Method, device and equipment for constructing digital twin model of physical entity system

Abstract

The invention relates to a method, a device and equipment for constructing a digital twin model of a physical entity system. The method comprises the following steps: receiving physical entity models sent by each computing device in the distributed system and state data of each physical entity model; the physical entity model is a simulation model of a physical entity associated with the computing device; performing fusion processing on each physical entity model and the corresponding state data to construct a digital twin model of the physical entity system; the physical entity system comprises physical entities associated with all computing devices. By adopting the method, fusion processing can be performed on each physical entity model and the corresponding state data in a centralized manner through the cloud equipment, so that the digital twin model of the physical entity system can be accurately and efficiently fused and constructed, and the problem of difficulty in constructing a plurality of physical entities in related technologies can be solved.

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and in particular to a method, device and equipment for constructing a digital twin model of a physical entity system. Background Art

[0002] With the continuous development of information processing technology, the application of digital twin technology is becoming more and more extensive. For example, in the complex scenes of digital engineering, there are a large number of physical entities. The digital twin technology can be used to build a digital twin model corresponding to a single physical entity. In practical applications, the constructed digital twin model can be used to realize data analysis, processing, prediction and other aspects of engineering systems in various application fields.

[0003] However, traditional digital twin model construction methods are difficult to build for multiple physical entities. Summary of the invention

[0004] Based on this, it is necessary to provide a method, device and equipment for constructing a digital twin model of a physical entity system in response to the above-mentioned technical problems, which can construct digital twin models for multiple physical entities.

[0005] In a first aspect, the present application provides a method for constructing a digital twin model of a physical entity system, which is applied to a cloud device in a distributed system, and the method includes:

[0006] Receiving a physical entity model sent by each computing device in the distributed system and state data of each physical entity model; the physical entity model is a simulation model of a physical entity associated with the computing device;

[0007] Each of the physical entity models and the corresponding state data are fused to construct a digital twin model of the physical entity system; the physical entity system includes all physical entities associated with the computing devices.

[0008] In one embodiment, the fusion processing of each of the physical entity models and the corresponding state data to construct a digital twin model of the physical entity system includes:

[0009] Performing fusion processing on all parameters of the physical entity models to obtain target model parameters;

[0010] Performing fusion processing on the state data corresponding to all the physical entity models to obtain fused data;

[0011] A digital twin model of the physical entity system is constructed based on the target model parameters and the fusion data.

[0012] In one embodiment, the fusion processing of all the parameters of the physical entity model to obtain the target model parameters includes:

[0013] Using a simulated annealing algorithm to optimize the parameters of all the physical entity models to obtain first parameters;

[0014] Using a genetic algorithm to optimize the parameters of all the physical entity models to obtain second parameters;

[0015] The first parameter and the second parameter are fused to obtain the target model parameter.

[0016] In one embodiment, the fusing of the state data corresponding to all the physical entity models to obtain fused data includes:

[0017] Performing semantic standardization processing on the state data corresponding to all the physical entity models to obtain standard state data corresponding to all the physical entity models;

[0018] A weighted accumulation operation is performed on the values ​​of all the standard state data to obtain the fused data.

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

[0020] Inputting the state data of all the physical entity models into the digital twin model for prediction, and obtaining the state data of the model state of the digital twin model at the next moment;

[0021] The digital twin model is updated according to the state data of all the physical entity models and the state data of the model state at the next moment.

[0022] In one embodiment, updating the digital twin model according to the state data of all the physical entity models and the state data of the model state at the next moment includes:

[0023] Determine a state matrix and a first covariance matrix according to the state data of all the physical entity models;

[0024] Determine a second covariance matrix according to the state data of the model state at the next moment;

[0025] determining a gain matrix according to the state matrix, the first covariance matrix and the second covariance matrix;

[0026] The model parameters of the digital twin model are adjusted according to the gain matrix, the state data of all the physical entity models, the state matrix and the state data of the model state at the next moment to obtain an updated digital twin model.

[0027] In a second aspect, the present application also provides a device for constructing a digital twin model of a physical entity system, comprising:

[0028] A receiving module, used for receiving a physical entity model sent by each computing device in the distributed system and state data of each physical entity model; the physical entity model is a simulation model of a physical entity associated with the computing device;

[0029] A construction module is used to fuse the physical entity models and the corresponding state data to construct a digital twin model of the physical entity system; the physical entity system includes all physical entities associated with the computing devices.

[0030] In a third aspect, the present application further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method in any one of the embodiments of the first aspect when executing the computer program.

[0031] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the method in any one of the embodiments of the first aspect are implemented.

[0032] In a fifth aspect, the present application further provides a computer program product, wherein the computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the method in any one of the embodiments of the first aspect are implemented.

[0033] The method, device and equipment for constructing the digital twin model of the above-mentioned physical entity system receive the physical entity models sent by each computing device in the distributed system, as well as the status data of each physical entity model; the physical entity model is a simulation model of the physical entity associated with the computing device; each physical entity model and the corresponding status data are fused to construct a digital twin model of the physical entity system; the physical entity system includes physical entities associated with all computing devices. The embodiment of the present application can obtain the physical entity model of the physical entity associated with the computing device and the status data of each physical entity model respectively through each computing device, and centrally fuse each physical entity model and the corresponding status data through a cloud device, so that the digital twin model of the physical entity system can be accurately and efficiently fused and constructed, which can solve the problem that it is difficult to construct for multiple physical entities in related technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 is an application environment diagram of a method for constructing a digital twin model of a physical entity system in an embodiment;

[0036] Figure 2 A schematic diagram of a process for constructing a digital twin model of a physical entity system in one embodiment;

[0037] Figure 3 A schematic diagram of a process for constructing a digital twin model in an embodiment;

[0038] Figure 4 is a flow chart of a method for constructing a digital twin model of a physical entity system in another embodiment;

[0039] Figure 5 A schematic flow chart of a method for constructing a digital twin model of a physical entity system in an optional embodiment;

[0040] Figure 6 is a structural block diagram of a device for constructing a digital twin model of a physical entity system in an embodiment;

[0041] Figure 7 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by technicians in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" in the specification and claims of this application and the above-mentioned figure descriptions and any variations thereof are intended to cover non-exclusive inclusions.

[0044] In the description of the embodiments of the present application, the technical terms "first", "second", etc. are only used to distinguish different objects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number, specific order or primary and secondary relationship of the indicated technical features. In the description of the embodiments of the present application, the meaning of "multiple" is more than two, unless otherwise clearly and specifically defined.

[0045] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0046] With the continuous development of information processing technology, the application of digital twin technology is becoming more and more extensive. For example, in complex scenarios of digital engineering, such as large-scale industrial production facilities, smart city infrastructure, complex transportation networks, etc., there are usually a large number of physical entities with complex structures, multiple operating states and rich physical properties. The digital twin technology can be used to build a digital twin model corresponding to a single physical entity. In practical applications, the constructed digital twin model can be used to realize data analysis, processing, prediction and other aspects of engineering systems in various application fields.

