Multi-agent cooperative communication and real-time resource optimization management method and system based on large model driving

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a multi-agent cooperative communication and real-time resource optimization management method and system based on large model driving, and the method comprises the following steps: semantic understanding and task analysis, communication protocol and mechanism design, resource demand prediction and allocation scheduling, information sharing and interaction, and feedback and evaluation mechanism. The method has the beneficial effects that efficient reliability allocation and feedback error correction are realized through a multi-agent reinforcement learning framework by utilizing powerful reasoning and generalization capabilities of a large model, so that the communication and cooperation capabilities among multiple agents are remarkably improved, and meanwhile, the method is combined with a real-time resource optimization management technology, and through reasonable resource allocation and scheduling, the real-time resource optimization management efficiency is improved. And maximum utilization of resources is realized.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and specifically to a multi-agent collaborative communication and real-time resource optimization management method and system driven by a large model. Background Art

[0002] A multi-agent system is a system composed of multiple independent but cooperative agents. Each agent has its own perception ability, decision-making ability and execution ability, and can communicate and coordinate with other agents. Such a system has characteristics such as distributed control, self-organization and system complexity, making it widely used in collaborative communication networks, such as sensor networks, mobile communication networks and the Internet of Things. Large models, especially large language models, have received extensive attention and research in recent years. In the field of embodied intelligence, large models were initially mainly used to solve the task planning problems of single agents. However, due to the misalignment between the knowledge of large models and specific embodied environments, the plans generated by large models are often difficult to execute in the environment. Therefore, researchers have begun to explore the application of large models in multi-agent systems to achieve more efficient and flexible collaboration and communication.

[0003] In a multi-agent system, the collaborative communication between agents is the key to achieving the overall goal. However, there are many challenges in this process:

[0004] Complexity and scalability. The associations between agents in a multi-agent system are complex. How to achieve efficient collaboration and decision-making between agents remains a difficult problem. As the scale of the system increases, how to maintain the scalability of the system is also an important issue. Environmental adaptability. Agents need to collaborate in a constantly changing environment. How to make agents adapt to different environments and complete tasks efficiently is also a challenge.

[0005] Moreover, the existing data sources are single. Existing model training data often mainly focuses on certain specific fields or types, such as text, images, etc., lacking the effective integration of richer information sources such as audio, video, and sensor data, resulting in limited understanding ability and insufficient adaptability of large models to complex real-world scenarios. There is a lack of a dynamic parameter adjustment mechanism. During the operation of the model, the model parameters cannot be dynamically adjusted according to real-time input data and task requirements, making the model unable to flexibly allocate computing resources when facing tasks of different scales and complexities, affecting the operation efficiency and accuracy of the model. Summary of the Invention

[0006] The purpose of the present invention is to provide a multi-agent collaborative communication and real-time resource optimization management method and system driven by a large model to solve the problems raised in the above background art.

[0007] To achieve the above object, the present invention provides the following technical solutions: A multi-agent collaborative communication and real-time resource optimization management method driven by a large model, the method comprising the following steps:

[0008] Semantic understanding and task parsing: The agent transmits task instructions and multi-modal information to the large model, which analyzes and processes the information, identifies the task objectives, key steps, and involved agents, and decomposes the task into specific subtasks; at the same time, the Internet of Things technology is used to obtain real-time environmental data, such as temperature, humidity, and pressure, to enrich the source of multi-modal information and improve the accuracy and comprehensiveness of the large model's semantic understanding;

[0009] Communication protocol and mechanism design: Customize the communication protocol, manage the communication frequency and bandwidth, build a distributed communication architecture, and achieve seamless communication connections between agents and devices; according to the system requirements and application scenarios, dynamically adjust the communication frequency of the agents to ensure communication efficiency while avoiding network congestion; and introduce intelligent routing technology to automatically select the optimal communication path and improve the timeliness and reliability of communication;

[0010] Resource demand prediction and allocation scheduling: The large model analyzes historical data and combines the real-time operation status information of the agents, uses the Internet of Things technology to obtain more real-time data on the environment and operation status of the agents, and dynamically adjusts the predicted resource demand; based on the urgency and importance of the task and the performance requirements of the agents, determine the priority of resource allocation, and use a distributed resource scheduling algorithm to enable the agents to autonomously perform resource scheduling according to the resource status of themselves and the surrounding environment;

[0011] Information sharing and interaction: Build a shared information platform, where the agents upload their status information, perception data, and task execution situations to the platform, and other agents obtain the required information from the platform; the large model integrates and analyzes the shared information to provide support for the agents' decision-making; design reasonable interaction mechanisms and collaboration strategies to promote effective collaboration between agents according to task requirements.

[0012] Feedback and evaluation mechanism: Introduce a multi-channel feedback mechanism for agents and users, build a comprehensive performance evaluation index system to evaluate the resource optimization effect; the Internet of Things technology provides richer data support for the feedback mechanism, real-time monitors the resource usage situation through Internet of Things sensors, feeds the real-time data back to the large model, and optimizes the resource management strategy based on the feedback information and the performance evaluation index system.

[0013] Preferably, in the communication protocol and mechanism design step, it further includes: an adaptive upgrade design of the communication protocol, enabling the communication protocol to automatically detect and adapt to changes in the network environment and new communication requirements. When network technology is updated or system functions are extended, the communication protocol can be automatically upgraded and optimized without manually adjusting the settings of the agents on a large scale.

