Digital Twin Approach for Data Center Optical Networks

Abstract

The present application relates to a digital twin method for data center optical networks. The method is applied to an optical network, in which multiple edge computing nodes and a central node are deployed, and corresponding digital twin models are deployed on the edge computing nodes, including: obtaining edge computing node data, the edge computing node data including optical signal status information and device status information; inputting the edge computing node data into the corresponding digital twin model to obtain node prediction results; transmitting the edge computing node data and the node prediction results to the central node, performing optimization calculations based on the edge computing node data and the node prediction results to obtain optimization results, which are used to indicate the operation of coordinating edge computing nodes. This method uses a distributed digital twin model, which, compared to traditional centralized processing, can directly obtain and process data because the edge computing nodes are at key positions in the optical network, reducing the delay in data transmission and enhancing the overall computing power.

Description

Technical Field

[0001] The present invention belongs to the field of computer technology, and in particular relates to a digital twin method for data center optical networks. Background Art

[0002] With the development of computer technology, digital twin technology has emerged. As an important simulation and simulation tool, digital twin technology is playing an increasingly important role in data center management, operation, maintenance, and optimization. Applying digital twin technology to data center optical networks can reflect the physical state and operating status of the data center in real time, providing accurate and comprehensive information support to decision makers. However, traditional data center architectures often rely on centralized processing, and their digital twin models are often deployed on central nodes, which faces problems such as data transmission delays and limited computing resources. Summary of the Invention

[0003] Based on this, it is necessary to provide a digital twin method for data center optical networks to address the above technical issues, so as to improve the operating efficiency and responsiveness of data center optical networks.

[0004] In a first aspect, the present application provides a digital twin method for a data center optical network, which is applied to an optical network. The optical network is deployed with multiple edge computing nodes and a central node, and corresponding digital twin models are deployed on the edge computing nodes. The method includes:

[0005] Obtain edge computing node data, including optical signal status information and device status information;

[0006] Input edge computing node data into the corresponding digital twin model to obtain node prediction results;

[0007] The edge computing node data and node prediction results are transmitted to the central node, and optimization calculations are performed based on the edge computing node data and node prediction results to obtain optimization results. The optimization results are used to indicate the operation of coordinating edge computing nodes.

[0008] In one embodiment, the digital twin model includes an optical device model and an optical network performance evaluation model;

[0009] The optical device model is used to simulate the working principle of the optical device and output the predicted device performance indicators of the optical device; the optical network performance evaluation model is used to simulate and analyze the performance parameters of the optical network and output the predicted optical network performance indicators.

[0010] In one embodiment, an optimization calculation is performed based on edge computing node data and node prediction results to obtain an optimization result, including:

[0011] Fuse the edge computing node data and node prediction results to obtain node fusion data;

[0012] The node fusion data is input into the deep learning-based optimization model to obtain the optimization results.

[0013] In one embodiment, edge computing node data and node prediction results are fused to obtain node fusion data, including:

[0014] Build a fusion model;

[0015] Determine the error function based on edge computing node data and node prediction results;

[0016] Calculate the optimal model parameters based on the error function;

[0017] Update the best model parameters to the fusion model to obtain the best fusion model;

[0018] The edge computing node data and node prediction results are input into the optimal fusion model to obtain node fusion data.

[0019] In one embodiment, the method further comprises:

[0020] Generate scheduling instructions based on the optimization results. The scheduling instructions include task allocation, resource allocation, and parameter adjustment requirements for edge computing nodes.

[0021] Adjust the working parameters of the edge computing nodes according to the scheduling instructions.

[0022] In one embodiment, the method further comprises:

[0023] The node prediction result is compared with the preset threshold to obtain the node evaluation result, which is used to characterize whether there is an abnormality in the edge computing node and indicate whether to execute local scheduling measures.

[0024] In one embodiment, the method further comprises:

[0025] The edge computing node data, node prediction results, and optimization results are converted into visualization parameters; the visualization parameters are used to display the edge computing node data, node prediction results, and optimization results in the display medium.

[0026] In a second aspect, the present application also provides a digital twin system for a data center optical network, which is applied to an optical network. Multiple edge computing nodes and a central node are deployed in the optical network, and corresponding digital twin models are deployed on the edge computing nodes. The system includes:

[0027] A node data acquisition module is used to obtain edge computing node data, including optical signal status information and device status information;

[0028] The node prediction module is used to input edge computing node data into the corresponding digital twin model to obtain node prediction results;

[0029] The global optimization module is used to transmit edge computing node data and node prediction results to the central node, perform optimization calculations based on the edge computing node data and node prediction results, and obtain optimization results. The optimization results are used to indicate the operation of coordinating edge computing nodes.

