Construction method and device of digital twin system

Abstract

The invention provides a construction method of a digital twin system, the method is applied to at least one agent, and the method comprises the following steps: obtaining mapping information associated with a physical system; analyzing the mapping information, and generating a resource configuration strategy meeting twin requirements; and generating the digital twin system in response to the resource configuration strategy. The mapping information of the physical system is analyzed through the intelligent agent, the resource configuration strategy adaptive to the twinborn demand is automatically generated, and the consistency of the digital twinborn system and the physical system is further ensured. The invention further provides electronic equipment, a computer program product and a construction device of the digital twin system.

Description

Technical Field

[0001] The present disclosure relates to the field of communication and information technology, and particularly relates to a method and apparatus for constructing a digital twin system. Background Art

[0002] A digital twin maps a physical object in real time through a virtual model. A digital twin network includes at least a physical system and a digital twin system. In order to better implement the virtualization mapping of the twin system, the digital twin system needs to be consistent with the physical system. However, currently, it is difficult to ensure such consistency. Summary of the Invention

[0003] The present disclosure provides a method and apparatus for constructing a digital twin system.

[0004] In a first aspect, an embodiment of the present disclosure provides a method for constructing a digital twin system, which is applied to at least one agent and includes: obtaining mapping information associated with a physical system; parsing the mapping information to generate a resource configuration policy that meets the twin requirements; and generating a digital twin system in response to the resource configuration policy.

[0005] In a second aspect, an embodiment of the present disclosure provides an electronic device, which includes a memory and a processor; the memory stores a computer program that can be executed by the processor, and when the computer program is executed by the processor, the construction method described in any one of the embodiments of the present disclosure is implemented.

[0006] In a third aspect, an embodiment of the present disclosure provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the construction method described in any one of the embodiments of the present disclosure is implemented.

[0007] In a fourth aspect, an embodiment of the present disclosure provides a device for constructing a digital twin system, which includes at least one agent, and the at least one agent is configured to: obtain mapping information associated with a physical system in real time; parse the mapping information to generate a resource configuration policy that meets the twin requirements; and generate a digital twin system in response to the resource configuration policy.

[0008] In the embodiments of the present disclosure, by parsing the mapping information of the physical system through an agent, a resource configuration policy adapted to the twin requirements is automatically generated, further ensuring the consistency between the digital twin system and the physical system. Brief Description of the Drawings

[0009] In the drawings of the embodiments of the present disclosure:

[0010] Figure 1 It is a schematic architecture diagram of a twin network to which the method for constructing a digital twin system provided by the embodiments of the present disclosure is applied;

[0011] Figure 2 Schematic diagram of another architecture of the twin network for the construction method of the digital twin system provided by the embodiments of the present disclosure;

[0012] Figure 3 Flowchart of the construction method of the digital twin system provided by the embodiments of the present disclosure;

[0013] Figure 4 In the construction method of the digital twin system provided by the embodiments of the present disclosure, flowchart of steps based on the multimodal perception algorithm;

[0014] Figure 5 Schematic diagram of the framework structure of the multimodal perception algorithm in another construction method of the digital twin system provided by the embodiments of the present disclosure;

[0015] Figure 6 Flowchart of the construction method of the digital twin system provided by another embodiment of the present disclosure;

[0016] Figure 7 In another construction method of the digital twin system provided by the embodiments of the present disclosure, schematic diagram of the step process of dual-mode verification;

[0017] Figure 8 Schematic diagram of the working principle of the construction device of the digital twin system provided by the embodiments of the present disclosure;

[0018] Figure 9 Flowchart of the construction method of the digital twin system provided by another embodiment of the present disclosure;

[0019] Figure 10 In the construction method of the digital twin system provided by the embodiments of the present disclosure, multimodal schematic diagram including the network domain, network element domain, and service domain;

[0020] Figure 11 Flowchart of the construction method of the digital twin system provided by another embodiment of the present disclosure;

[0021] Figure 12 Schematic block diagram of the composition of an electronic device provided by the embodiments of the present disclosure. Detailed implementation manners

[0022] To enable those skilled in the art to better understand the technical solutions of the present disclosure, the construction method of the digital twin system provided by the embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0023] The present disclosure will be described more fully hereinafter with reference to the accompanying drawings. However, the illustrated embodiments may be embodied in different forms, and the present disclosure should not be construed as limited to the embodiments set forth below. On the contrary, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0024] The accompanying drawings of the embodiments of the present disclosure are used to provide a further understanding of the embodiments of the present disclosure, and constitute a part of the specification. They are used to explain the present disclosure together with the detailed embodiments, and do not constitute a limitation to the present disclosure. By describing the detailed embodiments with reference to the accompanying drawings, the above and other features and advantages will become more apparent to those skilled in the art.

[0025] The present disclosure may be described with reference to plan views and / or cross-sectional views by means of ideal schematic diagrams of the present disclosure. Therefore, the example illustrations may be modified according to manufacturing techniques and / or tolerances.

[0026] In the case of no conflict, the embodiments of the present disclosure and the features in the embodiments may be combined with each other.

[0027] The terms used in the present disclosure are only for describing specific embodiments and are not intended to limit the present disclosure. As used in the present disclosure, the term "and / or" includes any and all combinations of one or more of the related listed items. As used in the present disclosure, the singular forms "a" and "the" are also intended to include the plural forms unless the context clearly indicates otherwise. As used in the present disclosure, the term "comprises" specifies the presence of the stated features, wholes, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their groups.

[0028] Unless otherwise defined, all terms (including technical and scientific terms) used in the present disclosure have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and the present disclosure, and will not be interpreted as having an idealized or overly formal meaning unless the present disclosure clearly so defines.

[0029] The present disclosure is not limited to the embodiments shown in the accompanying drawings, but includes modifications to the configurations formed based on the manufacturing process. Therefore, the regions illustrated in the accompanying drawings have exemplary properties, and the shapes of the regions shown in the figures illustrate the specific shapes of the regions of the elements, but are not intended to be restrictive.

[0030] Technical Term Explanation

[0031] 1. Digital Twin

[0032] A comprehensive digital technology method that realizes dynamic simulation, status monitoring, performance prediction, and decision optimization throughout the entire life cycle process by establishing a virtual mapping model of a physical entity and driving it based on real-time data. It uses multi-source heterogeneous sensors to collect the operating data of the physical entity, transmits it to the virtual space through a communication network, and constructs a virtual model relying on physical modeling, data fusion, and machine learning algorithms. This virtual model maintains a synchronous mapping relationship with the physical entity, has real-time interactivity, closed-loop feedback, and iterative evolution, and thus realizes the perception of the operating state of the physical entity, the diagnosis and early warning of abnormal behaviors, and the deduction and prediction of future trends, providing a digital support platform for the design verification, operation optimization, and maintenance decision-making of the physical entity.

[0033] 2. Digital Twin

[0034] Among them, the physical entity is a device or system in the real world, and the digital twin is the virtual mapping of the physical entity.

