A warehouse weighing, scanning code integrated automatic in-out warehouse registration method and system

Abstract

The application relates to the technical field of intelligent warehousing, and discloses a warehouse weighing and code scanning integrated automatic warehouse-in and warehouse-out registration method and system, which comprises the following steps: acquiring a multi-modal trigger event signal and constructing an observation window, reconstructing and calculating a regional quality gradient matrix and a spatial offset vector; inputting the spatial offset vector into the regional quality gradient matrix to obtain a topological offset diagnosis mark through comparison, extracting a coupled offset feature to execute an interrupt suspension in the skinning zero clearing process, and generating a basic zero point baseline; based on the basic zero point baseline, converting an actual material change amount and fusing warehouse-in and warehouse-out registration data, and encrypting a high-trust traceability block to solve a global inventory total. The application realizes non-inductive automatic registration and bottom-layer account tamper-proofing and tracing, and fundamentally eliminates global baseline drift and cumulative error through hardware adaptive calibration without stopping, thereby improving the accuracy of warehouse data.

Description

Technical Field

[0001] This invention relates to the field of intelligent warehousing technology, specifically to an automatic inbound and outbound registration method and system that integrates weighing and barcode scanning in warehousing. Background Technology

[0002] With the development of IoT and edge computing technologies, smart warehousing systems often rely on multi-source sensor fusion technology to achieve seamless registration of material entry and exit. The system typically deploys weighing sensors and barcode scanning or visual sensing components on each smart storage unit of the shelving support to acquire the physical weight data and identification information of the materials. When the target object accesses or retrieves materials, the edge control node monitors the aforementioned multi-dimensional data and then synthesizes the entry and exit registration data in the warehousing database.

[0003] However, existing systems typically process weight fluctuation time-series data based on a low-level tare and zeroing process. During long-term automatic registration, if the target object performs a trial return of materials or is misplaced across physical boundaries, the current target storage location node is prone to quality flow misjudgments due to center-of-gravity deviation. Adjacent storage location nodes also receive mechanical stress crosstalk transmitted through rack supports. This causes the global inventory total calculated by the warehouse database to be mixed with zero-point baseline drift and accumulated errors, thus affecting the accuracy of data processing. Therefore, how to reduce mechanical stress crosstalk and accumulated calculation errors between nodes during long-term inbound and outbound data processing in intelligent warehousing systems is a problem that urgently needs to be solved in this field. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide an automated inbound and outbound registration method integrating warehouse weighing and barcode scanning. This method achieves precise isolation of mechanical stress crosstalk caused by trial-and-error placement or misplacement by constructing a deep synergistic coupling of the observation window, spatial offset vector, and regional quality gradient matrix. Furthermore, it utilizes coupled offset features to interrupt and suspend the tare-clearing process, generating a baseline zero point. Combined with a high-reliability traceability block at the underlying level, it encrypts inbound and outbound registration data, achieving both seamless automated registration and tamper-proof traceability. Moreover, it enables downtime-free calibration and fatigue-resistant self-healing of the underlying hardware without affecting normal business operations, fundamentally eliminating global baseline drift and accumulated errors, and providing a high-confidence global inventory total for warehouse operations.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] In a first aspect, the present invention provides an automatic inbound and outbound registration method integrating warehouse weighing and barcode scanning, comprising:

[0007] Acquire multimodal trigger event signals and construct an observation window to collect weight fluctuation time series data. Perform spatial feature reconstruction calculation on the weight fluctuation time series data to obtain the regional mass gradient matrix. Extract continuous image frames based on multimodal trigger event signals. Perform feature displacement extraction and spatial projection difference calculation on the continuous image frames to obtain the spatial offset vector.

[0008] The spatial offset vector is input into the regional quality gradient matrix to perform spatial orientation feature addressing and coupling comparison operations to obtain the topology offset diagnostic identifier. The coupling offset feature is extracted from the topology offset diagnostic identifier to interrupt and suspend the tare zeroing process and generate the basic zero-point baseline.

[0009] Obtain the steady-state sampled digital value and perform algebraic difference conversion with the basic zero-point baseline to obtain the actual material change. Perform data fusion based on the actual material change to obtain the inbound and outbound registration data. Perform encryption processing calculation on the inbound and outbound registration data to obtain a high-reliability traceability block. Write the high-reliability traceability block into the traceability block storage chain for solution and output the global inventory summation.