[0030] In a third aspect, the present application also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a digital twin method for a data center optical network when executing the computer program.

[0031] In a fourth aspect, the present application also provides a computer-readable storage medium on which a computer program is stored, which, when executed by a processor, implements the steps of a digital twin method for a data center optical network.

[0032] The aforementioned digital twin method, system, computer device, and storage medium for data center optical networks utilize a distributed architecture to deploy digital twin models across multiple edge computing nodes within the data center optical network. This model collects edge computing node data in real time for simulation analysis, resulting in prediction results for each edge computing node. This reduces data transmission latency and improves the operational efficiency and resource utilization of the data center optical network. By optimizing the edge computing node data and corresponding node prediction results at the central node, the optimized results are obtained, enabling the data center optical network to coordinate and control each edge computing node based on real-time load and forecasted demand, thereby enhancing the overall decision-making support capabilities of the data center optical network. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0034] Figure 1 A flowchart of a digital twin method for a data center optical network is provided in one embodiment of the present application;

[0035] Figure 2A structural schematic diagram of a digital twin method for a data center optical network is provided in one embodiment of the present application. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0037] In one embodiment, Figure 1 As shown, a digital twin method for a data center optical network is provided. This embodiment uses the method applied to a terminal as an example. It is understandable that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method is applied to an optical network, in which multiple edge computing nodes and a central node are deployed. Corresponding digital twin models are deployed on the edge computing nodes. The method includes the following steps:

[0038] S101: Acquire edge computing node data, where the edge computing node data includes optical signal status information and device status information.

[0039] Specifically, a data center optical network refers to an optical communication network used for high-speed data transmission within or between data centers. By deploying sensors or real-time data stream processing frameworks on multiple edge computing nodes in a data center optical network, edge computing node data can be acquired, enabling real-time monitoring of the data center optical network status. This edge computing node data can include both optical signal status information (such as the transmission rate, bit error rate, and spectrum), and device status information (such as device type, device parameters, and power consumption). Data center optical networks typically consist of optical amplifiers, optical routers, optical fibers, optical detectors, and optical switches to enable data transmission.

[0040] S102: Input the edge computing node data into the corresponding digital twin model to obtain the node prediction result.

[0041] A digital twin model is a virtual representation of a physical object, process, or system. It can interact with the dynamic behavior of the physical system through real-time data updates to simulate the actual operating state of the system. In a data center optical network, each edge computing node has a corresponding digital twin model. Real-time edge computing node data is input into the corresponding digital twin model, and simulation analysis of the edge computing node is performed, such as optical signal performance evaluation, device status prediction, fault prediction, and resource optimization. The node prediction results are obtained. These node prediction results can be used as feedback to further adjust and optimize the digital twin model, help monitor changes in the data center optical network status, and provide a basis for future decision-making, such as resource allocation, troubleshooting, and network planning, thereby improving the reliability of the data center optical network.

[0042] S103: The edge computing node data and the node prediction results are transmitted to the central node, and optimization calculations are performed based on the edge computing node data and the node prediction results to obtain optimization results. The optimization results are used to indicate the operation of coordinating the edge computing nodes.

[0043] Data from multiple edge computing nodes in a data center's optical network and corresponding node prediction results are transmitted to a central node for centralized optimization calculations. This optimization calculation can include resource allocation optimization, potential fault prediction, and other aspects, resulting in optimized results. First, the optimization objective is determined, such as minimizing optical network latency, maximizing bandwidth utilization, and reducing energy consumption, to guide the optimization calculations. Second, a corresponding optimization algorithm is designed based on the edge computing node data and node prediction results. For example, methods such as genetic algorithms, particle swarm optimization, and reinforcement learning can be used to dynamically adjust the operating parameters and scheduling strategies of edge computing nodes. This centralized processing can further simulate and evaluate the overall status of the data center's optical network. The optimization results can be used to control and coordinate each edge computing node, for example, by changing signal transmission power, updating routing selection, or adjusting device parameters. This enables efficient management and optimization of the data center's optical network, significantly improving the performance and stability of the optical network.

