Load balancing method and device, electronic equipment, program product and storage medium

Abstract

The invention relates to the technical field of load balancing, and provides a load balancing method and device, electronic equipment, a program product and a storage medium. The method comprises the following steps: acquiring real-time target data in a system; the real-time target data comprises resource request data, resource response data and system state data; inputting the real-time target data into a pre-trained prediction model to obtain prediction data output by the pre-trained prediction model; the prediction data comprises resource request prediction data, resource response prediction data and system state prediction data; and controlling the intelligent agent network to carry out load balancing based on the prediction data. Through combination of the pre-trained prediction model and the agent network, the limitation of a traditional load balancing method in processing a complex and dynamic network environment can be broken through, efficient response and flexible allocation of resources are realized in the face of non-uniform request mode or resource state change, and the load balancing efficiency is improved. Therefore, the resource response speed and the resource allocation balance are improved.

Description

Technical Field

[0001] This application relates to the field of load balancing technology, specifically to a load balancing method, apparatus, electronic device, program product, and storage medium. Background Technology

[0002] Load balancing is a core concept in network technology, used to distribute network resources among multiple servers or network devices to optimize resource utilization and improve system response speed and reliability.

[0003] Traditional load balancing methods typically use pre-defined rules such as round-robin or weighted round-robin to allocate resources, which have limitations when dealing with complex and dynamic network environments. For example, when faced with uneven request patterns or changes in resource status, it often leads to uneven resource allocation and response delays. Summary of the Invention

[0004] This application provides a load balancing method, apparatus, electronic device, program product, and storage medium to solve the technical problems of traditional load balancing methods, which typically use preset rules such as round-robin or weighted round-robin to allocate resources. These methods have limitations when dealing with complex and dynamic network environments and often lead to uneven resource allocation and response delays when facing non-uniform request patterns or changes in resource status.

[0005] In a first aspect, embodiments of this application provide a load balancing method, including: Acquire real-time target data from the system; the real-time target data includes resource request data, resource response data, and system status data; The real-time target data is input into a pre-trained prediction model to obtain the prediction data output by the pre-trained prediction model; the prediction data includes resource request prediction data, resource response prediction data, and system status prediction data. The control agent network performs load balancing based on the predicted data.

[0006] In one embodiment, the agent network includes predictive agents, decision-making agents, and executive agents, and the control agent network performs load balancing based on the prediction data, including: The predictive agent is controlled to convert the predictive data into a recognizable form of data for the agent network. The decision-making agent is controlled to determine resource allocation strategies and task priorities based on the identifiable data. The execution agent is controlled to adjust resource allocation and schedule allocation tasks based on the resource allocation strategy and the allocation task priority.

[0007] In one embodiment, the agent network further includes a monitoring agent, and the load balancing method further includes: The monitoring agent is controlled to continuously monitor the status of the system and the operation of the predictive agent, the decision-making agent and the execution agent, and to obtain feedback target data in the system.

[0008] In one embodiment, after adjusting resource allocation and scheduling the allocated tasks, the process includes: The intelligent agent network is controlled to dynamically adjust the resource allocation strategy and the allocation task priority based on the feedback target data and the prediction data.

[0009] In one embodiment, after adjusting resource allocation and scheduling the allocated tasks, the process includes: The feedback target data is used as new real-time target data, and the process returns to the step of inputting the real-time target data into the pre-trained prediction model until the agent network is controlled again to perform load balancing based on the prediction data.

[0010] In one embodiment, the pre-trained prediction model is obtained based on the following method: Acquire historical target data from the system; Input the historical target data and the corresponding labels into the large model to obtain the historical prediction data output by the large model. With the goal of convergence of the historical prediction data, the parameters of the large model are fine-tuned, and the process returns to the step of obtaining historical target data in the system until the pre-trained prediction model is obtained.

[0011] Secondly, embodiments of this application provide a load balancing device, comprising: The data acquisition module is used to: acquire real-time target data in the system; the real-time target data includes resource request data, resource response data, and system status data; The data prediction module is used to: input the real-time target data into a pre-trained prediction model to obtain the prediction data output by the pre-trained prediction model; the prediction data includes resource request prediction data, resource response prediction data, and system status prediction data. The load balancing module is used to control the agent network to perform load balancing based on the predicted data.

