Simulation method and system of digital twin system, product, equipment and storage medium

Abstract

The invention discloses a simulation method and system of a digital twinning system, a product, equipment and a storage medium, and relates to the field of digital twinning, and the method comprises the steps: extracting each control target identifier and control intention parameter in a simulation control instruction through employing a preset large language model, and calling a preset quantum graph search algorithm based on the control target identifier to determine a target entity corresponding to the control target identifier in the twin map, and performing simulation control operation on the target entity based on the control intention parameter. According to the method, the preset large language model is configured to extract the control intention parameter containing the deep semantics and the control target identifier, and the quantum output state after the target entity executes the action is obtained by using the preset quantum feature extraction model during the simulation control operation, so that the extraction of the complex parameter relationship of the physical system on the quantum level is realized; and finally, the predicted operation state of the digital twin system is determined based on the quantum output state by using the preset multi-layer perceptron, so that the simulation precision and efficiency of the digital twin system are improved.

Description

Technical Field

[0001] This application relates to the field of digital twin technology, and in particular to a simulation method, system, product, device and storage medium for a digital twin system. Background Technology

[0002] A digital twin system is a virtual system that uses digital twin technology to model and simulate a physical system, thereby enabling the monitoring and control of the physical system. It is widely used in manufacturing, transportation, urban management and other fields.

[0003] To improve the modeling and control accuracy of digital twin systems on physical systems, existing digital twin methods often incorporate deep learning models to assist in modeling, simulation, and control. However, because existing deep learning models rely on gradient descent through a large number of samples and high-dimensional feature controls for model optimization, they are prone to local optima or gradient vanishing problems when the physical system exhibits high dimensionality, multiple variables, and nonlinear coupling. Consequently, digital twin systems built using existing methods fail to accurately capture the deep semantics of instructions and the complex parameter relationships within the physical system, leading to decreased simulation efficiency and accuracy. Summary of the Invention

[0004] In view of the above problems, this application provides a simulation method, system, product, device, and storage medium for a digital twin system to improve the simulation efficiency and accuracy of the digital twin system. The specific solution is as follows:

[0005] The first aspect of this application provides a simulation method for a digital twin system, comprising:

[0006] The obtained simulation control commands are input into a preset large language model to obtain the identification of each control target and the control intent parameters.

[0007] Based on the control target identifier, a preset quantum graph search algorithm is invoked to determine the target entity corresponding to the control target identifier in the twin graph, and simulation control operations are performed on the target entity based on the control intent parameters. The twin graph includes the entities and topological structures of the physical system simulated by the digital twin system, and the simulation control operations include:

[0008] Control the target entity in the twin graph to perform an action adapted to the control intent parameters, and obtain the temporal operation parameters of each entity in the twin graph at the current time;

[0009] Each of the aforementioned time-series operating parameters is input into a preset quantum feature extraction model to obtain the quantum output state corresponding to each of the aforementioned time-series operating parameters. The quantum output state characterizes the probability value of the entity in different operating states after the entity performs the action adapted by the control intention parameters. The preset quantum feature extraction model includes a preset time-series feature extraction model, a preset quantum state encoder, and a preset variable quantum circuit.

[0010] The measured quantum output states are input into a preset multilayer perceptron to obtain the predicted operating state of the digital twin system after the simulation control operation is executed.

[0011] In one possible implementation, the timing parameters are input into a preset quantum feature extraction model to obtain the quantum output state corresponding to each timing parameter, including:

[0012] The time-series features corresponding to each of the time-series operation parameters are extracted using the preset time-series feature extraction model.

[0013] The temporal features are mapped to quantum features using the preset quantum state encoder;

[0014] Each of the quantum features is input into the preset variable quantum circuit to obtain the quantum output state corresponding to each of the timing operation parameters.

[0015] In one possible implementation, the twin graph generation process includes: obtaining configuration information of each entity of the physical system and the topology;

[0016] The configuration information and the topology are input into a preset graph convolutional neural network to obtain an initial twin graph.

[0017] The initial twin map is encoded using a preset quantum tensor compressor, and the encoded initial twin map is rendered using a preset rendering engine to obtain the twin map.

[0018] In one possible implementation, the simulation method for the digital twin system further includes:

[0019] The loss values ​​of the preset temporal feature extraction model, the preset variable quantum circuit, and the preset rendering engine are obtained, and the weighted sum of each loss value is performed to obtain the joint loss value.

