An unmanned aerial vehicle performance monitoring platform and method based on digital twinning

Abstract

The application discloses a kind of based on digital twinning unmanned aerial vehicle performance monitoring platform and method, belong to digital twinning technical field.Method is: construct high-fidelity three-dimensional geometric model and agent model;Build digital twinning platform three-dimensional scene and visual environment;Flight data are collected and real-time performance prediction is carried out;Three-dimensional dynamic rendering and interactive feedback are executed.Through unmanned aerial vehicle performance monitoring platform is realized, with digital twinning platform as core, integrates physical entity and simulation module, data acquisition and communication module, data processing and conversion module, agent model calculation module, three-dimensional visualization and man-machine interaction module, each functional module carries out data interaction and collaborative work.The present application can realize the leap from limited point measurement to full field perception, reduce the dependence on complex sensor network and cost while ensuring monitoring accuracy, improve the integrity, real-time performance and intuitiveness of unmanned aerial vehicle performance monitoring, and provide an innovative solution for efficient flight safety monitoring and state evaluation.

Description

Technical Field

[0001] This invention belongs to the field of digital twin technology and relates to a performance monitoring platform and method for unmanned aerial vehicles based on digital twins. Background Technology

[0002] As a typical example of the integration of modern aviation technology and intelligent equipment, unmanned aerial vehicles (UAVs) have demonstrated wide application value in various fields such as geographic surveying, logistics transportation, and agricultural plant protection. Their mobility, ease of deployment, and adaptability to complex environments make them important tools for performing special tasks and improving operational efficiency. With the continuous expansion of UAV application scenarios and the increasing complexity of tasks, real-time and accurate monitoring and evaluation of their flight performance and structural health status has become a key technical requirement for ensuring flight safety, extending service life, and improving mission success rates.

[0003] Currently, UAV performance monitoring largely relies on sensor data acquisition and post-flight analysis, lacking real-time perception and visualization feedback of the dynamic response of the airframe structure in real flight environments. Although some research has attempted to apply structural health monitoring (SHM) methods to the UAV field, such as assessing local damage through vibration characteristic analysis and strain monitoring, these methods often suffer from strong model dependence, lagging data processing, and low system integration. Furthermore, existing technologies mostly focus on monitoring single physical quantities, failing to construct a closed-loop digital twin framework from sensor acquisition and model prediction to three-dimensional visualization feedback, thus limiting their promotion and application in practical engineering.

[0004] Chinese invention patent (publication number CN109711034B) provides a landing gear health monitoring method based on control theory, which extracts health characteristic parameters by establishing a physical model, demonstrating the advantages of the model-driven approach. Chinese invention patent (publication number CN119885438A), starting from aircraft structural integrity assessment, proposes a multi-index health status evaluation system, emphasizing the correlation between local and global states. While existing methods have achieved certain results in specific structures, they suffer from shortcomings in real-time monitoring, data comprehensiveness, intuitive visualization, and closed-loop decision support for early warning and lifespan prediction, limiting the real-time monitoring and early warning of the overall structural performance of unmanned aerial vehicles (UAVs).

[0005] Digital twin technology, as an important means to achieve bidirectional mapping and interaction between physical entities and virtual models, provides a new solution to the aforementioned problems. By constructing a high-fidelity virtual model of a drone and integrating multi-source sensor data and real-time simulation analysis, dynamic monitoring and predictive evaluation of multi-dimensional performance indicators such as the drone's structural state and aerodynamic loads can be achieved. Utilizing computer graphics and real-time rendering technology, the mechanical state information of the drone can be intuitively presented in the form of visualized cloud maps, significantly improving the efficiency of state perception and decision support. Summary of the Invention