[0012] Thirdly, embodiments of this application provide an electronic device, including a processor and a memory storing a computer program, wherein the processor executes the program to implement the steps of the load balancing method described in the first aspect.

[0013] Fourthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the steps of the load balancing method described in the first aspect.

[0014] Fifthly, embodiments of this application provide a non-transitory computer-readable storage medium, including a computer program, which, when executed by a processor, implements the steps of the load balancing method described in the first aspect.

[0015] The load balancing method, apparatus, electronic device, program product, and storage medium provided in this application acquire real-time target data in the system. This real-time target data includes resource request data, resource response data, and system status data. The real-time target data is input into a pre-trained prediction model to obtain predicted data output by the trained prediction model. This predicted data includes resource request prediction data, resource response prediction data, and system status prediction data. The application controls an intelligent agent network to perform load balancing based on the predicted data. This application utilizes a pre-trained prediction model to make forward-looking predictions based on real-time resource data and real-time system data. When facing diverse resource request patterns, it can accurately predict future changes in resource data and system data. Then, it utilizes an intelligent agent network to perform load balancing based on this constantly changing future data. It fully leverages the collaborative mechanism among multiple agents in the intelligent agent network to plan resource allocation strategies in advance, thereby achieving efficient resource response and flexible allocation. In summary, by combining a pre-trained prediction model and an intelligent agent network, this application overcomes the limitations of traditional load balancing methods in handling complex and dynamic network environments. When facing non-uniform request patterns or changes in resource status, it achieves efficient resource response and flexible allocation, thereby improving resource response speed and resource allocation balance. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this application 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 application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is one of the flowcharts illustrating the load balancing method provided in the embodiments of this application; Figure 2 This is a second schematic flowchart of the load balancing method provided in the embodiments of this application; Figure 3 This is the third flowchart illustrating the load balancing method provided in the embodiments of this application; Figure 4 This is a schematic diagram of the load balancing device provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] Figure 1 This is one of the flowcharts illustrating the load balancing method provided in the embodiments of this application. (Refer to...) Figure 1 This application provides a load balancing method, which may include: Step 101: Obtain real-time target data from the system; Real-time target data includes resource request data, resource response data, and system status data; Step 102: Input the real-time target data into the pre-trained prediction model to obtain the prediction data output by the pre-trained prediction model; The forecast data includes resource request forecast data, resource response forecast data, and system status forecast data; Step 103: Control the agent network to perform load balancing based on the predicted data.

[0020] In step 101, the resource request data in the real-time target data may include request type, request frequency, etc.; the resource response data in the real-time target data may include response duration, response resource consumption, etc.; the system status data in the real-time target data may include CPU utilization, memory usage, network bandwidth, etc.

[0021] It should be noted that in a two-tier load balancing scenario where northbound requests go to the gateway system and the gateway system goes to the southbound API (Application Programming Interface), the resource request data in the real-time target data can be northbound to southbound resource request data, and the resource response data in the real-time target data can be southbound to northbound resource response data.

[0022] Furthermore, these real-time target data can be preprocessed, such as data cleaning, formatting, and feature extraction. The extracted features can then be used as input data for the pre-trained prediction model. Since the preprocessed data has a uniform format, less interference, and prominent features, it can effectively improve the prediction efficiency and accuracy of the model.

[0023] In step 102, the pre-trained prediction model can be obtained based on the following method: Step 102a: Obtain historical target data from the system; Step 102b: Input the historical target data and the corresponding labels into the large model to obtain the historical prediction data output by the large model; Step 102c: With the goal of convergence of historical prediction data, fine-tune the parameters of the large model, and then return to step 102a until the pre-trained prediction model is obtained.

[0024] In step 102a, historical resource request data, historical resource response data, and historical system status data are obtained.