[0047] However, the traditional method of constructing digital twin models has the following problems: First, it can only construct models with limited accuracy for simple structures or physical entities under specific working conditions, which makes it difficult to meet the high-precision and high-fidelity requirements of digital twin models in complex scenarios. There is a large deviation between the digital twin model and the physical entity, and it is unable to accurately reflect the real state and change trend of the physical entity in real time, thereby limiting the application effect and decision-making support capabilities of digital twin technology in complex engineering scenarios; Second, in complex scenarios, physical entities may be affected by multiple factors such as changes in the external environment, internal component failures, and switching of operating conditions, resulting in real-time changes in the performance and state of the physical entity. The traditional method of constructing digital twin models is difficult to cope with this dynamic and uncertain nature, making it impossible for the digital twin model to adapt to the changes of the physical entity in a timely manner in actual applications, affecting the efficiency, reliability and practicality of the digital twin model; Third, in complex scenarios, the data collection for constructing digital twin models faces problems such as a wide collection range and a large amount of data. The centralized data processing method is prone to high transmission delays, high processing pressure on the processing center, and poor processing effects. In summary, the traditional method of constructing digital twin models is difficult to construct for multiple physical entities. Therefore, traditional technologies still have many shortcomings in terms of model accuracy, data fusion effect, dynamic update capability, and adaptability to complex scenarios. They cannot meet the actual needs of the growing complex engineering systems for digital twin technology. In other words, there is currently a lack of a complete, efficient, and reliable method for constructing digital twin models for multiple physical entities in complex digital engineering scenarios.

[0048] After introducing the background technology of the method for constructing a digital twin model of a physical entity system provided in the embodiment of the present application, the implementation environment involved in the method for constructing a digital twin model of a physical entity system provided in the embodiment of the present application will be briefly described below. The method for constructing a digital twin model of a physical entity system provided in the embodiment of the present application can be applied to Figure 1 In the cloud device in the distributed system shown. The distributed system includes a cloud device 11 and multiple computing devices 12, each computing device 12 is connected to the cloud device 11 in communication, the computing device 12 can be a terminal, or it can also be a server, and the computing device 12 includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities.

[0049] Those skilled in the art will understand that Figure 1The structure shown in the figure is only a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the distributed system to which the scheme of the present application is applied. The specific distributed system may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0050] In one embodiment, Figure 2 As shown in FIG, a method for constructing a digital twin model of a physical entity system is provided, and the method is applied to Figure 1 Take the cloud device in the distributed system in the example as an example to illustrate, including the following steps:

[0051] S201, receiving a physical entity model sent by each computing device in a distributed system and status data of each physical entity model; the physical entity model is a simulation model of a physical entity associated with the computing device.

[0052] Among them, each computing device can be associated with at least one physical entity, and the physical entity model refers to a simulation model of at least one physical entity associated with the computing device. The simulation model may include but is not limited to a geometric model, a physical model, and a behavioral model, etc. The state data refers to the state data corresponding to the physical entity model collected by the sensor.

[0053] In the embodiment of the present application, the cloud device in the distributed system can obtain each physical entity in the complex scene of the digital engineering according to the function, structural characteristics and data logical relationship, and associate each physical entity with the computing device, so that each computing device in the distributed system can obtain at least one physical entity associated with the computing device, and respectively build a physical entity model corresponding to the physical entity for at least one physical entity associated with the computing device, and use multiple sensors to obtain the state data of the physical entity model, and send the physical entity model and the state data of the physical entity model to the cloud device in the distributed system. Furthermore, the cloud device in the distributed system can receive the physical entity model sent by each computing device in the distributed system, as well as the state data of each physical entity model. Of course, the above-mentioned sending and receiving processes can be real-time, timed or triggered by instructions, and the embodiments of the present application do not limit the execution time of the above-mentioned sending and receiving processes.

[0054] Exemplarily, the process of obtaining each physical entity in a complex scene of digital engineering according to functions, structural characteristics and data logical relationships by a cloud device in a distributed system may be:

[0055] First, in terms of functional characteristics, cloud devices can modularly divide the power system, control system, and transmission system. For example, the physical entities in the complex scenes of digital engineering can be divided into physical entities according to functional characteristics such as energy transfer of the power system, feedback mechanism of the control system, and motion coupling of the transmission system. In this way, the independent functions of each system can be effectively captured and cross-interference can be avoided, ensuring that each physical entity can be independently optimized and can work together with other modules in the digital twin model.

[0056] Second, with respect to the division of structural features, the physical entities in the complex scenarios of digital engineering can be divided into various physical entities according to the connection methods and assembly levels of the mechanical structures corresponding to the physical entities. For example, in a piece of mechanical equipment, the power transmission module may be an integral structure, while the control system may be the result of the cooperation of multiple sub-modules. The embodiments of the present application also combine the principles of hierarchy and module division in engineering design when dividing the structure, and decompose according to the interface definition of each module, so as to achieve the purpose of standardization and modularization.

[0057] Third, in terms of the division of data logical relationships, since each physical entity will generate a large amount of data during operation, the flow direction and processing requirements of the data determine how to divide the physical entities and the interaction between the physical entities. Based on this, the physical entities in the complex scenes of digital engineering can be divided into various physical entities according to the data collection source, transmission path, storage and processing requirements. For example, the physical entities corresponding to temperature sensors and strain sensors can be divided into one category. This is because temperature sensors and strain sensors are installed in key parts of the transmission system and power system respectively. The data generated by temperature sensors and strain sensors are divided into corresponding modules according to functional characteristics, and each module will collect and process data according to the working status of the physical entity. The data flow direction and interaction relationship can ensure that the data of different modules can be seamlessly connected and can be comprehensively analyzed on cloud devices.

[0058] Exemplarily, the process of each computing device in the distributed system constructing a physical entity model corresponding to at least one physical entity associated with the computing device may be:

[0059] First, for the construction of geometric models, we can first use laser scanning technology to accurately capture the three-dimensional point cloud data of the surface of the physical entity, then process the above point cloud data through data processing algorithms, and use three-dimensional modeling software and the above point cloud data to accurately reconstruct the shape structure of the physical entity to obtain a high-precision three-dimensional model. The geometric model not only provides an accurate spatial layout for subsequent physical analysis, but also lays the foundation for the integration of various physical entity models. Through the establishment of the geometric model, the position and form of each physical entity in the physical space can be clearly defined, laying the foundation for accurate physical and behavioral simulation.

[0060] Second, for the construction of physical models, the mechanical, thermal, electrical and other physical properties of each physical entity can be described based on physical laws and material properties. For example, in mechanics, elastic mechanics or plastic mechanics theory can be used to describe the deformation and stress distribution of physical entities under force, that is, the mechanical model of the physical entity can be expressed as the deformation of the elastic body under the action of external force, and formula (1) is used to describe its deformation behavior:

[0061] (1)

[0062] Among them, F is the external force vector acting on the object, K is the stiffness matrix of the object, and the stiffness matrix K depends on the elastic modulus and geometric characteristics of the material, such as cross-sectional area, length, etc., and u is the displacement vector of the object. In this way, the displacement and stress distribution of the physical entity under different loads can be obtained, thereby predicting the behavior of the physical entity. The above physical model can not only optimize the structure of the physical entity, but also make accurate predictions on the response of the physical entity under different working conditions.

