospace, shipbuilding and automobiles, as well as process manufacturing fields such as chemicals, metallurgy and steel.

[0075] The method of instruction fine-tuning commonly used by industrial large models is to fine-tune the pre-trained industrial base large model on a set of pairs consisting of human instructions and expected outputs. In order to integrate industrial knowledge into the industrial large model, the industrial large model of the embodiment of the present application also requires industrial knowledge fusion and industrial mechanism embedding methods compared to the industrial large model of the prior art, namely: first, it is necessary to collect relevant industrial knowledge from multiple sources, including empirical knowledge such as process flow and technical specifications. Secondly, convert these industrial knowledge into a form that the industrial large model can understand and process, such as creating triple data in the knowledge graph (including head entity, relationship and tail entity), and using knowledge graph embedding technology to convert knowledge into embedded vector features for training. Finally, the industrial large model can be integrated with industrial knowledge by integrating the knowledge graph into the training target and fine-tuning the instructions of the knowledge graph.

[0076] In order to embed the equations of industrial mechanisms into large industrial models, the industrial mechanisms can be represented as feature information that can be recognized by the neural network and embedded in the network architecture. Secondly, in the model output stage, the model output is guided to conform to the laws of industrial physics, such as adding mechanism equations to the loss function to penalize output results that violate the laws of physics. In addition, the output of the constraint model can be checked to conform to the industrial mechanism according to the rules in the mechanism knowledge base, and the credibility and accuracy of the model output can also be verified by simulation tools or manual work.

[0077] In order to make the industrial big model more adaptable when applied to various industries, further industry knowledge embedding fine-tuning is required so that the industrial big model can be more proficient in the expertise of a certain industry and thus perform well in various industries. In the fine-tuning process of the industry domain big model, the use of adapter fine-tuning can enhance the adaptability of the industrial big model to different industries while retaining the task-oriented big model fine-tuning. Next, it is necessary to prepare detailed industry datasets and fine-tune the model performance on these industry-specific datasets to improve the performance of the industrial big model in industry-specific problems.

[0078] Furthermore, the model adaptation layer 230 is used to embed industrial mechanisms and fine-tune the pre-trained industrial base large model or the user-built large model through preset specific data sets and / or user-built data sets to obtain the industrial large model.

[0079] The industrial mechanism embedded fine-tuning provided in this embodiment has an implementation principle and technical effects similar to those of the industrial multimodal pre-training in the above-mentioned embodiments, and will not be described in detail in this embodiment.

[0080] The interaction layer 240 is used to improve the interaction capability of the industrial large model through industrial efficient reasoning; wherein the industrial large model is used to interact with technicians and cyber-physical systems respectively.

[0081] Specifically, in the interaction layer 240, technicians, industrial large models and cyber-physical systems are the three major interaction objects that interact and collaborate.

[0082] Industrial efficient reasoning enables industrial large models to process data quickly and accurately and make decisions in complex industrial environments. Industrial efficient reasoning can be achieved based on methods such as model compression, hardware acceleration, and industrial retrieval enhancement generation.

[0083] Furthermore, the interaction layer 240 trains the industrial large model into an intelligent agent, and the technicians, intelligent agents and cyber-physical systems interact and collaborate. Among them, the intelligent agent integrates several basic capabilities of the industrial large model and enhances memory, planning, action and perception to interact with the outside world. Compared with the industrial large model, the intelligent agent has a further ability to actively perceive and control the industrial environment, and can spontaneously remember, observe and influence the external environment.

[0084] The application layer 250 is used to express the functions of the industrial large model.

[0085] Specifically, the application layer 250 is the representation of the industrial big model on the server side. The functions of the industrial big model enable the industrial big model to not only answer complex queries, but also understand and analyze complex industrial environments, make scientific process decisions, directly control cyberphysical systems, and automatically generate technical documents and simulation designs.