[0035] A digital twin is a high-precision, real-time dynamic mapping of a physical entity (such as a device, system, factory, etc.) in the virtual space. It realizes real-time status monitoring, simulation analysis, and prediction optimization of physical objects through sensor data, artificial intelligence, modeling technology, and orchestration technology. A digital twin is a virtual model for a single entity (such as a machine, a building).

[0036] The application platform is the decision-making center, used for various application scenarios such as product design optimization, equipment health management, and fault prediction.

[0037] 3. Digital Twin Network (DTN, Digital Twin Network)

[0038] A digital twin network is a system composed of multiple interconnected digital twins, capable of overall modeling and collaborative management of complex physical systems (such as smart cities, industrial Internet, transportation networks, etc.). It not only includes the twins of individual entities but also covers the association relationships and networked behaviors between entities.

[0039] The digital twin is the basic unit. A single digital twin is the cornerstone for constructing a digital twin network, while the digital twin network is an extension. By interconnecting multiple digital twins, a higher-dimensional system-level mapping is formed. The twin can solve local problems such as equipment fault diagnosis, while the digital twin network can also solve global problems (such as optimizing urban traffic congestion and balancing energy networks).

[0040] To better achieve the virtualization mapping of the twin system, the digital twin system needs to be consistent with the physical system. However, currently, it is very difficult to ensure this consistency.

[0041] Based on this, on the one hand, the embodiments of the present disclosure propose a method for constructing a digital twin system.

[0042] This method is applied to a digital twin network architecture as shown in Figure 1 . The architecture includes a physical system, a digital twin system, and a network application platform, forming a closed-loop interaction through real-time data streams and control instructions. Among them:

[0043] (1) Physical system

[0044] The physical system is a collection of entity objects, devices, processes, or complex systems mapped and simulated by the digital twin system. As the objective entity to be mapped, it covers all hardware devices, connection topologies, and sensing and execution units. Its core functions include data acquisition and device control.

[0045] Among them, data acquisition means to capture multi-dimensional indicators such as device operation parameters, environmental status, and network traffic in real time through embedded sensors; device control means to receive the optimization decision instructions of the digital twin system and execute operations such as parameter adjustment and resource scheduling for the physical system to form a closed-loop control.

[0046] (2) Digital twin system

[0047] It includes a data storage module, a twin object simulation module, and a digital twin network ontology module, which respectively implement the following functions:

[0048] i) Data sharing warehouse

[0049] The data sharing warehouse is the central database of the DTN and the hub for data interaction between the physical network and digital twin nodes, implementing three functions: data collection and storage, data service and interface, and data management.

[0050] When executing the data collection and storage function, the data sharing warehouse real-time collects the operation status data (such as device performance, traffic load) and configuration data (such as network topology, parameter settings) in the physical network. The real-time acquisition and efficient storage of these data are crucial for the stable operation of the system, ensuring the accuracy and timeliness of the data source.

[0051] In terms of data service and interface, the data sharing warehouse provides a unified data interface, supporting other subsystems or external applications to call data through methods such as API. This design enhances the interoperability of the system, promotes the seamless flow of data between different platforms, and thus improves the data utilization rate and system compatibility.

[0052] Data management is another key function of the data sharing warehouse, including operations such as data cleaning, classification, and archiving to ensure the quality and security of data. Through strict data management and maintenance, the system guarantees the accuracy and reliability of data, providing strong data support for decision-makers.

[0053] ii) Twin mapping and management module

[0054] The twin mapping and management module is the core of the digital twin system. Based on the basic models (such as network element models, topology models) and functional models (covering all life cycle links of network planning, traffic modeling, scheduling optimization, fault diagnosis, security modeling, and quality assurance), it constructs or optimizes the digital twin network ontology, and through core components such as model management (version control, composition and orchestration), security management (access control, data encryption), etc., ensures the stable operation and collaborative control of the twin system;

[0055] Based on the data sharing warehouse and the twin mapping and management module, generate / optimize the digital twin network ontology module for storing the digital twin network ontology.

[0056] Among them, the digital twin system construction device of the present disclosure is set in the twin mapping and management module, and the digital twin system construction device includes at least one intelligent agent.

[0057] (3) Network application platform

[0058] The digital twin network DTN also includes a network application platform, which is the decision-making brain and user interface of the digital twin network DTN. Its core tasks are:

[0059] Requirement input: Receive business requirements from users or systems (such as "optimize network bandwidth utilization").

[0060] Model call: Generate solutions through the data and models of the twin network layer.

[0061] Business deployment: Map the solution to the physical network and drive the operation of the actual network (such as automatically adjusting the routing strategy).

[0062] Result feedback: Monitor the execution effect and continuously optimize through the twin layer.

[0063] For example, through the digital twin application platform, controls such as network innovation technology verification, network visualization, intent verification, and network maintenance and optimization (predictive maintenance, dynamic resource allocation) can be achieved.

[0064] In the embodiments of the present disclosure, the construction method of the digital twin system can be applied to at least one intelligent agent, that is, in Figure 1 the shown twin mapping and management module, the construction and / or optimization of the digital twin system is realized through one or more intelligent agents. Refer to Figure 2 , which shows a schematic diagram of a digital twin network including multiple intelligent agents.

[0065] Refer to Figure 3 , the embodiments of the construction method of the digital twin system of the present disclosure may include the following steps:

[0066] Step S310, obtain mapping information associated with the physical system;

[0067] Step S320, parse the mapping information to generate a resource configuration policy that meets the twin requirements;

[0068] Step S330, generate a digital twin system in response to the resource configuration policy.

[0069] In the embodiments of the present disclosure, an agent parses the mapping information of the physical system to automatically generate a resource configuration policy adapted to the twin requirements, further ensuring the consistency between the digital twin system and the physical system; the generated digital twin system provides a high-fidelity simulation environment for major operations (such as upgrades) and fault repairs of network elements, enabling users to identify potential risks (such as resource conflicts and configuration errors) in advance, so as to formulate solutions before actual operations and reduce operation and maintenance risks.

[0070] Step S301, obtain mapping information associated with the physical system.

[0071] In a preferred embodiment, mapping information associated with the physical system is obtained in real time.

[0072] In a preferred embodiment, this step can be executed by an analysis agent.

[0073] The physical system is a collection of entity objects, devices, processes, or complex systems mapped and simulated by the digital twin system.

[0074] Exemplarily, in the embodiments of the present disclosure, the physical system therein may be a physical system based on the 5G mobile communication core network and access network system. For example, the physical system may include physically deployed base stations (gNBs), core network elements (such as AMF, SMF, UPF), transmission devices (routers, switches), user terminals (mobile phones, Internet of Things devices), and physical optical fibers / wireless links connecting them. When the physical system operates, it involves multiple real-time interactive processes such as user access, session management, data forwarding, mobility management, policy control, and network slicing.