[0025] In step 102b: For historical resource request data, the historical prediction data output by the large model is the predicted value of the resource request data at the next historical moment, and the label corresponding to the historical resource request data is the classification of the true value of the resource request data at the next historical moment. For historical resource response data, the historical prediction data output by the large model is the predicted value of the resource response data at the next historical moment, and the label corresponding to the historical resource response data is the classification of the true value of the resource response data at the next historical moment. For historical system state data, the historical prediction data output by the large model is the predicted value of the system state data at the next historical moment, and the label corresponding to the historical system state data is the classification of the true value of the system state data at the next historical moment.

[0026] Then, after inputting the historical target data and the corresponding labels into the large model, the large model will make predictions on the historical target data based on fully learning the real information of the labels, thus obtaining relatively accurate historical prediction data.

[0027] In step 102c, the goal is to converge the historical prediction data, that is, to minimize the difference between the historical prediction data and the historical real data. When the difference is large, the parameters of the large model are continuously fine-tuned and repeated training is performed until the difference is reduced to meet the preset requirements or the preset number of training times is reached. At this time, the large model is the pre-trained prediction model, which can accurately predict future data based on real-time target data, providing forward-looking guidance for the subsequent collaborative load regulation of the intelligent agent network and helping the intelligent agent network make more accurate decisions.

[0028] In step 103, since the prediction data includes resource request prediction data, resource response prediction data, and system state prediction data, the agent network can plan resource allocation strategies in advance based on future resource requests and future resource responses, taking full account of the future state of the system, thereby achieving efficient resource response and flexible allocation.

[0029] The load balancing method provided in this embodiment acquires real-time target data from the system. This real-time target data includes resource request data, resource response data, and system status data. The real-time target data is input into a pre-trained prediction model to obtain predicted data output by the trained prediction model. This predicted data includes resource request prediction data, resource response prediction data, and system status prediction data. The intelligent agent network is then controlled to perform load balancing based on the predicted data. This embodiment utilizes a pre-trained prediction model to make forward-looking predictions based on real-time resource data and real-time system data. When facing diverse resource request patterns, it can accurately predict future changes in resource data and system data. Then, it utilizes an intelligent agent network to perform load balancing based on this constantly changing future data. By fully leveraging the collaborative mechanism among multiple agents in the intelligent agent network, resource allocation strategies can be planned in advance, thereby achieving efficient resource response and flexible allocation. In summary, this embodiment, by combining a pre-trained prediction model and an intelligent agent network, overcomes the limitations of traditional load balancing methods in handling complex and dynamic network environments. When facing non-uniform request patterns or changes in resource status, it achieves efficient resource response and flexible allocation, thereby improving resource response speed and resource allocation balance.

[0030] Furthermore, in the two-layer load balancing scenario of northbound requests to the gateway system and southbound APIs, this embodiment can achieve the most reasonable resource allocation by analyzing historical data and predicting future trends of the northbound, gateway, and southbound layer servers, thereby ensuring the security and high availability of the gateway system.

[0031] Figure 2 This is a second schematic flowchart of the load balancing method provided in the embodiments of this application. (Refer to...) Figure 2 In one embodiment, the agent network includes a predictive agent, a decision-making agent, and an executive agent, and step 103 may include: Step 201: Control the predictive agent to convert the prediction data into a recognizable form of data for the agent network; Step 202: The control decision-making agent determines the resource allocation strategy and task priority based on identifiable data. Step 203: The control execution agent adjusts resource allocation and schedules assigned tasks based on resource allocation strategies and task priorities.

[0032] In step 201, the control predictive agent receives the prediction data output by the pre-trained prediction model and converts it into a form that the agent network can understand, providing input to the decision agent.

[0033] In step 202, the control decision-making agent determines whether to adopt a static or dynamic allocation strategy, as well as the priority order of task allocation, based on the identifiable form of the predicted data.

[0034] One approach is to use reinforcement learning algorithms, such as Q-learning, to control the decision-making agent by continuously optimizing its decisions based on its determined resource allocation strategy and task priority, as well as the feedback after the execution of the resource allocation strategy and task priority, and outputting better resource allocation strategies and task priorities.