[0086] The industrial large model platform of the embodiment of the present application includes an infrastructure layer, a base layer, a model adaptation layer, an interaction layer and an application layer; the infrastructure layer is used to build basic resources for training industrial large models; the base layer is used to perform industrial multimodal pre-training to improve the processing ability of industrial large models on multimodal data; the model adaptation layer is used to perform embedded fine-tuning of industrial mechanisms to improve the processing ability of industrial large models on specific tasks and the generalization ability on unknown domain scenes; the interaction layer is used to improve the interaction ability of industrial large models; the application layer is used to express the functions of industrial large models. The following technical effects are achieved: the model adaptation layer improves the generalization ability of industrial large models in unknown domain scenes through industry domain large models, solving the problem of poor generalization ability of existing industrial large models; the model adaptation layer improves the processing ability of industrial large models on specific tasks through industrial base large models, solving the problem of lack of understanding of industrial mechanisms in existing industrial large models; the interaction layer improves the interaction ability of industrial large models through industrial efficient reasoning, solving the problem of low training efficiency of existing industrial large models.

[0087] In a possible design, the task-oriented large model includes: intelligent question-answering model, scenario cognition model, process decision model, terminal control model, content generation model and scientific discovery model;

[0088] The functions of the industrial big model include: intelligent question and answer, scenario recognition, process decision-making, terminal control, content generation and scientific discovery.

[0089] Specifically, the intelligent question-answering function means that the industrial big model can understand and answer complex questions related to the industry and provide instant information support. The industrial big model has strong text understanding and generation capabilities, combined with specific knowledge in the industrial field and related field technologies, and can provide question-answering services for the entire process anytime and anywhere.

[0090] Scene recognition function means that the industrial large model can recognize and understand various scenes and states in the industrial environment, providing a basis for further analysis and decision-making. The industrial large model has a strong perception ability of information such as vision and sensing, and incorporates it into the decision-making process to optimize the analysis and decision-making effect.

[0091] The process decision-making function refers to the ability of the industrial big model to make optimized decisions in complex industrial processes and improve production efficiency and quality. Based on its powerful logical reasoning and calculation capabilities, the industrial big model can provide reasonable solutions to various optimization problems abstracted from the industrial process.

[0092] The terminal control function refers to the ability of the industrial model to directly control industrial equipment and realize automated operation. Embodied intelligence is one of the important capabilities of the industrial model. Based on the decision results and processing plans, the industrial model can directly control the production equipment and complete industrial tasks.

[0093] The content generation function refers to the ability of the industrial big model to create technical documents, design plans, reports and other content to improve work efficiency. Based on its huge network architecture and through a large amount of data training, the industrial big model has a strong content generation capability and is widely used in applications such as code generation, image generation and time series data generation.

[0094] The scientific discovery function refers to the ability of industrial big models to have advanced association capabilities, thereby promoting the development and application of new materials, new processes, new products and new models. For example, through large-scale cross-disciplinary knowledge integration, industrial big models can identify the coupling mechanisms of complex products in multiple fields such as mechanics, electricity, fluids, heat, gas and magnetism, clarify the physical and chemical principles in new product design, and provide support for innovative process design solutions.

[0095] The technical effect of the solution in this embodiment is: through the task-oriented big model, the industrial big model has the functions of intelligent question answering, scene recognition, process decision-making, terminal control, content generation and scientific discovery.

[0096] In one possible design, the functions of the industrial big model are used to support R&D design, production and manufacturing, experimental testing, business management and operation and maintenance services.

[0097] Specifically, the functions of the industrial big model are aimed at the entire life cycle of the industrial manufacturing industry, and are used to support the five industrial links of R&D design, production and manufacturing, testing, business management, and operation and maintenance services.

[0098] R&D design involves the concept formation, principle design, prototyping, and simulation design of new products to determine the final product specifications and performance to ensure that the product can meet market demand and technical standards. In the R&D design stage, industrial large models mainly provide process assistance for design guidance, solution generation, and auxiliary design. The functions that can be realized by R&D design include intelligent question-and-answer assistants and industrial content generation.

[0099] For intelligent question-and-answer assistants, the industrial big model has a comprehensive and professional industrial knowledge base. It can serve as an industrial knowledge question-and-answer assistant, provide corresponding suggestions on the design processes, parameter types, and design criteria of various products, and combine industrial knowledge and mechanism cognition to demonstrate and check the rationality of design solutions, thereby simplifying the design process.

