A target scattering feature library dynamic construction and updating method based on digital twinning

Abstract

The application provides a target scattering feature library dynamic construction and updating method based on digital twinning, relates to the technical field of target scattering feature libraries, and comprises the following steps: constructing an integrated virtual scene; calibrating scattering data in the virtual scene; extracting and sorting scattering features and storing them in an initial library; continuously capturing target and environmental change data, synchronously updating a model and scattering data, comparing features, and dynamically updating the feature library. The application realizes the construction and updating of a dynamic target scattering feature library with precision, timeliness, integrity and high efficiency by means of innovative means such as building a digital twinning model consistent with a physical target, configuring virtual signal devices in modules, generating virtual scattering data in multiple dimensions, dynamically optimizing acquisition deployment, layering calibration data, standardizing and sorting stored features, dynamically adjusting the acquisition cycle, accurately comparing and updating features, and reasonably determining scene correlation, thereby providing reliable data support for the identification, analysis and application of target scattering features.

Description

Technical Field

[0001] This invention relates to the field of target scattering feature library technology, and more specifically, to a method for dynamically constructing and updating a target scattering feature library based on digital twins. Background Technology

[0002] With the technological advancements in radar detection, remote sensing monitoring, and target recognition, the requirements for the accuracy, timeliness, and completeness of target scattering feature data are becoming increasingly stringent. Digital twin technology, with its ability to accurately map physical entities, is widely used in industry and defense, providing a technological foundation for the virtual-real collaborative acquisition of scattering feature data. However, targets undergo structural changes and material aging throughout their lifecycle, and the environment experiences dynamic fluctuations such as occlusion and interference, necessitating a solution for constructing a scattering feature database that adapts to these changes.

[0003] Traditional methods for constructing target scattering feature libraries often employ static modeling, resulting in models that struggle to accurately reproduce the size, structure, and material properties of physical targets and are unable to adapt to structural changes in detachable components. Furthermore, the observation angles and polarization states of virtual scattering data are incomplete, and the actual data collection lacks dynamic adjustments to address obstructions, easily leading to data gaps. Data calibration lacks scientific hierarchical strategies, resulting in poor consistency between virtual and actual data. Feature processing lacks unified standards, storage methods are simplistic, and retrieval efficiency is low. Fixed collection cycles make it difficult to capture dynamic changes in targets and the environment, leading to delayed feature library updates and a large amount of redundant data.

[0004] Therefore, it is necessary to design a dynamic construction and update method for a target scattering feature library based on digital twins to solve the problem that the static construction of the scattering feature library in the existing technology cannot dynamically adapt to changes in target structure and environment. Summary of the Invention

[0005] In view of this, the present invention proposes a method for dynamically constructing and updating a target scattering feature library based on digital twins, aiming to solve the problem that static construction of scattering feature libraries in the prior art cannot dynamically adapt to changes in target structure and environment.

[0006] In one aspect, this invention proposes a method for dynamically constructing and updating a target scattering feature library based on digital twins, comprising: Build a digital twin model that matches the physical target in size, structure, and material; configure virtual signal transmitting and receiving devices in the digital twin model; and construct a virtual scene that integrates the physical target and signal equipment. The system performs signal transmission operations in a virtual scene, calculates scattering signal data under different observation angles and polarization states, and simultaneously acquires scattering signal data and target state data from the actual environment through acquisition components set on the physical target. The actual data is then transmitted to the digital twin model to calibrate the scattering data in the virtual scene. Scattering features are extracted from the calibrated virtual scattering data and the actual scattering data. The scattering features include signal waveform, scattering center position and signal intensity. The extracted scattering features are organized in a unified format and stored in the initial scattering feature library. The system continuously captures data on changes in the physical target's state and the environment through acquisition components. It then synchronizes this data to the digital twin model, recalculates the scattering signal data in the virtual scene based on the data, extracts new scattering features, compares these new features with existing data in the scattering feature library, replaces any discrepancies, and supplements the scattering feature data corresponding to the new scene.

[0007] Furthermore, when building a digital twin model, the size parameters, structural distribution information, and material attribute information of the physical target are first obtained, and the geometric structure and material distribution of the model are restored according to the actual shape of the physical target; when configuring virtual signal transmitting and receiving devices, the deployment positions of the devices are planned according to the signal transmission path; when constructing an integrated virtual scene, the operating parameters of the digital twin model and the virtual signal device are associated. When the physical target has detachable parts, the digital twin model is constructed by dividing it into modules, and the virtual signal devices are configured separately according to the module structure; when the physical target is an integral and indivisible structure, an integrated digital twin model is constructed, and the virtual signal devices are configured uniformly according to the overall structure.

[0008] Furthermore, when constructing the modular structure of the digital twin model by component breakdown, the connection relationship and assembly parameters of each module are recorded. When the virtual signal device is configured separately according to the modular structure, an independent signal transmitting sub-device and signal receiving sub-device are configured for each module, and signal interaction channels between modules are set up at the same time. When the connection relationship between modules changes, if the change is a relocation of connection location, the deployment coordinates of the signal transmitting sub-devices and signal receiving sub-devices are replanned according to the new connection location, and the signal coverage of the sub-devices is aligned with the new connection area of ​​the module. At the same time, the transmission bandwidth and delay parameters of the signal interaction channel are adjusted according to the new connection distance. If the change is an increase or decrease in the number of connections, the number of signal transmitting sub-devices and signal receiving sub-devices is increased or decreased according to the increased or decreased number of module connections. The transmission resources of the signal interaction channel are reallocated, and the number of parallel transmission paths of the channel is adjusted. When the connection relationship between the modules remains unchanged, the initial configuration of the virtual signal device and the parameters of the signal interaction channel are maintained. When configuring the operating parameters of the virtual signal device, set the frequency range and power range of the signal transmission.

[0009] Furthermore, when performing signal transmission operations in a virtual scene, the observation angle of the signal transmission device is adjusted at preset angle intervals, and the polarization state of the signal is switched according to a preset polarization mode, generating scattered signal data under different combinations of observation angles and polarization states one by one. When calculating the scattered signal data, the signal change data after the signal interacts with the model surface is derived based on the material properties and geometric structure of the digital twin model. When acquiring actual data through the acquisition component, the deployment position of the acquisition component corresponds to the deployment position of the virtual signal receiving device, and the acquisition frequency of the acquisition component matches the transmission frequency of the virtual signal. When there are obstructions in the actual environment, the number of data acquisition components deployed is increased to expand the acquisition range to cover the unobstructed target area; When there are no obstructions in the actual environment, the actual data is obtained according to the initial deployment plan.

[0010] Furthermore, when transmitting actual data to the digital twin model, the data is synchronized in real time, and any missing data during the transmission process is supplemented according to the correspondence between the actual data and the virtual data at adjacent moments. When calibrating scattering data in a virtual scene, the difference between the actual scattering data and the virtual scattering data is calculated, and the parameters of the virtual scene are adjusted based on the difference. When the difference is within the preset range, the first ratio of the difference to the upper limit of the preset range is calculated, and the attenuation coefficient of the virtual signal is linearly adjusted according to the first ratio. When the difference exceeds the preset range, if the difference is greater than the upper limit of the preset range, the real part of the material dielectric constant of the digital twin model is increased by the second ratio of the difference to the virtual scattering data, and the transmission power parameter of the virtual signal is increased proportionally to the absolute value of the difference; if the difference is less than the lower limit of the preset range, the real part of the material dielectric constant of the digital twin model is decreased by the second ratio, and the transmission power parameter of the virtual signal is decreased proportionally to the absolute value of the difference.

[0011] Furthermore, when extracting scattering features, the calibrated virtual scattering data and the actual scattering data are first denoised. Then, the peak value, valley value, period parameters of the signal waveform, the three-dimensional coordinate parameters of the scattering center, and the amplitude parameters of the scattering intensity are extracted from the denoised data. When organizing the scattering features in a unified format, different types of feature parameters are arranged in a preset order and each feature parameter is assigned a unique identifier. When there are duplicate extracted feature parameters, retain the feature parameters with the same value and remove duplicate records. When the extracted feature parameters are not duplicated, they are directly organized according to a uniform format; When storing the sorted scattering features into the initial scattering feature library, establish the correspondence between the feature parameters and the observation angle, polarization state, target state, and environmental conditions.

[0012] Furthermore, when establishing the correspondence between feature parameters and related conditions, an index is constructed according to the hierarchical structure of observation angle, polarization state, target state, and environmental conditions; when it is necessary to query scattering features under specific conditions, the corresponding feature parameters are located through the hierarchical index; the initial scattering feature library adopts a partitioned storage method, dividing the scattering features corresponding to different target states into different storage partitions; When the number of target state types is less than the type number threshold, a single storage partition is used to store all scattering features. When the number of target state types is greater than or equal to the threshold for the number of types, the storage partitions are split according to the target state type.

[0013] Furthermore, when continuously capturing data through the acquisition component, the captured physical target state change data includes material wear data and structural deformation data, and the environmental change data includes temperature and humidity data and interference signal data. The acquisition component captures data according to a preset cycle. When the captured data shows that the change in the physical target state is greater than the preset value, the acquisition cycle is shortened; when the captured data shows that the change in the physical target state is less than or equal to the preset value, the original acquisition cycle is maintained. When synchronizing changed data to the digital twin model, the timestamp and location identifier of the captured data are attached, and the digital twin model updates its own parameters according to the time sequence and location information; When recalculating the scattering signal data in the virtual scene based on the changed data, the model parameters and signal parameters corresponding to the changed data are called, and the scattering data calculation process is re-executed.

