Digital twin system construction method, modeling method, network optimization method and device

Abstract

Embodiments of the present application provide a digital twin system construction method, a modeling method, a network optimization method and device, the method comprising: obtaining first information from a first object, the first information comprising: data collected and / or perceived by one or more sensors in the first object, and / or data reported by entities in the first object, the first object being: a physical space or a physical entity or a physical system or a real system; inputting the first information into a second object, the second object comprising: a virtual entity and / or a digital twin model, the second object being: a digital space or a virtual entity or a twin system or a twin body or a digital system or a digital twin system or a virtual system; obtaining a first instruction through the digital twin model and the first information; inputting the first instruction into the first object and / or the second object, and controlling and / or optimizing the first object and / or the second object through the first instruction.

Description

Technical Field

[0001] This application relates to the field of communication technology, specifically to a method for constructing a digital twin system, a modeling method, a network optimization method, and an apparatus. Background Technology

[0002] With the construction of 5G mobile networks and the network requirements for enterprise (toB) business scenarios, in order to achieve the connection between field networks (such as Wi-Fi, Radio Frequency Identification (RFID), Zigbee, etc.) and 5G gateways or 5G base stations without modifying existing network hardware equipment in industry sites (such as factories), industry field networks have gradually become the last mile for 5G networks to enter vertical industries. Their multi-source heterogeneous network standards need to be tightly coupled and flexibly managed to meet the diverse needs of vertical industries such as industry and transportation in terms of field sensing, data, positioning, control, and management, especially the network function requirements of different industry application scenarios, such as high speed, low latency, high reliability, and high bandwidth, and ultimately achieve full life-cycle management of the network, including planning, construction, operation, maintenance, and optimization.

[0003] The networking configuration and access terminals of industry field networks are multi-source and heterogeneous. Coupled with the huge and personalized business needs, this will inevitably bring a certain impact to network operation and maintenance.

[0004] Therefore, how to maintain the network efficiently is an urgent problem to be solved. Summary of the Invention

[0005] This application provides a method for constructing a digital twin system, a modeling method, a network optimization method, and an apparatus to solve the problem of how to efficiently maintain a network.

[0006] Firstly, a method for constructing a digital twin system is provided, including:

[0007] First information is obtained from a first object, the first information including: data collected and / or sensed by one or more sensors in the first object, and / or data reported by an entity in the first object, the first object being: physical space or physical entity or physical system or real system;

[0008] The first information is input into the second object, which includes: a virtual entity and / or a digital twin model. The second object is: a digital space or a virtual entity or a twin system or a twin body or a digital system or a digital twin system or a virtual system.

[0009] The first instruction is obtained through the digital twin model and the first information;

[0010] The first instruction is input to the first object and / or the second object, and the first object and / or the second object are controlled and / or optimized through the first instruction.

[0011] Optionally, the data includes one or more of the following:

[0012] Sensor data, such as temperature and pressure;

[0013] Status data, such as: valve open or closed, engine on or off, etc.;

[0014] Event data, such as alarms caused by metrics exceeding thresholds;

[0015] Environmental data, such as weather data, maintenance logs, supply chain data, and inventory data.

[0016] Optionally, the first instruction includes one or more of the following:

[0017] Generate alarms;

[0018] Trigger analysis and / or decision-making and / or control and / or optimization of the corresponding workflow;

[0019] Driver fault repair;

[0020] Drive predictive analytics and / or maintenance.

[0021] Optionally, the method further includes:

[0022] Obtain the second information of the first object;

[0023] Based on the second information, construct a virtual entity and / or digital twin model of the second object, and / or the interaction logic between the various digital twin models;

[0024] The second information includes one or more of the following:

[0025] Information about each element in the first object;

[0026] Information about each entity in the first object;

[0027] The information of the elements includes one or more of the following:

[0028] Information about maintenance personnel;

[0029] Information on physical devices and / or systems;

[0030] Information about the network formed by the connections between physical entities;

[0031] Information on the operation of physical equipment and / or systems;

[0032] Knowledge of physical equipment and / or system operation and maintenance;

[0033] Physical equipment and / or system operation instructions;

[0034] Information about the site environment;

[0035] The information of the entity includes one or more of the following:

[0036] Topological information;

[0037] Performance information;

[0038] Measurement information;

[0039] Functional information.

[0040] Optionally, the digital twin model includes one or more of the following:

[0041] A geometric model, which is used to visualize one or more of the following: a first object, topological relationship, operating state, analysis process and / or result, decision process and / or result, control process and / or instruction, optimization process and / or effect;

[0042] An information model is used to provide a normalized semantic description of one or more of the attributes, capabilities, interfaces, and data flows of a first object.

[0043] The mechanism model has the following functions: knowledge mining of the first object; mining of the operational rules of the first object; analysis, decision-making, control and optimization of the entire life cycle of the first object.

[0044] Optionally, the type of the mechanistic model includes one or more of the following:

[0045] The operating mechanism of a device or system that has a clear principle and can be expressed by a numerical formula;

[0046] A rule set, rule engine, knowledge base, knowledge engine, or knowledge graph formed based on experience;

[0047] Mechanism models generated based on big data and / or AI algorithms.

[0048] Optionally, the step of obtaining the first instruction through the digital twin model and the first information includes:

[0049] Based on the first information, one or more of the following mechanisms—channel conflict detection mechanism model, network key performance indicator (KPI) prediction mechanism model, and network anomaly detection mechanism model—as well as the channel optimization strategy mechanism model, the first instruction is determined.

[0050] Optionally, it also includes:

[0051] The first information is input into one or more of the channel conflict detection mechanism model, the network KPI prediction mechanism model, and the network anomaly detection mechanism model;

[0052] The output results of one or more of the channel conflict detection mechanism model, network KPI prediction mechanism model, and network anomaly detection mechanism model are input into the channel optimization strategy mechanism model. The channel optimization strategy mechanism model includes: anomaly access point (AP) location mechanism model, channel replanning mechanism model, and radio frequency quality assessment mechanism model.

[0053] Optionally, the method further includes:

[0054] The abnormal AP location mechanism model is used to identify APs with channel conflicts.

[0055] Based on the load of the APs experiencing channel conflicts, one or more channel optimization schemes are determined through a channel replanning mechanism model.

[0056] The radio frequency quality of one or more channel optimization schemes is evaluated using a radio frequency quality assessment mechanism model, and the evaluation results are obtained.

[0057] Based on the evaluation results, the first instruction is determined, which is a control instruction for the optimal channel optimization scheme among the one or more channel optimization schemes.

[0058] Optionally, before the step of inputting the first instruction to the first object, the method further includes:

[0059] Apply the control quantity indicated by the first instruction to the second object;

[0060] Obtain the optimized state of the second object;

[0061] If the status of the second object indicates that the second object is operating normally, then the step of inputting the first instruction into the first object is executed.

[0062] Optionally, the state includes one or more of the following:

[0063] AP topology status;

[0064] AP load status;

[0065] AP channel status;

[0066] Network KPI status.

[0067] Optionally, the method further includes:

[0068] The optimal channel optimization scheme currently determined is executed on the first object, and one or more of the following are recorded: AP status, network KPI status, channel conflict information, and network anomaly information.

[0069] The radio frequency quality assessment mechanism model is trained by using one or more of the recorded AP status, network KPI status, channel conflict information, and network anomaly information as training samples.

[0070] Secondly, a digital twin system modeling method is provided, including:

[0071] Mathematical representations of a first object and a second object are created respectively. The first object is: physical space or physical entity or physical system or real system, and the second object is: digital space or virtual entity or twin system or twin body or digital system or digital twin system or virtual system.

[0072] Based on the mathematical representations of the first and second objects, create control optimization models for the system construction phase and / or system execution phase.

[0073] Optionally, the mathematical representation of the first object is:

[0074]

[0075] in, These are: the state of the first object at time t, the system state transition rules, the observation rules, the observed quantities, and the control quantities. These are the state and control variables of the first object at time t-1, respectively. The function represents the system state transition rule. This represents the observation rule function.

[0076] Optionally, the mathematical representation of the second object is:

[0077]

[0078] in, These respectively represent: the state mechanism model of the second object at time t, the system state transition rule mechanism model, the observation rule mechanism model, the observation mechanism model, and the control variable mechanism model. These respectively represent: the state mechanism model and the control quantity mechanism model of the second object at time t-1. This represents the system's state transition rule mechanism model. This represents the observation rule mechanism model.

[0079] Optionally, the method further includes:

[0080] Based on the mathematical representations of the first object and the second object and the first formula, a control optimization model for the system construction phase is constructed.

[0081] The first formula is:

[0082]

[0083] in, This indicates minimizing the observed distance measurement between the first object and the second object. This represents the distance measurement between the state of the second object at time t and its state at time t+T.

[0084] Optionally, the method further includes:

[0085] Based on the mathematical representations and second formulas of the first and second objects, a control optimization model for the system execution phase is constructed.

[0086] The second formula is:

[0087]

[0088] in, This represents the control pair between the first object and the second object;

[0089] This represents minimizing the observation distance metric between the first and second objects. This represents the distance metric between the state of the second object at time t and its state at time t+T.

[0090] Optionally, the method further includes:

[0091] according to The first object and the second object are controlled separately to minimize the distance metric between the state of the second object at time t and the state at time t+T.

[0092] Thirdly, a network optimization method is provided, the method comprising:

[0093] Mathematical representations of a first object and a second object are created respectively. The first object is: physical space or physical entity or physical system or real system, and the second object is: digital space or virtual entity or twin system or twin body or digital system or digital twin system or virtual system.

[0094] Based on the mathematical representations of the first and second objects and the network optimization objectives, create a control optimization model for the system construction phase and / or execution phase and / or simulation execution and verification process;

[0095] Based on the control optimization model of the system construction phase and / or execution phase and / or simulation execution and verification process, determine the control variables used to optimize the network.

[0096] Optionally, determining the control variables for optimizing the network based on the control optimization model from the system construction phase includes:

[0097] Based on the control optimization model from the system construction phase, the control variables used to optimize the network are determined to be the optimal solution set.

[0098] The control optimization model for the system construction phase is as follows:

[0099] in, For network performance evaluation functions, These respectively represent: the system state transition rule mechanism model, the observation rule mechanism model, the observed quantity, and the control quantity of the second object at time t. It is the control quantity of the first object at time t. These are the observations of the first object at time t. This represents minimizing the observation distance metric between the first object and the second object.

[0100] Optionally, determining the control variables for optimizing the network based on the control optimization model of the system execution phase includes:

[0101] Based on the control optimization model for the system execution phase, the control variables used to optimize the network are determined to be the optimal solution set.

[0102] The control optimization model for the system execution phase is as follows:

[0103]

[0104]

[0105]

[0106]

[0107]

[0108]

[0109] in, The AP topology, load, and channel state of the digital twin system at time t; The network KPI status of the digital twin system at time t; These are system observations of AP channel conflict in the virtual system and system observations of network performance anomalies in the virtual system, respectively. The digital twin system corresponds to time t. System transfer rules; These are the "channel conflict detection" mechanism model and the "network anomaly detection" mechanism model, respectively. These represent all system observations of the digital twin system at times t-1 and t, including channel conflicts and network performance anomalies for each AP.

[0110] Optionally, determining the control variables for optimizing the network based on the control optimization model of the system simulation execution and verification process includes:

[0111] Based on the control optimization model of the system simulation and verification process, the control quantity used to optimize the network is determined to be the optimal solution.

