An AI engine-based multi-signal digital twin processing method
By employing an AI engine-based multi-signal digital twin processing method, utilizing multi-source sensors and neural network models, and combining digital twin modeling technology, the control strategy of the air-supported dome stadium is dynamically adjusted, solving the problems of insufficient real-time performance and stability in existing technologies, and achieving efficient audio system control.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANJING PIONE HIGH TECH
- Filing Date
- 2025-07-17
- Publication Date
- 2026-07-07
AI Technical Summary
In the processing of high-fidelity audio systems in air-supported dome stadiums, existing technologies are insufficient to meet real-time requirements. Purely data-driven AI models lack physical constraints, resulting in control lag and an inability to adapt to time-varying states.
The method employs a multi-signal digital twin processing approach based on an AI engine. It collects physical state signals of the air-supported dome stadium through multiple source sensors, generates joint feature vectors using a trained neural network model, and constructs a multi-physics coupling model in conjunction with a digital twin modeling engine. The control strategy is then dynamically adjusted to output the target audio signal.
Real-time adaptive control was achieved, solving the problem of mismatch between spatiotemporal scales, reducing computational complexity, and ensuring system stability through embedded physical constraints.
Smart Images
Figure CN120764388B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing technology, and in particular to a multi-signal digital twin processing method based on an AI engine. Background Technology
[0002] Air-supported dome stadiums are a new type of architectural form that uses special architectural membrane materials as its outer shell, with the entire structure supported by internal air pressure. This type of architecture is increasingly widely used in sports stadiums, exhibition centers, and other fields. Acoustic air-supported dome stadiums are an innovative architectural form specifically designed to solve the acoustic problems of traditional air-supported dome stadiums. Compared with ordinary air-supported dome stadiums, acoustic air-supported dome stadiums use specialized acoustic materials and technologies to effectively control sound reflection, absorption, and diffusion, ensuring more uniform and clear audio transmission.
[0003] In the high-fidelity audio system processing, the air-supported membrane structure (ABS) achieves the final output function. Heterogeneous signals such as sound pressure, deformation, temperature, and airflow within the ABS all affect the final output, and multiple data points are interconnected. However, simulations based on physical equations are computationally complex and struggle to meet real-time requirements, while purely data-driven AI models lack physical constraints and are prone to failure under extreme conditions. Therefore, actuator command generation relies on manual parameter tuning, which cannot adapt to the time-varying state of the ABS, resulting in control lag. Summary of the Invention
[0004] This application provides a multi-signal digital twin processing method based on an AI engine to improve the above-mentioned problems.
[0005] To achieve the above objectives, this application adopts the following technical solution:
[0006] Firstly, this application proposes a multi-signal digital twin processing method based on an AI engine, applicable to an audio system. The audio system includes a control terminal, multiple source sensors, and an execution terminal, where the execution terminal includes an air-supported dome stadium. The method includes:
[0007] The control terminal controls the execution terminal to execute the first control strategy at the first time node, so that the execution terminal outputs the target audio signal and collects the physical state signals of the air-supported membrane venue based on multi-source sensors. The physical state signals include local strain signals, acoustic radiation signals, temperature gradient signals and airflow signals.
[0008] The control terminal inputs the physical state signal into the trained target neural network model, whereby the target neural network model outputs a joint feature vector based on the input physical state signal.
[0009] The control terminal constructs a multi-physics coupling model of the air film based on joint feature vectors and a digital twin modeling engine. The multi-physics coupling model includes multiple simulated physical state data.
[0010] The control terminal acquires multiple simulated physical state data at the second time node based on the multiphysics coupling model, and obtains control commands based on the simulated physical state data. The second time node is the time node after the first time node.
[0011] The control terminal sends control commands to the execution terminal, causing the execution terminal to execute the second control strategy and output the target audio signal.
[0012] In conjunction with the first aspect, in some embodiments, the control terminal inputs a physical state signal to a trained target neural network model, wherein the target neural network model is used to output a joint feature vector based on the input physical state signal, including:
[0013] The control terminal performs time-frequency domain encoding on the local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal, respectively.