[0010] Furthermore, the method for constructing the observation window includes:

[0011] Extract identity binding information and operation location coordinates from multimodal trigger event signals;

[0012] Obtain the warehouse database, perform spatial absolute addressing in the warehouse database based on the operation location coordinates to determine the target storage location node, and retrieve adjacent storage location nodes based on the network association physical pointers pre-registered for the target storage location node;

[0013] Extract the start timestamp from the multimodal trigger event signal, obtain the action duration, and add it to the start timestamp to get the end timestamp;

[0014] Extract the bounding box boundary parameters of the target storage location node and adjacent storage location nodes respectively, and use the bounding box boundary parameters to construct the global coordinate boundary;

[0015] A multidimensional data buffer is constructed, and the original sampled data is extracted from the multimodal trigger event signal. The original sampled data that falls within the global coordinate boundary and is in the interval from the start time stamp to the end time stamp is filled into the multidimensional data buffer to generate the observation window.

[0016] Furthermore, the method for acquiring the weight fluctuation time-series data includes:

[0017] Activate the underlying analog-to-digital conversion channels of the target cargo location node and all adjacent cargo location nodes located within the observation window, and collect analog voltage signals at preset equal time intervals to obtain sampled digital values.

[0018] The sampled digital values ​​continuously output from the start timestamp to the end timestamp are concatenated in chronological order to generate time-series data of weight fluctuations for the target cargo location node and each adjacent cargo location node, and are simultaneously stored in the multidimensional data cache.

[0019] Furthermore, the method for calculating the regional quality gradient matrix includes:

[0020] Extract the weight fluctuation time series data of the target cargo location node, calculate the algebraic difference between the sampled digital values ​​corresponding to the end timestamp and the start timestamp to obtain the central mass fluctuation amount, and simultaneously extract the weight fluctuation time series data of each adjacent cargo location node to calculate the algebraic difference to obtain the neighborhood stress fluctuation amount.

[0021] The stress transmission weight of adjacent cargo location nodes is calculated by obtaining the spatial damping attenuation coefficient and the three-dimensional geometric straight line distance. The stress transmission weight is then multiplied by the neighborhood stress fluctuation to obtain the normalized neighborhood stress value.

[0022] Perform matrix orthogonalization numerical filling operations on the central mass fluctuation and the normalized neighborhood stress values ​​to obtain the regional mass gradient matrix.

[0023] Furthermore, the method for calculating the spatial offset vector includes:

[0024] Optical features are extracted from multimodal trigger event signals, and optical features are continuously captured within the time interval from the start timestamp to the end timestamp to generate continuous image frames.

[0025] Calculate the two-dimensional displacement vector between adjacent image frames and integrate it along the time axis to obtain the continuous pixel movement trajectory. Obtain the depth mapping parameters and back-project the continuous pixel movement trajectory onto the three-dimensional coordinate system to generate the material movement trajectory.

[0026] Obtain the system's silent confirmation signal and extract the three-dimensional geometric coordinates of the material's movement trajectory, defining them as the actual landing point coordinates of the material.

[0027] Obtain the geometric center coordinates of the target storage location node, subtract the geometric center coordinates from the actual landing point coordinates of the material to obtain the lateral physical coordinate deviation, longitudinal physical coordinate deviation, and height physical coordinate deviation, and encapsulate them into a spatial offset vector.

[0028] Furthermore, the method for outputting the topology offset diagnostic identifier includes:

[0029] The spatial offset vector is used to perform directional cosine decomposition to obtain the spatial pointing angle. The spatial offset vector is then used to calculate the offset distance. The spatial pointing angle is mapped to a two-dimensional polar coordinate system with the center point index coordinate position of the preset region quality gradient matrix as the origin. The relative offset index coordinate position within the quadrant in which the spatial pointing angle points is selected.

[0030] The normalized neighborhood stress value is extracted from the relative offset index coordinates as the first comparison item, and the central mass fluctuation is extracted from the center point index coordinates as the second comparison item. The first comparison item and the second comparison item are algebraically added to obtain the stress closed-loop residual.

[0031] When the absolute value of the stress closed-loop residual is less than the preset bottom background noise threshold and the offset distance is greater than zero, the topology offset diagnosis flag of the logic true value status label is output; otherwise, the topology offset diagnosis flag of the logic false value status label is output.

[0032] Furthermore, the method for extracting the coupling offset features includes:

[0033] When the topology offset diagnostic flag is a logical truth status label, the abnormal interception logic is triggered, the absolute value of the center mass fluctuation is defined as the target weight reduction, and the absolute value of the first comparison item is defined as the neighborhood weight increase.

[0034] Obtain the spatial deviation multiplier factor and mechanical structure compensation coefficient, construct the stress decoupling transformation equation using the offset distance, perform algebraic mapping calculation on the target weight reduction and the neighborhood weight increase to obtain the compensation characteristic parameters;

[0035] The target weight reduction, neighborhood weight increase, and compensation feature parameters are encapsulated in a structured array to generate coupled offset features.

[0036] Furthermore, the method for obtaining the basic zero-point baseline includes:

[0037] Generate kernel-level interrupt control signaling and send it to the target cargo location node and the adjacent cargo location nodes that provide the spatial orientation of the first comparison item in the regional quality gradient matrix, forcing the tare and zeroing process to be interrupted and suspended.