[0112] The control optimization model for the system simulation execution and verification process is as follows:

[0113]

[0114] Subject to,

[0115]

[0116]

[0117]

[0118]

[0119]

[0120] in, The AP topology, load, and channel state of the digital twin system at time t; These are, respectively, system observations of AP channel conflict in the virtual system, system observations of network KPI prediction in the virtual system, and system observations of network performance anomalies in the virtual system; For the digital twin system at time t, corresponding to System transfer rules; These are the mechanism models for "channel conflict detection", "network KPI prediction", and "network anomaly detection". These represent all system observations of the digital twin system at times t-1 and t, including the channel conflict status of each AP, network KPI prediction status, and network performance anomaly status.

[0121] Optionally, the Includes one or more of the following:

[0122] AP topology change rules;

[0123] AP load variation rules;

[0124] AP channel variation rules;

[0125] Network KPI status change rules;

[0126] or,

[0127] The Includes one or more of the following:

[0128] Channel collision detection mechanism model;

[0129] Network KPI prediction mechanism model;

[0130] Network anomaly detection mechanism model.

[0131] Fourthly, a digital twin system construction apparatus is provided, comprising:

[0132] The first acquisition module is used to acquire first information from the first object, the first information including: data collected and / or sensed by one or more sensors in the first object, and / or data reported by entities in the first object, the first object being: physical space or physical entity or physical system or real system;

[0133] A first processing module is configured to input the first information into a second object, the second object including: a virtual entity and / or a digital twin model, the second object being: a digital space or virtual entity or twin system or twin body or digital system or digital twin system or virtual system; obtain a first instruction through the digital twin model and the first information; input the first instruction into the first object and / or the second object, and control and / or optimize the first object and / or the second object through the first instruction.

[0134] Fifthly, a digital twin system modeling apparatus is provided, comprising:

[0135] The sixth processing module is used to create mathematical representations of the first object and the second object respectively. The first object is: physical space or physical entity or physical system or real system, and the second object is: digital space or virtual entity or twin system or twin body or digital system or digital twin system or virtual system.

[0136] The seventh processing module is used to create a control optimization model for the system construction phase and / or system execution phase based on the mathematical representations of the first object and the second object.

[0137] Sixthly, a network optimization device is provided, comprising:

[0138] The eleventh processing module is used to create mathematical representations of a first object and a second object, respectively. The first object is a physical space, physical entity, physical system, or real system, and the second object is a digital space, virtual entity, twin system, twin body, digital system, digital twin system, or virtual system.

[0139] The twelfth processing module is used to create a control optimization model for the system construction phase and / or execution phase based on the mathematical representations of the first and second objects and the network optimization objective.

[0140] The thirteenth processing module is used to determine the control quantities for optimizing the network based on the control optimization model of the system construction phase and / or execution phase.

[0141] A seventh aspect provides an electronic device comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, performs the steps of the method as described in the first, second, or third aspect.

[0142] Eighthly, a readable storage medium is provided, characterized in that a program is stored on the readable storage medium, and when the program is executed by a processor, it implements the steps of the method as described in the first, second, or third aspect.

[0143] In this embodiment, physical entities, physical systems, or real systems are mapped to digital entities, virtual entities, twin systems, digital systems, digital twin systems, or virtual systems through data perception, and virtual entities and / or digital twin models are generated in the virtual system. Analysis and decision-making are performed through virtual entities and / or digital twin models to generate interactive commands to intervene in and regulate the physical space, physical entities, physical systems, or real systems, so as to keep the entire physical network system in good operating condition. Attached Figure Description

[0144] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0145] Figure 1 This is a flowchart of the method for constructing a digital twin system provided in the embodiments of this application;

[0146] Figure 2 This is a flowchart of the digital twin system modeling method provided in the embodiments of this application;

[0147] Figure 3 This is a flowchart of the network optimization method provided in the embodiments of this application;

[0148] Figure 4 This application provides a lightweight industry field network digital twin system.

[0149] Figure 5 This is a schematic diagram of the digital twin system construction pattern provided in the embodiments of this application;

[0150] Figure 6 This is a schematic diagram of the execution mode of the digital twin system provided in the embodiments of this application;

[0151] Figure 7 This is a schematic diagram illustrating mutual interference between APs caused by unreasonable channel planning, provided in an embodiment of this application.

[0152] Figure 8 This is a schematic diagram illustrating how three APs with channel 6 are too close together, causing mutual interference between them, according to an embodiment of this application.

[0153] Figure 9 This is a schematic diagram illustrating channel conflict and wasted channel 6 resources in a high-load AP according to an embodiment of this application.

[0154] Figure 10 This is a schematic diagram illustrating how high-load AP channels are staggered to ensure a good user experience, as provided in this embodiment of the application.

[0155] Figure 11 This is a schematic diagram of intelligent wireless radio frequency optimization based on an industry field network digital twin system provided in an embodiment of this application;

[0156] Figure 12 This is a schematic diagram of a neural network model;

[0157] Figure 13 This is a schematic diagram of multivariate time series anomaly detection based on the Variational Auto-Encoders (VAE) algorithm;

[0158] Figure 14 This is a schematic diagram of the apparatus for constructing a digital twin system provided in an embodiment of this application;

[0159] Figure 15 This is a schematic diagram of the digital twin system modeling device provided in the embodiments of this application;

[0160] Figure 16 This is a schematic diagram of the network optimization device provided in the embodiments of this application;

[0161] Figure 17 This is a schematic diagram of the communication device provided in the embodiments of this application. Detailed Implementation

[0162] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0163] The term "comprising," and any variations thereof, used in the specification and claims of this application, is intended to cover a non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or apparatus. Furthermore, the use of "and / or" in the specification and claims indicates at least one of the connected objects, such as A and / or B, indicating the inclusion of A alone, B alone, or both A and B.

[0164] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0165] It is worth noting that the technologies described in this application are not limited to Long Term Evolution (LTE) / LTE-Advanced (LTE-A) systems, but can also be used in other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency-Division Multiple Access (SC-FDMA), and other systems. The terms "system" and "network" in this application are often used interchangeably, and the described technologies can be used with the systems and radio technologies mentioned above, as well as with other systems and radio technologies. However, the following description describes New Radio (NR) systems for illustrative purposes, and the term NR is used in most of the following description, although these technologies can also be applied to applications other than NR systems, such as 6th Generation (6G) communication systems.

[0166] With the development of next-generation communication and information processing technologies, digital twin technology, as a new paradigm driving industrial automation, digitalization, and intelligence, mainly includes physical space and entities, virtual space and twins, and the interaction between them. On the one hand, the operating mechanisms of complex real systems in physical space are often unclear and unobservable, posing significant challenges to system management and optimization. Digital twin technology, through data-driven construction and solving of various models in digital space, achieves the desired control objectives while infinitely approximating the real system, thus optimizing the real system. On the other hand, by constructing digital twins for real-time simulation, dynamic optimization, and trial-and-error simulation, it enables the management and control of the entire product lifecycle, reducing operating and maintenance costs and improving operational efficiency. In summary, due to its intelligence, simulation capabilities, computability, controllability, and predictability, digital twins can address the management and optimization of complex systems throughout their entire lifecycle in a more efficient and cost-effective manner. Therefore, in the face of increasingly complex multi-source heterogeneous industry field networks, digital twin technology can be used to map network entities in the physical space into network twins in the virtual space through data perception. By constructing a network mechanism model in the virtual space for analysis and decision-making, interactive commands can be generated to intervene and regulate the physical space, so that the entire physical network system can maintain a good operating state.

[0167] On the one hand, existing digital twin system construction methods produce digital twin models and their interaction logic that are too complex to meet the needs of lightweight system design; moreover, they are often built for specific scenarios and lack universality. For example, the evolution mechanism and methods of digital twins for smart mining scenarios focus on 3D visualization modeling and virtual reality simulation, but do not build specific methods for closed-loop control and optimization of virtual and real systems from a general perspective.

[0168] On the other hand, existing digital twin systems for network scenarios mainly focus on constructing network digital twin models and optimizing models within the virtual space, without mathematical modeling from the perspective of system control to achieve closed-loop optimization of virtual-real interaction, nor are they verified in specific network operation and maintenance scenarios. For example, a network optimization model obtained in a communication network optimization scheme based on digital twins is obtained through simulation, optimization adjustment, and pre-action in the digital twin space, without joint optimization during the interaction between virtual and real systems. Another example is a network model based on digital twins that proposes a method for constructing the network digital twin model and provides its mathematical representation, but only lists the mathematical representations without mathematical modeling from the perspective of organic interaction between models and closed-loop feedback and optimization between virtual and real systems to achieve full lifecycle control and optimization of the real network.

[0169] Existing methods for constructing digital twin systems and network optimization based on digital twins have the following drawbacks:

[0170] (1) Lack of lightweight digital twin system design solutions;

[0171] (2) The interaction and control logic between physical space and digital space is still at the semantic description stage, lacking a closed-loop joint optimization method based on virtual-real interaction feedback and a computational framework based on mathematical modeling that is calculable and controllable.

[0172] (3) There is a lack of verification of the operating logic of the network digital twin system and the closed-loop control method of virtual-real interaction in network operation and maintenance and optimization scenarios.

[0173] See Figure 1 This application provides a method for constructing a digital twin system, with specific steps including: step 101, step 102, step 103 and step 104.

[0174] Step 101: Obtain first information from the first object, the first information including: data collected and / or sensed by one or more sensors in the first object, and / or data reported by entities in the first object, the first object being: physical space or physical entity or physical system or real system;

[0175] Step 102: Input the first information into the second object, the second object including: virtual entity (or twin, or digital twin) and / or digital twin model (or virtual model, or digital model, or twin model), the second object is: digital space or virtual entity or twin system or twin or digital system or digital twin system or virtual system;

[0176] Step 103: Obtain the first instruction using the digital twin model and the first information;

[0177] Step 104: Input the first instruction into the first object and / or the second object, and control and / or optimize the first object and / or the second object through the first instruction.

[0178] In one embodiment of this application, the data includes one or more of the following:

[0179] Sensor data, such as temperature and pressure;

[0180] Status data, such as: valve open or closed, engine on or off, etc.;

[0181] Event data, such as alarms caused by metrics exceeding thresholds;

[0182] Environmental data, such as weather data, maintenance logs, supply chain data, and inventory data.

[0183] In one embodiment of this application, the first instruction includes one or more of the following:

[0184] Generate alarms;

[0185] Trigger analysis and / or decision-making and / or control and / or optimization of the corresponding workflow;

[0186] Driver fault repair;

[0187] Drive predictive analytics and / or maintenance.

[0188] In one embodiment of this application, the method further includes:

[0189] Obtain the second information of the first object;

[0190] Based on the second information, construct a virtual entity (or twin, or digital twin) and / or a digital twin model (or described as a digital twin model, twin model) of the second object, and / or the interaction logic between the various digital twin models;

[0191] The second information includes one or more of the following:

[0192] (1) Information about the elements in the physical space;

[0193] (2) Information about physical entities in the physical space;

[0194] The information of the elements includes one or more of the following:

[0195] (1) Information on maintenance personnel;

[0196] (2) Information on physical equipment and / or systems;

[0197] (3) Information about the network formed by the connections between physical entities;

[0198] (4) Information on the operation of physical equipment and / or systems;

[0199] (5) Knowledge of physical equipment and / or system operation and maintenance;

[0200] (6) Information on the operation of physical equipment and / or systems (or described as operation instructions);

[0201] (7) Information about the site environment;

[0202] The information of the physical entity includes one or more of the following:

[0203] (1) Topological information of the physical entity;

[0204] (2) Performance information of the physical entity;

[0205] (3) Measurement information of the physical entity;

[0206] (4) Functional information of the physical entity.

[0207] The design of a minimalist 3D model for digital space greatly simplifies the complexity of digital space model construction and interaction, enabling lightweight system design.

