Multi-agent collaborative management method and system

Abstract

This invention provides a multi-agent collaborative management method and system, belonging to the field of artificial intelligence technology. The method includes: generating an agent directory based on the registration information of each initial agent; decomposing the composite task into multiple sub-tasks in response to a pending composite task; assigning at least one actual agent to each sub-task based on the registration information of each initial agent in the agent directory; scheduling all actual agents to collaboratively execute the composite task; and optimizing and updating the agent directory based on the performance data of each actual agent during collaborative execution. This invention improves the efficiency and stability of the entire task collaboration process by assigning suitable agents to each sub-task, scheduling all agents to collaboratively execute the task, and dynamically optimizing the agent directory for future task assignments using their actual performance data.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a multi-agent collaborative management method and system. Background Technology

[0002] With the rapid development of artificial intelligence technology, especially with the increasing popularity of large-scale model applications, various industries have begun to actively build intelligent agents to improve business automation and decision-making efficiency.

[0003] In existing technologies, each business unit typically operates independently, building its own intelligent agents to serve its specific needs. This decentralized approach leads to isolation between these agents, preventing them from perceiving each other or exchanging data effectively, thus creating severe information silos. Consequently, these agents cannot effectively share information or collaborate, resulting not only in redundant construction and waste of resources but also in poor efficiency and stability during overall task processing. Summary of the Invention

[0004] This invention provides a multi-agent collaborative management method, system, electronic device, storage medium, and computer program product to address the shortcomings of existing technologies in which agents cannot effectively share information and work collaboratively, thereby improving the efficiency and stability of overall task processing.

[0005] This invention provides a multi-agent collaborative management method, comprising the following steps:

[0006] A directory of agents is generated based on the registration information of each initial agent;

[0007] In response to the composite task to be processed, the composite task is decomposed into multiple sub-tasks;

[0008] Based on the registration information of each initial intelligent agent in the intelligent agent catalog, at least one actual intelligent agent is assigned to each subtask;

[0009] All the actual intelligent agents are scheduled to collaboratively execute the composite task, and the intelligent agent catalog is optimized and updated based on the performance data of each actual intelligent agent during the collaborative execution process.

[0010] According to a multi-agent collaborative management method provided by the present invention, the step of generating an agent directory based on the registration information of each initial agent includes:

[0011] Receive a registration request from any of the initial agents, the registration request carrying the registration information of the initial agent;

[0012] The registration request is validated for legality.

[0013] The registration information of the initial intelligent agents that have passed the legality verification is stored to generate the intelligent agent directory.

[0014] According to a multi-agent collaborative management method provided by the present invention, the registration information includes a functional description and a real-time status; the step of assigning at least one actual agent to each subtask based on the registration information of each initial agent in the agent directory includes:

[0015] Perform a requirements analysis on any of the sub-tasks to determine the agent capability requirements and agent state requirements required to execute the sub-tasks;

[0016] Based on the agent capability requirements and the agent state requirements, an initial agent whose functional description meets the agent capability requirements and whose real-time state meets the agent state requirements is selected from the agent directory and used as the actual agent.

[0017] According to a multi-agent cooperative management method provided by the present invention, the step of scheduling all the actual agents to cooperate in executing the composite task includes:

[0018] Determine the unified data format and interface standard to be followed for data interaction between the actual intelligent agents;

[0019] During the execution of the composite task, the actual intelligent agent is controlled to transmit and synchronize data in real time through a preset data sharing platform in accordance with the unified data format and the interface standard, so as to achieve collaborative execution of the composite task.

[0020] According to a multi-agent collaborative management method provided by the present invention, the method of controlling the actual intelligent agents to perform real-time data transmission and synchronization through a preset data sharing platform in accordance with the unified data format and the interface standard includes:

[0021] Before any of the actual intelligent agents transmits data through the preset data sharing platform, the data format of the original data to be transmitted by the actual intelligent agent is determined;

[0022] When the data format of the original data is inconsistent with the unified data format, the original data is converted into the unified data format to generate standard format data;

[0023] Real-time data transmission and synchronization are performed using the aforementioned standard format data.

[0024] According to a multi-agent collaborative management method provided by the present invention, the step of optimizing and updating the agent catalog based on the performance data of each actual agent during collaborative execution includes:

[0025] Real-time collection of performance data of each actual intelligent agent during the collaborative execution process, wherein the performance data includes at least one of response time, throughput or resource utilization;

[0026] The performance data of each entity are analyzed according to the preset evaluation indicators to generate an evaluation report that identifies the performance bottlenecks of each actual intelligent agent.