[0020] Based on the joint loss value, backpropagation optimization is performed on the preset temporal feature extraction model and the preset multilayer perceptron; based on the joint loss value, parameter gradient optimization is performed on the preset variable quantum circuit.

[0021] In one possible implementation, the generation process of the preset variable quantum circuit includes: obtaining the historical timing operation parameters of the physical system, and extracting features from each of the historical timing operation parameters to obtain each historical timing feature;

[0022] The preset quantum state encoder maps each of the historical time-series features to historical quantum features, and uses each of the historical quantum features to train the initial variable quantum circuit to obtain the preset variable quantum circuit. The input of the preset variable quantum circuit is the quantum feature, and the output of the preset variable quantum circuit is the quantum output state.

[0023] A second aspect of this application provides a simulation system for a digital twin system, the simulation system comprising:

[0024] The parameter extraction module is used to input the obtained simulation control commands into a preset large language model to obtain the identification of each control target and the control intent parameters.

[0025] The simulation control module is used to determine the target entity corresponding to the control target identifier in the twin graph by calling a preset quantum graph search algorithm based on the control target identifier, and to perform simulation control operations on the target entity based on the control intent parameters. The twin graph includes the entities and topological structures of the physical system simulated by the digital twin system, and the simulation control operations include:

[0026] Control the target entity in the twin graph to perform an action adapted to the control intent parameters, and obtain the temporal operation parameters of each entity in the twin graph at the current time;

[0027] Each of the aforementioned time-series operating parameters is input into a preset quantum feature extraction model to obtain the quantum output state corresponding to each of the aforementioned time-series operating parameters. The quantum output state characterizes the probability value of the entity in different operating states after the entity performs the action adapted by the control intention parameters. The preset quantum feature extraction model includes a preset time-series feature extraction model, a preset quantum state encoder, and a preset variable quantum circuit.

[0028] The measured quantum output states are input into a preset multilayer perceptron to obtain the predicted operating state of the digital twin system after the simulation control operation is executed.

[0029] In one possible implementation, the simulation control module is configured to input each of the time-series operating parameters into a preset quantum feature extraction model to obtain the quantum output state corresponding to each of the time-series operating parameters as follows:

[0030] The time-series features corresponding to each of the time-series operation parameters are extracted using the preset time-series feature extraction model.

[0031] The temporal features are mapped to quantum features using the preset quantum state encoder;

[0032] Each of the quantum features is input into the preset variable quantum circuit to obtain the quantum output state corresponding to each of the timing operation parameters.

[0033] In one possible implementation, the simulation system of the digital twin system further includes a twin map generation unit, which is configured during the generation of the twin map as follows:

[0034] Obtain the configuration information and topology of each entity in the physical system;

[0035] The configuration information and the topology are input into a preset graph convolutional neural network to obtain an initial twin graph.

[0036] The initial twin map is encoded using a preset quantum tensor compressor, and the encoded initial twin map is rendered using a preset rendering engine to obtain the twin map.

[0037] In one possible implementation, the simulation system of the digital twin system further includes:

[0038] The joint optimization unit is used to obtain the loss values ​​of the preset temporal feature extraction model, the preset variable quantum circuit, and the preset rendering engine, and to perform a weighted summation of the loss values ​​to obtain a joint loss value; to perform backpropagation optimization operation on the preset temporal feature extraction model and the preset multilayer perceptron based on the joint loss value; and to perform parameter gradient optimization operation on the preset variable quantum circuit based on the joint loss value.

[0039] In one possible implementation, the simulation system of the digital twin system further includes a variable quantum circuit generation unit, which is configured during the generation of the preset variable quantum circuit as follows:

[0040] The historical time-series operating parameters of the physical system are obtained, and features are extracted from each of the historical time-series operating parameters to obtain each historical time-series feature;

[0041] The preset quantum state encoder maps each of the historical time-series features to historical quantum features, and uses each of the historical quantum features to train the initial variable quantum circuit to obtain the preset variable quantum circuit. The input of the preset variable quantum circuit is the quantum feature, and the output of the preset variable quantum circuit is the quantum output state.

[0042] A third aspect of this application provides a computer program product including computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement a simulation method for a digital twin system as described in the first aspect or any implementation thereof.

[0043] A fourth aspect of this application provides an electronic device, including at least one processor and a memory connected to the processor, wherein:

[0044] The memory is used to store computer programs;

[0045] The processor is used to execute the computer program so that the electronic device can implement the simulation method of the digital twin system of the first aspect or any implementation thereof.