[0208] In this embodiment, the five elements of physical space—human-machine-network-law-environment—and the four-dimensional information of physical entities—topology, performance, telemetry, and function—as well as the actual operating mechanism of the network, are summarized into a three-dimensional model of digital space: a geometric model, an information model, and a mechanism model, along with their interaction logic. The elements, specific connotations, and construction methods of each model are clearly expressed mathematically, thus realizing the design of a lightweight digital twin system.

[0209] The main advantage of adopting digital twin technology is that once the physical system is mapped into the digital space, various artificial intelligence (AI) and control system analysis techniques, such as deep learning, optimal control, and adaptive optimization control, can be applied within this space. Specifically, knowledge graphs and AI algorithms are used to uncover the real-world mechanisms within a physical system that is organically integrated with physical entities, revealing complex operational logic and functions. Various mechanism models are then constructed in the digital space. Information models built from real data (including historical and real-time data) perceived and collected from the physical space drive and optimize these mechanism models. Operational instructions generated by these mechanism models manage and optimize the actual system in the physical space, and the results are visualized on the digital twin system based on geometric models. In short, algorithms are used to build mechanism models, and data drives and optimizes these models. Simulation, optimization, verification, and trial-and-error are then performed in the digital space, enabling automated, digitalized, and intelligent management and operation of complex systems throughout their entire lifecycle.

[0210] In one embodiment of this application, the digital twin model includes one or more of the following:

[0211] (1) Geometric model, wherein the geometric model is used to visualize one or more of the following: first object, topological relationship, operating status (including but not limited to fault, abnormality, etc.), analysis process and / or result, decision process and / or result, control process and / or instruction, optimization process and / or effect;

[0212] (2) Information model, wherein the information model is used to provide a standardized semantic description of one or more of the attributes, capabilities, interfaces and data flows of the first object;

[0213] (3) Mechanism model, wherein the functions of the mechanism model include one or more of the following: knowledge mining of the first object; mining of the operating rules of the first object; analysis, decision-making, control and optimization of the entire life cycle of the first object.

[0214] In one embodiment of this application, the type of the mechanism model includes one or more of the following:

[0215] (1) The operating mechanism (or operating law, operating mechanism) of a device or system that has a clear mechanism and can be expressed by a numerical formula;

[0216] (2) A set of rules, a rule engine, a knowledge base, a knowledge engine, or a knowledge graph formed based on experience;

[0217] (3) Mechanism models generated based on big data and / or AI algorithms (including machine learning and / or deep learning).

[0218] In one embodiment of this application, the step of obtaining the first instruction through the digital twin model and the first information includes:

[0219] Based on the first information, one or more of the channel conflict detection mechanism model, the network KPI prediction mechanism model, and the network anomaly detection mechanism model, as well as the channel optimization strategy mechanism model, the first instruction is determined.

[0220] In one embodiment of this application, it further includes:

[0221] The first information is input into one or more of the channel conflict detection mechanism model, the network KPI prediction mechanism model, and the network anomaly detection mechanism model;

[0222] The output results of one or more of the channel conflict detection mechanism model, network KPI prediction mechanism model, and network anomaly detection mechanism model are input into the channel optimization strategy mechanism model, which includes: anomaly AP location mechanism model, channel replanning mechanism model, and radio frequency quality assessment mechanism model.

[0223] In one embodiment of this application, the method further includes:

[0224] The abnormal AP location mechanism model is used to identify APs with channel conflicts.

[0225] Based on the load of the APs experiencing channel conflicts, one or more channel optimization schemes are determined through a channel replanning mechanism model.

[0226] The radio frequency quality of one or more channel optimization schemes is evaluated using a radio frequency quality assessment mechanism model, and the evaluation results are obtained.

[0227] Based on the evaluation results, the first instruction is determined, which is a control instruction (or control quantity, optimization instruction, or operation and maintenance instruction) for the optimal channel optimization scheme among the one or more channel optimization schemes.

[0228] In one embodiment of this application, prior to the step of inputting the first instruction to the first object, the method further includes:

[0229] Apply the control quantity indicated by the first instruction to the second object;

[0230] Obtain the optimized state of the second object;

[0231] If the status of the second object indicates that the second object is operating normally (or the fault is eliminated, or the fault is repaired, or the abnormality is eliminated, or the abnormality is repaired), then the step of inputting the first instruction to the first object is executed.

[0232] In one embodiment of this application, the state includes one or more of the following:

[0233] (1) AP topology status;

[0234] (2) AP load status;

[0235] (3) AP channel status;

[0236] (4) Network KPI status.

[0237] In one embodiment of this application, the method further includes:

[0238] The optimal channel optimization scheme currently determined is executed on the first object, and one or more of the following are recorded: AP status, network KPI status, channel conflict information, and network anomaly information.

[0239] The radio frequency quality assessment mechanism model is trained by using one or more of the recorded AP status, network KPI status, channel conflict information, and network anomaly information as training samples.

[0240] In this embodiment, physical entities, physical systems, or real systems are mapped to digital entities, virtual entities, twin systems, digital systems, digital twin systems, or virtual systems through data perception, and virtual entities and / or digital twin models are generated in the virtual system. Analysis and decision-making are performed through virtual entities and / or digital twin models to generate interactive commands to intervene in and regulate the physical space, physical entities, physical systems, or real systems, so as to keep the entire physical network system in good operating condition.

[0241] See Figure 2 This application provides a digital twin system modeling method, which includes steps 201 and 202.

[0242] Step 201: Create mathematical representations of the first object and the second object respectively. The first object is: physical space or physical entity or physical system or real system, and the second object is: digital space or virtual entity or twin system or twin body or digital system or digital twin system or virtual system.

[0243] Step 202: Based on the mathematical representations of the first and second objects, create a control optimization model (or control model, optimization model, closed-loop control model, joint optimization model, decision model, objective equation) for the system construction phase and / or system execution phase.

[0244] In one embodiment of this application, the mathematical representation of the first object is:

[0245]

[0246] in, These are: the state of the first object at time t, the system state transition rules, the observation rules, the observed quantities, and the control quantities. These are the state and control variables of the first object at time t-1, respectively. The function represents the system state transition rule. This represents the observation rule function.

[0247] In one embodiment of this application, the mathematical representation of the second object is:

[0248]

[0249] in, These respectively represent: the state mechanism model of the second object at time t, the system state transition rule mechanism model, the observation rule mechanism model, the observation mechanism model, and the control variable mechanism model. These respectively represent: the state mechanism model and the control quantity mechanism model of the second object at time t-1. This represents the system's state transition rule mechanism model. This represents the observation rule mechanism model.

[0250] In one embodiment of this application, the method further includes:

[0251] Based on the mathematical representations of the first object and the second object and the first formula, a control optimization model for the system construction phase is constructed.

[0252] The first formula is:

[0253]

[0254] in, This indicates minimizing the observed distance measurement between the first object and the second object. This represents the distance measurement between the state of the second object at time t and its state at time t+T.

[0255] In one embodiment of this application, the method further includes:

[0256] Based on the mathematical representations and second formulas of the first and second objects, a control optimization model for the system execution phase is constructed.

[0257] The second formula is:

[0258]

[0259] in, This represents the control pair between the first object and the second object;

[0260] This represents minimizing the observation distance metric between the first and second objects. This represents the distance metric between the state of the second object at time t and its state at time t+T.

[0261] In one embodiment of this application, the method further includes:

[0262] according to The first object and the second object are controlled separately to minimize the distance metric between the state of the second object at time t and the state at time t+T.

[0263] In this embodiment, by creating a control optimization model for the system construction phase and / or system execution phase, it is possible to map entities in physical space, physical entities, physical systems, or real systems into virtual entities and / or digital twin models in digital space, virtual entities, twin systems, digital twin systems, or virtual systems through data perception. Analysis and decision-making are then performed using virtual entities and / or digital twin models to generate interactive commands that intervene in and regulate the physical space, physical entities, physical systems, or real systems, thereby maintaining the entire physical network system in a good operating state.

[0264] See Figure 3 This application provides a network optimization method, which includes steps 301, 302 and 303.

[0265] Step 301: Create mathematical representations of the first object and the second object respectively. The first object is: physical space or physical entity or physical system or real system. The second object is: digital space or virtual entity or twin system or twin body or digital system or digital twin system or virtual system.

[0266] Step 302: Based on the mathematical representations of the first and second objects and the network optimization objective, create a control optimization model (or control model, optimization model, closed-loop control model, joint optimization model, decision model, objective equation) for the system construction phase and / or execution phase and / or simulation execution and verification process;

[0267] Step 303: Based on the control optimization model of the system construction phase and / or execution phase and / or simulation execution and verification process, determine the control quantities (or control instructions, optimization quantities, optimization instructions, operation and maintenance instructions, etc.) used to optimize the network.

[0268] In one embodiment of this application, determining the control variables for optimizing the network based on the control optimization model of the system construction phase includes:

[0269] Based on the control optimization model from the system construction phase, the control variables used to optimize the network are determined to be the optimal solution set.

[0270] The control optimization model for the system construction phase is as follows:

[0271] in, For network performance evaluation functions, These respectively represent: the system state transition rule mechanism model, the observation rule mechanism model, the observed quantity, and the control quantity of the second object at time t. It is the control quantity of the first object at time t. These are the observations of the first object at time t. This represents minimizing the observation distance metric between the first object and the second object.

[0272] In one embodiment of this application, determining the control variables for optimizing the network based on the control optimization model of the system execution phase includes:

[0273] Based on the control optimization model for the system execution phase, the control variables used to optimize the network are determined to be the optimal solution set.

[0274] The control optimization model for the system execution phase is as follows:

[0275]

[0276] Subject to the following conditions:

[0277]

[0278]

[0279]

[0280]

[0281]

[0282] in, The AP topology, load, and channel state of the digital twin system at time t; The network KPI status of the digital twin system at time t; These are system observations of AP channel conflict in the virtual system and system observations of network performance anomalies in the virtual system, respectively. The digital twin system corresponds to time t. System transfer rules; These are the "channel conflict detection" mechanism model and the "network anomaly detection" mechanism model, respectively. These represent all system observations of the digital twin system at times t-1 and t, including channel conflicts and network performance anomalies for each AP.

[0283] Optionally, determining the control variables for optimizing the network based on the control optimization model of the system simulation execution and verification process includes:

[0284] Based on the control optimization model of the system simulation and verification process, the control quantity used to optimize the network is determined to be the optimal solution.

[0285] The control optimization model for the system simulation execution and verification process is as follows:

[0286]

[0287] Subject to,

[0288]

[0289]

[0290]

[0291]

[0292]

[0293] in, The AP topology, load, and channel state of the digital twin system at time t; These are, respectively, system observations of AP channel conflict in the virtual system, system observations of network KPI prediction in the virtual system, and system observations of network performance anomalies in the virtual system; For the digital twin system at time t, corresponding to System transfer rules; These are the mechanism models for "channel conflict detection", "network KPI prediction", and "network anomaly detection". These represent all system observations of the digital twin system at times t-1 and t, including the channel conflict status of each AP, network KPI prediction status, and network performance anomaly status.

[0294] In the embodiments of this application, the Includes one or more of the following:

[0295] AP topology change rules;

[0296] AP load variation rules;

[0297] AP channel variation rules;

[0298] Network KPI status change rules;

[0299] or,

[0300] The Includes one or more of the following:

[0301] Channel collision detection mechanism model;

[0302] Network KPI prediction mechanism model;

[0303] Network anomaly detection mechanism model.

[0304] In this embodiment, physical entities, physical systems, or real systems are mapped to digital entities, virtual entities, twin systems, digital systems, digital twin systems, or virtual systems through data perception, and virtual entities and / or digital twin models are generated in the virtual system. Analysis and decision-making are performed through virtual entities and / or digital twin models to generate interactive commands to intervene in and regulate the physical space, physical entities, physical systems, or real systems, so as to keep the entire physical network system in good operating condition.