[0035] In step 203, the control execution agent executes the resource allocation strategy according to the instructions of the decision-making agent, prioritizing the allocated tasks. As the decision-making agent continuously optimizes its decisions, it achieves flexible adjustment of resource allocation and flexible scheduling of allocated tasks.

[0036] This embodiment achieves intelligent resource management and efficient load balancing by intelligently coordinating the predictive agent, decision-making agent, and execution agent in the agent network. This is achieved through the adaptive adjustment and execution of resource allocation strategies and task priorities from the northbound to the gateway system and from the gateway system to the southbound API. This ensures timely processing of critical tasks and efficient utilization of resources, thereby enhancing the system's adaptability and robustness.

[0037] In one embodiment, the agent network further includes a monitoring agent, and the load balancing method may further include: The control and monitoring agent continuously monitors the system's status and the operation of the prediction agent, decision agent, and execution agent, and obtains feedback target data in the system. This feedback target data is the real-time target data of the system under the influence of the decision after the control and execution agent executes the decision of the decision agent, which can reflect the execution effect of resource allocation strategy and task priority allocation.

[0038] Furthermore, in the intelligent agent network, the predictive agent, decision-making agent, execution agent, and monitoring agent can share information through a publish-subscribe messaging mechanism. This information includes prediction data, resource status, resource allocation strategies, and task priorities. Each agent has a clear division of labor, and the cooperation mechanism enables effective resource allocation and efficient task scheduling.

[0039] This embodiment, by monitoring intelligent agents, can obtain the system status and the operation status of each intelligent agent in real time, which helps to detect abnormal situations in the system or intelligent agents in time. Furthermore, by obtaining feedback target data in the system, it is possible to understand the execution effect of resource allocation strategies and task priority allocation in a timely manner, which facilitates subsequent continuous optimization and improvement.

[0040] In one embodiment, step 203 may be followed by: The control agent network dynamically adjusts resource allocation strategies and task priorities based on feedback target data and prediction data.

[0041] The feedback target data, along with the identifiable form of the latest prediction data, is input into the decision-making agent. The agent is then controlled to fully consider the feedback target data when determining a new resource allocation strategy and a new allocation task priority based on the identifiable form of the latest prediction data, so that the new resource allocation strategy and the new allocation task priority can be optimized towards achieving load balancing more quickly.

[0042] This embodiment feeds back the target data to the agent network, which can form a closed-loop control optimization of the agent network, thereby continuously improving the performance of the agent network, optimizing resource allocation strategies and task priorities, and achieving load balancing more quickly.

[0043] In one embodiment, step 203 may be followed by: The feedback target data is used as new real-time target data, and the process returns to the step of inputting the real-time target data into the pre-trained prediction model until the agent network is controlled again to perform load balancing based on the prediction data.

[0044] The feedback target data is then used as new real-time target data to input into the pre-trained prediction model, resulting in new prediction data output by the pre-trained prediction model. The control agent network then performs load balancing based on this new prediction data.

[0045] This embodiment feeds back the target data to the pre-trained prediction model, enabling the model to fully consider the target data when outputting prediction data. This allows for further fine-tuning of the model parameters, optimizing the output prediction data towards faster load balancing, improving the model's prediction accuracy, and simultaneously enhancing the decision-making performance of the agent network. This forms a closed-loop control optimization between the agent network and the prediction model. Combined with the aforementioned closed-loop control optimization within the agent network, this more effectively optimizes resource allocation strategies and task priorities, thereby improving load balancing efficiency and effectiveness.

[0046] Figure 3 This is the third flowchart illustrating the load balancing method provided in this application. (Refer to...) Figure 3 In one embodiment, the overall process of this application is briefly described as follows: Step 301: Obtain real-time target data from the system; Step 302: Preprocess the real-time target data to prepare it for use by the pre-trained prediction model; Step 303: Input the preprocessed data into the pre-trained prediction model to obtain the prediction data output by the model; Step 304: Control the agent network to perform load balancing based on the predicted data; Step 305: Obtain the feedback target data in the system and feed the feedback target data back to the agent network and the pre-trained prediction model to continuously iterate and optimize the agent network and the pre-trained prediction model.

