Method for determining translation rule of internet protocol version and related equipment

Abstract

The invention provides an internet protocol version translation rule determination method and related equipment, and the method comprises the steps: for each sub-translation node, in response to the determination of different internet protocol versions, each round of adjustment operation executes the following steps: based on network flow data, utilizing a model parameter and an intelligent agent in a federated learning model to determine the translation rule of each sub-translation node; determining a translation parameter combination corresponding to the internet protocol version of the client; the central translation node aggregates each model parameter; based on each model parameter and the aggregated model parameter, in response to determining that the federated learning model is not converged, taking the aggregated model parameter as a model parameter in a next round of operation, and performing the next round of operation; and in response to the determination of convergence of the federal learning model, obtaining the translation parameter combination corresponding to each sub-translation node, determining the translation rule corresponding to the client based on each translation parameter combination, and exiting at least one round of operation, thereby solving the technical problem of low adaptability of the translation rule of the internet protocol version in the prior art.

Description

Methods for determining translation rules for Internet Protocol versions and related equipment Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method and related equipment for determining translation rules for Internet Protocol versions. Background Technology

[0002] In the current network architecture, with the widespread adoption of IPv6 (Internet Protocol Version 6) and the depletion of IPv4 (Internet Protocol Version 4) addresses, the transition from IPv4 to IPv6 has become a crucial issue in global network operations and maintenance. In the power industry, with the rapid development of smart grids, Internet of Things (IoT) devices, and big data applications, network communication plays a vital role in the stability and reliability of power systems. However, the power industry's communication networks face unique challenges. Real-time data transmission between power terminal devices requires low latency and high reliability. Most of these devices use the IPv4 protocol; how to smoothly transition to IPv6 while ensuring system stability and efficiency has become an urgent technical problem to be solved.

[0003] In practical applications, the transition to IPv6 mainly relies on static configuration technology, that is, using fixed translation rules (such as fixed PSID, prefix and pattern). However, this method has the following limitations: (1) Lack of adaptability: Fixed translation parameters cannot dynamically adapt to network load and environmental changes, resulting in performance degradation. Especially in high-load network environments, dynamic network environments make the adaptability of translation rules composed of translation parameters low.

[0004] (2) Cross-domain scenario problem: The communication environment of the power industry usually involves multiple cross-domain data transmissions. However, the existing IPv4 / IPv6 translation technology does not perform well in such complex scenarios. The complex communication environment makes the adaptability of the translation rules composed of translation parameters low. Summary of the Invention

[0005] In view of this, the purpose of this application is to propose a method and related equipment for determining the translation rules of Internet Protocol versions, so as to overcome all or part of the deficiencies in the prior art.

[0006] To achieve the above objectives, this application provides a method for determining the translation rules of the Internet Protocol (IP) version, applied to a power system. The power system includes a central translation node and multiple sub-translation nodes. The method includes: for each sub-translation node, in response to determining that a communication request from a client has been received, the sub-translation node obtains the IP version of the client; in response to determining that the IP version differs from the IP version of the power system, the sub-translation node obtains network traffic data corresponding to the client; based on the network traffic data, the sub-translation node performs multiple rounds of adjustment operations, each round of adjustment operations as follows: based on the network traffic data, the sub-translation node uses model parameters from a pre-trained federated learning model and a pre-built intelligent agent to determine the client's IP version. The central translation node collects the translation parameter combinations corresponding to each version and uploads the model parameters to the central translation node. The central translation node aggregates the model parameters uploaded by each sub-translation node to obtain aggregated model parameters. Based on the model parameters uploaded by each sub-translation node and the aggregated model parameters, the central translation node determines whether the federated learning model has converged. In response to determining that the federated learning model has not converged, the central translation node uses the aggregated model parameters as the model parameters in the next round of adjustment operations. In response to determining that the federated learning model has converged, the central translation node obtains the translation parameter combinations corresponding to each sub-translation node and determines the translation rules corresponding to the client based on the translation parameter combinations corresponding to each sub-translation node, and exits at least one round of adjustment operations.

[0007] Optionally, based on the network traffic data, the sub-translation node uses model parameters in a pre-trained federated learning model and a pre-built agent to determine the translation parameter combination corresponding to the Internet Protocol version of the client, including: predicting the network state based on the network traffic data using model parameters in the federated learning model; and determining the translation parameter combination based on the network state using the agent.

[0008] Optionally, determining the translation parameter combination using the agent based on the network state includes: determining an initial translation parameter combination using the agent based on the network state; and adjusting the initial translation parameter combination using a reward function in the agent to obtain the translation parameter combination.

[0009] Optionally, determining the translation rule corresponding to the client based on the translation parameter combination corresponding to each sub-translation node includes: dividing parameters belonging to the same first predetermined type in all translation parameter combinations into the same sequence; for each sequence, counting the number of elements with the same name in the sequence, and determining the element corresponding to the largest number of elements in the sequence as the target translation parameter; and using all target translation parameters as the translation rule corresponding to the client.

[0010] Optionally, the central translation node aggregates the model parameters uploaded by each sub-translation node to obtain aggregated model parameters, including: using a federated averaging algorithm to calculate the average parameters of each type of model parameters uploaded by each sub-translation node that belong to the same second predetermined type; and using all average parameters as the aggregated model parameters.

[0011] Optionally, the step of using the aggregated model parameters as model parameters in the federated learning model in the next round of adjustment includes: the central translation node sending the aggregated model parameters to each sub-translation node, and each sub-translation node replacing the parameter in the federated learning model that belongs to the same second predetermined type as the average parameter with the average parameter for each average parameter in the aggregated model parameters.

[0012] Optionally, after exiting at least one round of adjustment operations, the method includes: the central translation node translating the Internet Protocol address corresponding to the client according to the translation rules.

