Real-time mapping digital twin archive warehouse management system

Abstract

The present application relates to a kind of real-time mapping digital twin archives management system, it is related to digital twin technical field, including the following steps, space point cloud is obtained by laser scanning, combined with radio frequency identification signal and is solved archive estimated coordinate and is registered to obtain initial space node;Based on point cloud reconstruction three-dimensional grid sheet, the initial node is connected and combined to build twin warehouse topology;Real-time acquisition current space node, calculate its relative displacement vector with initial node, according to this generate physical scheduling instruction and virtual update node, the continuous, accurate topology corresponding relationship between physical archives and virtual space node that traditional technology is difficult to establish is solved, it leads to the problem that accurate physical scheduling instruction cannot be generated to drive real-time synchronization of virtual and real state.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology, and in particular to a real-time mapping digital twin archive management system. Background Technology

[0002] With the advancement of information management of archives, digital twin technology is gradually being applied to the visual management of archive storage facilities. Existing technologies typically employ 3D modeling software to pre-construct a static 3D model of the storage facility, and combine this with radio frequency identification (RFID) technology to identify and record the location of archival entities, thereby achieving a digital display of the storage facility's status. In practical applications, this type of solution often relies on periodic manual inventory checks or area sensing by fixed readers to obtain archival location information, and then maps the acquired data to a pre-set static model for updates, thus achieving, to a certain extent, digital monitoring and basic management functions for the archive storage environment.

[0003] However, existing digital twin archive management methods, when dealing with frequent access to archives and dynamic changes in their physical location, lack a mechanism for synchronizing and integrating the spatial geometry of the archive with the real-time location of the archives. As a result, the archive nodes in the virtual model cannot automatically and accurately update their spatial coordinates as the physical entities move. They can only rely on discrete signal triggers for coarse-grained position updates, making it difficult to establish a continuous and accurate topological correspondence between physical archives and virtual spatial nodes. Consequently, they cannot generate accurate physical scheduling instructions to drive real-time synchronization between the virtual and physical states. Summary of the Invention

[0004] The main objective of this invention is to provide a real-time mapping digital twin archive management system, which solves the problem that traditional technologies struggle to establish a continuous and accurate topological correspondence between physical archives and virtual space nodes, resulting in the inability to generate accurate physical scheduling instructions to drive real-time synchronization between the virtual and physical states.

[0005] To achieve the above objectives, the present invention provides a scanning module for continuously scanning a physical archive storage room using a laser scanning device to obtain spatial point cloud data; The acquisition module is used to acquire signals from the physical archives in the physical archives storage room based on the radio frequency identification device, and obtain the archive radio frequency signals; The calculation module is used to perform spatial positioning calculation on the radio frequency signal of the archive to obtain the estimated coordinates of the archive, and to perform coordinate registration on the estimated coordinates of the archive based on the spatial coordinate system constructed by the spatial point cloud data to obtain the initial spatial nodes; The reconstruction module is used to perform surface triangulation reconstruction based on the spatial point cloud data to obtain a three-dimensional mesh patch, and to perform connectivity combination on the initial spatial nodes based on the three-dimensional mesh patch to obtain the twin warehouse topology. The acquisition module is used to acquire the current spatial node in real time, calculate the relative displacement between the initial spatial node and the current spatial node in the twin warehouse topology to obtain the spatial displacement vector, and generate physical scheduling instructions and virtual update nodes based on the spatial displacement vector.

[0006] Furthermore, the continuous scanning of the physical archive storage room using a laser scanning device to obtain spatial point cloud data includes: The ranging module is used to perform multi-view ranging on the physical surface of the physical archive storage room through the laser scanning device, obtain multi-view reflection points, and perform local coordinate mapping on the multi-view reflection points to obtain local three-dimensional coordinate points. The filtering module is used to perform spatial density filtering on the local three-dimensional coordinate points to obtain a skeleton feature point cloud, and to perform global coordinate stitching on the skeleton feature point cloud to obtain the spatial point cloud data.

[0007] Furthermore, the step of acquiring signals from the physical archives within the physical archive repository using radio frequency identification (RFID) equipment to obtain the archive radio frequency signals includes: The radio frequency identification device emits radio frequency excitation waves to the physical archives in the physical archives warehouse to obtain tag reflection waves, and performs multipath time delay separation on the tag reflection waves to obtain direct path waveforms; The direct path waveform is envelope-detected and demodulated to obtain the baseband payload code stream, and the carrier phase is extracted from the direct path waveform based on the baseband payload code stream to obtain the archive radio frequency signal.

[0008] Furthermore, the step of transmitting radio frequency excitation waves to the physical archives in the physical archive repository through the radio frequency identification device to obtain tag reflected waves includes: The radio frequency identification device transmits a single-frequency continuous wave to the physical archives in the physical archives repository to obtain a backscattered radio frequency stream, and then amplifies the backscattered radio frequency stream with low noise to obtain a primary radio frequency signal. The primary radio frequency signal is subjected to quadrature coherent demodulation to obtain a complex baseband waveform, and the complex baseband waveform is subjected to frame synchronization detection and signal amplitude normalization to obtain the tag reflection wave.

[0009] Furthermore, the step of performing spatial positioning calculations on the radio frequency signals of the archives to obtain the estimated coordinates of the archives includes: The phase difference of the radio frequency signal of the archive is calculated to obtain the signal phase difference value, and the angle of arrival is calculated based on the signal phase difference value to obtain the spatial incident angle. Obtain the array antenna coordinates of the radio frequency identification device, and perform ray projection on the array antenna coordinates based on the spatial incident angle to obtain a spatial intersecting ray; The intersection vertices of the spatial intersecting rays are extracted to obtain the estimated coordinates of the archive.

[0010] Furthermore, the surface triangulation reconstruction based on the spatial point cloud data to obtain a three-dimensional mesh patch includes: The spatial point cloud data is fitted with a local surface to obtain the point cloud surface normal vector, and a spatial neighborhood search is performed on the spatial point cloud data based on the point cloud surface normal vector to obtain a local neighboring point set. The local adjacent point set is spatially connected and projected to obtain a planar triangular profile, and the surface boundary is stitched based on the planar triangular profile to obtain the three-dimensional mesh patch.

[0011] Furthermore, the step of performing a spatial neighborhood search on the spatial point cloud data based on the point cloud surface normal vector to obtain a local neighboring point set includes: A spatial sphere domain search is performed on the spatial point cloud data to obtain initial neighborhood scattered points, and the normal angle of the initial neighborhood scattered points is extracted based on the surface normal vector of the point cloud to obtain the surface normal angle. Based on the surface normal angle, the initial neighborhood scattered points are filtered for orientation differences to obtain a set of candidate points in the same direction. Then, based on the surface normal vector of the point cloud, the candidate point set in the same direction is projected onto the normal vector to obtain the normal projection spacing. Based on the normal projection spacing, the same-direction candidate point set is subjected to cross-layer step rejection to obtain continuous surface nodes, and the continuous surface nodes are combined spatially to obtain the local adjacent point set.