[0027] Based on the evaluation report, adjust the registration information corresponding to the actual intelligent agents in the intelligent agent catalog.

[0028] According to a multi-agent collaborative management method provided by the present invention, adjusting the registration information corresponding to the actual agents in the agent catalog based on the evaluation report includes:

[0029] Based on the performance bottlenecks identified in the evaluation report, determine one or more configuration parameters of the actual intelligent agent to be optimized;

[0030] Update the configuration parameters to be optimized in the registration information corresponding to the actual intelligent agent in the intelligent agent directory, so that when assigning the actual intelligent agent to subsequent subtasks, the allocation is performed according to the updated configuration parameters.

[0031] According to a multi-agent collaborative management method provided by the present invention, after generating the agent catalog, it further includes:

[0032] Receive a discovery request from any of the initial agents for finding other collaborators, the discovery request carrying at least one query condition;

[0033] Based on the query criteria, a search is performed in the agent directory, and the registration information of one or more initial agents that meet the query criteria is returned to the initial agent that sent the discovery request.

[0034] The present invention also provides a multi-agent collaborative management system, comprising the following modules:

[0035] The first processing module is used to generate an agent directory based on the registration information of each initial agent;

[0036] The second processing module is used to decompose the composite task into multiple sub-tasks in response to the composite task to be processed.

[0037] The third processing module is used to assign at least one actual agent to each of the subtasks based on the registration information of each initial agent in the agent directory.

[0038] The fourth processing module is used to schedule all the actual intelligent agents to collaboratively execute the composite task, and to optimize and update the intelligent agent directory based on the performance data of each actual intelligent agent during the collaborative execution process.

[0039] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the multi-agent cooperative management method as described above.

[0040] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the multi-agent cooperative management method as described above.

[0041] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the multi-agent collaborative management method as described above.

[0042] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0043] By generating an agent directory based on the registration information of each initial agent, the previously scattered agent resources in the system are centrally managed, providing a foundation for subsequent task allocation and collaborative scheduling. Responding to complex tasks, these tasks are decomposed into multiple sub-tasks, transforming a complex, macroscopic request that is difficult to execute directly into a set of clearly defined executable units. This significantly reduces the difficulty of task processing and creates conditions for parallel collaboration. By assigning at least one actual agent to each sub-task based on the registration information of each initial agent in the agent directory, the most suitable executor can be precisely matched to each decomposed task unit, effectively ensuring a high correlation between task and agent capabilities and improving the success rate and efficiency of task execution. By scheduling all actual agents to collaboratively execute complex tasks and optimizing and updating the agent directory based on the performance data of each agent during collaborative execution, not only is orderly collaboration among multiple agents achieved to complete complex tasks, but the agent directory is also continuously improved based on real performance feedback. This significantly enhances the accuracy of future task allocation, thereby improving the efficiency and stability of the entire task collaboration process. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0045] Figure 1 This is one of the flowcharts of the multi-agent collaborative management method provided by the present invention.

[0046] Figure 2 This is the second flowchart of the multi-agent collaborative management method provided by the present invention.

[0047] Figure 3 This is the third flowchart of the multi-agent collaborative management method provided by the present invention.

[0048] Figure 4 This is the fourth flowchart of the multi-agent collaborative management method provided by the present invention.

[0049] Figure 5 This is the fifth flowchart of the multi-agent collaborative management method provided by the present invention.

[0050] Figure 6 This is the sixth flowchart of the multi-agent collaborative management method provided by the present invention.

[0051] Figure 7 This is the seventh flowchart of the multi-agent collaborative management method provided by the present invention.

[0052] Figure 8 This is the eighth flowchart of the multi-agent collaborative management method provided by the present invention.

[0053] Figure 9 This is a schematic diagram of the structure of the multi-agent collaborative management system provided by the present invention.

[0054] Figure 10 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0056] It should be noted that in the description of this invention, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. The terms "upper," "lower," etc., indicating orientation or positional relationships according to the accompanying drawings, are only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the system or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0057] The terms "first," "second," etc., used in this invention are used to distinguish similar objects, not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0058] The following is combined Figures 1 to 10 This invention describes the multi-agent collaborative management method, system, electronic device, storage medium, and computer program product provided by this invention.