[0100] For industrial content generation, based on the formation of the design scheme, the industrial big model uses its own content generation capabilities to generate corresponding engineering files from reasonable design results, such as Computer Aided Design (CAD) files or Computer Aided Engineering (CAE) files, etc., to form a seamless connection with other industrial software and prepare for the subsequent production stage. At the same time, based on its own learning of industrial field knowledge and mechanisms, the industrial big model can spontaneously and innovatively explore and generate new structures, new integration schemes and new concepts, and provide designers with innovative ideas. In addition, the industrial big model can design the corresponding simulation system, simulate and analyze the design scheme, verify the rationality of the parameters and structural design, and iterate the design scheme with the simulation analysis results.

[0101] Manufacturing is the process of transforming R&D designed products into actual items, including raw material procurement, production planning, production line optimization, processing and assembly, and quality inspection. In the manufacturing stage, industrial large models mainly assist in product process generation, production equipment control, production planning, and production defect identification. The functions that can be realized by manufacturing include operation process guidance and production process control.

[0102] For operation process guidance, the industrial model has the ability of professional processing technology knowledge, can guide the process design process, and interact with personnel to achieve operation process guidance, personnel training and status feedback interaction, etc. At the same time, the industrial model monitors the production process, and personnel can inquire about the production status of the industrial model in real time to control the production rhythm.

[0103] For production process control, terminal control is the key application of industrial large models in the manufacturing stage. Terminal control mainly includes two parts: production equipment motion control and production process control. Based on perception and decision-making capabilities, industrial large models form intelligent control solutions for equipment, such as robot arm movement speed, acceleration and direction, etc., to improve the performance, efficiency and flexibility of the motion system to better complete specific manufacturing tasks.

[0104] Experimental testing is to test and evaluate the performance, reliability and safety of manufactured products to ensure that the products meet the design requirements and industry standards and can meet the needs of users. In the experimental testing phase, the industrial large model mainly assists in the process of test plan design, test equipment control, test result analysis and test report generation. The functions that can be achieved by experimental testing include test environment recognition and test report generation.

[0105] In terms of test environment cognition, experimental testing involves various changing test environments. In this process, the industrial large model perceives environmental changes through various sensor data, further responds to the environment, and analyzes the corresponding test results. At the same time, the industrial large model also monitors the tested products in real time, perceives changes in product performance, and identifies product status characteristics.

[0106] For test report generation, the industrial big model receives detailed test requirements and design parameters, can generate specific test plans, and further generate test files executable by the equipment. After the test is completed, it generates a corresponding test analysis report to assist R&D personnel in product optimization design.

[0107] Operation management is the process of planning, organizing, coordinating, controlling and optimizing various aspects such as production, sales, finance and human resources in order to achieve its business goals. In the operation management stage, the industrial big model mainly assists in resource scheduling, production planning, document generation and report analysis. The functions that can be realized by operation management include operation status Q&A and operation report generation.

[0108] As for questions and answers about operational situations, the industrial big model is integrated into the entire production process. Managers can understand the production situation at any time through the industrial big model. From raw material supply to production line processing to product sales, they can interact with the industrial big model in the form of questions and answers. At the same time, it also includes auxiliary personnel to retrieve various industrial documents, which greatly facilitates managers' control over the entire process.

[0109] For the generation of operational reports, the industrial big model combines perception and decision-making capabilities to generate specific scheduling plans, which can be generated according to industrial documents or in conjunction with other industrial software requirements. In addition, managers can use the industrial big model to produce production management reports. The industrial big model can generate reports containing rich information according to the specific needs of managers, thereby improving management efficiency.

[0110] Operation and maintenance services are technical support and after-sales services such as installation, commissioning, repair, maintenance and upgrades provided by manufacturing enterprises to users after the products are sold to ensure that the products can operate normally and stably; at the same time, it also includes continuous monitoring of the quality of manufacturing equipment and products within the enterprise. In the operation and maintenance service stage, the industrial big model mainly assists in the process of fault diagnosis, after-sales service and intelligent customer service. The functions that can be realized by business management include intelligent after-sales service and intelligent fault-tolerant maintenance.

[0111] For intelligent after-sales service, the industrial big model monitors the operating status of equipment and products, provides fault warnings and maintenance questions and answers, and can interact with customers as an intelligent after-sales customer service, sort out various email summaries, and reduce labor costs.