[0012] Furthermore, the step of calculating the normal projection of the candidate point set in the same direction based on the normal vector of the point cloud surface to obtain the normal projection spacing includes: The mean of spatial coordinates is extracted from the candidate point set in the same direction to obtain the local spatial centroid, and a normal reference ray is constructed based on the local spatial centroid and the normal vector of the point cloud surface. Based on the normal reference ray, an orthogonal projection transformation is performed on the same-direction candidate point set to obtain the normal projection coordinates; The axial relative distance between the normal projection coordinates and the origin coordinates of the normal reference ray is calculated to obtain the normal projection spacing.

[0013] Furthermore, the step of performing cross-layer step rejection on the same-direction candidate point set based on the normal projection spacing to obtain continuous surface nodes includes: Based on the normal projection spacing, the same-direction candidate point set is sorted in one dimension to obtain an ordered projection point column, and the adjacent node difference is calculated on the ordered projection point column to obtain the local spacing gradient. The peak extrema of the local spacing gradient are extracted to obtain the step breakpoint index, and the ordered projection point sequence is truncated and separated based on the step breakpoint index to obtain the continuous surface nodes.

[0014] This invention provides a real-time mapping digital twin archive storage management system, comprising: a scanning module for continuously scanning a physical archive storage facility using a laser scanning device to obtain spatial point cloud data; an acquisition module for acquiring signals from physical archives within the physical archive storage facility using a radio frequency identification (RFID) device to obtain archive radio frequency signals; a calculation module for performing spatial positioning calculations on the archive radio frequency signals to obtain estimated archive coordinates, and performing coordinate registration on the estimated archive coordinates using a spatial coordinate system constructed based on the spatial point cloud data to obtain initial spatial nodes; and a reconstruction module for performing surface triangulation reconstruction based on the spatial point cloud data to obtain three-dimensional mesh patches, and reconstructing the initial spatial nodes using the three-dimensional mesh patches. The nodes are combined for connectivity to obtain a twin warehouse topology. The acquisition module is used to acquire the current spatial nodes collected in real time, calculate the relative displacement between the initial spatial nodes and the current spatial nodes in the twin warehouse topology to obtain a spatial displacement vector, and generate physical scheduling instructions and virtual update nodes based on the spatial displacement vector. This solves the problem that traditional technologies have difficulty in establishing a continuous and accurate topological correspondence between physical archives and virtual spatial nodes, which leads to the inability to generate accurate physical scheduling instructions to drive real-time synchronization of virtual and physical states. It realizes strong spatiotemporal coupling and topological adaptive synchronization between physical archive entities and digital twin models in the dynamic scheduling process, ensuring the continuity of virtual and physical state mapping and the accuracy of instruction execution. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of a real-time mapping digital twin archive management system according to an embodiment of the present invention; Figure 2 This is one embodiment of the present invention. Figure 1 A schematic diagram of the scanning module.

[0017] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] The embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0019] The following describes in detail, with reference to the accompanying drawings, a real-time mapping digital twin archive management system according to an embodiment of the present invention. First, the real-time mapping digital twin archive management system according to an embodiment of the present invention will be described with reference to the accompanying drawings.

[0020] like Figure 1 As shown, Figure 1 This invention provides a real-time mapping digital twin archive management system, comprising: Scanning module 1 is used to continuously scan the physical archive storage room using a laser scanning device to obtain spatial point cloud data.

[0021] Specifically, scanning module 1 is used to deploy a 3D laser scanner within the physical archive storage room. By setting scanning resolution and frame rate parameters, it continuously acquires data from the storage room walls, mobile shelving, and aisles along a preset trajectory, obtaining a raw point cloud sequence containing spatial location information. During the acquisition process, device pose data is recorded simultaneously. A multi-frame point cloud registration algorithm is used to eliminate accumulated errors, unifying the discrete point clouds into a global coordinate system to form complete spatial point cloud data. This data accurately represents the internal geometry and obstacle distribution of the storage room. In this embodiment, the above scheme provides a high-precision spatial reference framework for subsequent archive coordinate registration and twin topology construction.

[0022] Acquisition module 2 is used to acquire signals from the physical archives in the physical archives storage room based on radio frequency identification equipment, and obtain the archive radio frequency signals.

[0023] Specifically, acquisition module 2 is used to deploy an ultra-high frequency radio frequency identification (RFID) reader and a distributed antenna array within the physical archive storage room. It continuously transmits excitation signals at a preset polling period to activate passive electronic tags attached to the surface of the archive boxes. After receiving the modulated signal backscattered from the tag, the reader demodulates and extracts the tag's unique code, received signal strength indicator value, and phase information. These parameters are then bound and encapsulated into a structured data stream along with the acquisition time and antenna port number. To address multipath interference caused by the metal environment of the mobile shelving unit, a sliding window is used to perform mean filtering on the continuous signal strength, retaining stable feature components as effective output. In this embodiment, the above scheme achieves real-time and continuous acquisition of the physical archive identity and signal characteristics within the storage room, providing reliable raw data support for subsequent spatial positioning calculations.

[0024] Solution module 3 is used to perform spatial positioning solution on the radio frequency signal of the archive to obtain the estimated coordinates of the archive, and to perform coordinate registration on the estimated coordinates of the archive based on the spatial coordinate system constructed based on the spatial point cloud data to obtain the initial spatial nodes.

[0025] Specifically, the calculation module 3 utilizes the strength and phase difference of the archive's radio frequency signal received by multiple antennas to calculate the estimated three-dimensional coordinates of the physical archive in the local coordinate system of the radio frequency device using an angle-of-arrival estimation algorithm. Subsequently, it extracts preset calibration reference points from the spatial point cloud data and employs an iterative nearest-point algorithm to rigidly transform and register the radio frequency local coordinate system with the point cloud global spatial coordinate system, eliminating translation and rotation deviations between the two. This ensures that the estimated archive coordinates are accurately mapped to the point cloud coordinate system, generating initial spatial nodes representing the archive's true spatial location. In this embodiment, the above scheme achieves precise alignment between the radio frequency positioning results and the three-dimensional geometric space, providing a reliable spatial reference for twin topology construction.

[0026] Reconstruction module 4 is used to perform surface triangulation reconstruction based on the spatial point cloud data to obtain a three-dimensional mesh patch, and to perform connectivity combination on the initial spatial nodes based on the three-dimensional mesh patch to obtain a twin warehouse topology.

[0027] Specifically, reconstruction module 4 performs a Poisson surface reconstruction algorithm on the spatial point cloud data, generating closed 3D mesh patches based on the consistency of point normal vectors to accurately represent the geometric boundaries of the warehouse walls and mobile shelving. Then, using the initial spatial nodes as vertices, an adjacency matrix is ​​constructed based on the physical barrier relationships defined by the 3D mesh patches. Connectivity is determined only when the line connecting two nodes does not penetrate any mesh patch, thus establishing a topological connection relationship reflecting the actual passage path. Finally, a twin warehouse topology structure containing node positions and connected edge information is output. In this embodiment, the above scheme achieves semantic fusion between the spatial location of the archives and the physical structure of the warehouse, providing a reliable topological foundation for path planning.