[0014] Furthermore, after extracting new scattering features, a feature comparison algorithm is used to calculate the similarity coefficient between the new scattering features and the existing data in the scattering feature database. The similarity coefficient is calculated as the ratio of the sum of the absolute values ​​of the differences between the corresponding parameters of the new features and the existing features to the sum of the absolute values ​​of the parameters of the existing features. When the similarity coefficient is greater than or equal to the preset threshold, it is determined that there is a difference, and the existing data is replaced with the new scattering feature. When the similarity coefficient is less than the preset threshold, it is determined that there is no difference and the existing data is retained; When replacing data, record the replacement time, the reason for the change, and related parameters; When supplementing scattering feature data for new scenes, first determine the degree of correlation between the new scene and the existing scene, and then classify the new feature data into the corresponding category or create a new category according to the degree of correlation.

[0015] Furthermore, when determining the degree of correlation between the new scene and the existing scene, the matching degree between the environmental conditions, observation angle, polarization state of the new scene and the corresponding parameters of the existing scene is calculated. The matching degree is calculated as the ratio of the number of identical parameters to the total number of parameters. When the matching degree is greater than the preset ratio, the new feature data will be classified into the corresponding existing scene category. When the matching degree is less than or equal to the preset ratio, a new scene category is created and new feature data is stored. The updated scattering feature library is validated to check the completeness and format correctness of the feature parameters. If a parameter is missing or the format is incorrect, the corresponding scattering feature is re-extracted and corrected. If the parameter is complete and the format is correct, the update process is completed.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: The method for dynamically constructing and updating a target scattering feature library based on digital twins, by building a digital twin model that is completely consistent with the size, structure, and material of the physical target and configuring modular virtual signal devices, ensures a high degree of fit between the virtual and physical scenes while also possessing flexible adaptability to changes in the structure of detachable target components. By generating virtual scattering data according to preset angle intervals and polarization methods, and dynamically adjusting the deployment of acquisition components based on the degree of obstruction, the coverage integrity and acquisition effectiveness of the scattering data are effectively improved. Through a layered calibration strategy based on a preset difference interval, accurate correction of the virtual scattering data is achieved, ensuring the consistency between virtual and actual data. By organizing scattering features in a preset order and using a partitioned storage method based on a threshold for the number of target state types, the standardization and retrieval efficiency of feature data are improved. By dynamically adjusting the acquisition cycle based on preset values ​​of state changes, the system can capture changes in the target and environment in a timely manner, ensuring the timeliness of data acquisition and resource utilization. Through a feature comparison and update mechanism based on a preset threshold of similarity coefficients, the system achieves accurate iteration of feature library data and avoids redundant data retention. By using scene association judgment based on preset ratios, the system ensures the rationality and completeness of feature library classification. The resulting dynamic scattering feature library has the advantages of accuracy, timeliness, completeness, and efficiency, and can provide reliable data support for the identification, analysis, and application of target scattering features. Attached Figure Description

[0017] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1A flowchart illustrating the method for dynamically constructing and updating a target scattering feature library based on digital twins, as provided in this embodiment of the invention. Detailed Implementation

[0018] Exemplary embodiments of the present invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the invention to those skilled in the art. It should be noted that, without conflict, the embodiments and features described herein can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] Reference Figure 1 As shown in some embodiments of this application, a method for dynamically constructing and updating a target scattering feature library based on digital twins includes: Build a digital twin model that matches the physical target in size, structure, and material; configure virtual signal transmitting and receiving devices in the digital twin model; and construct a virtual scene that integrates the physical target and signal equipment. The system performs signal transmission operations in a virtual scene, calculates scattering signal data under different observation angles and polarization states, and simultaneously acquires scattering signal data and target state data from the actual environment through acquisition components set on the physical target. The actual data is then transmitted to the digital twin model to calibrate the scattering data in the virtual scene. Scattering features are extracted from the calibrated virtual scattering data and the actual scattering data. The scattering features include signal waveform, scattering center position and signal intensity. The extracted scattering features are organized in a unified format and stored in the initial scattering feature library. The system continuously captures data on changes in the physical target's state and the environment through acquisition components. It then synchronizes this data to the digital twin model, recalculates the scattering signal data in the virtual scene based on the data, extracts new scattering features, compares these new features with existing data in the scattering feature library, replaces any discrepancies, and supplements the scattering feature data corresponding to the new scene.

[0020] Specifically, by comprehensively measuring and collecting information from the physical target, its three-dimensional dimensions, the structural composition and connection methods of each component, and the material types and distribution of its surface and interior are obtained. Based on this collected information, a three-dimensional modeling method is used to reconstruct the geometric contours, component assembly relationships, and material distribution of the physical target, ensuring that the constructed digital twin model completely corresponds to the physical target in terms of dimensions, structural form, and material properties. Subsequently, based on the requirements for capturing the signal scattering characteristics of the target and combined with the morphological structure of the digital twin model, the deployment locations of the virtual signal transmitting and receiving devices are planned, ensuring that the signal coverage of the transmitting devices can cover the entire surface of the digital twin model and different observation points. The orientation is determined by setting the signal type and frequency range of the signal transmitting device and the signal acquisition parameters of the receiving device to ensure that the device parameters are compatible with the signal interaction scenario in actual applications. Finally, the operation association between the digital twin model and the virtual signal transmitting and receiving devices is established, signal transmission path rules are set, and the complete interaction process of device startup, signal transmission, signal action on the model surface, and receiving device acquisition of scattered signals is clarified. This ensures that changes in the structural state of the digital twin model can be synchronously associated with the adjustment of the operating parameters of the signal device, and that the signal transmission and reception process of the signal device can accurately correspond to the physical form of the model. This results in the construction of an integrated virtual scene in which the physical target and signal equipment operate collaboratively and interact effectively.

[0021] Specifically, in the virtual scene, the observation position of the virtual signal transmitting device is gradually switched according to preset angle adjustment rules to achieve full-angle coverage from the front, side, back, and various tilted positions of the target. Simultaneously, the polarization state of the signal is switched sequentially according to preset polarization modes, including horizontal polarization, vertical polarization, and other cross-polarization forms. The signal transmitting device is then activated to transmit a preset type of signal. This signal acts on the surface and internal structure of the digital twin model according to a set transmission path. Based on the model's material dielectric constant, permeability, and other property parameters, as well as the spatial distribution of its geometric structure, the physical interaction processes of the signal with the model surface—including incident, reflected, refracted, and internally scattered—are deduced to obtain the signal at different observation angles. The system collects multi-dimensional data on the amplitude, phase, waveform, and frequency response of the scattered signal under various polarization states. Simultaneously, acquisition components are deployed in key scattering areas, different structural parts, and critical locations in the surrounding environment of the physical target. These components are configured with the same signal acquisition frequency and parameters as the virtual signal transmitter to capture real-time data on the scattered signal generated after the signal acts on the physical target in the actual environment. This includes detailed information such as the peak value, valley value, period, and phase shift of the scattered signal. Simultaneously, real-time status data of the physical target is acquired, covering parameters such as surface wear, structural deformation, and component connection tightness, as well as environmental temperature, humidity, air pressure, interference signal strength, and the distribution of obstructions. The system collects environmental data such as morphology; then, it packages the actual scattering signal data and target state data in real time via a data transmission link, adding a timestamp of data acquisition, the deployment location identifier of the acquisition component, and an environmental condition label during data acquisition. An anti-interference transmission protocol is used for data transmission, and data integrity is verified during transmission. Any transmission delays or packet loss are retransmitted to ensure the actual data is transmitted completely and accurately to the digital twin model system. After receiving the data, the system first precisely aligns the timestamps, observation angles, and polarization states of the actual data with the scattering data calculated in the virtual scene, ensuring that each set of virtual scattering data can find a corresponding actual data sample. The deviation values ​​of the two in various dimensions such as amplitude, phase, and waveform characteristics are calculated, including the absolute deviation, relative deviation, and overall data distribution deviation of individual parameters. Based on the magnitude and distribution of the deviation values, the material property parameters, geometric dimensions, signal attenuation coefficient of the virtual signal transmitting device, and transmission power compensation parameters of the digital twin model are adjusted. If the deviation is due to differences in material properties, the dielectric constant of the corresponding part of the model is corrected. If the deviation is due to differences in signal transmission path, the propagation loss parameters of the virtual signal are adjusted. Through multiple rounds of iterative adjustments, the scattering data calculated in the virtual scene is made consistent with the scattering data actually collected in all dimensions, thus completing the accurate calibration of the scattering data of the virtual scene.

[0022] Specifically, the calibrated virtual scattering data and actual scattering data are preprocessed separately. Low-pass filtering and adaptive filtering remove environmental interference noise, equipment operating noise, and transmission clutter. A data smoothing algorithm is then used to correct abrupt changes in the data, ensuring its continuity and stability. For signal waveform feature extraction, the preprocessed data is analyzed frame by frame to identify and record specific parameters such as rise time, fall time, peak amplitude, valley amplitude, peak interval, period duration, duty cycle, phase offset, and waveform distortion. Waveform similarity analysis is then used to classify... For each waveform type, the coordinates of the characteristic inflection points and the curve fitting parameters are labeled. For the extraction of scattering center locations, hotspot areas of concentrated signal energy are identified based on the signal amplitude distribution heatmap. A three-dimensional coordinate system is established with the geometric center of the physical target as the origin. The three-dimensional coordinates (precise values ​​of the x, y, and z axes) of each hotspot area are calculated by combining the triangulation method with the signal arrival time difference under different observation angles. False energy hotspots caused by noise are eliminated, and the number, spatial distribution, and signal contribution weight of each true scattering center are determined. For signal intensity feature extraction, the mean signal amplitude corresponding to each scattering center is calculated. The system first extracts peak intensity, RMS value, signal attenuation coefficient, and gradients of intensity parameters under different observation angles and polarization states. It then correlates the scattering center location with these intensity parameters to form a position-intensity correspondence. Subsequently, all extracted scattering features are organized according to a pre-defined, unified format. This format includes a feature data header identifier (target number, data source (virtual / real), calibration identifier, and acquisition timestamp), feature classification fields (waveform parameter class, position parameter class, intensity parameter class), specific parameter items (arranged in a pre-defined order, each parameter item including parameter name, value, unit, and accuracy level), and association condition fields (observation angle, polarization state, target state, and environmental conditions). After organization, the feature data undergoes format verification, checking field completeness, numerical range rationality, and unit consistency. Verified data is compressed and encoded. A hierarchical storage directory for the initial scattering feature library is established according to target type, observation angle range, and polarization state type. The encoded feature data is written to the corresponding storage files in the specified directories. Simultaneously, a feature data index table is created, containing information such as feature identifier, storage path, key association conditions, and data length, enabling rapid retrieval and retrieval of feature data. This completes the initial scattering feature library storage operation.