[0046] The fifth aspect of this application provides a computer storage medium carrying one or more computer programs, which, when executed by an electronic device, enable the electronic device to simulate a digital twin system of the first aspect or any implementation thereof.

[0047] By employing the aforementioned technical solutions, this application provides a simulation method, system, product, device, and storage medium for a digital twin system. Through configuration and utilization of a preset large language model, it outputs control target identifiers and control intent parameters based on the obtained simulation control commands. This utilizes the preset large language model to capture the deep semantics of the simulation control commands, improving the accuracy of the obtained control intent parameters and control target identifiers. Furthermore, for high-dimensional, multi-variable, and nonlinear physical systems, their corresponding twin graphs contain numerous transition relationships. Quantum graph search algorithms, through quantum walks, efficiently verify quantum states and can capture long-distance relationships between different entities. Therefore, this application, by configuring a preset quantum graph search algorithm, determines the target entities corresponding to the control target identifiers in the twin graph, improving the search efficiency and accuracy for target entities. Subsequently, by configuring the target entities in the control twin graph to execute actions adapted to the control intent parameters, a visual representation of the simulation process is achieved. Simultaneously, by configuring actions that adapt to the control intent parameters of the target entity, the temporal operating parameters of each entity in the current twin graph are obtained. These temporal operating parameters are then input into a preset quantum feature extraction model to obtain the corresponding quantum output state. Since the preset quantum feature extraction model includes a preset temporal feature extraction model, a preset quantum state encoder, and a preset variable quantum circuit, it allows for the analysis and extraction of complex parameter relationships between entities in the physical system at the quantum level. Finally, by configuring the measured quantum output states to be input into a preset multilayer perceptron, the predicted operating state of the digital twin system after the simulation control operation is obtained using the preset multilayer perceptron output, thus improving the simulation accuracy of the digital twin system. Therefore, this application improves the simulation efficiency and accuracy of the digital twin system. Attached Figure Description

[0048] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent when taken in conjunction with the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0049] Figure 1 A flowchart of a simulation method for a digital twin system provided in this application;

[0050] Figure 2 A block diagram illustrating the simulation of a digital twin system provided in this application;

[0051] Figure 3 This is a schematic diagram of the structure of an electronic device provided in this application. Detailed Implementation

[0052] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is for explaining specific embodiments only and is not intended to limit the scope of this application.

[0053] The embodiments of this application will now be described with reference to the accompanying drawings. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are equally applicable to similar technical problems.

[0054] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0055] The first aspect of this application provides a simulation method for a digital twin system, comprising:

[0056] S101. Input the obtained simulation control commands into the preset large language model to obtain the identification of each control target and the control intention parameters.

[0057] It should be noted that in practical application scenarios, the above simulation control commands can be voice commands issued by the user or irregular text commands in natural language form input by the user.

[0058] It should be noted that, in practical application scenarios, the aforementioned control target identifier is the unique identifier of the entity in the digital twin system that requires the corresponding simulation control command to execute actions.

[0059] It should be noted that in practical application scenarios, the aforementioned control intent parameters can be parameters that characterize the control intent of the simulation control command. For example, in a smart traffic scenario at an intersection, the content of the simulation control command might be: simulate congestion paths at the intersection at 8:00 AM. The extracted command control intent parameters could then be information such as the illumination status of each traffic light at the intersection and the congestion status of each path.

[0060] It should be noted that in practical applications, existing digital twin systems require users to input corresponding simulation control commands to trigger entity actions under specific operating conditions during simulation. These commands are often input by operators using rule-based commands or physical triggers, resulting in high operational difficulty and configuration complexity. This application, however, utilizes a pre-defined large language model to identify the control target identifiers and control intent parameters within the input simulation control commands. This reduces operational difficulty and system configuration complexity while improving the accuracy of extracting deep semantics from the commands, thereby enhancing the accuracy of the obtained control target identifiers and control intent parameters.

[0061] It should be noted that in practical application scenarios, the aforementioned preset large language model can be a large language model (LLM) service provided by a partner, or it can be a large language model that has been trained using historical simulation control commands and then deployed locally.