[0059] Reference Figure 1 , Figure 1 This is one of the flowcharts illustrating the multi-agent collaborative management method provided by the present invention, such as... Figure 1 As shown, steps 101 to 104 are included:

[0060] Step 101: Generate an agent directory based on the registration information of each initial agent.

[0061] To address the technical problem of independent construction and isolation of individual agents in existing technologies, a mechanism is needed to uniformly manage and identify all agents. This mechanism is fundamental for subsequent task allocation and collaborative work. Therefore, one embodiment of the present invention first generates an agent directory based on the registration information of each initial agent. This step aims to centrally register and manage all dispersed agent resources in the system.

[0062] In this step, an initial agent refers to any independent agent unit existing in the system and capable of being scheduled. These agents can come from different business units. Registration information is a set of data describing the characteristics of an initial agent. This data provides the necessary basis for system management and scheduling of the agent. The agent catalog is a centralized repository. It is used to systematically record and manage the registration information of all initial agents, forming a global view of agent resources.

[0063] In one specific implementation, the multi-agent collaborative management system actively or passively acquires the registration information of each initial agent. The acquired registration information is then processed and organized. Finally, this structured information is aggregated to generate a complete agent directory. This directory provides a unified, queryable pool of agent resources for all entities in the system.

[0064] Step 102: In response to the composite task to be processed, decompose the composite task into multiple subtasks.

[0065] After generating the agent catalog, a method is needed to handle complex requests that cannot be completed independently by a single initial agent. These requests typically require multiple agents with different capabilities to work collaboratively. Therefore, one embodiment of the present invention, upon receiving such a request, performs a step in response to the complex task to be processed, decomposing the complex task into multiple subtasks. This aims to transform a macroscopically complex processing objective into a series of manageable and assignable execution units.

[0066] In this step, the composite task to be processed refers to a comprehensive task. Completing this task requires invoking multiple different functions or executing several different steps, exceeding the capabilities of a single initial agent. Subtasks, on the other hand, are smaller, more clearly defined independent task units formed after the composite task has been decomposed. Each subtask can be considered an independently assignable and executable work package.

[0067] In one specific implementation, in response to the emergence of a complex task to be processed, the overall goal, constraints, and required capabilities of the complex task are first analyzed. Based on the analysis results, the various logical stages or functional links necessary to complete the complex task can be identified. Subsequently, the complex task is decomposed into multiple subtasks. Each generated subtask explicitly corresponds to one of the identified logical stages or functional links.

[0068] Step 103: Assign at least one actual agent to each subtask based on the registration information of each initial agent in the agent catalog.

[0069] After a complex task is decomposed into multiple subtasks, specific executors need to be assigned to these independent subtasks. If tasks cannot be effectively associated with executors, the task decomposition itself becomes meaningless, and the entire collaborative process cannot continue. Therefore, one embodiment of the present invention performs a step after task decomposition to assign at least one actual intelligent agent to each subtask based on the registration information of each initial intelligent agent in the intelligent agent catalog. This step aims to establish a precise mapping relationship from tasks to execution resources.

[0070] In this step, the actual agent refers to one or more initial agents selected from the agent catalog and assigned to perform a specific subtask. It is a specific execution unit selected from the pool of potential initial agents for a particular task.

[0071] In one specific implementation, the generated agent directory is queried for each subtask. Then, the registration information of each initial agent in the agent directory is analyzed and matched. Through this analysis and matching process, one or more most suitable initial agents can be selected for each subtask. Finally, the operation of assigning at least one actual agent to each subtask is completed, thereby clarifying the responsible party for each subtask.

[0072] Step 104: Schedule all actual agents to collaboratively execute the composite task, and optimize and update the agent catalog based on the performance data of each actual agent during the collaborative execution process.

[0073] After assigning actual agents to subtasks, these independent agents need to be organized and managed to collaboratively complete a composite task. However, the registration information used for the initial assignment may not fully reflect the agents' actual performance in collaboration. Therefore, one embodiment of the present invention performs a comprehensive execution and optimization step. This step schedules all actual agents to collaboratively execute the composite task and optimizes and updates the agent catalog based on the performance data of each agent during collaborative execution.

[0074] In this step, performance data refers to information collected during task execution that reflects the working efficiency and status of each actual agent. Optimizing and updating the agent catalog involves modifying the registration information associated with the corresponding actual agents stored in the catalog based on the analysis results of the performance data.