[0013] Based on the same inventive concept, this application also provides a device for determining the translation rules of Internet Protocol (IP) versions, applied to a power system. The power system includes a central translation node and multiple sub-translation nodes. The device includes: a first acquisition module configured to, for each sub-translation node, acquire the IP version of the client in response to determining that a communication request from a client has been received; a second acquisition module configured to, in response to determining that the IP version is different from the IP version of the power system, acquire network traffic data corresponding to the client; based on the network traffic data, the sub-translation nodes perform multiple rounds of adjustment operations, each round of adjustment operations being executed as follows: the first determination module is configured to, based on the network traffic data, use model parameters in a pre-trained federated learning model and a pre-built intelligent agent to determine the IP version corresponding to the client. The process involves: a translation parameter combination module, which uploads the model parameters to the central translation node; an aggregation module, configured to aggregate the model parameters uploaded by each sub-translation node to obtain aggregated model parameters; a second determination module, configured to determine whether the federated learning model has converged based on the model parameters uploaded by each sub-translation node and the aggregated model parameters; a third determination module, configured to, in response to determining that the federated learning model has not converged, use the aggregated model parameters as model parameters in the next round of adjustment operations; and the third determination module, further configured to, in response to determining that the federated learning model has converged, obtain the translation parameter combination corresponding to each sub-translation node, determine the translation rule corresponding to the client based on the translation parameter combination corresponding to each sub-translation node, and exit at least one round of adjustment operations.

[0014] Based on the same inventive concept, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method described above when executing the computer program.

[0015] Based on the same inventive concept, this application also provides a non-transitory computer-readable storage medium that stores computer instructions for causing a computer to perform the method described above.

[0016] As can be seen from the above, the method and related equipment for determining the translation rules of the Internet Protocol version provided in this application include, for each sub-translation node: in response to determining that a communication request from a client has been received, the sub-translation node obtains the Internet Protocol version of the client; in response to determining that the Internet Protocol version is different from the Internet Protocol version of the power system, the sub-translation node obtains the network traffic data corresponding to the client; based on the network traffic data, the sub-translation node performs multiple rounds of adjustment operations, each round of adjustment operations being performed as follows: based on the network traffic data, the sub-translation node uses the model parameters in a pre-trained federated learning model and a pre-built intelligent agent to determine the combination of translation parameters corresponding to the Internet Protocol version of the client, and The model parameters are uploaded to the central translation node; the central translation node aggregates the model parameters uploaded by each sub-translation node to obtain aggregated model parameters; based on the model parameters uploaded by each sub-translation node and the aggregated model parameters, the central translation node determines whether the federated learning model has converged; in response to determining that the federated learning model has not converged, the aggregated model parameters are used as the model parameters in the federated learning model in the next round of adjustment, and the next round of adjustment is performed; in response to determining that the federated learning model has converged, the central translation node obtains the translation parameter combination corresponding to each sub-translation node, and based on the translation parameter combination corresponding to each sub-translation node, determines the translation rule corresponding to the client, and exits at least one round of adjustment. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 is a flowchart illustrating the method for determining translation rules for Internet Protocol versions according to an embodiment of this application; Figure 2 is a flowchart illustrating the training and aggregation process of the federated learning model according to an embodiment of this application; Figure 3 is a comparison chart of translation parameter optimization effects according to an embodiment of this application; Figure 4 is a flowchart illustrating the optimization process of translation parameter combination based on PPO according to an embodiment of this application; Figure 5 is a schematic diagram of the system architecture for IPv4 / IPv6 translation optimization based on federated learning according to an embodiment of this application; Figure 6 is a schematic diagram of the structure of the device for determining translation rules for Internet Protocol versions according to an embodiment of this application; Figure 7 is a schematic diagram of the hardware structure of the electronic device according to an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0020] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this application should have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0021] As described in the background section, in the current network architecture, with the promotion of IPv6 and the depletion of IPv4 addresses, the transition from IPv4 to IPv6 has become an important issue in global network operation and maintenance. In the power industry, with the rapid development of smart grids, Internet of Things (IoT) devices and big data applications, network communication plays a crucial role in the stability and reliability of the power system. With the widespread deployment of smart grids, automation equipment and IoT devices, the requirements for network communication have become more complex. In particular, real-time data transmission between power terminal equipment (such as substation automation systems, power dispatch centers, power plants, etc.) needs to ensure low latency and high reliability. The communication characteristics of the power industry are as follows: (1) High load and time-varying traffic: The load of the power system fluctuates greatly, especially during peak periods, and network traffic may increase sharply; (2) High reliability requirements: The communication of the power system requires high availability and low failure rate. Any network failure may affect the normal operation of the power system; (3) Cross-domain communication: The power network usually covers multiple regions and involves different autonomous systems (AS) and devices. Cross-domain connections of IPv6 are more challenging.

[0022] However, the power industry's communication networks face some unique challenges, especially during the transition from IPv4 to IPv6. The main difference between IPv4 and IPv6 lies in the size of the address space: IPv4 uses 32-bit addresses (approximately 4.2 billion addresses), while IPv6 uses 128-bit addresses (a virtually unlimited address space). Although IPv6 offers a viable solution to the address exhaustion problem, the transition is complex because many network devices and applications still rely on the IPv4 protocol. Since most of these devices use the IPv4 protocol, ensuring a smooth transition and maintaining system stability and efficiency after the introduction of IPv6 has become a pressing technical issue.

[0023] In practical applications, the transition to IPv6 mainly relies on the following technologies: (1) NAT64 (Network Address Translation from IPv6 to IPv4): an IPv6 / IPv4 translation technology that allows IPv6 clients to access IPv4 resources.

[0024] (2) IVI (Internet Virtualization Infrastructure, IPv4 and IPv6 interoperability technology based on IVI technology): a technology for IPv6 transition, which enables IPv6 addresses to communicate with IPv4 devices through IPv4 / IPv6 translation.