[0112] For intelligent fault-tolerant maintenance, when equipment fails or performance degrades, the industrial large model can perform corresponding operational fault-tolerant control to ensure that the equipment can still maintain its specified functions or continue to work at the expense of performance when a failure occurs. Based on intelligent detection and diagnosis of perception and decision-making, corresponding control measures are taken to respond to system failures to ensure the safety and reliability of equipment operation, such as adjusting equipment operating parameters, slowing down production pace, switching to backup systems and timely warning of failures, etc., to provide time for maintenance and reduce the accident rate.

[0113] The technical effect of the solution in this embodiment is to apply the functions of the industrial large model to specific applications.

[0114] In a possible design, the model adaptation layer 230 is also used to perform scenario knowledge internalization and enhanced self-training on the industrial base large model after industry domain knowledge embedding and adapter fine-tuning through preset limited annotated data, so as to improve the processing capability of the industrial large model in industrial segmentation scenarios.

[0115] Specifically, the internalization of scene knowledge and enhanced self-training are designed to address the problem of limited labeled data for some segmented scenes in industrial industry knowledge. Through preliminary training on small-scale labeled data, the model is used to generate pseudo-labels to label large-scale unlabeled data, and reinforcement learning is introduced to set a reward mechanism to optimize the quality of pseudo-labels, thereby improving model performance. In the case of limited labeled data, the internalization of scene knowledge and enhanced self-training can support the fine-tuning of large scene models that solve industrial segmented scene problems, so as to improve the processing capabilities of large industrial models in industrial segmented scenes.

[0116] The technical effect of the solution in this embodiment is: strengthening self-training through internalization of scenario knowledge and improving the processing capabilities of large industrial models in industrial segmentation scenarios.

[0117] In one possible design, the base layer 220 is pre-installed with respective tool chains for industrial multimodal pre-training, industrial mechanism embedded fine-tuning, and industrial efficient reasoning, as well as multiple industrial template large models; wherein the industrial base large model is one of the multiple industrial template large models.

[0118] Specifically, the base layer 220 is pre-installed with an industrial multimodal pre-training tool chain, so that the base layer 220 can call it at any time when pre-training the pre-set industrial base large model; the base layer 220 is also pre-installed with an industrial mechanism embedded fine-tuning tool chain, so that the model adaptation layer 230 can call it at any time when performing secondary training on the pre-trained industrial base large model; the base layer 220 is also pre-installed with an industrial efficient reasoning tool chain, so that the interaction layer 240 can call it at any time when improving the interaction ability of the industrial large model. In addition, the base layer 220 is also pre-installed with multiple industrial template large models, and technicians can select an industrial template large model from these industrial template large models as the industrial base large model according to their needs.

[0119] The technical effect of the solution in this embodiment is: by presetting multiple tool chains and multiple industrial template large models on the base layer, the problem of industrial large model platform support is solved.

[0120] In one possible design, the basic resources include: industrial data, computing resources, and industrial knowledge;

[0121] Industrial data is used to provide training data for large industrial models;

[0122] Computing resources are used to provide computing power support for large industrial models;

[0123] Industrial knowledge is used to provide deep logic for large industrial models.

[0124] Specifically, industrial data includes computer-aided design software (CAX) files, industrial time series data, machine instructions, industrial documents, and multimodal data (such as images, videos, and audio). Industrial data is the basis for model training and operation. Computing resources include cloud-edge computing power and storage for industrial large-scale model training and reasoning, as well as chips designed for artificial intelligence (AI) operations. Industrial knowledge includes general industrial knowledge and enterprise-specific knowledge, covering structured files and dedicated knowledge graphs such as industry specifications, operating documents, machine operation principles, and maintenance experience, providing deep logic for decision analysis for industrial large-scale models.

[0125] The technical effect of the solution in this embodiment is: by pre-setting industrial data, computing resources and industrial knowledge at the infrastructure layer, the problem of basic resources for large industrial models is solved.

[0126] Figure 3 Structural schematic diagram of the industrial large model platform provided in the embodiment of the present application Figure 2 .like Figure 3 As shown, the industrial large model platform provided in this embodiment is further refined on the basis of the industrial large model platform provided in the previous embodiment of this application. The industrial large model platform 200 provided in this embodiment also includes: a pre-training module 221 supported by a base layer 220;

[0127] The pre-training module 221 is used to perform industrial multimodal pre-training on the industrial base large model through at least one of a plurality of pre-training methods; wherein the plurality of pre-training methods include: a distributed training method, a transfer learning training method, a self-supervised learning method and a multi-task training method.