[0014] Preferably, in the resource demand prediction and allocation scheduling steps, it further includes: encouraging resource sharing and reuse among agents to improve the overall utilization rate of resources; through Internet of Things technology, achieving more refined perception and management of resources. For example, in a logistics distribution system, according to the information tracked in real time by Internet of Things devices, transportation resources can be scheduled more accurately.

[0015] Preferably, the method further includes data diversification and continuous update steps: continuously collecting and integrating diverse data, including task instructions, multi-modal information, and operation data of agents in different fields and scenarios, introducing richer information sources such as audio, video, and sensor data; establishing a data update mechanism to regularly incorporate new data into the training and update of the large model to ensure that the large model can adapt to the changing environment and task requirements.

[0016] Preferably, the method further includes model architecture optimization and adaptive adjustment, as well as reinforcement learning and online optimization steps: designing a flexible and efficient model architecture, adopting a modular and dynamic parameter adjustment mechanism, optimizing it in combination with the characteristics of Internet of Things application scenarios, and adaptively adjusting according to different task and data characteristics; applying reinforcement learning technology to enable the large model to continuously learn and optimize its own strategies during the interaction with the multi-agent system, realizing the online optimization of the large model, enabling it to receive feedback information in real time during the system operation and quickly adjusting parameters and updating strategies according to the feedback.

[0017] A multi-agent collaborative communication and real-time resource optimization management system driven by a large model is applied to a multi-agent collaborative communication and real-time resource optimization management method driven by a large model. The system includes:

[0018] A semantic understanding and task parsing module. Agents transmit task instructions and multi-modal information to the large model, and the large model analyzes and processes the information, identifies task objectives, key steps, and involved agents, and decomposes the task into specific subtasks; at the same time, real-time environmental data such as temperature, humidity, and pressure is obtained through Internet of Things technology to enrich the source of multi-modal information and improve the accuracy and comprehensiveness of the large model's semantic understanding.

[0019] A communication protocol and mechanism design module. Customize communication protocols, manage communication frequencies and bandwidths, build a distributed communication architecture to achieve seamless communication connections between agents and devices; dynamically adjust the communication frequencies of agents according to system requirements and application scenarios to ensure communication efficiency while avoiding network congestion; and introduce intelligent routing technology to automatically select the optimal communication path to improve the timeliness and reliability of communication.

[0020] Resource Requirement Prediction and Allocation Scheduling Module: The large model analyzes historical data and combines it with the real-time operating status information of agents. Using Internet of Things technology, it obtains more real-time data on the environment and operating status of agents, and dynamically adjusts the predicted resource requirements. Based on the urgency, importance of tasks, and the performance requirements of agents, it determines the priority of resource allocation, and applies a distributed resource scheduling algorithm to enable agents to autonomously perform resource scheduling according to the resource conditions of themselves and the surrounding environment.

[0021] Information Sharing and Interaction Module: It builds a shared information platform where agents upload their status information, perception data, and task execution situations to the platform, and other agents obtain the required information from the platform. The large model integrates and analyzes the shared information to support the decision-making of agents. It designs reasonable interaction mechanisms and cooperation strategies to promote effective cooperation among agents according to task requirements.

[0022] Feedback and Evaluation Mechanism Module: It introduces a multi-channel feedback mechanism for agents and users, constructs a comprehensive performance evaluation index system to evaluate the resource optimization effect. Internet of Things technology provides richer data support for the feedback mechanism. By using Internet of Things sensors to monitor the resource usage situation in real time and feeding the real-time data back to the large model, it optimizes the resource management strategy based on the feedback information and the performance evaluation index system.

[0023] Preferably, the Communication Protocol and Mechanism Design Module further includes: an adaptive upgrade design of the communication protocol, enabling the communication protocol to automatically detect and adapt to changes in the network environment and new communication requirements. When network technology is updated or system functions are extended, the communication protocol can be automatically upgraded and optimized without manually making large-scale adjustments to the settings of agents.

[0024] Preferably, the Resource Requirement Prediction and Allocation Scheduling Module further includes: encouraging resource sharing and reuse among agents to improve the overall utilization rate of resources; through Internet of Things technology, achieving more refined perception and management of resources. For example, in a logistics distribution system, according to the information tracked in real time by Internet of Things devices, transportation resources can be scheduled more precisely.

[0025] Preferably, the system further includes a Data Diversification and Continuous Update Module, which continuously collects and integrates diverse data, including task instructions, multi-modal information, and the operating data of agents in different fields and scenarios, introducing richer information sources such as audio, video, and sensor data; establishing a data update mechanism to regularly incorporate new data into the training and update of the large model to ensure that the large model can adapt to the changing environment and task requirements.

[0026] Preferably, the system further includes a model architecture optimization and adaptive adjustment module, and a reinforcement learning and online optimization module, which designs a flexible and efficient model architecture, adopts a modular and dynamic parameter adjustment mechanism, optimizes in combination with the characteristics of the Internet of Things application scenarios, and adaptively adjusts according to different tasks and data characteristics; applies reinforcement learning technology to enable the large model to continuously learn and optimize its own strategies during the interaction with the multi-agent system, realizes the online optimization of the large model, enables it to receive feedback information in real time during the system operation process, and performs rapid parameter adjustment and policy update according to the feedback.