[0075] In one specific implementation, all actual agents are first scheduled to collaboratively execute a complex task. This scheduling process coordinates and manages the behavior of all assigned agents, ensuring orderly interaction and cooperation to jointly advance the completion of the complex task. During or after collaborative execution, performance data for each agent is collected. This performance data is then analyzed to evaluate the performance of each agent. Finally, based on the evaluation results, the agent directory is optimized and updated, applying adjustments reflecting their actual performance to their registration information.

[0076] This step ensures that the assigned agents form an organic whole, working together effectively to successfully complete complex, multi-faceted tasks. More importantly, it introduces a dynamic feedback optimization mechanism. By updating the agent catalog using actual performance data, the information in the catalog more accurately reflects the agents' true capabilities and states. This directly improves the accuracy of subsequent task allocation and continuously enhances the efficiency and stability of the entire collaborative process.

[0077] To ensure the accuracy and reliability of the generated agent directory, a standardized process is needed to manage the initial access of agents, thereby preventing unauthorized or incorrectly formatted agent information from contaminating the directory and ensuring the effectiveness of all subsequent collaborative operations. Therefore, in one possible implementation, step 101 specifically includes the following steps:

[0078] Step 201: Receive a registration request from any initial agent, the registration request carrying the registration information of the initial agent.

[0079] Step 202: Verify the legitimacy of the registration request.

[0080] Step 203: Store the registration information of the initial agents that have passed the legality verification and generate an agent directory.

[0081] Specifically, the system first receives a registration request from any initial agent. This registration request carries the initial agent's registration information, which details the agent's various characteristics.

[0082] Upon receiving a registration request, its information is not immediately entered. Instead, the registration request is first verified for legitimacy. This verification process may include checking the source of the request, the integrity and correctness of the digital signature, or the format of the information to confirm that the identity of the initial agent is trustworthy and that the information provided is valid. Only when a registration request passes the legitimacy verification will its subsequent processing be permitted.

[0083] Finally, the registration information of the initial agents that have passed the legitimacy verification is stored. By continuously storing this verified registration information, a complete and trustworthy directory of agents is eventually generated.

[0084] By following the steps described above, illegal or non-compliant agents can be effectively prevented from being included in the management scope. This ensures the accuracy and security of every registration entry in the agent directory.

[0085] To improve the accuracy of task allocation from simple functional matching to precise matching that comprehensively considers the agent's capabilities and state, a more refined allocation method is needed. This method ensures that the agent assigned to a subtask not only possesses the required capabilities but is also in its optimal state for efficiently performing the task. Therefore, step 103 specifically includes the following steps:

[0086] Step 301: Perform requirements analysis on any subtask to determine the agent capability requirements and agent state requirements for executing the subtask.

[0087] Step 302: Based on the agent capability requirements and agent state requirements, select the initial agents from the agent directory whose functional descriptions meet the agent capability requirements and whose real-time states meet the agent state requirements, and use them as the actual agents.

[0088] Specifically, a requirements analysis is performed on any subtask, that is, the specific conditions required to execute the subtask are clarified, thereby determining the agent's capability requirements and agent state requirements for executing the subtask. The registration information includes a functional description and a real-time state. The functional description defines the agent's capabilities; therefore, the agent capability requirements correspond to the requirements for the functional description. The real-time state represents the agent's immediate availability; therefore, the agent state requirements correspond to the requirements for the real-time state.

[0089] Once these two requirements are clarified, individuals meeting the criteria are selected from the generated agent catalog based on these agent capability and state requirements. The selection criteria are twofold: first, their functional descriptions must meet the agent capability requirements; second, their real-time states must meet the agent state requirements. All initial agents that simultaneously meet both conditions are ultimately determined as the actual agents for this subtask.

[0090] By implementing such an allocation process that includes detailed analysis and dual-condition filtering, more precise and efficient utilization of agent resources can be achieved. It not only ensures that the selected agents possess the necessary functionalities to complete the task, but also guarantees that they are in an idle or suitable working state when allocated. This effectively avoids assigning tasks to busy or faulty agents, thereby significantly improving the immediacy and success rate of task execution and optimizing the overall system's resource allocation efficiency.