[0014] The control terminal dynamically fuses heterogeneous signal features through attention weights, reducing the dimensionality of the fused high-dimensional features to a joint feature vector of fixed dimensions.
[0015] In conjunction with the first aspect, in some implementations, the control terminal dynamically fuses heterogeneous signal features through attention weights, reducing the fused high-dimensional features to a joint feature vector of fixed dimensions, including:
[0016] The control terminal unifies the time reference by interpolating the acoustic radiation signal and the temperature gradient signal in the time domain, and aligns the local strain signal and the airflow signal in the coordinate system by spatial remapping.
[0017] The control terminal acquires the first spatiotemporal coupling strength between the acoustic radiation signal and the local strain signal, and the second spatiotemporal coupling strength between the temperature gradient signal and the airflow signal;
[0018] Based on the first and second spatiotemporal coupling strengths, the control terminal assigns different attention weights to local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals. Among them, the physical state signal with the larger attention weight is more correlated with the state of the air-supported membrane structure.
[0019] The control terminal superimposes the weighted local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal, and outputs a joint feature vector of fixed dimensions.
[0020] In conjunction with the first aspect, in some implementations, the control terminal superimposes the weighted local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal, and outputs a joint feature vector of fixed dimension, including:
[0021] The control terminal divides the air film surface into a grid array corresponding to the spatial location, and superimposes the acoustic radiation signal, temperature gradient signal and airflow signal, and outputs a fixed dimension mapping to the grid array. Among them, the physical state signal with higher attention weight occupies a larger channel dimension in the grid array.
[0022] The control terminal adds synchronization timestamps to local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals.
[0023] The control terminal acquires data from a fixed-dimensional grid as a joint feature vector.
[0024] In conjunction with the first aspect, in some implementations, the control terminal divides the air film surface into a grid array corresponding to spatial locations, superimposes the acoustic radiation signal, temperature gradient signal, and airflow signal, and outputs a fixed-dimensional mapping onto the grid array. The physical state signal with a higher attention weight occupies a larger channel dimension in the grid array, including:
[0025] When any one or more of the local strain signal, acoustic radiation signal, and temperature gradient signal are abnormal, the dimension corresponding to the abnormal physical state signal increases.
[0026] In conjunction with the first aspect, in some implementations, the control terminal acquires multiple simulated physical state data at a second time node based on a multiphysics coupling model, and acquires control commands based on the simulated physical state data, wherein the second time node is a time node following the first time node, including:
[0027] The control terminal obtains the multiphysics coupling model at the second time node based on the number of simulated physical states, and obtains the output audio signal of the multiphysics coupling model according to the multiphysics coupling model;
[0028] The control terminal, based on a multiphysics coupling model, acquires the target physical signal of the air-supported membrane structure at the second time node when it outputs the target audio signal.
[0029] Based on the target physical signal and multiple simulated physical state data at the second time node, the control command is determined.
[0030] In conjunction with the first aspect, in some implementations, control commands are determined based on the target physical signal and multiple simulated physical state data at a second time point, wherein the control commands include tension regulation, damping compensation, or drive signal correction commands for the air-supported membrane structure.
[0031] Secondly, this application proposes an audio system, which includes a control terminal, multi-source sensors, and an execution terminal, wherein the execution terminal includes an air-supported membrane structure. The system is configured as follows:
[0032] The control terminal controls the execution terminal to execute the first control strategy at the first time node, so that the execution terminal outputs the target audio signal and collects the physical state signals of the air-supported membrane venue based on multi-source sensors. The physical state signals include local strain signals, acoustic radiation signals, temperature gradient signals and airflow signals.
[0033] The control terminal inputs the physical state signal into the trained target neural network model, whereby the target neural network model outputs a joint feature vector based on the input physical state signal.
[0034] The control terminal constructs a multi-physics coupling model of the air film based on joint feature vectors and a digital twin modeling engine. The multi-physics coupling model includes multiple simulated physical state data.
[0035] The control terminal acquires multiple simulated physical state data at the second time node based on the multiphysics coupling model, and obtains control commands based on the simulated physical state data. The second time node is the time node after the first time node.
[0036] The control terminal sends control commands to the execution terminal, causing the execution terminal to execute the second control strategy and output the target audio signal.