[0006] To address the shortcomings of existing UAV structural health monitoring technologies, such as incomplete data-driven approaches, insufficient real-time performance, and weak visualization and interactivity, this invention provides a UAV performance monitoring platform and method based on digital twins. This invention integrates real-time sensor data with a high-performance proxy model to achieve real-time bidirectional mapping and driving between the physical entity and the virtual model. By collecting key flight parameters of the UAV and using a data-driven proxy model, it achieves efficient and rapid prediction of structural performance. Within the digital twin platform, it realizes real-time synchronous mapping of the UAV's multi-dimensional mechanical properties, dynamic marking of extreme value regions, and immersive visual interaction.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A method for monitoring the performance of unmanned aerial vehicles (UAVs) based on digital twins, the method comprising the following steps:

[0009] Step 1: Construct a high-fidelity 3D geometric model and proxy model. Specifically:

[0010] Step 1.1: Based on the physical structural characteristics of the UAV, a three-dimensional geometric model of the UAV is established using three-dimensional modeling software. By adjusting parameters and optimizing the design, the three-dimensional geometric model can accurately reflect the real structure of the UAV.

[0011] Step 1.2: Simplify the 3D geometric model established in Step 1.1, divide the simplified 3D geometric model into regional meshes to obtain the UAV mesh model, and export the number and spatial coordinates of each mesh node to the local data file.

[0012] Step 1.3: Determine the input variables and load parameter domain of the surrogate model by combining the inertial loads and aerodynamic loads experienced by the UAV in actual flight.

[0013] The surrogate model is an efficient approximation model built using machine learning algorithms, used to quickly predict the full-field mechanical response of an unmanned aerial vehicle (UAV) structure without performing a complete finite element method (FEM) solution. The input variable of the surrogate model is the equivalent load vector acting on the UAV, which consists of load components along the longitudinal, transverse, and vertical axes of the aircraft body, used to characterize the force characteristics under different flight conditions.

[0014] The value range of each component of the load vector is determined based on the typical flight attitude and aerodynamic range of the UAV, forming a load parameter domain. A characteristic load sample set is then generated within the load parameter domain using a spatial sampling method to cover the typical flight conditions of the UAV.

[0015] Step 1.4: Using the finite element analysis method, the multidimensional load parameters obtained in Step 1.3 are applied to the UAV mesh model divided in Step 1.2, and the full-field mechanical response data of the UAV under multiple working conditions are output.

[0016] Step 1.5: Based on the feature load sample set in Step 1.3 and the full-field mechanical response data calculated in Step 1.4, construct a training sample database and use the training sample database to train the surrogate model.

[0017] Step 1.6: Deploy an inertial measurement unit on the physical entity of the UAV to collect three-axis acceleration and pitch angle data.

[0018] Step 2: Build the 3D scene and visualization environment for the digital twin platform. Specifically:

[0019] Step 2.1: Build a digital twin platform and construct a 3D scene within the digital twin platform.

[0020] Step 2.2: Import the data file of the UAV mesh model exported in Step 1.2 into the digital twin platform, and construct the UAV 3D model in the 3D scene through parametric modeling.

[0021] Step 2.3: Create a user interface in the 3D scene, integrating mode switching buttons, real-time data charts, network settings panel, data management panel, and function buttons to complete the interface construction of the 3D visualization and human-computer interaction module.

[0022] Step 2.4: Configure camera control, data-driven rendering, and interface event response logic for the 3D visualization and human-computer interaction module. The camera control logic is bound to user input, the data-driven rendering logic is bound to the output data of the proxy model trained in Step 1.5, and the interface event response logic is bound to mode switching and extreme value marker switches, thus forming a data-driven real-time visualization closed loop.

[0023] Step 3: Collect flight data and perform real-time performance prediction. Specifically:

[0024] Step 3.1: During the flight of the UAV, the inertial measurement unit deployed in step 1.6 collects the three-axis acceleration component data in the body coordinate system in real time. , , With pitch angle data This yields the raw data. This contains the raw acceleration data along the longitudinal axis of the fuselage (head direction). This contains the raw acceleration data along the transverse axis of the aircraft (wing direction, usually right side is positive). The raw acceleration data is along the vertical axis of the fuselage (vertical to the downward direction of the fuselage). This is the angle between the longitudinal axis of the aircraft and the horizontal plane, i.e., the original data of the pitch angle.