[0305] The embodiments of this application can be applied to the optimization of industry field networks. Industry field networks mainly refer to multi-source heterogeneous network architectures serving vertical industries, including 5G gateways or 5G base stations and below, such as Wireless Fidelity (WIFI) networks, Programmable Controllers (PLCs), short-range communication (Zigbee, new short-range communication, etc.), Passive Radio Frequency Identification (RFID), millimeter wave, Ultra Wide Band (UWB), Bluetooth, etc. In industry field networks, digital twin modeling maps physical entities such as people (network operation and maintenance personnel), machines (network elements), networks (5G networks and field networks), laws (network knowledge such as operation and maintenance manuals and operation guides), and environments (field environment) to atomic digital models in virtual space. These atomic models are then arranged and combined according to network topology and business requirements to support the intelligent, visible, manageable, and controllable requirements of industry field networks. Network visualization can be completed directly in virtual space, while network management and control require feeding back virtual space analysis and decision-making instructions to the physical space to achieve control over network physical entities (such as fault alarms, work order dispatch, network repair, and network optimization). To meet the demand for lightweight network management in industry settings and to achieve the construction of a simplified digital twin system for industry field networks, this invention first proposes a lightweight digital twin system for industry field networks, such as... Figure 4 As shown.

[0306] The industry field network digital twin system includes a physical space system (also known as physical space) and a digital twin system (also known as digital space, virtual space, twin space, or digital twin space). The perception data of the physical space (sensor acquisition or internal status reporting) is input into the digital space, where it is calculated and analyzed. The system outputs operation and maintenance commands to intervene in the physical space, thereby realizing intelligent management of the industry field network.

[0307] The physical space consists of physical entities such as humans, machines, networks, laws, and environments, while the digital space consists of digital models mapping these physical entities. Unlike the more complex five-dimensional models and intricate internal interaction logic proposed in the industry, this system architecture defines the industry field network digital twin model as a three-dimensional model, including a geometric model, an information model, and a mechanistic model. The geometric model represents the entities in the physical space and the interactions between them in two-dimensional or three-dimensional graphics in the virtual space, enabling visualization of information such as network devices, network topology, and network faults. The information model provides standardized semantic descriptions of the attributes, capabilities, interfaces, and data flows of physical entities. Through the information model, data collected and perceived from the physical space can be transmitted to the virtual space in a standardized manner for visualization and intelligent analysis. The purpose of constructing the mechanistic model is to realistically reflect the objective laws of the physical space, and on this basis, manage and control the entities in the physical space. The physical space and digital space, as well as their respective modules, will be described in detail below.

[0308] (I) Physical Space

[0309] Physical space is the space corresponding to the real world, that is, the space where physical entities and systems actually exist. The physical space of the Industry Field Network digital twin system mainly includes five elements: Person (P... P ), machine (NC) P ), network (N) P ), law (R) P ), ring (E) P As shown in equation (1), S P A system representing the organic combination and interaction of real physical entities existing in physical space (also known as a "physical system" or "real system"); ψ P It represents the operating rules and logic of a physical system, the interaction rules between physical entities, etc.

[0310] S P =ψ P (P P NC P N P R P E P (1)

[0311] "People" refers to network operation and maintenance personnel; "Machines" refers to the various network element devices in the industry field network, including: Wireless Local Area Network (WLAN), PLC, short-range communication (Zigbee, new short-range communication, etc.), passive RFID, millimeter wave, UWB, Near Field Communication (NFC), Bluetooth, and other network infrastructure such as base stations, access points, gateways, and terminals; "Network" refers to 5G network and field network (a multi-source heterogeneous network composed of network element devices in the industry field network, including WLAN, short-range communication, etc.), that is, the network composed of the connection relationships between network element devices in the field network; "Law" refers to the objective laws of the operation (including planning, construction, operation, maintenance, and optimization, referred to as "planning," "construction," "operation," "maintenance," and "optimization") of various multi-source heterogeneous networks in the industry field network, as well as network knowledge such as operation and maintenance manuals and operation guides based on long-term data accumulation and expert experience; "Environment" refers to the field environment, including field temperature, humidity, etc.

[0312] From an object-oriented perspective, that is, from the perspective of a single physical entity (mainly network element devices) in the physical space, it contains the following main information:

[0313]

[0314] 1. Topological information: geometric shape (size, shape, dimensions, etc.), location (Geographic Information System (GIS) location, etc.), attributes (inherent information of the entity itself, such as device model, manufacturing information, Medium Access Control (MAC) address, etc.), and relationships (relationships with other entities, including subordinate, adjacent, and connected relationships).

[0315] 2. Performance information: Central Processing Unit (CPU) utilization, channel utilization, etc.;

[0316] 3. Measurement (or telemetry) information: bandwidth, latency, packet loss rate, stability, signal-to-noise ratio, signal strength, etc.

[0317] 4. Functional Information: Operations or actions that an entity can perform based on instructions or configurations, such as turning the AP device on and off, resetting the AP channel and transmit power, etc.

[0318] Typically, attributes here refer to information inherent to an entity that doesn't easily change with the environment; telemetry and performance information represent measurement results or events, whose specific values ​​may change at any time. Telemetry information, in particular, is more like an event stream, a set of short-lived data that needs to be sensed and processed in real time.

[0319] (II) Digital Space

[0320] Digital space (or "virtual space") consists of digital models that map physical entities, including geometric models (M). geometry Information Model (M) information ) and mechanism model (M mechanism There are three types of models, and the models interact with each other through the information model, as shown in equation (3) below.

[0321] By summarizing the five elements of physical space—human-machine-network-law-environment—as well as the four-dimensional information of physical entities—topology, performance, telemetry, and function—into a three-dimensional model of digital space and its interaction logic, a lightweight digital twin system design was realized.

[0322]

[0323] Geometric models are used to visualize network elements, network topology, and network faults and anomalies; information models provide standardized semantic descriptions of network element attributes, capabilities, interfaces, and data flows; mechanistic models enable intelligent analysis and decision-making throughout the entire lifecycle of network planning, construction, operation, maintenance, and optimization in industry settings. Mathematical theorems, physical principles, rule sets, knowledge bases, and AI models all fall under the category of mechanistic models. The following section will provide a detailed introduction to 3D models from the perspectives of their mathematical representation, construction techniques, and specific connotations.

[0324] (1) Geometric model

[0325]

[0326] The mathematical representation of the digital twin's geometric model is shown in equation (4) above, which includes the twin's dimensions (S). i ), position (L) i ), shape (Sh i ), connection (C i Information such as [data missing] is used as input, and this information is organically combined according to certain rules and logic (i.e., the G function) to form the geometric model of the digital twin. Geometric model construction technologies include 3D visualization and GIS mapping. The specific geometric models of the virtual space in the industry field network digital twin system include: equipment geometric models, network geometric models, fault geometric models, anomaly geometric models, alarm geometric models, work order geometric models, repair plan geometric models, and network optimization plan geometric models.

[0327] (2) Information Model

[0328]

[0329] The mathematical representation of the digital twin information model is shown in equation (5) above, which includes the type (T) i ), attributes (P) i This includes device model, manufacturing information, MAC address, etc., and telemetry (Te i (Including bandwidth, latency, packet loss rate, signal-to-noise ratio, etc.) and relation (R) i This includes relationships such as subordination, proximity, and connection, as well as performance (Pe). i This includes CPU utilization, channel utilization, etc., and functions (F). i This includes accepting instructions to reconfigure itself or perform operations on downstream devices. Similarly, the above information, organically combined according to certain established rules (i.e., I-functions), forms the information model of the digital space. The construction of the information model mainly relies on the design and implementation of modeling languages, such as Microsoft DTDL, Huawei DTML, and ThingML. The information model of the virtual space of the industry field network digital twin system specifically includes: equipment information model, network information model, personnel information model, anomaly information model, fault information model, root cause information model, alarm information model, work order information model, repair plan information model, and network optimization plan information model.

[0330] (3) Mechanism Model

[0331]

[0332] The mathematical representation of the digital twin mechanism model is shown in equation (6) above. The digital twin system proposed in this invention defines the mechanism model as follows:

[0333] 1) The operating mechanism of a device or system that has a clear mechanism and can be expressed by a mathematical formula, such as the mathematical formula F=m*a that describes the physical relationship between force and mass;

[0334] 2) Based on the rule set or knowledge graph formed by domain experts in the course of their work, for example, if the signal strength detected in a WIFI network is less than -70dBm, it can be initially judged that the network may have some kind of anomaly. At this time, it is necessary to combine other indicators for further confirmation.

[0335] 3) AI mechanism models trained using machine learning or deep learning algorithms based on big data, such as deep neural networks (DNN), convolutional neural networks (CNN, which is good at handling image problems), and recurrent neural networks (RNN, which is good at handling time-series data such as speech).

[0336] The principle of constructing digital space mechanism models is to approximate the objective laws governing the operation of physical space systems as closely as possible, and to express the complex physical laws and embedded mechanisms that are difficult to observe directly in an explicit way. This allows the models to act on the digital twin in the virtual space and be organically scheduled or manipulated. Methods for constructing mechanism models include mathematical formulas, physical theorems, rule sets, knowledge graphs, and AI algorithms (including machine learning and deep learning algorithms). Specifically, the mechanism models for the virtual space of the industry field network digital twin system include: anomaly detection mechanism models, root cause localization mechanism models, and fault repair mechanism models.

[0337] The following section explains the specific process by which the constructed industry field network digital twin system achieves network management and optimization from the perspective of closed-loop control and interactive feedback between virtual and physical systems. The interaction between the virtual and physical systems here includes... Figure 2 The digital space in the model perceives data from the physical space and generates and feeds back operational instructions to the physical space. To solve the optimization problem, the virtual and physical systems are mathematically represented separately, and an objective function for optimization is constructed.

[0338] Let S be the real system in physical space. P (Hereinafter referred to as the "physical system"), its mathematical representation is as follows:

[0339]

[0340] In the above formula, These are the state of the physical system at time t, the system state transition rules, the observation rules, the observed quantities, and the control quantities.

[0341] Let S be the twin system of digital space. D (Hereinafter referred to as the "virtual system"), its mathematical representation is as follows:

[0342]

[0343] In the above formula, These are the state of the digital twin system at time t, the system state transition rules, the observation rules, the observed quantities, and the control quantities.

[0344] Based on the mathematical representation of the virtual and real systems in the digital twin system above, we will model the system from the two stages of system construction and execution to reflect the optimization and solution objectives at different stages, as well as the joint optimization process based on the closed-loop control and interactive feedback of the virtual and real systems.

[0345] (1) Digital Twin System Construction Model: Driven by the mechanisms and data of the actual physical system, and utilizing knowledge representation and learning methods, the objects, relationships, processes, and functions of complex systems are combined into a software-defined virtual digital system, such as... Figure 5 As shown.

[0346] In this model, the mathematical essence of the digital twin system problem is that at each time t, in the real system (here denoted as S) P (represented by) and virtual system (here denoted by S) D The transfer rule (F) represents t ), observation rules (H) t ) and control rules (T t Constructing a twin system under the constraints of its own changes, etc. and Make the observations of the real system and observations of virtual systems By minimizing a certain distance metric, the virtual system in digital space can be focused on the observation of the real system in physical space. Therefore, the construction process of a twin system is essentially the construction (or solution) of a solution. and The process. Due to the complexity of actual systems in physical space, and Typically unobservable, therefore, digital twin systems are needed to model complex system problems, the mathematical form of which is:

[0347]

[0348]

[0349] in and All are mechanistic models, and their construction methods can refer to the three methods for constructing mechanistic models in the previous chapter. Specifically, when there is sufficient mechanistic knowledge, models can be constructed directly based on physical laws or mathematical formulas. When mechanistic knowledge is lacking but historical data is sufficient, AI algorithms can be used for construction. For example, in the joint training of system functions... and In the process, reinforcement learning techniques based on deep networks can be used to correctly model the complex behavior of the system based on data-driven approaches, even when there is a lack of knowledge about the system's underlying mechanisms.