[0128] Specifically, in the pre-training module 221, a variety of pre-training methods are supported, including distributed training methods, transfer learning training methods, self-supervised learning methods, and multi-task training methods, etc. They can be combined arbitrarily to meet the training needs of different technicians and scenarios, ensuring that technicians can choose appropriate pre-training methods according to their own needs.

[0129] The pre-training module 221 is also used to set the industrial data called by the industrial base large model and the model parameters of the industrial base large model.

[0130] Specifically, the execution steps of the pre-training module 221 include data pre-processing, model building and model training.

[0131] For data preprocessing, the pre-training module 221 supports unified management of industrial data in the infrastructure layer 210, and technicians can view basic information of industrial data, including version number (ID), data volume, and import log. Technicians can customize industrial data, set data names, select data types, and set data save locations. When customizing data, technicians can use industrial data in combination with functions such as data annotation, data cleaning, and data enhancement to build industrial data suitable for generative large model scenarios. Among them, data cleaning includes operations such as converting traditional Chinese to simplified Chinese, converting uppercase to lowercase, and deleting abnormal characters in industrial data. For the technician's field-related corpus, distribution statistics and quality checks of text data can be performed. This function can output the overall distribution of the field corpus and the quality assessment results, including the distribution statistics of the field type and task type of the data, as well as preliminary data quality checks. In addition, in order to help technicians get started quickly and verify the training effect, the pre-training module 221 can preset some industrial industry-specific industrial data, and technicians can directly use these preset industrial data for training and testing to accelerate the model development process.

[0132] During the process of model building and training, technicians can also set the model parameters of the industrial large model. The pre-training module 221 supports a variety of optimization algorithms and parallel computing architectures to ensure efficiency and accuracy in large-scale data processing. In addition, technicians can also use visualization tools to monitor the progress of training in real time, analyze the performance of the model, and perform necessary tuning operations. After the training of the industrial large model is completed, the pre-training module 221 provides a one-click deployment function, and technicians can quickly apply the trained industrial large model to the production environment to achieve intelligent decision-making, process optimization or other specific business needs. At the same time, the pre-training module 221 also supports online updates and iterative optimization of industrial large models to ensure that they always maintain efficiency and accuracy in ever-changing industrial scenarios.

[0133] The technical effect of the solution in this embodiment is: through a pre-training module supported by a base layer, a large industrial base model is pre-trained using a variety of training methods.

[0134] In one possible design, the industrial large model platform 200 further includes: an adaptation fine-tuning module 231 supported by the model adaptation layer 230;

[0135] The adaptation fine-tuning module 231 is used to perform industrial mechanism embedding fine-tuning on the pre-trained industrial base large model through one of a variety of fine-tuning training methods; wherein the various fine-tuning training methods include: industrial knowledge fusion and industrial mechanism embedding method, supervised fine-tuning method, direct preference optimization method and human feedback reinforcement learning method.

[0136] Specifically, after the pre-training module 221 completes the pre-training of the industrial base large model, the adaptation fine-tuning module 231 will fine-tune the model to enhance its performance. In the adaptation fine-tuning module 231, a variety of fine-tuning training methods are supported, including industrial knowledge fusion and industrial mechanism embedding methods, supervised fine-tuning methods, direct preference optimization methods, and human feedback reinforcement learning methods, etc. Technicians can choose a suitable fine-tuning training method according to their own needs.

[0137] The industrial knowledge fusion and industrial mechanism embedding method creates triple data in the knowledge graph and uses the knowledge graph embedding technology to convert the knowledge into embedded vector features for training, so that the industrial model can integrate industrial knowledge. Secondly, in order to embed the model of industrial mechanism into the industrial model, the industrial mechanism can be represented as feature information that can be recognized by the neural network and embedded in the network architecture. In the model output stage, the model output is guided to conform to the laws of industrial physics, such as adding mechanism equations to the loss function to punish output results that violate the laws of physics.

[0138] Supervised fine-tuning uses a pre-trained neural network model and retrains it on a small amount of supervised data to adapt it to specific task requirements. The process of performing additional training on the basis of a pre-trained industrial base model is usually to make the industrial base model better suited to a specific domain or task.