[0075] Exemplarily, in the embodiments of the present disclosure, the physical system therein may also be an industrial Internet of Things (IIoT) edge computing gateway cluster system. For example, in this system, a group of physical edge gateway devices are deployed in the factory workshop or a specific area. The gateways therein are responsible for connecting and managing a large number of on-site sensors (temperature, pressure, vibration, images), actuators (robotic arms, valves), and PLC controllers, performing data collection, local preprocessing (filtering, aggregation), protocol conversion, edge computing tasks (such as real-time quality detection, equipment predictive maintenance analysis), and communication with the cloud or upper-layer systems.

[0076] Exemplarily, in an embodiment of the present disclosure, the physical system therein may also be a physical system based on a smart grid regional substation and a distribution network system. For example, in this system, a power supply network composed of substations (transformers, circuit breakers, disconnectors, protective relays), power transmission and distribution lines, smart meters, distributed energy sources (such as rooftop photovoltaics), energy storage devices, etc. is included. And when the system operates, it involves the conversion, transmission, distribution, metering of high-voltage / low-voltage electric energy, as well as fault detection and isolation.

[0077] It can be understood that the specific form, application field, etc. of the physical system in the embodiments of the present disclosure are not limited.

[0078] In an embodiment of the present disclosure, the mapping information provides data basis for the construction of the digital twin system, and it is the key data and knowledge for creating, describing, and associating physical system entities, states, behaviors, functions, and operating environments in the digital space. The digital twin model needs to reflect which aspects of the physical system and how to associate the two.

[0079] Preferably, the mapping information may be multimodal information, including traditional user instructions (such as user instructions input in natural language, etc.), and can also integrate multi-source data from IoT sensors, monitoring alarms, historical operation and maintenance data records, etc. to achieve deep environmental perception, timely capture early warning information such as performance degradation and fault alarms, as well as network element operation information planned by users.

[0080] Step S302, parse the mapping information to generate a resource configuration strategy that meets the twin requirements.

[0081] In a preferred embodiment, this step may be executed by an analysis agent.

[0082] Exemplarily, in a preferred embodiment, to generate a resource configuration strategy based on the mapping information, a multimodal perception algorithm may be used, specifically including Figure 4 The following steps are shown:

[0083] Step S3021, after obtaining user instructions input in natural language, IoT sensor data, and historical operation and maintenance data, perform keyword matching on the mapping information based on a preset rule library to obtain key features;

[0084] Step S3022, encode and map the key features to a high-dimensional feature space for fusion, capture temporal associations based on a cross-attention mechanism, and generate a semantic representation vector;

[0085] Step S3023: Based on the semantic representation vector, perform semantic reasoning to generate a resource configuration policy, which is embodied as a resource loading list. The resource configuration policy embodied as a resource loading list is for realizing specific twin requirements. For example, for high-precision simulation, dynamic optimization, service chain orchestration, etc., for the virtualization components constituting the twin system, including but not limited to the planning or decision-making on the structural division, function encapsulation, interface definition, and deployment method for network element function modules and their encapsulated microservices.

[0086] Further, in the embodiments of the present disclosure, exemplarily, the framework of the multimodal perception algorithm refers to Figure 5 as shown.

[0087] In Figure 5 the multimodal perception algorithm framework structure shown:

[0088] (1) The input data includes the following three categories, namely:

[0089] Natural language data: user instructions (such as natural language commands).

[0090] IoT sensor data: Internet of Things sensing data such as real-time network traffic and device status.

[0091] Operation and maintenance data: structured data such as historical logs, monitoring alarms, and configuration records.

[0092] (2) The processing process

[0093] The input data goes through three stages to achieve multimodal fusion through rule filtering → deep context parsing → semantic reasoning, which are respectively:

[0094] Stage 1: Data preprocessing and rule filtering

[0095] Preprocess the three categories of input data respectively (such as data cleaning and standardization).

[0096] Stage 2: Feature encoding and fusion

[0097] The entire stage requires the assistance of three encoders, a text encoder, a time series encoder, and a structured data encoder. Among them, the text encoder is used to process natural language to generate text feature vectors; the time series encoder is used to process IoT sensor data to extract time series features. The structured data encoder is used to process operation and maintenance data to generate structured features.

[0098] After obtaining the text feature vector, temporal feature, and structured feature, the three types of features are also mapped to a high-dimensional feature space for feature fusion. Further, through a multi-layer Transformer structure combined with a cross-attention mechanism, cross-modal feature interaction is achieved (such as the association between text semantics and real-time network status).

[0099] Phase 3: Semantic Reasoning and Decision Making

[0100] Perform multi-task learning (text parsing, alarm analysis, fault prediction) based on the fused features; decompose the user's intention into specific operation requirements (such as "adding QoS module resources to the UPF network element").

[0101] (3) Output Data

[0102] Output a resource loading list. The resource loading list includes the list of activated services (such as "QoS policy module"), corresponding network elements (such as "UPF network element"), and operation suggestions (such as "resource expansion"), and then pass the resource loading list to the orchestration agent for the orchestration agent to implement atomic-level service hot plugging and cluster-level reorganization.

[0103] It can be seen that through the multi-modal horizontal fusion mechanism, this embodiment breaks through the limitations of traditional single data sources, synchronously integrates natural language instructions, real-time IoT sensor data, and historical operation and maintenance logs, and realizes the dynamic perception of the network environment. For example, when the user instruction "optimize video service latency" is combined with real-time high-traffic alarms, the system accurately parses it into a network element-level operation of "adding QoS resources to the UPF network element", significantly improving the response accuracy.

[0104] Moreover, its three-stage parsing engine further strengthens the parsing. Among them, the rule filtering layer quickly filters out noise instructions based on lightweight regular expressions to improve processing efficiency; the deep context parsing layer introduces the Transformer cross-attention mechanism to associate text sentiment, context, and real-time data (such as the instruction of "urgent fault" automatically triggers a high-priority response);

[0105] The semantic reasoning layer can preferably rely on the pre-trained large model in the telecommunications field to refine fuzzy requirements into specific operations (such as locking the dependency relationship of "the QoS module needs to be bound to the UPF network element") to ensure the executability of the decision.

[0106] Therefore, through the above multi-modal perception algorithm, multi-source data can be fused in real time to quickly respond to environmental changes such as sudden traffic; moreover, the accuracy of intention parsing: transform abstract instructions into atomic-level network element operations, reducing manual intervention.

[0107] Step S303, in response to the resource configuration policy, construct a digital twin system.

[0108] In one embodiment of the present disclosure, the resource configuration policy obtained according to the foregoing step S302 is executed by the orchestration agent, including the following steps:

[0109] a. In response to the resource configuration policy, identify network element function modules.

[0110] The core of this step is to start the deployment process according to the specific plan for the identification and division of network element function modules in the resource configuration policy. According to the resource configuration policy, determine the target physical network elements to be incorporated into the digital twin system (for example, the SMF session management function and the UPF user plane forwarding function in the 5G core network, or the relay protection logic module in the smart grid), and according to the rule settings in the resource configuration policy, decompose the functions of the network elements into appropriately granular, independently manageable and orchestratable network element function modules.

[0111] b. Split each network element function module into independent microservices and provide application programming interfaces for each microservice.