[0027] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0028] The method and system for multi-agent collaborative communication and real-time resource optimization management driven by a large model proposed by the present invention utilize the powerful reasoning and generalization capabilities of the large model, and realize efficient belief assignment and feedback error correction through a multi-agent reinforcement learning framework, thereby significantly improving the communication and collaboration capabilities between multi-agents. At the same time, this method combines real-time resource optimization management technology, and realizes the maximization of resource utilization through reasonable resource allocation and scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0030] In order to clearly and completely describe the objectives, technical solutions of the present invention, and make the advantages more clear, the following further details the embodiments of the present invention with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are part of the embodiments of the present invention, rather than all of the embodiments, and are only used to explain the embodiments of the present invention, and are not used to limit the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0031] Embodiment 1, please refer to Figure 1 The present invention provides a technical solution: a method for multi-agent collaborative communication and real-time resource optimization management driven by a large model, the method includes the following steps:

[0032] Step 1:

[0033] Semantic Understanding and Task Parsing: The intelligent agent transmits the task instructions and multi-modal information to the large model. The large model analyzes and processes the information, identifies the target, steps, and intelligent agents, decomposes the task, and uses the Internet of Things to obtain environmental data to improve the accuracy of semantic understanding. The intelligent agent transmits the received task instructions and related multi-modal information to the large model. The large model analyzes and processes the information, accurately identifies the task target, key steps, and involved intelligent agents, and decomposes the task into specific subtasks to prepare for subsequent collaborative work. Among them, using Internet of Things technology, the intelligent agent obtains richer real-time environmental data, such as temperature, humidity, pressure, etc. through sensors, to enrich the source of multi-modal information and further improve the accuracy and comprehensiveness of the large model's semantic understanding. It can improve the accuracy of semantic understanding. By integrating the rich real-time environmental data obtained by Internet of Things technology, the source of multi-modal information is greatly enriched, enabling the large model to understand task instructions more accurately and comprehensively, laying a solid foundation for the accurate decomposition of subsequent tasks and the collaborative work of intelligent agents, and reducing task allocation errors and low collaborative efficiency caused by semantic understanding deviations.

[0034] Communication Protocol and Mechanism Design: Customize the communication protocol, manage the frequency bandwidth, build a distributed architecture, and use the Internet of Things to achieve communication connections between intelligent agents and devices. Optimize the communication strategy. According to the system requirements and application scenarios, develop a suitable customized communication protocol, clarify the message format, encoding method, transmission path, etc.; manage the communication frequency and bandwidth, dynamically adjust the communication frequency of intelligent agents according to the task priority and network conditions, ensure communication efficiency while avoiding network congestion; build a distributed communication architecture to achieve point-to-point communication between intelligent agents, enhancing the reliability and scalability of the system; and, using Internet of Things technology, achieve seamless communication connections between intelligent agents and various Internet of Things devices, enabling the communication protocol to more intelligently adjust the communication strategy, optimize communication efficiency and quality. It can optimize communication efficiency and quality. The customized communication protocol combined with intelligent frequency and bandwidth management, and the construction of a distributed communication architecture, ensure communication efficiency, avoid network congestion, and ensure that information can be transmitted quickly and accurately between intelligent agents. At the same time, using the Internet of Things to achieve seamless communication connections with various devices further optimizes the communication strategy and improves the reliability and quality of communication.

[0035] Enhance collaborative communication capabilities: Introduce intelligent routing to select the optimal path, optimize cross-platform compatibility, achieve adaptive upgrade of communication protocols, ensure the improvement of communication performance and collaborative capabilities. Introduce intelligent routing technology: Deploy intelligent routing algorithms in the communication network. According to the real-time network status and the communication requirements of agents, automatically select the optimal communication path. When multiple agents need to transmit a large amount of data simultaneously, ensure that the data reaches the destination quickly and accurately, improving the timeliness and reliability of communication. Ensure that communication protocols and mechanisms can be compatible with a variety of different types of agents and operating system platforms, enabling agents from different sources and functions to communicate seamlessly, expanding the application scope and collaborative capabilities of multi-agent systems. Adaptive upgrade of communication protocols: Design communication protocols to have the ability to automatically detect and adapt to changes in the network environment and new communication requirements. When network technology is updated or system functions are extended, the communication protocol can be automatically upgraded and optimized without manually making large-scale adjustments to the settings of agents, ensuring that the multi-agent system always maintains good communication performance and room for improvement in collaborative capabilities during long-term operation. It can enhance collaborative communication capabilities. Intelligent routing technology ensures that data reaches the destination quickly and accurately, improving the timeliness and reliability of communication, especially with obvious advantages when multiple agents transmit a large amount of data simultaneously. Optimized cross-platform compatibility expands the application scope and collaborative capabilities of the system, enabling different types of agents to communicate and collaborate seamlessly. The adaptive upgrade of communication protocols ensures that the system always maintains good communication performance and room for improvement in collaborative capabilities during long-term operation, without frequent manual adjustment, reducing maintenance costs.

[0036] Information sharing and interaction: Build a platform for agents to upload and obtain information, integrate and analyze with large models, enrich data sources with the Internet of Things, promote the interaction between agents and the physical world and the adjustment of collaboration strategies. Build a shared information platform where agents upload their own status information, perception data, and task execution status to the platform; other agents can obtain the required information from the platform. The large model integrates and analyzes the shared information to provide support for the decision-making of agents; Design reasonable interaction mechanisms and collaboration strategies, such as using the contract net protocol, etc., to promote effective collaboration between agents according to task requirements. Among them, the Internet of Things technology provides a wider data source for information sharing. Agents obtain a large amount of real-time data through Internet of Things devices and upload it to the shared information platform, while realizing a closer interaction between agents and the physical world, and adjusting collaboration strategies in a timely manner according to actual environmental changes, improving collaborative efficiency and effectiveness. It can promote information sharing and collaboration. The built shared information platform combined with the Internet of Things technology provides agents with a wider data source and a closer interaction ability with the physical world. Agents can upload and obtain accurate information in a timely manner. The integration and analysis of the large model provide strong support for decision-making. Reasonable interaction mechanisms and collaboration strategies promote efficient collaboration between agents according to task requirements, significantly improving collaborative efficiency and effectiveness.