[0028] The acquisition module 5 is used to acquire the current spatial node in real time, calculate the relative displacement between the initial spatial node and the current spatial node in the twin warehouse topology to obtain the spatial displacement vector, and generate physical scheduling instructions and virtual update nodes based on the spatial displacement vector.

[0029] Specifically, module 5 acquires the current location of the archive in real time using an RFID device as the current spatial node. It then performs a three-dimensional coordinate difference operation between this current location and the initial spatial node recorded in the twin warehouse topology to obtain a spatial displacement vector representing the direction and distance of the position offset. When the magnitude of this vector exceeds a preset trigger threshold, the system generates a physical scheduling command to move the mobile shelving unit based on its directional component. Simultaneously, it superimposes this displacement vector using the initial spatial node as a reference to generate a virtual update node reflecting the latest state of the archive, ensuring that the digital model and the physical state evolve synchronously. In this embodiment, the above scheme achieves the linkage update of physical scheduling and digital twin state, ensuring consistency between the virtual and physical systems.

[0030] In a specific embodiment, such as Figure 2 As shown, the step of continuously scanning the physical archive storage room using a laser scanning device to obtain spatial point cloud data includes: The ranging module 11 is used to perform multi-view ranging on the physical surface of the physical archive room through the laser scanning device, obtain multi-view reflection points, and perform local coordinate mapping on the multi-view reflection points to obtain local three-dimensional coordinate points. The filtering module 12 is used to perform spatial density filtering on the local three-dimensional coordinate points to obtain a skeleton feature point cloud, and to perform global coordinate stitching on the skeleton feature point cloud to obtain the spatial point cloud data.

[0031] Specifically, a 3D laser scanning device is deployed within the physical archive storage room. A turntable drives the laser to continuously rotate and scan the interior walls, mobile shelving, and file box surfaces at a preset angular resolution, acquiring the time-of-flight and echo intensity at each measurement point. Due to blind spots in single-station scanning, multiple scanning stations are planned within the storage room. Each station collects a set of multi-view reflection points, initially containing only polar coordinate distance and angle information relative to the current scanner center. To transform the raw observations into computable geometric data, the system maps each set of multi-view reflection points from the polar coordinate system to a local Cartesian coordinate system with the current station as the origin, based on the scanner's internal calibration parameters. This generates local 3D coordinate points, whose spatial position is uniquely determined by three orthogonal components: the horizontal axis, the vertical axis, and the elevation.

[0032] Considering the specular reflection noise easily generated on the surface of metal mobile shelving units in the archive storage room, and the potential introduction of outliers by airborne particles, a spatial density filtering operation needs to be performed on local 3D coordinate points. Specifically, the number of valid points is counted within a spherical neighborhood with a radius of five centimeters centered on any local 3D coordinate point. If the number of points in this neighborhood is lower than a preset density threshold, the point is determined to be isolated noise and is removed; otherwise, the point cloud retains the skeletal feature points representing key geometric structures such as the mobile shelving units' columns, beams, or wall transitions. This filtering mechanism ensures that subsequent processing relies only on high-confidence structural features, avoids redundant data interfering with registration accuracy, and significantly reduces the amount of point cloud data to improve stitching efficiency.

[0033] After extracting the skeleton feature point clouds for each station, the global coordinate stitching stage begins. Utilizing the shared skeleton feature point clouds of overlapping areas between adjacent stations, the optimal rigid transformation parameters between the two local coordinate systems are calculated using an iterative nearest-point algorithm. These parameters include three-axis translation and three-axis rotation angles, ensuring precise spatial alignment of corresponding feature points. This stitching process is sequentially passed, uniformly transforming all local 3D coordinate points to the predefined origin of the warehouse's global coordinate system, ultimately fusing them to form a complete spatial point cloud covering the entire physical archive warehouse. Throughout the stitching chain, each transformation is based on the registration result of the previous station. Since errors accumulate, the registration residual for each station must be strictly controlled within two millimeters to ensure that the deviation at the end of the warehouse, spanning tens of meters, still meets the engineering acceptance standards. Furthermore, the origin of the global coordinate system is typically set at a corner of the warehouse entrance, with its axis parallel to the building's main axis, facilitating subsequent cross-modal alignment with architectural drawings or RFID positioning systems. In this embodiment, the above scheme achieves a high-density, low-noise, and coordinate-unified 3D geometric representation of the warehouse, providing an accurate spatial basis for twin modeling.

[0034] In a specific embodiment, the step of acquiring signals from physical archives within the physical archive repository using a radio frequency identification (RFID) device to obtain archive radio frequency signals includes: The radio frequency identification device emits radio frequency excitation waves to the physical archives in the physical archives warehouse to obtain tag reflection waves, and performs multipath time delay separation on the tag reflection waves to obtain direct path waveforms; The direct path waveform is envelope-detected and demodulated to obtain the baseband payload code stream, and the carrier phase is extracted from the direct path waveform based on the baseband payload code stream to obtain the archive radio frequency signal.

[0035] Specifically, an ultra-high frequency radio frequency identification (RFID) reader is deployed within the physical archive storage room. Its antenna array emits a 920 MHz radio frequency excitation wave at a fixed tilt angle towards the mobile shelving aisle. When this excitation wave illuminates a passive electronic tag affixed to the spine of the archive box, the tag chip loads its stored information onto the reflected signal via backscatter modulation, forming a tag reflection wave carrying coded data that returns to the reader receiver. Due to the complex electromagnetic environment created by the metal mobile shelving, walls, and floor within the storage room, the received signal is actually a composite waveform of the direct path and multiple reflected paths, each exhibiting different time delays due to varying propagation distances. The system first performs multipath delay separation on the tag reflection wave, using channel impulse response estimation technology to identify the peak component with the strongest energy and shortest delay. This component corresponds to the straight-line propagation path between the reader antenna and the tag, and is extracted as the direct path waveform. The remaining delayed reflection components are filtered out to avoid inter-symbol interference affecting subsequent demodulation accuracy.

[0036] After obtaining a clean direct waveform, the system proceeds to the baseband signal recovery stage. Envelope detection and demodulation are performed on this waveform. By detecting the amplitude variation profile of the radio frequency carrier, the low-frequency modulation signal is extracted and then low-pass filtered and thresholded to restore it into a binary baseband payload stream. This stream fully carries the unique identifier of the file tag and the content of the user data area.

[0037] Meanwhile, to obtain high-precision RF features suitable for phase positioning, the system uses the same baseband payload code stream as a synchronization reference to perform coherent carrier phase extraction on the original direct path waveform. Specifically, a local reference oscillator is constructed using known pilot sequences or preamble segments in the code stream, and orthogonally mixed with the direct path waveform to output two baseband components: in-phase and quadrature. The instantaneous phase value is then calculated using arctangent operations. This phase value eliminates the phase jump introduced by tag backscatter modulation, retaining only continuous phase information linearly related to the propagation distance. The final output is a complete RF signal containing amplitude, phase, and timestamp. The entire processing chain strictly follows a unidirectional causal sequence of "first separating the direct path, then demodulating the code stream, and finally using the code stream to assist in phase extraction," ensuring that phase extraction is not affected by multipath effects and modulation noise coupling.

