An embodied intelligent agent system for industrial scenarios and an implementation method thereof

Abstract

The application discloses an embodied intelligent Agent system for an industrial scene and an implementation method thereof, and the system comprises: a user interaction layer: used for providing diversified interfaces and communicating with a backend service through an API gateway; an embodied intelligent layer: used for adopting a micro-service architecture based on container arrangement to realize task management, task planning and resource scheduling functions; and a physical device layer: used for integrating with the upper layer system through an industrial Internet of Things protocol and processing time-sensitive applications by adopting an edge computing technology; wherein the user interaction layer and the embodied intelligent layer are bidirectionally connected, and the embodied intelligent layer and the physical device layer are bidirectionally connected. The application can independently perform a series of operations such as order analysis, material planning, production planning, task distribution, material preparation, production execution, quality control and abnormality processing, greatly reduces manual intervention and improves production efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent manufacturing, and in particular to an embodied intelligent Agent system for industrial scenarios and an implementation method thereof. BACKGROUND

[0002] In today's rapidly developing field of intelligent manufacturing, the application of artificial intelligence technology has become a key driving force for improving production efficiency and product quality. With the continuous progress of big data, machine learning, computer vision and other technologies, intelligent manufacturing systems are gradually developing towards intelligence, automation and integration. However, despite the significant advantages exhibited by existing technologies, their development and application still face many challenges and limitations.

[0003] The prior art discloses an intelligent manufacturing system based on artificial intelligence, which includes a manufacturing module, a management module, a quality module and a power supply module. The quality module includes a shooting unit, an analysis unit and a material taking unit. The management module controls the operation of the manufacturing module and the quality module. The manufacturing module automatically assembles power equipment. The shooting unit captures and collects pictures of power equipment to obtain shooting data. The analysis unit identifies the shooting data and obtains analysis information by combining preset defective product parameters. The material taking unit screens out defective products in the power equipment based on the analysis information. This system manages the power equipment production line through artificial intelligence technology, and captures pictures of power equipment during production. Image recognition technology analyzes and judges defective products. Compared with the randomness of traditional manual sampling inspection, defective products can be eliminated one by one, greatly improving the qualified quality of power equipment.

[0004] However, the above system is only applicable to a specific scenario and lacks generalization. Moreover, it simply divides modules for the system without considering the planning and scheduling between modules, and does not have general artificial intelligence capabilities.

[0005] Existing intelligent manufacturing systems usually rely on multiple digital systems and devices, but lack unified intelligent control and coordination mechanisms, making it difficult to achieve full automation and intelligentization of the production process. Therefore, there is an urgent need for an intelligent system that can integrate various digital systems and devices and has autonomous decision-making and execution capabilities to improve production efficiency and flexibility. SUMMARY

[0006] To solve the above technical problems, the present application proposes an embodied intelligent Agent system for industrial scenarios and an implementation method thereof, which utilizes all digital systems and devices of a factory, combines relevant systems and databases, process files and other prior information to realize autonomous production process planning and execution after order input.

[0007] In one aspect, to achieve the above object, the present application provides an embodied intelligent Agent system for industrial scenarios, comprising:

[0008] User Interaction Layer: for providing diversified interfaces and communicating with backend services through API Gateway;

[0009] Embodied Intelligence Layer: for adopting container orchestration-based microservice architecture to realize task management, task planning, and resource scheduling functions;

[0010] Physical Device Layer: for integrating with upper-layer systems through industrial Internet of Things protocols and processing time-sensitive applications using edge computing technology;

[0011] The User Interaction Layer and the Embodied Intelligence Layer are bidirectionally connected, and the Embodied Intelligence Layer and the Physical Device Layer are bidirectionally connected.

[0012] Preferably, the diversified interfaces in the User Interaction Layer include:

[0013] order subsystem interfaces based on modern Web frameworks, cross-platform mobile application interfaces, voice interaction subsystem interfaces integrated with deep learning, and AR / VR monitoring interfaces.