[0037] In conjunction with the second aspect, in some implementations, the system is configured as follows:
[0038] The control terminal inputs the physical state signal into the trained target neural network model, whereby the target neural network model outputs a joint feature vector based on the input physical state signal, including:
[0039] The control terminal performs time-frequency domain encoding on the local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal, respectively.
[0040] The control terminal dynamically fuses heterogeneous signal features through attention weights, reducing the dimensionality of the fused high-dimensional features to a joint feature vector of fixed dimensions.
[0041] In conjunction with the second aspect, in some implementations, the system is configured as follows:
[0042] The control terminal dynamically fuses heterogeneous signal features through attention weights, reducing the fused high-dimensional features to a fixed-dimensional joint feature vector, including:
[0043] The control terminal unifies the time reference by interpolating the acoustic radiation signal and the temperature gradient signal in the time domain, and aligns the local strain signal and the airflow signal in the coordinate system by spatial remapping.
[0044] The control terminal acquires the first spatiotemporal coupling strength between the acoustic radiation signal and the local strain signal, and the second spatiotemporal coupling strength between the temperature gradient signal and the airflow signal;
[0045] Based on the first and second spatiotemporal coupling strengths, the control terminal assigns different attention weights to local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals. Among them, the physical state signal with the larger attention weight is more correlated with the state of the air-supported membrane structure.
[0046] The control terminal superimposes the weighted local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal, and outputs a joint feature vector of fixed dimensions.
[0047] In conjunction with the second aspect, in some implementations, the system is configured as follows:
[0048] The control terminal superimposes the weighted local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal, and outputs a fixed-dimensional joint feature vector, including:
[0049] The control terminal divides the air film surface into a grid array corresponding to the spatial location, and superimposes the acoustic radiation signal, temperature gradient signal and airflow signal, and outputs a fixed dimension mapping to the grid array. Among them, the physical state signal with higher attention weight occupies a larger channel dimension in the grid array.
[0050] The control terminal adds synchronization timestamps to local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals.
[0051] The control terminal acquires data from a fixed-dimensional grid as a joint feature vector.
[0052] In conjunction with the second aspect, in some implementations, the system is configured as follows:
[0053] The control terminal divides the air film surface into a grid array corresponding to spatial locations, superimposes the acoustic radiation signal, temperature gradient signal, and airflow signal, and outputs a fixed-dimensional mapping onto the grid array. The physical state signal with higher attention weight occupies a larger channel dimension in the grid array, including:
[0054] When any one or more of the local strain signal, acoustic radiation signal, and temperature gradient signal are abnormal, the dimension corresponding to the abnormal physical state signal increases.
[0055] In conjunction with the second aspect, in some implementations, the system is configured as follows:
[0056] The control terminal acquires multiple simulated physical state data at the second time node based on a multiphysics coupling model, and obtains control commands based on the simulated physical state data. The second time node is a time node following the first time node, including:
[0057] The control terminal obtains the multiphysics coupling model at the second time node based on the number of simulated physical states, and obtains the output audio signal of the multiphysics coupling model according to the multiphysics coupling model;
[0058] The control terminal, based on a multiphysics coupling model, acquires the target physical signal of the air-supported membrane structure at the second time node when it outputs the target audio signal.
[0059] Based on the target physical signal and multiple simulated physical state data at the second time node, the control command is determined.
[0060] In conjunction with the second aspect, in some implementations, the system is configured as follows:
[0061] Based on the target physical signal and multiple simulated physical state data at the second time node, control commands are determined, including tension regulation, damping compensation, or drive signal correction commands for the air-supported membrane structure.
[0062] A third aspect of this invention provides an electronic device, which includes:
[0063] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method proposed in the first aspect of the present invention.
[0064] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method as described in the first aspect of the present invention.
