Fault simulation and network performance improvement method and device based on digital twin model

Abstract

The invention discloses a fault simulation and network performance improvement method and device based on a digital twinning model, and belongs to the technical field of network communication and digital twinning. The method comprises the following steps: collecting and preprocessing real network physical entity data, and constructing a network digital twin model for realizing three-dimensional mapping; establishing a fault scene library based on the model, and injecting fault parameters for simulation to obtain a fault simulation result; setting an evaluation index system, quantitatively evaluating network performance and identifying a bottleneck; and formulating a performance improvement strategy, issuing the performance improvement strategy to a real network after simulation verification, and establishing a closed-loop feedback mechanism for continuous optimization. The device comprises a data acquisition module, a model construction module, a fault simulation module, a performance evaluation module, a strategy making and implementation module and a data interaction module, and can be additionally provided with a model updating module. According to the method, accurate fault simulation and efficient network performance optimization are realized, the network operation stability and the intelligent management level are improved, and the method adapts to a dynamic network environment and has good adaptability and expansibility.

Description

Technical Field

[0001] This invention relates to the intersection of network communication technology and digital twin technology, and more specifically, to a method and apparatus for fault simulation and network performance improvement based on a digital twin model. Background Technology

[0002] With the rapid development of technologies such as 5G, IoT, and cloud computing, network scale is constantly expanding, network topology is becoming increasingly complex, and the number of network devices is surging. This significantly increases the probability of network failures and places higher demands on the stability and efficiency of network performance. Network failures can lead to data transmission interruptions, service quality degradation, and even significant economic losses and adverse social impacts; while insufficient network performance will restrict the normal operation of various services and fail to meet users' needs for high-speed, low-latency network services.

[0003] In existing technologies, network fault simulation often employs mathematical models or static simulations. These methods frequently fail to accurately replicate the dynamic characteristics and complex interactions of real networks, leading to significant discrepancies between simulation results and actual conditions. Consequently, they cannot provide reliable references for fault diagnosis and response. Regarding network performance improvement, traditional methods rely heavily on manual experience for parameter adjustments or equipment upgrades. They lack real-time perception and dynamic optimization capabilities of network operating status, resulting in limited optimization effects and delayed responses, making it difficult to adapt to dynamically changing network environments.

[0004] Digital twin technology, capable of real-time mapping and bidirectional interaction between physical entities and virtual models, offers a new approach to solving the aforementioned problems. By constructing a digital twin model of a network, its operational status, topology, and device characteristics can be accurately simulated, enabling precise simulation and prediction of network faults. Simultaneously, leveraging the real-time data interaction and simulation analysis capabilities of the digital twin model, scientific and efficient decision support can be provided for network performance optimization. However, current solutions for applying digital twin technology to network fault simulation and performance improvement are still imperfect, suffering from insufficient model construction accuracy, limited fault simulation scenarios, and a lack of targeted performance optimization strategies.

[0005] Therefore, there is an urgent need for an efficient and accurate method and device for fault simulation and network performance improvement based on digital twin models. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and apparatus for fault simulation and network performance improvement based on a digital twin model. This method and apparatus achieve accurate simulation and real-time prediction of network faults by constructing a high-precision network digital twin model. At the same time, based on the simulation analysis results of the model, targeted performance optimization strategies are formulated to improve the stability and efficiency of network operation.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] The method for fault simulation and network performance improvement based on digital twin models includes the following steps:

[0009] Steps for building a network digital twin model:

[0010] Collect physical entity data of real networks, preprocess the collected data, and construct a digital twin model of the network based on the preprocessed data using multiphysics coupling simulation technology to realize geometric mapping, physical mapping and behavioral mapping between real networks and virtual models;

[0011] Fault simulation steps:

[0012] A fault scenario library is established based on the constructed network digital twin model. Target fault types are selected and fault parameters are set. After the fault parameters are injected into the model, simulation is started to simulate the occurrence and propagation of faults and obtain fault simulation results.

[0013] Network performance evaluation steps:

[0014] Establish a network performance evaluation index system, and quantitatively evaluate network performance based on real-time operating data and fault simulation results of the network digital twin model, generate a network performance evaluation report, and identify network performance bottlenecks.