[0037] Step 2:

[0038] Resource demand prediction: The large model analyzes historical data and combines it with the operating status information of agents. It uses the Internet of Things to obtain more environmental and operating data, dynamically adjusts the predicted resource demand. The large model analyzes the historical data of the multi-agent system to mine the patterns and rules of resource demand; combines the operating status information of agents collected by real-time monitoring devices, such as CPU usage rate, memory occupancy rate, etc., to dynamically adjust the prediction of resource demand; uses Internet of Things technology to obtain more real-time data on the environment and operating status of agents, such as the energy consumption and working temperature of devices, further enriching the basis for resource demand prediction, making the prediction more accurate, enabling precise resource demand prediction. The large model combines historical data and the real-time operating status information of agents, and uses the Internet of Things to obtain more real-time environmental and device data, which can more deeply mine the patterns and rules of resource demand, realize dynamic and accurate prediction of resource demand, provide a reliable basis for reasonable resource allocation and scheduling, and avoid resource waste and unreasonable allocation situations.

[0039] Resource allocation and scheduling: Determine the priority according to the tasks and agents, use a distributed algorithm for scheduling, encourage resource sharing and reuse, and the Internet of Things realizes fine-grained perception and management of resources, optimizing the resource configuration of systems such as logistics. Based on the urgency and importance of tasks and the performance requirements of agents, determine the priority of resource allocation; use a distributed resource scheduling algorithm to enable agents to autonomously perform resource scheduling according to the resource status of themselves and the surrounding environment; encourage resource sharing and reuse among agents to improve the overall utilization rate of resources; through Internet of Things technology, achieve more fine-grained perception and management of resources. For example, in the logistics distribution system, according to the information real-time tracked by Internet of Things devices, more precisely schedule transportation resources, reasonably arrange vehicle driving routes and cargo distribution, and at the same time promote more convenient resource sharing and reuse among agents, realizing the efficient utilization and optimized configuration of resources. It can achieve efficient resource allocation and scheduling. Based on scientific evaluation, determine the priority of resource allocation, use a distributed scheduling algorithm combined with the fine-grained perception and management capabilities of the Internet of Things, enable agents to autonomously schedule resources according to the resource status of themselves and the surrounding environment, encourage resource sharing and reuse, greatly improve the overall utilization rate of resources, optimize system performance, and ensure that tasks can be executed efficiently.

[0040] Feedback and Evaluation of Resource Optimization: The agent and the user provide feedback on resource usage and system operation, build an index system to evaluate the optimization effect, the Internet of Things supports the feedback mechanism, optimize the resource management strategy based on the feedback, introduce the feedback mechanism, and the agent provides real-time feedback: During the process of executing tasks and using resources, the agent immediately feeds back information such as the usage effect, performance, and problems encountered by the resources to the large model. For example, the agent reports the difference between the actual resource consumption and the expected value when executing a specific task, or the signal interference situation encountered during communication, etc. User feedback integration: In addition to the feedback from the agent itself, the feedback information of the user is also integrated. The user can evaluate and provide feedback on aspects such as the overall operation effect and task completion quality of the multi-agent system. For example, in an intelligent customer service system, the user provides feedback on the satisfaction with the answers of the intelligent customer service robot. These feedback information are taken into account for resource optimization. Based on the feedback information, a performance evaluation index system is constructed to evaluate the resource optimization effect, such as resource utilization rate, task completion time, system response speed, user satisfaction, etc. Through multi-dimensional evaluation indicators, comprehensively understand the operation status of the system and the effectiveness of resource optimization. According to the evaluation results, continuously improve the optimization methods and strategies to achieve the continuous optimization management of resources. For example, if it is found that a certain agent often lacks resources when executing tasks, adjust its resource allocation strategy or optimize the task allocation method; if the user's satisfaction with a certain function is low, specifically improve the collaborative working mode or algorithm of the relevant agent; among them, the Internet of Things technology provides richer data support for the feedback mechanism, real-time monitors the resource usage through Internet of Things sensors, feeds the real-time data back to the large model, and at the same time facilitates the user to provide feedback information more conveniently. Based on the feedback information and the performance evaluation index system, more effectively optimize the resource management strategy, improve the overall performance of the system and user satisfaction, can improve the feedback and evaluation mechanism, introduce the multi-channel feedback mechanism of the agent and the user, and build a comprehensive performance evaluation index system, can comprehensively understand the system operation status and the effectiveness of resource optimization, based on the rich data support provided by the Internet of Things technology, can obtain feedback information more timely and accurately, quickly improve the optimization methods and strategies according to the evaluation results, achieve the continuous optimization management of resources, and improve the stability of the system and user satisfaction.