[0025] Step 3.2: The digital twin platform built in Step 2.1 receives the raw data collected in Step 3.1 and extracts the effective triaxial acceleration component data within the digital twin platform. , , , and pitch angle data , thus obtaining valid data.

[0026] Step 3.3: Filter and denoise the valid data extracted in Step 3.2 to obtain the processed data. , , , .in, The processed acceleration data along the longitudinal axis of the machine body. The processed acceleration data along the transverse axis of the aircraft body. The processed acceleration data along the vertical axis of the aircraft. The data represents the processed pitch angle. Subsequently, based on flight dynamics principles, the processed data is converted into an equivalent payload vector acting on the UAV, yielding the real-time payload vector. ,in The conversion formula is as follows:

[0027] (1)

[0028] (2)

[0029] (3)

[0030] in, This represents the equivalent aerodynamic load component along the X-axis of the aircraft. The equivalent aerodynamic load component along the Y-axis of the fuselage. This represents the equivalent aerodynamic load component along the Z-axis of the fuselage. For predefined load conversion factors, This is the acceleration due to gravity.

[0031] Step 3.4: Convert the real-time load vector obtained in Step 3.3 into a vector. Input the data into the agent model trained in step 1.5 to quickly calculate and output the full-field mechanical response data of the UAV under the current flight state.

[0032] Step 4: Perform 3D dynamic rendering and interactive feedback. Specifically:

[0033] Step 4.1: Based on the 3D visualization and human-computer interaction module built in Step 2.3, the user can select the current display mode through the mode switching button. The digital twin platform will process the data group corresponding to the current mode in the UAV full-field mechanical response data output in Step 3.4, and render a dynamically changing performance cloud map in real time on the UAV 3D model constructed in Step 2.2 through the color mapping algorithm.

[0034] Step 4.2: Based on the performance cloud map rendered in real time in the current display mode in Step 4.1, the user can use the extreme value marker switch to enable the digital twin platform to traverse the full-field mechanical response data output in Step 3.4, automatically identify the node with the largest value in the data group corresponding to the current mode, and instantiate or update a colored marker sphere at the spatial location of the node in the three-dimensional scene for dynamic highlighting.

[0035] Step 4.3: Based on the camera control logic configured in the 3D visualization and human-computer interaction module built in Step 2.3, users can rotate and zoom the viewpoint within the digital twin platform using a mouse.

[0036] A digital twin-based drone performance monitoring platform is provided. This platform implements the aforementioned digital twin-based drone performance monitoring method. The platform is centered on a digital twin platform and integrates five functional modules: a physical entity and simulation module, a data acquisition and communication module, a data processing and conversion module, a proxy model calculation module, and a 3D visualization and human-computer interaction module. Each functional module interacts and collaborates through the digital twin platform. Specifically:

[0037] The physical entity and simulation module consists of a UAV physical entity, an inertial measurement unit deployed on it, and a simulation calculation part for finite element analysis. On the one hand, the physical entity and simulation module uses the inertial measurement unit to collect key parameters such as the three-axis acceleration and pitch angle of the UAV in real time during flight; on the other hand, it performs finite element analysis to obtain full-field mechanical response data under multiple operating conditions.

[0038] The data acquisition and communication module establishes a connection with the physical entity of the UAV through a stable network communication protocol. It is used to receive raw data from the sensors in real time and transmit it to the digital twin platform in a low-latency and high-reliability manner. It also has the function of local data storage and management.

[0039] The data processing and conversion module is integrated into the digital twin platform. The data processing and conversion module filters and denoises the effective sensor data transmitted by the data acquisition and communication module, and based on the principles of flight dynamics, converts the triaxial acceleration and pitch angle data into real-time load vectors acting on the UAV structure, providing an input interface for the proxy model calculation module.

[0040] The surrogate model calculation module is the core calculation module in the digital twin platform. This module is trained based on a sample database generated by the physical entity and simulation modules to construct a surrogate model. The surrogate model receives real-time load vectors and quickly outputs the UAV's full-field mechanical response data under the current flight state.