[0112] This step executes the decisions regarding microservice transformation and interface standardization in the resource configuration policy. For each network element function module identified and orchestrated in step S1031, according to the decomposition principles specified in the policy (such as single responsibility, high cohesion and low coupling), implement it as an independent and self - contained microservice. Each microservice encapsulates the core logic of the function module, has an independent code library, running process and data management (or access interface).

[0113] At the same time, strictly follow the interface specifications defined in the policy to design and implement standardized application programming interfaces for each newly created microservice.

[0114] c. Package the microservices into microservice containers.

[0115] This step executes the requirements regarding containerized packaging in the resource configuration policy. Package and encapsulate each independent microservice developed in step S1032 using the selected container technology (such as Docker). The encapsulation process is carried out according to the container construction specifications specified in the policy, including selecting a suitable base operating system image, installing necessary runtime dependency libraries, copying the microservice code and its configuration files into the container, setting environment variables, defining the container startup command, etc.

[0116] d. Orchestrate and deploy the microservice containers through a container orchestration tool to build a digital twin system.

[0117] This step, according to the detailed container orchestration and deployment strategy in the resource configuration policy, uses a container orchestration tool (such as Kubernetes) to schedule, deploy, combine and manage multiple encapsulated microservice container instances, and finally form or optimize the digital twin system.

[0118] Therefore, after each network element function module (such as the QoS policy module of 5G UPF) is independently encapsulated into a microservice container and provides a standardized API interface, containerized deployment is implemented. It uses Docker to encapsulate services into lightweight containers and realizes the automated deployment, domain scaling, and expansion of services through Kubernetes cluster orchestration. Then, based on Kubernetes rolling updates, readiness probes, and liveness probes, hot replacement during the operation of the module is achieved, ensuring that the service operation is not affected.

[0119] In an embodiment of the present disclosure, a monitoring agent can also be set up. By collecting service performance metrics in real time and combining with a big data prediction model, the future load trend and bottlenecks are predicted, so that the orchestration agent can dynamically adjust the service deployment strategy according to the prediction results and achieve preventive resource scheduling.

[0120] In the above embodiment of the present disclosure, the above three steps S301, S302, and S303 are executed by the construction system of the digital twin system. This system includes two agents, namely: an analysis agent and an orchestration agent. Among them, the analysis agent executes real-time acquisition of mapping information associated with the physical system and parses the mapping information to generate a resource configuration policy that meets the twin requirements; the orchestration agent executes in response to the resource configuration policy to generate a digital twin system. However, the embodiments of the present disclosure are not limited thereto. These three steps can be executed by only one agent, or multiple agents can share the operations of the analysis agent, or multiple agents jointly undertake the work of the orchestration agent. The embodiments of the present disclosure do not make a limitation on this.

[0121] The digital twin system can achieve risk prediction and decision-making optimization by accurately simulating the state of the physical system. For this purpose, according to the twin requirements, the twin system can be configured to achieve:

[0122] Network simulation rehearsal: Before performing major operations (such as network element upgrade, topology adjustment) on the real network, the operation process is simulated through the digital twin system to predict potential failures (such as resource conflicts, configuration errors).

[0123] Fault repair deduction: Encapsulate historical fault scenarios (such as service crashes caused by traffic overload), reproduce and verify the effectiveness of the repair plan in the twin environment, and avoid the risk of trial and error in the real environment.

[0124] Service change verification: When a service chain is added or adjusted (such as deploying a 5G QoS module), verify the compatibility between the service and network element resources to ensure the stability of the system after the change.

[0125] It can be understood that

[0126] It can be understood that through the above steps S301, S302, and S303, the embodiments of the present disclosure realize the virtual mapping of the twin system based on the automated construction of the digital twin system, better achieving the consistency between the digital twin system and the physical system and narrowing the gap between the digital twin system and the physical system.

[0127] Referring to Figure 6 , another embodiment of the method for constructing the digital twin system of the present disclosure is shown.

[0128] It may include the following steps:

[0129] Step S610, obtaining mapping information associated with the physical system;

[0130] Step S620, parsing the mapping information to generate a resource configuration policy that meets the twin requirements;

[0131] Step S630, generating a digital twin system in response to the resource configuration policy;

[0132] Step S640, performing consistency verification on the digital twin system and the physical system and confirming the verification response deviation;

[0133] Step S650, generating a resource configuration optimization policy based on the verification response deviation, where the resource configuration optimization policy is used to narrow the deviation between the digital twin system and the physical system.

[0134] In the embodiments of the present disclosure, first, the mapping information of the physical system is obtained; then, this information is parsed and a resource configuration policy adapted to the twin requirements is generated; subsequently, a digital twin system is constructed based on the policy; then, consistency verification is performed to compare the running states of the twin system and the physical system in real time (such as traffic processing results, resource occupancy rate), and the response deviation is quantified (such as latency difference, configuration error); finally, a resource configuration optimization policy (such as dynamically adjusting service deployment, resource reallocation) is automatically generated based on the deviation to drive the dynamic correction of the twin system.

[0135] The embodiments of the present disclosure ensure that the twin system continuously and accurately maps the dynamic changes of the physical system (such as network topology updates, device state fluctuations) through the closed-loop optimization driven by "verification - deviation confirmation - optimization policy generation", and solve the problem of inaccurate static models. Obviously, the embodiments of the present disclosure provide a real-time calibrated simulation environment for major operations of network elements (such as upgrade and expansion), expose resource conflicts or performance bottlenecks in advance, and further reduce the operation risk in the real environment.

[0136] It can be seen that after generating the digital twin system, the embodiments of the present disclosure further include operations of consistency verification and optimizing the digital twin system.

[0137] In specific implementation, the construction method can be throughFigure 8 implemented by the construction system shown, which includes an analysis agent (i.e., Figure 8 the analysis Agent in Figure 8 ), an orchestration agent (i.e., Figure 8 the orchestration Agent in Figure 8 ), a verification agent (i.e.,

[0138] ), and an optimization agent (i.e.,

[0139] Step S610: The analysis agent obtains mapping information associated with the physical system;

[0140] Step S620: The analysis agent parses the mapping information and generates a resource configuration policy that meets the twin requirements;

[0141] Step S630: The orchestration agent generates a digital twin system in response to the resource configuration policy;

[0142] Step S640: The verification agent performs consistency verification on the digital twin system and the physical system and confirms the verification response deviation;

[0143] Step S650: The optimization agent generates a resource configuration optimization policy based on the verification response deviation, and the resource configuration optimization policy is used to narrow the deviation between the digital twin system and the physical system.

[0144] It can be seen that in the embodiment of the present disclosure, the construction link, verification link, and optimization link of the digital twin system are included. The construction link is implemented by the analysis agent and the orchestration agent according to steps S610 - S630, the verification link is implemented by the verification agent according to step S640, and the optimization link is implemented by the optimization agent according to step S650.