[0015] Performance improvement strategy formulation and implementation steps:

[0016] Based on the network performance evaluation report and fault simulation results, performance improvement strategies are formulated, and the strategies are input into the model for simulation verification. After successful verification, the strategies are deployed to real network devices to achieve performance optimization, and a closed-loop feedback mechanism is established to continuously optimize the strategies.

[0017] Furthermore, the physical entity data includes network topology data, device parameter data, link status data, and service traffic data; preprocessing includes removing noise data, redundant data, and filling in missing data.

[0018] Furthermore, in the construction steps of the network digital twin model, a collaborative approach of edge computing and cloud computing is adopted. Edge nodes collect data with high real-time requirements, while the cloud computing platform performs big data processing and model construction and optimization.

[0019] Furthermore, in the fault simulation step, machine learning algorithms are used to analyze the fault simulation results, predict the probability and trend of fault occurrence, and provide support for fault early warning.

[0020] Furthermore, the network performance evaluation index system includes bandwidth utilization, transmission latency, packet loss rate, and service availability; the performance improvement strategies include route optimization, dynamic bandwidth allocation, device load balancing, and fault redundancy backup.

[0021] A fault simulation and network performance enhancement device based on a digital twin model includes:

[0022] Data acquisition module: Used to collect physical entity data from the real network;

[0023] Model building module: Connected to the data acquisition module, it is used to preprocess the acquired data and build a network digital twin model;

[0024] Fault simulation module: Connected to the model building module, it is used to establish a fault scenario library, inject fault parameters to perform fault simulation, and obtain fault simulation results;

[0025] Performance evaluation module: Connected to the model building module and the fault simulation module respectively, it is used to set the evaluation index system, quantitatively evaluate network performance and generate evaluation reports;

[0026] Strategy formulation and implementation module: Connected to the performance evaluation module, it is used to formulate performance improvement strategies, perform simulation verification, and issue verified strategies.

[0027] Data interaction module: Connects to each module to enable real-time data interaction between modules and between the model and real network devices.

[0028] Furthermore, the data acquisition module includes multiple edge acquisition nodes and a data aggregation unit. The edge acquisition nodes are deployed close to the network devices, and the data aggregation unit is used to summarize and initially filter the data before transmitting it to the model building module.

[0029] Furthermore, it also includes a model update module, which is connected to the data interaction module and the model building module respectively, and is used to dynamically update and calibrate the network digital twin model based on real network data and optimization results.

[0030] Beneficial effects of this invention:

[0031] (1) By constructing a network digital twin model, this invention realizes accurate mapping and two-way interaction between the real network and the virtual model. It can accurately simulate the occurrence and propagation process of various network faults. The fault simulation results are highly consistent with the actual situation, providing a reliable basis for fault investigation, early warning and response, and effectively reducing the losses caused by network faults.

[0032] (2) Based on the real-time data interaction and simulation analysis capabilities of the digital twin model, this invention can comprehensively and quantitatively evaluate network performance, accurately identify network performance bottlenecks, and formulate targeted performance improvement strategies. The effectiveness and feasibility of the strategies are verified through simulation before being applied to real networks, thus avoiding the risks of blind optimization and improving the efficiency and effectiveness of network performance optimization.

[0033] (3) This invention uses edge computing and cloud computing to collect data and build models, taking into account the real-time nature of data collection and the efficiency and accuracy of model building. At the same time, it establishes a performance optimization closed-loop feedback mechanism, which can dynamically update and calibrate the network digital twin model, continuously optimize performance improvement strategies, and enable the device to adapt to the dynamically changing network environment, with good adaptability and scalability.

[0034] (4) The device structure of the present invention is reasonable, and the modules work together to realize the full-process automation from data acquisition, model building, fault simulation, performance evaluation to strategy implementation, which reduces the cost of manual intervention and improves the level of network management intelligence. Attached Figure Description

[0035] Figure 1 This is a flowchart illustrating the implementation of the fault simulation and network performance improvement method based on a digital twin model according to the present invention.