[0041] The 3D visualization and human-computer interaction module uses a color mapping algorithm to render the full-field mechanical response data output by the proxy model calculation module into a dynamically changing performance cloud map in real time on the digital twin platform, and provides interactive functions including viewpoint control, extreme value marking, and mode switching.

[0042] The beneficial effects of this invention are as follows:

[0043] (1) This invention constructs a real-time calculation system for the overall performance of UAVs based on lightweight sensor data. By collecting a few key parameters such as acceleration and pitch angle during UAV flight, a high-performance proxy model can be driven to quickly obtain the overall mechanical response data of the UAV under the current flight state. This achieves a leap from finite point measurement to full-field perception, and significantly reduces the dependence on complex sensor networks and system hardware costs while ensuring monitoring accuracy.

[0044] (2) This invention creates a digital twin platform that integrates real-time driving, high-performance computing, and immersive interaction. The platform acquires real-time status through the data acquisition and communication module, completes performance prediction through the proxy model calculation module, and finally presents it intuitively in the three-dimensional visualization and human-computer interaction module in the form of multi-mode dynamic cloud map, extreme value marking, and flexible view control. It completes the full-link connection from physical signals to decision information and provides a core technical platform for real-time evaluation and interactive analysis of UAV status.

[0045] (3) This invention upgrades the monitoring and maintenance mode from post-flight analysis to process monitoring. Traditional methods often rely on downloading and analyzing data after flight, which has serious lag. Through the above system, this invention enables the structural performance changes of UAVs during flight to be captured, calculated and visualized in real time, providing operators with immediate status feedback, making real-time status-based decision-making and intervention possible, and effectively improving the timeliness and proactivity of flight safety monitoring.

[0046] In summary, this invention significantly improves the completeness, real-time performance, and intuitiveness of UAV performance monitoring, providing an innovative solution for achieving efficient flight safety monitoring and status assessment. Attached Figure Description

[0047] Figure 1 This is a flowchart of the method of the present invention.

[0048] Figure 2 This is a schematic diagram of the wing mesh model of a drone according to an embodiment of the present invention.

[0049] Figure 3 This is a schematic diagram of a local mesh model of a drone wing according to an embodiment of the present invention.

[0050] Figure 4 This is a schematic diagram of a three-dimensional model of a drone wing within a three-dimensional scene according to an embodiment of the present invention. Detailed Implementation

[0051] To provide a more detailed explanation of the present invention, specific embodiments will be described in detail below. These descriptions are only for explaining the principles and implementation of the present invention and do not constitute any limitation on the scope of protection of the present invention.

[0052] This embodiment provides a digital twin-based unmanned aerial vehicle (UAV) performance monitoring platform and method. The invention integrates real-time sensor data with a high-performance proxy model to achieve real-time bidirectional mapping and driving between the physical entity and the virtual model. By collecting and analyzing key flight parameters of the UAV, and based on the data-driven proxy model, it achieves efficient and rapid prediction of structural performance. The digital twin platform enables real-time synchronous mapping of the UAV's multi-dimensional mechanical performance, dynamic marking of extreme value regions, and immersive visual interaction. Specifically:

[0053] A method for monitoring the performance of unmanned aerial vehicles (UAVs) based on digital twins, the method comprising the following steps:

[0054] Step 1: Construct a high-fidelity 3D geometric model and proxy model. Specifically:

[0055] Step 1.1: Based on the physical structural characteristics of the UAV, a three-dimensional geometric model of the UAV is established using three-dimensional modeling software. By adjusting parameters and optimizing the design, the three-dimensional geometric model can accurately reflect the real structure of the UAV.

[0056] Step 1.2: The 3D geometric model established in Step 1.1 is simplified. Finite element preprocessing software is used to perform regional mesh generation on the simplified 3D geometric model. In this embodiment, the UAV wing is used as an example, resulting in the following UAV wing mesh model: Figure 2 As shown, due to the fine mesh, the local mesh model of the UAV wing is as follows: Figure 3 As shown, the number and spatial coordinates of each grid node are exported to a local data file.