[0145] The operations of steps S610 - S630 are similar to the above steps S310 - S330, and the relevant parts can be referred to each other, so they will not be elaborated here. The following will describe steps S640 and S660.

[0146] Verification link:

[0147] Step S640: After the orchestration agent generates a digital twin system according to the resource configuration policy, the verification agent performs consistency verification on the digital twin system and the physical system and confirms the verification response deviation.

[0148] In the embodiment of the present disclosure, it is executed in the following manner:

[0149] The verification agent injects the operation state data of the physical system into the digital twin system;

[0150] The verification agent respectively obtains the responses of the physical system and the digital twin system based on the operation status data, and confirms the verification response deviation.

[0151] More specifically, the verification agent can analyze the digital twin system and the physical system in the following ways:

[0152] a. Inject the operation status data of the physical system into the digital twin system;

[0153] b. Respectively obtain the responses of the physical system and the digital twin system based on the operation status data, and confirm the verification response deviation.

[0154] In an embodiment of the present disclosure, when the operation status data of the physical system is the traffic of the current real-time operation data, for performing mirror verification; exemplarily, the mirror verification includes the following steps:

[0155] First, inject the traffic of the current real-time operation data of the physical system into the digital twin system according to a preset ratio;

[0156] Then, respectively obtain the actual response of the physical system and the mirror response of the digital twin system, and confirm the first verification response deviation based on the mirror response and the actual response.

[0157] Exemplarily, in an embodiment of the present disclosure, the verification agent can perform mirror verification in the following ways:

[0158] (1) Environmental traffic replication: Use a network monitoring tool to replicate the on-site environmental traffic (1% sampling rate) and inject it into the twin system.

[0159] (2) Compare key indicators: Set the key indicators for mirror verification (network domain, network element domain, service domain state migration, network request latency, node resource occupancy), and compare the real traffic with the processing results of the twin system in real time. It is required that the state migrations of each domain are the same, the latency deviation is less than 50 microseconds, and the node resource occupancy deviation is less than 10%.

[0160] In another embodiment of the present disclosure, the physical system can be historical fault operation data; at this time, the verification agent performs playback verification. Exemplarily, it can include the following steps:

[0161] First, inject the fault cause unit for encapsulating the historical fault operation data into the digital twin system; then, determine the second verification response deviation based on the parallel time axis playback including the fault handling result and the delay time; wherein, the executable gene unit can include at least one of a fault cause, environmental resource configuration, network element node dependency, fault manifestation, and fault recovery strategy.

[0162] Exemplarily, in one embodiment of the present disclosure, the verification agent can perform replay verification in the following manner:

[0163] (1) Fault scenario encapsulation: Encapsulate historical faults into executable network gene units. Specifically, encapsulate the fault cause (the direct cause and root cause of the fault, for example, the direct cause is that the request quantity exceeds 1,000,000 / s, and the root cause is the lack of a current limiting protection service for the interface), environmental resources (the on-site resource configuration when the fault occurs), dependent nodes (the dependencies of the current network element and service when the fault occurs), fault manifestations (the basis for the verification agent to check whether a fault occurs), and solutions (the methods for fault recovery) into a configuration file, so that the verification agent can modify the current system configuration according to the configuration file, trigger the generation of faults, and automatically solve the faults.

[0164] (2) Parallel timeline replay: Build a parallel timeline in the twin system (backup the current data and configuration of the twin system, and adjust the time of the twin system to the time when the fault occurred at the current time). Replay the network gene units on the parallel timeline, verify the deviation of the fault response time and the response result, and after the verification is completed, restore the backup configuration and data.

[0165] In one embodiment of the present disclosure, the verification agent performs verification operations through the following verification process:

[0166] Sandbox isolation: The verification agent calls the environment deployment interface to create an isolated sandbox. This sandbox realizes memory isolation and filters system calls, and implants eBPF probes (Extended Berkeley Packet Filter, used for new system tracking, security monitoring, etc. When the Linux kernel monitors the arrival of network packets and system calls occur, the eBPF program will be automatically called) inside the twin system for monitoring the behavior of the twin system.

[0167] Comprehensive testing: The verification agent calls the test deployment interface to deploy test data sets (on-site data traffic + historical fault network gene units), the Logstash log collection tool, the NTP clock synchronization network, the log comparison engine, the fault injection engine, the fault detection engine, and the system status detection tool.

[0168] Dual-mode verification: The verification agent calls the test interface to perform verification tests. For the mirror verification method, it calls the log comparison engine and the system monitoring engine to monitor the differences between the physical environment and the twin system in real time; for the replay verification, it calls the network clock engine and the fault injection engine to dynamically inject historical faults, and then calls the fault detection engine to check the response result and latency of the twin system. Refer to Figure 7 , Figure 7Schematically shows a flowchart of the embodiment of the present disclosure based on dual-mode verification.

[0169] Optimization phase:

[0170] Step S650: The optimization agent generates a resource configuration optimization strategy based on the verification response deviation, and the resource configuration optimization strategy is used to achieve the consistency between the physical system and the digital twin system.

[0171] In an embodiment of the present disclosure, when the verification phase detects a deviation or a fault, the optimization agent analyzes the cause of the abnormality based on the verification result and historical data, and generates a globally optimal adjustment strategy through distributed negotiation of agents (such as a voting mechanism), automatically issues an optimization instruction, and implements dynamic resource adjustment and fault rollback.

[0172] Exemplarily, after constructing and verifying the twin system, for the problems found in the verification phase, combined with the resource status of the twin system, a negotiation mechanism is adopted to generate an optimization decision, and the optimization agent executes the optimization action through API calls, which is mainly divided into the following three steps:

[0173] Deployment: Deploy the optimization agent on different domains (network domain, network element domain, service domain), and each domain continuously collects key performance data (traffic metrics, resource utilization rate) to determine whether there is an abnormal state or a performance bottleneck. Set abnormal trigger conditions (performance degradation, resource occupation), and detect whether optimization is required according to the preset threshold.

[0174] Analysis: When the optimization agent receives an abnormal trigger signal, it quickly combines historical operation data and current real-time data to analyze the cause of the abnormality, determine whether it is a temporary fluctuation or a long-term impact, and then decide whether optimization is required.

[0175] Optimization: When optimization is required, the optimization agents share optimization decision information with each other, reach an agreement through a voting mechanism / negotiation mechanism to ensure that each optimization will not cause the global system to crash. After the optimization is completed, each optimization decision and policy adjustment are recorded on the chain to ensure traceability in case of problems after optimization.

[0176] It can be understood that the embodiment of the present disclosure realizes a closed-loop mechanism of digital twin network construction-verification-optimization through corresponding verification and optimization operations, further ensuring the consistency between the twin system and the physical system.

[0177] In an embodiment of the present disclosure, the analysis agent, the orchestration agent, the verification agent, and the optimization agent perform loading dependency verification on the required resources and / or operations before executing the operations to eliminate dependency conflicts between the agents.