[0065] In summary, the above method has the following technical effects:
[0066] This application proposes a multi-signal digital twin processing method based on an AI engine. It collects physical state signals of an air-supported membrane structure from multiple sources, including local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals. These physical state signals are input into a trained target neural network model, which outputs a joint feature vector. A multi-physics coupling model of the air-supported membrane is constructed based on this joint feature vector and a digital twin modeling engine. Control commands are then obtained from the simulated physical state data and finally sent to the execution terminal to execute a second control strategy and output the target audio signal. This AI engine-based multi-signal digital twin processing method solves the problem of spatiotemporal scale mismatch by dynamically fusing multi-source signals through an attention mechanism. Furthermore, feature compression technology significantly reduces computational complexity, and embedded physical constraints ensure system stability. Attached Figure Description
[0067] Figure 1 This is a flowchart illustrating a multi-signal digital twin processing method based on an AI engine proposed in this application. Detailed Implementation
[0068] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0069] This application proposes a multi-signal digital twin processing method based on an AI engine, applicable to an audio system. The audio system includes a control terminal, multi-source sensors, and an execution terminal. The execution terminal includes an air-supported dome stadium. Please refer to [link to relevant documentation]. Figure 1 The method includes steps S101-S105. It should be noted that the execution subject in this embodiment is the execution terminal. In other embodiments, it can be other subjects, such as cloud servers, etc., which are not limited here.
[0070] S101: At the first time node, control the execution terminal to execute the first control strategy so that the execution terminal outputs the target audio signal and collects the physical state signal of the air-supported membrane venue based on multi-source sensors, wherein the physical state signal includes local strain signal, acoustic radiation signal, temperature gradient signal and airflow signal.
[0071] Understandably, physical state signals comprehensively reflect the overall performance of the air-supported membrane structure under different working conditions. Through high-precision acquisition by multi-source sensors, the local strain of the air-supported membrane structure can be obtained in real time, that is, the minute deformation it produces when driven by audio signals. At the same time, the acoustic radiation signal reveals the characteristics of the radiated sound waves of the air-supported membrane structure, the temperature gradient signal reflects the heat distribution of the air-supported membrane structure during operation, which helps to identify potential overheating risks, and the airflow signal reveals the airflow dynamics around the air-supported membrane structure.
[0072] S102: Input the physical state signal into the trained target neural network model, wherein the target neural network model is used to output a joint feature vector based on the input physical state signal.
[0073] Understandably, heterogeneous signals refer to multi-source sensor data with different physical properties. Dynamic fusion refers to assigning differentiated weights to different signal features in real time based on the physical correlation between signals, generating a unified representation vector.
[0074] Specifically, in this embodiment, the local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal are encoded in the time-frequency domain. For example, the acoustic radiation signal (high-frequency sampling) and the temperature gradient signal (low-frequency sampling) are unified in time reference using a time-domain interpolation module. Of course, the local strain signal (spatial grid distribution) and the airflow signal (point detection) can also be aligned in coordinate system using a spatial remapping module.
[0075] Specifically, the first spatiotemporal coupling strength between the acoustic radiation signal and the local strain signal is obtained, and the second spatiotemporal coupling strength between the temperature gradient signal and the airflow signal is obtained. For example, the first spatiotemporal coupling strength is whether the vibration at a certain point of the air-supported membrane structure is accompanied by a peak sound pressure, and the second spatiotemporal coupling strength is whether the injection of cold airflow causes a change in the stiffness of the air-supported membrane structure.
[0076] Then, by dynamically fusing heterogeneous signal features through attention weights, the fused high-dimensional features are reduced to a joint feature vector of fixed dimensions.
[0077] For example, based on the first and second spatiotemporal coupling strengths, different attention weights are assigned to local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals. The physical state signal with the higher its attention weight is, the stronger its correlation with the state of the air-supported membrane structure. For instance, features strongly correlated with the current state of the air-supported membrane (such as strain signals in areas of severe deformation) receive higher weights.
[0078] Noise interference or weakly correlated features (such as thermal signals in temperature-stable regions) will have lower weights. Of course, the weight values can be updated in real time according to the operating status of the air-supported membrane structure.
[0079] The weighted local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal are superimposed, and a joint feature vector with a fixed dimension is output.
[0080] Understandably, the weighted local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal are superimposed to ensure that each signal is adjusted according to a predetermined weight to reflect its importance in the overall system. After weighting and superposition, a fixed-dimensional joint feature vector is generated. This joint feature vector is the product of fusing signals from different physical quantities, and it can comprehensively reflect the system's state and performance. In this way, multi-source information can be effectively integrated, providing a comprehensive and accurate data foundation for subsequent decision-making and control.