[0044] This method deploys multiple edge computing nodes and a central node within a data center optical network. After acquiring edge computing node data, a corresponding digital twin model is deployed on each edge node for simulation analysis. Compared to traditional centralized processing, which sends all raw data to a central node for analysis and decision-making, this significantly reduces the amount of data required to be transmitted to the central node, improving data transmission efficiency. By receiving edge computing node data and node prediction results, the central node can perform centralized optimization calculations, thereby more effectively coordinating the operations of each edge computing node. This optimization not only improves resource utilization and reduces energy consumption, but also enhances overall network performance.

[0045] In an exemplary embodiment, the digital twin model may include an optical device model and an optical network performance evaluation model;

[0046] The optical device model is used to simulate the working principle of the optical device and output the predicted device performance indicators of the optical device; the optical network performance evaluation model is used to simulate and analyze the performance parameters of the optical network and output the predicted optical network performance indicators.

[0047] Specifically, the optical device model can capture the input-output relationship of the optical device and its impact on signal processing by simulating the working mechanism of various optical devices (such as optical amplifiers, optical switches, optical routers, etc.) in the data center optical network based on the physical model, and obtain the predicted device performance indicators by inputting the acquired edge computing node data, such as the maximum data transmission rate of the optical device under different working conditions, the power consumption of the optical device under different load conditions, and the bit error rate during signal transmission.

[0048] The optical network performance evaluation model can adopt a hierarchical structure to construct a physical layer, a network layer, and a service layer to simulate the performance parameters of the data center optical network. The physical layer is used to simulate the propagation of optical signals in optical fibers, including signal attenuation, dispersion, nonlinear effects, etc.; the network layer is used to model the optical network topology, analyze the connection relationship between nodes, and calculate the bandwidth and delay of the path; the service layer is used to simulate the traffic pattern and service quality requirements of user requests, and evaluate the performance of the network under specific service requirements. Exemplarily, a numerical simulation of the propagation of simulated optical signals can be performed based on the finite difference method or the finite element method. The acquired edge computing node data is input into the optical network performance evaluation model to obtain predicted optical network performance indicators, such as throughput, bandwidth utilization, transmission delay, etc.

[0049] By leveraging optical device models and optical network performance evaluation models, multiple edge computing nodes within a data center optical network can more effectively understand and optimize optical network performance. This not only improves the adaptability and resource utilization of edge computing nodes, but also significantly enhances network reliability and service quality, thereby improving the operational efficiency of the entire optical network.

[0050] In an exemplary embodiment, an optimization calculation is performed based on edge computing node data and node prediction results to obtain an optimization result, including:

[0051] Fuse the edge computing node data and node prediction results to obtain node fusion data;

[0052] The node fusion data is input into the deep learning-based optimization model to obtain the optimization results.

[0053] By fusing edge computing node data with node prediction results at the central node, the resulting node-fused data provides a more comprehensive understanding of the operational status of the data center's optical network. This reduces potential errors in individual data sources, eliminates redundant data, and improves the accuracy of performance analysis and decision-making for the entire optical network. Deep learning, a machine learning technology, is well-suited for complex pattern recognition and prediction problems. By feeding this node-fused data into a deep learning-based optimization model, it maintains efficient prediction capabilities even in the face of nonlinear and complex network behaviors, such as traffic bursts and equipment failures. This optimizes the coordinated scheduling of edge computing nodes, improving the scientific nature and accuracy of decision-making and ensuring efficient and stable operation of the optical network.

[0054] In an exemplary embodiment, edge computing node data and node prediction results are fused to obtain node fusion data, including:

[0055] Build a fusion model;

[0056] Determine the error function based on edge computing node data and node prediction results;

[0057] Calculate the optimal model parameters based on the error function;

[0058] Update the best model parameters to the fusion model to obtain the best fusion model;

[0059] The edge computing node data and node prediction results are input into the optimal fusion model to obtain node fusion data.