[0025] (3) MAP-T / E (Mapping of Address and Port translation, MAP-T: address and port mapping technology with translation, MAP-E: address and port mapping technology with encapsulation): a technology based on address and port mapping that enables bidirectional communication between IPv4 and IPv6. Traditional IPv4 / IPv6 translation technologies (such as NAT64, IVI, MAP-T / E, etc.) often use fixed translation parameters and cannot effectively cope with changes in dynamic network environments.

[0026] Existing IPv4 / IPv6 translation technologies typically employ static configuration, which means using fixed translation rules (such as fixed PSID, prefix, and pattern). However, this approach is prone to the following problems when facing high traffic and cross-domain scenarios: (1) Port conflicts: Due to the limitation of PSID (source port identifier) ​​length, multiple users may share the same port, leading to collisions and conflicts in network traffic.

[0027] (2) Delay and jitter: Fixed translation parameters may not be able to adapt to changes in network conditions, resulting in delay and jitter, which affects the power system's demand for real-time data.

[0028] (3) Throughput bottleneck: Under high load and high concurrency, the static translation mechanism cannot effectively utilize network bandwidth, resulting in a decrease in throughput.

[0029] Existing IPv4 / IPv6 translation technologies (such as NAT64, IVI, MAP-T / E) have the following limitations: (1) Lack of adaptability: Fixed translation parameters cannot dynamically adapt to changes in network load and environment, resulting in performance degradation. Especially in high-load network environments, dynamic network environments make the adaptability of translation rules composed of translation parameters low.

[0030] (2) Cross-domain scenario problem: The communication environment of the power industry usually involves multiple cross-domain data transmissions. However, the existing IPv4 / IPv6 translation technology does not perform well in such complex scenarios. The complex communication environment makes the adaptability of the translation rules composed of translation parameters low.

[0031] Furthermore, existing IPv4 / IPv6 translation technologies cannot meet privacy and compliance requirements. Data privacy and security are critical issues facing the power industry during network optimization. Traditional optimization methods typically rely on centralized data collection and processing, but protecting data privacy becomes a challenge in a strict compliance environment.

[0032] In view of this, this application proposes a method for determining the translation rules of Internet Protocol version. Referring to Figure 1, it is applied to a power system, which includes a central translation node and multiple sub-translation nodes. The method includes the following steps: Step 101, for each sub-translation node, in response to determining that a communication request from a client has been received, the sub-translation node obtains the Internet Protocol version of the client.

[0033] In this step, devices with different Internet Protocol (IP) versions cannot communicate directly; IP address translation is required to ensure that the IP versions of the devices are consistent before communication can occur. To ensure that external clients can communicate with the power system, it is first necessary to determine whether the client's IP version is the same as the power system's. For example, the client may be a power terminal, sensor, or inverter. The power system includes a central translation node, multiple sub-translation nodes, and multiple functional nodes, which perform core functions such as data acquisition, processing, analysis, control, and communication. To ensure the stability of IP version translation, this application introduces federated learning. Federated learning, through local training at each sub-translation node and uploading only model parameters, avoids centralized data storage and processing, effectively protecting data privacy while ensuring the stability of IP version translation. Upon receiving a communication request from an external client, each sub-translation node obtains the client's IP version.

[0034] Step 102: In response to determining that the Internet Protocol version is different from the Internet Protocol version of the power system, the sub-translation node obtains the network traffic data corresponding to the client.

[0035] In this step, the client's network traffic data is dynamically changing, and this data affects the determination of translation parameters. Therefore, if it is determined that the client's Internet Protocol version is different from the power system's Internet Protocol version, for example, if the client sends IPv4 packets with Internet Protocol version 4, while the power system requires Internet Protocol version 6, then it is determined that the client's Internet Protocol version is different from the power system's. Similarly, if the client sends IPv6 packets with Internet Protocol version 6, while the power system requires Internet Protocol version 4, then it is determined that the client's Internet Protocol version is different from the power system's.

[0036] Each translation node independently monitors the traffic of each connected client (whether IPv4 or IPv6) to obtain corresponding network traffic data. This network traffic data reflects the communication quality and status between the client and the power system. It includes both current and historical network traffic data, encompassing at least current and historical network load, latency, packet loss, and window conflicts. Network load reflects the amount of data transmitted or the intensity of traffic. By acquiring network load data, the translation node can dynamically perceive network resource occupancy. Latency is a key indicator of network response speed; by acquiring latency data, the translation node can understand the current transmission efficiency. Packet loss rate reflects the reliability of network transmission; by acquiring packet loss data, the translation node can assess the current stability of the network. Window conflicts are typically related to the TCP protocol's flow control mechanism; by acquiring window conflict data, the translation node can understand the current network congestion status.

[0037] Step 103: Based on the network traffic data, the sub-translation nodes perform multiple rounds of adjustment operations. Each round of adjustment operations is performed as follows: Based on the network traffic data, the sub-translation nodes use the model parameters in the pre-trained federated learning model and the pre-built agent to determine the translation parameter combination corresponding to the Internet Protocol version of the client, and upload the model parameters to the central translation node.

[0038] In this step, to determine the optimal combination of translation parameters, multiple rounds of adjustment operations are performed on the sub-translation nodes based on network traffic data. Each round of adjustment is executed as follows: The sub-translation nodes use the model parameters in a pre-trained federated learning model and a pre-built agent to determine the combination of translation parameters corresponding to the client's Internet Protocol version. Each sub-translation node deploys a federated learning model and an agent. The federated learning model is used to predict the network state of the power system, where the network state characterizes the changing trend of network traffic data. The agent is used to determine the combination of translation parameters corresponding to the client's Internet Protocol version based on the network state. The combination of translation parameters consists of multiple translation parameters used to translate the client's Internet Protocol version, and includes at least the source port identifier (PSID) length, prefix length, and translation mode (such as IVI, MAP-T, NAT64).