[0023] Specifically, data acquisition components continuously deployed in key material areas, structural connection points, easily deformable locations, and sensitive points in the surrounding environment of the physical target capture real-time data on the physical target's state changes (including material surface wear depth, wear area, changes in dielectric constant / permeability due to material aging, deformation displacement of structural components, deformation angle, degree of connection looseness, and addition or removal of components) and environmental changes (including real-time values ​​and rates of change of ambient temperature and humidity, air pressure fluctuations, frequency, amplitude, duration of interference signals, type and size of obstructions). (Coverage range, movement trajectory, etc.) During the data collection process, each set of data is appended with a precise timestamp, a unique identifier for the collection component, the target location coordinates corresponding to the data, and environmental condition labels. The captured data undergoes preprocessing, including noise reduction, deduplication, and outlier removal. Then, the preprocessed change data is transmitted to the digital twin model system via an encrypted transmission link using a real-time streaming protocol. During transmission, data integrity is verified using a checksum, and a retransmission mechanism is initiated for lost or delayed data to ensure accurate data synchronization. After receiving the change data, the digital twin model aligns the data with the current state of the model according to the timestamp order and updates the corresponding parameters based on the data type: material change data is used to correct the corresponding areas of the model. The material properties and structural change data of the domain are used to adjust the geometric dimensions, component connection relationships, and assembly parameters of the model. Environmental change data is used to update the environmental parameters of the virtual scene (such as the signal propagation loss coefficient corresponding to temperature and humidity, interference signal parameters, and virtual models of obstructions). Based on the updated digital twin model parameters and virtual scene environmental parameters, the observation angle coverage, polarization state type, and combination method from the initial calculation are retained. The signal transmitting device is restarted to transmit signals. The signals act on the adjusted model structure and material surface according to the updated transmission path. By reproducing the physical processes of signal incident, reflection, refraction, and internal scattering, combined with the updated material properties and structural parameters, the signal transmission is further refined. Based on the spatial distribution and environmental impact parameters, the amplitude, phase, waveform, frequency response, and other full-dimensional data of the scattered signal under various observation angles and polarization state combinations were recalculated. Subsequently, the recalculated virtual scattering data was processed according to the same preprocessing standards (low-pass filtering, adaptive filtering, and data smoothing) as the initial feature extraction. The same feature extraction rules were used: the data was analyzed frame by frame to obtain parameters such as the rising / falling edge time, peak / valley amplitude, period, duty cycle, and phase shift of the signal waveform. The three-dimensional coordinates of the scattering center were calculated by heatmap analysis and triangulation method, and false hot spots were removed. The mean, peak, effective value, and gradient of the signal intensity were statistically analyzed to extract new scattering features.After organizing the newly extracted scattering features in a unified format, the index table of the feature library is used to locate existing feature data corresponding to the same observation angle, polarization state, target basic state, and environmental conditions. A comparison is performed using a "parameter-by-parameter difference calculation + overall similarity weighting" method: first, the absolute and relative deviations of each corresponding parameter between the new feature and the existing feature are calculated; then, the overall similarity is calculated using the importance weight of each parameter as a coefficient (similarity = 1 - Σ(parameter deviation × weight) / Σ weight). When the overall similarity is lower than a preset threshold, it is determined that there is a difference, and the data is replaced according to the rule of "new features covering existing features, retaining historical version records of existing features (including update time, reasons for change, and original parameters)". When there is no matching existing data in the feature library for the observation angle, polarization state, target state, or environmental condition combination corresponding to the new feature, it is determined to be a new scene. A classification directory is created for the new scene according to the partitioned storage rules of the feature library, a unique identifier is assigned to the new feature data, an associated index is established, the organized new feature data is stored in the corresponding directory, and the index table and classification statistics of the feature library are updated simultaneously, completing the supplementation of feature data for the new scene.

[0024] Understandably, by building a digital twin model that perfectly matches the size, structure, and material of the physical target and configuring modular virtual signal devices, a high degree of consistency between the virtual and physical scenes is ensured, while also providing flexible adaptability to changes in the structure of the target's detachable components. By generating virtual scattering data at preset angle intervals and polarization patterns, and dynamically adjusting the deployment of acquisition components based on the degree of obstruction, the coverage integrity and acquisition effectiveness of the scattering data are effectively improved. Through a layered calibration strategy based on preset difference intervals, precise correction of the virtual scattering data is achieved, ensuring consistency between virtual and actual data. Furthermore, by organizing scattering characteristics in a preset order and based on the target's state type... The partitioned storage method based on quantity thresholds improves the standardization and retrieval efficiency of feature data; the dynamic adjustment of the acquisition cycle based on preset values ​​of state change enables timely capture of changes in the target and environment, ensuring the timeliness of data acquisition and resource utilization; the feature comparison and update mechanism based on preset thresholds of similarity coefficients achieves accurate iteration of feature library data, avoiding redundant data retention; and the scene association judgment based on preset ratios ensures the rationality and completeness of feature library classification. The final dynamic scattering feature library has the advantages of accuracy, timeliness, completeness, and efficiency, and can provide reliable data support for the identification, analysis, and application of target scattering features.

[0025] In some embodiments of this application, when building a digital twin model, the size parameters, structural distribution information, and material attribute information of the physical target are first obtained, and the geometric structure and material distribution of the model are restored according to the actual shape of the physical target; when configuring virtual signal transmitting and receiving devices, the deployment positions of the devices are planned according to the signal transmission path; when constructing an integrated virtual scene, the operating parameters of the digital twin model and the virtual signal device are associated. When the physical target has detachable parts, the digital twin model is constructed by dividing it into modules, and the virtual signal devices are configured separately according to the module structure; when the physical target is an integral and indivisible structure, an integrated digital twin model is constructed, and the virtual signal devices are configured uniformly according to the overall structure.

[0026] Specifically, the following information was obtained regarding the physical target's dimensions (5m length, 3m width, 2m height), structural distribution (main frame + 3 detachable functional modules; module 1: 1.5m × 1m × 0.8m; module 2: 1m × 1m × 0.8m; module 3: 0.8m × 1m × 0.8m; each module is connected to the main frame via bolts), and material properties (main frame is aluminum alloy with a conductivity of 2.5 × 10⁷ S / m; functional module shells are made of engineering plastic with a relative permittivity of 3.2-j0.03; internal core components are made of copper with a density of 8960 kg / m³). 3The electrical conductivity is set according to the ideal conductor. The model's geometry and material distribution are restored according to the actual shape of the physical target (the main frame is a cuboid, module 1 is located on the front left side of the main frame, module 2 on the front right side, and module 3 in the rear middle). The main frame area is given aluminum alloy properties, the module shell area is given engineering plastic properties, and the core component area is given copper properties. When configuring virtual signal transmitting and receiving devices, the deployment positions of the devices are planned according to the signal transmission path (straight transmission from the transmitting device to the model surface, reflection from the model surface to the receiving device). Three transmitting devices are deployed 5m directly in front of the model, 4m to the left front, and 4m to the right front; six receiving devices are evenly distributed on a 3m circle around the model, with adjacent devices spaced 60° apart. When constructing an integrated virtual scene, the operating parameters of the digital twin model and the virtual signal devices are linked (when the structural deformation of a certain area of ​​the model exceeds 0.5cm, the corresponding area...). The signal frequency of the transmitting device in the domain is adjusted by ±5%; when the physical target has detachable parts (module 1, module 2, module 3, meeting the criteria of bolt connection, independent disassembly and independent function), the digital twin model is constructed in a modular structure according to the parts (main frame module + module 1 sub-module + module 2 sub-module + module 3 sub-module, recording the bolt connection position and tightening torque parameters of each sub-module), and the virtual signal device is configured according to the modular structure (the main frame module is configured with 1 transmitting device and 2 receiving devices, and each functional sub-module is configured with 1 transmitting device and 1 receiving device); when the physical target is an integral and indivisible structure (such as an integral cast main frame with no detachable parts), an integrated digital twin model is constructed (the geometric structure is a complete cuboid, without split modules), and the virtual signal device is uniformly configured according to the overall structure (3 transmitting devices and 6 receiving devices are deployed in the above positions, covering the entire model surface).

[0027] Understandably, by accurately acquiring the size, structure, and material information of the physical target and reconstructing the model, a high degree of fit between the virtual scene and the physical scene is ensured, providing model support for the accuracy of subsequent scattering data calculations. At the same time, by distinguishing whether the physical target is a detachable structure, modular or integrated model construction and device configuration methods are adopted respectively, which initially realizes the adaptation to targets with different structural types and improves the versatility of the method.