[0063] For example, in thermal models, the heat conduction equation can be used to describe the temperature distribution of a physical entity under the action of different heat sources. The heat conduction process inside the physical entity is described by formula (2):

[0064] (2)

[0065] Where T is the temperature, is the thermal diffusivity, is the Laplace operator, which represents the diffusion of heat. Thermal conductivity Depending on the material of the object, it can accurately describe the flow of heat under different temperature gradients. Thermal models are essential for predicting the stability of physical entities under high or low temperature conditions.

[0066] For example, an electrical model can include the relationship between current, voltage and resistance, such as Ohm's law, which can be used to describe the electrical behavior in a circuit. For a complex circuit containing resistors, capacitors and inductors, a circuit model can be established to predict the electrical response of the physical entity under different operating conditions.

[0067] Third, for the construction of the behavior model, the machine learning model can be trained through historical operation data to obtain a trained machine learning model for inferring the behavior pattern and response law of the physical entity under different working conditions. For example, for a transmission system, the behavior model can infer the behavior of the system under different load, speed, temperature and other working conditions. Exemplarily, the behavior model can be described by a linear regression model as shown in formula (3):

[0068] (3)

[0069] Among them, y is the output behavior of the physical entity (such as position, velocity, etc.), X is the input feature matrix (such as working condition parameters), is the regression coefficient, is the error term. The regression coefficient obtained through training Able to predict the specific behavior of physical entities under different operating conditions.

[0070] Exemplarily, the process of each computing device in the distributed system using multiple sensors to acquire the state data of the physical entity model may be:

[0071] Low-power, high-sensitivity sensors with adaptive sampling frequency adjustment functions can be deployed at key locations of physical entities, and multiple sensors can be used to obtain the status data of the physical entity model. Therefore, in a complex digital engineering environment, sensors can be used to adjust the sampling frequency according to the real-time working conditions and operating status of the physical entity to ensure the representativeness and accuracy of the data, which can ensure that the collected data not only reflects the real state of the physical entity, but also provides effective support in data analysis. Among them, sensor types can include but are not limited to position sensors, pressure sensors, temperature sensors, strain sensors, etc. Among them, position sensors can monitor the position changes of physical entities in real time, temperature sensors and pressure sensors can monitor changes in the physical environment, and strain sensors can measure structural stress and deformation. Different types of sensors can capture different properties of physical entities, and the combination of different types of sensors can fully reflect the health status and working efficiency of physical entities.

[0072] Regarding the adaptive sampling frequency adjustment function, for example, when the physical entity is in a stable operating state, the sampling frequency can be reduced to save energy; when the working conditions change drastically, such as a sudden increase in load or equipment failure, the sampling frequency can be quickly increased to capture more detailed information. The above dynamic adjustment process can effectively avoid data oversampling or undersampling, ensuring that energy efficiency is maximized while ensuring data accuracy. For example, formula (4) can be used to dynamically adjust the sampling frequency f of the sensor according to the workload L of the physical entity and the degree of environmental change E:

[0073] (4)

[0074] Where f is the sampling frequency of the sensor, L is the load of the physical entity, and E is the degree of environmental change. , , is the adjustment coefficient, which reflects the impact of load and environmental changes on the sampling frequency. Through the above sampling frequency adjustment mechanism, the relationship between the sampling frequency and the physical entity load and environmental changes can be quantified to ensure that when the load is high or the environment changes greatly, the sampling frequency is increased to capture more data details and ensure that important operating data is not missed at critical moments; while under low load and stable environment, the sampling frequency will be appropriately reduced to reduce energy consumption and unnecessary data redundancy. In addition, by adjusting the sampling frequency, the sampling frequency can be adjusted to reduce the load and environmental changes. , , The dynamic adjustment of equal coefficients can optimize the sampling strategy according to different physical entities or application scenarios, thereby improving the representativeness and accuracy of the data and enhancing the real-time and responsiveness of the entire twin system.

[0075] In addition, in order to efficiently integrate the above data and ensure that the data between modules can be transmitted and processed quickly and accurately, a mathematical model of data flow and interaction can be established. Assume that the data generated by each physical entity model in a certain period of time is represented as a vector ,in is all the data generated by the i-th physical entity model at time t. Then, according to the data generation source and flow relationship, the data will be transmitted through a unified interface. If the data flows from physical entity model i to physical entity model j, the transmission time and delay of the data flow can be expressed by formula (5):

[0076] (5)

[0077] in, represents the transmission time of the data flow from physical entity model i to physical entity model j, is the amount of data generated by the physical entity model i, is the data transmission link rate, It is the network bandwidth limitation. In this way, the time delay in the data transmission process can be calculated, and the data transmission link can be further analyzed and optimized. When the data transmission speed is slow and / or the delay is high, the data acquisition frequency can be adjusted, the data transmission path and network bandwidth can be optimized, and the data transmission speed and delay between each physical entity model can be optimized to improve the efficiency of data processing, thereby achieving fast data synchronization and real-time update in the digital twin model, ensuring the efficiency and timeliness of the data flow, and ensuring that the digital twin model can continuously and accurately reflect the status of the physical entity.

[0078] In addition, after obtaining the status data of the physical entity model through sensors, each computing device in the distributed system can also perform real-time analysis and processing on the status data of the physical entity model, thereby improving the efficiency of data collection and processing while filtering out useless noise data, retaining only valuable information for digital twin model optimization and analysis, and sending only processed status data to cloud devices, thereby significantly reducing data processing delays, alleviating the burden on cloud devices, reducing dependence on central computing resources, and improving the scalability and reliability of the system.

[0079] The data processing process performed by each computing device on the status data may include but is not limited to operations such as data caching, real-time data processing, denoising, feature extraction, and anomaly detection.

[0080] For example, Kalman filtering can be used to denoise the state data. Assume that the data sequence collected by a sensor is , where data It may be interfered by noise of different frequencies. Through the Kalman filtering method, the original data can be filtered using formula (6):

[0081] (6)

[0082] in, is the filtered value at the current moment, is the sensor reading at the current moment, is the filtered value at the previous moment, It is the Kalman gain that determines the trade-off between the current reading and the filtered value. Kalman filtering can effectively suppress noise and extract the true trend of sensor signals while ensuring computational efficiency. In this way, the computing device can remove noise while collecting data in real time, providing more accurate data for subsequent analysis and processing.

[0083] For example, the state data can be feature extracted to obtain valuable features, such as the frequency domain features in the vibration signal and the fluctuation trend in the temperature signal. These features are crucial to the optimization and accuracy improvement of the digital twin model. Assume that the collected temperature data sequence is , it can be converted from the time domain to the frequency domain through the fast Fourier transform (FFT) to extract the frequency component. The formula of FFT is shown in the following equation (7):

[0084] (7)

[0085] in, is the frequency domain representation of the signal, is the temperature data at the kth moment, f is the frequency, and j is the imaginary unit. Through frequency domain analysis, the computing device can identify periodic changes in the signal and extract features that are helpful for model optimization.

[0086] For example, in order to promptly detect sudden changes or anomalies in sensor data during the operation of a physical entity, the computing device can perform anomaly detection on the state data by setting a threshold or using a machine learning model. Assume that a threshold method based on standard deviation is used to detect anomalies in temperature data:

[0087] (8)

[0088] in, is the temperature data at the current moment, is the mean of historical data, is the standard deviation of the historical data, is a multiple of the set threshold. When the difference between a data point and the mean exceeds a certain multiple of the standard deviation, the computing device can determine that the temperature data is abnormal. In this way, the computing device can detect potential equipment failures or operating condition changes in real time and report them to the cloud device in a timely manner, thereby providing a basis for fault warning.