[0350] (2) Digital Twin System Execution Mode: The goal of this approach is to achieve the control and optimization of complex systems in physical space through virtual-physical interaction, such as... Figure 6As shown, a digital twin system and a real system are simultaneously implemented, interacting in a certain way. The virtual system is focused on the observation of the real system, and the twin model guides and manages the real system. Optimization of the virtual system leads to optimization of the real system. The twin system minimizes the observation distance measurement between the virtual and real systems. And distance measurement between the current state and the desired state of the system To solve for the control variables together with the target and (They are usually considered equal), achieving mutual feedback and driving between the virtual and real systems, as well as continuous iterative optimization; this mathematical process is the execution process. Specifically, the objective equation of this process is as follows:

[0351]

[0352]

[0353] In this mode, The system can be trained jointly or assumed to be known; the objective is to find the control input of the virtual and real systems. and Its optimization has two objectives, one of which is... in The second is the system's preset target state; The above mathematical formula represents the control quantity of the virtual and real systems through joint solution. This enables the virtual system to move towards a preset system objective. Convergence achieves the goal of "optimization"; simultaneously, it minimizes the difference between the appearances of the virtual and real systems. This allows the real system to converge to a pre-defined virtual system state, thus achieving the "guidance objective." In summary, the real system is guided to the desired optimal state by utilizing a solvable virtual system.

[0354] In this digital twin system architecture, the core of system guidance and optimization lies in finding the optimal solution set. The search for [the value]. Combined with [the desired value]. Figure 5 and Figure 6 "Virtual-real interaction" is reflected in the following two aspects: Firstly, through the control variables of the virtual system and the real system... Real-time control of the virtual and real systems to minimize the distance between the current state and the desired state of the virtual system. This achieves a predetermined system objective and optimizes the system. On the other hand, by introducing real data from the real system (i.e., system observations) into the virtual system, the observation distance metric between the real and virtual systems is minimized. To solve for the control variables of the virtual and real systems together, we need to consider the target. and To achieve mutual feedback and driving between virtual and real systems.

[0355] The following section uses the intelligent wireless radio frequency optimization of the industry field network digital twin system (taking the AP channel planning of the WIFI network - the optimal channel selection) as an example.

[0356] As a crucial component of multi-source heterogeneous industry field networks, Wi-Fi networks play a vital role in network services across various scenarios, including office buildings, large conference venues, and industrial parks. This section, building upon the industry field network digital twin system constructed in the previous section and its global optimization computing framework based on virtual-real interaction feedback, will use the AP channel planning scenario (i.e., optimal channel selection in multi-AP scenarios) of a Wi-Fi network as an example to further illustrate how the industry field network digital twin system achieves intelligent wireless radio frequency optimization.

[0357] Traditional Wi-Fi network access point (AP) channel planning relies on manual optimization and planning. This method is extremely costly, requiring multiple adjustments over several days, and the final result may worsen with each adjustment, leading to issues such as channel conflicts and interference from APs installed too close together. Figure 7 and Figure 8 As shown. Once environmental changes occur, such as the deployment of new access points (APs) or the addition of a new wall, the original plan needs to be readjusted, potentially causing widespread disruption and making the task extremely difficult. Furthermore, radio frequency channel resources are limited; in high-density scenarios or those with third-party interference, channel conflicts are unavoidable. Figure 9 and Figure 10 As shown, ensuring a good user experience in scenarios with limited resources is a challenge.

[0358] To address the aforementioned problem of unreasonable AP channel planning and achieve dynamic, accurate, efficient, globally optimized, trial-and-error-capable, and continuously learning-based WIFI network AP channel planning, this invention constructs an intelligent wireless radio frequency optimization system based on digital twins. For example... Figure 11As shown in the diagram (the black and gray solid lines represent the process), after the physical AP devices are deployed, the digital space, on one hand, senses the raw data from the physical devices and generates the AP topology (including the spatial location and distance of each AP), AP load status, and channel status based on the geometric and information models of the twin. This status information is input into the "channel conflict detection" mechanism model, which determines which APs may have channel conflicts based on rules such as whether the distance between APs with the same channel is too close. On the other hand, numerical information of multiple network KPI indicators (including signal-to-noise ratio (SNR), signal-to-interference-plus-noise ratio (SINR), received signal strength (RSRP), throughput, packet loss, latency, and bandwidth) obtained from the raw data of the physical devices is input into the "network anomaly detection" mechanism model. From the perspective of network performance and user experience, the network conflict situation is reconstructed in an intuitive scoring manner. Finally, the "channel optimization strategy" mechanism model outputs high-precision network optimization suggestions. Specifically, the outputs of the "channel conflict detection" mechanism model and the "network anomaly detection" mechanism model are put into the "channel optimization strategy" mechanism model group, which mainly includes three mechanism models. The "abnormal AP location" mechanism model finds the root cause of network conflict, that is, locates the AP with unreasonable channel planning. Combined with the load of the AP with channel conflict, the "channel replanning" mechanism model gives a variety of optional channel optimization schemes. Finally, the "RF quality assessment" mechanism model is used to predict the various RF quality related indicators (i.e. network performance indicators) under different optimization schemes, and calculates the user experience score. The scheme with the highest score is the optimal channel planning scheme that can reasonably allocate air interface resources.

[0359] Based on the above interaction and control process, the physical space real system mathematical representation of the intelligent wireless radio frequency optimization system (i.e., WiFi network AP channel planning system) based on digital twins can be given as follows:

[0360]

[0361] here, It is the state of the WiFi network AP channel planning system at time t, including AP topology state, AP load state, AP channel state, network KPI state (or wireless radio frequency indicator state, including SNR, signal to interference plus noise ratio (SINR), reference signal receiving power (RSRP), throughput, packet loss, delay, etc.) and other multi-dimensional data. It is the control quantity applied to the actual AP device of the physical space WIFI network at time t. The corresponding system state transition rules are complex and not fully known under limited resources. This represents the system's observation at time t. In this case, the observations are the channel conflict status of each AP and the network performance anomaly status. The corresponding system observation rule is complex and not fully known due to numerous factors that cause channel conflicts, degraded radio frequency quality, and affect user experience. Therefore, it is necessary to construct a twin model of it in the digital space. By leveraging the construction and solution methods of the mechanistic model in the digital twin system, and using data-driven and continuous learning and training methods, the solution can be obtained to achieve optimal control of the real system in the physical space and achieve the desired control objectives and effects.

[0362] Given the complex system characteristics of the real-world Wi-Fi network AP channel planning system, a corresponding digital twin system for Wi-Fi network AP channel planning is constructed using the digital twin system method. Its mathematical representation is as follows:

[0363]

[0364] in, It is the state of the digital twin system at time t, mainly composed of AP topology state, AP load state, AP channel state, network KPI state (or wireless radio frequency indicator state, including multi-dimensional data such as SNR, SINR, RSRP, throughput, packet loss, and latency). It is the control quantity applied to the digital twin system at time t, in this case... Distribute control variables to each AP channel. This refers to the corresponding system state transition rules. This represents the system observation value of the digital twin system at time t, which is taken as the channel conflict status of each AP in the virtual system and the abnormal network performance status. This corresponds to the system observation rules.

[0365] Based on the digital representations of the real and virtual systems in Wi-Fi network AP channel planning and the control objectives in this scenario, the general objective equation for solving the virtual system in this digital twin system can be constructed as follows:

[0366]

[0367] in, This is a radio frequency quality evaluation function (or a network performance evaluation function).

[0368] In the digital twin system approach, the core of system guidance, optimization, and control lies in finding the optimal solution set. The method for solving the above mathematical quantities in the research scenario of this case is as follows.

[0369] 1) System state transition rules Depend on Figure 11 The overall control process of the system and the interaction mode between the virtual and physical systems can be known (i.e.) Figure 10 (The process is represented by black and gray solid lines). The state transition rules in the virtual system mainly involve changes in AP topology, AP load, and channel conditions. Therefore, by constructing corresponding transition rule models, the overall system state transition rules can be solved. The AP topology change model can be obtained based on geometric and information models, while AP load and channel variables objectively exist; therefore, the system state transition rules... Yes, it can be obtained.

[0370] 2) System observation rules In this case, the virtual system observation rules include a "channel collision detection" mechanism model. "Network KPI Prediction" Mechanism Model and the mechanism model of "network anomaly detection" The observations include channel conflict status for each AP, predicted values ​​of multiple network KPIs, and network performance anomalies for each AP.

[0371] For the mechanism model of "channel collision detection" The following rules can be used to build the system (the output includes the load information for each AP):

[0372] if c i ==c j &&d ij <d th

[0373] then output = 1

[0374] Where c i d represents the channel used by the i-th AP. ij d represents the spatial distance between the i-th AP and the j-th AP. th This represents the distance threshold at which signal interference occurs between two access points (APs). This rule means that if the spatial distance between two APs on the same channel is less than the set threshold, signal interference is considered to have occurred, i.e., a channel conflict has been determined. In this model d... th Neural networks (AI algorithms) can be trained, optimized, and dynamically adjusted based on historical data to make more accurate judgments that are more in line with reality.

[0375] For the mechanism model of "network KPI prediction" A data-driven approach can be adopted, using AI algorithms to build neural networks (such as...). Figure 12 The representation and solution are performed as shown in the figure. Where f NN The function expression corresponding to the constructed neural network, These are AP status parameters (including AP topology status, AP load, and channel status, etc.). To predict multiple network KPI values ​​(including SNR, SINR, RSRP, etc.), the neural network takes the time-series AP state parameters as input and its network KPIs as output. For each AP to be studied, a multi-layer feedforward network is trained using the backpropagation algorithm. w and b are the weight vector and bias vector, respectively, which are parameters that need to be iteratively solved during the training of the neural network. The solution process usually uses gradient descent to backpropagate the error until the error converges to a preset threshold range. This ensures that the neural network model built based on real historical data is infinitely close to the actual but unobservable and unobtainable mechanisms in the real physical space. The more accurate the mechanism model built in the digital space, the more realistically it can reflect the actual situation in the physical space.

[0376] For the mechanism model of "network anomaly detection" Its function is to comprehensively judge and give the probability of network performance anomalies based on the input values ​​of multiple network KPI indicators for each AP. The construction of this mechanism model can be based on a rule set and knowledge base summarized by expert experience, such as setting thresholds for different network KPI parameters for anomaly detection, with the threshold selection based on expert experience; alternatively, it can use data-driven AI algorithms for threshold model training and prediction, and finally adjust the sensitivity based on experience, such as VAE-based anomaly detection algorithms (e.g.,...). Figure 13 (as shown in the image) etc.

[0377] 3) Control variables Based on the modeling and solving of the above system state transition rules and system observation rules, the objective equation for control and optimization in this case during the system execution phase can be further transformed into:

[0378]

[0379] Subject to,

[0380]

[0381]

[0382]

[0383]

[0384]

[0385] In this case, A baseline control variable is set for each AP channel. Therefore, in the actual execution phase, only the control variables in the virtual system need to be solved, and then directly applied to the real system. There is no need to perform conversion calculations between virtual and real control variables.