[0081] For example, the surface of the air film is divided into a grid array corresponding to spatial locations. Then, the acoustic radiation signal, temperature gradient signal, and airflow signal are superimposed, and the output is mapped to the grid array with a fixed dimension. The physical state signal with a higher attention weight occupies a larger channel dimension in the grid array.
[0082] Synchronization timestamps are added to local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals. Data occupied by a fixed-dimensional grid is obtained as a joint feature vector.
[0083] S103: Construct a multi-physics coupling model of the air film based on joint feature vectors and a digital twin modeling engine. The multi-physics coupling model includes multiple simulated physical state data.
[0084] By combining joint feature vectors with a digital twin modeling engine, a multiphysics coupling model for simulating air-supported dome stadiums was constructed. This multiphysics coupling model not only includes simulation data of the air-supported dome stadium under different physical conditions, but also accurately reflects the physical behavior and interactions of the diaphragm during actual operation.
[0085] For example, a two-dimensional digital grid coordinate system can be established based on the surface of the audio air film. Taking a 100×100 grid as an example, four types of signals can be mapped to the grid: local strain signals correspond to marking the degree of deformation at the grid points, acoustic radiation signals correspond to marking the sound pressure intensity at the grid points, temperature gradient signals correspond to recording the temperature values at the grid points, and airflow signals correspond to marking the airflow velocity direction at the grid points. For example, when the acoustic radiation signal detects vibration, the temperature signal generates a corresponding time estimate using data from adjacent time points.
[0086] S104: Based on the multiphysics coupling model, obtain multiple simulated physical state data at the second time node, and obtain control commands based on the simulated physical state data, wherein the second time node is the time node after the first time node.
[0087] Understandably, a multiphysics coupling model is used to obtain multiple simulated physical state data at the second time point. This data reflects the physical state of the system at the second time point, i.e., after the first time point. Based on this simulated physical state data, corresponding control commands are further calculated and generated. These control commands aim to adjust or optimize the system's behavior to ensure it operates according to predetermined parameters.
[0088] Specifically, the multiphysics coupling model for the second time node is obtained based on the number of simulated physical states, and the output audio signal of the multiphysics coupling model is obtained based on the multiphysics coupling model. Then, based on the multiphysics coupling model, the target physical signal of the air-supported membrane stadium when it outputs the target audio signal at the second time node is obtained. Finally, the control command is determined based on the target physical signal and the multiple simulated physical state data of the second time node.
[0089] In some implementations, control commands include tension regulation, damping compensation, or drive signal correction commands for the air-supported membrane structure.
[0090] S105: Send a control command to the execution terminal so that the execution terminal executes the second control strategy and outputs the target audio signal.
[0091] This application proposes a multi-signal digital twin processing method based on an AI engine. It collects physical state signals of an air-supported membrane structure from multiple sources, including local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals. These physical state signals are input into a trained target neural network model, which outputs a joint feature vector. A multi-physics coupling model of the air-supported membrane is constructed based on this joint feature vector and a digital twin modeling engine. Control commands are then obtained from the simulated physical state data and finally sent to the execution terminal to execute a second control strategy and output the target audio signal. This AI engine-based multi-signal digital twin processing method solves the problem of spatiotemporal scale mismatch by dynamically fusing multi-source signals through an attention mechanism. Furthermore, feature compression technology significantly reduces computational complexity, and embedded physical constraints ensure system stability.
[0092] Secondly, this application proposes an audio system, which includes a control terminal, multi-source sensors, and an execution terminal, wherein the execution terminal includes an air-supported membrane structure. The system is configured as follows:
[0093] The control terminal controls the execution terminal to execute the first control strategy at the first time node, so that the execution terminal outputs the target audio signal and collects the physical state signals of the air-supported membrane venue based on multi-source sensors. The physical state signals include local strain signals, acoustic radiation signals, temperature gradient signals and airflow signals.
[0094] The control terminal inputs the physical state signal into the trained target neural network model, whereby the target neural network model outputs a joint feature vector based on the input physical state signal.