[0062] S102. Based on the control target identifier, a preset quantum graph search algorithm is invoked to determine the target entity corresponding to the control target identifier in the twin graph, and simulation control operations are performed on the target entity based on the control intent parameters. The twin graph includes the entities and topological structures of the physical system simulated by the digital twin system, and the simulation control operations include:

[0063] Control the target entities in the twin graph to perform actions that are adapted to the control intent parameters, and obtain the temporal operation parameters of each entity in the twin graph at the current moment;

[0064] Each time-series operating parameter is input into a preset quantum feature extraction model to obtain the quantum output state corresponding to each time-series operating parameter. The quantum output state represents the probability value of the entity in different operating states after the entity performs the action adapted to the control intention parameters. The preset quantum feature extraction model includes a preset time-series feature extraction model, a preset quantum state encoder, and a preset variable quantum circuit.

[0065] The measured quantum output states are input into a preset multilayer perceptron to obtain the predicted operating state of the digital twin system after the simulation control operation is performed.

[0066] It should be noted that in practical applications, the entities mentioned above are virtual components simulating elements in a physical system. The aforementioned twin graph is a graph generated based on digital twin technology to simulate the topological relationships between entities in a physical system.

[0067] It should be noted that, in practical applications, the aforementioned pre-defined quantum graph search algorithm utilizes quantum properties to rapidly search for nodes corresponding to target entities within twin graphs. Because this algorithm is based on the quantum walk principle, it uses superposition states to identify multiple search paths in parallel, thereby calculating the correlation probability between entities and ultimately determining the target entity corresponding to the control target identifier. Compared to existing graph search algorithms (such as Dijkstra's algorithm) that traverse nodes one by one, this pre-defined quantum graph search algorithm utilizes quantum entanglement to capture long-distance nonlinear coupling relationships (such as implicit dependencies between devices in a system). While improving search efficiency and accuracy, it avoids getting trapped in local optima, thus improving the accuracy and efficiency of determining the target entity corresponding to the control target identifier, and consequently improving simulation efficiency.

[0068] It should be noted that, in practical applications, the aforementioned time-series operating parameters can be the operating parameters of various entities collected by sensors in the physical system. Since there are correlations between the operating states of different entities in the physical system, and these correlations are usually carried within the aforementioned time-series operating parameters, this application configures a preset quantum feature extraction model, including a preset time-series feature extraction model, a preset quantum state encoder, and a preset variable quantum circuit. The preset quantum feature extraction model then generates the quantum output state corresponding to each input time-series operating parameter, thereby enabling the extraction of correlations between entities in the physical system at the quantum level, thus improving the data quality used for operating state prediction.

[0069] It should be noted that, in practical application scenarios, this application configures a preset quantum feature extraction model, including a preset temporal feature extraction model, a preset quantum state encoder, and a preset variable quantum circuit. Based on each temporal operating parameter, the preset quantum feature extraction model is used to extract and analyze the complex parameter relationships between entities in the physical system at the quantum level, thereby outputting a quantum output state that represents the probability value of the entity in different operating states after the entity performs the action of adapting the control intention parameters.

[0070] It should be noted that, in practical applications, the aforementioned pre-defined multilayer perceptron (MLP) is a feedforward neural network that predicts the operating state of the digital twin system based on a nonlinear mapping of forward propagation and error backpropagation, according to the input quantum output state. Because the input to the pre-defined MLP in this application is a high-dimensional quantum output state, it avoids the overfitting problem caused by traditional MLPs that use low-dimensional features as input during the simulation of nonlinear systems, thus improving the prediction accuracy of the digital twin system's operating state after executing simulation control operations.

[0071] It should be noted that, in practical applications, the aforementioned measurement of the quantum output state of the input preset multilayer perceptron refers to the operation of collapsing the quantum output state of the quantum state to construct observable classical values. Specifically, since the quantum output state output by the preset quantum feature extraction model is a quantum state in a superposition state, and because the quantum state contains probabilistic information about the complex parameter relationships of the physical system existing in the form of qubit superposition or entanglement, the quantum output state cannot be directly processed or recognized by the preset multilayer perceptron, thus causing the subsequent simulation process to be interrupted. This application, by configuring the measurement of the quantum output state input to the preset multilayer perceptron, collapses the quantum output state into observable classical values ​​(such as bit sequences or probability values), thereby ensuring that the subsequent simulation prediction function is executed normally.