[0038] In practical deployments, considering that the spacing between mobile shelving units is typically 0.5 to 0.7 meters, the radio frequency wavelength is approximately 32 centimeters, and the time delay difference between adjacent reflection paths is on the order of 1 nanosecond, the time resolution of the multipath separation algorithm needs to be better than 0.5 nanoseconds to effectively distinguish direct paths. This parameter setting conforms to the engineering practice standards for UHF RFID systems in dense metal environments. In this embodiment, the above scheme achieves reliable extraction of high-purity phase information from the archive's radio frequency signal, providing high-quality raw observation data for subsequent phase-based precise positioning.

[0039] In a specific embodiment, the step of transmitting radio frequency excitation waves to the physical archives in the physical archive repository through the radio frequency identification device to obtain tag reflected waves includes: The radio frequency identification device transmits a single-frequency continuous wave to the physical archives in the physical archives repository to obtain a backscattered radio frequency stream, and then amplifies the backscattered radio frequency stream with low noise to obtain a primary radio frequency signal. The primary radio frequency signal is subjected to quadrature coherent demodulation to obtain a complex baseband waveform, and the complex baseband waveform is subjected to frame synchronization detection and signal amplitude normalization to obtain the tag reflection wave.

[0040] Specifically, the RFID device continuously transmits a single-frequency continuous wave with a power of 30 dBmW at a center frequency of 920 MHz into the mobile shelving aisle within the physical archive storage room. When this electromagnetic wave strikes a passive electronic tag affixed to the surface of the file box, the tag's antenna impedance periodically switches according to the internal encoded data, thereby modulating the amplitude of the incident wave and forming a backscattered RF stream carrying information that returns to the reader's receiver. Because the backscattered signal's energy is significantly attenuated after propagation through space and obstruction by the metal shelving, its power reaching the receiving antenna is typically as low as -70 dBmW. Therefore, it must first be boosted by a low-noise amplifier with a fixed gain of 25 dB and a noise figure of 1.2 dB. This amplifier amplifies the weak signal to a level suitable for subsequent processing while minimizing its own thermal noise introduction, outputting a primary RF signal with a signal-to-noise ratio that meets demodulation requirements.

[0041] After acquiring the primary RF signal, the system employs an orthogonal coherent demodulation architecture to downconvert it to the baseband domain. Specifically, using a local oscillator signal that is co-originating with and phase-locked to the transmitted single-frequency continuous wave as a reference, it is mixed in phase and quadraturely with the primary RF signal, respectively. The two mixed outputs are low-pass filtered to form a complex baseband waveform with mutually orthogonal real and imaginary parts. This waveform completely preserves the amplitude and phase information of the original backscattered signal, and its center frequency is zero, facilitating digital domain processing. Subsequently, frame synchronization detection is performed on the complex baseband waveform. By calculating the cross-correlation peak value between the received waveform and the locally pre-stored preamble template within a sliding window, the frame start position is determined when the correlation value exceeds a preset decision threshold, achieving precise alignment of symbol boundaries. Next, signal amplitude normalization is performed. Based on the average power of the preamble segment after frame synchronization confirmation, the instantaneous amplitude of the entire complex baseband waveform is scaled to a uniform dimension to eliminate amplitude fluctuations caused by tag distance differences or antenna pattern fluctuations. Finally, the timing-aligned and amplitude-stable tag reflection wave is output for subsequent decoding and phase extraction.

[0042] The entire processing chain strictly follows a unidirectional causal sequence of amplification, demodulation, and synchronization normalization. Parameter settings for each stage are based on measured characteristics of the UHF RFID system in a dense metal environment. For example, the gain margin of the low-noise amplifier ensures a demodulation signal-to-noise ratio above 10 dB at a maximum reading distance of 8 meters, while the frame synchronization threshold is dynamically set according to the statistical distribution of background noise to avoid false alarms and missed detections. In this embodiment, the above scheme achieves high-fidelity recovery and standardized preprocessing of weak backscattered signals, laying a reliable signal foundation for subsequent accurate decoding and phase feature extraction.

[0043] In a specific embodiment, the step of performing spatial positioning calculation on the archive radio frequency signal to obtain the archive's estimated coordinates includes: The phase difference of the radio frequency signal of the archive is calculated to obtain the signal phase difference value, and the angle of arrival is calculated based on the signal phase difference value to obtain the spatial incident angle. Obtain the array antenna coordinates of the radio frequency identification device, and perform ray projection on the array antenna coordinates based on the spatial incident angle to obtain a spatial intersecting ray; The intersection vertices of the spatial intersecting rays are extracted to obtain the estimated coordinates of the archive.

[0044] Specifically, after acquiring the radio frequency signal of the archive, the system first performs phase difference calculation to extract spatial angle information. Specifically, for the reflected signal from the same tag received by two adjacent array elements in the array antenna, the carrier phase value is measured and the difference is calculated to obtain the signal phase difference. This difference is positively correlated with the path difference of the electromagnetic wave reaching the two array elements, and the path difference is determined by the cosine function of the angle between the incident direction and the line connecting the array elements. Since the array element spacing is fixed at half a wavelength, approximately 16 centimeters, the variation range of the signal phase difference is strictly limited to between negative π and positive π radians, thus avoiding phase ambiguity. Based on this physical constraint, the system maps the signal phase difference to a spatial incident angle through inverse cosine calculation. This angle characterizes the azimuth offset of the tag relative to the normal direction of the array antenna, with an accuracy within ±3 degrees, meeting the lateral resolution requirements of archives within the mobile shelving aisle.

[0045] After obtaining the spatial incident angle, it needs to be converted into geometric positioning elements. The system pre-calibrates the three-dimensional position and attitude parameters of the RFID device array antenna in the global coordinate system, which serve as the array antenna coordinates. Starting from these coordinates, an infinitely extending ray is constructed along the direction indicated by the spatial incident angle, called a spatial intersecting ray. When multiple array antennas are deployed in the warehouse, each antenna independently calculates one or more spatial intersecting rays. Theoretically, these rays should converge at the tag location in three-dimensional space. However, due to measurement noise and multipath residue, the actual rays often cannot precisely intersect at one point, but instead form a cluster of spatial straight lines that are close to each other but do not overlap. To address this, the system uses a least-squares intersection algorithm to extract the intersection vertex of all spatial intersecting rays, i.e., to find a spatial point that minimizes the sum of the squares of the perpendicular distances to each ray. This optimal estimated point is the file's estimated coordinate. In the implementation of the project, considering that the width of the aisle of the mobile shelving unit is typically 0.6 meters and the thickness of adjacent file boxes is approximately 5 centimeters, the required lateral resolution for positioning is no less than 3 centimeters. Therefore, the standard deviation of the phase differential measurement must be controlled within 0.1 radians, corresponding to a signal-to-noise ratio of no less than 20 dB. This specification is consistent with the performance settings of the aforementioned low-noise amplification and normalization processing stages, ensuring end-to-end accuracy matching. Furthermore, installation error compensation has been performed on the array antenna coordinates before ray projection to eliminate the systematic impact of mechanical assembly deviations on angle calculation. In this embodiment, the above solution achieves high-precision reconstruction of the three-dimensional position of files from pure phase information without additional ranging, providing reliable coordinate support for non-contact inventory checks in dense storage environments.