[0095] The control terminal constructs a multi-physics coupling model of the air film based on joint feature vectors and a digital twin modeling engine. The multi-physics coupling model includes multiple simulated physical state data.
[0096] The control terminal acquires multiple simulated physical state data at the second time node based on the multiphysics coupling model, and obtains control commands based on the simulated physical state data. The second time node is the time node after the first time node.
[0097] The control terminal sends control commands to the execution terminal, causing the execution terminal to execute the second control strategy and output the target audio signal.
[0098] In some implementations, the system is configured as follows:
[0099] The control terminal inputs the physical state signal into the trained target neural network model, whereby the target neural network model outputs a joint feature vector based on the input physical state signal, including:
[0100] The control terminal performs time-frequency domain encoding on the local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal, respectively.
[0101] The control terminal dynamically fuses heterogeneous signal features through attention weights, reducing the dimensionality of the fused high-dimensional features to a joint feature vector of fixed dimensions.
[0102] In some implementations, the system is configured as follows:
[0103] The control terminal dynamically fuses heterogeneous signal features through attention weights, reducing the fused high-dimensional features to a fixed-dimensional joint feature vector, including:
[0104] The control terminal unifies the time reference by interpolating the acoustic radiation signal and the temperature gradient signal in the time domain, and aligns the local strain signal and the airflow signal in the coordinate system by spatial remapping.
[0105] The control terminal acquires the first spatiotemporal coupling strength between the acoustic radiation signal and the local strain signal, and the second spatiotemporal coupling strength between the temperature gradient signal and the airflow signal;
[0106] Based on the first and second spatiotemporal coupling strengths, the control terminal assigns different attention weights to local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals. Among them, the physical state signal with the larger attention weight is more correlated with the state of the air-supported membrane structure.
[0107] The control terminal superimposes the weighted local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal, and outputs a joint feature vector of fixed dimensions.
[0108] In some implementations, the system is configured as follows:
[0109] The control terminal superimposes the weighted local strain signal, acoustic radiation signal, temperature gradient signal, and airflow signal, and outputs a fixed-dimensional joint feature vector, including:
[0110] The control terminal divides the air film surface into a grid array corresponding to the spatial location, and superimposes the acoustic radiation signal, temperature gradient signal and airflow signal, and outputs a fixed dimension mapping to the grid array. Among them, the physical state signal with higher attention weight occupies a larger channel dimension in the grid array.
[0111] The control terminal adds synchronization timestamps to local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals.
[0112] The control terminal acquires data from a fixed-dimensional grid as a joint feature vector.
[0113] In some implementations, the system is configured as follows:
[0114] The control terminal divides the air film surface into a grid array corresponding to spatial locations, superimposes the acoustic radiation signal, temperature gradient signal, and airflow signal, and outputs a fixed-dimensional mapping onto the grid array. The physical state signal with higher attention weight occupies a larger channel dimension in the grid array, including:
[0115] When any one or more of the local strain signal, acoustic radiation signal, and temperature gradient signal are abnormal, the dimension corresponding to the abnormal physical state signal increases.
[0116] In some implementations, the system is configured as follows:
[0117] The control terminal acquires multiple simulated physical state data at the second time node based on a multiphysics coupling model, and obtains control commands based on the simulated physical state data. The second time node is a time node following the first time node, including:
[0118] The control terminal obtains the multiphysics coupling model at the second time node based on the number of simulated physical states, and obtains the output audio signal of the multiphysics coupling model according to the multiphysics coupling model;
[0119] The control terminal, based on a multiphysics coupling model, acquires the target physical signal of the air-supported membrane structure at the second time node when it outputs the target audio signal.
[0120] Based on the target physical signal and multiple simulated physical state data at the second time node, the control command is determined.
[0121] In some implementations, the system is configured as follows:
[0122] Based on the target physical signal and multiple simulated physical state data at the second time node, control commands are determined, including tension regulation, damping compensation, or drive signal correction commands for the air-supported membrane structure.