[0091] To enable multiple real-world agents assigned to perform different sub-tasks to work effectively together, potential communication barriers between them must be addressed. These barriers typically stem from different agents using heterogeneous data formats, preventing information exchange. Therefore, in one possible implementation, step 104, scheduling all real-world agents to collaboratively execute the composite task, specifically includes the following steps:

[0092] Step 401: Determine the unified data format and interface standard to be followed for data interaction between actual intelligent agents.

[0093] Step 402: During the execution of the composite task, the actual intelligent agent is controlled to transmit and synchronize data in real time through a preset data sharing platform in accordance with a unified data format and interface standard, so as to achieve collaborative execution of the composite task.

[0094] Specifically, the first step is to determine the unified data format and interface standard that should be followed for data interaction between actual intelligent agents. The unified data format is a predefined data structure used to standardize the content and form of all interactive information. The interface standard defines the protocol and invocation method for data transmission. Together, these constitute the communication contract that all intelligent agents must adhere to.

[0095] Once this standard is established, during the execution of composite tasks, the controlling agents transmit and synchronize data in real time through a pre-defined data sharing platform, adhering to this unified data format and interface standard. The data sharing platform is a centralized data exchange center, providing a common data interaction channel for all agents. This control process ensures that all data flows pass through this platform and strictly follow the established standards, thereby achieving collaborative execution of composite tasks.

[0096] Through the above steps, all actual intelligent agents can communicate in a common language and channel, thereby ensuring seamless data flow and real-time synchronization during the collaboration process, which greatly improves the collaboration efficiency among multiple intelligent agents and the success rate of task execution.

[0097] To ensure that all real-world agents strictly adhere to a unified data interaction standard, especially when dealing with agents using non-standard data formats, a compatibility handling mechanism is needed. This mechanism can automatically handle differences in data formats, lowering the technical barrier for agents to access the collaborative system. Therefore, in one possible implementation, step 402 specifically includes the following steps:

[0098] Step 501: Before any actual intelligent agent transmits data through a preset data sharing platform, the data format of the original data to be transmitted by the actual intelligent agent is determined.

[0099] Step 502: When the original data format is inconsistent with the unified data format, convert the original data into the unified data format to generate standard format data.

[0100] Step 503: Use standard format data for real-time data transmission and synchronization.

[0101] Specifically, before any actual intelligent agent transmits data through a pre-defined data sharing platform, it is necessary to determine the data format of the raw data to be transmitted. Raw data refers to the initial data generated internally by the intelligent agent that has not yet undergone any processing. This determination step aims to identify whether the raw data format is consistent with the pre-defined unified data format.

[0102] When the original data format is inconsistent with the unified data format, a conversion process is triggered. This process converts the original data into the unified data format, thereby generating standard format data. Standard format data is standardized data that conforms to the requirements of the unified data format. If the original data format itself is consistent with the unified data format, no conversion is needed; it can be considered standard format data on its own.

[0103] Finally, whether the data is generated through conversion or is already compliant, it will be transmitted and synchronized in real time using standard format data.

[0104] To ensure the agent catalog dynamically reflects the agents' actual performance in real-world tasks, rather than solely relying on their initial registration information, a continuous optimization mechanism based on performance feedback is needed. This mechanism can identify and correct discrepancies between agent information and their actual capabilities. Therefore, in one possible implementation, step 104, optimizing and updating the agent catalog based on the performance data of each actual agent during collaborative execution, specifically includes the following steps:

[0105] Step 601: Collect performance data of each actual intelligent agent in real time during the collaborative execution process. The performance data includes at least one of response time, throughput or resource utilization.

[0106] Step 602: Analyze the performance data according to the preset evaluation indicators and generate an evaluation report that identifies the performance bottlenecks of each actual intelligent agent.

[0107] Step 603: Based on the evaluation report, adjust the registration information corresponding to the actual intelligent agents in the intelligent agent catalog.

[0108] Specifically, the first step is to collect performance data of each actual intelligent agent in real time during the collaborative execution process. This performance data is a key indicator for quantifying the performance of the intelligent agents, and it may include at least one of response time, throughput, or resource utilization.

[0109] After collecting this data, the performance data is analyzed according to preset evaluation metrics. This analysis process aims to transform the raw data into meaningful evaluation conclusions, thereby generating an evaluation report that identifies the performance bottlenecks of each actual agent. This evaluation report clearly points out which agents perform poorly when performing tasks and where their performance bottlenecks lie.