[0014] Preferably, the Embodied Intelligence Layer includes:

[0015] Central Agent Module: for realizing order management, task planning, resource scheduling, and embodied intelligence Agent group management functions based on container orchestration-based microservice architecture;

[0016] Embodied Intelligence Agent Group: for adopting multi-agent architecture to endow each Agent with perception, reasoning, planning, and learning capabilities;

[0017] Integration Module: for integrating subsystems through service bus technology and ensuring real-time data synchronization using change data capture technology;

[0018] Intelligent Function Module: for fusing AI and big data technologies to control the Central Agent Module.

[0019] Preferably, the Embodied Intelligence Layer further includes:

[0020] Data Analysis Module: for conducting data analysis based on distributed computing frameworks combined with deep learning libraries;

[0021] Digital Twin Module: for synchronizing with real-time data of the Physical Device Layer through an IoT platform;

[0022] Knowledge Management Module: for storing knowledge graphs using graph databases and implementing natural language understanding combined with pre-trained language models;

[0023] Decision Support Module: for making optimized decisions in complex environments.

[0024] Preferably, the physical device layer is integrated with upper system through industrial Internet of Things protocol, and adopts edge computing method to process time-sensitive application.

[0025] Preferably, the physical device layer further comprises a sensor network, which is used to connect with large-scale IoT devices through lightweight communication protocol.

[0026] In another aspect, to achieve the above-mentioned purpose, the application further provides an implementation method of embodied intelligent Agent system for industrial scene, comprising:

[0027] inputting a to-be-processed order, performing semantic analysis on the to-be-processed order through a central Agent module, extracting order key information, and performing order management, task planning and resource scheduling according to the key information, starting an embodied intelligent Agent group after the central Agent module completes planning, and completing corresponding tasks respectively.

[0028] Preferably, performing order management, task planning and resource scheduling according to the key information comprises:

[0029] performing semantic analysis through a central Agent module, extracting order key information, and performing order analysis, material planning, production planning and task distribution according to the key information;

[0030] the order analysis utilizes knowledge graph and graph computing technology in the knowledge management module to perform knowledge reasoning, estimates production time and cost;

[0031] the material planning queries the inventory management system in real time based on the digital twin module, and triggers the intelligent procurement system when necessary;

[0032] the production planning considers production efficiency, energy consumption and delivery date factors simultaneously through the embodied intelligent Agent group using multi-objective optimization algorithm,

[0033] the task distribution uses a decentralized task allocation mechanism through a decision support module.

[0034] Preferably, AGV and device Agent will perform material preparation and production execution after receiving the task, wherein the material preparation and production execution utilize intelligent scheduling algorithm and edge computing technology to realize near real-time device control and parameter adjustment, and also predict possible device failure through a predictive maintenance model.

[0035] Preferably, in the implementation method, the quality inspection agent and the maintenance agent in the embodied intelligent agent group perform throughout the production process, wherein the quality inspection agent performs real-time quality inspection and analyzes the results throughout the production process using detection technology, and if an abnormality occurs, the abnormality processing module of the maintenance agent autonomously processes the abnormality in combination with case reasoning and a rule engine, and provides decision support when necessary.

[0036] Compared with the prior art, the present application has the following advantages and technical effects:

[0037] (1) Improve production efficiency: by integrating various advanced technologies such as natural language processing, knowledge graph, reinforcement learning, edge computing and computer vision, the embodied intelligent agent system of the present application can realize full automation and intelligentization of the production process. The system can autonomously perform a series of operations such as order analysis, material planning, production planning, task distribution, material preparation, production execution, quality control and abnormality processing, greatly reducing manual intervention and improving production efficiency;

[0038] (2) Improve production quality: the present application uses advanced detection technology for real-time quality inspection, combined with predictive maintenance algorithms, to predict and handle equipment failures in advance, ensuring the stability of the production process and the consistency of product quality. Through real-time monitoring and automatic adjustment of production parameters, the system can continuously optimize during the production process, improving product quality;