[0139] Direct preference optimization methods can directly optimize the output of language models to precisely control the output of industrial large models without reinforcement learning. This method can accurately understand and learn the preferences of technicians and achieve more outstanding results.

[0140] Human feedback reinforcement learning refers to incorporating human feedback into the training process so that large industrial models can quickly master human experience, use human preferences as reward signals to guide model training, and thus enhance the model's understanding and satisfaction of human intentions. The success of human feedback reinforcement learning depends on the quality of feedback provided by humans. Adaptation fine-tuning module 231 has developed an effective and scalable method to collect and process such feedback. Adaptation fine-tuning module 231 accesses multi-round dialogue and sorting, prompt corpus datasets as human feedback for reinforcement learning from human feedback (RLHF) training, further improving the intelligence and effectiveness of large industrial models.

[0141] The technical effect of the solution in this embodiment is: through the adaptation fine-tuning module supported by the model adaptation layer, efficient model training means are adopted to achieve fast training of large industrial models.

[0142] In one possible design, the industrial large model platform 200 further includes: an interactive reasoning module 241 supported by the interactive layer 240;

[0143] The interactive reasoning module 241 is used to improve the interactive ability of the industrial large model through a variety of interactive reasoning optimization methods; wherein the various interactive reasoning optimization methods include: reasoning acceleration, industrial knowledge base retrieval enhancement generation, model orchestration and plug-in mechanism.

[0144] Specifically, in the interactive reasoning module 241, the fine-tuned industrial big model can infer the technician's expected conclusion or predict future events based on the input information and existing knowledge. The interactive reasoning module 241 improves the ability of the industrial big model in specific industrial fields and optimizes the performance of knowledge-intensive tasks through a variety of interactive reasoning optimization methods such as reasoning acceleration, industrial knowledge base retrieval-augmented generation (RAG), model orchestration and plug-in mechanism.

[0145] In terms of reasoning acceleration, various technologies such as model compression, pruning, and quantization can be used to reduce the computing and storage requirements of large industrial models, thereby improving the reasoning speed. At the same time, high-performance hardware such as Graphics Processing Unit (GPU) and Tensor Processing Unit (TPU) are supported to speed up the reasoning process. These hardware acceleration technologies enable large industrial models to efficiently process large amounts of data and meet industrial application scenarios with high real-time requirements, such as real-time monitoring of production lines and equipment fault detection. Reasoning acceleration can also reduce energy consumption and operating costs, making large industrial models more usable and popular in industrial sites with limited resources.

[0146] In terms of RAG, the interactive reasoning module 241 combines retrieval and generation technologies, which is particularly suitable for knowledge-intensive tasks in industrial scenarios. RAG first builds an industrial knowledge base covering product design, process flow, and equipment maintenance; then, it uses efficient retrieval algorithms, such as best match and intensive retrieval, to find relevant information from the knowledge base. Based on the retrieved results, the industrial big model further generates more accurate and targeted answers. In the field of industrial manufacturing, RAG can be used to quickly provide technical support, generate maintenance manuals, and optimize production processes. For example, in equipment maintenance, technicians can use RAG to quickly find solutions to faults and automatically generate detailed repair steps to improve maintenance efficiency and accuracy.

[0147] Model orchestration involves model selection, task allocation, and result integration to achieve complex industrial tasks. Model selection selects a suitable large or small model based on specific task requirements; task allocation decomposes complex tasks into subtasks and assigns them to the most suitable model; result integration summarizes the output of sub-models through methods such as weighted average or logistic regression to generate the final solution. For example, in a smart factory, the large model is responsible for production planning and scheduling, while the small model handles operation monitoring and quality inspection. Through model orchestration, the overall production process is optimized and the system efficiency is improved.

[0148] In terms of the plug-in mechanism, the interactive reasoning module 241 supports the installation and use of customized plug-ins to expand the functions of the industrial large model. These plug-ins can be used to add new reasoning algorithms, support more data types or formats, provide visual interfaces, etc. The interactive reasoning module 241 also provides a rich plug-in resource library, such as knowledge base plug-ins and web page parsing plug-ins, and technicians can search, select and install corresponding plug-ins as needed. In addition, technicians can integrate internal services into service orchestration through custom plug-ins. The plug-in mechanism enhances the scalability and customization capabilities of the interactive reasoning module 241, allowing it to demonstrate flexibility and adaptability in more application scenarios and meet the needs of a wide range of technicians.