[0028] In some embodiments of this application, when constructing the modular structure of the digital twin model by component breakdown, the connection relationship and assembly parameters of each module are recorded. When the virtual signal device is configured separately according to the modular structure, an independent signal transmitting sub-device and signal receiving sub-device are configured for each module, and a signal interaction channel between modules is set up at the same time. When the connection relationship between modules changes, if the change is a relocation of connection location, the deployment coordinates of the signal transmitting sub-devices and signal receiving sub-devices are replanned according to the new connection location, and the signal coverage of the sub-devices is aligned with the new connection area of ​​the module. At the same time, the transmission bandwidth and delay parameters of the signal interaction channel are adjusted according to the new connection distance. If the change is an increase or decrease in the number of connections, the number of signal transmitting sub-devices and signal receiving sub-devices is increased or decreased according to the increased or decreased number of module connections. The transmission resources of the signal interaction channel are reallocated, and the number of parallel transmission paths of the channel is adjusted. When the connection relationship between the modules remains unchanged, the initial configuration of the virtual signal device and the parameters of the signal interaction channel are maintained. When configuring the operating parameters of the virtual signal device, set the frequency range and power range of the signal transmission.

[0029] Specifically, when constructing the modular structure of the digital twin model by component breakdown, the connection relationships of each module (module 1 is bolted to the main frame, module 2 is snap-fitted to the main frame, and module 3 is welded to the main frame) and assembly parameters (bolt connection torque 20 N·m, snap-fit ​​connection engagement depth 5 mm, welded connection weld width 3 mm) are recorded. When the virtual signal device is configured according to the modular structure, one independent signal transmitting sub-device and one signal receiving sub-device are configured for each of module 1, module 2, module 3, and the main frame. At the same time, signal interaction channels between modules are set up (initial transmission bandwidth 10 MHz, delay 5 ms, and 3 parallel transmission channels). When each module... When the connection relationship changes, if the change is a relocation of the connection position (e.g., module 1 is moved 20cm to the right from its original installation position on the main frame), then the deployment coordinates of the signal transmitting sub-device and the signal receiving sub-device will be replanned according to the new connection position (the center coordinates of module 1 are adjusted from (10cm, 20cm, 30cm) to (30cm, 20cm, 30cm)). The transmitting sub-device will be moved 20cm to the right with module 1 and deployed 3m directly in front of the center of module 1, and the receiving sub-device will be deployed 3m directly behind the center of module 1. The signal coverage area of ​​the sub-device (a circular area with a diameter of 5m) will be compared with the new connection area of ​​the module (the new mating surface between module 1 and the main frame, with an area of ​​0.2m²). 2Correspondingly, the transmission bandwidth of the signal interaction channel is adjusted (from 10MHz to 12MHz) and the delay parameter (from 5ms to 4ms) according to the new connection distance (the original connection distance between module 1 and the main frame was 0cm, and there is no change after adjustment; the connection distance between adjacent module 1 and module 2 increases from 10cm to 30cm). If the change type is an increase or decrease in the number of connections (such as adding module 4 to connect to the main frame, increasing the number of modules from 3 to 4), then the number of corresponding signal transmitting sub-devices and signal receiving sub-devices is increased or decreased according to the increased or decreased number of module connections (4 modules) (adding 1 transmitting sub-device and 1 receiving sub-device). The signal interaction channel transmission resources are reallocated (the original 3 parallel transmission channels are increased to 4, with an average bandwidth of 8MHz per channel), and the number of parallel transmission channels is adjusted (4 channels). When the connection relationship between each module remains unchanged, the initial configuration of the virtual signal device (1 transmitting sub-device and 1 receiving sub-device per module, with the deployment location unchanged) and the parameters of the signal interaction channel (bandwidth 10MHz, delay 5ms, number of parallel channels 3) are maintained. When configuring the operating parameters of the virtual signal device, the frequency range (1GHz-10GHz) and power range (0.1W-1W) of the signal transmission are set.

[0030] Understandably, by recording the module connection relationships and assembly parameters, a basis is provided for the precise deployment of virtual signal sub-devices and the reasonable setting of signal interaction channels, making the sub-module configuration more targeted. At the same time, differentiated device adjustment and channel optimization strategies are formulated for different types of changes in module connection relationships (position movement, quantity increase or decrease), which significantly improves the flexible adaptability to changes in the structure of the target detachable components, ensuring that the virtual scene can still accurately reflect the physical target state after structural changes.

[0031] In some embodiments of this application, when performing signal transmission operations in a virtual scene, the observation angle of the signal transmission device is adjusted at preset angle intervals, and the polarization state of the signal is switched according to a preset polarization mode, thereby generating scattered signal data under different combinations of observation angles and polarization states one by one. When calculating the scattered signal data, the signal change data after the signal interacts with the model surface is derived based on the material properties and geometric structure of the digital twin model. When acquiring actual data through the acquisition component, the deployment position of the acquisition component corresponds to the deployment position of the virtual signal receiving device, and the acquisition frequency of the acquisition component matches the transmission frequency of the virtual signal. When there are obstructions in the actual environment, the number of data acquisition components deployed is increased to expand the acquisition range to cover the unobstructed target area; When there are no obstructions in the actual environment, the actual data is obtained according to the initial deployment plan.

[0032] Specifically, the preset angle interval refers to the fixed angle adjustment step size used when adjusting the observation angle of the signal transmitting device in the virtual scene. It is an angle increment set to ensure that the transmitted signal can cover all observable directions of the physical target without omission. Its value directly determines the density of the observation angles. The smaller the angle interval, the denser the distribution of observation angles, and the higher the coverage of the target direction by the scattered signal data. At the same time, the amount of calculation of the scattered data will also increase accordingly. The larger the angle interval, the sparser the distribution of observation angles, and the higher the calculation efficiency of the scattered data. At the same time, it is necessary to avoid the omission of scattering features of some key directions due to excessively large intervals. The preset polarization mode refers to the specific type of electric field vibration direction used by the virtual signal transmitting device when transmitting signals. Common polarization modes include horizontal polarization, vertical polarization, 45° cross polarization, etc. The signal under different polarization modes will produce different interaction effects with the material and structure of the target, and the collected scattering signal features will have significant differences. Setting multiple polarization modes is to comprehensively capture the scattering characteristics of the target under different polarization conditions, thereby meeting the storage requirements of the scattering feature library for multi-dimensional scattering data. The process of obtaining the interval involves first fully modeling the three-dimensional geometry of the physical target to accurately determine its maximum outline size and the location distribution of key scattering regions. Then, in a virtual scene, the coverage of the key scattering regions by the signal transmitter under different angular intervals is simulated. The minimum angle increment required to cover all key scattering regions is calculated. Combined with the computational efficiency requirements of the scattered signal data in the virtual scene, the minimum angle increment is repeatedly verified and adjusted. Redundant angle increments that do not affect coverage integrity are eliminated, ultimately determining the specific value of the angle interval that balances coverage integrity and computational efficiency. The process of obtaining the preset polarization mode involves first investigating the actual application scenarios of the target's scattering characteristics to clarify the specific requirements for signal polarization types in the application scenarios. Then, by reviewing relevant literature on the response characteristics of target materials and structures to different polarized signals, polarization types that can effectively distinguish the target's scattering characteristics are selected. Subsequently, multi-polarization signal scattering simulation tests are conducted in a virtual scene, comparing the discriminability and effectiveness of the scattered signal data under different polarization modes. Polarization types with low discriminability and poor effectiveness are eliminated, ultimately determining the preset polarization mode set that includes all key polarization types.

[0033] Specifically, when performing signal transmission operations in the virtual scene, the observation angle of the signal transmitting device is adjusted at preset 15° intervals to achieve full-angle coverage from 0° to 360°. The polarization state of the signal is switched according to preset polarization modes of horizontal polarization, vertical polarization, and 45° linear polarization, generating scattered signal data under different combinations of observation angles and polarization states. When calculating the scattered signal data, the signal change data after the interaction between the signal and the model surface is derived based on the material properties of the digital twin model (aluminum alloy for the main frame, relative permittivity 9.8; engineering plastic for the module shell, relative permittivity 3.2) and geometric structure (5m×3m×2m for the main frame, 1.5m×1m×0.8m for module 1). When acquiring actual data through the acquisition component, the deployment position of the acquisition component and the virtual signal receiving device are aligned around the model. The deployment positions within a 3m radius correspond one-to-one, and the 2GHz-18GHz acquisition frequency of the acquisition components matches the 2GHz-18GHz transmission frequency of the virtual signal. When there are obstructions in the actual environment, the coverage area and degree of obstruction of the obstruction are first determined. If the proportion of the obstructed area to the total target area exceeds a preset threshold of 30%, the number of acquisition components deployed is increased proportionally. The deployment positions of the newly added acquisition components are planned to be the unobstructed areas surrounding the obstructed area, and the signal coverage ranges of adjacent acquisition components overlap by 10%. If the proportion of the obstructed area to the total target area does not exceed the preset threshold of 30%, acquisition components are deployed at the edge of the obstructed area to ensure that the unobstructed target area is completely within the acquisition range. When there are no obstructions in the actual environment, the actual data is obtained according to the initial deployment of 3 acquisition components.

[0034] Understandably, by presetting angle intervals and polarization methods, full coverage of observation angles and multi-dimensional coverage of polarization states are achieved, effectively improving the azimuth integrity and dimensional richness of virtual scattering data. At the same time, by keeping the deployment position and acquisition frequency of the acquisition components consistent with the virtual signal device, the comparability of actual and virtual data is ensured. Furthermore, the deployment of acquisition components is dynamically adjusted according to the situation of obstructions, avoiding data loss caused by obstructions and significantly improving the effectiveness of actual scattering data acquisition.