[0041] Step Three:

[0042] Data Diversification and Continuous Update: Collect and integrate diverse data, establish an update mechanism, incorporate real-time IoT data, adapt to environmental and task changes, and enhance the adaptability of the large model to the IoT environment. Continuously collect and integrate diverse data, including task instructions, multimodal information, and the operation data of agents in different fields and scenarios. In addition to common text and image data, introduce richer information sources such as audio, video, and sensor data to enhance the large model's understanding of various complex situations. Establish a data update mechanism to regularly incorporate new data into the training and update of the large model, ensuring that the large model can adapt to the changing environment and task requirements. With the development of technology and the expansion of application scenarios, update the data in a timely manner to cover new types of agents, communication protocols, and resource management models, etc. Utilize IoT technology to obtain a large amount of real-time data from the physical world, further enriching data diversity. At the same time, require the data update mechanism to be able to adapt to the changes in IoT data in a timely manner and incorporate new IoT data into the training and update of the large model, continuously enhancing the large model's adaptability to various complex situations in the IoT environment. It can enhance data diversity and adaptability, continuously collect and integrate diverse data, introduce a variety of rich information sources, utilize IoT to obtain a large amount of real-time data from the physical world, establish a flexible data update mechanism, and timely cover changes such as new types of agents, communication protocols, and resource management models, significantly enhancing the large model's understanding of various complex situations and the IoT environment, enabling it to quickly adapt to the changing environment and task requirements.

[0043] Model Architecture Optimization and Adaptive Adjustment: Design a flexible architecture, adopt a modular and dynamic parameter adjustment mechanism, optimize the architecture considering the characteristics of the Internet of Things, ensure efficient operation and accurate prediction in different Internet of Things environments, research and design a more flexible and efficient large model architecture that can be adaptively adjusted according to different tasks and data characteristics. For example, adopt a modular design concept, let the large model select the appropriate module combination to work according to the specific application scenario, improve the adaptability and efficiency of the model; introduce a dynamic parameter adjustment mechanism, during the model operation, dynamically adjust the model parameters according to the real-time input data and task requirements. When dealing with large-scale data or complex tasks, automatically increase the model complexity and computing resource allocation, while for simple tasks, appropriately simplify the model structure to improve the processing speed and reduce resource consumption; consider the diversity and complexity of Internet of Things application scenarios, further optimize the model architecture to adapt to the characteristics and processing requirements of Internet of Things data. For example, aiming at the real-time and continuous characteristics of Internet of Things data, design special model modules to process time series data and streaming data, combine the computing power and resource limitations of Internet of Things devices, optimize the model deployment and operation mode, through the dynamic parameter adjustment mechanism, flexibly adjust the model parameters according to the real-time status and data transmission situation of Internet of Things devices, ensure efficient operation and accurate prediction in different Internet of Things environments, which can optimize the model architecture and performance. Design a flexible and efficient model architecture, adopt a modular and dynamic parameter adjustment mechanism, optimize in combination with the characteristics of Internet of Things application scenarios, can be adaptively adjusted according to different tasks and data characteristics, reasonably allocate computing resources when dealing with tasks of different scales and complexities, improve the operation efficiency and accuracy of the model, and ensure efficient operation and accurate prediction in various Internet of Things environments.

[0044] Reinforcement Learning and Online Optimization: Apply reinforcement learning to enable large models to learn optimization strategies and achieve online optimization. In the Internet of Things (IoT) environment, utilize data for learning, adjust according to feedback, explore optimal strategies, improve system performance and adaptability. Apply reinforcement learning techniques to enable large models to continuously learn and optimize their own strategies during interactions with multi-agent systems. Through the reward mechanism, encourage large models to explore more effective methods for semantic understanding, task parsing, and resource optimization to adapt to different system states and task objectives; achieve online optimization of large models, enabling them to receive feedback information in real-time during system operation and perform rapid parameter adjustment and policy updates based on the feedback. For example, when an agent feedbacks that the execution effect of a certain task is not good, the large model can immediately adjust the relevant processing strategy to improve the execution efficiency and quality of subsequent similar tasks; in the IoT environment, reinforcement learning better utilizes the large amount of real-time data generated by IoT devices for learning and optimization. Through interactions with IoT agents, the large model adjusts strategies based on the real-time state of the devices and environmental changes to achieve more accurate resource management and task allocation. Combining with the real-time communication capabilities of the IoT, the large model receives feedback information more timely, conducts online optimization, and quickly adapts to the dynamic changes in the IoT environment. Utilize the reward mechanism of reinforcement learning to motivate the large model to continuously explore the optimal strategies suitable for IoT scenarios, improve the overall performance and adaptability of multi-agent systems in IoT applications, and can strengthen the learning and optimization capabilities. Apply reinforcement learning techniques to enable large models to continuously learn and optimize strategies during interactions with multi-agent systems, and encourage them to explore more effective methods through the reward mechanism. In the IoT environment, it can better utilize real-time data for learning and optimization, adjust strategies in a timely manner according to device status and environmental changes, achieve more accurate resource management and task allocation, and realize online optimization to enable large models to receive feedback in real-time and make rapid adjustments, improving the overall performance and adaptability of multi-agent systems in IoT applications and enabling them to better cope with different system states and task objectives.

[0045] This multi-agent collaborative communication and real-time resource optimization management method driven by large models can improve the accuracy of semantic understanding. By integrating the rich real-time environmental data obtained through IoT technology, it greatly enriches the multi-modal information sources, enabling the large model to more accurately and comprehensively understand task instructions, laying a solid foundation for the precise decomposition of subsequent tasks and the collaborative work of agents, reducing task allocation errors and low collaborative efficiency caused by semantic understanding deviations. It can also optimize communication efficiency and quality. The customized communication protocol combines intelligent frequency and bandwidth management, as well as the construction of a distributed communication architecture, ensuring communication efficiency, avoiding network congestion, and ensuring that information can be transmitted quickly and accurately between agents. At the same time, seamless communication connections with various devices are achieved through the IoT, further optimizing communication strategies and improving the reliability and quality of communication.