[0046] In a specific embodiment, the step of performing surface triangulation reconstruction based on the spatial point cloud data to obtain a three-dimensional mesh patch includes: The spatial point cloud data is fitted with a local surface to obtain the point cloud surface normal vector, and a spatial neighborhood search is performed on the spatial point cloud data based on the point cloud surface normal vector to obtain a local neighboring point set. The local adjacent point set is spatially connected and projected to obtain a planar triangular profile, and the surface boundary is stitched based on the planar triangular profile to obtain the three-dimensional mesh patch.

[0047] Specifically, after completing a 3D laser scan of the physical archive and acquiring spatial point cloud data in a unified coordinate system, the system first performs local surface fitting on the discrete point set to recover geometric and topological information. For any target point in space, a spherical search domain with a radius of three centimeters is defined centered on that point, and at least six effective sampling points are selected within this neighborhood to construct a local point cluster. Principal component analysis is used to perform eigenvalue decomposition on the covariance matrix of this point cluster, where the eigenvector corresponding to the smallest eigenvalue represents the surface orientation of the local region, and this vector is normalized and used as the surface normal vector of the point cloud.

[0048] The direction of this normal vector is strictly perpendicular to the tangent plane of the object being measured, such as the upright of the mobile shelving unit or the spine of the file box. Its directional consistency is corrected through global viewpoint constraints to ensure that all normal vectors point outward from the scanner rather than inward from the object. After obtaining the normal vector for each point, the system again performs a spatial neighborhood search with each point as the center and the same 3-centimeter radius. However, this time the filtering criteria include a threshold limit for the angle between the normal vectors, retaining only neighboring points with an angle less than 30 degrees to the center point's normal vector to form a local adjacent point set. This avoids erroneous connections across edges or different component surfaces, ensuring that the adjacency relationship truly reflects the continuous curved surface structure of the object.

[0049] Based on the establishment of reliable local adjacency relationships, the system proceeds to the triangular patch generation stage. The system orthogonally projects all points in the local adjacency set, along with the center point, onto a two-dimensional tangent plane passing through the center point and perpendicular to the normal vector, obtaining a set of planar coordinate points. Within this two-dimensional plane, the Delaunay triangulation algorithm is used to connect the projected points, generating a planar triangular contour that satisfies the maximum and minimum angle criterion. This contour has no intersecting edges in the two-dimensional domain and is as close as possible to an equilateral triangle, effectively avoiding elongated degenerate units. Subsequently, the vertices in the planar triangular contour are mapped back to their original three-dimensional spatial coordinates, and the triangular patches generated from adjacent local regions are stitched together along a common boundary. During the stitching process, when two adjacent point sets share two or more common points, the system forcibly merges the corresponding edges and unifies the normal direction, eliminating duplicate vertices and gaps. For areas with only partial overlap, the system determines whether to expand the stitching range based on curvature continuity, ultimately forming a watertight three-dimensional mesh patch covering the entire warehouse interior wall, mobile shelving, and file box surface.

[0050] The parameter settings in the above reconstruction process closely match the actual physical characteristics of the archive storage room. For example, the width of the crossbeams of the mobile shelving unit is usually four centimeters, and a search radius of three centimeters ensures that at least half of the component cross-section is covered without exceeding the boundary. The 30-degree normal angle threshold is adapted to the minimum curvature radius of the metal bending parts, avoiding cross-surface misconnections at acute angles. In this embodiment, the above scheme achieves a high-fidelity conversion from disordered scattered points to a structured mesh, providing a geometrically accurate and topologically continuous visualization basis for the twin model.

[0051] In a specific embodiment, the step of performing a spatial neighborhood search on the spatial point cloud data based on the point cloud surface normal vector to obtain a local neighboring point set includes: A spatial sphere domain search is performed on the spatial point cloud data to obtain initial neighborhood scattered points, and the normal angle of the initial neighborhood scattered points is extracted based on the surface normal vector of the point cloud to obtain the surface normal angle. Based on the surface normal angle, the initial neighborhood scattered points are filtered for orientation differences to obtain a set of candidate points in the same direction. Then, based on the surface normal vector of the point cloud, the candidate point set in the same direction is projected onto the normal vector to obtain the normal projection spacing. Based on the normal projection spacing, the same-direction candidate point set is subjected to cross-layer step rejection to obtain continuous surface nodes, and the continuous surface nodes are combined spatially to obtain the local adjacent point set.

[0052] Specifically, after acquiring spatial point cloud data and the corresponding point cloud surface normal vectors, the system first performs a search within a spatial sphere centered on the current point to be processed, with a radius of three centimeters. All sampled points within this sphere are included in the initial neighborhood scatter set. This radius is determined based on the dimensions of typical warehouse components such as the thickness of the shelving unit and the width of the file box spine. This ensures sufficient coverage of effective neighboring points to support subsequent geometric analysis while avoiding the inclusion of irrelevant points on discontinuous surfaces due to an excessively large search range. Subsequently, the system calculates the angle between the normal vector of each point in the initial neighborhood scatter set and the normal vector of the center point's point cloud surface, obtaining the surface normal angle. The magnitude of this angle directly reflects the degree of orientation consistency of the local surfaces where the two points are located. When the angle approaches zero, it indicates that the two points are located on the same smooth, continuous surface; conversely, it suggests the presence of edges or surface boundaries. Based on this physical relationship, the system sets 20 degrees as the orientation difference filtering threshold, and retains only points with a surface normal angle less than the threshold to form a candidate point set with the same orientation, thereby eliminating misconnected points that cross the turning point between the side and top of the compact shelving column, and ensuring that the orientation of the curved surfaces within the candidate point set is highly uniform.

[0053] To further eliminate interference points that, although oriented similarly, are actually at different depth levels, the system performs normal projection calculations on each point in the candidate point set. Specifically, the position vector of the candidate point relative to the center point is orthogonally projected along the direction of the normal vector of the center point's point cloud surface. The resulting scalar value is the normal projection spacing. This spacing characterizes the distance of the candidate point from the tangent plane of the center point in the normal direction; the smaller the absolute value, the closer the point is to the ideal continuous surface where the center point is located.

[0054] Considering the close arrangement of file boxes and the typical five-millimeter gap between layers, the system sets the upper limit of the normal projection spacing to two millimeters. Points exceeding this limit are considered cross-layer step points and rejected, while the remaining points are confirmed as continuous surface nodes. This threshold setting provides sufficient engineering margin, allowing for minor normal offsets caused by scanning noise or slight deformation of objects, while strictly preventing erroneous connections between file boxes in adjacent layers due to accidental fulfillment of the normal angle condition.