[0057] Step 1.3: Determine the input variables and load parameter domain of the surrogate model by combining the inertial loads and aerodynamic loads experienced by the UAV in actual flight.

[0058] The surrogate model is an efficient approximation model built using machine learning algorithms, used to quickly predict the full-field mechanical response of an unmanned aerial vehicle (UAV) structure without performing a complete finite element method (FEM) solution. The input variable of the surrogate model is the equivalent load vector acting on the UAV, which consists of load components along the longitudinal, transverse, and vertical axes of the aircraft body, used to characterize the force characteristics under different flight conditions.

[0059] The value range of each component of the load vector is determined based on the typical flight attitude and aerodynamic range of the UAV, forming a load parameter domain. A characteristic load sample set is then generated within the load parameter domain using a spatial sampling method to cover typical UAV flight conditions.

[0060] Step 1.4: Using the finite element analysis method, the multidimensional load parameters obtained in Step 1.3 are applied to the UAV wing mesh model divided in Step 1.2, and the full-field mechanical response data of the UAV wing under multiple working conditions are output.

[0061] Step 1.5: Based on the feature load sample set in Step 1.3 and the full-field mechanical response data calculated in Step 1.4, construct a training sample database and use the training sample database to train the surrogate model.

[0062] Step 1.6: Deploy an inertial measurement unit on the physical entity of the UAV wing to collect three-axis acceleration and pitch angle data.

[0063] Step 2: Build the 3D scene and visualization environment for the digital twin platform. Specifically:

[0064] Step 2.1: Build a digital twin platform and construct a 3D scene within the digital twin platform.

[0065] Step 2.2: Import the data file of the UAV wing mesh model exported in Step 1.2 into the digital twin platform. Through parametric modeling, construct a 3D model of the UAV wing in the 3D scene, as shown below. Figure 4 As shown.

[0066] Step 2.3: Create a user interface in the 3D scene, integrating mode switching buttons, real-time data charts, network settings panel, data management panel, and function buttons to complete the interface construction of the 3D visualization and human-computer interaction module.

[0067] Step 2.4: Configure camera control, data-driven rendering, and interface event response logic for the 3D visualization and human-computer interaction module. The camera control logic is bound to user input, the data-driven rendering logic is bound to the output data of the proxy model trained in Step 1.5, and the interface event response logic is bound to mode switching and extreme value marker switches, thus forming a data-driven real-time visualization closed loop.

[0068] Step 3: Collect flight data and perform real-time performance prediction. Specifically:

[0069] Step 3.1: During the flight of the UAV, the inertial measurement unit deployed in step 1.6 collects the three-axis acceleration component data in the body coordinate system in real time. , , With pitch angle data .in, This contains the raw acceleration data along the longitudinal axis of the fuselage (head direction). This contains the raw acceleration data along the transverse axis of the aircraft (wing direction, usually right side is positive). The raw acceleration data is along the vertical axis of the fuselage (vertical to the downward direction of the fuselage). This is the angle between the longitudinal axis of the aircraft and the horizontal plane, i.e., the original data of the pitch angle.

[0070] Step 3.2: The digital twin platform built in Step 2.1 receives the raw data collected in Step 3.1 and extracts the effective triaxial acceleration component data within the platform. , , , and pitch angle data , thus obtaining valid data.

[0071] Step 3.3: Filter and denoise the valid data extracted in Step 3.2 to obtain the processed data. , , , .in, The processed acceleration data along the longitudinal axis of the machine body. The processed acceleration data along the transverse axis of the aircraft body. The processed acceleration data along the vertical axis of the aircraft. The data represents the processed pitch angle. Subsequently, based on flight dynamics principles, the processed data is converted into an equivalent payload vector acting on the UAV, yielding the real-time payload vector. ,in The conversion formula is as follows:

[0072] (1)

[0073] (2)

[0074] (3)

[0075] in, This represents the equivalent aerodynamic load component along the X-axis of the aircraft. The equivalent aerodynamic load component along the Y-axis of the fuselage. This represents the equivalent aerodynamic load component along the Z-axis of the fuselage. For predefined load conversion factors, Let gravitational acceleration be (take) ).