[0046] The multi-agent collaborative communication and real-time resource optimization management method driven by large models can enhance collaborative communication capabilities. The intelligent routing technology ensures that data arrives at the destination quickly and accurately, improving the timeliness and reliability of communication. It has obvious advantages especially when multiple agents transmit a large amount of data simultaneously. The cross-platform compatibility optimization broadens the application scope and collaborative capabilities of the system, enabling different types of agents to seamlessly dock and cooperate. The adaptive upgrade of the communication protocol ensures that the system always maintains good communication performance and room for improvement in collaborative capabilities during long-term operation, without the need for frequent manual adjustment, reducing maintenance costs, and promoting information sharing and collaboration. The built shared information platform combined with Internet of Things technology provides agents with a wider data source and a closer ability to interact with the physical world. Agents can upload and obtain accurate information in a timely manner. The integrated analysis of large models provides strong support for decision-making. The reasonable interaction mechanism and collaboration strategy promote agents to cooperate efficiently according to task requirements, significantly improving the collaborative efficiency and effect.

[0047] The multi-agent collaborative communication and real-time resource optimization management method driven by large models can accurately predict resource requirements. The large model combines historical data and the real-time operating status information of agents, and uses the Internet of Things to obtain more real-time data on the environment and devices, enabling a deeper exploration of resource demand patterns and rules, realizing dynamic and accurate prediction of resource requirements, providing a reliable basis for reasonable resource allocation and scheduling, avoiding resource waste and unreasonable allocation. It can perform efficient resource allocation and scheduling. Based on scientific evaluation, it determines the priority of resource allocation, and uses distributed scheduling algorithms combined with the fine perception and management capabilities of the Internet of Things, enabling agents to autonomously schedule resources according to their own and surrounding environmental resource conditions, encouraging resource sharing and reuse, greatly improving the overall utilization rate of resources, optimizing system performance, and ensuring that tasks can be executed efficiently. It can also improve the feedback and evaluation mechanism, introduce a multi-channel feedback mechanism for agents and users, and build a comprehensive performance evaluation index system, enabling a comprehensive understanding of the system operation status and the effectiveness of resource optimization. Based on the rich data support provided by Internet of Things technology, it can obtain feedback information more timely and accurately, quickly improve optimization methods and strategies according to the evaluation results, realize continuous optimization management of resources, and improve the stability of the system and user satisfaction.

[0048] The multi-agent collaborative communication and real-time resource optimization management method driven by large models can enhance data diversity and adaptability, continuously collect and integrate diverse data, introduce various rich information sources, and use the Internet of Things to obtain a large amount of real-time data in the physical world. It establishes a flexible data update mechanism to promptly cover changes such as new agent types, communication protocols, and resource management modes, significantly enhancing the large model's understanding ability of various complex situations and the Internet of Things environment, enabling it to quickly adapt to the changing environment and task requirements. It can optimize the model architecture and performance, design a flexible and efficient model architecture, adopt a modular and dynamic parameter adjustment mechanism, and optimize in combination with the characteristics of Internet of Things application scenarios. It can adaptively adjust according to different task and data characteristics, reasonably allocate computing resources when dealing with tasks of different scales and complexities, improve the model's operation efficiency and accuracy, ensure efficient operation and accurate prediction in various Internet of Things environments, and also strengthen the learning and optimization ability. By applying reinforcement learning technology, the large model continuously learns and optimizes strategies in the interaction with the multi-agent system, and is motivated to explore more effective methods through a reward mechanism. In the Internet of Things environment, it can better utilize real-time data for learning and optimization, adjust strategies in a timely manner according to device status and environmental changes, achieve more precise resource management and task allocation, and realize online optimization so that the large model can receive feedback in real time and quickly adjust, improving the overall performance and adaptability of the multi-agent system in Internet of Things applications, enabling it to better handle different system states and task objectives.

[0049] Embodiment 2, based on Embodiment 1, proposes a multi-agent collaborative communication and real-time resource optimization management system driven by large models, which is applied to the multi-agent collaborative communication and real-time resource optimization management method driven by large models. The system includes:

[0050] A semantic understanding and task parsing module. The agent transmits task instructions and multi-modal information to the large model, and the large model analyzes and processes the information, identifies the task objective, key steps, and involved agents, and decomposes the task into specific subtasks. At the same time, it uses Internet of Things technology to obtain real-time environmental data such as temperature, humidity, and pressure to enrich the source of multi-modal information and improve the accuracy and comprehensiveness of the large model's semantic understanding.

[0051] Communication protocol and mechanism design module, customize communication protocols, manage communication frequencies and bandwidths, build a distributed communication architecture, and achieve seamless communication connections between agents and devices; according to system requirements and application scenarios, dynamically adjust the communication frequencies of agents to ensure communication efficiency while avoiding network congestion; and introduce intelligent routing technology to automatically select the optimal communication path, improving the timeliness and reliability of communication; also includes: adaptive upgrade design of communication protocols, enabling communication protocols to automatically detect and adapt to changes in the network environment and new communication requirements. When network technology is updated or system functions are extended, the communication protocol can be automatically upgraded and optimized without manually making large-scale adjustments to the settings of agents.