[0055] Finally, the selected continuous surface nodes are combined based on spatial connectivity. This involves constructing a graph structure with each node as a vertex, establishing an edge connection only when the Euclidean distance between two points is less than a preset connectivity threshold, and extracting the largest connected subgraph as the local set of adjacent nodes using breadth-first traversal. This connectivity threshold is consistent with the aforementioned spatial sphere search radius, both being three centimeters, ensuring geometric consistency in adjacency relationships. In this embodiment, the above scheme refines the neighborhood point set through multi-level geometric constraints, effectively suppressing cross-surface and cross-layer misconnections caused by the compact arrangement of components in dense storage scenarios, providing a reliable input with topological purity and surface continuity for subsequent triangulation reconstruction.

[0056] In a specific embodiment, the step of calculating the normal projection of the same-direction candidate point set based on the normal vector of the point cloud surface to obtain the normal projection spacing includes: The mean of spatial coordinates is extracted from the candidate point set in the same direction to obtain the local spatial centroid, and a normal reference ray is constructed based on the local spatial centroid and the normal vector of the point cloud surface. Based on the normal reference ray, an orthogonal projection transformation is performed on the same-direction candidate point set to obtain the normal projection coordinates; The axial relative distance between the normal projection coordinates and the origin coordinates of the normal reference ray is calculated to obtain the normal projection spacing.

[0057] Specifically, after obtaining the candidate point set in the same direction, the system first performs an arithmetic mean operation on the three-dimensional spatial coordinates of all points in the set to obtain the local spatial centroid. This centroid is not a sampling point on a physical entity, but rather a virtual reference position representing the center of the current neighborhood geometric distribution. Its coordinate value is equal to the sum of the horizontal, vertical, and vertical coordinate components of each point in the candidate point set, divided by the number of points. Using this local spatial centroid as the origin, an infinitely extending normal reference ray is constructed along the direction pointed to by the normal vector of the point cloud surface. This ray is strictly perpendicular to the currently fitted surface in space and passes through the centroid position, serving as the reference axis for subsequent projection calculations. Since the normal vector of the point cloud surface has been globally corrected in the previous step, the direction of the normal reference ray always points outward from the scanner, ensuring that the projection direction is consistent with the actual outer surface of the object and avoiding errors in the spacing sign due to normal reversal.

[0058] After establishing the normal reference ray, the system performs orthogonal projection transformation calculations on each point in the candidate point set. Specifically, the position vector of each candidate point relative to the local centroid is decomposed into two orthogonal components: one parallel to the normal reference ray and the other perpendicular to it. The endpoint of the parallel component is the foot of the perpendicular to the normal reference ray, and the three-dimensional position of this foot in the global coordinate system is defined as the normal projection coordinates. This projection process essentially compresses the three-dimensional point into a one-dimensional normal axis, retaining its depth information along the surface normal while filtering out lateral offsets in the tangent plane. Because the projection operation strictly follows orthogonal constraints, the line connecting the normal projection coordinates and the original candidate point must be perpendicular to the normal reference ray, thus ensuring that the extracted depth quantity purely reflects the surface undulations rather than tilt errors.

[0059] After obtaining the normal projection coordinates of each candidate point, the system further calculates the axial relative distance between it and the origin of the normal reference ray. This distance is obtained by calculating the length of the directed line segment of the normal projection coordinates and the local spatial centroid in the direction of the normal reference ray. Its value directly represents the degree to which the corresponding candidate point deviates from the local fitting reference plane. A positive value indicates that the point is located on the side pointed to by the normal vector, and a negative value indicates the opposite. In the scenario of a mobile shelving unit in an archive warehouse, the flatness of the shelf surface is usually better than 0.5 mm. Therefore, the normal projection spacing of points on a normal continuous surface should be concentrated in a small range near zero. If the absolute value of the normal projection spacing of a point exceeds two millimeters, it is highly suspected to be a mistaken entry point of an adjacent layer or a component behind it. This threshold setting not only accommodates the millimeter-level measurement noise inherent in laser scanning, but also effectively blocks cross-layer interference caused by dense arrangement, providing a precise quantitative basis for subsequent cross-layer step rejection.

[0060] In this embodiment, the above scheme constructs a local adaptive projection reference, transforming the depth deviation of discrete points into a comparable scalar spacing, which significantly improves the ability to resolve small curved surface steps in dense storage environments.

[0061] In a specific embodiment, the step of performing cross-layer step rejection on the same-direction candidate point set based on the normal projection spacing to obtain continuous surface nodes includes: Based on the normal projection spacing, the same-direction candidate point set is sorted in one dimension to obtain an ordered projection point column, and the adjacent node difference is calculated on the ordered projection point column to obtain the local spacing gradient. The peak extrema of the local spacing gradient are extracted to obtain the step breakpoint index, and the ordered projection point sequence is truncated and separated based on the step breakpoint index to obtain the continuous surface nodes.

[0062] Specifically, after obtaining the normal projection spacing of each point in the candidate point set, the system first performs a one-dimensional ascending sort on all candidate points based on this spacing value, forming an ordered projection point sequence. This sorting operation reorganizes the originally scattered neighborhood points in three-dimensional space into a linear sequence according to the degree to which they deviate from the local centroid along the normal reference ray. This ensures that geometrically adjacent points on the same continuous surface are close to each other in the sequence, while points belonging to different depth levels are naturally separated. After sorting, the system calculates the difference in normal projection spacing between each pair of adjacent points in the ordered projection point sequence, obtaining the local spacing gradient. This gradient value characterizes the rate of change of surface depth along the normal direction. When two adjacent points are located on the same smooth surface, their spacing difference is usually maintained below the millimeter level, caused only by scanning noise or small curvature. Once the sequence crosses to another physical level, such as jumping from the surface of a shelf in a mobile shelving unit to the bottom of an upper file box, the spacing difference will increase sharply, exhibiting a significant numerical jump.

[0063] Based on this abrupt change characteristic of local spacing gradients, the system performs peak extremum extraction. Specifically, it traverses the entire gradient sequence, identifying all local maxima points significantly higher than the background noise level. The sequence positions corresponding to these maxima points are marked as step breakpoint indices. In a real-world archive scenario, considering that the standard gap between shelves and file boxes is approximately five millimeters, while the point cloud accuracy of laser scanning is typically between 0.1 and 0.3 millimeters, the system sets the step detection threshold to one millimeter. This threshold is much higher than the measurement noise amplitude, ensuring that false breakpoints are not triggered by random errors, while being significantly smaller than the actual interlayer gap, leaving sufficient engineering margin to reliably capture all physical steps. Each step breakpoint index precisely indicates the location where the continuous surface in the ordered projected point sequence is interrupted.

[0064] After obtaining the step breakpoint index, the system performs sequence truncation and separation on the ordered projection point sequence accordingly. That is, using each step breakpoint index as a dividing boundary, the original ordered sequence is divided into several non-overlapping sub-sequence segments. Points within each segment satisfy the continuity condition that the local spacing gradient is below a preset threshold, thus being identified as continuous surface nodes belonging to the same physical surface. These continuous surface nodes not only retain the three-dimensional coordinates and normal vector information of the original points, but more importantly, their internal topological relationships have been rigorously verified through the aforementioned sorting and truncation process, completely eliminating the possibility of cross-layer misconnections. For example, when a scan point simultaneously covers the front edge of the second-layer shelf of a mobile shelving unit and the spine of the third-layer file box, although the angle between their normals may occasionally satisfy the same-direction screening condition, their normal projection spacing will inevitably differ by several millimeters. After sorting, a clear gradient peak will form in the middle of the sequence, which will then be truncated into two independent sets of continuous surface nodes, each corresponding to its own actual physical component.