[0035] In some embodiments of this application, when transmitting actual data to the digital twin model, the data is synchronized in real time, and any missing data during the transmission process is supplemented according to the correspondence between the actual data and the virtual data at adjacent times. When calibrating scattering data in a virtual scene, the difference between the actual scattering data and the virtual scattering data is calculated, and the parameters of the virtual scene are adjusted based on the difference. When the difference is within the preset range, the first ratio of the difference to the upper limit of the preset range is calculated, and the attenuation coefficient of the virtual signal is linearly adjusted according to the first ratio. When the difference exceeds the preset range, if the difference is greater than the upper limit of the preset range, the real part of the material dielectric constant of the digital twin model is increased by the second ratio of the difference to the virtual scattering data, and the transmission power parameter of the virtual signal is increased proportionally to the absolute value of the difference; if the difference is less than the lower limit of the preset range, the real part of the material dielectric constant of the digital twin model is decreased by the second ratio, and the transmission power parameter of the virtual signal is decreased proportionally to the absolute value of the difference.

[0036] Specifically, the preset interval is a pre-defined allowable fluctuation range of the difference between the actual and virtual scattering data during the virtual scene scattering data calibration process. This interval consists of a clearly defined lower and upper limit value and is the core basis for determining the degree of deviation in the virtual scene and selecting the corresponding calibration strategy. When the difference between the actual and virtual scattering data falls within this interval, it is determined that the virtual scene has only a slight deviation, and calibration can be completed by adjusting the attenuation coefficient of the virtual signal. When the difference exceeds the upper limit or falls below the lower limit of the interval, it is determined that the virtual scene has a significant deviation, and the material property parameters of the digital twin model and the emission parameters of the virtual signal must be adjusted simultaneously to achieve accurate calibration. The numerical range of the preset interval directly determines the rigor and frequency of calibration. The smaller the interval, the higher the calibration accuracy, but the calibration operation frequency will increase accordingly. The larger the interval, the higher the calibration efficiency, but it is necessary to avoid a decrease in the consistency of scattering data due to an excessively wide interval. The preset interval is obtained by first using a data acquisition component at an ambient temperature of 25°C. Under standard operating conditions of ℃, 50% humidity, and no obstruction, 100 repeated data acquisitions were performed on the physical target to obtain 100 sets of actual scattering signal amplitude data, with the data range concentrated between 9.5V and 10.5V. Simultaneously, 100 sets of matching virtual scattering signal amplitude data were calculated in the corresponding virtual scene, with the data range concentrated between 9.6V and 10.4V. Then, the difference between the actual scattering data and the virtual scattering data was calculated for each set, and the difference range was found to be between -0.4V and +0.4V. The distribution pattern of all differences was statistically analyzed, and the concentrated distribution range of the difference was determined to be between -0.3V and +0.3V. Then, combined with the accuracy requirements of the scattering feature library for data consistency, the allowable deviation range was expanded to -0.5V to +0.5V. Subsequently, 50 rounds of calibration tests were performed in the virtual scene to verify the rationality of the adjustment strategy switching within this range. It was confirmed that within this range, only adjusting the attenuation coefficient is sufficient to meet the accuracy requirements. When the range is exceeded, adjusting the material and signal parameters can effectively correct the deviation. Finally, the specific numerical range of the preset range was determined.

[0037] Specifically, when transmitting actual data to the digital twin model, data is synchronized in real-time. For any missing data during transmission, it is supplemented by matching the actual scattering data (e.g., 10.2V at the 10th second, 9.7V at the 12th second) with the virtual scattering data (e.g., 10.0V at the 10th second, 9.8V at the 12th second) at adjacent moments. When calibrating the scattering data in the virtual scene, the differences between the actual scattering data (e.g., amplitudes of 10.3V, 9.2V) and the virtual scattering data (e.g., amplitudes of 10.0V, 10.0V) are calculated, which are +0.3V and -0.8V respectively. A preset range is set as [-0.5V, +0.5V], and the parameters of the virtual scene are adjusted based on these differences. When the difference of +0.3V falls within the preset range of [-0.5V, +0.5V], the first difference between this difference and the upper limit of the preset range, +0.5V, is calculated. The ratio, 0.3 ÷ 0.5 = 0.6, is used to linearly adjust the attenuation coefficient of the virtual signal (initial value 0.8) based on this first ratio of 0.6. After adjustment, the attenuation coefficient is 0.8 + (0.8 × 0.6) = 1.28. When the difference is -0.8V, exceeding the preset range of [-0.5V, +0.5V], since this difference is less than the lower limit of the preset range (-0.5V), the absolute value of this difference is compared with the virtual scattering data amplitude of 10.0. The second ratio of V, i.e. 0.8 ÷ 10.0 = 0.08, reduces the dielectric constant of the digital twin model's materials (aluminum alloy from 9.8 to 9.8 - 9.8 × 0.08 = 9.016, engineering plastic from 3.2 to 3.2 - 3.2 × 0.08 = 2.944), and at the same time reduces the virtual signal's transmission power parameter by the absolute value of the difference, 0.8V (from 0.5W to 0.5 - 0.5 × 0.08 = 0.46W).

[0038] Understandably, the adoption of real-time transmission and integrity verification and retransmission mechanisms ensures the integrity and accuracy of actual data transmission. Precise alignment of dimensions such as timestamps and observation angles provides a reliable premise for data calibration. The hierarchical calibration strategy based on the preset difference interval (adjusting the attenuation coefficient within the interval and synchronously adjusting the model and signal parameters outside the interval) achieves accurate correction of virtual scattering data, effectively reducing the deviation between virtual and actual data and ensuring the consistency of the two types of data.

[0039] In some embodiments of this application, when extracting scattering features, the calibrated virtual scattering data and the actual scattering data are first denoised, and then the peak value, valley value, period parameters of the signal waveform, the three-dimensional coordinate parameters of the scattering center, and the amplitude parameters of the scattering intensity are extracted from the denoised data. When organizing the scattering features in a unified format, different types of feature parameters are arranged in a preset order and each feature parameter is assigned a unique identifier. When there are duplicate extracted feature parameters, retain the feature parameters with the same value and remove duplicate records. When the extracted feature parameters are not duplicated, they are directly organized according to a uniform format; When storing the sorted scattering features into the initial scattering feature library, establish the correspondence between the feature parameters and the observation angle, polarization state, target state, and environmental conditions.

[0040] Specifically, the preset order is a fixed arrangement order followed when organizing extracted scattering feature parameters in a unified format. This order is set based on the importance of the scattering feature parameters in characterizing the target's scattering characteristics and the need for convenient retrieval and comparison of feature library data. The core principle is to place key parameters that directly reflect the essential attributes of the target's scattering at the beginning, followed by auxiliary parameters. Specifically, signal waveform parameters (peak, valley, period) are arranged first, then scattering center location parameters (three-dimensional coordinates), and finally signal intensity parameters (amplitude). The fixed order ensures that scattering feature data from different batches and sources have a consistent structure after processing, avoiding problems such as feature comparison errors and low data retrieval efficiency caused by disordered parameter arrangement. At the same time, it lays a unified data structure foundation for the subsequent hierarchical storage and rapid indexing of the scattering feature library. The operation of obtaining the preset order first involves a comprehensive review of the data from the calibrated virtual... The types of scattering feature parameters that can be extracted from scattering data and actual scattering data were identified. The specific sub-items included in the three main categories of parameters—signal waveform parameters, scattering center location parameters, and signal intensity parameters—were clearly defined. Then, by investigating the actual application scenarios of the target scattering feature library, the frequency and priority of different types of parameters in feature comparison, target recognition, and other applications were statistically analyzed. A priority ranking was determined, with signal waveform parameters having the highest priority for representing target scattering characteristics, followed by scattering center location parameters, and signal intensity parameters as a supplement. Subsequently, 100 sets of scattering feature data from different observation angles and polarization states were selected, and organized, stored, and retrieved according to a pre-determined order. The retrieval time and comparison accuracy for each type of parameter were recorded. Adjustments and optimizations were made to the positions of parameters with long retrieval times and prone to errors in comparison. Redundant parameter arrangements were eliminated, and similar parameter sub-items were merged. After multiple rounds of testing and verification, a preset order that balances representation priority and ease of use was finally determined.

[0041] Specifically, when extracting scattering features, the calibrated virtual scattering data (amplitude range 5.0V-5.5V) and the actual scattering data (amplitude range 4.9V-5.4V) are first denoised using a 5th-order low-pass filter. Then, the peak (5.2V), valley (4.8V), and envelope period (20ms) parameters of the signal waveform, as well as the three-dimensional coordinate parameters of the scattering center (x: 15cm, y: 20cm, z: 10cm), are extracted from the denoised data. When organizing the scattering features in a unified format, different types of feature parameters are arranged in a preset order of "signal waveform parameters → scattering center position parameters → signal strength parameters," and each feature parameter is assigned a unique identifier (identification rule: target number T001 - acquisition time 2). 02501011200 – Parameter type code W01 / B01 / Q01, where W01 is the waveform peak value, W02 is the waveform valley value, W03 is the envelope period, B01 is the scattering center coordinates, and Q01 is the scattering intensity amplitude; when the extracted feature parameters are duplicated (e.g., the difference between the peak values ​​of two extractions is ≤0.1V), the feature parameters with the same value are retained, and duplicate records are removed; when the extracted feature parameters are not duplicated, they are directly sorted according to the preset order; when the sorted scattering features are stored in the initial scattering feature library, the correspondence between the feature parameters and the observation angle (30°), polarization state (vertical polarization), target state (material without wear, structure without deformation), and environmental conditions (temperature 25℃, humidity 50%) is established.

[0042] Understandably, by removing data noise and outliers through preprocessing, the accuracy of scattering feature extraction is improved. The specific extraction parameters of core features such as signal waveform, scattering center location, and signal intensity are clarified, ensuring the comprehensiveness and standardization of feature extraction. At the same time, features are organized in a preset order and assigned unique identifiers, establishing a correspondence between features and key conditions, making the feature data structure unified and the association clear, which facilitates the storage and retrieval of the subsequent feature library.