[0036] Figure 2 This is a structural framework diagram of the fault simulation and network performance improvement device based on the digital twin model of the present invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0038] This invention proposes a method for fault simulation and network performance improvement based on a digital twin model, comprising the following steps:

[0039] Steps for building a network digital twin model:

[0040] By deploying intelligent acquisition terminals at core and edge nodes of the network, multi-dimensional physical entity data of the real network is collected. This physical entity data includes network topology data (including device connection relationships and port mapping relationships), device parameter data (including hardware configuration, firmware version, and operating thresholds), link status data (including real-time bandwidth, transmission rate, and bit error rate), and service traffic data (including traffic peaks, transmission protocols, and service types). The collected data is processed using an adaptive filtering algorithm to remove noise, a cosine similarity algorithm to eliminate redundant data, and an LSTM neural network-based missing value completion algorithm to accurately complete missing data. Based on the preprocessed data, a digital twin model of the network is constructed by integrating finite element analysis and multiphysics coupling simulation technology. Geometric mapping between the real network and the virtual model is achieved through 3D modeling technology, physical mapping through parametric modeling, and behavioral mapping through a dynamic simulation engine, ensuring that the real-time synchronization accuracy error between the model and the real network is less than 1%.

[0041] Fault simulation steps:

[0042] Based on the constructed network digital twin model, a fault scenario library is established by combining fault tree analysis and historical fault data mining. This library includes not only common fault types such as link interruption, node failure, bandwidth congestion, and high latency, but also scenarios involving multiple faults overlapping. Each fault type is labeled with its triggering conditions, propagation path characteristics, and impact weight. Target fault types are selected from the fault scenario library according to actual network operation and maintenance needs. Fault parameters covering dimensions such as fault location, trigger time, fault level, and duration are set. These standardized fault parameters are injected into the network digital twin model through a model interface. The model's dynamic simulation engine is then activated to simulate the occurrence and propagation of the target fault in the network at a 1:1 time scale. Network operation data is collected in real time during the simulation process. The final output includes the fault's impact range (accurate to specific devices and services), fault duration, impact on service quality (quantified as a percentage decrease in service quality indicators), and fault propagation path.

[0043] Network performance evaluation steps:

[0044] Based on the requirements of the Network Service Level Agreement (SLA), a multi-dimensional network performance evaluation index system is established. The evaluation indexes include bandwidth utilization (quantification threshold 0-100%), transmission delay (subdivided into one-way delay and round-trip delay), packet loss rate (thresholds are set according to service type), and service availability (the percentage of available time is statistically analyzed according to the time dimension). Based on the real-time operation data and fault simulation results of the network digital twin model, the weight of each evaluation index is determined by the analytic hierarchy process (AHP). The network performance is quantitatively evaluated by the fuzzy comprehensive evaluation method, generating a network performance evaluation report that includes index scores, bottleneck location, and risk level. The performance bottleneck identification can accurately locate specific device ports or link segments.

[0045] Performance improvement strategy formulation and implementation steps:

[0046] Based on network performance evaluation reports and fault simulation results, and considering the service priority requirements and operational cost budget of the real network, targeted performance improvement strategies are formulated. These strategies include routing optimization based on the shortest path first algorithm, dynamic bandwidth allocation based on traffic prediction, device load scheduling based on load balancing algorithms, and fault redundancy backup based on dual-machine hot standby / cluster deployment. The formulated performance improvement strategies are converted into standardized configuration commands and input into the network digital twin model. Simulation verification scenarios are built to simulate the execution effects of the strategies under different network loads and service scenarios, evaluating the improvement in network performance indicators and the impact on other services. If the simulation verification shows that the performance indicator improvement meets the preset targets and has no negative impact, the performance improvement strategies are distributed to real network devices through the network management interface to achieve automated deployment. If the simulation verification fails, the parameter combinations of the performance improvement strategies are adjusted using a genetic algorithm based on the verification results, and the simulation verification is repeated until it passes, ensuring the security and effectiveness of the strategies.

[0047] Furthermore, in the construction of the network digital twin model, a collaborative approach of edge computing and cloud computing is adopted for data collection and model building. Edge nodes are responsible for collecting real-time equipment operation data and business traffic data, while the cloud computing platform is responsible for big data processing and model building and optimization, thereby improving the efficiency and accuracy of model building.