[0038] After confirming that the mechanical system is at rest, the sampled digital values ​​output by the target storage location node and the affected adjacent storage location nodes under steady state are read and defined as the static weight register values.

[0039] The compensation feature parameter is extracted from the coupling offset feature. The static weight register value of the target storage location node is added to the compensation feature parameter, and the static weight register value of the affected adjacent storage location node is subtracted from the compensation feature parameter. Algebraic cancellation calculation is then performed to obtain the updated value.

[0040] The updated values ​​will be burned back to the baseline configuration memory sector of the corresponding microcontroller unit and uniformly defined as the baseline zero point.

[0041] Furthermore, the method for calculating the actual material change includes:

[0042] The system acquires a silent confirmation signal and reads the single valid sampled data output by the target cargo location node in a state of absolute mechanical stillness through the underlying communication bus, which is defined as the steady-state sampled digital value.

[0043] Subtract the baseline zero-point value from the steady-state sampled digital value and perform an algebraic difference conversion operation to obtain the net weight fluctuation value;

[0044] Obtain the standard weight and physical tolerance compensation factor of the single item, construct a discrete quantity conversion equation using the net weight fluctuation value and the standard weight of the single item, and perform a floor function mapping calculation with the physical tolerance compensation factor to obtain the actual material change with positive and negative mathematical signs.

[0045] Furthermore, the method for generating the inbound / outbound registration data and the high-reliability traceability block includes:

[0046] Extract the physical identification code of the target storage location node from the warehouse database, and combine it with the identity binding information and the actual material change to generate inbound and outbound registration data;

[0047] Obtain the global physical timestamp, extract the power supply voltage characteristics and signal-to-noise ratio characteristics of the weighing sensor, perform bit concatenation to generate the underlying sensor signal-to-noise ratio characteristics, and perform bitwise XOR logic operation on the binary sequence of the global physical timestamp and the underlying sensor signal-to-noise ratio characteristics to obtain the dynamic salting key;

[0048] The data entry and exit registration data is converted into a one-dimensional basic byte stream, and then subjected to cross-obfuscation and cyclic left shift mathematical operations with a dynamically salted key to obtain a scrambled data stream.

[0049] The standard hash algorithm is called to perform hash calculation on the scrambled data stream to obtain the tamper-proof status hash value. The tamper-proof status hash value is logically encapsulated with the inbound and outbound registration data to obtain a highly reliable traceability block.

[0050] Furthermore, the method for outputting the global inventory summation includes:

[0051] Extract the traceability block storage chain from the warehouse database, obtain the most recent previous historical traceability block and the memory logical pointer used to indicate the data association relationship;

[0052] Extract the tamper-proof state hash value from the most recent preceding historical trace block, and obtain the characteristic bits of the tamper-proof state hash value;

[0053] Write the high-reliability traceability block to the end of the storage queue of the traceability block storage chain, and use a memory logical pointer to link it unidirectionally to the feature bit;

[0054] Extract the historical inventory data cached in the warehouse database before the start timestamp of the target storage location node, and perform a signed algebraic cumulative iterative calculation operation on the historical inventory data and the actual material change to obtain the latest inventory quantity at a single point.

[0055] Extract all registered smart storage units from the warehouse database, perform global traversal and arithmetic addition on the latest inventory of each smart storage unit, and output the total global inventory.

[0056] Secondly, the present invention provides an automated warehouse entry and exit registration system integrating weighing and barcode scanning, which is used to implement the aforementioned automated warehouse entry and exit registration method integrating weighing and barcode scanning, the system comprising:

[0057] Feature reconstruction module: used to acquire multimodal trigger event signals, construct observation windows to collect weight fluctuation time series data, perform spatial feature reconstruction calculation on weight fluctuation time series data to obtain regional mass gradient matrix, extract continuous image frames based on multimodal trigger event signals, and perform feature displacement extraction and spatial projection difference calculation on continuous image frames to obtain spatial offset vector;

[0058] Offset calibration module: It is used to input the spatial offset vector into the regional quality gradient matrix to perform spatial orientation feature addressing and coupling comparison operations, obtain the topology offset diagnostic identifier, extract the coupling offset feature from the topology offset diagnostic identifier, perform an interruption and suspension operation on the tare zeroing process, and generate the basic zero-point baseline.

[0059] Trusted Traceability Module: This module is used to acquire steady-state sampled digital values ​​and perform algebraic difference conversion with the baseline zero point to obtain the actual material change. Based on the actual material change, it performs data fusion to obtain inbound and outbound registration data. It performs encryption processing on the inbound and outbound registration data to obtain a high-trust traceability block. The high-trust traceability block is written into the traceability block storage chain for calculation and outputs the global inventory total.