[0089] In addition, before each computing device in the distributed system sends the physical entity model and the status data of the physical entity model to the cloud device in the distributed system, it is necessary to pre-standardize the structured design of each physical entity model, including the division of functional modules, the definition of data interfaces, and the communication protocols between modules. For example, the data output of each physical entity model can be configured in JSON format.

[0090] In order to achieve the rapid combination and seamless connection of physical entity models, the definition of the interface must not only consider the data transmission format, but also ensure the timeliness and consistency of data exchange. Therefore, the data interaction protocol must have an efficient message transmission mechanism. For example, the publish / subscribe mode and the asynchronous communication mode based on the message queue can be used to handle the information exchange between different physical entity models to ensure that the input and output of each physical entity model can be synchronized in real time. Specifically, the computing device can send the processed state data corresponding to the physical entity model to other physical entity models or cloud devices through the message queue according to the source, processing requirements and timeliness of the physical entity model. Other physical entity models or cloud devices subscribe to the corresponding message queue and can receive the processed state data in real time or regularly according to the requirements. In this way, the direct dependence between physical entity models can be eliminated, so that each physical entity model can remain relatively independent in terms of functional logic, message transmission rate and data format, which not only enhances the flexibility and scalability between physical entity models, but also greatly improves the stability of the physical entity system under high load.

[0091] In addition, by setting priorities for messages, we can ensure that critical data and emergency messages are delivered and processed first, thus reducing the delay of important tasks during transmission. For example, the delay model shown in formula (9) can be used to describe the performance of message transmission:

[0092] (9)

[0093] in, Indicates the total delay from message publishing to subscription completion. is the time when the publisher pushes the message to the queue, It is the time it takes for a message to be queued and transmitted. It is the time it takes for a subscriber to receive and process a message from a queue. To reduce the overall latency, you can optimize the queue scheduling strategy and use a priority-based scheduling algorithm to put high-priority messages at the front of the queue, thereby reducing the transmission delay of critical data.

[0094] In addition, in order to avoid queue blocking or message accumulation under high load conditions, a dynamic load balancing algorithm can be introduced to distribute different messages to multiple parallel queues for processing. The load balancing algorithm can be expressed as formula (10):

[0095] (10)

[0096] in, represents the message load assigned to the i-th queue, is the weight of the ith queue, representing the processing capacity of the queue, and R is the total message request rate. Through the above load balancing strategy, the distribution of messages can be dynamically adjusted according to the processing capacity of different queues to avoid a message delivery bottleneck caused by overloading of a certain queue.

[0097] In addition, the asynchronous communication middleware of each message (i.e., each physical entity model and the corresponding state data) can use a persistent storage mechanism to ensure that when a system failure or network interruption occurs, the messages that have been transmitted but not consumed by the subscriber can be safely stored and redelivered after the system is restored to prevent data loss.

[0098] S202, fusing each physical entity model and the corresponding state data to construct a digital twin model of the physical entity system; the physical entity system includes all physical entities associated with the computing devices.

[0099] Among them, the physical entity system includes all physical entities associated with the computing devices. In the embodiment of the present application, optionally, the cloud device in the distributed system can simultaneously fuse each physical entity model and the corresponding state data to construct a digital twin model of the physical entity system; or, the cloud device in the distributed system can also first fuse each physical entity model to obtain a fused physical entity model, and then fuse the state data corresponding to each physical entity model to obtain the fused state data, thereby constructing a digital twin model of the physical entity system based on the fused physical entity model and the fused state data; or, the cloud device in the distributed system can also first fuse the state data corresponding to each physical entity model to obtain the fused state data, and then fuse each physical entity model to obtain the fused physical entity model, thereby constructing a digital twin model of the physical entity system based on the fused physical entity model and the fused state data. Of course, the embodiment of the present application does not limit the specific implementation method and sequence of the fusion processing.

[0100] In the method for constructing the digital twin model of the above-mentioned physical entity system, the physical entity models sent by each computing device in the distributed system and the status data of each physical entity model are received; the physical entity model is a simulation model of the physical entity associated with the computing device; each physical entity model and the corresponding status data are fused to construct a digital twin model of the physical entity system; the physical entity system includes physical entities associated with all computing devices. In the embodiment of the present application, the physical entity model of the physical entity associated with the computing device and the status data of each physical entity model can be obtained by each computing device respectively, and each physical entity model and the corresponding status data can be fused centrally by a cloud device. In this way, the digital twin model of the physical entity system can be accurately and efficiently fused and constructed, which can solve the problem that it is difficult to construct for multiple physical entities in related technologies.

[0101] In one embodiment, a method for constructing a digital twin model is provided, namely, "combining each physical entity model and the corresponding state data to construct a digital twin model of the physical entity system" in the above S202. Figure 3 As shown, including:

[0102] S301, fusing the parameters of all physical entity models to obtain target model parameters.

[0103] In the embodiments of the present application, optionally, the cloud device in the distributed system can directly fuse the parameters of all physical entity models to obtain the target model parameters; or, the cloud device in the distributed system can also use different optimization algorithms to optimize the parameters respectively, obtain the parameter optimization results corresponding to each optimization algorithm, and then fuse the parameter optimization results corresponding to each optimization algorithm to obtain the target model parameters. Of course, the embodiments of the present application do not limit the specific implementation method for obtaining the target model parameters. Among them, the target model parameters refer to the fused model parameters used to construct the digital twin model.

[0104] S302, fusing the state data corresponding to all physical entity models to obtain fused data.

[0105] In an embodiment of the present application, optionally, the cloud device in the distributed system can directly perform fusion processing on the state data corresponding to all physical entity models to obtain fused data; or, the cloud device in the distributed system can first perform data processing on the state data corresponding to all physical entity models to obtain the state data after data processing, and then perform fusion processing on the state data after data processing to obtain fused data. Of course, the embodiment of the present application does not limit the specific implementation method for obtaining fused data. Among them, fused data refers to the fused state data used to construct the digital twin model.

[0106] It should be noted that the embodiment of the present application does not limit the sequence of S301 and S302.

[0107] S303, constructing a digital twin model of the physical entity system based on the target model parameters and the fusion data.

[0108] In an embodiment of the present application, the cloud device in the distributed system can predetermine the model framework, and configure the target model parameters and fusion data in the model framework, so as to construct a digital twin model of the physical entity system.

[0109] In this embodiment, the parameters of all physical entity models can be fused to accurately obtain the target model parameters, and the state data corresponding to all physical entity models can be fused to accurately obtain the fused data. Therefore, based on the accurate target model parameters and fused data, the digital twin model of the physical entity system can be accurately and efficiently constructed.

[0110] In one embodiment, a method for implementing model parameter fusion is provided, namely, the step of “fusion processing of parameters of all physical entity models to obtain target model parameters” in S301 above, including:

[0111] The simulated annealing algorithm is used to optimize the parameters of all physical entity models to obtain the first parameters.