[0043] In some embodiments of this application, when establishing the correspondence between feature parameters and related conditions, an index is constructed according to the hierarchical structure of observation angle, polarization state, target state, and environmental conditions; when it is necessary to query scattering features under specific conditions, the corresponding feature parameters are located through the hierarchical index; the storage structure of the initial scattering feature library adopts a partitioned storage method, dividing the scattering features corresponding to different target states into different storage partitions; When the number of target state types is less than the type number threshold, a single storage partition is used to store all scattering features. When the number of target state types is greater than or equal to the threshold for the number of types, the storage partitions are split according to the target state type.

[0044] Specifically, the type number threshold is a critical value used to determine whether to split the storage partitions according to the target state type when the initial scattering feature library adopts a partitioned storage method. This threshold is the core judgment indicator for balancing the storage efficiency and retrieval speed of the feature library. When the number of target state types is less than this threshold, a single storage partition can meet the data management needs and avoid index redundancy caused by too many partitions. When the number of target state types is greater than or equal to this threshold, the storage partitions need to be split according to the target state type to prevent the retrieval time from increasing due to the large amount of data in a single partition. The value of this threshold needs to be determined comprehensively in combination with the hardware storage capacity of the feature library and the response time requirements of data retrieval. Different application scenarios correspond to different threshold values. The operation of obtaining the type number threshold first involves comprehensively sorting out the physical targets that may appear throughout their entire life cycle. All state types, including no wear, slight wear, moderate wear, heavy wear, structural deformation, and component damage, are counted to obtain the total number of target state types. Then, a feature library test environment is built to simulate the performance of single-partition storage and multi-partition storage under different threshold values. Taking a total of 9 target state types as an example, threshold candidate values ​​of 3, 5, and 8 are set respectively. The data write speed and retrieval response time of single partition or multi-partition storage corresponding to different candidate values ​​are tested in turn. The performance indicators under each partition strategy are recorded. Then, combined with the requirement that the feature library retrieval response time does not exceed 1 second in practical applications, the partition strategy that meets the performance indicators is selected. Finally, the stability of the partition strategy is verified through multiple rounds of repeated testing. Candidate values ​​corresponding to performance fluctuations caused by accidental factors are eliminated, and the specific value of the type number threshold is finally determined.

[0045] Specifically, when establishing the correspondence between feature parameters and related conditions, an index is constructed according to a hierarchical structure of observation angle (0°, 15°, 30°, 45°, 60°), polarization state (horizontal polarization, vertical polarization, 45° cross-line polarization), target state (no wear, slight wear, moderate wear), and environmental conditions (temperature 20°C, humidity 40%RH; temperature 25°C, humidity 50%RH; temperature 30°C, humidity 60%RH). A threshold of 5 types is set. When querying scattering features under specific conditions, the corresponding feature parameters are located through the hierarchical index. (For example, to query feature parameters under the conditions of 30° observation angle, vertical polarization, no wear target state, and 25°C / 50%RH environmental conditions, the index first searches the 30° observation angle level, then filters the vertical polarization sub-level, then locates the no-wear sub-level for the target state, and finally matches the environmental condition as 25°C / 50%RH.) (The corresponding feature parameters can be obtained from the sub-level of H). The storage structure of the initial scattering feature library adopts a partitioned storage method, dividing the scattering features corresponding to different target states into different storage partitions. When the number of target state types is 3 (no wear, slight wear, moderate wear) less than the type number threshold of 5, a single storage partition is used to store all scattering features, and all feature parameters are integrated into the partition numbered FZ001. When the number of target state types is 6 (no wear, slight wear, moderate wear, heavy wear, structural deformation, component loss) greater than or equal to the type number threshold of 5, the storage partitions are split according to the target state type, and partitions numbered FZ001-no wear, FZ002-slight wear, FZ003-moderate wear, FZ004-heavy wear, FZ005-structural deformation, and FZ006-component loss are established respectively, and the scattering features of the corresponding target states are stored in their respective partitions.

[0046] Understandably, building an index based on a hierarchical structure of key conditions such as observation angle and polarization state significantly improves the retrieval speed of scattering features under specific conditions. The partitioned storage strategy based on the threshold of the number of target state types, by distinguishing the number of types and using single partitions or split partitions for storage, avoids the inefficiency of retrieval caused by the large amount of data in a single partition, and also prevents index redundancy caused by too many partitions. This achieves a balance between storage efficiency and retrieval efficiency, and further improves the standardization and efficiency of feature data management.

[0047] In some embodiments of this application, when data is continuously captured by the acquisition component, the captured physical target state change data includes material wear data and structural deformation data, and the environmental change data includes temperature and humidity data and interference signal data; the acquisition component captures data according to a preset cycle, and when the captured data shows that the change in the physical target state is greater than a preset value, the acquisition cycle is shortened; when the captured data shows that the change in the physical target state is less than or equal to the preset value, the original acquisition cycle is maintained; When synchronizing changed data to the digital twin model, the timestamp and location identifier of the captured data are attached, and the digital twin model updates its own parameters according to the time sequence and location information; When recalculating the scattering signal data in the virtual scene based on the changed data, the model parameters and signal parameters corresponding to the changed data are called, and the scattering data calculation process is re-executed.

[0048] Specifically, the preset value is a critical value used to determine whether the acquisition cycle of the acquisition component needs to be adjusted during the continuous acquisition of physical target state change data. This value is determined based on the performance stability range of the physical target and the sensitivity of scattering characteristic changes. It is a core indicator for balancing data acquisition timeliness and hardware resource consumption. When the amount of physical target state change is greater than the preset value, it indicates that the target state has changed significantly, and the scattering characteristics are likely to fluctuate significantly. The acquisition cycle needs to be shortened to obtain denser state and scattering data. When the amount of state change is less than or equal to the preset value, it indicates that the target state is in a relatively stable range, the scattering characteristics fluctuate less, and maintaining the original acquisition cycle can meet the needs of feature library updates. The magnitude of the preset value directly affects the frequency of adjustment of the acquisition strategy. The smaller the value, the more sensitive the acquisition strategy, but the greater the hardware resource consumption. The larger the value, the less hardware resource consumption, but some key state change data may be missed. The acquisition of the preset value is as follows: First, by consulting the material performance manual and structural design drawings of the physical target, the performance stability parameters such as the wear critical depth of the target material and the deformation critical displacement of the structure are determined. Then, through accelerated aging tests and structural mechanics tests, the scattering characteristic data of the target under different wear depths and different deformation displacements are collected. The critical value of the state change amount corresponding to the start of significant changes in scattering characteristics is statistically analyzed. Next, an acquisition strategy simulation test platform is built, and multiple different candidate preset values ​​are set to simulate the entire process of the target from stability to change. The amount of data collected, the hardware resource utilization rate, and the scattering characteristic update accuracy under different candidate values ​​are recorded. Candidate values ​​with a balance between the amount of data collected and the hardware resource utilization rate and the characteristic update accuracy meet the standards are selected. Finally, the applicability of the candidate values ​​is verified through multiple rounds of field tests, and abnormal critical values ​​caused by environmental interference are eliminated. Finally, the specific value of the preset value is determined.

[0049] Specifically, when continuously capturing data through the acquisition component, the captured physical target state change data includes material wear data (wear depth 0-5mm), structural deformation data (deformation displacement 0-3mm), and environmental change data including temperature and humidity data (temperature 10-40℃, humidity 30%-70%RH), and interference signal data (interference frequency 1-10GHz, interference amplitude 0-2V). The acquisition component captures data at a preset cycle of 10 minutes, with a preset value of 0.5mm. When the captured data shows a physical target state change of 0.8mm, which is greater than the preset value of 0.5mm, the acquisition cycle is shortened to 5 minutes. When the captured data shows a physical target state change of 0.3mm, which is less than or equal to the preset value of 0.5mm, the original 10-minute cycle is maintained. The data acquisition cycle involves synchronizing the changed data to the digital twin model, including the timestamp of the data capture (e.g., 20250101120000) and location identifiers (e.g., acquisition points C001 and C002). The digital twin model updates its parameters according to the time sequence and location information (e.g., the relative permittivity of the aluminum alloy in the corresponding area of ​​the model decreases from 9.8 to 9.5 when the wear depth is 0.8mm, and the geometric coordinates of the corresponding area of ​​the model are adjusted by 0.8mm along the x-axis when the deformation displacement is 0.8mm). When recalculating the scattering signal data in the virtual scene based on the changed data, the model parameters (relative permittivity 9.5, geometric coordinate adjustment 0.8mm) and signal parameters (transmission frequency 5GHz, transmission power 0.5W) corresponding to the changed data are called, and the scattering data calculation process is re-executed.

[0050] Understandably, by comprehensively capturing data on changes in the physical target's state (material wear, structural deformation) and environmental changes (temperature, humidity, interference signals, etc.), a comprehensive perception of changes in the target and environment is ensured. The acquisition cycle is dynamically adjusted based on the preset value of the state change. When the state changes significantly, the cycle is shortened to ensure data timeliness, while the cycle is maintained when the state is stable to save hardware resources. This achieves an optimized balance between data acquisition timeliness and resource utilization. At the same time, the changing data is synchronously updated to the digital twin model and the scattering data is recalculated, ensuring real-time synchronization between virtual data and changes in the target and environment.