[0386] The channel optimization strategy mainly includes three parts: abnormal AP location, channel replanning, and radio frequency quality assessment. Abnormal APs are located by combining channel collision detection results and network anomaly detection results. The channel of the abnormal AP is then replanned by comprehensively considering factors such as the channel, distance, and load of neighboring APs, and a "radio frequency quality assessment" mechanism model is used. The radio frequency quality (or network performance) after channel replanning is evaluated to ensure... The highest-scoring channel distributed control output This is what we are looking for. Among them, the "RF quality assessment" mechanism model... The construction can employ a rule base or knowledge graph based on expert experience, or an AI algorithm built and optimized based on historical data. Furthermore, after each execution, the status of each AP, the network KPI status, and the corresponding channel conflicts and network anomalies are recorded as a mechanism model for subsequent "RF quality assessment." The continuous learning and optimization of training samples makes the model's predictions more and more accurate with increasing usage frequency, and the channel optimization strategy becomes more and more reasonable, so that the whole system becomes smarter with use.

[0387] Additionally, it should be noted that, Figure 11 The gray dashed line represents the simulation execution and verification process. Based on the "network KPI prediction" mechanism model, anomaly detection is performed according to the predicted network KPI values. Combined with the output of the "channel conflict detection" mechanism model, channel optimization is carried out. The control quantity generated by the optimization strategy is applied to the virtual AP device in the digital space to obtain the state after channel planning. This process is repeated until there are no channel conflicts or network performance anomalies, and the radio frequency quality has a high score. Finally, the optimal channel planning scheme is deployed on real AP devices in the physical space (that is, the generated channel control quantity is applied to the physical AP device). Through simulation execution and verification in the digital space, low-cost trial and error and label data generation are achieved (which is beneficial for the continuous training and optimization of the mechanism model built based on data-driven AI algorithms).

[0388] Finally, based on the above construction and solution process, let's summarize the main features and advantages of the intelligent wireless RF tuning system based on digital twins:

[0389] 1) Real-time simulation and interactive feedback: Based on the AP distribution topology, the distance between APs on the same channel, and AP load, combined with network KPI status, it analyzes network interference and anomalies, locates abnormal APs, and implements channel replanning for proactive avoidance and adjustment. The simulation effect is evaluated through RF quality assessment scores, eliminating the need for administrators to repeatedly adjust configurations and verify on-site. This significantly improves operational efficiency and enhances the user's wireless network experience.

[0390] 2) Dynamic real-time global optimization: Prioritize the allocation of "clean" channels for high-load APs to improve the overall network throughput and maximize the utilization of spectrum resources. For example... Figure 9 As shown, in an environment where only channels 6 and 11 are available, traditional channel allocation methods may result in AP2 and AP3, which are under heavy load, sharing the same channel, leading to severe conflicts and significantly impacting user experience. Figure 10 As shown, priority is given to ensuring that the channels of the two high-load APs, AP2 and AP3, are staggered to avoid a large number of collisions. Some collision with the lightly loaded AP1 is allowed to maximize the user experience under high-load APs (AP1 has fewer users under light load, resulting in less actual interference). Furthermore, since the load status of each AP changes dynamically in real time, channel planning should also be real-time to guarantee the user experience under high-load APs. Clearly, traditional channel planning methods struggle to meet this requirement.

[0391] 3) Low-cost trial and error and validation: Figure 11 The gray dashed line represents the simulation execution process. When the network is not busy (with idle computing resources and without affecting the current task), the generated channel optimization strategy is simulated and deployed in virtual space. The current network KPI status is predicted using a "network KPI prediction" mechanism model, and network anomaly detection is performed based on the prediction results. Simultaneously, an "AP anomaly detection" mechanism model is used to evaluate the abnormal states of APs in the current network. If anomalies still exist, channel planning is re-performed, and the corresponding RF quality assessment score is calculated. The channel optimization strategy with the highest score is selected and simulated and deployed on the virtual AP twin. This process is repeated until there are no abnormal APs. Finally, the control quantity generated by the channel optimization strategy is applied to the physical devices in the real space. The objective equation for system control and optimization in this process can be expressed as:

[0392]

[0393] Subject to,

[0394]

[0395]

[0396]

[0397]

[0398]

[0399] 4) Intelligent Analysis and Decision Making: For real systems whose mechanisms cannot be accurately obtained or described, AI algorithms are introduced and trained based on data to build neural network models. For example, models such as "network KPI prediction", "network anomaly detection", and "radio frequency quality assessment" can all be built using AI algorithms to achieve intelligent analysis and decision making.

[0400] 5) Continuous learning and optimization: Store historical data after each system execution (including actual data after execution by real devices in the physical space and simulation data after execution by virtual devices in the digital space). This process is called accumulating labeled data. Based on historical data, the mechanism model in the virtual system is continuously trained and optimized. By continuously learning from actual data and simulation data, a more accurate mechanism model is obtained, making the operation of the digital space virtual system more consistent with the operation law of the real physical space system. This makes the virtual system infinitely close to the actual system, while obtaining more intelligent analysis and decision-making, achieving the desired control goals and effects, and making the entire system smarter with use.

[0401] See Figure 14 This application provides a digital twin system construction apparatus, the apparatus 1400 comprising:

[0402] The first acquisition module 1401 is used to acquire first information from the first object. The first information includes: data collected and / or sensed by one or more sensors in the first object, and / or data reported by an entity in the first object. The first object is: physical space or physical entity or physical system or real system.

[0403] The first processing module 1402 is configured to input the first information to a second object, the second object including: a virtual entity and / or a digital twin model, the second object being: a digital space or virtual entity or twin system or twin body or digital system or digital twin system or virtual system; obtain a first instruction through the digital twin model and the first information; input the first instruction to the first object and / or the second object, and control and / or optimize the first object and / or the second object through the first instruction.

[0404] In one embodiment of this application, the data includes one or more of the following:

[0405] Sensor data;

[0406] Status data;

[0407] Event data;

[0408] Environmental data.

[0409] In one embodiment of this application, the first instruction includes one or more of the following:

[0410] Generate alarms;

[0411] Trigger analysis and / or decision-making and / or control and / or optimization of the corresponding workflow;

[0412] Driver fault repair;

[0413] Drive predictive analytics and / or maintenance.

[0414] In one embodiment of this application, the apparatus further includes:

[0415] The second acquisition module is used to acquire the second information of the first object;

[0416] The second processing module is used to construct a virtual entity and / or digital twin model of the second object, and / or interaction logic between the various digital twin models, based on the second information.

[0417] The second information includes one or more of the following:

[0418] Information about the elements in the physical space;

[0419] Information about physical entities in the physical space;

[0420] The information of the elements includes one or more of the following:

[0421] Information about maintenance personnel;

[0422] Information on physical devices and / or systems;

[0423] Information about the network formed by the connections between physical entities;

[0424] Information on the operation of physical equipment and / or systems;

[0425] Knowledge of physical equipment and / or system operation and maintenance;

[0426] Physical equipment and / or system operation instructions;

[0427] Information about the site environment;

[0428] The information of the physical entity includes one or more of the following:

[0429] The topological information of the physical entity;

[0430] The performance information of the physical entity;

[0431] Measurement information of the physical entity;

[0432] The functional information of the physical entity.

[0433] In one embodiment of this application, the digital twin model includes one or more of the following:

[0434] A geometric model, which is used to visualize one or more of the following: physical entities, topological relationships, operating states, analysis processes and / or results, decision-making processes and / or results, control processes and / or instructions, optimization processes and / or effects.

[0435] An information model is used to provide a standardized semantic description of one or more of the attributes, capabilities, interfaces, and data flows of a physical entity.

[0436] The mechanism model has the following functions: knowledge mining of the first object; mining of the operational rules of the first object; analysis, decision-making, control and optimization of the entire life cycle of the first object.

[0437] In one embodiment of this application, the type of the mechanism model includes one or more of the following:

[0438] The operating mechanism of a device or system that has a clear principle and can be expressed by a numerical formula;

[0439] A set of rules, a rule engine, a knowledge base, a knowledge engine, or a knowledge graph formed based on experience;

[0440] Mechanism models generated based on big data and / or AI algorithms.

[0441] In one embodiment of this application, the first processing module 1403 is further configured to:

[0442] Based on the first information, one or more of the channel conflict detection mechanism model, the network KPI prediction mechanism model, and the network anomaly detection mechanism model, as well as the channel optimization strategy mechanism model, the first instruction is determined.

[0443] In one embodiment of this application, the first processing module 1403 is further configured to:

[0444] The first information is input into one or more of the channel conflict detection mechanism model, the network KPI prediction mechanism model, and the network anomaly detection mechanism model;

[0445] The output results of one or more of the channel conflict detection mechanism model, network KPI prediction mechanism model, and network anomaly detection mechanism model are input into the channel optimization strategy mechanism model, which includes: anomaly AP location mechanism model, channel replanning mechanism model, and radio frequency quality assessment mechanism model.

[0446] In one embodiment of this application, the apparatus further includes:

[0447] The third processing module is used to determine the AP with channel conflict through the abnormal AP location mechanism model; determine one or more channel optimization schemes based on the load of the AP with channel conflict through the channel replanning mechanism model; evaluate the radio frequency quality of one or more channel optimization schemes through the radio frequency quality evaluation mechanism model to obtain the evaluation result; and determine the first instruction based on the evaluation result, wherein the first instruction is a control instruction for the optimal channel optimization scheme among the one or more channel optimization schemes.

[0448] In one embodiment of this application, the apparatus further includes:

[0449] The fourth processing module is used to apply the control quantity indicated by the first instruction to the second object; obtain the optimized state of the second object; if the state of the second object indicates that the second object is operating normally, then the step of inputting the first instruction to the first object is executed.

[0450] In one embodiment of this application, the state includes one or more of the following:

[0451] AP topology status;

[0452] AP load status;

[0453] AP channel status;

[0454] Network KPI status.

[0455] In one embodiment of this application, the apparatus further includes:

[0456] The fifth processing module is used to execute the currently determined optimal channel optimization scheme on the first object, record one or more of the AP status, network KPI status, channel conflict information, and network anomaly information; and use one or more of the recorded AP status, network KPI status, channel conflict information, and network anomaly information as training samples to train the radio frequency quality assessment mechanism model.

[0457] The apparatus provided in this application embodiment can achieve... Figure 1The various processes implemented in the method embodiments shown achieve the same technical effects, and will not be described again here to avoid repetition.

[0458] See Figure 15 This application provides a digital twin system modeling apparatus, the apparatus 1500 comprising:

[0459] The sixth processing module 1501 is used to create mathematical representations of the first object and the second object respectively. The first object is: physical space or physical entity or physical system or real system, and the second object is: digital space or virtual entity or twin system or twin body or digital system or digital twin system or virtual system.

[0460] The seventh processing module 1502 is used to create a control optimization model for the system construction phase and / or system execution phase based on the mathematical representations of the first object and the second object.

[0461] In one embodiment of this application, the mathematical representation of the first object is:

[0462]

[0463] in, These are: the state of the first object at time t, the system state transition rules, the observation rules, the observed quantities, and the control quantities. These are the state and control variables of the first object at time t-1, respectively.

[0464] In one embodiment of this application, the mathematical representation of the second object is:

[0465]

[0466] in, These respectively represent: the state mechanism model of the second object at time t, the system state transition rule mechanism model, the observation rule mechanism model, the observation mechanism model, and the control variable mechanism model. These respectively represent the state mechanism model and the control quantity mechanism model of the second object at time t-1.

[0467] In one embodiment of this application, the apparatus further includes:

[0468] The eighth processing module is used to construct a control optimization model for the system construction phase based on the mathematical representations of the first object and the second object and the first formula.

[0469] The first formula is:

[0470]

[0471] in, This indicates minimizing the observed distance measurement between the first object and the second object. This represents the distance measurement between the state of the second object at time t and its state at time t+T.

[0472] In one embodiment of this application, the apparatus further includes:

[0473] The ninth processing module is used to construct a control optimization model for the system execution phase based on the mathematical representations and second formulas of the first and second objects.