[0039] By utilizing a federated learning model to process dynamic network traffic data and then employing an agent to determine the optimal combination of translation parameters under the current network conditions, the dynamic nature of power system network traffic data is taken into account when determining the translation parameter combination. This allows for flexible determination of the translation parameter combination, rather than adhering to fixed rules, thus enhancing the adaptability of the translation parameter combination. Simultaneously, using an agent to determine the translation parameter combination also ensures its accuracy. The scheme combining federated learning and reinforcement learning can adjust the IPv4 / IPv6 translation parameter combination in real time to cope with the complex network environment in the power industry, improving the adaptability of the translation parameter combination and enhancing system stability and performance. For example, the federated learning model incorporates LSTM (Long Short-Term Memory Networks) or TFT (Temporal Fusion Transformer, a hybrid architecture combining LSTM and self-attention). To ensure more accurate prediction results from the federated learning model, model parameters are uploaded to a central translation node, where model parameters include at least model gradients and model weights.

[0040] Step 104: The central translation node aggregates the model parameters uploaded by each sub-translation node to obtain aggregated model parameters.

[0041] In this step, as shown in Figure 2, which is a schematic diagram of the federated learning model training and aggregation process in this embodiment, the central translation node aggregates the model parameters uploaded by each sub-translation node to obtain aggregated model parameters. The main role of the central translation node is to indirectly influence the decisions of all sub-translation nodes through the aggregation of model parameters. Through the aggregation and feedback mechanism of model parameters, it ensures that the translation combinations of each sub-translation node gradually tend towards the global optimum, thereby optimizing the overall network performance. On the one hand, the sub-translation nodes only upload model parameters, not the original data, and the central node cannot directly access the original data, protecting data privacy. On the other hand, the aggregated model parameters, compared with the original model parameters, improve the ability to capture global features, thereby improving the prediction accuracy of the model.

[0042] Step 105: Based on the model parameters uploaded by each sub-translation node and the aggregated model parameters, the central translation node determines whether the federated learning model has converged.

[0043] In this step, the core of convergence judgment is monitoring the changing trend of model parameters or performance indicators. When the change is lower than a preset threshold, the federated learning model is considered to have converged. In federated learning, the aggregated model parameters obtained after each aggregation are gradually optimized. If the change of each identical parameter in the model parameters uploaded by each sub-translation node and the aggregated model parameters is less than a predetermined threshold (e.g., the L2 norm difference is lower than the threshold), it indicates that the federated learning model is close to a stable state. Therefore, based on the model parameters uploaded by each sub-translation node and the aggregated model parameters, the central translation node determines whether the federated learning model has converged, ensuring that the current federated learning model has the optimal prediction result and avoiding overfitting.

[0044] Step 1061: In response to determining that the federated learning model has not converged, the aggregated model parameters are used as the model parameters in the federated learning model in the next round of adjustment operations, and the next round of adjustment operations is performed.

[0045] In this step, if the federated learning model fails to converge, it indicates that the model has not yet reached its optimal prediction result. The aggregated model parameters are then used as the model parameters for the next round of adjustment. The central translation node distributes the aggregated model parameters to each sub-translation node, and all sub-translation nodes continue local training based on these parameters. The aggregated model parameters reflect the "trend" of the optimal translation combination across all sub-translation nodes; that is, each node adjusts its translation parameters based on the guidance of the trained federated learning model to determine the current optimal combination. Through multiple rounds of adjustment, the federated learning model learns the complex patterns and dynamic changes in power system network traffic data. This capability enables the federated learning model to accurately predict future trends in network conditions, such as latency, packet loss rate, and window conflicts, enhancing the adaptability of the determined translation parameter combinations and providing a reliable basis for their determination.

[0046] Step 1062: In response to determining that the federated learning model has converged, the central translation node obtains the translation parameter combination corresponding to each sub-translation node, and determines the translation rule corresponding to the client based on the translation parameter combination corresponding to each sub-translation node, and exits at least one round of adjustment operation.

[0047] In this step, once the federated learning model has converged, it indicates that the model has reached its optimal prediction performance. The agent uses the network state predicted by the federated learning model to determine the translation parameter combinations, which are accurate. At this point, the central translation node obtains the translation parameter combinations corresponding to each sub-translation node and, based on these combinations, determines the translation rules for the client, exiting at least one round of adjustment. Through multiple rounds of adjustment, the sub-translation nodes can progressively optimize the translation parameter combinations. Each round of adjustment is based on the results of the previous round and network state feedback, forming a closed-loop optimization that eventually converges to the globally optimal or near-optimal solution. Finally, the central translation node determines the translation rules for the client based on the translation parameter combinations corresponding to each sub-translation node. By dynamically determining the translation rules, end-to-end latency can be significantly reduced (P95 latency reduced by 15%–25%), while throughput is increased (increased by 10%–18%), demonstrating the enhanced adaptability of the translation rules, thereby improving network performance and reducing latency. This application can adaptively adjust to different business scenarios, effectively avoiding port conflicts, ensuring smooth business operations during IPv6 migration, and exhibiting high availability and fault tolerance. It also effectively improves the adaptability of defined translation rules. Furthermore, this application employs federated learning to ensure that data privacy is fully protected while performing network-wide optimization, meeting the stringent compliance requirements of the power industry.