[0052] Resource demand prediction and allocation scheduling module, the large model analyzes historical data and combines it with the real-time operation status information of agents, uses Internet of Things technology to obtain more real-time data on the environment and operation status of agents, and dynamically adjusts the predicted resource demand; based on the urgency, importance of tasks, and the performance requirements of agents, determine the priority of resource allocation, and use a distributed resource scheduling algorithm to enable agents to autonomously perform resource scheduling according to the resource status of themselves and the surrounding environment; also includes: encouraging resource sharing and reuse among agents to improve the overall utilization rate of resources; through Internet of Things technology, achieve more refined perception and management of resources. For example, in a logistics distribution system, according to the information tracked in real-time by Internet of Things devices, more accurately schedule transportation resources.

[0053] Information sharing and interaction module, build a shared information platform, agents upload status information, perception data, and task execution status to the platform, and other agents obtain the required information from the platform; the large model integrates and analyzes the shared information to provide support for the decision-making of agents; design reasonable interaction mechanisms and collaboration strategies to promote effective collaboration among agents according to task requirements.

[0054] Feedback and evaluation mechanism module, introduce a multi-channel feedback mechanism for agents and users, build a comprehensive performance evaluation index system to evaluate the resource optimization effect; Internet of Things technology provides richer data support for the feedback mechanism, real-time monitor the resource usage situation through Internet of Things sensors, and feedback the real-time data to the large model, and optimize the resource management strategy based on the feedback information and the performance evaluation index system.

[0055] The system also includes a data diversification and continuous update module, continuously collect and integrate diverse data, including task instructions, multi-modal information, and operation data of agents in different fields and scenarios, introduce richer information sources such as audio, video, and sensor data; establish a data update mechanism, regularly incorporate new data into the training and update of the large model to ensure that the large model can adapt to the changing environment and task requirements.

[0056] The system also includes a model architecture optimization and adaptive adjustment module, as well as a reinforcement learning and online optimization module, which designs a flexible and efficient model architecture, adopts a modular and dynamic parameter adjustment mechanism, optimizes in combination with the characteristics of the Internet of Things application scenarios, and adaptively adjusts according to different tasks and data characteristics; applies reinforcement learning technology to enable the large model to continuously learn and optimize its own strategies during the interaction with the multi-agent system, realizes the online optimization of the large model, enables it to receive feedback information in real time during the system operation, and performs rapid parameter adjustment and policy update according to the feedback.

[0057] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-agent collaborative communication and real-time resource optimization management method based on a large model, characterized by: The method comprises the following steps: Semantic understanding and task analysis: The intelligent agent transmits the task instructions and multimodal information to the big model, which analyzes and processes the information, identifies the task objectives, key steps and the intelligent agents involved, and decomposes the task into specific subtasks. At the same time, the Internet of Things technology is used to obtain real-time environmental data, such as temperature, humidity, and pressure, to enrich the source of multimodal information and improve the accuracy and comprehensiveness of the semantic understanding of the big model. Communication protocol and mechanism design: customize communication protocols, manage communication frequency and bandwidth, build distributed communication architecture, and achieve seamless communication between agents and devices; dynamically adjust the communication frequency of agents according to system requirements and application scenarios to ensure communication efficiency while avoiding network congestion; and introduce intelligent routing technology to automatically select the optimal communication path to improve the timeliness and reliability of communication; Resource demand prediction and allocation scheduling: The big model analyzes historical data and combines it with the real-time operating status information of the intelligent agent, uses the Internet of Things technology to obtain more real-time data about the environment and operating status of the intelligent agent, and dynamically adjusts the predicted resource demand; based on the urgency and importance of the task and the performance requirements of the intelligent agent, it determines the priority of resource allocation, and uses a distributed resource scheduling algorithm to enable the intelligent agent to autonomously schedule resources based on its own and the surrounding environment's resource conditions; Information sharing and interaction: Build a shared information platform, where intelligent agents upload status information, perception data, and task execution status to the platform, and other intelligent agents obtain the required information from the platform; the large model integrates and analyzes shared information to provide support for the decision-making of intelligent agents; design reasonable interaction mechanisms and collaboration strategies to promote effective collaboration between intelligent agents based on task requirements. Feedback and evaluation mechanism: Introduce a multi-channel feedback mechanism for intelligent agents and users, build a comprehensive performance evaluation indicator system, and evaluate the effect of resource optimization; IoT technology provides richer data support for the feedback mechanism, monitors resource usage in real time through IoT sensors, feeds real-time data back to the big model, and optimizes resource management strategies based on feedback information and performance evaluation indicator system.

2. The multi-agent collaborative communication and real-time resource optimization management method based on large model drive according to claim 1 is characterized by: The communication protocol and mechanism design steps also include: adaptive upgrade design of the communication protocol, so that the communication protocol has the ability to automatically detect and adapt to changes in the network environment and new communication requirements. When the network technology is updated or the system function is expanded, the communication protocol can be automatically upgraded and optimized without manual large-scale adjustment of the intelligent agent settings.

3. The method for multi-agent collaborative communication and real-time resource optimization management based on large model drive according to claim 1 is characterized in that: The resource demand forecasting and allocation scheduling steps also include: encouraging resource sharing and reuse among intelligent agents to improve the overall utilization of resources; achieving more refined perception and management of resources through Internet of Things technology. For example, in the logistics distribution system, transportation resources can be more accurately scheduled based on the information tracked in real time by Internet of Things devices.