[0123] This application proposes an audio system that collects physical state signals of an air-supported membrane structure using multi-source sensors. These physical state signals include local strain signals, acoustic radiation signals, temperature gradient signals, and airflow signals. The physical state signals are input into a trained target neural network model, which outputs a joint feature vector. Based on this joint feature vector and a digital twin modeling engine, a multiphysics coupling model of the air-supported membrane is constructed. Control commands are obtained from the simulated physical state data, and finally, control commands are sent to an execution terminal to execute a second control strategy and output the target audio signal. This proposed audio system solves the problem of spatiotemporal scale mismatch by dynamically fusing multi-source signals through an attention mechanism. Furthermore, feature compression technology significantly reduces computational complexity, and embedded physical constraints ensure system stability.
[0124] Based on the same inventive concept, embodiments of this application also propose an electronic device, which includes:
[0125] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the AI engine-based multi-signal digital twin processing method of the present application embodiments.
[0126] Furthermore, to achieve the above objectives, embodiments of this application also propose a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the AI engine-based multi-signal digital twin processing method of this application.
[0127] The following is a detailed introduction to the various components of the electronic device:
[0128] In this context, the processor is the control center of the electronic device. It can be a single processor or a collective term for multiple processing elements. For example, a processor can be one or more central processing units (CPUs), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).
[0129] Alternatively, the processor can perform various functions of the electronic device by running or executing software programs stored in memory and by calling data stored in memory.
[0130] The memory is used to store the software program that executes the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can be referred to the above method embodiment, which will not be repeated here.
[0131] Optionally, the memory can be read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions, random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory can be integrated with the processor or exist independently and coupled to the processor through the interface circuit of the electronic device; the embodiments of the present invention do not specifically limit this.
[0132] A transceiver is used to communicate with network devices or with terminal devices.
[0133] Optionally, the transceiver may include a receiver and a transmitter. The receiver is used to implement the receiving function, and the transmitter is used to implement the sending function.
[0134] Optionally, the transceiver can be integrated with the processor or exist independently and coupled to the processor through the router's interface circuit. This embodiment of the invention does not specifically limit this.
[0135] Furthermore, the technical effects of the electronic device can be referred to the technical effects of the data transmission method in the above method embodiments, and will not be repeated here.
[0136] It should be understood that the processor in the embodiments of the present invention can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.
[0137] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).
[0138] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.
[0139] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.
[0140] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.
[0141] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.
[0142] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
Claims
1. A multi-signal digital twin processing method based on an AI engine, characterized in that, An audio system is applicable, the audio system including a control terminal, a multi-source sensor, and an execution terminal, the execution terminal including an air-supported dome stadium, the method comprising: The control terminal controls the execution terminal to execute a first control strategy at a first time point, so that the execution terminal outputs a target audio signal and collects the physical state signal of the air-supported membrane structure based on the multi-source sensor, wherein the physical state signal includes local strain signal, acoustic radiation signal, temperature gradient signal and airflow signal; The control terminal inputs the physical state signal into the trained target neural network model, wherein the target neural network model is used to output a joint feature vector based on the input physical state signal, specifically including: The control terminal performs time-frequency domain encoding on the local strain signal, the acoustic radiation signal, the temperature gradient signal, and the airflow signal, respectively. The control terminal dynamically fuses heterogeneous signal features through attention weights, reducing the fused high-dimensional features to a fixed-dimensional joint feature vector, including: The control terminal unifies the acoustic radiation signal and the temperature gradient signal into a unified time reference through time-domain interpolation, and aligns the local strain signal and the airflow signal into a coordinate system through spatial remapping. The control terminal acquires the first spatiotemporal coupling strength between the acoustic radiation signal and the local strain signal, and acquires the second spatiotemporal coupling strength between the temperature gradient signal and the airflow signal; Based on the first spatiotemporal coupling strength and the second spatiotemporal coupling strength, the control terminal assigns different attention weights to the local strain signal, the acoustic radiation signal, the temperature gradient signal and the airflow signal. The physical state signal with a larger attention weight is more correlated with the state of the air-supported membrane structure. The control terminal superimposes the weighted local strain signal, the acoustic radiation signal, the temperature gradient signal, and the airflow signal, and outputs the joint feature vector of fixed dimension. The control terminal constructs a multi-physics coupling model of the air film based on the joint feature vector and the digital twin modeling engine. The multi-physics coupling model includes multiple simulated physical state data. The control terminal acquires multiple simulated physical state data at a second time node based on the multiphysics coupling model, and acquires control commands based on the simulated physical state data, wherein the second time node is a time node after the first time node; The control terminal sends the control command to the execution terminal so that the execution terminal executes the second control strategy and outputs the target audio signal.