[0112] The genetic algorithm is used to optimize the parameters of all physical entity models to obtain the second parameters.

[0113] The first parameter and the second parameter are fused to obtain the target model parameter.

[0114] Among them, the simulated annealing algorithm is a global optimization algorithm based on randomization, which is suitable for large-scale and complex structure optimization problems. The simulated annealing algorithm is good at finding the global optimal solution in a large-scale solution space. The genetic algorithm can effectively handle complex parameter spaces and optimization problems. The first parameter and the second parameter can include but are not limited to the physical properties, boundary conditions and interactive interfaces of each physical entity model.

[0115] In the embodiment of the present application, first, the cloud device in the distributed system can use the simulated annealing algorithm to optimize the parameters of all physical entity models to obtain the first parameter. For example, the optimal solution can be searched by simulating the particle state change in the physical annealing process. The core idea is to guide the search process by gradually lowering the system temperature, thereby avoiding falling into the local optimal solution. The formula of the simulated annealing algorithm is shown in the following formula (11):

[0116] (11)

[0117] Among them, E and are the objective function values ​​of the current solution and the new solution respectively, T is the current temperature, represents the probability of accepting a new solution. When it is less than the current solution E, the new solution is accepted; otherwise, the temperature is used to determine whether to accept the new solution. As the temperature gradually decreases, the simulated annealing algorithm will gradually converge to the global optimal solution.

[0118] Secondly, the cloud devices in the distributed system can use genetic algorithms to optimize the parameters of all physical entity models through natural selection, crossover and mutation operations, and find a set of optimal parameter configurations to obtain the second parameters. For example, when there are N physical entity models, assuming that the parameters of each physical entity model are vectors ,in, Including the parameters of the physical entity model in terms of geometry, physics, and behavior, in the optimization process, a genetic algorithm can be used to calculate the fitness function corresponding to each set of parameter configurations, and generate new parameter combinations through operations such as selection, crossover, and mutation to gradually improve the performance of the overall system. The goal of the genetic algorithm is to find an optimal solution that minimizes the error of the global digital twin model. The fitness function can be defined as the weighted sum of the model error, as shown in formula (12):

[0119] (12)

[0120] in, represents the geometric model error, represents the physical property error, represents the prediction error of the behavioral model, is the weight coefficient, which is used to balance the impact of different errors on the optimization results. Through iterative optimization by genetic algorithm, the goal is to make the fitness function The value of is minimized, thus ensuring the accuracy and consistency of the global model.

[0121] Afterwards, the cloud device in the distributed system can perform weighted fusion processing on the first parameter and the second parameter to obtain the target model parameter. It should be noted that the embodiment of the present application does not limit the order of the two parameter optimization processes.

[0122] In this embodiment, through the comprehensive application of genetic algorithm and simulated annealing algorithm, the parameters and connection relationships between physical entity models can be automatically adjusted to achieve coordinated optimization of physical entity model parameters, and each verified and optimized physical entity model can be effectively integrated to form a unified global model of physical structure, functional relationship and data interaction. The consistency and high precision of the global digital twin model in geometric shape, physical properties and behavior mode are ensured, and the integration of each physical entity model can meet the performance requirements of the overall system, and ensure that the global digital twin model has good scalability and flexibility, can adapt to digital engineering scenes of different scales and complexities, and provide a solid guarantee for the long-term stable operation of the digital twin model. Thus, under dynamically changing working conditions, the optimized digital twin model can accurately reflect the state of the physical entity, and then be used in application scenarios such as real-time monitoring, fault diagnosis, and performance optimization. Specifically, the genetic algorithm can effectively search for the optimal parameter configuration in the global solution space, while the simulated annealing algorithm can help jump out of the local optimum and further improve the overall performance of the integrated model.

[0123] In one embodiment, a method for implementing state data fusion is provided, namely, the step of “fusion processing of state data corresponding to all physical entity models to obtain fused data” in S302 above, including:

[0124] Semantic standardization is performed on the state data corresponding to all physical entity models to obtain standard state data corresponding to all physical entity models.

[0125] Perform weighted accumulation operation on the values ​​of all standard state data to obtain fused data.

[0126] In the embodiments of the present application, firstly, due to the differences in functions, structures and data interaction logics between different physical entities, the state data generated by them have significant differences at the semantic level. In order to eliminate such differences, the embodiments of the present application can construct a data semantic description system covering all aspects of physical entities through an ontological method, so that the data generated by different physical entity models can be accurately understood and used correctly during the transmission and fusion process across physical entity models. The ontological method is to formally describe the various attributes, characteristics and interrelationships of physical entities to form a unified semantic model to ensure that information is not lost due to semantic ambiguity during data transmission, parsing and fusion. For example, the semantic model of a physical entity can be represented as a triple , where C is a concept set, including core concepts such as the functional characteristics, structural features, sensor types, and data sources of physical entities; R is a relationship set, describing the associations between different concepts, such as the hierarchical relationship between physical entities, and the mapping relationship between sensor data and physical model parameters; I is an instance set, including specific instance data of physical entities and sensors in actual operation. That is, the cloud devices in the distributed system can perform semantic standardization on the state data corresponding to all physical entity models, and obtain the standard state data corresponding to all physical entity models. Based on the triples after semantic standardization, all data sources can be labeled and parsed under a unified semantic framework to ensure that data from different sources have a consistent semantic basis when fused across models.

[0127] For example, the specific process of semantic standardization can be: for each data set generated by a sensor or data node , it needs to be mapped and annotated according to the semantic model O. The semantic mapping process can be formally expressed as shown in formula (13):

[0128] (13)

[0129] in, represents the standardized data after semantic mapping, f is the semantic mapping function, which converts the original data Converted into a data representation that conforms to the unified semantic model O. In this way, the state data not only retains the original numerical information, but also has semantic information such as its source, unit, characteristic meaning, and association with other entities. This semantic annotation process enables data to be accurately parsed and understood during transmission and fusion, thereby avoiding information loss or misinterpretation due to differences in data source and meaning.

[0130] Afterwards, the cloud devices in the distributed system can use big data technology and the distributed computing architecture Apache Spark to conduct in-depth data mining and fusion analysis. Spark's in-memory computing capabilities enable it to maintain high efficiency when processing state data, thereby ensuring the real-time processing of state data and adapting to the needs of real-time monitoring and decision-making. When the sensor data of multiple physical entities need to be fused, Spark can apply a variety of data fusion algorithms through its built-in machine learning library MLlib to achieve collaborative analysis of different data sources. For example, the cloud devices in the distributed system can perform weighted accumulation operations on the values ​​of all standard state data to obtain fused data, which can be understood as, for data sets from different sensors , each data set contains different features from different physical entities, such as position, temperature, strain, etc. Spark's distributed processing capability can perform data fusion through the weighted average method of the following formula (14):

[0131] (14)

[0132] in, is the fused dataset, is the sensor data of the ith physical entity, is the weight of the data set, indicating the reliability or importance of the data source. In the process of data fusion, Spark can adjust the weight of each data source according to factors such as the trust or validity of the data source, the sensor accuracy of the data source, the time synchronization error, the accuracy of historical data, etc., so as to generate more accurate fusion data and provide more accurate input for the optimization of the digital twin model. Through the above fusion process, the cloud can integrate the status data from different physical entities and sensors in real time and comprehensively, eliminate data redundancy, and improve the accuracy and consistency of the data. In addition, Spark also further analyzes the data, performs pattern recognition, anomaly detection and trend prediction, etc. For example, through semantic rules and threshold settings, it can quickly identify abnormal data points that may appear in the cross-model fusion process to avoid interference with the overall model. In this way, when the data collected by a sensor has a significant deviation from the physical model parameters associated with it, this abnormal situation can be detected by setting a threshold, and then a data backtracking or compensation mechanism is adopted to ensure the stability and consistency of the overall data fusion.