[0076] Step 3.4: Convert the real-time load vector obtained in Step 3.3 into a vector. Input the data into the agent model trained in step 1.5 to quickly calculate and output the full-field mechanical response data of the UAV under the current flight state.

[0077] Step 4: Perform 3D dynamic rendering and interactive feedback. Specifically:

[0078] Step 4.1: Based on the 3D visualization and human-computer interaction module built in Step 2.3, the user can select the current display mode through the mode switching button. The digital twin platform will process the data group corresponding to the current mode in the UAV full-field mechanical response data output in Step 3.4, and render a dynamically changing performance cloud map in real time on the UAV 3D model constructed in Step 2.2 through the color mapping algorithm.

[0079] Step 4.2: Based on the performance cloud map rendered in real time in the current display mode in Step 4.1, the user can use the extreme value marker switch to enable the digital twin platform to traverse the full-field mechanical response data output in Step 3.4, automatically identify the node with the largest value in the data group corresponding to the current mode, and instantiate or update a colored marker sphere at the spatial location of the node in the three-dimensional scene for dynamic highlighting.

[0080] Step 4.3: Based on the camera control logic configured in the 3D visualization and human-computer interaction module built in Step 2.3, users can rotate and zoom the viewpoint within the digital twin platform using a mouse.

[0081] A digital twin-based drone performance monitoring platform is provided. This platform implements the aforementioned digital twin-based drone performance monitoring method. The platform is centered on a digital twin platform and integrates five functional modules: a physical entity and simulation module, a data acquisition and communication module, a data processing and conversion module, a proxy model calculation module, and a 3D visualization and human-computer interaction module. Each functional module interacts and collaborates through the digital twin platform. Specifically:

[0082] The physical entity and simulation module consists of a UAV physical entity, an inertial measurement unit deployed on it, and a simulation calculation part for finite element analysis. This module utilizes the inertial measurement unit to collect key parameters such as the UAV's three-axis acceleration and pitch angle in real time during flight, and performs finite element analysis to obtain full-field mechanical response data under multiple operating conditions.

[0083] The data acquisition and communication module establishes a connection with the physical entity of the UAV through a stable network communication protocol. It is used to receive raw data from the sensors in real time and transmit it to the digital twin platform in a low-latency and high-reliability manner. It also has the function of local data storage and management.

[0084] The data processing and conversion module is integrated into the digital twin platform. The data processing and conversion module filters and denoises the effective sensor data transmitted by the data acquisition and communication module, and based on the principles of flight dynamics, converts the triaxial acceleration and pitch angle data into real-time load vectors acting on the UAV structure, providing an input interface for the proxy model calculation module.

[0085] The surrogate model calculation module is the core calculation module in the digital twin platform. This module is trained based on a sample database generated by the physical entity and simulation modules to construct a surrogate model. The surrogate model receives real-time load vectors and quickly outputs the UAV's full-field mechanical response data under the current flight state.

[0086] The 3D visualization and human-computer interaction module uses a color mapping algorithm to render the full-field mechanical response data output by the proxy model calculation module into a dynamically changing performance cloud map in real time on the digital twin platform, and provides interactive functions including viewpoint control, extreme value marking, and mode switching.

[0087] The above embodiments are merely illustrative of the implementation methods of the present invention, but should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the protection scope of the present invention.