[0072] This application improves the accuracy of the obtained control intent parameters and outputs control target identifiers and control intent parameters based on the obtained simulation control commands by configuring a pre-set large language model. This allows the pre-set large language model to capture the deep semantics of the simulation control commands. Furthermore, for high-dimensional, multivariable, and nonlinear physical systems, their corresponding twin graphs contain numerous transition relationships. Quantum graph search algorithms, which efficiently verify quantum states through quantum walks, can capture long-distance relationships between different entities. Therefore, this application improves the search efficiency and accuracy of target entities by configuring a pre-set quantum graph search algorithm to determine the target entities corresponding to the control target identifiers in the twin graph. Subsequently, by configuring the target entities in the control twin graph to perform actions adapted to the control intent parameters, the simulation process is visualized. Simultaneously, by configuring actions that adapt to the control intent parameters of the target entity, the temporal operating parameters of each entity in the current twin graph are obtained. These temporal operating parameters are then input into a preset quantum feature extraction model to obtain the corresponding quantum output state. Since the preset quantum feature extraction model includes a preset temporal feature extraction model, a preset quantum state encoder, and a preset variable quantum circuit, it allows for the analysis and extraction of complex parameter relationships between entities in the physical system at the quantum level. Finally, by configuring the measured quantum output states to be input into a preset multilayer perceptron, the predicted operating state of the digital twin system after the simulation control operation is obtained using the preset multilayer perceptron output, thus improving the simulation accuracy of the digital twin system. Therefore, this application improves the simulation efficiency and accuracy of the digital twin system.

[0073] In one possible implementation, the various time-series operating parameters are input into a preset quantum feature extraction model to obtain the quantum output states corresponding to each time-series operating parameter, including:

[0074] The temporal features corresponding to each temporal operating parameter are extracted using a pre-defined temporal feature extraction model.

[0075] A pre-defined quantum state encoder is used to map each temporal feature into a quantum feature;

[0076] Each quantum feature is input into a preset variable quantum circuit to obtain the quantum output state corresponding to each timing operation parameter.

[0077] It should be noted that in practical applications, the aforementioned time-series data refers to structured time-series features that do not contain nonlinear relationships. The preset time-series feature extraction model can be a neural network model built based on Long Short-Term Memory (LSTM) or Transformer models. Since the aforementioned time-series operating parameters refer to the high-dimensional, continuous, and unstructured time-series data (such as velocity, position, energy, etc.) output by an entity in a physical system after performing actions adapted to control intent parameters over a period of time, this data may include noisy data, redundant data, or nonlinear relationships. Traditional quantum state encoders, for unstructured time-series data, require the superposition and entanglement of qubits to process the nonlinear relationships in the data, leading to increased processing time. This is detrimental to improving the mapping efficiency of quantum features and to the accurate capture of the entity's dependencies in the time dimension by the quantum state encoder. Therefore, this application uses a preset temporal feature extraction model to extract temporal features corresponding to each temporal operating parameter to obtain structured features that do not contain nonlinear relationships. This improves the output efficiency of the preset quantum state encoder for quantum features and enables the preset quantum state encoder to accurately capture the time-dimensional dependencies between entities based on temporal features, thereby improving the accuracy of the output quantum features.

[0078] In one possible implementation, the above-described method of mapping each temporal feature to a quantum feature using a preset quantum state encoder may include the following steps A1 to A3.

[0079] Step A1: Initialize the timing characteristics to the corresponding initial qubits. Then trigger step A2.

[0080] Step A2: Use a quantum gate sequence to convert the initial qubit into the target state, and trigger step A3.

[0081] In one possible implementation, the quantum gate sequence in step A2 above can be a rotation gate sequence or a control gate sequence.

[0082] Step A3: Output the target state as a quantum feature.

[0083] In one possible implementation, the process of generating the aforementioned twin map includes:

[0084] Obtain the configuration information and topology of each entity in the physical system;

[0085] The configuration information and topology are input into a pre-defined graph convolutional neural network to obtain an initial twin graph.

[0086] The initial twin map is encoded using a preset quantum tensor compressor, and the encoded initial twin map is rendered using a preset rendering engine to obtain the twin map.

[0087] It should be noted that in practical applications, the aforementioned pre-defined graph convolutional neural network can embed entity features into the topological structure through domain aggregation to obtain a high-dimensional initial twin graph. Subsequently, a pre-defined quantum tensor compressor is used to compress and encode the initial twin graph to reduce its dimensionality. Finally, a pre-defined rendering engine is used to visualize the compressed and encoded initial twin graph based on rasterization or ray tracing principles, thus obtaining the twin graph. Since the obtained twin graph can be used not only for subsequent simulations but also for visualizing digital twin systems, the above two compressions allow the obtained twin graph to retain quantum detail features while reducing dimensionality, thereby improving the visualization and interactive display effects and real-time performance.