[0048] Furthermore, in the fault simulation step, machine learning algorithms are used to analyze the fault simulation results, predict the probability and trend of fault occurrence, and provide support for fault early warning.

[0049] Furthermore, in the process of formulating and implementing performance improvement strategies, a closed-loop feedback mechanism for performance optimization is established. Real-time operational data after optimization of the actual network is collected and fed back to the network digital twin model to dynamically update and calibrate the model, thereby continuously optimizing the performance improvement strategy.

[0050] The present invention also provides a fault simulation and network performance improvement device based on a digital twin model, comprising:

[0051] Data acquisition module: Composed of edge acquisition nodes, core acquisition nodes and data preprocessing sub-module, it is used to collect physical entity data of the real network through various protocols such as TCP / IP and SNMP, including network topology data, device parameter data, link status data and service traffic data. The acquisition frequency can be dynamically adjusted according to business needs (range 1-60 seconds / time).

[0052] Model building module: Connected to the data acquisition module via high-speed Ethernet, it has a built-in data cleaning engine, simulation modeling engine and model calibration sub-module, which are used to perform full-process preprocessing of the acquired data and build a network digital twin model based on the preprocessed data fusion multi-physics coupling technology. It also has a real-time model accuracy calibration function.

[0053] Fault simulation module: Connected to the model building module via API interface, it includes a fault scenario library management submodule, a fault parameter configuration submodule, and a simulation operation submodule. It is used to build and maintain the fault scenario library, receive fault parameters input by users and standardize them for injection into the model, start the simulation process, and collect and analyze fault simulation results.

[0054] Performance evaluation module: It is connected to the model building module and the fault simulation module via a data bus. It has a built-in indicator system management sub-module and a quantitative evaluation sub-module. It is used to preset and maintain the network performance evaluation indicator system, receive real-time model running data and fault simulation results, complete the quantitative evaluation of network performance and generate a detailed evaluation report.

[0055] Strategy formulation and implementation module: communicates bidirectionally with the performance evaluation module, and includes a strategy generation submodule, a simulation verification submodule, and a strategy distribution submodule. It is used to generate performance improvement strategies based on evaluation reports and fault simulation results, verify the effectiveness of the strategies through model simulation, and convert the verified strategies into configuration instructions that can be executed by the device and distribute them.

[0056] Data interaction module: Using the MQTT IoT communication protocol, communication links are established with the model building module, fault simulation module, performance evaluation module, and strategy formulation and implementation module to achieve low-latency data transmission and interaction between the modules. At the same time, real-time data interaction and command issuance between the network digital twin model and real network devices are realized through network management protocols (such as NETCONF).

[0057] Furthermore, the data acquisition module includes multiple edge acquisition nodes and a data aggregation unit. The edge acquisition nodes are deployed close to the network devices to collect real-time running data. The data aggregation unit is used to summarize and preliminarily filter the data collected by each edge acquisition node before transmitting it to the model building module.

[0058] Furthermore, the device also includes a model update module, which is connected to the data interaction module and the model building module respectively. The model update module is used to dynamically update and calibrate the network digital twin model based on the real-time operating data and performance optimization results of the real network, thereby improving the accuracy and adaptability of the model. Specific implementation examples:

[0060] The present invention will be further described in detail below with reference to specific embodiments.

[0061] This embodiment provides a method and apparatus for fault simulation and network performance improvement based on a digital twin model, which is applied to network management of large enterprise local area networks.

[0062] I. Construction of Network Digital Twin Model

[0063] 1. Data Acquisition: By deploying edge acquisition nodes at key nodes of the enterprise LAN, network topology data (including the connection relationships of devices such as routers, switches, and servers), device parameter data (such as device model, number of ports, and maximum bandwidth), link status data (such as link bandwidth utilization, transmission latency, and packet loss rate), and service traffic data (such as the traffic volume, transmission direction, and peak periods of each service) are collected. The edge acquisition nodes transmit the collected data to the data aggregation unit in real time. The data aggregation unit performs preliminary filtering of the data, removing obvious noise data (such as abnormally large traffic data) and duplicate data, before transmitting it to the cloud computing platform.