Claims

1. A method for monitoring the performance of unmanned aerial vehicles (UAVs) based on digital twins, characterized in that, The method for monitoring the performance of unmanned aerial vehicles includes the following steps: Step 1: Construct a high-fidelity 3D geometric model and proxy model; A 3D geometric model of the UAV is established and processed to obtain a UAV mesh model. The input variables and load parameter domain of the surrogate model are determined, and a set of characteristic load samples is generated within the load parameter domain. These samples are then applied to the UAV mesh model to output full-field mechanical response data. A training sample database is constructed to train the surrogate model. Specifically: Step 1.1: Based on the physical structural characteristics of the UAV, a three-dimensional geometric model of the UAV is established using three-dimensional modeling software. By adjusting parameters and optimizing the design, the three-dimensional geometric model can accurately reflect the real structure of the UAV. Step 1.2: Simplify the 3D geometric model established in Step 1.1, divide the simplified 3D geometric model into regional meshes to obtain the UAV mesh model, and export the number and spatial coordinates of each mesh node to the local data file; Step 1.3: Determine the input variables and load parameter domain of the surrogate model by combining the inertial loads and aerodynamic loads experienced by the UAV in actual flight; The input variable of the proxy model is the equivalent load vector acting on the UAV, which is used to characterize the force characteristics under different flight conditions. The value range of each component of the load vector is determined according to the typical flight attitude and aerodynamic range of the UAV, forming a load parameter domain. A feature load sample set is generated in the load parameter domain using a spatial sampling method to cover the typical flight conditions of the UAV. Step 1.4: Using the finite element analysis method, the multidimensional load parameters obtained in Step 1.3 are applied to the UAV mesh model divided in Step 1.2, and the full-field mechanical response data of the UAV under multiple working conditions are output. Step 1.5: Based on the feature load sample set in Step 1.3 and the full-field mechanical response data calculated in Step 1.4, construct a training sample database and use the training sample database to train the surrogate model; Step 1.6: Deploy an inertial measurement unit on the physical entity of the UAV to collect three-axis acceleration and pitch angle data; Step 2: Build the 3D scene and visualization environment of the digital twin platform; A 3D model of a drone is built in a 3D scene constructed by a digital twin platform, and an interface for 3D visualization and human-computer interaction modules is built, ultimately forming a data-driven real-time visualization closed loop. Step 3: Collect flight data and perform real-time performance prediction; the equivalent load vector consists of load components along the longitudinal, transverse, and vertical axes of the aircraft; specifically: Step 3.1: During the flight of the UAV, the inertial measurement unit deployed in step 1.6 collects the three-axis acceleration component data in the body coordinate system in real time. , , With pitch angle data The original data is obtained; among which, This contains the raw acceleration data along the longitudinal axis of the aircraft. This is the raw acceleration data along the transverse axis of the aircraft. This contains the raw acceleration data along the vertical axis of the aircraft. This is the angle between the longitudinal axis of the aircraft and the horizontal plane, i.e., the original data of the pitch angle; Step 3.2: The digital twin platform built in Step 2.1 receives the raw data collected in Step 3.1 and extracts the effective triaxial acceleration component data within the digital twin platform. , , , and pitch angle data To obtain valid data; Step 3.3: Filter and denoise the valid data extracted in Step 3.2 to obtain the processed data. , , , ;in, The processed acceleration data along the longitudinal axis of the machine body. The processed acceleration data along the transverse axis of the aircraft body. The processed acceleration data along the vertical axis of the aircraft. The processed pitch angle data is then converted into an equivalent payload vector acting on the UAV, yielding the real-time payload vector. ,in The conversion formula is as follows: （1）； （2）； （3）； in, This represents the equivalent aerodynamic load component along the X-axis of the aircraft. The equivalent aerodynamic load component along the Y-axis of the fuselage. This represents the equivalent aerodynamic load component along the Z-axis of the aircraft. For predefined load conversion factors, It is the acceleration due to gravity; Step 3.4: Convert the real-time load vector obtained in Step 3.3 into a vector. Input the data into the agent model trained in step 1.5 to calculate and output the full-field mechanical response data of the UAV under the current flight state; Step 4: Perform 3D dynamic rendering and interactive feedback.

2. The method for monitoring the performance of a UAV based on digital twins according to claim 1, characterized in that, In step 1.3, the surrogate model is an approximate model established by a machine learning algorithm, used to predict the full-field mechanical response of the UAV structure without performing a complete finite element solution.