[0047] The load balancing device provided in the embodiments of this application is described below. The load balancing device described below and the load balancing method described above can be referred to in correspondence.

[0048] Figure 4 This is a schematic diagram of the load balancing device provided in an embodiment of this application. (Refer to...) Figure 4 This application provides a load balancing device, which may include: The data acquisition module 401 is used to: acquire real-time target data in the system; the real-time target data includes resource request data, resource response data and system status data; The data prediction module 402 is used to: input the real-time target data into a pre-trained prediction model to obtain the prediction data output by the pre-trained prediction model; the prediction data includes resource request prediction data, resource response prediction data and system status prediction data; The load balancing module 403 is used to control the agent network to perform load balancing based on the predicted data.

[0049] The load balancing device provided in this embodiment acquires real-time target data from the system. This real-time target data includes resource request data, resource response data, and system status data. The real-time target data is input into a pre-trained prediction model to obtain predicted data output by the trained prediction model. This predicted data includes resource request prediction data, resource response prediction data, and system status prediction data. The device then controls an intelligent agent network to perform load balancing based on this predicted data. This embodiment utilizes a pre-trained prediction model to make forward-looking predictions based on real-time resource data and real-time system data. When facing diverse resource request patterns, it can accurately predict future changes in resource and system data. Then, it utilizes an intelligent agent network to perform load balancing based on this constantly changing future data. By fully leveraging the collaborative mechanism among multiple agents in the intelligent agent network, it can plan resource allocation strategies in advance, thereby achieving efficient resource response and flexible allocation. In summary, this embodiment, by combining a pre-trained prediction model and an intelligent agent network, overcomes the limitations of traditional load balancing methods in handling complex and dynamic network environments. When facing non-uniform request patterns or changes in resource status, it achieves efficient resource response and flexible allocation, thereby improving resource response speed and resource allocation balance.

[0050] In one embodiment, the agent network includes a predictive agent, a decision-making agent, and an execution agent, and the load balancing module 403 is specifically used for: The predictive agent is controlled to convert the predictive data into a recognizable form of data for the agent network. The decision-making agent is controlled to determine resource allocation strategies and task priorities based on the identifiable data. The execution agent is controlled to adjust resource allocation and schedule allocation tasks based on the resource allocation strategy and the allocation task priority.

[0051] In one embodiment, the agent network further includes a monitoring agent, and the load balancing module 403 is further configured to: The monitoring agent is controlled to continuously monitor the status of the system and the operation of the predictive agent, the decision-making agent and the execution agent, and to obtain feedback target data in the system.

[0052] In one embodiment, a data feedback module (not shown in the figure) is further included for: The intelligent agent network is controlled to dynamically adjust the resource allocation strategy and the allocation task priority based on the feedback target data and the prediction data.

[0053] In one embodiment, the data feedback module is further configured to: The feedback target data is used as new real-time target data, and the process returns to the step of inputting the real-time target data into the pre-trained prediction model until the agent network is controlled again to perform load balancing based on the prediction data.

[0054] In one embodiment, a model building module (not shown in the figure) is also included for: Acquire historical target data from the system; Input the historical target data and the corresponding labels into the large model to obtain the historical prediction data output by the large model. With the goal of convergence of the historical prediction data, the parameters of the large model are fine-tuned, and the process returns to the step of obtaining historical target data in the system until the pre-trained prediction model is obtained.

[0055] Figure 5 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application, such as... Figure 5As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other via the communication bus 540. The processor 510 can call a computer program stored in the memory 530 to execute the steps of a load balancing method, such as including: Acquire real-time target data from the system; the real-time target data includes resource request data, resource response data, and system status data; The real-time target data is input into a pre-trained prediction model to obtain the prediction data output by the pre-trained prediction model; the prediction data includes resource request prediction data, resource response prediction data, and system status prediction data. The control agent network performs load balancing based on the predicted data.

[0056] Furthermore, the logical instructions in the aforementioned memory 530 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 this application, in essence, or the part that contributes to the prior art, or a portion 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 described in the various embodiments of this application. 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.