[0149] The technical effect of the solution in this embodiment is: through the interactive reasoning module supported by the interactive layer, the processing speed and resource efficiency of the industrial large model are improved, the application scenarios are expanded, and the adaptability and solution capabilities of the industrial large model to complex industrial tasks are improved.

[0150] In one possible design, the industrial large model platform 200 further includes: an evaluation module 251 supported by the application layer 250;

[0151] The evaluation module 251 is used to evaluate the function of the industrial large model by at least one of a plurality of evaluation methods; wherein the plurality of evaluation methods include: automatic evaluation and manual evaluation.

[0152] Specifically, the evaluation module 251 includes a series of evaluation indicators and methods for performance evaluation of trained industrial large models. The evaluation module 251 provides two evaluation methods: automatic evaluation and manual evaluation, which aims to comprehensively measure the performance of semantic understanding, knowledge reasoning, professional ability, application ability, instruction following, robustness, bias, hallucination and safety. In addition, it also evaluates industrial specific tasks, such as product design, process design and quality control.

[0153] Automatic evaluation refers to the objective evaluation of the output generated by industrial large models through computer algorithms and statistical methods. This method is widely used in various tasks and can quickly and accurately evaluate model performance. Automatic evaluation usually relies on predefined evaluation indicators and algorithms, mainly by calculating indicators such as accuracy, recall and perplexity to quantify the performance of industrial large models in specific tasks. For industrial specific tasks, automatic evaluation also involves the application effect and practicality in the actual production environment. These tasks cover all aspects of industrial production, from product design to after-sales service, including but not limited to product design, process design, quality control, marketing management, after-sales service, planning and scheduling, production operations, warehousing and logistics, equipment management, safety and environmental protection, energy management and supply chain and other manufacturing application capabilities. This type of evaluation not only focuses on technical indicators, but also pays attention to the performance and value of industrial large models in actual applications. Through evaluation, the applicability and benefits of industrial large models in industrial environments can be fully understood, thereby providing a basis for the improvement and optimization of industrial large models.

[0154] Manual evaluation refers to the subjective evaluation of the output generated by the industrial large model by the evaluator. In order to ensure the consistency and accuracy of the evaluation, a set of evaluation indicators and scoring criteria are usually designed. For example, the fluency, accuracy, relevance and innovation of the text generated by the model can all be used as evaluation dimensions. The evaluator scores the performance of the industrial large model based on these criteria and provides detailed feedback and improvement suggestions.

[0155] Evaluation module 251 comprehensively measures the performance of industrial large models in terms of semantic understanding, knowledge reasoning, and industrial application by combining automatic evaluation and manual evaluation. Automatic evaluation provides efficient and objective quantitative analysis, while manual evaluation supplements the detailed consideration of industrial large models in actual applications through subjective feedback from evaluators. The two complement each other and provide comprehensive data support and practical reference for the optimization and improvement of industrial large models, ensuring that industrial large models can better adapt to complex industrial application scenarios.

[0156] The technical effect of the solution in this embodiment is: performance evaluation of the trained industrial large model is performed through an evaluation module supported by the application layer.

[0157] The embodiment of the present application also provides an industrial large model platform system, the industrial large model platform system comprising: an industrial large model, and a cyber physics system interacting with the industrial large model;

[0158] The industrial large model is a model trained by the industrial large model platform provided in the above embodiments of the present application.

[0159] The industrial large model platform system provided in this embodiment has similar implementation principles and technical effects to the industrial large model platform in the above-mentioned embodiments, and will not be described in detail in this embodiment.