[0048] The adaptive optimization method for IPv4 / IPv6 translation parameters based on federated reinforcement learning provided in this application aims to solve problems such as latency, port conflicts, and throughput bottlenecks encountered during IPv6 translation in the power industry, especially in high-load and cross-domain environments. As shown in Figure 3, which compares the translation parameter optimization effects of embodiments of this application, the translation rules of this invention outperform statically configured translation technologies in terms of end-to-end latency, throughput, and port collision rate. This application effectively solves the static configuration problem existing in current translation technologies, dynamically adjusting translation parameters to adapt to complex scenarios such as high load, cross-domain, and time-varying traffic in the power industry, thus enhancing the adaptability of translation rules. Federated reinforcement learning (FRL) adopts a collaborative approach of distributed training and centralized decision-making. Through federated learning, each sub-translation node (such as a translation gateway, router, etc.) can train the model locally, calculate updated model parameters, and then upload the model parameters to the central translation node. The central translation node aggregates the model parameters of all sub-translation nodes to obtain the optimal model parameters for the entire network. By employing a reinforcement learning-based agent, the optimization strategy can dynamically adjust translation parameters based on the real-time state of the network (such as load, latency, packet loss, etc.). Through the combination of federated learning and reinforcement learning, this application not only enhances the adaptability of translation rules while ensuring privacy, but also improves the performance and stability of power industry networks, providing a feasible technical solution for the smooth transition to IPv6.

[0049] In the above scheme, for each sub-translation node, in response to determining that a communication request from a client has been received, the sub-translation node obtains the Internet Protocol version of the client; in response to determining that the Internet Protocol version is different from the Internet Protocol version of the power system, the sub-translation node obtains the network traffic data corresponding to the client; based on the network traffic data, the sub-translation node performs multiple rounds of adjustment operations, each round of adjustment operations being performed as follows: based on the network traffic data, the sub-translation node uses the model parameters in a pre-trained federated learning model and a pre-built intelligent agent to determine the translation parameter combination corresponding to the Internet Protocol version of the client, and uploads the model parameters to the central translation node. The central translation node aggregates the model parameters uploaded by each sub-translation node to obtain aggregated model parameters. Based on the model parameters uploaded by each sub-translation node and the aggregated model parameters, the central translation node determines whether the federated learning model has converged. In response to determining that the federated learning model has not converged, the aggregated model parameters are used as the model parameters in the federated learning model in the next round of adjustment, and the next round of adjustment is performed. In response to determining that the federated learning model has converged, the central translation node obtains the translation parameter combination corresponding to each sub-translation node, and based on the translation parameter combination corresponding to each sub-translation node, determines the translation rule corresponding to the client, and exits at least one round of adjustment.

[0050] In some embodiments, based on the network traffic data, the sub-translation node determines the translation parameter combination corresponding to the Internet Protocol version of the client using model parameters in a pre-trained federated learning model and a pre-built agent. This includes: predicting the network state based on the network traffic data using model parameters in the federated learning model; and determining the translation parameter combination based on the network state using the agent. In this embodiment, the network traffic data corresponding to the client is input into the federated learning model, and the network traffic data is processed using model parameters in the federated learning model to determine and output the network state of the power system. The network state reflects the trend of network traffic changes in the power system. This refers to the comprehensive reflection of key performance indicators of the network at time step k, such as load, latency, packet loss rate, and port conflicts. It is not merely described by a single network load or latency, but rather a network performance state derived from a combination of factors (e.g., load, latency, packet loss rate). It should be noted that each Internet Protocol version (IPv4 or IPv6) may have different network and transmission characteristics; therefore, the translation nodes will train corresponding federated learning models based on the Internet Protocol version of the power system. For example, the federated learning model might be a load-latency prediction model.

[0051] Based on network state, an agent is used to determine the combination of translation parameters. The architecture of the agent based on Deep Reinforcement Learning (DRL) includes: a state space: a description of the network state input to the federated learning output; an action space: for adjusting translation parameters; and a reward function: a multi-objective optimization that integrates packet loss rate, port collision rate, and throughput bottlenecks. Since network traffic data is dynamically changing, the federated learning model can determine the trend of network traffic changes in the power system. Then, under this trend, the agent is used to determine the combination of translation parameters, making the determination of the translation parameter combination both flexible and accurate.

[0052] This application combines Federated Learning (FL) and Reinforcement Learning (RL) to adaptively optimize core parameters in the IPv4 / IPv6 translation process, such as source port identifier (PSID) length, prefix length, and translation mode (e.g., IVI, MAP-T, NAT64), to minimize network latency, packet loss rate, and port collision rate, while simultaneously improving throughput, ultimately ensuring network performance and stability during the IPv6 transition. For example, the reinforcement learning is PPO (Proximal Policy Optimization) / CRL (Continuous Reinforcement Learning), and constraints are handled using Lagrange multipliers.

[0053] It should be noted that before inputting the network traffic data corresponding to the client into the federated learning model, the network traffic data is preprocessed. Specifically, the network traffic data is denoised, normalized, missing values ​​are filled in, and timestamps are added to synchronize multi-source data.

[0054] In some embodiments, determining the translation parameter combination based on the network state using the agent includes: determining an initial translation parameter combination based on the network state using the agent; and adjusting the initial translation parameter combination through a reward function in the agent to obtain the translation parameter combination. In this embodiment, the agent can quickly generate the initial translation parameter combination by inputting the network state, demonstrating the agent's rapid decision-making capability. The initial translation parameter combination may not be the globally optimal solution. In this case, the reward function in the agent will perform closed-loop adjustment of the initial translation parameter combination based on real-time feedback. This process, through the trial-and-error and feedback mechanism in reinforcement learning, allows the initial translation parameter combination to gradually converge to the optimal state in a dynamic environment. Reinforcement learning algorithms (such as PPO, Proximal Policy Optimization) are used to optimize the initial translation parameter combination. As shown in Figure 4, Figure 4 is a flowchart of the translation parameter combination optimization based on PPO in this embodiment. Each sub-translation node selects an action according to the current network state and evaluates the result through a reward function. Through multiple training and adjustments, the optimal translation parameter combination setting is gradually converged. By interacting with the network environment, the agent adaptively adjusts the combination of translation parameters to optimize network performance, thereby reducing latency, increasing throughput, and lowering port collision rates. It should be noted that the PSID mapping in the translation parameter combination follows the extended modulo algorithm and sets port security zones.