[0474] The second formula is:

[0475]

[0476] in, This represents the control pair between the first object and the second object;

[0477] This represents minimizing the observation distance metric between the first and second objects. This represents the distance metric between the state of the second object at time t and its state at time t+T.

[0478] In one embodiment of this application, the apparatus further includes:

[0479] The tenth processing module is used to... The first object and the second object are controlled separately to minimize the distance metric between the state of the second object at time t and the state at time t+T.

[0480] The apparatus provided in this application embodiment can achieve... Figure 2 The various processes implemented in the method embodiments shown achieve the same technical effects, and will not be described again here to avoid repetition.

[0481] See Figure 16 This application provides a network optimization device, the device 1600 including:

[0482] The eleventh processing module 1601 is used to create mathematical representations of a first object and a second object respectively. The first object is: physical space or physical entity or physical system or real system, and the second object is: digital space or virtual entity or twin system or twin body or digital system or digital twin system or virtual system.

[0483] The twelfth processing module 1602 is used to create a control optimization model for the system construction phase and / or execution phase and / or simulation execution and verification process based on the mathematical representation of the first object and the second object and the network optimization objective.

[0484] The thirteenth processing module 1603 is used to determine the control quantity for optimizing the network based on the control optimization model of the system construction phase and / or execution phase and / or simulation execution and verification process.

[0485] In one embodiment of this application, the thirteenth processing module is further configured to:

[0486] Based on the control optimization model from the system construction phase, the control variables used to optimize the network are determined to be the optimal solution set.

[0487] The control optimization model for the system construction phase is as follows:

[0488] in, For network performance evaluation functions, These respectively represent: the system state transition rule mechanism model, the observation rule mechanism model, the observed quantity, and the control quantity of the second object at time t. It is the control quantity of the first object at time t. These are the observations of the first object at time t. This represents minimizing the observation distance metric between the first and second objects.

[0489] In one embodiment of this application, the thirteenth processing module is further configured to:

[0490] Based on the control optimization model for the system execution phase, the control variables used to optimize the network are determined to be the optimal solution set.

[0491] The control optimization model for the system execution phase is as follows:

[0492]

[0493]

[0494]

[0495]

[0496]

[0497]

[0498] in, The AP topology, load, and channel state of the digital twin system at time t; The network KPI status of the digital twin system at time t; These are system observations of AP channel conflict in the virtual system and system observations of network performance anomalies in the virtual system, respectively. The digital twin system corresponds to time t. System transfer rules; These are the "channel conflict detection" mechanism model and the "network anomaly detection" mechanism model, respectively. These represent all system observations of the digital twin system at times t-1 and t, including channel conflicts and network performance anomalies for each AP.

[0499] In one embodiment of this application, determining the control variables for optimizing the network based on the control optimization model of the system simulation execution and verification process includes:

[0500] Based on the control optimization model of the system simulation and verification process, the control quantity used to optimize the network is determined to be the optimal solution.

[0501] The control optimization model for the system simulation execution and verification process is as follows:

[0502]

[0503] Subject to,

[0504]

[0505]

[0506]

[0507]

[0508]

[0509] in, The AP topology, load, and channel state of the digital twin system at time t; These are, respectively, system observations of AP channel conflict in the virtual system, system observations of network KPI prediction in the virtual system, and system observations of network performance anomalies in the virtual system; For the digital twin system at time t, corresponding to System transfer rules; These are the mechanism models for "channel conflict detection", "network KPI prediction", and "network anomaly detection". These represent all system observations of the digital twin system at times t-1 and t, including the channel conflict status of each AP, network KPI prediction status, and network performance anomaly status.

[0510] In one embodiment of this application, the Includes one or more of the following:

[0511] AP topology change rules;

[0512] AP load variation rules;

[0513] AP channel variation rules;

[0514] Network KPI status change rules;

[0515] or,

[0516] The Includes one or more of the following:

[0517] Channel collision detection mechanism model;

[0518] Network KPI prediction mechanism model;

[0519] Network anomaly detection mechanism model.

[0520] The apparatus provided in this application embodiment can achieve... Figure 3 The various processes implemented in the method embodiments shown achieve the same technical effects, and will not be described again here to avoid repetition.

[0521] like Figure 17 As shown, this application embodiment also provides a communication device 1700, including a processor 1701, a memory 1702, and a program or instructions stored in the memory 1702 and executable on the processor 1701. When the program or instructions are executed by the processor 1701, they implement the above-mentioned... Figure 1 or Figure 2 or Figure 3 The various processes of the illustrated method embodiments can achieve the same technical effect. To avoid repetition, they will not be described again here.

[0522] This application embodiment also provides a readable storage medium storing a program or instructions that, when executed by a processor, implement the above-described functionality. Figure 1 or Figure 2 or Figure 3 The various processes of the method embodiments shown can achieve the same technical effect, and will not be described again here to avoid repetition.

[0523] The processor mentioned above is the processor in the terminal described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0524] The steps of the methods or algorithms described in this application can be implemented in hardware or by executing software instructions on a processor. The software instructions can consist of corresponding software modules, which can be stored in RAM, flash memory, ROM, EPROM, EEPROM, registers, hard disk, portable hard disk, read-only optical disk, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and storage medium can be housed in an ASIC. Alternatively, the ASIC can be housed in a core network interface device. Of course, the processor and storage medium can also exist as discrete components in the core network interface device.

[0525] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in this application can be implemented using hardware, software, firmware, or any combination thereof. When implemented in software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transfer of a computer program from one place to another. Storage media can be any available medium accessible to a general-purpose or special-purpose computer.

[0526] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this application. It should be understood that the above description is only a specific embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of this application should be included within the scope of protection of this application.

[0527] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, embodiments of this application can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of this application can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0528] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0529] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0530] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0531] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims of this application and their equivalents, this application also intends to include these modifications and variations.

Claims

1. A method for constructing a digital twin system, characterized in that, include: First information is obtained from a first object, the first information including: data collected and / or sensed by one or more sensors in the first object, and / or data reported by a physical entity in the first object, the first object being: a physical entity or a physical system or a real system; the physical system or real system includes the physical entity, the physical entity being a network element device; The first information is input into a second object, which includes: a virtual entity and a digital twin model; The first instruction is obtained through the digital twin model and the first information; The first instruction is input to the first object and / or the second object, and the first object and / or the second object are controlled and / or optimized through the first instruction; The step of obtaining the first instruction through the digital twin model and the first information includes: Based on the first information, one or more of the following: channel conflict detection mechanism model, network key performance indicator (KPI) prediction mechanism model, and network anomaly detection mechanism model, as well as the channel optimization strategy mechanism model, the first instruction is determined. The method further includes: The first information is input into one or more of the channel conflict detection mechanism model, the network KPI prediction mechanism model, and the network anomaly detection mechanism model; The output results of one or more of the channel conflict detection mechanism model, network KPI prediction mechanism model, and network anomaly detection mechanism model are input into the channel optimization strategy mechanism model. The channel optimization strategy mechanism model includes: anomaly access point (AP) location mechanism model, channel replanning mechanism model, and radio frequency quality assessment mechanism model. The method further includes: The abnormal AP location mechanism model is used to identify APs with channel conflicts. Based on the load of the APs experiencing channel conflicts, one or more channel optimization schemes are determined through a channel replanning mechanism model. The radio frequency quality of one or more channel optimization schemes is evaluated using a radio frequency quality assessment mechanism model, and the evaluation results are obtained. Based on the evaluation results, the first instruction is determined, which is a control instruction for the optimal channel optimization scheme among the one or more channel optimization schemes.

2. The method according to claim 1, characterized in that, The data includes one or more of the following: Sensor data; Status data; Event data; Environmental data.

3. The method according to claim 1, characterized in that, The first instruction includes one or more of the following: Generate alarms; Trigger analysis and / or decision-making and / or control and / or optimization of the corresponding workflow; Driver fault repair; Drive predictive analytics and / or maintenance.

4. The method according to claim 1, characterized in that, The method further includes: Obtain the second information of the first object; Based on the second information, construct the virtual entity and digital twin model of the second object, and / or the interaction logic between the various digital twin models; The second information includes one or more of the following: Information about each element in the first object; Information about each entity in the first object; The information of the elements includes one or more of the following: Information about maintenance personnel; Information on physical devices and / or systems; Information about the network formed by the connections between physical entities; Information on the operation of physical equipment and / or systems; Knowledge of physical equipment and / or system operation and maintenance; Physical equipment and / or system operation instructions; Information about the site environment; The information of the entity includes one or more of the following: Topological information; Performance information; Measurement information; Functional information.

5. The method according to claim 1 or 4, characterized in that, The digital twin model includes one or more of the following: A geometric model, which is used to visualize one or more of the following: a first object, topological relationship, operating state, analysis process and / or result, decision process and / or result, control process and / or instruction, optimization process and / or effect; An information model is used to provide a normalized semantic description of one or more of the attributes, capabilities, interfaces, and data flows of a first object. The mechanism model has the following functions: knowledge mining of the first object; mining of the operational rules of the first object; analysis, decision-making, control and optimization of the entire life cycle of the first object.

6. The method according to claim 5, characterized in that, The types of the mechanistic models include one or more of the following: The operating mechanism of a device or system that has a clear principle and can be expressed by a numerical formula; A set of rules, a rule engine, a knowledge base, a knowledge engine, or a knowledge graph formed based on experience; Mechanism models generated based on big data and / or artificial intelligence (AI) algorithms.

7. The method according to claim 1, characterized in that, Prior to the step of inputting the first instruction to the first object, the method further includes: Apply the control quantity indicated by the first instruction to the second object; Obtain the optimized state of the second object; If the status of the second object indicates that the second object is operating normally, then the step of inputting the first instruction into the first object is executed.

8. The method according to claim 7, characterized in that, The state includes one or more of the following: AP topology status; AP load status; AP channel status; Network KPI status.

9. The method according to claim 7, characterized in that, The method further includes: The optimal channel optimization scheme currently determined is executed on the first object, and one or more of the following are recorded: AP status, network KPI status, channel conflict information, and network anomaly information. The radio frequency quality assessment mechanism model is trained by using one or more of the recorded AP status, network KPI status, channel conflict information, and network anomaly information as training samples.

10. A method for constructing a digital twin system, characterized in that, include: Mathematical representations of a first object and a second object are created respectively. The first object is a physical entity, physical system, or real system, and the second object is a digital twin system. The physical system or real system includes the physical entity, which is a network element device. Based on the mathematical representations of the first object and the second object and the first formula, a control optimization model for the system construction phase is constructed; the control optimization model is used to determine the control quantities used to optimize the network, and the control quantities are the control quantities acting on network element devices; The first formula is: in, The second object has a system state transition rule mechanism model, an observation rule mechanism model, and a control rule mechanism model at time t. The system state transition rule mechanism model includes rules for changes in AP topology, AP load, AP channel, and / or network KPI state. The observation rule mechanism model includes a channel conflict detection mechanism model, a network KPI prediction mechanism model, and / or a network anomaly detection mechanism model. The control rule mechanism model is used to determine network performance problems based on network KPI prediction values, locate the root cause of network performance problems by combining the channel conflict situation of each AP, provide channel optimization suggestions, and generate channel control quantities acting on APs. This is used to represent the optimal solution set of the control optimization model in the system construction stage obtained by the first formula, that is, the parameter set of the optimal solution of the system state transition rule mechanism model, the optimal solution of the observation rule mechanism model, and the optimal solution of the control rule mechanism model of the second object at time t; This represents the optimal solution of the system state transition rule mechanism model for the second object at time t. This represents the optimal solution of the observation rule mechanism model for the second object at time t. This represents the optimal solution of the control rule mechanism model for the second object at time t; This indicates the measurement of the observed distance between the first object and the second object. This represents the set of observations of the first object from time 1 to time t. This represents the set of observable mechanism models for the second object from time 1 to time t; This represents the distance measurement between the current state and the desired state of the second object. This represents the set of state mechanism models of the second object from time t to t+T, i.e., the current state of the second object; This represents the set of expected values ​​of the state mechanism model of the second object from time t to t+T, i.e., the preset target state of the second object.