[0039] (3) Reduce production cost: through intelligent scheduling and optimization algorithms, the present application can effectively allocate and utilize resources, reducing material waste and equipment idle time. The AGV scheduling system and intelligent logistics system can optimize material transportation and finished product distribution paths, reducing logistics costs. The application of edge computing technology reduces data transmission delay and improves real-time response capability, further reducing operating costs;

[0040] (4) Enhance system flexibility: the embodied intelligent agent system adopts a microservice architecture, with high scalability and flexibility. The system can dynamically adjust resource allocation and task planning according to production needs, adapting to production tasks of different scales and complexities. Through federated learning technology, the system can realize cross-factory knowledge sharing and continuous optimization while protecting data privacy, enhancing the adaptability and flexibility of the system. BRIEF DESCRIPTION OF DRAWINGS

[0041] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application and are incorporated in and constitute a part of this application. The illustrations, together with the description, serve to explain the application, but do not limit the application. In the drawings:

[0042] Figure 1A structure diagram of an embodiment of the present application for an industrial scene-oriented embodied intelligent Agent system;

[0043] Figure 2 A flow chart of an embodiment of the present application for an implementation method of an industrial scene-oriented embodied intelligent Agent system. DETAILED DESCRIPTION

[0044] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

[0045] It should be noted that the steps shown in the flow chart of the accompanying drawings can be executed in a computer system such as a group of computer executable instructions, and although the logical order is shown in the flow chart, in some cases, the steps shown or described herein can be executed in an order different from that shown herein.

[0046] The present application proposes an industrial scene-oriented embodied intelligent Agent system, as shown in Figure 1 , comprising:

[0047] A user interaction layer for providing diversified interfaces and communicating with backend services through an API gateway;

[0048] An embodied intelligent layer for adopting a micro-service architecture based on container orchestration to realize task management, task planning and resource scheduling functions;

[0049] A physical device layer for integrating with the upper layer system through industrial Internet of Things protocols and adopting edge computing technology to process time-sensitive applications;

[0050] Among them, the user interaction layer and the embodied intelligent layer are bidirectionally connected, and the embodied intelligent layer and the physical device layer are bidirectionally connected.

[0051] The diversified interfaces in the user interaction layer include:

[0052] An order subsystem interface based on a modern Web framework, a cross-platform mobile application interface, a voice interaction subsystem interface integrated with deep learning, and an AR / VR monitoring interface.

[0053] Specifically, the user interaction layer provides diversified interfaces, including an order system based on a modern Web framework, a cross-platform mobile application, a voice interaction system integrated with deep learning, and an AR / VR monitoring interface. All front-end applications communicate with backend services through an API gateway to ensure the efficiency and flexibility of data transmission.

[0054] The embodied intelligent layer includes:

[0055] Central Agent Module: For container orchestration-based microservice architecture, realizing order management, task planning, resource scheduling, and embodied intelligent Agent group management functions.

[0056] Embodied Intelligent Agent Group: For multi-agent architecture, giving each Agent the ability to perceive, reason, plan, and learn.

[0057] Integration Module: For integrating subsystems through service bus technology and ensuring real-time data synchronization using change data capture technology.

[0058] Intelligent Function Module: For controlling the Central Agent Module by integrating AI and big data technologies.

[0059] Data Analysis Module: For data analysis based on a distributed computing framework combined with deep learning libraries.

[0060] Digital Twin Module: For synchronizing with real-time data from the physical device layer through an IoT platform.

[0061] Knowledge Management Module: For storing knowledge graphs using graph databases and implementing natural language understanding with pre-trained language models.

[0062] Decision Support Module: For making optimized decisions in complex environments.

[0063] The Embodied Intelligence Layer is the core layer of the system, including the Central Agent Module, Embodied Intelligent Agent Group, Integration Module, Intelligent Function Module, Data Analysis Module, Data Twin Module, Knowledge Management Module, and Decision Support Module.

[0064] The Central Agent Module is a first-level embodied intelligent Agent, using a container orchestration-based microservice architecture to realize core functions such as order management, task planning, resource scheduling, and embodied intelligent Agent group management.