[0064] 2. Data preprocessing: The cloud computing platform further preprocesses the received data, using interpolation to complete missing link status data and principal component analysis to remove redundant data, resulting in standardized preprocessed data.

[0065] 3. Model Construction: Based on the preprocessed data, a digital twin model of the enterprise local area network is constructed using ANSYS simulation software combined with multiphysics coupling technology. During model construction, geometric mapping (restoring device appearance and link layout), physical mapping (replicating device performance parameters and link transmission characteristics), and behavioral mapping (simulating device operating status and business data transmission processes) are achieved between real network devices and links and corresponding elements in the virtual model. Simultaneously, real-time data interaction between edge computing nodes and the cloud computing platform enables synchronous updates between the virtual model and the real network.

[0066] II. Fault Simulation

[0067] 1. Fault Scenario Database Establishment: Based on common fault types in enterprise LANs, establish a fault scenario database, including fault scenarios such as link interruption, switch failure, bandwidth congestion, and excessive latency, and define fault parameters (such as fault occurrence time, scope of impact, and fault severity) for each fault scenario.

[0068] 2. Fault Parameter Injection: Assuming a scenario of core switch failure needs to be simulated, select the "Node Failure - Core Switch" scenario from the fault scenario library, set the fault parameters as "Fault Occurrence Time: 10:00 AM on weekdays (peak business hours), Fault Severity: Complete Downtime", and inject these fault parameters into the constructed network digital twin model.

[0069] 3. Simulation and Result Acquisition: Start the model simulation to simulate the propagation process of the fault in the network after the core switch goes down, such as the interruption of the connection between each access switch and the core network, and the obstruction of communication between various business systems; obtain the fault simulation results through the model output, including the scope of the fault impact (involving 20 access nodes and 5 business systems), the duration of the fault (expected to last 4 hours if no intervention is taken), and the impact on the quality of business services (the order processing system cannot operate normally, and the customer access latency increases by 80%).

[0070] III. Network Performance Evaluation

[0071] 1. Evaluation Metrics Setting: Establish a performance evaluation metric system for the enterprise LAN, including bandwidth utilization (threshold: ≤80%), transmission latency (threshold: ≤50ms), packet loss rate (threshold: ≤1%), and service availability (threshold: ≥99.9%).

[0072] 2. Performance Evaluation: Based on real-time operational data from the network digital twin model (normal operation data before fault simulation and data during fault simulation), a quantitative evaluation of network performance was conducted. The evaluation results showed that before the fault simulation, network bandwidth utilization was 65%, transmission latency was 30ms, packet loss rate was 0.5%, and service availability was 99.95%, all meeting the threshold requirements. During the fault simulation, the bandwidth utilization in the affected area plummeted to 0 (due to connection interruption), transmission latency became unmeasurable, packet loss rate reached 100%, and service availability dropped to 0, failing to meet the threshold requirements. Simultaneously, the core switch was identified as the network performance bottleneck (high risk of single-point failure).

[0073] IV. Performance Improvement Strategy Formulation and Implementation

[0074] 1. Strategy Formulation: Based on the performance evaluation report and fault simulation results, formulate performance improvement strategies:

[0075] ① Route optimization: Enable backup routing and switch access nodes connected to the faulty core switch to the backup core switch;

[0076] ② Redundancy backup: Add redundant equipment to the core switch to achieve dual-machine hot standby and avoid single point of failure;

[0077] ③ Dynamic bandwidth allocation: During peak business periods, bandwidth for non-core businesses (such as video surveillance) is temporarily allocated to core businesses (such as order processing) to improve the transmission efficiency of core businesses;

[0078] 2. Simulation Verification: Input the formulated performance improvement strategy into the network digital twin model for simulation verification, and simulate the network operation state after the strategy is implemented;

[0079] Verification results show that after enabling the backup route, the scope of the fault was reduced to zero, and all access nodes resumed normal connection; after adding redundant equipment, the service availability of the core switch was improved to 99.99%; after dynamic bandwidth allocation, the transmission latency of the order processing system was reduced to 45ms, meeting the threshold requirements, and the strategy was effective and feasible.