[0060] Specifically, the fusion model can be a linear model, a nonlinear model or other complex machine learning model. Preferably, when the relationship between the data is more complex, compared with the linear model, nonlinear models such as polynomial regression and support vector machines can be used to capture more subtle and complex patterns. At the same time, these models can also avoid the high computational requirements that may be brought about by certain machine learning models to a certain extent. The error function is used to measure the difference between the output of the fusion model and the true value, and is usually evaluated using mean square error (MSE) and absolute error. Because the mean square error is more sensitive to the penalty of small deviations, it is more suitable for scenarios that require high precision and high sensitivity. Preferably, the mean square error can be selected as the error function, and its calculation formula is as follows:

[0061]

[0062] Where N is the number of edge computing nodes; M is the number of prediction values ​​provided by each edge computing node; y ij is the true value of the j-th predicted value on the i-th node; is the fusion model prediction value of the j-th prediction value on the i-th node.

[0063] A dataset is generated based on edge computing node data and node prediction results to train the fusion model. The dataset is divided into a training set and a validation set. The error function can be minimized through optimization algorithms such as Stochastic Gradient Descent (SGD) to update the model parameters. For example, the parameter update calculation process for each step using SGD is as follows:

[0064]

[0065] Among them, θ is the model parameter; α is the learning rate, a hyperparameter that controls the parameter update step size; is the partial derivative of MSE with respect to the parameter θ, i.e. the gradient.

[0066] When model training is complete, the optimal model parameters are obtained and updated to the fusion model to obtain the optimal fusion model. The performance of the training set and validation set is compared to ensure that the model has good predictive capabilities. The edge computing node data and node prediction results are input into the optimal fusion model to generate node fusion data. This data can include the fusion of optical signal status information collected on each edge computing node and its predicted value, the real-time status information of the devices on each edge computing node and its predicted status information, and the predicted values ​​of optical device performance and network performance.

[0067] In an exemplary embodiment, the method further includes:

[0068] Generate scheduling instructions based on the optimization results. The scheduling instructions include task allocation, resource allocation, and parameter adjustment requirements for edge computing nodes.

[0069] Adjust the working parameters of the edge computing nodes according to the scheduling instructions.

[0070] After the central node obtains the optimization results through optimization calculations, the system generates corresponding scheduling instructions based on the results. Scheduling instructions include task allocation, resource allocation, and parameter adjustment requirements for edge computing nodes. Specifically, task allocation determines which edge computing nodes will undertake specific computing or processing tasks to improve the optical network's responsiveness. Resource allocation dynamically adjusts the resource allocation of each edge computing node, such as bandwidth, storage, and computing power, based on the optimization results to ensure efficient resource utilization. Parameter adjustment requirements include specific adjustments to the operating parameters of each device, such as modifying the device's operating frequency, changing the optical signal transmission mode, and optimizing algorithm parameters, to improve device performance and the optical network's operating efficiency. Finally, the edge computing node adjusts its operating parameters accordingly based on the scheduling instructions. For example, during task allocation, the edge computing node identifies the specific tasks to be executed, including data processing, computing, and network communication tasks, based on the task allocation information in the scheduling instructions. Based on the importance and urgency of the tasks, the edge computing node uses a priority mechanism to adjust its task execution queue to ensure that high-priority tasks are prioritized.

[0071] In an exemplary embodiment, the method further includes:

[0072] The node prediction result is compared with the preset threshold to obtain the node evaluation result, which is used to characterize whether there is an abnormality in the edge computing node and indicate whether to execute local scheduling measures.

[0073] The edge computing node first acquires and collects its own status data, including optical signal status information and device status information. This data is then input into the corresponding digital twin model to generate node prediction results. These results reflect the expected performance and status of the edge computing node over the next period of time. Preset thresholds are various performance indicators developed by professionals based on historical data, industry standards, or operational experience to determine whether the node is operating normally. At the edge computing node, the node prediction results are compared with the preset thresholds to obtain the node evaluation results.

[0074] Indicatively, if the node prediction result is within the preset threshold, it indicates that the node is working properly. If the node prediction result exceeds the preset threshold (for example, the signal strength is lower than the acceptable level or the equipment load is too high), it indicates that the node has a potential abnormality. Based on the node evaluation results, the system can take corresponding local scheduling measures. For example, if the node has a potential abnormality, it will be optimized, which may include adjusting the signal strength, reallocating bandwidth, or rerouting the network. This step enables local decision-making of edge computing nodes, enhances the flexibility of the system, and allows edge computing nodes to make independent adjustments when encountering local failures or changes without waiting for global instructions. This autonomous capability improves the robustness and self-healing capabilities of the optical network, and can provide higher adaptability and stability in complex and dynamic optical network environments.