[0065] The Embodied Intelligent Agent Group uses a multi-agent system architecture, with each Agent having the ability to perceive, reason, plan, and learn. Different types of Agents (such as scheduling, equipment, AGV, and quality inspection) use specialized algorithms to complete their respective tasks.

[0066] The Integration Module uses enterprise service bus technology to seamlessly integrate ERP, MES, WMS, and other systems, and uses change data capture technology to ensure real-time data synchronization.

[0067] The Intelligent Function Module is the brain of the Agent, integrating AI and big data technologies.

[0068] The data analysis module is based on a distributed computing framework and combines deep learning libraries for data analysis, enabling complex machine learning.

[0069] The digital twin module synchronizes real-time data with physical devices through an IoT platform.

[0070] The knowledge management module uses a graph database to store a knowledge graph and combines pre-trained language models to achieve efficient natural language understanding.

[0071] The decision support module is based on reinforcement learning algorithms and can make optimized decisions in complex environments.

[0072] The physical device layer integrates with the upper layer system through industrial IoT protocols and uses edge computing methods to handle time-sensitive applications.

[0073] Specifically, the physical device layer integrates with the upper layer system through industrial IoT protocols and uses edge computing technology to handle time-sensitive applications.

[0074] The physical device layer also includes a sensor network that connects to a large-scale IoT device through a lightweight communication protocol.

[0075] The workflow of this system is a highly automated and intelligent closed-loop process that combines advanced artificial intelligence and automation technology. Therefore, the present application also provides an implementation method for an embodied intelligent Agent system for industrial scenarios, as shown in the technical flowchart Figure 2 .

[0076] The input order is processed by the central Agent module for semantic analysis, extracting key information, and managing orders, task planning, and resource scheduling based on the key information. After the central Agent module completes planning, the embodied intelligent Agent group is started and completes the corresponding tasks.

[0077] Specifically, the process begins with order input. Order information first passes through the central Agent module, which uses natural language processing technology to perform semantic analysis on the order, extract key information, and perform upper-level planning such as order management, task planning, and resource scheduling. After the central Agent module completes planning, the embodied intelligent Agent group is started and completes the corresponding tasks.

[0078] Firstly, the scheduling Agent will conduct order analysis, material planning, production planning and task distribution. Order analysis uses knowledge graphs and graph computing techniques in the knowledge management module for knowledge reasoning, estimating production time and cost; material planning is based on real-time queries to the inventory management system through the digital twin module, and if necessary, triggers the intelligent procurement system; production planning uses multi-objective optimization algorithms through embodied intelligent Agent groups, considering production efficiency, energy consumption and delivery date factors, and task distribution uses a decentralized task allocation mechanism through the decision support module.

[0079] Next, AGV and equipment Agents will prepare materials and execute production after receiving the task. The material preparation and production execution stage uses intelligent scheduling algorithms and edge computing techniques to achieve near-real-time device control and parameter adjustment. The system also integrates predictive maintenance models that can predict potential device failures in advance.

[0080] Once the task is started, quality inspection Agents and maintenance Agents in the embodied intelligent Agent group will execute throughout the production process. The quality inspection Agent runs through the entire production process, using advanced detection technology for real-time quality inspection and analysis. Once an anomaly occurs, the exception handling module of the maintenance Agent will handle most of the exceptions autonomously, and provide decision support when necessary, combining case reasoning and rule engines.

[0081] After production is completed, AGV and logistics Agents will automatically execute finished product warehousing and logistics delivery tasks. The finished product warehousing and logistics delivery stage uses optimization algorithms for cargo placement and transportation route planning. The entire process is recorded through distributed ledger technology to ensure the transparency and traceability of the supply chain.

[0082] Finally, both the central Agent and the embodied intelligent Agent group have their own data analysis and optimization functions. The data analysis and optimization stage uses big data architecture combined with batch processing and stream processing techniques. The machine learning module supports model version control and automated deployment. The entire process also integrates explainable AI modules to enhance the credibility and explainability of the system. At the same time, the application of digital twin technology allows the pre-visualization and optimization of production processes in a virtual environment, further improving the accuracy of decision-making and the efficiency of production.