[0178] Exemplarily, load dependency conflicts are eliminated based on a distributed negotiation protocol. Specifically, before each agent performs an operation, it needs to broadcast the metadata of the upcoming operation (resource requirements, operation object, list of dependent objects). After receiving the broadcast message, each agent checks whether there is a conflict with its own operation metadata. If there is a conflict, a conflict message is returned, and the agent abandons the conflicting operation after receiving the broadcast message.

[0179] In one embodiment of the present disclosure, it further includes: recording the configuration information corresponding to the resource configuration policy and the resource configuration optimization policy to achieve dynamic fault rollback.

[0180] Exemplarily, the configurations and states of networks, network elements, and services can be defined as digital DNA, and version control technology is used to record each change in detail.

[0181] More preferably, a deployment comparison agent can also be set in the digital twin system. The deployment comparison agent is used to compare the deviation between the digital twin system in the current state and the digital twin system corresponding to the baseline DNA. When the deviation exceeds the threshold, a rollback signal is triggered. After the rollback signal is triggered, the system configuration is restored to the baseline DNA version, and the verification agent performs secondary verification on the rolled-back system to ensure rollback safety and prevent new risks introduced by the rollback.

[0182] In another embodiment of the digital twin system construction method of the present disclosure, the following steps are included:

[0183] Obtain mapping information associated with the physical system, where the mapping information at least includes mapping information for determining loaded resources and loading timings; parse the mapping information to generate a resource configuration policy that meets the twin requirements; and generate a digital twin system in response to the resource configuration policy.

[0184] Exemplarily, the mapping information for determining loaded resources and loading timings can be multi-domain real-time status information, and the multi-domain real-time status information includes network domain information, network element domain information, and service domain information.

[0185] Refer to Figure 9 , Figure 9 which schematically shows another embodiment of the digital twin system construction method of the present disclosure, including the following steps:

[0186] Step S910, obtain mapping information associated with the physical system, where the mapping information includes first information and second information;

[0187] Step S920, generate a resource loading list based on the first information;

[0188] Step S930: parsing the loading list according to the second information, determining associated loading resources and loading sequence, and generating a resource configuration strategy;

[0189] Step S940: Build a digital twin system in response to the resource allocation strategy.

[0190] In the embodiment of the present disclosure, the user demand is dynamically converted into an executable resource loading list by fusing the first information including the user's loading intention and the second information dynamically representing the loading resources and loading sequence, and the associated loading resources and loading sequence are determined based on the analysis of the second information to achieve on-demand and collaborative loading of resources, thereby reducing the risk of conflicts caused by mismatched inter-domain status (such as the dependent network element is not ready when the service is started).

[0191] The following is a further explanation of each of the above steps.

[0192] Step S910: Acquire mapping information associated with the physical system, where the mapping information includes first information and second information.

[0193] Exemplarily, the first information includes at least natural language-based user instructions, and may also include other information such as historical operation and maintenance logs.

[0194] Exemplarily, the second information includes at least multi-domain real-time status information, and the multi-domain real-time status information includes network domain information, network element domain information and service domain information.

[0195] Before describing the network domain information, network element domain information, and service domain information, the definitions of the network domain, network element domain, and service domain are described.

[0196] Network domain: Through Internet of Things (IoT) sensors, traffic collection devices, and network monitoring systems, physical network structure and traffic data are obtained in real time to build a mapping of the digital twin system.

[0197] Network element domain: Temporal Graph Convolutional Networks (TGCN) are used to dynamically restore the dependencies and traffic changes between NFVs in the network element domain, ensuring data synchronization and real-time feedback between the physical network and the digital twin.

[0198] The network element domain is defined as follows:

[0199] 1) Real-time collection of NFV basic information, real-time configuration and interdependencies through network management

[0200] 2) Use network monitoring tools to collect network traffic, latency, and resource usage

[0201] 3) Each NFV is a node in the graph, and the node features include configuration, resources, and dependencies.

[0202] 4) NFV dependencies are edges, and the edge features include latency, link, and bilateral signaling.

[0203] 5) Segment the collected data according to time windows to establish timestamps and time series data for the dynamic graph model.

[0204] 6) Deploy an online update module to continuously receive new graph data and dynamically infer the status of each node and network dependencies, enabling the model to restore the dependency relationships and traffic changes between NFVs based on the input temporal features.

[0205] Service domain: modular encapsulation of network element functions to decouple the network element domain and the service domain, making each functional module an independent and schedulable unit.

[0206] The service domain is defined as follows:

[0207] 1) Divide the service domain functions into independent microservices, each microservice encapsulating a specific function, having an independent life cycle, and supporting individual service updates and replacements.

[0208] 2) Define clear API interfaces for each service to ensure that service communication follows a unified specification.

[0209] 3) Use container orchestration tools to manage the services, supporting automatic service deployment, scaling, etc.

[0210] Exemplarily, the network domain information may include network topology information; the network element domain information may include network function virtualization NFV dependency relationship information and / or constraint conditions; the service domain information may include service performance metric information.

[0211] Step S920, generate a resource loading list according to the first information.

[0212] Exemplarily, the resource loading list can be generated in the following manner:

[0213] Step b1, perform keyword matching on the first information based on a preset rule to obtain key features.

[0214] Step b2, encode and map the key features into a high-dimensional feature space for fusion, capture temporal associations based on the cross-attention mechanism, and generate a semantic representation vector.

[0215] Step b3, perform semantic reasoning based on the semantic representation vector to generate the resource loading list.

[0216] Step S930: Parse the loading list according to the second information, determine the associated loading resources and loading timing, and generate a resource configuration policy.

[0217] When performing this operation, it is necessary to first establish a dependency graph, and then, based on the dependency graph and the currently obtained second information, parse the loading list to determine the associated loading resources and loading timing.

[0218] (1) Construct a dependency graph

[0219] Refer to Figure 10 , construct a dependency graph, including nodes (node attributes include node unique identifier, type, metadata (metadata is the requirement of the node itself, such as ports, resource requirements, etc.)) and edges (edge attributes include source node unique identifier, destination node unique identifier, dependency type, constraint conditions (upper limit of delay, lower limit of IO)). There are two types of dependency relationships: dependency and interconnection. A depends on B means that B needs to be loaded first when loading A, and B can only be unloaded after A is unloaded; A is interconnected with B means that A and B need to be loaded and unloaded simultaneously.

[0220] (2) Parse the loading list according to the dependency graph

[0221] First, retrieve nodes from the dependency model library based on the loading list generated by the analysis agent, then retrieve edges according to the nodes, and output the topological loading list order based on the source node, destination node, and dependency type attributes in the edge attributes; if no dependency edge is retrieved, load synchronously.

[0222] (3) Eliminate dependency conflicts

[0223] Eliminate loading dependency conflicts based on the distributed negotiation protocol. Specifically, each agent needs to broadcast the metadata (resource requirements, operation object, dependency object list) of the upcoming operation before performing the operation. After receiving the broadcast message, each agent checks whether there is a conflict with its own operation metadata. If there is a conflict, a conflict message is returned, and the agent abandons the current conflict operation after receiving the broadcast message.