[0051] In some embodiments of this application, after extracting new scattering features, a feature comparison algorithm is used to calculate the similarity coefficient between the new scattering features and existing data in the scattering feature library. The similarity coefficient is calculated as the ratio of the sum of the absolute values ​​of the differences between the corresponding parameters of the new features and the existing features to the sum of the absolute values ​​of the parameters of the existing features. When the similarity coefficient is greater than or equal to the preset threshold, it is determined that there is a difference, and the existing data is replaced with the new scattering feature. When the similarity coefficient is less than the preset threshold, it is determined that there is no difference and the existing data is retained; When replacing data, record the replacement time, the reason for the change, and related parameters; When supplementing scattering feature data for new scenes, first determine the degree of correlation between the new scene and the existing scene, and then classify the new feature data into the corresponding category or create a new category according to the degree of correlation.

[0052] Specifically, the preset threshold is a critical value used to determine whether there is a significant difference between newly extracted scattering features and existing features in the scattering feature library when comparing their similarity. This value is the core indicator for measuring the consistency between new and old scattering features. Its magnitude is set comprehensively based on the accuracy requirements for data consistency in the scattering feature library and the efficiency requirements for feature updates. When the similarity coefficient is less than the preset threshold, the new feature is considered to have high consistency with the existing feature, and there is no need to replace the existing data. When the similarity coefficient is greater than or equal to the preset threshold, the new feature is considered to have a significant difference with the existing feature, and the existing data needs to be replaced. The value of the preset threshold directly determines the update frequency and data purity of the feature library data. The smaller the value, the stricter the requirement for feature consistency, and the higher the data update frequency, but it may introduce redundant update operations. The larger the value, the more relaxed the requirement for feature consistency, and the higher the data update efficiency, but it may miss some effective feature changes. The preset threshold is obtained by first selecting 200 groups of different observation angles, polarization states, ... The scattering feature sample data under target conditions were used to split each group of sample data into baseline features and test features. The similarity coefficient of each group of samples was calculated using the method of "the ratio of the sum of the absolute values ​​of the differences between corresponding parameters to the sum of the absolute values ​​of the existing feature parameters". The distribution pattern of all similarity coefficients was statistically analyzed, and it was determined that the similarity coefficients were concentrated in the range of 0.75-0.9. Combined with the actual needs of the scattering feature library to support target recognition, the accuracy standard of feature comparison was determined to be a target recognition accuracy of not less than 95%. Then, four candidate preset thresholds of 0.7, 0.75, 0.8 and 0.85 were set, and feature comparison and data update tests were carried out respectively. The performance indicators such as feature update frequency and target recognition accuracy under different thresholds were recorded. When the candidate threshold was 0.8, the target recognition accuracy reached 96% and the update frequency was within a reasonable range. Finally, through 50 rounds of field collection and feature comparison verification tests, abnormal indicators caused by equipment noise and environmental interference were eliminated, and the specific value of the preset threshold was finally determined to be 0.8.

[0053] Specifically, after extracting new scattering features, the new feature parameters are determined as follows: signal waveform peak value 5.2V, valley value 1.8V, period 2ms, scattering center three-dimensional coordinates x15cm, y20cm, z10cm, and signal intensity amplitude 3.5V. The existing feature parameters are: signal waveform peak value 5.0V, valley value 2.0V, period 2.1ms, scattering center three-dimensional coordinates x14.5cm, y20.2cm, z9.8cm, and signal intensity amplitude 3.4V. Feature comparison is then performed. The algorithm calculates the similarity coefficient by first summing the absolute values ​​of the differences between corresponding parameters: |5.2-5.0|+|1.8-2.0|+|2-2.1|+|15-14.5|+|20-20.2|+|10-9.8|+|3.5-3.4|=0.2+0.2+0.1+0.5+0.2+0.2+0.1=1.5. Then, it calculates the sum of the absolute values ​​of existing feature parameters: |5.0|+|2.0|+|2.1|+|14.5|+|2 0.2|+|9.8|+|3.4|=5.0+2.0+2.1+14.5+20.2+9.8+3.4=57, yielding a similarity coefficient of 1.5÷57≈0.026. A preset threshold of 0.8 is set. Since the similarity coefficient 0.026 is less than the preset threshold 0.8, it is determined that there is no difference, and the existing data is retained. If the similarity coefficient calculated between the new feature and the existing feature is 0.85, which is greater than or equal to the preset threshold 0.8, it is determined that there is a difference, and the existing data is discarded. Replace with new scattering features; when replacing data, record the replacement time as 20250101143000, the reason for the change as the physical target material wear depth reaching 0.8mm, and the original parameters as the values ​​of the existing features; when supplementing the scattering feature data corresponding to the new scene, first determine the degree of correlation between the new scene (observation angle 75°, horizontal polarization, target severe wear, temperature 35℃ / humidity 65%RH) and the existing scene, and classify the new feature data into the corresponding category or create a new category according to the degree of correlation.

[0054] Understandably, clarifying the calculation method of the similarity coefficient provides a quantitative standard for feature difference judgment. Based on the difference judgment mechanism of preset threshold (replace if less than the threshold, retain if greater than or equal to the threshold), accurate iteration of feature library data is achieved. This ensures timely updates of differentiated features, avoids redundant update operations of indifferent features, reduces storage redundancy in the feature library, and records key information of the replaced data, providing support for feature library data traceability.

[0055] In some embodiments of this application, when determining the degree of correlation between a new scene and an existing scene, the matching degree between the environmental conditions, observation angle, polarization state of the new scene and the corresponding parameters of the existing scene is calculated. The matching degree is calculated as the ratio of the number of identical parameters to the total number of parameters. When the matching degree is greater than the preset ratio, the new feature data will be classified into the corresponding existing scene category. When the matching degree is less than or equal to the preset ratio, a new scene category is created and new feature data is stored. The updated scattering feature library is validated to check the completeness and format correctness of the feature parameters. If a parameter is missing or the format is incorrect, the corresponding scattering feature is re-extracted and corrected. If the parameter is complete and the format is correct, the update process is completed.

[0056] Specifically, the preset ratio is a critical value used to determine whether new feature data should be classified into an existing scene category or a newly created scene category when judging the correlation between a new scene and an existing scene. This value is a core judgment indicator that balances the rationality of feature library classification and retrieval convenience. Its value is determined based on the influence weight of scene parameters on scattering features and the coverage of existing scene categories. When the parameter matching degree between the new scene and the existing scene is greater than the preset ratio, the correlation between the two is considered high, and the new feature data can be directly classified into the corresponding existing scene category. When the matching degree is less than or equal to the preset ratio, the correlation between the two is considered low, and a dedicated category needs to be created for the new scene and the new feature data should be stored. The size of the preset ratio directly affects the number of categories and retrieval efficiency of the feature library. The larger the value, the stricter the requirements for scene correlation. More categories will increase the matching time during retrieval. The smaller the value, the more lenient the scene correlation judgment. Fewer categories may result in the coexistence of features from different scenes. The operation of obtaining the preset ratio is to first comprehensively sort out the factors affecting scattering features. The scene parameters for the feature were defined, with observation angle, polarization state, and environmental conditions as the core parameters. The distribution of each parameter dimension in existing scene classifications was then statistically analyzed to determine the influence weight of each core parameter on the scattering features (observation angle 30%, polarization state 40%, environmental conditions 30%). Next, 100 sets of parameter combination samples of new and existing scenes were selected, and the parameter matching degree of each set was calculated using a weighted similarity calculation method (rather than simple counting). The distribution range of the matching degree was statistically analyzed to be 40%-80%. Subsequently, three candidate preset ratios of 50%, 60%, and 70% were set, and classification tests were conducted at each ratio. The number of classifications, feature retrieval response time, classification accuracy, and other indicators were recorded under different ratios. Combined with the requirements that the feature library retrieval response time should not exceed 2 seconds and the classification accuracy should not be lower than 90%, the candidate ratio of 60% was selected as the one that met the requirements of all indicators and had the best overall performance. Finally, through multiple rounds of real-world scene feature classification verification, test deviations caused by parameter anomalies were eliminated, and the specific value of the preset ratio was finally determined.

[0057] Specifically, when determining the correlation between a new scene (environmental conditions: temperature 35℃, humidity 65%RH; observation angle: 75°; polarization state: horizontal polarization) and an existing scene, a weighted similarity is used to calculate the matching degree between the environmental conditions, observation angle, and polarization state of the new scene and the corresponding parameters of the existing scene. The similarity calculation method for each parameter is as follows: a temperature difference ≤ 2℃ is considered similar (similarity 1), otherwise the difference is linearly reduced; a humidity difference ≤ 5%RH is considered similar; a difference in observation angle ≤ 5° is considered similar; if the polarization states are the same, the similarity is 1, otherwise it is 0. The weighted matching degree calculation formula is: matching degree = 0.3 × temperature similarity + 0.3 × humidity similarity + 0.4 × polarization state similarity. The preset ratio is set to 60%. When the number of identical parameters between the new scene and the existing scene A (environmental conditions: temperature 35℃, humidity 60%RH; observation angle: 60°; polarization state: horizontal polarization) is 2, the matching degree is 2÷3≈66.7%. This matching degree is greater than the preset ratio of 60%, so the new feature data is classified into the corresponding existing scene A category. When the number of identical parameters between the new scene and the existing scene B (environmental conditions: temperature 25℃, humidity 50%RH; observation angle: 30°; polarization state: vertical polarization) is 0, the matching degree is 0÷3=0%. This matching degree is less than or equal to the preset ratio of 60%, so a new scene category is created and the new feature data is stored. The updated scattering feature library is validated to check the completeness and format correctness of the feature parameters. If any parameters are missing (e.g., the z-axis coordinate parameter of the scattering center is missing) or the format is incorrect (e.g., the period parameter unit is mistakenly written as s instead of ms), the corresponding scattering features are re-extracted (the z-axis coordinate is supplemented by 10cm, and the period unit is corrected to ms) and then corrected. When the parameters are complete (including waveform peak value of 5.2V, valley value of 1.8V, period of 2ms, scattering center x15cm, y20cm, z10cm, intensity amplitude of 3.5V) and the format is correct (all parameter units are consistent and the identification is standardized), the update process is completed.