[0083] This embodied intelligent system, through the integration of AI, big data, IoT and automation technologies, realizes the intelligentization of the entire process from order to delivery, greatly improving production efficiency and flexibility, while maintaining high scalability and security.

[0084] The above merely provides the preferred embodiments of the present application, and the protection scope of the present application is not limited thereto, and any changes or substitutions within the technical scope disclosed by the present application should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. An embodied intelligent agent system for industrial scenarios, characterized in that, include: User interaction layer: used to provide diverse interfaces and communicate with backend services through API gateway; Embodied Intelligence Layer: Used to implement task management, task planning, and resource scheduling functions in a container-based microservice architecture; Physical device layer: Used to integrate with upper-layer systems via industrial IoT protocols and to handle time-sensitive applications using edge computing technology; The user interaction layer is bidirectionally connected to the embodied intelligence layer, and the embodied intelligence layer is bidirectionally connected to the physical device layer. The diverse interfaces in the user interaction layer include: The interface includes an order subsystem based on a modern web framework, a cross-platform mobile application interface, a voice interaction subsystem interface integrating deep learning, and an AR / VR monitoring interface. The embodied intelligence layer includes: Central Agent Module: Used in container orchestration-based microservice architectures to implement order management, task planning, resource scheduling, and embodied intelligent agent group management functions; Embodied intelligence agent swarm: used to employ a multi-agent architecture, endowing each agent with the ability to perceive, reason, plan, and learn; Integration module: Used to integrate subsystems through service bus technology and to ensure real-time data synchronization using change data capture technology; Intelligent function module: used to integrate AI and big data technologies to control the central agent module; The embodied intelligence layer also includes: Data Analysis Module: Used for data analysis based on a distributed computing framework and deep learning libraries; Digital twin module: used to synchronize real-time data with the physical device layer via an IoT platform; Knowledge Management Module: Used to store knowledge graphs using graph databases and combine them with pre-trained language models to achieve natural language understanding; Decision support module: Used to make optimization decisions in complex environments; The physical device layer integrates with the upper-layer system through industrial IoT protocols and uses edge computing methods to handle time-sensitive applications; The physical device layer also includes a sensor network for connecting to a large number of IoT devices via a lightweight communication protocol.

2. A method for implementing the embodied intelligent agent system for industrial scenarios as described in claim 1, characterized in that, include: Input an order to be processed. The order is semantically analyzed by the central agent module to extract key information. Based on the key information, the central agent module performs order management, task planning, and resource scheduling. After the central agent module completes the planning, it starts the embodied intelligent agent group and each agent completes its corresponding task.

3. The implementation method according to claim 2, characterized in that, Based on the aforementioned key information, order management, task planning, and resource scheduling are performed, including: Semantic analysis is performed through the central Agent module to extract key order information, and order analysis, material planning, production planning, and task distribution are carried out based on the key information. The order analysis utilizes knowledge graphs and graph computing technology in the knowledge management module to perform knowledge reasoning and estimate production time and costs. The material planning is based on a digital twin module that queries the inventory management system in real time, and triggers the intelligent procurement system when necessary; The production planning utilizes a swarm of embodied intelligent agents and employs a multi-objective optimization algorithm, simultaneously considering factors such as production efficiency, energy consumption, and delivery time. The task distribution uses a decentralized task allocation mechanism through the decision support module.

4. The implementation method according to claim 2, characterized in that, After receiving a task, the AGV and the equipment agent will prepare materials and execute production. The material preparation and production execution utilize intelligent scheduling algorithms and edge computing technology to achieve near real-time equipment control and parameter adjustment. It also uses a predictive maintenance model to predict possible equipment failures.

5. The implementation method according to claim 2, characterized in that, In the implementation method, the quality inspection agent and maintenance agent in the embodied intelligent agent group will be executed throughout the production process. The quality inspection agent runs through the entire production process, uses detection technology to perform real-time quality detection and analyze the results. If an anomaly occurs, the anomaly handling module of the maintenance agent combines case reasoning and rule engine to autonomously handle the anomaly and provide decision support when necessary.

Images