[0055] Reinforcement learning is used to adaptively optimize IPv4 / IPv6 translation parameter combinations to reduce end-to-end latency, port conflicts, and improve throughput. The optimization process includes the following steps: Action Space: The selection of translation parameter combinations (such as PSID length, prefix length, translation mode, etc.) serves as the action space in reinforcement learning.

[0056] State space: The state consists of metrics such as current network load, latency, and packet loss rate. For example, the state can be represented as: .

[0057] Reward Function: The reward function measures the effectiveness of the current translation parameter settings. The design of the reward function needs to consider various network metrics. The specific formula for the reward function is as follows: Formula 1, where, The network state at time step k. For the time delay at time step k, The weights corresponding to network performance metrics. The weights corresponding to the packet loss rate. The weights corresponding to the port conflict rate. The weights are the weights corresponding to the throughput bottlenecks, and these weights are used to adjust the impact of various indicators on the reward function.

[0058] In some embodiments, determining the translation rule corresponding to the client based on the translation parameter combination corresponding to each sub-translation node includes: grouping parameters belonging to the same first predetermined type in all translation parameter combinations into the same sequence; for each sequence, counting the number of elements with the same name in the sequence, and determining the element corresponding to the largest number of elements in the sequence as the target translation parameter; and using all target translation parameters as the translation rule corresponding to the client. In this embodiment, the parameters belonging to the same first predetermined type in the translation parameter combinations are grouped into the same sequence. The first predetermined type includes at least a PSID length type, a prefix length type, and a translation mode type. The PSID length corresponding to each sub-translation node may vary between [16, 32], the prefix length may vary between [64, 96], and the translation mode may be one of IVI, MAP-T, and NAT64. For each sequence, counting the number of elements with the same name in the sequence, and determining the element corresponding to the largest number of elements in the sequence as the target translation parameter. For example, for the translation mode sequence {IVI, MAP-T, IVI}, the target translation parameter is IVI. It should be noted that when the maximum number of elements corresponds to multiple elements, the target translation parameter is randomly determined from among these multiple elements. For example, for the translation pattern sequence {IVI, MAP-T, IVI, MAP-T, NAT64}, the target translation parameter is either IVI or MAP-T. All target translation parameters corresponding to the first predetermined type are used as the translation rules for the client. In the optimal combination of translation parameters across all translation nodes, the element that appears most frequently within the same type is used as the target translation parameter, ensuring the reliability of the target translation parameter and thus the reliability of the translation rules.

[0059] In some embodiments, the central translation node aggregates the model parameters uploaded by each sub-translation node to obtain aggregated model parameters, including: calculating the average parameter corresponding to each type of model parameters uploaded by each sub-translation node using a federated averaging algorithm; and using all average parameters as the aggregated model parameters. In this embodiment, the federated averaging algorithm is used to calculate the average parameter corresponding to each type of model parameters uploaded by each sub-translation node using the following formula: Formula 2, where, It is the average parameter after aggregation of the predefined type t. It is the first The model parameters of the predefined type t in each translation node. It is the total number of model parameters of the predefined type t.

[0060] By aggregating model parameters belonging to the same second predetermined type in all sub-translation nodes, aggregated model parameters are obtained, effectively fusing multi-source parameters, so that the aggregated model parameters can accurately optimize the prediction trend of all sub-translation nodes.

[0061] In some embodiments, using the aggregated model parameters as model parameters in the federated learning model in the next round of adjustment includes: the central translation node sending the aggregated model parameters to each sub-translation node, and each sub-translation node replacing each average parameter in the aggregated model parameters with the average parameter in the federated learning model that belongs to the same second predetermined type. In this embodiment, the central translation node sends the aggregated model parameters to each sub-translation node, and each sub-translation node uses the aggregated model parameters to replace the parameters of its local federated learning model. Specifically, for each average parameter in the aggregated model parameters, the average parameter in the federated learning model is replaced with the average parameter. It should be noted that each average parameter in the aggregated model parameters has a one-to-one correspondence with the parameters in the federated learning model. The average parameter in the aggregated model parameters and the parameter that belongs to the same second predetermined type in the federated learning model have the same functional position in the federated learning model. For example, the model weights of the federated learning model are replaced with the average model weights that have the same position. By replacing model parameters in federated learning models, the model's generalization and stability can be effectively improved, balancing individual and global needs and adapting to dynamic environments.

[0062] In some embodiments, after exiting at least one round of adjustment operations, the method includes: the central translation node translating the Internet Protocol (IP) address corresponding to the client according to the translation rules. In this embodiment, after determining the translation rules for the client, the central translation node obtains the IP address corresponding to the client, translates the IP address according to the translation rules, and the IP version corresponding to the translated IP address is the same as the IP version of the power system. The translation rules are highly adaptable, and the client communicates with the power system using the IP address translated according to the translation rules, ensuring stable communication between the client and the power system.

[0063] In another embodiment provided in this application, as shown in Figure 5, which is a schematic diagram of the IPv4 / IPv6 translation optimization system architecture based on federated learning according to this application embodiment, each sub-translation node publishes optimized translation rules after local training and federated learning aggregation. After the power system deploys the optimized translation rules corresponding to the client, a canary release strategy is adopted to gradually cover different areas of the power system, ensuring that performance improvement does not cause other network problems. The optimized translation rules corresponding to the client are promoted throughout the entire network through canary release, ensuring that the optimization process does not negatively affect network performance. The power system generates and saves complete operation logs to ensure compliance with the power industry's requirements for data privacy and compliance. If performance anomalies occur in a certain area, the power system can promptly roll back the optimization strategy to ensure network stability.

[0064] Because of the use of a federated learning framework, data is always kept locally, avoiding privacy leaks caused by large-scale centralized data processing. All network optimization processes are strictly carried out in accordance with the compliance requirements of the power industry, and all operation logs (including translation parameter adjustments, policy switching, etc.) are encrypted and stored for at least 180 days for auditing and traceability.