[0060] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0061] This invention achieves precise isolation of spatiotemporal interference and irreversible underlying data tracing by constructing an observation window and a high-reliability traceability block. It solves the technical defects of traditional independent tare algorithms that cause global zero-point baseline drift when faced with invalid touches, random interference touches, or erroneous actions. The synergy between the regional quality gradient matrix and the spatial offset vector transforms the traditional passive filtering of weight anomalies into a multi-dimensional reconstruction and extraction of local weight redistribution and coupled offset features. It accurately quantifies the degree of quality loss and stress compensation direction caused by exploratory actions or cross-boundary misplacement, and realizes hardware-level adaptive repair. The actual material change quantity combined with the anti-tampering status hash value and the signal-to-noise ratio characteristics of the underlying sensors realizes the logical encapsulation of the core payload. This not only eliminates application-layer replay attacks and forgery and tampering of historical data from the source, but also provides the application layer with objective and true business indicators, greatly improving the accuracy and data security of the global inventory summation. Attached Figure Description

[0062] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0063] Figure 1 A flowchart of an automatic warehouse entry and exit registration method integrating weighing and barcode scanning is provided in an embodiment of the present invention.

[0064] Figure 2 A front view diagram of the distributed node physical deployment structure of the underlying hardware of the smart warehouse provided in this embodiment of the invention;

[0065] Figure 3 This is a schematic diagram illustrating the top-down relationship between the global coordinate boundary and the node topology provided in an embodiment of the present invention.

[0066] Figure 4 A schematic diagram of two-dimensional continuous pixel trajectory tracking based on multimodal optical features provided in an embodiment of the present invention;

[0067] Figure 5 This is a schematic diagram illustrating the feature extraction of spatial offset and force crosstalk caused by misplaced materials across boundaries, provided in an embodiment of the present invention.

[0068] Figure 6 This is a functional module diagram of an automatic warehouse entry and exit registration system that integrates weighing and barcode scanning, provided as an embodiment of the present invention. Detailed Implementation

[0069] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0070] Example 1

[0071] Please see Figure 1 As shown, this embodiment provides an automatic inbound and outbound registration method integrating warehouse weighing and barcode scanning, including:

[0072] Step S10: Acquire multimodal trigger event signals, construct an observation window to collect weight fluctuation time series data, perform spatial feature reconstruction calculation on the weight fluctuation time series data to obtain the regional mass gradient matrix, extract continuous image frames based on multimodal trigger event signals, and perform feature displacement extraction and spatial projection difference calculation on the continuous image frames to obtain the spatial offset vector.

[0073] Further, step S10 includes:

[0074] Step S11: Obtain multimodal trigger event signals and construct an observation window based on the multimodal trigger event signals.

[0075] In a seamless warehouse entry and exit registration scenario integrating weighing and barcode scanning, when a target object performs material retrieval or tentative placement actions in front of the shelf, not only do visual characteristics change, but complex stress fluctuations in the underlying metal mesh frame are also triggered. Traditional independent weighing environments rely solely on passive weight threshold triggers, lacking multi-dimensional physical space perception and temporal isolation mechanisms. This makes it highly susceptible to misinterpreting unmotivated touches, vibration interference from adjacent locations, or misplacement across boundaries as genuine material outbound actions. To filter out such spatiotemporally irrelevant interference signals at the source and establish a high-fidelity dynamic event tracking benchmark, a logical monitoring container is constructed before extracting the underlying weight data. This container tightly anchors multimodal visual features, personnel identity, and the three-dimensional spatial coordinates of the physical shelf. The aim is to transform discrete physical retrieval actions into a digitally isolated space with strict time start and end boundaries and an absolute three-dimensional spatial addressing range.

[0076] Specifically, multimodal trigger event signals are acquired. These multimodal trigger event signals are continuous digital electrical signal streams containing optical and barcode parsing features generated in the underlying data link when scanning components or visual sensing components deployed in front of the material storage shelves detect personnel approaching, barcode exposure, or physical actions related to material storage or retrieval. Based on these multimodal trigger event signals, the target object's identity binding information and operation position coordinates are parsed and extracted in the processing core of the edge control node. The edge control node refers to an embedded computing hardware unit physically deployed on the local side of the warehouse rack, possessing real-time data acquisition and on-site algorithm computation capabilities. It maintains a continuous data link connection via wired or wireless communication bus with the strain gauge weighing sensor array at the bottom of the rack, the barcode scanning component at the front of the rack, and the visual perception component. It is responsible for independently completing the entire real-time computing task, from multimodal signal acquisition to the generation of inbound / outbound registration data, without relying on a remote cloud server. The identity binding information refers to a digital identity tag or mapping code for personnel with unique corresponding permissions, determined after hash feature comparison. The operation position coordinates refer to the three-dimensional absolute spatial physical coordinates of the material picking or tentative placement action captured by the visual perception component. These operation position coordinates include lateral position components, longitudinal position components, and vertical height position components, which together constitute a complete three-dimensional geometric positioning description of the target object's hand movements in real physical space. See also... Figure 2 This is a front view schematic diagram of the distributed node physical deployment of the underlying hardware of the intelligent warehouse provided in this embodiment of the invention. The diagram exemplarily shows the rack support, indicated by a thick black solid line outline and its internal vertical dividing lines, forming the basic physical framework of the rack. This divides the space into multiple intelligent storage unit units filled with light gray, forming independent material-carrying cavities. Simultaneously, the diagram exemplarily shows a weighing sensor, indicated by a dark gray flat rectangle, positioned flat at the bottom center of each intelligent storage unit to maintain consistent measurement standards.