[0088] In one possible implementation, the simulation method for the digital twin system provided in the first aspect of this application further includes:

[0089] Obtain the loss values ​​of the preset temporal feature extraction model, preset variable quantum circuit and preset rendering engine, and sum the weighted values ​​to obtain the joint loss value;

[0090] Backpropagation optimization is performed on a preset temporal feature extraction model and a preset multilayer perceptron based on the joint loss value; parameter gradient optimization is performed on a preset variable quantum circuit based on the joint loss value.

[0091] It should be noted that the aforementioned backpropagation can be implemented using classical gradient descent and / or quantum variational algorithms. This application configures backpropagation optimization operations based on the joint loss value for a preset temporal feature extraction model and a preset multilayer perceptron; and performs parameter gradient optimization operations on a preset variational quantum circuit based on the joint loss value. This allows this application to achieve global error minimization compared to the independent optimization methods of existing technologies, thereby improving overall simulation efficiency.

[0092] In one possible implementation, the generation process of the aforementioned preset variable quantum circuit includes:

[0093] The historical time-series operating parameters of the physical system are obtained, and features are extracted from each historical time-series operating parameter to obtain the historical time-series features.

[0094] The preset quantum state encoder maps each historical time sequence feature to a historical quantum feature, and uses each historical quantum feature to train the initial variable quantum circuit to obtain the preset variable quantum circuit. The input of the preset variable quantum circuit is the quantum feature, and the output of the preset variable quantum circuit is the quantum output state.

[0095] It should be noted that, in practical applications, the aforementioned preset quantum state encoder is used to map historical temporal characteristics to quantum states, enabling subsequent preset variable quantum circuits to perform probabilistic calculations in quantum space based on historical quantum characteristics. Specifically, the aforementioned preset quantum state encoder embeds historical temporal characteristics into the amplitude of the quantum state, using qubits to represent these historical temporal characteristics. Through the exponential storage capability of quantum superposition, high-dimensional historical temporal characteristics are compressed into low-dimensional quantum space, facilitating the preset variable quantum circuit to analyze nonlinear coupling relationships between entities (e.g., latent fault correlations between devices) in quantum parallelism. This improves the accuracy of the quantum output state of the preset variable quantum circuit, thereby enhancing the accuracy of subsequent simulation predictions.

[0096] It should be noted that, in practical applications, the aforementioned Preset Variational Quantum Circuit (VQC) is a quantum circuit that uses adjustable-parameter quantum gates to process the input quantum state through quantum state evolution. By adjusting the parameters of the quantum gates, the target loss is minimized (the probability distribution of the output is adjusted). Because the Preset Variational Quantum Circuit can utilize quantum parallelism and quantum entanglement to efficiently simulate the nonlinear coupling and uncertainties of physical systems, it improves the quantum output state, thereby enhancing the final simulation and prediction accuracy.

[0097] A second aspect of this application provides a simulation system for a digital twin system, the simulation system comprising:

[0098] The parameter extraction module 201 is used to input the obtained simulation control commands into a preset large language model to obtain the identification of each control target and the control intention parameters.

[0099] The simulation control module 202 is used to determine the target entity corresponding to the control target identifier in the twin graph by calling a preset quantum graph search algorithm based on the control target identifier, and to perform simulation control operations on the target entity based on the control intent parameters. The twin graph includes the entities and topological structures of the physical system simulated by the digital twin system. The simulation control operations include:

[0100] Control the target entities in the twin graph to perform actions that are adapted to the control intent parameters, and obtain the temporal operation parameters of each entity in the twin graph at the current moment;

[0101] Each time-series operating parameter is input into a preset quantum feature extraction model to obtain the quantum output state corresponding to each time-series operating parameter. The quantum output state represents the probability value of the entity in different operating states after the entity performs the action adapted to the control intention parameters. The preset quantum feature extraction model includes a preset time-series feature extraction model, a preset quantum state encoder, and a preset variable quantum circuit.

[0102] The measured quantum output states are input into a preset multilayer perceptron to obtain the predicted operating state of the digital twin system after the simulation control operation is performed.