[0110] Finally, based on the issues revealed in this assessment report, the registration information corresponding to the actual intelligent agents in the intelligent agent catalog was directly adjusted.

[0111] Through the above steps, the information in the agent directory is no longer static, but can dynamically evolve based on actual performance data. This fact-based adjustment ensures that the registration information more accurately reflects the true working capabilities and status of each agent, thereby enabling more precise and efficient task allocation in the future.

[0112] After identifying performance bottlenecks in the agents through the evaluation report, a specific method is needed to translate these evaluation results into practical and actionable modifications to the agent catalog. This method ensures that optimization measures can be applied precisely and take effect immediately in subsequent task assignments. Therefore, in one possible implementation, step 603 specifically includes the following steps:

[0113] Step 701: Based on the performance bottlenecks identified in the evaluation report, determine one or more configuration parameters of the actual agent to be optimized.

[0114] Step 702: Update the configuration parameters to be optimized in the registration information corresponding to the actual agents in the agent directory, so that when assigning actual agents to subsequent subtasks, the allocation is performed according to the updated configuration parameters.

[0115] Specifically, the first step is to identify one or more configuration parameters to be optimized for the actual agent based on the performance bottlenecks identified in the evaluation report. These parameters are specific fields in the registration information that directly affect agent performance or resource scheduling, such as task processing concurrency and memory allocation limits. This process pinpoints macroscopic performance issues to specific parameter items.

[0116] Once these parameters are determined, an update operation is performed. This operation updates the configuration parameters to be optimized within the registration information corresponding to the actual agents in the agent directory. This update action directly modifies the values ​​or settings stored in the directory database. The purpose of this is to ensure that when assigning actual agents to subsequent subtasks, the assignments are made based on the updated configuration parameters.

[0117] Besides the collaborative mode where tasks are assigned by a higher layer, in some scenarios, an initial agent performing a task may need to actively seek help from other unknown agents. To support this dynamic, agent-initiated collaborative need, one embodiment of the present invention provides an optional discovery method after generating an agent catalog. In one possible implementation, after step 101, the method further includes the following steps:

[0118] Step 801: Receive a discovery request from any initial agent to find other collaborators, the discovery request carrying at least one query condition.

[0119] Step 802: Based on the query conditions, search the agent directory and return the registration information of one or more initial agents that meet the query conditions to the initial agent that sent the discovery request.

[0120] Specifically, the system first receives a discovery request from any initial agent to find other collaborators. This discovery request is a specific query message issued by the initial agent, carrying at least one query condition. The query condition is a specific criterion used to describe the capabilities or states that the desired collaborators should possess.

[0121] Upon receiving the request, a search is performed in the agent directory based on the query criteria included in the request. This search process matches entries in the complete agent directory that meet all query criteria. Finally, the registration information of one or more initial agents that match the query criteria is returned to the initial agent that sent the discovery request.

[0122] By implementing this method, individual initial agents are endowed with the ability to proactively discover and identify potential partners. This breaks the limitation of relying entirely on centralized allocation and introduces a more flexible peer-to-peer collaboration model. This mechanism greatly enhances the dynamism and adaptability of the entire multi-agent network, enabling it to handle complex and sudden tasks requiring ad-hoc, immediate collaboration, thereby improving overall problem-solving efficiency.

[0123] Reference Figure 9 , Figure 9 This is a schematic diagram of the structure of the multi-agent collaborative management system provided by the present invention. The system includes:

[0124] The first processing module is used to generate an agent directory based on the registration information of each initial agent;

[0125] The second processing module is used to decompose the composite task into multiple sub-tasks in response to the composite task to be processed.

[0126] The third processing module is used to assign at least one actual agent to each of the subtasks based on the registration information of each initial agent in the agent directory.

[0127] The fourth processing module is used to schedule all the actual intelligent agents to collaboratively execute the composite task, and to optimize and update the intelligent agent directory based on the performance data of each actual intelligent agent during the collaborative execution process.

[0128] In one possible implementation, the first processing module is further configured to:

[0129] Receive a registration request from any of the initial agents, the registration request carrying the registration information of the initial agent;

[0130] The registration request is validated for legality.

[0131] The registration information of the initial intelligent agents that have passed the legality verification is stored to generate the intelligent agent directory.