[0077] The warehouse database is obtained. The warehouse database refers to a digital relational data table pre-installed in the non-volatile memory of the edge control node, used to map the one-to-one correspondence between the three-dimensional geometric network of the physical shelf space structure and the physical addresses of the underlying sensor hardware communication. Each storage location unit in the warehouse database is pre-registered with geometric center coordinates and a network-associated physical pointer. The geometric center coordinates are static physical position vectors in three-dimensional absolute space, uniquely representing the geometric centroid of the intelligent storage location unit, calibrated by three-dimensional laser scanning during system initialization and referenced to a fixed origin in the warehouse space. The network-associated physical pointer is a structured index field pre-configured in the warehouse database for each storage location node, recording the database primary key set of adjacent storage location units sharing the same shelf support. This is used to directly retrieve physically adjacent storage location nodes during the addressing phase without traversing the entire warehouse. Each intelligent storage location unit is a basic physical grid entity constituting the shelf storage array. It has a fixed weighing sensor at its bottom and is surrounded by metal partitions to form a material-bearing cavity with defined three-dimensional spatial boundaries. Each intelligent storage location unit has a unique physical identifier code in the warehouse database. Based on the operation position coordinates, spatial absolute addressing is performed in the warehouse database to determine the target storage location node that is directly touched. Specifically, the process of performing spatial absolute addressing is as follows: extracting the lateral position component, longitudinal position component, and vertical height position component from the operation position coordinates; traversing all pre-registered geometric center coordinates in the warehouse database, and calculating the three-dimensional geometric straight-line distance between the operation position coordinates and each geometric center coordinate using the three-dimensional Euclidean distance formula; extracting the intelligent storage unit corresponding to the geometric center coordinate with the smallest three-dimensional geometric straight-line distance value, and defining it as the target storage location node. The target storage location node refers to a specific intelligent storage location unit with a fixedly deployed weighing sensor at the bottom layer and whose three-dimensional geometric straight-line distance between its geometric center coordinates and the operation position coordinates is the minimum value among all storage location nodes, i.e., the core physical grid area identified by the visual perception component as the core physical grid area where the target object's hand intrusion interaction occurs; wherein, the core physical grid area refers to the three-dimensional spatial determination domain used to determine whether the physical contour of the target object's hand has made effective contact with the storage location unit, with the geometric center coordinates of the target storage location node as the reference center and the three-dimensional spatial bounding box boundary of the corresponding intelligent storage location unit as the constraint range.

[0078] Simultaneously, based on the network association physical pointers pre-registered in the warehousing database for the target storage location node, intelligent storage location units adjacent to the target storage location node on the left, right, directly above, and directly below the same shelf support on the same floor are retrieved and extracted, defined as adjacent storage location nodes. The start timestamp is extracted from the multimodal trigger event signal. The start timestamp refers to the physical time recorded by the underlying clock chip in the multimodal trigger event signal when the physical outline of the target object's hand first crosses the vertical projection plane of the shelf's front edge. The action duration is obtained. The action duration is a pre-set, longest physiological and physical time constant for a human to complete a standard material retrieval and completely remove the physical outline of the hand from the shelf space. The purpose of the action duration is to provide a clear time convergence boundary for the underlying monitoring range of the action event, preventing the accumulation of computing power and memory overflow of the edge control node due to indefinitely listening to meaningless environmental fluctuations. The start timestamp is superimposed with the action duration, and algebraic addition is performed to obtain the end timestamp. The time interval between the start timestamp and the end timestamp is matrix-encapsulated with the physical three-dimensional spatial coordinate range of the target storage location node and adjacent storage location nodes to obtain the observation window. Specifically, the bounding box boundary parameters of the target storage location node and each adjacent storage location node pre-registered in the warehousing database are extracted. These bounding box boundary parameters refer to the extreme coordinates of the six faces of the smallest orthogonal cuboid that completely encloses the entire physical volume of the corresponding intelligent storage location unit, written into the database after 3D laser scanning calibration during system initialization. These parameters are described on the horizontal, vertical, and height axes, including the maximum and minimum boundary values ​​along the horizontal, vertical, and height axes. The extracted bounding box boundary parameters are used to independently calculate the maximum and minimum coordinate values ​​of all nodes along the three dimensions of the horizontal axis, horizontal axis, and vertical axis. The maximum coordinate value is the maximum value among the maximum boundary values ​​of the bounding box boundary parameters of all nodes along the corresponding coordinate axis, and the minimum coordinate value is the minimum value among the minimum boundary values ​​of the bounding box boundary parameters of all nodes along the corresponding coordinate axis. The maximum and minimum coordinate values ​​of each of the three coordinate axes are combined in a union operation to form a global coordinate boundary covering the entire area affected by physical stress of the target storage location node and all adjacent storage location nodes.