[0065] It should be noted that the method in this embodiment can be executed by a single device, such as a computer or server. The method can also be applied in a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method in this embodiment, and the multiple devices will interact with each other to complete the method described.

[0066] It should be noted that the above description describes some embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0067] Based on the same inventive concept, corresponding to any of the above embodiments, this application also provides an Internet Protocol version translation rule determination device.

[0068] Referring to Figure 6, the Internet Protocol version translation rule determination device is applied to a power system, which includes a central translation node and multiple sub-translation nodes. The device includes a first acquisition module 10, configured to acquire the Internet Protocol version of the client for each sub-translation node in response to determining that a communication request from a client has been received.

[0069] The second acquisition module 20 is configured to, in response to determining that the Internet Protocol version is different from the Internet Protocol version of the power system, acquire the network traffic data corresponding to the client.

[0070] Based on the network traffic data, the sub-translation nodes perform multiple rounds of adjustment operations. Each round of adjustment operations is performed as follows: The first determination module 30 is configured to, based on the network traffic data, use the model parameters in the pre-trained federated learning model and the pre-built intelligent agent to determine the translation parameter combination corresponding to the Internet Protocol version of the client, and upload the model parameters to the central translation node.

[0071] The aggregation module 40 is configured to aggregate the model parameters uploaded by each sub-translation node by the central translation node to obtain aggregated model parameters.

[0072] The second determining module 50 is configured to determine whether the federated learning model has converged based on the model parameters uploaded by each sub-translation node and the aggregated model parameters.

[0073] The third determining module 601 is configured to, in response to determining that the federated learning model has not converged, use the aggregated model parameters as the model parameters in the federated learning model in the next round of adjustment operations, and perform the next round of adjustment operations.

[0074] The third determining module 602 is configured to, in response to determining that the federated learning model has converged, obtain the translation parameter combination corresponding to each sub-translation node, and determine the translation rule corresponding to the client based on the translation parameter combination corresponding to each sub-translation node, and exit at least one round of adjustment operation.

[0075] Using the aforementioned apparatus, for each sub-translation node, in response to determining that a communication request from a client has been received, the sub-translation node obtains the Internet Protocol version of the client; in response to determining that the Internet Protocol version is different from the Internet Protocol version of the power system, the sub-translation node obtains the network traffic data corresponding to the client; based on the network traffic data, the sub-translation node performs multiple rounds of adjustment operations, each round of adjustment operations being performed as follows: based on the network traffic data, the sub-translation node uses the model parameters in a pre-trained federated learning model and a pre-built intelligent agent to determine the translation parameter combination corresponding to the client's Internet Protocol version, and uploads the model parameters to the central translation node. The central translation node aggregates the model parameters uploaded by each sub-translation node to obtain aggregated model parameters. Based on the model parameters uploaded by each sub-translation node and the aggregated model parameters, the central translation node determines whether the federated learning model has converged. In response to determining that the federated learning model has not converged, the aggregated model parameters are used as the model parameters in the federated learning model in the next round of adjustment, and the next round of adjustment is performed. In response to determining that the federated learning model has converged, the central translation node obtains the translation parameter combination corresponding to each sub-translation node, and based on the translation parameter combination corresponding to each sub-translation node, determines the translation rule corresponding to the client, and exits at least one round of adjustment.

[0076] In some embodiments, the first determining module 30 is further configured to predict the network state based on the network traffic data using model parameters in the federated learning model; and to determine the translation parameter combination using the agent based on the network state.

[0077] In some embodiments, the first determining module 30 is further configured to determine an initial translation parameter combination based on the network state using the agent; and to adjust the initial translation parameter combination through a reward function in the agent to obtain the translation parameter combination.

[0078] In some embodiments, the third determining module 602 is further configured to divide parameters belonging to the same first predetermined type in all translation parameter combinations into the same sequence; for each sequence, count the number of elements with the same name in the sequence, determine the element corresponding to the largest number of elements in the sequence as the target translation parameter; and use all target translation parameters as the translation rules corresponding to the client.

[0079] In some embodiments, the aggregation module 40 is further configured to use a federated averaging algorithm to calculate the average parameters of the same second predetermined type among the model parameters uploaded by each sub-translation node, and obtain the average parameters corresponding to each type; and use all the average parameters as the aggregated model parameters.

[0080] In some embodiments, the third determining module 601 is further configured such that the central translation node sends the aggregated model parameters to each sub-translation node, and each sub-translation node replaces the parameter in the federated learning model that belongs to the same second predetermined type as the average parameter with the average parameter for each average parameter in the aggregated model parameters.

[0081] In some embodiments, a translation module is also included, which is configured to, after exiting at least one round of adjustment operations, have the central translation node translate the Internet Protocol address corresponding to the client according to the translation rules.

[0082] For ease of description, the above devices are described in terms of function, divided into various modules. Of course, in implementing this application, the functions of each module can be implemented in one or more software and / or hardware.

[0083] The apparatus described above is used to implement the method for determining translation rules for the corresponding Internet Protocol version in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0084] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this application 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 Internet Protocol version translation rule determination method as described in any of the above embodiments.

[0085] Figure 7 shows a more specific hardware structure diagram of an electronic device provided in this embodiment. The device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected internally via the bus 1050.

[0086] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0087] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0088] The input / output interface 1030 is used to connect input / output modules to realize information input and output. The input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, speakers, vibrators, indicator lights, etc.

[0089] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0090] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0091] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0092] The electronic devices described above are used to implement the method for determining translation rules of the corresponding Internet Protocol version in any of the foregoing embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0093] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this application also provides a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the Internet Protocol version translation rule determination method as described in any of the above embodiments.