[0058] Understandably, by calculating the parameter matching degree between new and existing scenarios and combining it with a preset ratio, the degree of scenario correlation is quantitatively determined. This avoids the redundancy caused by misclassifying highly correlated new scenarios as new categories, and also prevents the feature pooling caused by classifying low-correlation new scenarios into existing categories. This ensures the rationality and completeness of the feature library classification. Furthermore, the updated feature library verification mechanism further improves the quality of the feature library data by checking the completeness and format correctness of parameters and correcting them in a timely manner, providing reliable data support for the identification, analysis, and application of target scattering features.

[0059] It should be noted that: Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of this application may be practiced without these specific details. In some instances, well-known structures and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0060] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of this application and form different embodiments.

[0061] The above description is merely a preferred embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for dynamically constructing and updating a target scattering feature library based on digital twins, characterized in that, include: Build a digital twin model that matches the physical target in size, structure, and material; configure virtual signal transmitting and receiving devices in the digital twin model; and construct a virtual scene that integrates the physical target and signal equipment. The system performs signal transmission operations in a virtual scene, calculates scattering signal data under different observation angles and polarization states, and simultaneously acquires scattering signal data and target state data from the actual environment through acquisition components set on the physical target. The actual data is then transmitted to the digital twin model to calibrate the scattering data in the virtual scene. Scattering features are extracted from the calibrated virtual scattering data and the actual scattering data. The scattering features include signal waveform, scattering center position and signal intensity. The extracted scattering features are organized in a unified format and stored in the initial scattering feature library. The system continuously captures data on the state changes of physical targets and environmental changes through acquisition components, synchronizes the changes to the digital twin model, recalculates the scattering signal data in the virtual scene based on the changes, extracts new scattering features, compares the new scattering features with the existing data in the scattering feature library, replaces the existing data that differs, and supplements the scattering feature data corresponding to the new scene. When transmitting actual data to the digital twin model, the data is synchronized in real time. Any missing data during the transmission process is supplemented according to the correspondence between the actual data and the virtual data at adjacent moments. When calibrating scattering data in a virtual scene, the difference between the actual scattering data and the virtual scattering data is calculated, and the parameters of the virtual scene are adjusted based on the difference. When the difference is within the preset range, the first ratio of the difference to the upper limit of the preset range is calculated, and the attenuation coefficient of the virtual signal is linearly adjusted according to the first ratio. When the difference exceeds the preset range, if the difference is greater than the upper limit of the preset range, the real part of the material dielectric constant of the digital twin model is increased by the second ratio of the difference to the virtual scattering data, and the transmission power parameter of the virtual signal is increased proportionally to the absolute value of the difference; if the difference is less than the lower limit of the preset range, the real part of the material dielectric constant of the digital twin model is decreased by the second ratio, and the transmission power parameter of the virtual signal is decreased proportionally to the absolute value of the difference.

2. The method for dynamically constructing and updating a target scattering feature library based on digital twins according to claim 1, characterized in that, When building a digital twin model, first obtain the size parameters, structural distribution information, and material attribute information of the physical target, and then restore the geometric structure and material distribution of the model according to the actual shape of the physical target; when configuring virtual signal transmitting and receiving devices, plan the deployment positions of the devices according to the signal transmission path; when constructing an integrated virtual scene, associate the operating parameters of the digital twin model and the virtual signal device. When the physical target has detachable parts, the digital twin model is constructed by dividing it into modules, and the virtual signal devices are configured separately according to the module structure; when the physical target is an integral and indivisible structure, an integrated digital twin model is constructed, and the virtual signal devices are configured uniformly according to the overall structure.

3. The method for dynamically constructing and updating a target scattering feature library based on digital twins according to claim 2, characterized in that, When constructing a modular structure for a digital twin model by component breakdown, the connection relationships and assembly parameters of each module are recorded. When the virtual signal device is configured separately according to the modular structure, an independent signal transmitting sub-device and signal receiving sub-device are configured for each module, and signal interaction channels between modules are set up at the same time. When the connection relationship between modules changes, if the change is a relocation of connection location, the deployment coordinates of the signal transmitting sub-devices and signal receiving sub-devices are replanned according to the new connection location, and the signal coverage of the sub-devices is aligned with the new connection area of ​​the module. At the same time, the transmission bandwidth and delay parameters of the signal interaction channel are adjusted according to the new connection distance. If the change is an increase or decrease in the number of connections, the number of signal transmitting sub-devices and signal receiving sub-devices is increased or decreased according to the increased or decreased number of module connections. The transmission resources of the signal interaction channel are reallocated, and the number of parallel transmission paths of the channel is adjusted. When the connection relationship between the modules remains unchanged, the initial configuration of the virtual signal device and the parameters of the signal interaction channel are maintained. When configuring the operating parameters of the virtual signal device, set the frequency range and power range of the signal transmission.

4. The method for dynamically constructing and updating a target scattering feature library based on digital twins according to claim 3, characterized in that, When performing signal transmission operations in a virtual scene, the observation angle of the signal transmission device is adjusted at preset angle intervals, and the polarization state of the signal is switched according to a preset polarization mode, generating scattered signal data under different combinations of observation angles and polarization states one by one. When calculating the scattered signal data, the signal change data after the signal interacts with the model surface is derived based on the material properties and geometric structure of the digital twin model. When acquiring actual data through the acquisition component, the deployment position of the acquisition component corresponds to the deployment position of the virtual signal receiving device, and the acquisition frequency of the acquisition component matches the transmission frequency of the virtual signal. When there are obstructions in the actual environment, the number of data acquisition components deployed is increased to expand the acquisition range to cover the unobstructed target area; When there are no obstructions in the actual environment, the actual data is obtained according to the initial deployment plan.

5. The method for dynamically constructing and updating a target scattering feature library based on digital twins according to claim 4, characterized in that, When extracting scattering features, the calibrated virtual scattering data and the actual scattering data are first denoised. Then, the peak, valley, and period parameters of the signal waveform, the three-dimensional coordinate parameters of the scattering center, and the amplitude parameters of the scattering intensity are extracted from the denoised data. When organizing the scattering features in a unified format, different types of feature parameters are arranged in a preset order and each feature parameter is assigned a unique identifier. When there are duplicate extracted feature parameters, retain the feature parameters with the same value and remove duplicate records. When the extracted feature parameters are not duplicated, they are directly organized according to a uniform format; When storing the sorted scattering features into the initial scattering feature library, establish the correspondence between the feature parameters and the observation angle, polarization state, target state, and environmental conditions.

6. The method for dynamically constructing and updating a target scattering feature library based on digital twins according to claim 5, characterized in that, When establishing the correspondence between feature parameters and related conditions, an index is constructed according to the hierarchical structure of observation angle, polarization state, target state, and environmental conditions; when it is necessary to query scattering features under specific conditions, the corresponding feature parameters are located through the hierarchical index; the initial scattering feature library adopts a partitioned storage method, dividing the scattering features corresponding to different target states into different storage partitions; When the number of target state types is less than the type number threshold, a single storage partition is used to store all scattering features. When the number of target state types is greater than or equal to the threshold for the number of types, the storage partitions are split according to the target state type.

7. The method for dynamically constructing and updating a target scattering feature library based on digital twins according to claim 6, characterized in that, When continuously capturing data through the acquisition component, the captured physical target state change data includes material wear data and structural deformation data, and the environmental change data includes temperature and humidity data and interference signal data. The acquisition component captures data according to a preset cycle. When the captured data shows that the change in the physical target state is greater than the preset value, the acquisition cycle is shortened. When the captured data shows that the change in the physical target state is less than or equal to the preset value, the original acquisition cycle is maintained. When synchronizing changed data to the digital twin model, the timestamp and location identifier of the captured data are attached, and the digital twin model updates its own parameters according to the time sequence and location information; When recalculating the scattering signal data in the virtual scene based on the changed data, the model parameters and signal parameters corresponding to the changed data are called, and the scattering data calculation process is re-executed.

8. The method for dynamically constructing and updating a target scattering feature library based on digital twins according to claim 7, characterized in that, After extracting new scattering features, a feature comparison algorithm is used to calculate the similarity coefficient between the new scattering features and the existing data in the scattering feature database. The similarity coefficient is calculated as the ratio of the sum of the absolute values ​​of the differences between the corresponding parameters of the new features and the existing features to the sum of the absolute values ​​of the parameters of the existing features. When the similarity coefficient is greater than or equal to the preset threshold, it is determined that there is a difference, and the existing data is replaced with the new scattering feature. When the similarity coefficient is less than the preset threshold, it is determined that there is no difference and the existing data is retained; When replacing data, record the replacement time, the reason for the change, and related parameters; When supplementing scattering feature data for new scenes, first determine the degree of correlation between the new scene and the existing scene, and then classify the new feature data into the corresponding category or create a new category according to the degree of correlation.

9. A method for dynamically constructing and updating a target scattering feature library based on digital twins according to claim 8, characterized in that, When determining the degree of correlation between a new scene and an existing scene, the matching degree between the environmental conditions, observation angle, polarization state of the new scene and the corresponding parameters of the existing scene is calculated. The matching degree is calculated as the ratio of the number of identical parameters to the total number of parameters. When the matching degree is greater than the preset ratio, the new feature data will be classified into the corresponding existing scene category. When the matching degree is less than or equal to the preset ratio, a new scene category is created and new feature data is stored. The updated scattering feature library is validated to check the completeness and format correctness of the feature parameters. If a parameter is missing or the format is incorrect, the corresponding scattering feature is re-extracted and corrected. If the parameter is complete and the format is correct, the update process is completed.

Images