3. The method for monitoring the performance of a UAV based on digital twins according to claim 2, characterized in that, Step 2 specifically refers to: Step 2.1: Build a digital twin platform and construct a 3D scene within the digital twin platform; Step 2.2: Import the data file of the UAV mesh model exported in Step 1.2 into the digital twin platform, and construct the UAV 3D model in the 3D scene through parametric modeling; Step 2.3: Create a user interface in the 3D scene, integrating mode switching buttons, real-time data charts, network settings panel, data management panel and function buttons, to complete the interface construction of the 3D visualization and human-computer interaction module; Step 2.4: Configure camera control, data-driven rendering, and interface event response logic for the 3D visualization and human-computer interaction module; The camera control logic is bound to user input, the data-driven rendering logic is bound to the output data of the proxy model trained in step 1.5, and the interface event response logic is bound to mode switching and extreme value marker switch, thus forming a data-driven real-time visualization closed loop.

4. The method for monitoring the performance of a UAV based on digital twins according to claim 3, characterized in that, In step 3.1, the longitudinal axis of the fuselage represents the direction of the nose, the transverse axis of the fuselage represents the direction of the wings, and the vertical axis of the fuselage represents the direction perpendicular to the fuselage and downward.

5. The method for monitoring the performance of a UAV based on digital twins according to claim 4, characterized in that, Step 4 specifically includes: Step 4.1: Based on the 3D visualization and human-computer interaction module built in Step 2.3, select the current display mode through the mode switching button, so that the digital twin platform processes the data group corresponding to the current mode in the UAV full-field mechanical response data output in Step 3.4, and renders the dynamically changing performance cloud map in real time on the UAV 3D model built in Step 2.2 through the color mapping algorithm. Step 4.2: Based on the performance cloud map rendered in real time in the current display mode in Step 4.1, the digital twin platform traverses the full-field mechanical response data output in Step 3.4 through the extreme value marker switch, automatically identifies the node with the largest value in the data group corresponding to the current mode, and instantiates or updates a colored marker sphere at the spatial location of the node in the three-dimensional scene for dynamic highlighting. Step 4.3: Based on the camera control logic configured in the 3D visualization and human-computer interaction module built in Step 2.3, the rotation and scaling of the viewpoint within the digital twin platform are realized through the mouse.

6. A performance monitoring platform for unmanned aerial vehicles (UAVs) based on digital twins, characterized in that, The UAV performance monitoring platform described in claims 1-5 implements the UAV performance monitoring method based on digital twins. The UAV performance monitoring platform is centered on a digital twin platform and integrates five functional modules, including: a physical entity and simulation module, a data acquisition and communication module, a data processing and conversion module, a proxy model calculation module, and a 3D visualization and human-computer interaction module. Each functional module interacts and collaborates through the digital twin platform.

7. A UAV performance monitoring platform based on digital twins according to claim 6, characterized in that, The drone performance monitoring platform is specifically: The physical entity and simulation module consists of a UAV physical entity, an inertial measurement unit deployed on it, and a simulation calculation part for finite element analysis; the inertial measurement unit collects the parameters of the UAV in real time during flight and obtains full-field mechanical response data under multiple working conditions. The data acquisition and communication module establishes a connection with the physical entity of the UAV through a stable network communication protocol, which is used to receive raw data from the sensors in real time and transmit it to the digital twin platform. It also has the function of local data storage and management. The data processing and conversion module is integrated into the digital twin platform. The data processing and conversion module filters and denoises the effective sensor data transmitted by the data acquisition and communication module, and converts the triaxial acceleration and pitch angle data into real-time load vectors acting on the UAV structure, providing an input interface for the proxy model calculation module. The proxy model calculation module is the core calculation module in the digital twin platform. It is trained and constructed based on the sample database generated by the physical entity and simulation module. The proxy model receives real-time load vectors and outputs the full-field mechanical response data of the UAV under the current flight state. The 3D visualization and human-computer interaction module uses a color mapping algorithm to render the full-field mechanical response data output by the proxy model calculation module into a dynamically changing performance cloud map in real time on the digital twin platform, and provides interactive functions including viewpoint control, extreme value marking, and mode switching.

Images