[0065] In this embodiment, the above-mentioned scheme transforms the three-dimensional spatial step problem into one-dimensional sequence mutation detection, thereby achieving accurate layer-by-layer stripping of multi-layer stacked components in dense storage scenarios and providing unambiguous pure surface node inputs for subsequent triangulation reconstruction.

[0066] In a specific embodiment, the step of combining the connectivity of the initial spatial nodes based on the three-dimensional mesh patches to obtain the twin warehouse topology includes: Based on the three-dimensional mesh patch, the initial spatial node is subjected to vertical normal projection to obtain patch projection coordinates. Based on the patch projection coordinates, the initial spatial node is assigned to a surface to obtain a set of nodes on the same layer. For the three-dimensional mesh patch, surface mesh pathfinding is performed on the same layer node set to obtain the surface connectivity path; The twin warehouse topology is obtained by combining the same-layer node set through the surface connectivity path.

[0067] Specifically, after obtaining the 3D mesh patches and initial spatial nodes, the system first orthogonally projects each initial spatial node along its own normal vector direction onto the plane containing the nearest 3D mesh patch, obtaining the patch projection coordinates. This projection operation is strictly limited to the hemisphere pointed to by the node's normal vector. A projection is considered valid only when the angle between the node and the normal vector of a patch is less than 45 degrees and the vertical distance does not exceed 5 millimeters; otherwise, the node is determined not to belong to the current patch's coverage area. This constraint ensures that the projection relationship conforms to physical fit logic, avoiding the incorrect assignment of nodes on the sides of the mobile shelving column to adjacent shelf surfaces. Based on the patch projection coordinates, the system performs surface assignment partitioning on all initial spatial nodes: nodes successfully projected onto the same 3D mesh patch and whose projection points are within the patch's boundary are grouped into the same set of nodes on the same floor. All nodes within this set not only share the same geometric bearing surface but also, due to the previous cross-floor step rejection step eliminating depth anomalies, ensure that the nodes within the set physically belong to the same continuous component surface; for example, they are all located on the upper surface of the third shelf of the second row of mobile shelving.

[0068] For each set of nodes at the same level, the system performs surface mesh pathfinding on its corresponding 3D mesh facets. The pathfinding process uses the triangular mesh facets as the travel basis, taking the facet projection coordinates of the nodes as the path endpoints, and employing Dijkstra's algorithm to search for the shortest geodesic path on the graph formed by the edges of the triangular facets. Since the 3D mesh facets themselves have eliminated gaps and overlaps through a stitching operation and completely cover the surface of the target component, the generated surface connectivity path will strictly conform to the actual geometry of the object, without crossing air gaps or traversing non-connected regions. For example, when two nodes are located at opposite ends of the spine of the same file box, the pathfinding result will extend naturally along the curved surface of the spine, rather than passing through the interior of the box in a straight line. The length and direction of this path are entirely determined by the facet topology and are unaffected by the density of the original point cloud distribution, thus providing a stable and reliable connection basis.

[0069] After obtaining the surface connectivity paths, the system performs topological connection combinations on the set of nodes on the same layer. Specifically, the edges of the triangular facets traversed by each surface connectivity path are extracted sequentially, and the initial spatial nodes corresponding to the path's start and end points are established with directed connections through this edge sequence. All such connections are then aggregated to form a complete graph structure describing the internal surfaces of various components within the warehouse and the adjacency relationships between components—the twin warehouse topology. This topology not only preserves the spatial location attributes of the nodes but also encodes the continuity and accessibility information of the object's surface, enabling subsequent digital twin applications to accurately understand key semantics such as "which points belong to the same physical surface" and "how two points on the surface travel along the surface." For example, in a virtual inventory scenario, the system can use this topology to determine whether a file box is obscured by other boxes or calculate the optimal trajectory for a robotic arm to move along the shelf surface without repeatedly querying the original point cloud or reconstructing the geometric model. In this embodiment, the above scheme realizes a reliable transformation from an unordered set of points to a structured semantic topology by anchoring discrete nodes to continuous mesh patches and establishing connections along the patch topology, providing a low-level association framework for twin systems that combines geometric accuracy and physical consistency.

[0070] In a specific embodiment, the step of generating physical scheduling instructions and virtual update nodes based on the spatial displacement vector includes: The spatial displacement vector is decomposed into three-dimensional coordinates to obtain displacement direction components, and the spatial position of the displacement direction components is mapped based on the spatial coordinate system to obtain the physical target coordinates. The physical target coordinates are encapsulated into a scheduling message to obtain the physical scheduling instruction; Based on the physical target coordinates, the topology nodes of the twin warehouse are replaced to obtain the virtual update node.

[0071] Specifically, after acquiring the spatial displacement vector, the system first performs vector decomposition along the three orthogonal axes of a preset spatial coordinate system to obtain displacement direction components representing the lateral, longitudinal, and vertical motion. This decomposition process strictly follows the coordinate system definition, where the lateral component corresponds to the movement direction between the rows of the compact shelving, the longitudinal component corresponds to the shelf depth direction, and the vertical component corresponds to the shelf height direction. Together, these three components constitute a complete three-dimensional motion description. Subsequently, the system superimposes the values ​​of each component onto the initial spatial node coordinates of the currently operated object, completing the mapping from relative displacement to absolute position, and obtaining the physical target coordinates. This mapping result directly reflects the actual spatial position that the equipment actuator needs to reach. For example, when a file box needs to be moved from the second row, third layer to the fourth row, fifth layer, the physical target coordinates precisely point to the center point of the corresponding storage location on the fourth row, fifth layer. The values ​​are calculated from the global origin of the warehouse, in millimeters, and each axis component is rounded to match the mechanical positioning accuracy.

[0072] After obtaining the physical target coordinates, the system encapsulates them into a scheduling message according to a predefined communication protocol. The encapsulation includes the three-axis values ​​of the target coordinates, a unique device identifier, motion priority, and a checksum. All data is arranged in byte order and frame headers and trailers are added to form a physical scheduling instruction conforming to industrial bus specifications. This instruction is sent to the mobile shelving unit's main control unit via Ethernet or fieldbus, driving the servo motor to perform the handling action along the specified trajectory. To ensure instruction reliability, the system performs boundary checks on the physical target coordinates before encapsulation. If any axis component exceeds the warehouse's physical limits or falls into a restricted area, encapsulation is immediately stopped and an alarm is triggered to prevent equipment collisions or out-of-bounds operation due to calculation errors.