11. The method according to claim 10, characterized in that, The mathematical representation of the first object is: in, , , , , These are: the state of the first object at time t, the system state transition rules, the observation rules, the observed quantities, and the control quantities. These are the state and control variables of the first object at time t-1, respectively. The function represents the system state transition rule. This represents the observation rule function.

12. The method according to claim 10, characterized in that, The mathematical representation of the second object is: in, , , , , These respectively represent: the state mechanism model of the second object at time t, the system state transition rule mechanism model, the observation rule mechanism model, the observation mechanism model, and the control variable mechanism model. These respectively represent: the state mechanism model and the control quantity mechanism model of the second object at time t-1. This represents the system's state transition rule mechanism model. () represents the observation rule mechanism model.

13. The method according to claim 10, characterized in that, The method further includes: Based on the mathematical representations and second formulas of the first and second objects, a control optimization model for the system execution phase is constructed. The second formula is: in, This represents the control pair between the first object and the second object; This represents the control mechanism model of the second object at time t. This represents the optimal solution of the control mechanism model for the second object at time t. This represents the control quantity of the first object at time t. This represents the optimal solution of the control quantity for the first object at time t; This indicates the measurement of the observed distance between the first object and the second object. This represents the set of observations of the first object from time 1 to time t. This represents the set of observable mechanism models for the second object from time 1 to time t; This represents the distance measurement between the current state and the desired state of the second object. This represents the set of state mechanism models of the second object from time t to t+T, i.e., the current state of the second object; This represents the set of expected values ​​of the state mechanism model of the second object from time t to t+T, i.e., the preset target state of the second object.

14. The method according to claim 13, characterized in that, The method further includes: according to The first object and the second object are controlled respectively to minimize the distance metric between the state of the second object at time t and the state at time t+T.

15. A network optimization method, characterized in that, The method includes: Mathematical representations of a first object and a second object are created respectively. The first object is a physical entity, physical system, or real system, which includes the physical entity, which is a network element device. The second object is a digital twin system. Based on the mathematical representations of the first and second objects and the network optimization objectives, create a control optimization model for the system construction phase and / or execution phase and / or simulation execution and verification process; Based on the control optimization model of the system construction phase and / or execution phase and / or simulation execution and verification process, the control quantities used to optimize the network are determined, and the control quantities are the control quantities acting on network element devices; The step of determining the control variables for optimizing the network based on the control optimization model from the system construction phase includes: Based on the control optimization model from the system construction phase, the control variables used to optimize the network are determined to be the optimal solution set. ; The control optimization model for the system construction phase is as follows: );in, For network performance evaluation functions, , , , These respectively represent: the system state transition rule mechanism model, observation rule mechanism model, observation mechanism model, and control quantity mechanism model of the second object at time t; wherein, the system state transition rule mechanism model includes AP topology change rules, AP load change rules, AP channel change rules, and / or network KPI state change rules; the observation rule mechanism model includes channel conflict detection mechanism model, network KPI prediction mechanism model, and / or network anomaly detection mechanism model; the observation values ​​corresponding to the observation mechanism model include channel conflict status of each AP, network KPI prediction values, and / or network performance anomaly status under each AP; and the control quantity mechanism model includes channel optimization strategy mechanism model. and Let represent the control quantity and the optimal solution of the control quantity for the first object at time t, respectively. These are the observations of the first object at time t. This indicates the measurement of the observed distance between the first object and the second object; These represent the optimal solutions of the system state transition rule mechanism model, the observation rule mechanism model, and the control variable mechanism model for the second object at time t, respectively.

16. The method according to claim 15, characterized in that, The step of determining the control variables for network optimization based on the control optimization model of the system execution phase includes: Based on the control optimization model for the system execution phase, the control variables used to optimize the network are determined to be the optimal solution set. ; The control optimization model for the system execution phase is as follows: in, and These represent the control mechanism models of the second object at times t and t-1, respectively. and These represent the optimal solutions of the control mechanism model for the second object at times t and t-1, respectively. and Let represent the control quantity and the optimal solution of the control quantity for the first object at time t, respectively. and Let represent the network performance evaluation functions for the second object at times t and t-1, respectively; and These represent the AP topology, load, and channel status of the digital twin system at times t and t-1, respectively. and These represent the network KPI states of the digital twin system at times t and t-1, respectively. and These are system observations of AP channel conflict in the virtual system and system observations of network performance anomalies in the virtual system, respectively. and The digital twin system corresponds to time t. , System transfer rules; and These are the "channel conflict detection" mechanism model and the "network anomaly detection" mechanism model, respectively. These represent all system observations of the digital twin system at times t-1 and t, including channel conflicts and network performance anomalies for each AP.

17. The method according to claim 15, characterized in that, The step of determining the control variables for optimizing the network based on the control optimization model of the system simulation execution and verification process includes: Based on the control optimization model of the system simulation and verification process, the control quantity used to optimize the network is determined to be the optimal solution. ; The control optimization model for the system simulation execution and verification process is as follows: Subject to the following conditions: in, and Let represent the control input and the optimal solution of the control input used by the second object to optimize the network at time t, respectively. and Let represent the control input and the optimal solution of the control input used by the second object to optimize the network at time t-1, respectively. and Let represent the network performance evaluation functions for the second object at times t and t-1, respectively; and These represent the AP topology, load, and channel status of the digital twin system at times t and t-1, respectively. , and These are, respectively, system observations of AP channel conflict in the virtual system, system observations of network KPI prediction in the virtual system, and system observations of network performance anomalies in the virtual system; For the digital twin system at time t, corresponding to System transfer rules; , and These are the mechanism models for "channel conflict detection", "network KPI prediction", and "network anomaly detection". and These represent all system observations of the digital twin system at times t-1 and t, including the channel conflict status of each AP, network KPI prediction status, and network performance anomaly status.

18. A digital twin system construction apparatus, characterized in that, include: The first acquisition module is used to acquire first information from a first object, the first information including: data collected and / or sensed by one or more sensors in the first object, and / or data reported by a physical entity in the first object, the first object being: a physical entity or a physical system or a real system; the physical system or real system includes the physical entity, the physical entity being a network element device; A first processing module is configured to input the first information into a second object, the second object including a virtual entity and a digital twin model; obtain a first instruction through the digital twin model and the first information; input the first instruction into the first object and / or the second object, and control and / or optimize the first object and / or the second object through the first instruction; The process of obtaining the first instruction through the digital twin model and the first information includes: Based on the first information, one or more of the following: channel conflict detection mechanism model, network key performance indicator (KPI) prediction mechanism model, and network anomaly detection mechanism model, as well as the channel optimization strategy mechanism model, the first instruction is determined. The first processing module is further configured to: input the first information into one or more of the channel conflict detection mechanism model, the network KPI prediction mechanism model, and the network anomaly detection mechanism model; The output results of one or more of the channel conflict detection mechanism model, network KPI prediction mechanism model, and network anomaly detection mechanism model are input into the channel optimization strategy mechanism model. The channel optimization strategy mechanism model includes: anomaly access point (AP) location mechanism model, channel replanning mechanism model, and radio frequency quality assessment mechanism model. The device further includes: a third processing module, configured to: determine the AP with channel conflict through the abnormal AP location mechanism model; determine one or more channel optimization schemes based on the load status of the AP with channel conflict through the channel replanning mechanism model; evaluate the radio frequency quality of one or more channel optimization schemes through the radio frequency quality evaluation mechanism model to obtain the evaluation result; and determine the first instruction based on the evaluation result, wherein the first instruction is a control instruction for the optimal channel optimization scheme among the one or more channel optimization schemes.

19. A digital twin system construction apparatus, characterized in that, include: The sixth processing module is used to create mathematical representations of a first object and a second object, respectively. The first object is a physical entity, physical system, or real system, wherein the physical system or real system includes the physical entity, and the physical entity is a network element device; the second object is a digital twin system. The seventh processing module is used to construct a control optimization model for the system construction phase based on the mathematical representations of the first object and the second object and the first formula; the control optimization model is used to determine the control quantities used to optimize the network, and the control quantities are the control quantities acting on network element devices; The first formula is: The second object has a system state transition rule mechanism model, an observation rule mechanism model, and a control rule mechanism model at time t. The system state transition rule mechanism model includes rules for changes in AP topology, AP load, AP channel, and / or network KPI state. The observation rule mechanism model includes a channel conflict detection mechanism model, a network KPI prediction mechanism model, and / or a network anomaly detection mechanism model. The control rule mechanism model is used to determine network performance problems based on network KPI prediction values, locate the root cause of network performance problems by combining the channel conflict situation of each AP, provide channel optimization suggestions, and generate channel control quantities acting on APs. This is used to represent the optimal solution set of the control optimization model in the system construction stage obtained by the first formula, that is, the parameter set of the optimal solution of the system state transition rule mechanism model, the optimal solution of the observation rule mechanism model, and the optimal solution of the control rule mechanism model of the second object at time t; This represents the optimal solution of the system state transition rule mechanism model for the second object at time t. This represents the optimal solution of the observation rule mechanism model for the second object at time t. This represents the optimal solution of the control rule mechanism model for the second object at time t. This indicates the measurement of the observed distance between the first object and the second object. This represents the set of observations of the first object from time 1 to time t. This represents the set of observable mechanism models for the second object from time 1 to time t; This represents the distance measurement between the current state and the desired state of the second object. This represents the set of state mechanism models of the second object from time t to t+T, i.e., the current state of the second object; This represents the set of expected values ​​of the state mechanism model of the second object from time t to t+T, i.e., the preset target state of the second object.

20. A network optimization device, characterized in that, include: The eleventh processing module is used to create mathematical representations of a first object and a second object, respectively. The first object is a physical entity, physical system, or real system, wherein the physical system or real system includes the physical entity, and the physical entity is a network element device; the second object is a digital twin system. The twelfth processing module is used to create a control optimization model for the system construction phase and / or execution phase based on the mathematical representations of the first and second objects and the network optimization objective. The thirteenth processing module is used to determine the control variables for optimizing the network based on the control optimization model of the system construction phase and / or execution phase; The step of determining the control variables for optimizing the network based on the control optimization model from the system construction phase includes: Based on the control optimization model from the system construction phase, the control variables used to optimize the network are determined to be the optimal solution set. ; The control optimization model for the system construction phase is as follows: );in, For network performance evaluation functions, , , , These respectively represent: the system state transition rule mechanism model, observation rule mechanism model, observation mechanism model, and control quantity mechanism model of the second object at time t; wherein, the system state transition rule mechanism model includes AP topology change rules, AP load change rules, AP channel change rules, and / or network KPI state change rules; the observation rule mechanism model includes channel conflict detection mechanism model, network KPI prediction mechanism model, and / or network anomaly detection mechanism model; the observation values ​​corresponding to the observation mechanism model include channel conflict status of each AP, network KPI prediction values, and / or network performance anomaly status under each AP; and the control quantity mechanism model includes channel optimization strategy mechanism model. and Let represent the control quantity and the optimal solution of the control quantity for the first object at time t, respectively. These are the observations of the first object at time t. This indicates the measurement of the observed distance between the first object and the second object; These represent the optimal solutions of the system state transition rule mechanism model, the observation rule mechanism model, and the control variable mechanism model for the second object at time t, respectively.

21. An electronic device, characterized in that, include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the method as described in any one of claims 1-17.

22. A readable storage medium, characterized in that, The readable storage medium stores a program that, when executed by a processor, implements the steps of the method as described in any one of claims 1 to 17.

Images