[0075] In an exemplary embodiment, the method further includes:

[0076] The edge computing node data, node prediction results, and optimization results are converted into visualization parameters; the visualization parameters are used to display the edge computing node data, node prediction results, and optimization results in the display medium.

[0077] Specifically, the process of converting edge computing node data, node prediction results, and optimization results into visual parameters involves transforming complex data information into easy-to-understand graphics or charts, such as trend charts, pie charts, or heat maps, to help relevant personnel quickly identify system status, performance indicators, and potential problems. The display of visual parameters can help operators quickly obtain the real-time status of each edge computing node in the optical network and understand the relationships and dependencies between nodes, thereby better supporting decision-making and operational coordination. In addition, through visual display, users can intuitively see the effects of node prediction results and optimization calculations, allowing timely adjustments to network configurations or maintenance. This not only improves system transparency but also enhances the efficiency of management and operation and maintenance, enabling faster and more effective responses and adjustments in complex situations.

[0078] Based on the same inventive concept, the present application also provides a digital twin system 200 for data center optical networks, such as Figure 2 As shown, it is applied to an optical network, in which multiple edge computing nodes and a central node are deployed, and corresponding digital twin models are deployed on the edge computing nodes. The system 200 includes:

[0079] Node data acquisition module 201, used to obtain edge computing node data, the edge computing node data includes optical signal status information and device status information;

[0080] The node prediction module 202 is used to input the edge computing node data into the corresponding digital twin model to obtain the node prediction result;

[0081] The global optimization module 203 is used to transmit edge computing node data and node prediction results to the central node, perform optimization calculations based on the edge computing node data and node prediction results, and obtain optimization results. The optimization results are used to indicate the operation of coordinating edge computing nodes.

[0082] Specifically, the system can deploy edge computing nodes at key locations in the optical network of the data center, which are closer to the data source than the central node. Therefore, the optical signal status information and device status information can be acquired and processed in real time at the edge computing node through the node data acquisition module 201 and the node prediction module 202. The local data is predicted and analyzed by utilizing the digital twin model deployed on the edge computing node to obtain the node prediction result. This step not only helps the edge computing node to self-adjust and optimize, but also provides valuable input for global optimization. Finally, the global optimization module 203 performs optimization calculations to obtain optimization results, which provide guidance on how to coordinate and adjust the operation of edge computing nodes, ensure that network resources are reasonably allocated, optimize overall performance, and achieve global decision-making. In summary, the system realizes efficient management and optimization of the optical network of the data center through the collaboration between multiple modules based on edge computing and digital twin technology.

[0083] In one embodiment, the digital twin model in the node prediction module 202 includes an optical device model and an optical network performance evaluation model;

[0084] The optical device model is used to simulate the working principle of the optical device and output the predicted device performance indicators of the optical device; the optical network performance evaluation model is used to simulate and analyze the performance parameters of the optical network and output the predicted optical network performance indicators.

[0085] In one embodiment, the global optimization module 203 is further configured to:

[0086] Fuse the edge computing node data and node prediction results to obtain node fusion data;

[0087] The node fusion data is input into the deep learning-based optimization model to obtain the optimization results.

[0088] Exemplarily, the global optimization module 203 is further configured to:

[0089] Build a fusion model;

[0090] Determine the error function based on edge computing node data and node prediction results;

[0091] Calculate the optimal model parameters based on the error function;

[0092] Update the best model parameters to the fusion model to obtain the best fusion model;

[0093] The edge computing node data and node prediction results are input into the optimal fusion model to obtain node fusion data.

[0094] In one embodiment, the system further includes a global scheduling module 204 for:

[0095] Generate scheduling instructions based on the optimization results. The scheduling instructions include task allocation, resource allocation, and parameter adjustment requirements for edge computing nodes.

[0096] Adjust the working parameters of the edge computing nodes according to the scheduling instructions.

[0097] In one embodiment, the system further includes a local decision module 205 for:

[0098] The node prediction result is compared with the preset threshold to obtain the node evaluation result, which is used to characterize whether there is an abnormality in the edge computing node and indicate whether to execute local scheduling measures.

[0099] In one embodiment, the system further includes a visualization module 206 for:

[0100] The edge computing node data, node prediction results, and optimization results are converted into visualization parameters; the visualization parameters are used to display the edge computing node data, node prediction results, and optimization results in the display medium.