[0133] Optionally, the sensor state data can also be fused based on the Dempster-Shafer evidence theory and / or fuzzy logic fusion method. In this way, multi-source information can be effectively integrated, the bias caused by a single data source can be avoided, and the interference of redundant data can be eliminated to a certain extent.

[0134] For example, Dempster-Shafer evidence theory is an important method for dealing with uncertainty information, which realizes data fusion by assigning confidence and combining evidence from different data sources. Assume that there are multiple physical entity models that provide confidence assignments about a certain event or parameter. is the set of possible events, and They represent the confidence allocation functions of two different data sources respectively, and their application formulas can be integrated through the following formula (15):

[0135] (15)

[0136] in, is the comprehensive confidence of event A, B and C are the confidence distribution sets from two data sources. It represents the conflict coefficient, which reflects the degree of conflict between different data sources in the confidence distribution. Through the above fusion formula, the conflict impact of different physical entity models at the data level can be effectively reduced, and the confidence information corresponding to the state data of each physical entity model can be reasonably integrated to obtain a more reliable fusion result. This formula is particularly effective when there is a large uncertainty or partial conflict in the data, ensuring that the fusion result can achieve the optimal balance between different confidence distributions.

[0137] In addition, to ensure the security and integrity of data, the data processing center can also implement backup and redundancy mechanisms. In the cloud data center, important data will be backed up multiple times and stored in different physical locations. Redundancy mechanisms include regular data backup, real-time synchronous data storage, and disaster recovery strategies. The data synchronization mechanism ensures that data will not be lost when problems occur in computing devices or networks, and the status data of all sensors can be recovered and processed in the cloud.

[0138] In this embodiment, the state data corresponding to all physical entity models can be semantically standardized to obtain the standard state data corresponding to all physical entity models, and then the values ​​of all standard state data can be weighted and accumulated to obtain fused data. In this way, the multi-source information fusion theory can be introduced in the data fusion process to combine the characteristics of data from different sources to generate fused data with higher credibility and integrity, so as to support the accuracy, real-time update and precise optimization of the digital twin model.

[0139] In one embodiment, a method for updating a digital twin model is provided, that is, a method for constructing a digital twin model of the above-mentioned physical entity system, such as Figure 4 As shown, it also includes:

[0140] S203, inputting the state data of all physical entity models into the digital twin model for prediction, and obtaining the state data of the model state of the digital twin model at the next moment.

[0141] S204: Update the digital twin model according to the status data of all physical entity models and the status data of the model status at the next moment.

[0142] In an embodiment of the present application, the cloud device in the distributed system can input the state data of all physical entity models into the digital twin model for prediction, and obtain the state data of the model state of the digital twin model at the next moment. Afterwards, optionally, the cloud device in the distributed system can update the digital twin model based on the state data of all physical entity models at the current moment and the state data of the model state at the next moment; or, the cloud device in the distributed system can also obtain external environment change data, and update the digital twin model based on the state data of all physical entity models at the current moment, the external environment change data, and the state data of the model state at the next moment. Of course, the embodiment of the present application does not limit the specific implementation method for updating the digital twin model.

[0143] In addition, a physical entity model library for managing physical entity models can be set up in the cloud device in the distributed system. The physical entity model library is not only used to store models, but also to provide model classification, version control and update management functions. In this way, each physical entity model can be classified and stored according to its function and purpose, such as mechanical model, electrical model, thermal model, etc. And by controlling the version of the physical entity model library, it can be ensured that the update of each model can be tracked and managed to avoid conflicts and inconsistencies between different versions of models.

[0144] In addition, event-triggered synchronization strategies can be designed in cloud devices in distributed systems. Event-triggered synchronization strategies are used to ensure that the digital twin model can respond quickly and synchronize accurately at critical moments in the state changes of physical entities, that is, when important events occur. Important events may include but are not limited to sudden changes in the state of the physical entity system, drastic changes in the external environment, or abnormal performance of key components. When the event triggering conditions are met, the cloud device can immediately start the model synchronization mechanism to update the state of the digital twin model in real time to ensure a high degree of consistency between the twin and the physical entity in time and state space. For example, an event triggering condition function can be defined , when the following condition (16) is met, state synchronization is triggered:

[0145] (16)

[0146] in, is a preset threshold, and the vector corresponding to the state data of the digital twin model at time t is , the state data collected by the sensor at time t is , is the state deviation function, which is defined as the Euclidean distance or the difference in specific key indicators between the state data corresponding to the digital twin model and the state data collected by the sensor. When the above formula (16) is satisfied, the cloud device can immediately perform forced synchronization of the model state, with priority over the model update mechanism. In this way, the above event-driven synchronization strategy can be used to ensure the real-time consistency between the digital twin model and the physical entity in abnormal states or key changes.

[0147] In the process of dynamic update and real-time synchronization, a state weight adjustment mechanism can also be introduced, that is, different update weights can be assigned to various parts of the digital twin model according to the importance of different parameters. For example, for the key structures or high-risk areas of the physical entity, a higher weight coefficient can be set to ensure that the key structures or high-risk areas of the physical entity get higher accuracy and faster response speed during the synchronization process.

[0148] In addition, the effect of dynamic model update can be verified through the following aspects: first, the state prediction error of the model is significantly reduced, indicating that the model can accurately reflect the actual state of the physical entity; second, the event-triggered synchronization response time is shortened, and the synchronous adjustment of the model can be quickly completed at the critical moment; third, the model shows good stability and consistency in long-term continuous operation, avoiding model drift caused by error accumulation.

[0149] In this embodiment, the digital twin model can be updated according to the state data of all physical entity models and the state data of the model state at the next moment. In this way, by establishing a data-driven model update mechanism and an event-triggered synchronization strategy, it is possible to ensure that the digital twin model is always highly consistent with the physical entity, providing accurate model support for real-time monitoring, analysis, and decision-making in complex digital engineering scenarios.

[0150] In one embodiment, a method for updating a digital twin model is provided, namely, “updating the digital twin model according to the state data of all physical entity models and the state data of the model state at the next moment” in the above S204, including:

[0151] A state matrix and a first covariance matrix are determined according to the state data of all physical entity models.

[0152] The second covariance matrix is ​​determined according to the state data of the model state at the next moment.

[0153] A gain matrix is ​​determined according to the state matrix, the first covariance matrix, and the second covariance matrix.

[0154] The model parameters of the digital twin model are adjusted according to the gain matrix, the state data of all physical entity models, the state matrix and the state data of the model state at the next moment to obtain an updated digital twin model.