[0132] In one possible implementation, the third processing module is further configured to:

[0133] Perform a requirements analysis on any of the sub-tasks to determine the agent capability requirements and agent state requirements required to execute the sub-tasks;

[0134] Based on the agent capability requirements and the agent state requirements, an initial agent whose functional description meets the agent capability requirements and whose real-time state meets the agent state requirements is selected from the agent directory and used as the actual agent.

[0135] In one possible implementation, the fourth processing module is further configured to:

[0136] Determine the unified data format and interface standard to be followed for data interaction between the actual intelligent agents;

[0137] During the execution of the composite task, the actual intelligent agent is controlled to transmit and synchronize data in real time through a preset data sharing platform in accordance with the unified data format and the interface standard, so as to achieve collaborative execution of the composite task.

[0138] In one possible implementation, the fourth processing module is further configured to:

[0139] Before any of the actual intelligent agents transmits data through the preset data sharing platform, the data format of the original data to be transmitted by the actual intelligent agent is determined;

[0140] When the data format of the original data is inconsistent with the unified data format, the original data is converted into the unified data format to generate standard format data;

[0141] Real-time data transmission and synchronization are performed using the aforementioned standard format data.

[0142] In one possible implementation, the fourth processing module is further configured to:

[0143] Real-time collection of performance data of each actual intelligent agent during the collaborative execution process, wherein the performance data includes at least one of response time, throughput or resource utilization;

[0144] The performance data of each entity are analyzed according to the preset evaluation indicators to generate an evaluation report that identifies the performance bottlenecks of each actual intelligent agent.

[0145] Based on the evaluation report, adjust the registration information corresponding to the actual intelligent agents in the intelligent agent catalog.

[0146] In one possible implementation, the fourth processing module is further configured to:

[0147] Based on the performance bottlenecks identified in the evaluation report, determine one or more configuration parameters of the actual intelligent agent to be optimized;

[0148] Update the configuration parameters to be optimized in the registration information corresponding to the actual intelligent agent in the intelligent agent directory, so that when assigning the actual intelligent agent to subsequent subtasks, the allocation is performed according to the updated configuration parameters.

[0149] In one possible implementation, the system further includes a fifth processing module for:

[0150] Receive a discovery request from any of the initial agents for finding other collaborators, the discovery request carrying at least one query condition;

[0151] Based on the query criteria, a search is performed in the agent directory, and the registration information of one or more initial agents that meet the query criteria is returned to the initial agent that sent the discovery request.

[0152] It should be noted that the multi-agent collaborative management system provided by the present invention can execute the multi-agent collaborative management method of any of the above embodiments during specific operation, which will not be elaborated in this embodiment.

[0153] Figure 10 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 10 As shown, the electronic device may include: a processor 1010, a communication interface 1020, a memory 1030, and a communication bus 1040, wherein the processor 1010, the communication interface 1020, and the memory 1030 communicate with each other through the communication bus 1040. The processor 1010 can call logical instructions in the memory 1030 to execute a multi-agent cooperative management method, which includes: generating an agent directory based on the registration information of each initial agent; decomposing the composite task into multiple sub-tasks in response to the composite task to be processed; assigning at least one actual agent to each sub-task based on the registration information of each initial agent in the agent directory; scheduling all actual agents to cooperate in executing the composite task; and optimizing and updating the agent directory based on the performance data of each actual agent during the cooperative execution process.

[0154] Furthermore, the logical instructions in the aforementioned memory 1030 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0155] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, and when the program instructions are executed by the computer, the computer is able to execute the multi-agent cooperative management method provided in the above embodiments.

[0156] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the multi-agent cooperative management method provided in the above embodiments.

[0157] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0158] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0159] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-agent collaborative management method, characterized in that, include: A directory of agents is generated based on the registration information of each initial agent; In response to the composite task to be processed, the composite task is decomposed into multiple sub-tasks; Based on the registration information of each initial intelligent agent in the intelligent agent catalog, at least one actual intelligent agent is assigned to each subtask; All the actual intelligent agents are scheduled to collaboratively execute the composite task, and the directory of intelligent agents is optimized and updated based on the performance data of each actual intelligent agent during the collaborative execution process. The optimization and update of the agent catalog based on the performance data of each actual agent during the collaborative execution process includes: Real-time collection of performance data of each actual intelligent agent during the collaborative execution process, wherein the performance data includes at least one of response time, throughput or resource utilization; The performance data of each entity are analyzed according to the preset evaluation indicators to generate an evaluation report that identifies the performance bottlenecks of each actual intelligent agent. Based on the evaluation report, adjust the registration information corresponding to the actual intelligent agents in the intelligent agent directory; the adjustment of the registration information corresponding to the actual intelligent agents in the intelligent agent directory based on the evaluation report includes: Based on the performance bottlenecks identified in the evaluation report, one or more configuration parameters of the actual intelligent agent to be optimized are determined; the configuration parameters to be optimized include the number of concurrent task processing and memory allocation limits; Update the configuration parameters to be optimized in the registration information corresponding to the actual intelligent agent in the intelligent agent directory, so that when assigning the actual intelligent agent to subsequent subtasks, the allocation is performed according to the updated configuration parameters.