[0160] So far, the technical solution of the present application has been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present application is obviously not limited to these specific embodiments, and the above embodiments are only used to illustrate the technical solution of the present application rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. An industrial large model platform, characterized in that: The industrial large model platform is used to train the industrial large model, and the industrial large model platform includes: an infrastructure layer, a base layer, a model adaptation layer, an interaction layer and an application layer; The infrastructure layer is used to construct basic resources for training the industrial large model; The base layer is used to perform industrial multimodal pre-training on the preset industrial base large model through the preset multimodal data set to improve the processing capability of the industrial large model on multimodal data; wherein the industrial base large model is the base of the industrial large model, and the industrial base large model has the general solution capability of industrial tasks; The model adaptation layer is used to embed and fine-tune the pre-trained industrial base big model with industrial mechanisms through a preset specific data set to obtain the industrial big model; wherein the specific data set includes a task-oriented big model and an industry domain big model; the task-oriented big model is used to fine-tune the pre-trained industrial base big model with multi-task instructions to improve the processing capability of the industrial big model on specific tasks; the industry domain big model is used to embed industry domain knowledge and fine-tune the adapter on the industrial base big model after multi-task instruction fine-tuning to improve the generalization capability of the industrial big model in unknown domain scenarios; The interaction layer is used to improve the interaction capability of the industrial large model through industrial efficient reasoning; wherein the industrial large model is used to interact with technicians and cyber-physical systems respectively; The application layer is used to express the functions of the industrial large model.

2. The industrial large model platform according to claim 1, characterized in that: The task-oriented large model includes: intelligent question-answering model, scenario recognition model, process decision model, terminal control model, content generation model and scientific discovery model; The functions of the industrial big model include: intelligent question and answer, scenario recognition, process decision-making, terminal control, content generation and scientific discovery.

3. The industrial large model platform according to claim 2, characterized in that: The functions of the industrial big model are used to support the application of R&D design, production and manufacturing, experimental testing, business management and operation and maintenance services.

4. The industrial large model platform according to any one of claims 1 to 3, characterized in that: The model adaptation layer is also used to perform scenario knowledge internalization and enhanced self-training on the industrial base large model after industry domain knowledge embedding and adapter fine-tuning through preset limited annotated data, so as to improve the processing capability of the industrial large model in industrial segmentation scenarios.

5. The industrial large model platform according to claim 1, characterized in that: The base layer is pre-installed with tool chains for industrial multimodal pre-training, industrial mechanism embedded fine-tuning and industrial efficient reasoning, as well as multiple industrial template large models; wherein the industrial base large model is one of the multiple industrial template large models.

6. The industrial large model platform according to claim 1, characterized in that: The basic resources include: industrial data, computing resources and industrial knowledge; The industrial data is used to provide training data for the industrial big model; The computing resources are used to provide computing power support for the industrial large model; The industrial knowledge is used to provide deep logic for the industrial big model.

7. The industrial large model platform according to claim 6, characterized in that: The industrial large model platform also includes: a pre-training module supported by the base layer; The pre-training module is used to perform industrial multimodal pre-training on the industrial base large model by at least one of a plurality of pre-training methods; wherein the plurality of pre-training methods include: a distributed training method, a transfer learning training method, a self-supervised learning method, and a multi-task training method; The pre-training module is also used to set the industrial data called by the industrial base large model and the model parameters of the industrial base large model.

8. The industrial large model platform according to claim 1, characterized in that: The industrial large model platform also includes: an adaptation fine-tuning module supported by the model adaptation layer; The adaptation fine-tuning module is used to perform industrial mechanism embedding fine-tuning on the pre-trained industrial base large model through one of a variety of fine-tuning training methods; wherein the multiple fine-tuning training methods include: industrial knowledge fusion and industrial mechanism embedding method, supervised fine-tuning method, direct preference optimization method and human feedback reinforcement learning method.

9. The industrial large model platform according to claim 1, characterized in that: The industrial large model platform also includes: an interactive reasoning module supported by the interactive layer; The interactive reasoning module is used to improve the interactive capability of the industrial large model through a variety of interactive reasoning optimization methods; wherein the various interactive reasoning optimization methods include: reasoning acceleration, industrial knowledge base retrieval enhancement generation, model orchestration and plug-in mechanism.

10. The industrial large model platform according to claim 1, characterized in that: The industrial large model platform also includes: an evaluation module supported by the application layer; The evaluation module is used to evaluate the function of the industrial large model through at least one of multiple evaluation methods; wherein the multiple evaluation methods include: automatic evaluation and manual evaluation.

11. An industrial large model platform system, characterized in that: The industrial large model platform system includes: an industrial large model, and a cyber physics system interacting with the industrial large model; The industrial large model is a model trained by the industrial large model platform described in any one of claims 1 to 10.

Images