[0057] On the other hand, embodiments of this application also provide a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the steps of the load balancing method provided in the above embodiments, such as including: Acquire real-time target data from the system; the real-time target data includes resource request data, resource response data, and system status data; The real-time target data is input into a pre-trained prediction model to obtain the prediction data output by the pre-trained prediction model; the prediction data includes resource request prediction data, resource response prediction data, and system status prediction data. The control agent network performs load balancing based on the predicted data.

[0058] On the other hand, embodiments of this application also provide a non-transitory computer-readable storage medium storing a computer program thereon, the computer program being used to cause a processor to execute the steps of the load balancing methods provided in the above embodiments, for example including: Acquire real-time target data from the system; the real-time target data includes resource request data, resource response data, and system status data; The real-time target data is input into a pre-trained prediction model to obtain the prediction data output by the pre-trained prediction model; the prediction data includes resource request prediction data, resource response prediction data, and system status prediction data. The control agent network performs load balancing based on the predicted data.

[0059] The non-transitory computer-readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to magnetic memory (e.g., floppy disk, hard disk, magnetic tape, magneto-optical disk (MO)), optical memory (e.g., CD, DVD, BD, HVD), and semiconductor memory (e.g., ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)).

[0060] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. 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.

[0061] 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., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0062] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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. Such 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 this application.

Claims

1. A load balancing method, characterized in that, include: Obtain real-time target data from the system; The real-time target data includes resource request data, resource response data, and system status data; The real-time target data is input into a pre-trained prediction model to obtain the prediction data output by the pre-trained prediction model; the prediction data includes resource request prediction data, resource response prediction data, and system status prediction data. The control agent network performs load balancing based on the predicted data.

2. The load balancing method according to claim 1, characterized in that, The agent network includes predictive agents, decision-making agents, and execution agents. The control agent network performs load balancing based on the prediction data, including: The predictive agent is controlled to convert the predictive data into a recognizable form of data for the agent network. The decision-making agent is controlled to determine resource allocation strategies and task priorities based on the identifiable data. The execution agent is controlled to adjust resource allocation and schedule allocation tasks based on the resource allocation strategy and the allocation task priority.

3. The load balancing method according to claim 2, characterized in that, The agent network also includes a monitoring agent, and the load balancing method further includes: The monitoring agent is controlled to continuously monitor the status of the system and the operation of the predictive agent, the decision-making agent and the execution agent, and to obtain feedback target data in the system.

4. The load balancing method according to claim 3, characterized in that, After adjusting resource allocation and scheduling assigned tasks, the process includes: The intelligent agent network is controlled to dynamically adjust the resource allocation strategy and the allocation task priority based on the feedback target data and the prediction data.

5. The load balancing method according to claim 3, characterized in that, After adjusting resource allocation and scheduling assigned tasks, the process includes: The feedback target data is used as new real-time target data, and the process returns to the step of inputting the real-time target data into the pre-trained prediction model until the agent network is controlled again to perform load balancing based on the prediction data.

6. The load balancing method according to claim 1, characterized in that, The pre-trained prediction model is obtained based on the following method: Acquire historical target data from the system; Input the historical target data and the corresponding labels into the large model to obtain the historical prediction data output by the large model. With the goal of convergence of the historical prediction data, the parameters of the large model are fine-tuned, and the process returns to the step of obtaining historical target data in the system until the pre-trained prediction model is obtained.

7. A load balancing device, characterized in that, include: The data acquisition module is used to: acquire real-time target data in the system; the real-time target data includes resource request data, resource response data, and system status data; The data prediction module is used to: input the real-time target data into a pre-trained prediction model to obtain the prediction data output by the pre-trained prediction model; the prediction data includes resource request prediction data, resource response prediction data, and system status prediction data. The load balancing module is used to control the agent network to perform load balancing based on the predicted data.

8. An electronic device comprising a processor and a memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the load balancing method according to any one of claims 1 to 6.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the load balancing method according to any one of claims 1 to 6.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the load balancing method according to any one of claims 1 to 6.

Images