[0094] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0095] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the Internet Protocol version translation rule determination method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0096] Based on the same concept, corresponding to the methods of any of the above embodiments, this application also provides a computer program product, including computer program instructions, which, when run on a computer, cause the computer to execute the Internet Protocol version translation rule determination method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0097] It should be noted that the embodiments of this application can be further described in the following ways: It is understood that before using the technical solutions of the various embodiments in this disclosure, the user will be informed of the type, scope of use, and usage scenarios of the personal information involved in an appropriate manner, and the user's authorization will be obtained.

[0098] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose, based on the prompt message, whether to provide personal information to the software or hardware such as electronic devices, applications, servers, or storage media performing the operations of this disclosed technical solution.

[0099] As an optional but not limited implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0100] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0101] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this application is limited to these examples; under the concept of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this application as described above, which are not provided in detail for the sake of brevity.

[0102] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this application, the well-known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided drawings. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this application, and this also takes into account the fact that the details of the implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this application will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuits) have been set forth to describe exemplary embodiments of this application, it will be apparent to those skilled in the art that the embodiments of this application can be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0103] Although this application has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0104] The embodiments of this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of this application.

Claims

1. A method for determining translation rules for Internet Protocol versions, characterized in that, Applied to a power system, the power system includes a central translation node and multiple sub-translation nodes. The method includes: for each sub-translation node, in response to determining that a communication request from a client has been received, the sub-translation node obtains the Internet Protocol (IP) version of the client; in response to determining that the IP version is different from the IP version of the power system, the sub-translation node obtains network traffic data corresponding to the client; based on the network traffic data, the sub-translation node performs multiple rounds of adjustment operations, each round of adjustment operations being performed as follows: based on the network traffic data, the sub-translation node uses model parameters in a pre-trained federated learning model and a pre-built agent to determine the combination of translation parameters corresponding to the client's IP version, and then... Model parameters are uploaded to the central translation node; the central translation node aggregates the model parameters uploaded by each sub-translation node to obtain aggregated model parameters; based on the model parameters uploaded by each sub-translation node and the aggregated model parameters, the central translation node determines whether the federated learning model has converged; in response to determining that the federated learning model has not converged, the aggregated model parameters are used as model parameters in the federated learning model in the next round of adjustment, and the next round of adjustment is performed; in response to determining that the federated learning model has converged, the central translation node obtains the translation parameter combination corresponding to each sub-translation node, and based on the translation parameter combination corresponding to each sub-translation node, determines the translation rule corresponding to the client, and exits at least one round of adjustment.

2. The method according to claim 1, characterized in that, Based on the network traffic data, the sub-translation node uses model parameters in a pre-trained federated learning model and a pre-built agent to determine the translation parameter combination corresponding to the client's Internet Protocol version, including: predicting the network state based on the network traffic data using model parameters in the federated learning model; and determining the translation parameter combination based on the network state using the agent.

3. The method according to claim 2, characterized in that, The step of determining the translation parameter combination using the agent based on the network state includes: determining an initial translation parameter combination using the agent based on the network state; and adjusting the initial translation parameter combination through a reward function in the agent to obtain the final translation parameter combination.

4. The method according to claim 1, characterized in that, The step of determining the translation rules corresponding to the client based on the translation parameter combination corresponding to each sub-translation node includes: dividing parameters belonging to the same first predetermined type in all translation parameter combinations into the same sequence; for each sequence, counting the number of elements with the same name in the sequence, and determining the element corresponding to the largest number of elements in the sequence as the target translation parameter; and using all target translation parameters as the translation rules corresponding to the client.

5. The method according to claim 1, characterized in that, The central translation node aggregates the model parameters uploaded by each sub-translation node to obtain aggregated model parameters, including: using a federated averaging algorithm to calculate the average parameter corresponding to each type of model parameters uploaded by each sub-translation node. The average parameter of all parameters is then used as the aggregated model parameters.

6. The method according to claim 5, characterized in that, The step of using the aggregated model parameters as model parameters in the federated learning model in the next round of adjustment includes: the central translation node sending the aggregated model parameters to each sub-translation node, and each sub-translation node replacing each average parameter in the federated learning model that belongs to the same second predetermined type as the average parameter with the average parameter for each average parameter in the aggregated model parameters.

7. The method according to claim 1, characterized in that, After exiting at least one round of adjustment operations, the method includes: the central translation node translating the Internet Protocol address corresponding to the client according to the translation rules.

8. A device for determining translation rules for an Internet Protocol version, characterized in that, Applied to a power system, the power system includes a central translation node and multiple sub-translation nodes. The device includes: a first acquisition module configured to, for each sub-translation node, in response to determining that a communication request from a client has been received, acquire the Internet Protocol version of the client; a second acquisition module configured to, in response to determining that the Internet Protocol version is different from the Internet Protocol version of the power system, acquire network traffic data corresponding to the client; based on the network traffic data, the sub-translation nodes perform multiple rounds of adjustment operations, each round of adjustment operations being executed as follows: a first determination module configured to, based on the network traffic data, use model parameters in a pre-trained federated learning model and a pre-built intelligent agent to determine the combination of translation parameters corresponding to the client's Internet Protocol version, and upload the model parameters. The process involves: a central translation node; an aggregation module configured to aggregate the model parameters uploaded by each sub-translation node to obtain aggregated model parameters; a second determination module configured to determine whether the federated learning model has converged based on the model parameters uploaded by each sub-translation node and the aggregated model parameters; a third determination module configured to, in response to determining that the federated learning model has not converged, use the aggregated model parameters as model parameters in the next round of adjustment operations; and the third determination module further configured to, in response to determining that the federated learning model has converged, obtain the translation parameter combination corresponding to each sub-translation node, determine the translation rule corresponding to the client based on the translation parameter combination corresponding to each sub-translation node, and exit at least one round of adjustment operations.

9. 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 program, it implements the method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method described in any one of claims 1 to 7.

Images