[0155] In an embodiment of the present application, since the data-driven model update mechanism relies on the data stream transmitted from the physical entity to the twin system in real time, these state data may include but are not limited to various state parameters such as position, pressure, temperature, stress, etc. collected by sensors, as well as historical operation data and external environment change data. Based on this, the cloud devices in the distributed system can drive the dynamic update of model parameters and status based on the state data of all the above-mentioned physical entity models and the state data of the predicted model state at the next moment.

[0156] Exemplarily, the cloud device in the distributed system can determine the state matrix and the first covariance matrix according to the state data of all physical entity models, and determine the second covariance matrix according to the state data of the model state at the next moment, so that the gain matrix can be determined according to the state matrix, the first covariance matrix and the second covariance matrix using formula (17). The first covariance matrix can be the observation noise covariance matrix, the second covariance matrix can be the model state prediction error covariance matrix, and the gain matrix can be the Kalman gain matrix.

[0157] (17)

[0158] in, is the Kalman gain matrix, H is the state matrix, is the model state prediction error covariance matrix, and R is the observation noise covariance matrix.

[0159] Thus, the cloud device in the distributed system can adjust the model parameters of the digital twin model according to the gain matrix, the state data of all physical entity models, the state matrix and the state data of the model state at the next moment to obtain an updated digital twin model. For example, assuming that the vector corresponding to the state data of the digital twin model at time t is , the state data collected by the sensor at time t is , the vector corresponding to the state data of the predicted model state is , the model parameters can be calibrated in real time through the formula (18) corresponding to the Kalman Filter algorithm:

[0160] (18)

[0161] in, represents the vector corresponding to the calibrated model state, is the vector corresponding to the model state predicted by the model, It is the state data collected by the sensor. In this way, the state data of the sensor can be weighted and fused with the state data predicted by the model, and the weight can be determined by the covariance of the observation noise and the prediction error, so that the parameters and state of the digital twin model can be adjusted in real time to make it closer to the actual operation of the physical entity.

[0162] In this embodiment, by adjusting the parameters and status of the digital twin model in real time, the digital twin model can be highly consistent with the actual operation of the physical entity in the time and space dimensions, ensuring that the digital twin model can reflect the state changes of the physical entity in real-time operation and respond promptly to external environmental disturbances to support real-time monitoring, analysis and decision-making.

[0163] In an optional embodiment, if Figure 5 As shown, a method for constructing a digital twin model of a physical entity system is provided, which is applied to cloud devices in a distributed system. The physical entity system includes physical entities associated with all computing devices, including:

[0164] S401, receiving a physical entity model sent by each computing device in a distributed system and state data of each physical entity model; the physical entity model is a simulation model of a physical entity associated with the computing device;

[0165] S402, optimizing the parameters of all physical entity models by using a simulated annealing algorithm to obtain first parameters;

[0166] S403, optimizing the parameters of all physical entity models using a genetic algorithm to obtain second parameters;

[0167] S404, fusing the first parameter and the second parameter to obtain a target model parameter;

[0168] S405, performing semantic standardization processing on the state data corresponding to all physical entity models to obtain standard state data corresponding to all physical entity models;

[0169] S406, performing weighted accumulation operation on the values ​​of all standard state data to obtain fused data;

[0170] S407, constructing a digital twin model of the physical entity system based on the target model parameters and the fusion data;

[0171] S408, inputting the state data of all physical entity models into the digital twin model for prediction, and obtaining the state data of the model state of the digital twin model at the next moment;

[0172] S409, determining a state matrix and a first covariance matrix according to the state data of all physical entity models;

[0173] S410, determining a second covariance matrix according to state data of the model state at the next moment;

[0174] S411, determining a gain matrix according to the state matrix, the first covariance matrix and the second covariance matrix;

[0175] S412, adjusting the model parameters of the digital twin model according to the gain matrix, the state data of all physical entity models, the state matrix and the state data of the model state at the next moment to obtain an updated digital twin model.

[0176] In addition, cloud devices in distributed systems can also perform performance evaluation on digital twin models after building a global digital twin model. The performance evaluation indicators should cover multiple aspects such as model accuracy, real-time performance, stability, and scalability. The selection and quantification of each indicator needs to be closely related to the application scenarios and objectives of the digital twin model.

[0177] For example, first, the validity and credibility of the digital twin model can be evaluated by determining the model accuracy. Specifically, the model accuracy can be measured by comparing the deviation rate between the actual physical entity and the geometric model, the error range of the physical model, and the prediction accuracy of the behavioral model. For the deviation rate of the geometric model, it can be quantified by calculating the distance error between the digital twin model and the physical entity. The geometric shape of the digital twin model is represented by a series of point clouds, while the actual geometric shape of the physical entity is obtained by laser scanning or other measurement methods, so the geometric error can be calculated by the following formula (19):

[0178] (19)

[0179] in, represents the position of the i-th point in the digital twin model, represents the position of the corresponding point in the physical entity, N is the total number of points, Represents the Euclidean distance. This formula measures the deviation of the model in space. A value indicates a high accuracy of the twin geometry model.

[0180] For physical models, the error range can be used to measure the degree of match between the digital twin model and the actual physical behavior, such as mechanical, thermal or electrical errors. The temperature calculated by the physical model is , while the actual temperature measurement is , the error range is expressed as absolute error or relative error, as shown in the following formula (20):

[0181] (20)

[0182] The above indicators can be used to evaluate the performance of physical models under different working conditions. The smaller the error, the more accurate the prediction of the physical properties of the model.

[0183] The prediction accuracy of the behavior model can be evaluated by the error between the predicted value and the actual observed value. The output of a certain working condition predicted by the model is , and the actual output is , the prediction error is expressed by the formula (21) corresponding to the mean square error (MSE):

[0184] (twenty one)

[0185] Where M is the number of samples, and Represent the predicted value and true value of the jth sample respectively. The smaller This shows that the behavioral model has strong predictive ability.

[0186] Second, real-time indicators can be determined by data update delay and model response time. Data update delay refers to the time interval from obtaining data from the physical entity and transmitting it to the model, and model response time refers to the response time of the digital twin model to external inputs (such as sensor data or control instructions). To quantify these delays, the following formula (22) can be used for calculation:

[0187] (twenty two)

[0188] in, is the data receiving time, is the data sending time. For the model response time, the input signal is The output response is , the response time is expressed as the following formula (23):

[0189] (twenty three)

[0190] in, and Represent the timestamps of output and input respectively, and N is the number of processed data. Reducing latency and response time can improve the real-time performance of the model, making it more suitable for dynamic monitoring and control systems.

[0191] Third, the stability index can also be used to evaluate the operational stability and reliability of the digital twin model under different working conditions. The stability index is quantified by evaluating the model's adaptability to environmental changes, input fluctuations, or system failures during long-term operation. In order to evaluate the stability of the model, a change rate index is defined as shown in the following formula (24):

[0192] (twenty four)

[0193] in, is the change in the model output, is the time interval. The smaller the value, the more stable the model is under various conditions.

[0194] Fourth, the scalability index can be used to evaluate the digital twin model's ability to expand to new physical entities or functions. As the complexity of the system increases, the digital twin model should be able to flexibly incorporate new physical entity models and ensure that the overall performance is not negatively affected. Scalability can be measured by the performance change of the system after adding a new physical entity model. The specific formula is shown in the following formula (25):

[0195] (25)

[0196] like A smaller value indicates that the digital twin model has strong scalability.