[0079] A multidimensional data buffer is created in the random access memory of the edge control node. This multidimensional data buffer is a structured data buffer allocated in the random access memory in the form of a multi-axis index array. It has three spatial dimension index axes and one time dimension index axis, forming a four-dimensional index structure. The three spatial dimension index axes correspond to the horizontal axis, horizontal axis, and vertical height axis in the physical shelf coordinate system, respectively. The physical boundaries of their index value ranges are determined by a direct linear mapping of the global coordinate boundaries along the corresponding coordinate axes. The mapping relationship between the global coordinate boundaries and the physical limits of the spatial dimension indexes in the multidimensional data buffer is as follows: the maximum coordinate value of the corresponding axis in the global coordinate boundary is mapped to the maximum physical limit value of the spatial dimension index axis, and the minimum coordinate value is mapped to the minimum physical limit value of the spatial dimension index axis. A corresponding physical spatial position is allocated for each index bit between the two physical limits through linear interpolation. The start and end indexes of the time dimension index axis correspond to the start and end timestamps, respectively, and the time axis index step size is determined by the underlying sampling clock cycle. The raw sampled data of all multimodal triggered event signals within the time interval from the start timestamp to the end timestamp, where the coordinates fall entirely within the global coordinate boundary, are extracted and structured. The extracted raw sampled data is then integrated and filled into the corresponding index positions of the multidimensional data buffer according to its timestamp and three-dimensional spatial coordinates, ultimately completing the splicing to generate an observation window. This observation window is a logically isolated monitoring container, strictly defined in both the time and physical space dimensions, specifically designed to capture the transient physical response characteristics of a single material access. It fundamentally eliminates external time-independent fluctuations and space-independent interference. See also... Figure 3 This is a schematic diagram illustrating the top-down relationship between the global coordinate boundary and the node topology provided in an embodiment of the present invention. Figure 3 As shown in the figure, the dark gray rectangles exemplarily mark the target storage location nodes where physical contact interactions occur. The four light gray rectangles adjacent to it on the left, right, directly above, and directly below represent adjacent storage location nodes affected by physical stress transmission. The four unfilled blank rectangles on the diagonal represent other non-adjacent storage location nodes. The thin solid-line box surrounding each storage location unit exemplarily reflects the pre-registered bounding box boundary parameters. By combining the bounding box boundary parameters of the nodes within the central cross region, the system exemplarily generates the outermost thick black dashed box, i.e., the global coordinate boundary. This boundary exemplarily covers the central node and its associated cross region, thus defining a tight data capture range for the observation window in physical space.

[0080] Step S12: Based on the observation window, activate the underlying analog-to-digital conversion channel to generate weight fluctuation time series data, and call the stress extraction algorithm to perform spatial feature reconstruction calculation on the weight fluctuation time series data to obtain the regional quality gradient matrix.

[0081] After obtaining the observation window, when the target object performs a trial return of materials or experiences physical spatial misplacement, the physical center of gravity of the returned materials is easily deviated from the geometric center coordinates of the intelligent storage unit, or even directly rests against the metal partition of the adjacent storage node. This cross-boundary offset will generate mechanical stress crosstalk through the shelf supports, causing traditional independent tare zeroing algorithms based on a single storage location to misjudge it as real weight loss, thus triggering global zero-point baseline drift. In order to accurately isolate the mechanical deformation crosstalk caused by the trial return action, the approach is to transform from simple weight threshold comparison to constructing a global mass conservation topology with physical stress coupling.

[0082] Specifically, the underlying analog-to-digital conversion (ADC) channels of the target storage location node and all adjacent storage location nodes within the physical three-dimensional spatial coordinate range of the observation window are activated. The underlying ADC channel refers to the hardware data reading interface opened by the analog-to-digital converter integrated in the weighing sensor control circuit at the bottom of each intelligent storage location unit. After receiving the activation command from the edge control node, it begins to collect the analog voltage signal at each sampling moment at preset equal time intervals, performing continuous synchronous sampling and quantization operations on the analog voltage signal to obtain the sampled digital value. The equal time interval is set based on the Nyquist sampling theorem and the upper limit extreme value of the inherent mechanical vibration frequency of the warehouse metal mesh structure, ensuring that the sampling frequency can completely capture the envelope characteristics of transient stress fluctuations without spectral aliasing.