[0103] In one possible implementation, the simulation control module 202 is configured to input each time-series operating parameter into a preset quantum feature extraction model and obtain the quantum output state corresponding to each time-series operating parameter as follows:

[0104] The temporal features corresponding to each temporal operating parameter are extracted using a pre-defined temporal feature extraction model.

[0105] A pre-defined quantum state encoder is used to map each temporal feature into a quantum feature;

[0106] Each quantum feature is input into a preset variable quantum circuit to obtain the quantum output state corresponding to each timing operation parameter.

[0107] In one possible implementation, the simulation system of the digital twin system provided in the second aspect of this application further includes a twin map generation unit, which is configured during the twin map generation process as follows:

[0108] Obtain the configuration information and topology of each entity in the physical system;

[0109] The configuration information and topology are input into a pre-defined graph convolutional neural network to obtain an initial twin graph.

[0110] The initial twin map is encoded using a preset quantum tensor compressor, and the encoded initial twin map is rendered using a preset rendering engine to obtain the twin map.

[0111] In one possible implementation, the simulation system for the digital twin system provided in the second aspect of this application further includes:

[0112] The joint optimization unit is used to obtain the loss values ​​of the preset temporal feature extraction model, the preset variable quantum circuit, and the preset rendering engine, and to perform a weighted summation of the loss values ​​to obtain the joint loss value; to perform backpropagation optimization operation on the preset temporal feature extraction model and the preset multilayer perceptron based on the joint loss value; and to perform parameter gradient optimization operation on the preset variable quantum circuit based on the joint loss value.

[0113] In one possible implementation, the simulation system of the digital twin system further includes a variable quantum circuit generation unit, which is configured during the generation of a preset variable quantum circuit as follows:

[0114] The historical time-series operating parameters of the physical system are obtained, and features are extracted from each historical time-series operating parameter to obtain the historical time-series features.

[0115] By using a preset quantum state encoder, each historical time sequence feature is mapped to a historical quantum feature, and the initial variable quantum circuit is trained using each historical quantum feature to obtain the preset variable quantum circuit. The input of the preset variable quantum circuit is the quantum feature, and the output of the preset variable quantum circuit is the quantum output state.

[0116] A third aspect of this application provides a computer program product including computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement a simulation method for a digital twin system as described in the first aspect or any implementation thereof.

[0117] A fourth aspect of this application provides an electronic device, including at least one processor and a memory connected to the processor, wherein:

[0118] Memory is used to store computer programs;

[0119] The processor is used to execute computer programs to enable electronic devices to implement the simulation method of a digital twin system of the first aspect or any implementation thereof.

[0120] The fifth aspect of this application provides a computer storage medium carrying one or more computer programs, which, when executed by an electronic device, enable the electronic device to simulate a digital twin system of the first aspect or any implementation thereof.

[0121] The structural schematic diagram of the electronic device provided in the third aspect of this application is as follows: Figure 3 As shown. The electronic devices in the embodiments of this application may include, but are not limited to, fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 3 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0122] like Figure 3As shown, the electronic device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. When the electronic device is powered on, the RAM 303 also stores various programs and data required for the operation of the electronic device. The processing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0123] Typically, the following devices can be connected to I / O interface 305: input devices 306 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 307 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 308 including, for example, memory cards, hard drives, etc.; and communication devices 309. Communication device 309 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have alternatively.