[0101] Based on the same inventive concept, an embodiment of the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the steps of a digital twin method for a data center optical network. Specifically, the method may include steps such as data acquisition, model analysis, and optimization calculations, thereby utilizing distributed digital twin technology to monitor and manage the data center optical network, improve the performance and reliability of the optical network, and achieve flexible and efficient resource management.

[0102] Based on the same inventive concept, the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a digital twin method for a data center optical network. The computer storage medium may include random access memory, read-only memory, flash memory or other solid-state storage technology, digital versatile discs or other optical storage, tape cassettes, magnetic tape, disk storage or other magnetic storage devices.

[0103] The above-described embodiments merely represent several implementation methods of the embodiments of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that a person skilled in the art may make various modifications and improvements without departing from the concept of the embodiments of the present application, and these modifications and improvements fall within the scope of protection of the embodiments of the present application.

Claims

1. A digital twin method for data center optical networks, characterized in that: Applied to an optical network, wherein a plurality of edge computing nodes and a central node are deployed in the optical network, and corresponding digital twin models are deployed on the edge computing nodes, the method comprising: Acquire edge computing node data, wherein the edge computing node data includes optical signal status information and device status information; Input the edge computing node data into the corresponding digital twin model to obtain a node prediction result; Transmitting the edge computing node data and the node prediction result to the central node, performing optimization calculation based on the edge computing node data and the node prediction result to obtain an optimization result, wherein the optimization result is used to indicate and coordinate the operation of the edge computing node; At the edge computing node, the node prediction result is compared with a preset threshold to obtain a node evaluation result. The node evaluation result is used to characterize whether there is an abnormality in the edge computing node, and local scheduling measures are executed according to the node evaluation result. The local scheduling measures include adjusting signal strength and reallocating bandwidth.

2. The digital twin method according to claim 1, characterized in that: The digital twin model includes an optical device model and an optical network performance evaluation model; The optical device model is used to simulate the working principle of the optical device and output the predicted device performance index of the optical device; The optical network performance evaluation model is used to simulate and analyze the performance parameters of the optical network and output predicted optical network performance indicators.

3. The digital twin method according to claim 1, characterized in that: The performing optimization calculation based on the edge computing node data and the node prediction result to obtain the optimization result includes: Fusing the edge computing node data and the node prediction results to obtain node fusion data; The node fusion data is input into an optimization model based on deep learning to obtain an optimization result.

4. The digital twin method according to claim 3, characterized in that: The step of fusing the edge computing node data and the node prediction result to obtain node fusion data includes: Build a fusion model; Determine an error function based on the edge computing node data and the node prediction result; Calculating based on the error function to obtain optimal model parameters; Updating the optimal model parameters to the fusion model to obtain an optimal fusion model; The edge computing node data and the node prediction result are input into the optimal fusion model to obtain the node fusion data.

5. The digital twin method according to claim 1, characterized in that: Also includes: Generate a scheduling instruction based on the optimization result, wherein the scheduling instruction includes task allocation, resource allocation, and parameter adjustment requirements of the edge computing node; Adjust the operating parameters of the edge computing node according to the scheduling instruction.

6. The digital twin method according to claim 1, characterized in that: Also includes: Converting the edge computing node data, the node prediction results, and the optimization results into visualization parameters; The visualization parameters are used to display the edge computing node data, the node prediction results, and the optimization results in a display medium.

7. A digital twin system for data center optical networks, characterized by: Applied to an optical network, wherein a plurality of edge computing nodes and a central node are deployed in the optical network, and corresponding digital twin models are deployed on the edge computing nodes, the system includes: A node data acquisition module is used to acquire edge computing node data, wherein the edge computing node data includes optical signal status information and device status information; A node prediction module, configured to input the edge computing node data into the corresponding digital twin model to obtain a node prediction result; a global optimization module, configured to transmit the edge computing node data and the node prediction results to the central node, perform optimization calculations based on the edge computing node data and the node prediction results, and obtain an optimization result, wherein the optimization result is used to indicate and coordinate operations of the edge computing nodes; A local decision-making module is used to compare the node prediction result with a preset threshold at the edge computing node to obtain a node evaluation result, which is used to characterize whether there is an abnormality in the edge computing node, and to execute local scheduling measures based on the node evaluation result, wherein the local scheduling measures include adjusting signal strength and reallocating bandwidth.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the digital twin method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the digital twin method according to any one of claims 1 to 6 are implemented.

Images