4. The method for multi-agent collaborative communication and real-time resource optimization management based on large model drive according to claim 1 is characterized in that: The method also includes data diversification and continuous updating steps: continuously collecting and integrating diverse data, including task instructions, multimodal information and intelligent agent operation data in different fields and scenarios, and introducing richer information sources such as audio, video and sensor data; establishing a data update mechanism to regularly incorporate new data into the training and updating of the large model to ensure that the large model can adapt to the ever-changing environment and task requirements.

5. The method for multi-agent collaborative communication and real-time resource optimization management based on large model drive according to claim 1 is characterized in that: The method also includes model architecture optimization and adaptive adjustment as well as reinforcement learning and online optimization steps: designing a flexible and efficient model architecture, adopting a modular and dynamic parameter adjustment mechanism, optimizing in combination with the characteristics of IoT application scenarios, and adaptively adjusting according to different tasks and data characteristics; applying reinforcement learning technology to allow the large model to continuously learn and optimize its own strategies during the interaction with the multi-agent system, realize online optimization of the large model, enable it to receive feedback information in real time during system operation, and perform rapid parameter adjustments and strategy updates based on the feedback.

6. A multi-agent collaborative communication and real-time resource optimization management system driven by a large model, applied to a multi-agent collaborative communication and real-time resource optimization management method driven by a large model as described in any one of claims 1 to 5, characterized in that: The system comprises: In the semantic understanding and task parsing module, the intelligent agent transmits the task instructions and multimodal information to the large model, which analyzes and processes the information, identifies the task objectives, key steps and the intelligent agents involved, and decomposes the task into specific subtasks. At the same time, the Internet of Things technology is used to obtain real-time environmental data, such as temperature, humidity, and pressure, to enrich the source of multimodal information and improve the accuracy and comprehensiveness of the semantic understanding of the large model. The communication protocol and mechanism design module customizes the communication protocol, manages the communication frequency and bandwidth, builds a distributed communication architecture, and realizes seamless communication between the intelligent agent and the device; dynamically adjusts the communication frequency of the intelligent agent according to system requirements and application scenarios to ensure communication efficiency while avoiding network congestion; and introduces intelligent routing technology to automatically select the optimal communication path to improve the timeliness and reliability of communication; Resource demand prediction and allocation scheduling module: The big model analyzes historical data and combines it with the real-time operating status information of the intelligent agent, using the Internet of Things technology to obtain more real-time data about the environment and operating status of the intelligent agent, and dynamically adjusts the predicted resource demand; based on the urgency and importance of the task and the performance requirements of the intelligent agent, it determines the priority of resource allocation, and uses a distributed resource scheduling algorithm to enable the intelligent agent to autonomously schedule resources according to its own and the surrounding environment's resource conditions; The information sharing and interaction module builds a shared information platform. The intelligent agent uploads status information, perception data and task execution status to the platform, and other intelligent agents obtain the required information from the platform; the large model integrates and analyzes the shared information to provide support for the decision-making of the intelligent agent; and reasonable interaction mechanisms and collaboration strategies are designed to promote effective collaboration between intelligent agents according to task requirements. The feedback and evaluation mechanism module introduces a multi-channel feedback mechanism for intelligent agents and users, builds a comprehensive performance evaluation indicator system, and evaluates the effect of resource optimization. The Internet of Things technology provides richer data support for the feedback mechanism, monitors resource usage in real time through Internet of Things sensors, feeds real-time data back to the big model, and optimizes resource management strategies based on feedback information and performance evaluation indicator system.

7. The multi-agent collaborative communication and real-time resource optimization management system based on large model drive according to claim 6 is characterized by: The communication protocol and mechanism design module also includes: adaptive upgrade design of the communication protocol, which enables the communication protocol to automatically detect and adapt to changes in the network environment and new communication requirements. When network technology is updated or system functions are expanded, the communication protocol can be automatically upgraded and optimized without manual large-scale adjustment of the settings of the intelligent agent.

8. The multi-agent collaborative communication and real-time resource optimization management system based on large model drive according to claim 6 is characterized by: The resource demand forecasting and allocation scheduling module also includes: encouraging resource sharing and reuse among intelligent agents to improve the overall utilization of resources; achieving more refined perception and management of resources through Internet of Things technology. For example, in the logistics distribution system, transportation resources can be more accurately scheduled based on the information tracked in real time by Internet of Things devices.

9. The multi-agent collaborative communication and real-time resource optimization management system based on large model drive according to claim 6 is characterized by: The system also includes a data diversification and continuous update module, which continuously collects and integrates diverse data, including task instructions, multimodal information and intelligent agent operation data in different fields and scenarios, and introduces richer information sources such as audio, video and sensor data; establishes a data update mechanism to regularly incorporate new data into the training and updating of the large model to ensure that the large model can adapt to the ever-changing environment and task requirements.

10. The multi-agent collaborative communication and real-time resource optimization management system based on large model drive according to claim 6 is characterized by: The system also includes model architecture optimization and adaptive adjustment as well as reinforcement learning and online optimization modules. It designs a flexible and efficient model architecture, adopts modular and dynamic parameter adjustment mechanisms, optimizes in combination with the characteristics of IoT application scenarios, and adaptively adjusts according to different tasks and data characteristics. It applies reinforcement learning technology to allow the large model to continuously learn and optimize its own strategies during the interaction with the multi-agent system, realize online optimization of the large model, receive feedback information in real time during system operation, and quickly adjust parameters and update strategies based on the feedback.

Images