2. The multi-signal digital twin processing method based on an AI engine according to claim 1, characterized in that, The control terminal superimposes the weighted local strain signal, the acoustic radiation signal, the temperature gradient signal, and the airflow signal, and outputs the joint feature vector of fixed dimension, including: The control terminal divides the air film surface into a grid array corresponding to the spatial location, and superimposes the acoustic radiation signal, the temperature gradient signal and the airflow signal, and outputs a fixed dimension mapping to the grid array. The physical state signal with a higher attention weight occupies a larger channel dimension in the grid array. The control terminal adds a synchronization timestamp to the local strain signal, the acoustic radiation signal, the temperature gradient signal, and the airflow signal; The control terminal acquires the data occupied by the grid of fixed dimensions as the joint feature vector.
3. The multi-signal digital twin processing method based on an AI engine according to claim 2, characterized in that, The control terminal divides the air film surface into a grid array corresponding to spatial locations, and superimposes the acoustic radiation signal, the temperature gradient signal, and the airflow signal, outputting a fixed-dimensional mapping onto the grid array. The physical state signal with a higher attention weight occupies a larger channel dimension in the grid array, including: When any one or more of the local strain signal, the acoustic radiation signal, and the temperature gradient signal become abnormal, the dimension corresponding to the abnormal physical state signal increases.
4. The multi-signal digital twin processing method based on an AI engine according to claim 1, characterized in that, The control terminal acquires multiple simulated physical state data at a second time node based on the multiphysics coupling model, and obtains control commands according to the simulated physical state data. The second time node is a time node following the first time node, including: The control terminal obtains the simulated physical state data of the second time node based on the multiphysics coupling model, and obtains the output audio signal of the multiphysics coupling model according to the multiphysics coupling model; The control terminal, based on the multiphysics coupling model, obtains the target physical signal of the air-supported membrane structure at the second time node when it outputs the target audio signal corresponding to the second control strategy. The control command is determined based on the target physical signal and multiple simulated physical state data at the second time node.
5. The multi-signal digital twin processing method based on an AI engine according to claim 4, characterized in that, Based on the target physical signal and multiple simulated physical state data at the second time node, the control command is determined, wherein the control command includes tension adjustment, damping compensation, or drive signal correction command for the air-supported membrane structure.
6. An audio system, characterized in that, For executing the AI engine-based multi-signal digital twin processing method as described in claim 1, the audio system includes a control terminal, multi-source sensors, and an execution terminal, wherein the execution terminal includes an air-supported dome stadium, and the system is configured to: The control terminal controls the execution terminal to execute a first control strategy at a first time point, so that the execution terminal outputs a target audio signal and collects the physical state signal of the air-supported membrane structure based on the multi-source sensor, wherein the physical state signal includes local strain signal, acoustic radiation signal, temperature gradient signal and airflow signal; The control terminal inputs the physical state signal into the trained target neural network model, wherein the target neural network model is used to output a joint feature vector based on the input physical state signal; The control terminal constructs a multi-physics coupling model of the air film based on the joint feature vector and the digital twin modeling engine. The multi-physics coupling model includes multiple simulated physical state data. The control terminal acquires multiple simulated physical state data at a second time node based on the multiphysics coupling model, and acquires control commands based on the simulated physical state data, wherein the second time node is a time node after the first time node; The control terminal sends the control command to the execution terminal so that the execution terminal executes the second control strategy and outputs the target audio signal.
7. An electronic device, characterized in that, Includes at least one processor; And a memory communicatively connected to at least one of the processors; wherein the memory stores instructions executable by at least one of the processors, the instructions being executed by at least one of the processors to enable at least one of the processors to perform the method as claimed in any one of claims 1-5.
8. A computer-readable storage medium, characterized in that... It stores a computer program that, when executed by a processor, implements the method as claimed in any one of claims 1-5.