[0224] Exemplarily, based on the above description, in an embodiment of the present disclosure, the NFV dependency relationship information and / or constraint conditions are parsed in the following manner:

[0225] Step a1: Real-time capture the spatio-temporal characteristics of each node and the relationships between nodes in the dependency graph, and update the dependency graph to obtain the dynamic graph model at the current moment.

[0226] Preferably, the spatio-temporal features include spatial dependency features and temporal features; the real-time capturing of the spatio-temporal features of each node and the relationships between nodes in the dependency graph includes: obtaining continuous temporal data associated with the NFV nodes in the current time window; based on the continuous temporal data, extracting spatial dependency features characterizing the dependency relationships and / or interconnection relationships between the various nodes in the dependency graph, and temporal features characterizing the temporal changes of the nodes in the dependency graph. The dependency graph herein is a graph with NFV as nodes and the dependency relationships and / or interconnection relationships between the nodes as edges.

[0227] Step a2, based on the dynamic graph model, determine the NFV dependency relationship information and / or constraint conditions of the target network element.

[0228] That is, at least parse the loading list according to one of the multi-domain real-time status information including network domain information, network element domain information, and service domain information, determine the associated loading resources and loading timings, and generate a resource configuration policy.

[0229] It should be added that network elements include traditional physical network elements and virtualized network elements corresponding to the physical network elements, and the NFV of the target network element refers to the virtualized network element corresponding to the physical network element.

[0230] Step S940, in response to the resource configuration policy, construct a digital twin system.

[0231] It can be seen from the above steps that the embodiments of the present disclosure implement a cross-domain feedback mechanism, that is, by constructing a feedback mechanism across the network domain, network element domain, and service domain, the performance monitoring and optimization results of the service domain can be instantaneously fed back to the network element domain and the network domain.

[0232] Among them, by deploying monitoring tools in the network domain, network element domain, and service domain, the CPU and memory data of each node are collected in real time; edge data aggregation nodes are deployed within each domain to perform preliminary processing (data cleaning, normalization, temporal synchronization) on the collected data of the same type of data within the domain (for example, the network element domain processes all network elements within the domain, and the service domain processes all services within the domain).

[0233] Moreover, a unified data warehouse is built within the network domain, and the edge aggregation nodes of each domain upload the preprocessed data to the unified data warehouse using standardized interfaces. The warehouse aligns and integrates the real-time monitoring data and resource metrics from each domain to construct the entire network topology, the resource occupancy rates of the nodes in each network element domain and service domain, and the network load.

[0234] Perform real-time monitoring on the network topology, the data of each network element domain, and the service domain, and be able to give decision-making suggestions in a timely manner for network failures and system alarms. In one embodiment, it can be executed by a monitoring agent.

[0235] Refer toFigure 11 , Figure 11 is a schematic flowchart of a method for constructing a digital twin system provided by another embodiment of the present disclosure, including the following steps:

[0236] Step S1110, obtaining mapping information associated with the physical system; the mapping information includes first information and second information;

[0237] Step S1120, generating a resource loading list according to the first information; parsing the loading list according to the second information to determine associated loading resources and loading timings, and generating a resource configuration policy;

[0238] Step S1130, constructing a digital twin system in response to the resource configuration policy;

[0239] Step S1140, performing consistency verification on the digital twin system and the physical system, and confirming the verification response deviation;

[0240] Step S1150, generating a resource configuration optimization policy according to the verification response deviation, and the resource configuration optimization policy is used to achieve the consistency between the physical system and the digital twin system.

[0241] In the embodiments of the present disclosure, the dual-information parsing in step S1120 directly solves the problems of manual configuration deviation and cross-domain collaboration deficiency; the verification deviation-driven optimization in steps S1140 - S1150 overcomes the distortion problem caused by environmental changes in the static model. Specifically:

[0242] First, in the embodiments of the present disclosure, an initial loading list is generated through the first information, and then the loading timings and resource associations are dynamically parsed in combination with the second information (for example, adjusting the service deployment order according to the network element resource occupancy rate), realizing cross-domain resource collaborative scheduling and avoiding loading conflicts caused by mismatched inter-domain states (such as the network element on which a service depends not being ready when the service starts).

[0243] Second, in the consistency verification link, the running states of the twin system and the physical system are compared in real time (such as traffic processing results and resource occupancy deviation), and the response deviation is quantified (such as the delay difference exceeding the threshold); an optimization policy (such as dynamically adjusting service deployment and resource reallocation) is automatically generated based on the deviation, forming a "construction-verification-optimization" closed loop to ensure that the twin system continuously and accurately maps the dynamic changes of the physical system (such as network topology update).

[0244] The above two aspects ensure a high degree of consistency between the digital twin system and the physical system.

[0245] It should be noted that the specific implementation manners of the above steps S1110 - S1150 are the same as the operations in the foregoing embodiments, and the embodiments of the present disclosure will not be elaborated herein again. Refer to the foregoing description for details.

[0246] In a second aspect, referring to Figure 12 , an embodiment of the present disclosure provides an electronic device, which includes a memory and a processor; the memory stores a computer program that can be executed by the processor, and when the computer program is executed by the processor, it implements any one of the digital twin system construction methods of the embodiments of the present disclosure.

[0247] In a third aspect, an embodiment of the present disclosure provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements any one of the digital twin system construction methods of the embodiments of the present disclosure.

[0248] Among them, the processor is a device with data processing capabilities, which includes but is not limited to a central processing unit (CPU), etc.; the memory is a device with data storage capabilities, which includes but is not limited to a random access memory (RAM, more specifically such as SDRAM, DDR, etc.), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory (FLASH); the I / O interface (read / write interface) is connected between the processor and the memory and can realize the information interaction between the memory and the processor, and it includes but is not limited to a data bus (Bus), etc.

[0249] In a fourth aspect, an embodiment of the present disclosure further provides a digital twin system construction device, which includes at least one intelligent agent, and the at least one intelligent agent is used to implement:

[0250] Obtain mapping information associated with the physical system in real time;

[0251] Analyze the mapping information to generate a resource configuration policy that meets the twin requirements;

[0252] In response to the resource configuration policy, construct a digital twin system.

[0253] In the embodiments of the present disclosure, the intelligent agent includes an analysis intelligent agent, an orchestration intelligent agent, a verification intelligent agent, and an optimization intelligent agent; the analysis intelligent agent is used to obtain mapping information associated with the physical system in real time; analyze the mapping information to generate a resource configuration policy that meets the twin requirements; the orchestration intelligent agent is used to form a digital twin system in response to the resource configuration policy; the verification intelligent agent is used to perform consistency verification on the digital twin system and the physical system and confirm the verification response deviation; the optimization intelligent agent is used to generate a resource configuration optimization policy based on the verification response deviation, and the resource configuration optimization policy is used to achieve the consistency between the physical system and the digital twin system.

[0254] Wherein, the mapping information includes at least one of a user instruction based on natural language and / or multi-domain real-time status information; wherein, the multi-domain real-time status information includes network domain information, network element domain information, and service domain information.