[0197] Through comprehensive analysis of the above indicators, it is possible to identify the deficiencies of the digital twin model in terms of accuracy, real-time performance, stability and scalability, and provide data support for further optimization and improvement. The above evaluation method can not only ensure the reliability of the model under different working conditions, but also improve the model's adaptability in dealing with complex and dynamically changing digital engineering environments.

[0198] In the method for constructing the digital twin model of the above-mentioned physical entity system, the physical entity models sent by each computing device in the distributed system and the status data of each physical entity model are received; the physical entity model is a simulation model of the physical entity associated with the computing device; each physical entity model and the corresponding status data are fused to construct a digital twin model of the physical entity system; the physical entity system includes physical entities associated with all computing devices. In the embodiment of the present application, the physical entity model of the physical entity associated with the computing device and the status data of each physical entity model can be obtained by each computing device respectively, and each physical entity model and the corresponding status data can be fused centrally by a cloud device. In this way, the digital twin model of the physical entity system can be accurately and efficiently fused and constructed, which can solve the problem that it is difficult to construct for multiple physical entities in related technologies.

[0199] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0200] Based on the same inventive concept, the embodiment of the present application also provides a device for constructing a digital twin model of a physical entity system for implementing the method for constructing a digital twin model of a physical entity system involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of the device for constructing a digital twin model of one or more physical entity systems provided below can be found in the above limitations on the method for constructing a digital twin model of a physical entity system, and will not be repeated here.

[0201] In an exemplary embodiment, Figure 6 As shown, a device for constructing a digital twin model of a physical entity system is provided, including: a receiving module 31 and a constructing module 32, wherein:

[0202] The receiving module 31 is used to receive the physical entity model sent by each computing device in the distributed system, as well as the status data of each physical entity model; the physical entity model is a simulation model of the physical entity associated with the computing device.

[0203] The construction module 32 is used to fuse each physical entity model and the corresponding state data to construct a digital twin model of the physical entity system; the physical entity system includes all physical entities associated with the computing devices.

[0204] In one embodiment, the building block 32 includes:

[0205] The first fusion unit is used to fuse the parameters of all physical entity models to obtain target model parameters;

[0206] The second fusion unit is used to perform fusion processing on the state data corresponding to all physical entity models to obtain fused data;

[0207] A construction unit is used to construct a digital twin model of a physical entity system based on target model parameters and fusion data.

[0208] In one embodiment, the first fusion unit comprises:

[0209] A first parameter optimization subunit is used to optimize the parameters of all physical entity models by using a simulated annealing algorithm to obtain first parameters;

[0210] A second parameter optimization subunit is used to optimize the parameters of all physical entity models using a genetic algorithm to obtain second parameters;

[0211] The fusion subunit is used to fuse the first parameter and the second parameter to obtain the target model parameter.

[0212] In one embodiment, the second fusion unit comprises:

[0213] The standardization processing subunit is used to perform semantic standardization processing on the state data corresponding to all physical entity models to obtain standard state data corresponding to all physical entity models;

[0214] The accumulation operation subunit is used to perform weighted accumulation operation on the values ​​of all standard state data to obtain fused data.

[0215] In one embodiment, the device for constructing the digital twin model of the physical entity system further includes:

[0216] The prediction module is used to input the state data of all physical entity models into the digital twin model for prediction, and obtain the state data of the model state of the digital twin model at the next moment;

[0217] The update module is used to update the digital twin model according to the status data of all physical entity models and the status data of the model state at the next moment.

[0218] In one embodiment, the update module includes:

[0219] A first determining unit, used for determining a state matrix and a first covariance matrix according to state data of all physical entity models;

[0220] A second determining unit, configured to determine a second covariance matrix according to state data of the model state at a next moment;

[0221] A third determining unit, configured to determine a gain matrix according to the state matrix, the first covariance matrix, and the second covariance matrix;

[0222] The updating unit is used to adjust the model parameters of the digital twin model according to the gain matrix, the state data of all physical entity models, the state matrix and the state data of the model state at the next moment to obtain an updated digital twin model.

[0223] Each module in the device for constructing the digital twin model of the physical entity system can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0224] In an exemplary embodiment, a computer device is provided. The computer device may be a server, or the computer device may be a terminal. The internal structure diagram thereof may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be realized through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a method for constructing a digital twin model of a physical entity system is realized. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0225] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0226] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0227] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0228] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0229] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0230] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0231] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0232] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for constructing a digital twin model of a physical entity system, characterized in that: Applied to a cloud device in a distributed system, the method comprises: Receiving a physical entity model sent by each computing device in the distributed system and state data of each physical entity model; the physical entity model is a simulation model of a physical entity associated with the computing device; Each of the physical entity models and the corresponding state data are fused to construct a digital twin model of the physical entity system; the physical entity system includes all physical entities associated with the computing devices.

2. The method according to claim 1, characterized in that The fusing of the physical entity models and the corresponding state data to construct a digital twin model of the physical entity system includes: Performing fusion processing on all parameters of the physical entity models to obtain target model parameters; Performing fusion processing on the state data corresponding to all the physical entity models to obtain fused data; A digital twin model of the physical entity system is constructed based on the target model parameters and the fusion data.

3. The method according to claim 2, characterized in that The fusion processing of all the parameters of the physical entity model to obtain the target model parameters includes: Using a simulated annealing algorithm to optimize the parameters of all the physical entity models to obtain first parameters; Using a genetic algorithm to optimize the parameters of all the physical entity models to obtain second parameters; The first parameter and the second parameter are fused to obtain the target model parameter.

4. The method according to claim 2, characterized in that: The fusing of the state data corresponding to all the physical entity models to obtain fused data includes: Performing semantic standardization processing on the state data corresponding to all the physical entity models to obtain standard state data corresponding to all the physical entity models; A weighted accumulation operation is performed on the values ​​of all the standard state data to obtain the fused data.

5. The method according to any one of claims 1 to 4, characterized in that The method further comprises: Inputting the state data of all the physical entity models into the digital twin model for prediction, and obtaining the state data of the model state of the digital twin model at the next moment; The digital twin model is updated according to the state data of all the physical entity models and the state data of the model state at the next moment.

6. The method according to claim 5, characterized in that The updating of the digital twin model according to the state data of all the physical entity models and the state data of the model state at the next moment includes: Determine a state matrix and a first covariance matrix according to the state data of all the physical entity models; Determine a second covariance matrix according to the state data of the model state at the next moment; determining a gain matrix according to the state matrix, the first covariance matrix and the second covariance matrix; The model parameters of the digital twin model are adjusted according to the gain matrix, the state data of all the physical entity models, the state matrix and the state data of the model state at the next moment to obtain an updated digital twin model.

7. A device for constructing a digital twin model of a physical entity system, characterized in that: The device comprises: A receiving module, used for receiving a physical entity model sent by each computing device in the distributed system and state data of each physical entity model; the physical entity model is a simulation model of a physical entity associated with the computing device; A construction module is used to fuse the physical entity models and the corresponding state data to construct a digital twin model of the physical entity system; the physical entity system includes all physical entities associated with the computing devices.

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

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

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images