[0083] Within the observation window's start and end timestamp interval, the continuously output sampled digital values ​​of the target cargo location node and all adjacent cargo location nodes are concatenated into a one-dimensional discrete array according to the chronological order of their corresponding sampling times. This generates time-series data on weight fluctuations for the target cargo location node and each of its adjacent cargo location nodes. This weight fluctuation time-series data is then written in real-time to the multi-dimensional data buffer in the edge control node's random access memory. Specifically, when the weight fluctuation time-series data is stored in the multi-dimensional data cache, its spatial dimension position is determined by mapping the coordinate information of the target storage location node or adjacent storage location node to which the weight fluctuation time-series data belongs onto three spatial dimension index axes, thus obtaining the corresponding spatial dimension index axis coordinates. Its time dimension position is determined by mapping the sampling time corresponding to each sampled digital value onto the time dimension index axis, thus obtaining the corresponding time dimension index axis coordinates. This ensures that the weight fluctuation time-series data of the target storage location node and each adjacent storage location node are strictly aligned in the multi-dimensional data cache, providing a physical consistency basis for performing cross-node synchronous comparison operations according to the time dimension index axis. In order to accurately remove the mechanical deformation crosstalk caused by the tentative cross-boundary return action, the stress extraction algorithm is called to perform spatial feature reconstruction calculation on the weight fluctuation time-series data in the multi-dimensional data cache to obtain the regional quality gradient matrix. Specifically, the execution flow of the stress extraction algorithm is as follows: Extract the weight fluctuation time-series data of the target storage location node within the observation window, calculate the algebraic difference between the sampled digital value corresponding to the end timestamp and the sampled digital value corresponding to the start timestamp, and define it as the central mass fluctuation quantity. The central mass fluctuation quantity is a scalar value characterizing the degree of increase or decrease in the underlying physical gravity of the target storage location node before and after a single material storage / retrieval operation. A negative sign indicates that the material has left the target storage location node, and a positive sign indicates that the material has been placed into the target storage location node. Simultaneously extract the weight fluctuation time-series data of each adjacent storage location node, and calculate the algebraic difference of each adjacent storage location node using the same time-series subtraction rule, i.e., subtracting the sampled digital value corresponding to the start timestamp from the sampled digital value corresponding to the end timestamp. This difference is defined as the neighborhood stress fluctuation quantity. The neighborhood stress fluctuation quantity is a scalar value characterizing the degree of change in the underlying gravity of the corresponding adjacent storage location node caused by the mechanical stress transmission received through the metal support frame due to the material storage / retrieval operation at the target storage location node.

[0084] Let N represent the total number of adjacent storage location nodes within the observation window, and define node indices j for adjacent storage location nodes, ranging from 1 to N; solve for the stress transmission weight W of each adjacent storage location node, using the stress transmission weight of the j-th adjacent storage location node as an example. For example, the stress transmission weight of the j-th adjacent storage location node The calculation formula is: ,in, denoted as an exponential function with the natural constant as its base; D represents the three-dimensional geometric straight-line distance. This represents the three-dimensional geometric straight-line distance between the geometric center coordinates of the j-th adjacent storage location node and the geometric center coordinates of the target storage location node. By extracting the pre-registered geometric center coordinates of the j-th adjacent storage location node and the pre-registered geometric center coordinates of the target storage location node from the warehousing database, and performing algebraic solutions on the coordinate values ​​of the two using the three-dimensional Euclidean distance formula, the three-dimensional geometric straight-line distance is obtained. The spatial damping attenuation coefficient is a dimensionless scalar used to characterize the inherent energy dissipation characteristics of the rack support for mechanical stress transmission. Its setting is based on: applying a standard step impact torque to an unloaded intelligent storage unit, measuring the transient physical attenuation rate of this standard step impact torque as it is transmitted to neighboring physical nodes in the metal support frame, and obtaining the value through fitting and calibration based on this transient physical attenuation rate. For example, it is set to 0.45. K represents the traversal index variable used for algebraic summation, and its value ranges from 1 to N, increasing by integers. This represents an algebraic summation operation on the exponential terms of all N adjacent storage location nodes. The formula for calculating the stress transmission weight is based on the physical law of forced vibration stress diffusion in solid mechanics. When a local physical node in a metal space frame deforms under stress, the crosstalk stress potential energy transmitted to adjacent physical nodes exhibits a nonlinear exponential dissipation decay characteristic as the three-dimensional geometric straight-line distance in physical space increases. The node indices of adjacent storage location nodes are traversed, and the stress transmission weight of each adjacent storage location node is multiplied by the corresponding neighborhood stress fluctuation to obtain the normalized neighborhood stress value of the adjacent storage location node. A matrix orthogonalization numerical filling operation is performed using the central mass fluctuation and the normalized neighborhood stress values ​​corresponding to all adjacent storage location nodes to obtain the regional mass gradient matrix. Specifically, the process of performing matrix orthogonalization numerical filling is as follows: A two-dimensional floating-point numerical array is allocated and initialized in the random access memory of the edge control node; the central mass fluctuation is extracted and filled into the center point index coordinates of the two-dimensional floating-point numerical array; based on the pre-registered two-dimensional spatial physical arrangement orientation of each adjacent storage location node relative to the target storage location node in the warehousing database (e.g., left side, right side, directly above, or directly below on the same floor), the corresponding calculated normalized neighborhood stress value is filled into the two-dimensional floating-point numerical array and placed in the relative offset index coordinates with the center point index coordinates as the spatial reference origin; for the relative offset index coordinates of adjacent storage location nodes that are actually missing due to being at the physical edge boundary of the shelf, a data completion operation is performed on these coordinates using a constant zero value. The output of the two-dimensional floating-point numerical array after data splicing and filling is defined as the regional mass gradient matrix. The regional mass gradient matrix is ​​a structured data array used to accurately map the local physical shelf stress deformation coupling state and the nonlinear topological distribution law of bottom crosstalk weight loss in two-dimensional mathematical space.