[0255] Since the technical content involved in the construction device of the digital twin system has been described above, it will not be repeated here, and the relevant parts can be referred to the foregoing description.

[0256] Those of ordinary skill in the art can understand that all or some of the steps, systems, and functional modules / units in the devices disclosed above can be implemented as software, firmware, hardware, and their appropriate combinations.

[0257] In the hardware implementation, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component can have multiple functions, or a function or step can be executed by several physical components in cooperation.

[0258] Some or all physical components can be implemented as software executed by a processor, such as a central processing unit (CPU), a digital signal processor, or a microprocessor, or can be implemented as hardware, or can be implemented as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include a computer storage medium (or non-transitory medium) and a communication medium (or transitory medium). As is well known to those of ordinary skill in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory (FLASH), or other disk memories; compact disc read-only memory (CD-ROM), digital versatile disc (DVD), or other optical disc memories; magnetic cartridges, tapes, magnetic disk storage, or other magnetic memories; any other medium that can be used to store the desired information and can be accessed by a computer. In addition, as is well known to those of ordinary skill in the art, communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and can include any information delivery medium.

[0259] The present disclosure has disclosed example embodiments, and although specific terms are employed, they are used and should be interpreted only in a general illustrative sense and not for purposes of limitation. In some instances, it will be apparent to those skilled in the art that, unless otherwise expressly indicated, features, characteristics, and / or elements described in conjunction with a particular embodiment may be used alone or in combination with features, characteristics, and / or elements described in conjunction with other embodiments. Therefore, it will be understood by those skilled in the art that various changes in form and detail may be made without departing from the scope of the present disclosure as set forth in the appended claims.

Claims

1. A construction method of a digital twin system, applied to at least one intelligent agent, includes: Obtain mapping information associated with the physical system; Parse the mapping information to generate a resource configuration policy that meets the twin requirements; In response to the resource configuration policy, construct a digital twin system.

2. The method according to claim 1, wherein, After generating the digital twin system, it further includes: Perform consistency verification on the digital twin system and the physical system, and confirm the verification response deviation; Generate a resource configuration optimization policy based on the verification response deviation.

3. The method according to claim 1 or 2, wherein, The mapping information includes at least one of a user instruction based on natural language and / or multi-domain real-time status information; Wherein, the multi-domain real-time status information includes network domain information, network element domain information, and service domain information.

4. The method according to claim 3, wherein, The parsing of the mapping information to generate a resource configuration policy that meets the twin requirements includes: Perform keyword matching on the mapping information based on a preset rule library to obtain key features; Encode and map the key features into a high-dimensional feature space for fusion, capture temporal correlations based on a cross-attention mechanism, and generate a semantic representation vector; Perform semantic reasoning based on the semantic representation vector to generate a resource configuration policy.

5. The method according to claim 1 or 2, wherein, The mapping information includes first information and second information; The parsing of the mapping information to generate a resource configuration policy includes: Generate a resource loading list based on the first information; Parse the loading list according to the second information to determine the associated loaded resources and loading timings, and generate a resource configuration policy.

6. The method according to claim 5, wherein, The generating of the resource loading list based on the first information includes: Perform keyword matching on the first information based on a preset rule to obtain key features; Encode and map the key features into a high-dimensional feature space for fusion, capture temporal correlations based on a cross-attention mechanism, and generate a semantic representation vector; Perform semantic reasoning based on the semantic representation vector to generate a resource loading list.

7. The method according to claim 6, wherein, The first information includes a user instruction based on natural language; and / or The second information includes multi-domain real-time status information, and the multi-domain real-time status information includes network domain information, network element domain information, and service domain information.

8. The method according to claim 7, wherein, The network domain information includes network topology information; The network element domain information includes network function virtualization NFV dependency information and / or constraint conditions; The service domain information includes service performance metric information.

9. The method according to claim 8, wherein, The NFV dependency information and / or constraint conditions are determined by the following method: Capture the spatio-temporal features of each node and the relationships between nodes in the dependency graph in real time, and update the dependency graph to obtain a dynamic graph model at the current moment; Based on the dynamic graph model, determine the NFV dependency information and / or constraint conditions of the target network element; Wherein, the dependency graph is a dynamic graph with NFVs as nodes and the dependency relationships and / or interconnection relationships between the nodes as edges.

10. The method according to claim 9, wherein the spatio-temporal features include spatial dependency features and temporal features; the real-time capture of the spatio-temporal features of each node and the relationships between the nodes in the dependency graph includes: obtaining continuous temporal data associated with the NFV nodes in the current time window; based on the continuous temporal data, extracting spatial features characterizing the dependency relationships and / or interconnection relationships of each node in the dependency graph, and temporal features characterizing the temporal changes of each node in the dependency graph.

11. The method according to claim 1, further comprising: optimizing the resource allocation strategy according to the load trend prediction of the physical system.

12. The method according to claim 2, wherein the consistency verification of the digital twin system and the physical system and the confirmation of the verification response deviation include: injecting the operation state data of the physical system into the digital twin system; respectively obtaining the responses of the physical system and the digital twin system based on the operation state data, and confirming the verification response deviation.

13. The method according to claim 2, wherein generating a resource allocation optimization strategy based on the verification response deviation includes: obtaining the performance metrics of the digital twin system in the network domain, network element domain or service domain; in the case of abnormal performance metrics, identifying the cause of the abnormality and generating a resource allocation optimization strategy.

14. The method according to any one of claims 1-13, further comprising: Before each agent executes the corresponding operation, perform a loading dependency check on the required resources and / or operations to eliminate the dependency conflicts of each agent.

15. An electronic device, comprising a memory and a processor; the memory stores a computer program executable by the processor, and when the computer program is executed by the processor, the construction method according to any one of claims 1 to 14 is implemented.

16. A computer program product, comprising a computer program, and when the computer program is executed by a processor, the construction method according to any one of claims 1 to 14 is implemented.

17. A construction device for a digital twin system, comprising at least one agent, and the at least one agent is used to implement: real-time obtain mapping information associated with the physical system; parse the mapping information to generate a resource allocation strategy that meets the twin requirements; generate a digital twin system in response to the resource allocation strategy.

18. The construction device according to claim 17, wherein the agents include an analysis agent, an orchestration agent, a verification agent and an optimization agent; the analysis agent is used to real-time obtain mapping information associated with the physical system; parse the mapping information to generate a resource allocation strategy that meets the twin requirements; the orchestration agent is used to form a digital twin system in response to the resource allocation strategy; the verification agent is used to perform consistency verification on the digital twin system and the physical system and confirm the verification response deviation; The optimization agent is used to generate a resource configuration optimization strategy according to the verification response deviation, and the resource configuration optimization strategy is used to reduce the state deviation between the physical system and the digital twin system.

19. The construction device according to claim 17 or 18, wherein The mapping information includes at least one of user instructions based on natural language and / or multi-domain real-time state information; wherein the multi-domain real-time state information includes network domain information, network element domain information, and service domain information.

Images