[0124] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0125] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0126] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0127] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A simulation method of a digital twin system, characterized by, The method comprises the following steps: inputting the obtained simulation control instruction into a preset large language model to obtain control target identifiers and control intent parameters; based on the control target identifiers, calling a preset quantum graph search algorithm to determine target entities corresponding to the control target identifiers in a twin graph, and performing simulation control operations on the target entities based on the control intent parameters, wherein the twin graph comprises entities and topological structures of the entities of a physical system simulated by the digital twin system, and the simulation control operations comprise: controlling the target entities in the twin graph to perform actions adapted to the control intent parameters, and obtaining time sequence running parameters of the entities in the twin graph at the current time; inputting each time sequence running parameter into a preset quantum feature extraction model to obtain a quantum output state corresponding to each time sequence running parameter, wherein the quantum output state represents probability values of different running states of the entities after the entities perform actions adapted to the control intent parameters, and the preset quantum feature extraction model comprises a preset time sequence feature extraction model, a preset quantum state encoder, and a preset variational quantum circuit; inputting each measured quantum output state into a preset multilayer perceptron to obtain a predicted running state of the digital twin system after the simulation control operation is performed. 2.The simulation method of a digital twin system according to claim 1, characterized in that, inputting each time sequence running parameter into a preset quantum feature extraction model to obtain a quantum output state corresponding to each time sequence running parameter, comprising: extracting time sequence features corresponding to each time sequence running parameter using the preset time sequence feature extraction model; mapping each time sequence feature to a quantum feature using the preset quantum state encoder; inputting each quantum feature into the preset variational quantum circuit to obtain the quantum output state corresponding to each time sequence running parameter. 3.The simulation method of a digital twin system according to claim 1, characterized in that, The generation process of the twin graph comprises: obtaining configuration information and the topological structure of each entity of the physical system; inputting the configuration information and the topological structure into a preset graph convolutional neural network to obtain an initial twin graph; encoding the initial twin graph using a preset quantum tensor compressor, and rendering the encoded initial twin graph using a preset rendering engine to obtain the twin graph. 4.The simulation method of the digital twin system according to claim 3, characterized in that, The simulation method of the digital twin system further comprises: obtaining loss values of the preset time sequence feature extraction model, the preset variational quantum circuit, and the preset rendering engine, and performing weighted summation on each loss value to obtain a joint loss value; based on the joint loss value, performing a backpropagation optimization operation on the preset time sequence feature extraction model and the preset multilayer perceptron; based on the joint loss value, performing a parameter gradient optimization operation on the preset variational quantum circuit. 5.The simulation method of a digital twin system according to claim 1, wherein, The generation process of the preset variational quantum circuit comprises: obtaining historical time sequence running parameters of the physical system, and performing feature extraction on each historical time sequence running parameter to obtain historical time sequence features; Each of the historical time sequence features is mapped into a historical quantum feature by using the preset quantum state encoder, and an initial variational quantum circuit is trained by using each of the historical quantum features to obtain the preset variational quantum circuit, an input of the preset variational quantum circuit being a quantum feature, and an output of the preset variational quantum circuit being a quantum output state.

6. A simulation system of a digital twin system, characterized by, The simulation system of the digital twin system comprises: The parameter extraction module is configured to input the obtained simulation control instruction into a preset large language model to obtain each control target identifier and a control intent parameter; The simulation control module is configured to determine a target entity corresponding to the control target identifier in a twin graph based on the control target identifier by using a preset quantum graph search algorithm, and perform a simulation control operation on the target entity based on the control intent parameter, wherein the twin graph comprises each entity and a topological structure of each entity of a physical system simulated by the digital twin system, and the simulation control operation comprises: controlling the target entity in the twin graph to perform an action adapted to the control intent parameter, and obtaining time sequence running parameters of each entity in the twin graph at a current time; inputting each of the time sequence running parameters into a preset quantum feature extraction model to obtain a quantum output state corresponding to each of the time sequence running parameters, the quantum output state representing probability values of different running states of the entity after the entity performs the action adapted to the control intent parameter, and the preset quantum feature extraction model comprising a preset time sequence feature extraction model, a preset quantum state encoder, and a preset variational quantum circuit; inputting each of the measured quantum output states into a preset multilayer perception machine to obtain a predicted running state of the digital twin system after the simulation control operation is performed.

7. The simulation system of claim 6, wherein, When the simulation control module inputs each of the time sequence running parameters into the preset quantum feature extraction model to obtain the quantum output state corresponding to each of the time sequence running parameters, the simulation control module is configured to: extract a time sequence feature corresponding to each of the time sequence running parameters by using the preset time sequence feature extraction model; map each of the time sequence features into a quantum feature by using the preset quantum state encoder; input each of the quantum features into the preset variational quantum circuit to obtain the quantum output state corresponding to each of the time sequence running parameters.

8. A computer program product, characterised in that, The computer readable instructions, when executed on an electronic device, cause the electronic device to implement the simulation method of the digital twin system according to any one of claims 1 to 5.

9. An electronic device, comprising: The memory is configured to store a computer program, and the processor is configured to execute the computer program to enable the electronic device to implement the simulation method of the digital twin system according to any one of claims 1 to 5. The storage medium carries one or more computer programs, and when the one or more computer programs are executed by an electronic device, the electronic device can implement the simulation method of the digital twin system according to any one of claims 1 to 5. ​ 10. A computer storage medium, characterized in that, ​

Images