[0073] Meanwhile, to maintain real-time synchronization between the digital twin model and the physical entity, the system performs topology node replacement operations on the twin warehouse topology based on the same physical target coordinates. Specifically, it locates virtual objects associated with the original initial spatial nodes in the topology graph, updates their spatial attributes to the new positions indicated by the physical target coordinates, and recalculates their connections with neighboring nodes to generate virtual update nodes. These nodes not only inherit the semantic labels of the original objects (such as file number and category) but also occupy new adjacent positions in the topology, enabling subsequent twin applications such as path planning and occupancy analysis to reflect the latest layout status in real time. For example, when a file box completes a cross-row movement, the virtual update node automatically detaches from the connected subgraph of the original row and connects to the surface grid path of the new row, ensuring that the accessibility analysis in the digital space is completely consistent with the physical world. The entire replacement process is executed atomically, preventing intermediate data from being read by other modules and ensuring the consistency and timing correctness of the twin model.

[0074] In this embodiment, the above scheme drives physical execution and virtual updates simultaneously through a unified coordinate source, ensuring strict alignment of entity scheduling and digital mapping in the spatiotemporal dimension, and providing a real-time and reliable state benchmark for the twin system.

Claims

1. A real-time mapping digital twin archive management system, characterized in that, Includes the following steps: The scanning module is used to continuously scan the physical archive storage room using a laser scanning device to obtain spatial point cloud data; The acquisition module is used to acquire signals from the physical archives in the physical archives storage room based on the radio frequency identification device, and obtain the archive radio frequency signals; The calculation module is used to perform spatial positioning calculation on the radio frequency signal of the archive to obtain the estimated coordinates of the archive, and to perform coordinate registration on the estimated coordinates of the archive based on the spatial coordinate system constructed by the spatial point cloud data to obtain the initial spatial nodes; The reconstruction module is used to perform surface triangulation reconstruction based on the spatial point cloud data to obtain a three-dimensional mesh patch, and to perform connectivity combination on the initial spatial nodes based on the three-dimensional mesh patch to obtain the twin warehouse topology. The acquisition module is used to acquire the current spatial node in real time, calculate the relative displacement between the initial spatial node and the current spatial node in the twin warehouse topology to obtain the spatial displacement vector, and generate physical scheduling instructions and virtual update nodes based on the spatial displacement vector.

2. The real-time mapping digital twin archive management system according to claim 1, characterized in that, The process of continuously scanning the physical archive storage room using a laser scanning device to obtain spatial point cloud data includes: The ranging module is used to perform multi-view ranging on the physical surface of the physical archive storage room through the laser scanning device, obtain multi-view reflection points, and perform local coordinate mapping on the multi-view reflection points to obtain local three-dimensional coordinate points. The filtering module is used to perform spatial density filtering on the local three-dimensional coordinate points to obtain a skeleton feature point cloud, and to perform global coordinate stitching on the skeleton feature point cloud to obtain the spatial point cloud data.

3. The real-time mapping digital twin archive management system according to claim 1, characterized in that, The radio frequency identification (RFID) device is used to collect signals from the physical archives in the physical archive storage room to obtain the archive radio frequency signals, including: The radio frequency identification device emits radio frequency excitation waves to the physical archives in the physical archives warehouse to obtain tag reflection waves, and performs multipath time delay separation on the tag reflection waves to obtain direct path waveforms; The direct path waveform is envelope-detected and demodulated to obtain the baseband payload code stream, and the carrier phase is extracted from the direct path waveform based on the baseband payload code stream to obtain the archive radio frequency signal.

4. The real-time mapping digital twin archive management system according to claim 3, characterized in that, The step of transmitting radio frequency excitation waves to the physical archives in the physical archive storage room through the radio frequency identification device to obtain tag reflected waves includes: The radio frequency identification device transmits a single-frequency continuous wave to the physical archives in the physical archives repository to obtain a backscattered radio frequency stream, and then amplifies the backscattered radio frequency stream with low noise to obtain a primary radio frequency signal. The primary radio frequency signal is subjected to quadrature coherent demodulation to obtain a complex baseband waveform, and the complex baseband waveform is subjected to frame synchronization detection and signal amplitude normalization to obtain the tag reflection wave.

5. The real-time mapping digital twin archive management system according to claim 1, characterized in that, The step of performing spatial positioning calculations on the radio frequency signals of the archives to obtain the estimated coordinates of the archives includes: The phase difference of the radio frequency signal of the archive is calculated to obtain the signal phase difference value, and the angle of arrival is calculated based on the signal phase difference value to obtain the spatial incident angle. Obtain the array antenna coordinates of the radio frequency identification device, and perform ray projection on the array antenna coordinates based on the spatial incident angle to obtain a spatial intersecting ray; The intersection vertices of the spatial intersecting rays are extracted to obtain the estimated coordinates of the archive.

6. The real-time mapping digital twin archive management system according to claim 1, characterized in that, The surface triangulation reconstruction based on the spatial point cloud data to obtain a three-dimensional mesh patch includes: The spatial point cloud data is fitted with a local surface to obtain the point cloud surface normal vector, and a spatial neighborhood search is performed on the spatial point cloud data based on the point cloud surface normal vector to obtain a local neighboring point set. The local adjacent point set is spatially connected and projected to obtain a planar triangular profile, and the surface boundary is stitched based on the planar triangular profile to obtain the three-dimensional mesh patch.

7. The real-time mapping digital twin archive management system according to claim 6, characterized in that, The step of performing a spatial neighborhood search on the spatial point cloud data based on the point cloud surface normal vector to obtain a local neighboring point set includes: A spatial sphere domain search is performed on the spatial point cloud data to obtain initial neighborhood scattered points, and the normal angle of the initial neighborhood scattered points is extracted based on the surface normal vector of the point cloud to obtain the surface normal angle. Based on the surface normal angle, the initial neighborhood scattered points are filtered for orientation differences to obtain a set of candidate points in the same direction. Then, based on the surface normal vector of the point cloud, the candidate point set in the same direction is projected onto the normal vector to obtain the normal projection spacing. Based on the normal projection spacing, the same-direction candidate point set is subjected to cross-layer step rejection to obtain continuous surface nodes, and the continuous surface nodes are combined spatially to obtain the local adjacent point set.

8. The real-time mapping digital twin archive management system according to claim 7, characterized in that, The step of calculating the normal projection distance of the candidate point set in the same direction based on the surface normal vector of the point cloud includes: The mean of spatial coordinates is extracted from the candidate point set in the same direction to obtain the local spatial centroid, and a normal reference ray is constructed based on the local spatial centroid and the normal vector of the point cloud surface. Based on the normal reference ray, an orthogonal projection transformation is performed on the same-direction candidate point set to obtain the normal projection coordinates; The axial relative distance between the normal projection coordinates and the origin coordinates of the normal reference ray is calculated to obtain the normal projection spacing.

9. The real-time mapping digital twin archive management system according to claim 7, characterized in that, The step rejection of the same-direction candidate point set based on the normal projection spacing to obtain continuous surface nodes includes: Based on the normal projection spacing, the same-direction candidate point set is sorted in one dimension to obtain an ordered projection point column, and the adjacent node difference is calculated on the ordered projection point column to obtain the local spacing gradient. The peak extrema of the local spacing gradient are extracted to obtain the step breakpoint index, and the ordered projection point sequence is truncated and separated based on the step breakpoint index to obtain the continuous surface nodes.

Images