[0080] 3. Strategy Implementation: The validated performance improvement strategies are distributed to real network devices in the enterprise LAN, such as configuring backup routing parameters, deploying redundant core switches, and setting dynamic bandwidth allocation rules. At the same time, the data interaction module collects real-time operational data of the optimized real network and feeds it back to the model update module to dynamically update and calibrate the network digital twin model.

[0081] Through the method and apparatus of this embodiment, the fault simulation accuracy of enterprise local area networks reaches over 95%, the network fault handling time is shortened by 60%, the network service availability is increased to 99.99%, and the bandwidth utilization is optimized to around 70%, effectively improving the stability and efficiency of network operation.

[0082] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A method for fault simulation and network performance improvement based on a digital twin model, characterized in that, The method comprises the following steps: Network digital twin model construction step: Collect physical entity data of the real network, preprocess the collected data, and construct a network digital twin model based on the preprocessed data using multi-physics field coupling simulation technology to realize geometric mapping, physical mapping, and behavior mapping between the real network and the virtual model; Fault simulation step: Based on the constructed network digital twin model, a fault scenario library is established, a target fault type is selected and fault parameters are set, the fault parameters are injected into the model and simulation is started to simulate the fault occurrence and propagation process, and fault simulation results are obtained; Network performance evaluation step: Set up a network performance evaluation index system, based on the real-time running data of the network digital twin model and the fault simulation results, quantitatively evaluate the network performance, generate a network performance evaluation report and identify the network performance bottleneck; Performance improvement strategy formulation and implementation step: According to the network performance evaluation report and the fault simulation results, formulate a performance improvement strategy, input the strategy into the model for simulation verification, and if the verification is passed, issue it to the real network equipment to realize performance optimization and establish a closed-loop feedback mechanism to continuously optimize the strategy.

2. The method of claim 1, wherein the method further comprises: The physical entity data includes network topology structure data, device parameter data, link state data, and traffic data; preprocessing includes removing noise data, redundant data, and completing missing data.

3. The method of claim 1, wherein: In the network digital twin model construction step, an edge computing and cloud computing collaborative method is used, the edge node collects data with high real-time requirements, and the cloud computing platform performs big data processing and model construction and optimization.

4. The method of claim 1, wherein: In the fault simulation step, a machine learning algorithm is used to analyze the fault simulation results to predict the probability and trend of fault occurrence and provide support for fault early warning.

5. The method of claim 1, wherein: The network performance evaluation index system includes bandwidth utilization, transmission delay, packet loss rate, and service availability; the performance improvement strategy includes route optimization, bandwidth dynamic allocation, device load balancing, and fault redundancy backup.

6. The device for fault simulation and network performance improvement based on digital twin model, characterized in that, It comprises: Data collection module: used for collecting physical entity data of the real network; Model construction module: connected with the data collection module, used for preprocessing the collected data and constructing a network digital twin model; Fault simulation module: connected with the model construction module, used for establishing a fault scenario library, injecting fault parameters for fault simulation and obtaining fault simulation results; Performance evaluation module: connected with the model construction module and the fault simulation module respectively, used for setting up an evaluation index system, quantitatively evaluating network performance and generating an evaluation report; Strategy formulation and implementation module: connected with the performance evaluation module, used for formulating a performance improvement strategy, performing simulation verification and issuing the verified strategy; Data interaction module: connected with each module respectively, realizing real-time data interaction between modules and between the model and real network equipment.

7. The digital twin model based fault simulation and network performance improvement apparatus according to claim 6, characterized in that: The data collection module comprises multiple edge collection nodes and a data aggregation unit, the edge collection nodes are deployed near the network equipment, and the data aggregation unit is used to aggregate and preliminarily filter data before transmitting it to the model construction module.

8. The digital twin model based fault simulation and network performance improvement apparatus according to claim 6, characterized in that: The model updating module is connected with the data interaction module and the model construction module respectively, and is used for dynamically updating and calibrating the network digital twin model according to the real network data and the optimization result.

Images