2. The multi-agent collaborative management method according to claim 1, characterized in that, The step of generating an agent directory based on the registration information of each initial agent includes: Receive a registration request from any of the initial agents, the registration request carrying the registration information of the initial agent; The registration request is validated for legality. The registration information of the initial intelligent agents that have passed the legality verification is stored to generate the intelligent agent directory.

3. The multi-agent collaborative management method according to claim 1, characterized in that, The registration information includes a functional description and real-time status; the step of assigning at least one actual agent to each subtask based on the registration information of each initial agent in the agent directory includes: Perform a requirements analysis on any of the sub-tasks to determine the agent capability requirements and agent state requirements required to execute the sub-tasks; Based on the agent capability requirements and the agent state requirements, an initial agent whose functional description meets the agent capability requirements and whose real-time state meets the agent state requirements is selected from the agent directory and used as the actual agent.

4. The multi-agent cooperative management method according to claim 1, characterized in that, The scheduling of all the actual intelligent agents to collaboratively execute the composite task includes: Determine the unified data format and interface standard to be followed for data interaction between the actual intelligent agents; During the execution of the composite task, the actual intelligent agent is controlled to transmit and synchronize data in real time through a preset data sharing platform in accordance with the unified data format and the interface standard, so as to achieve collaborative execution of the composite task.

5. The multi-agent collaborative management method according to claim 4, characterized in that, The control of the actual intelligent agent to perform real-time data transmission and synchronization according to the unified data format and the interface standard through a preset data sharing platform includes: Before any of the actual intelligent agents transmits data through the preset data sharing platform, the data format of the original data to be transmitted by the actual intelligent agent is determined; When the data format of the original data is inconsistent with the unified data format, the original data is converted into the unified data format to generate standard format data; Real-time data transmission and synchronization are performed using the aforementioned standard format data.

6. The multi-agent cooperative management method according to claim 1, characterized in that, Following the generation of the agent catalog, the following is also included: Receive a discovery request from any of the initial agents for finding other collaborators, the discovery request carrying at least one query condition; Based on the query criteria, a search is performed in the agent directory, and the registration information of one or more initial agents that meet the query criteria is returned to the initial agent that sent the discovery request.

7. A multi-agent collaborative management system, characterized in that, include: The first processing module is used to generate an agent directory based on the registration information of each initial agent; The second processing module is used to decompose the composite task into multiple sub-tasks in response to the composite task to be processed. The third processing module is used to assign at least one actual agent to each of the subtasks based on the registration information of each initial agent in the agent directory. The fourth processing module is used to schedule all the actual intelligent agents to collaboratively execute the composite task, and to optimize and update the intelligent agent directory based on the performance data of each actual intelligent agent during the collaborative execution process. The step of optimizing and updating the agent directory based on the performance data of each actual agent during collaborative execution includes: real-time collection of performance data of each actual agent during collaborative execution, the performance data including at least one of response time, throughput, or resource utilization; analysis of each performance data according to preset evaluation indicators to generate an evaluation report identifying the performance bottlenecks of each actual agent; and adjustment of the registration information corresponding to the actual agents in the agent directory based on the evaluation report. The step of adjusting the registration information corresponding to the actual agents in the agent directory based on the evaluation report includes: determining one or more configuration parameters to be optimized for each actual agent based on the performance bottlenecks identified in the evaluation report; the configuration parameters to be optimized include task processing concurrency and memory allocation limits; and updating the configuration parameters to be optimized in the registration information corresponding to the actual agents in the agent directory so that when allocating the actual agents for subsequent subtasks, the allocation is performed according to the updated configuration parameters.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the multi-agent cooperative management method as described in any one of claims 1 to 6.

Images