[0085] Step S13: Extract continuous image frames based on multimodal trigger event signals, perform feature displacement extraction operation on the continuous image frames to obtain the material movement trajectory, and perform spatial projection difference calculation operation on the material movement trajectory to obtain the spatial offset vector.

[0086] In the process of generating the regional mass gradient matrix, in order to provide external confirmatory evidence independent of weight fluctuations for possible tentative replacement or cross-boundary misplacement of the target object from a kinematic perspective, the optical features in the multimodal trigger event signal are transformed into a high-precision spatial misplacement vector.

[0087] Specifically, optical features are extracted from the multimodal trigger event signal. These optical features refer to the two-dimensional pixel array features captured by the visual sensing component at the moment of light exposure, reflecting the physical contour of the target object's hand and the intensity and color gradient of light reflection on the material surface. Continuous image frames are obtained based on these optical features. Specifically, within the time interval from the start timestamp to the end timestamp corresponding to the observation window, optical features are continuously extracted according to the inherent video sampling frame rate of the visual sensing component and arranged in an array according to time order to generate continuous image frames. A continuous image frame refers to a set of two-dimensional static image sequences that, within a set time interval, discretely and at equal time intervals record the complete evolution of the material's movement from being grasped and leaving the target storage location node to finally returning to a physically static state. Feature displacement extraction is performed on the continuous image frames to obtain the material's movement trajectory. Specifically, the process of performing feature displacement extraction is as follows: the preceding image frame and the next adjacent image frame arranged in chronological order in the continuous image frames are used as input pairs; the two-dimensional displacement vector of the pixel representing the material's edge contour in the preceding image frame is calculated in the next adjacent image frame. The material edge contour refers to the closed or semi-closed pixel geometric boundary line formed between the surface of the target material entity and the surrounding background environment in a two-dimensional static image due to significant abrupt changes in light reflectivity, color gradient, or texture features. Traversing all adjacent image frame pairs in a continuous image frame, all two-dimensional displacement vectors are accumulated along the time axis to obtain the continuous pixel movement trajectory on the two-dimensional plane. Simultaneously, depth mapping parameters generated by the visual perception component's ranging are acquired. Combined with the camera intrinsic parameter matrix of the visual perception component, the continuous pixel movement trajectory is back-projected along the camera's optical center into a three-dimensional coordinate system. The depth mapping parameters are then used to assign depth dimension scales and perform absolute coordinate transformations on the back-projected rays, thereby mapping and restoring the two-dimensional pixel movement trajectory to the actual three-dimensional physical space position, resulting in a series of spatial three-dimensional geometric coordinate points representing the continuous physical displacement of the material at different sampling times. These spatial three-dimensional geometric coordinate points are combined and encapsulated in chronological order to generate a set of spatial three-dimensional geometric coordinate points, defined as the material movement trajectory. See also... Figure 4This diagram illustrates a two-dimensional continuous pixel trajectory tracking method based on multimodal optical features, as provided in an embodiment of the present invention. The diagram exemplarily shows three consecutive image frames arranged in chronological order, represented by three overlapping two-dimensional static image frames with light gray shaded borders. In each consecutive image frame, the material entity held by the target object is represented by a dark gray-filled geometric block, with a closed thick black outline indicating the material's edge contour extracted from the visual image. A thin solid line vector with a black arrow represents the two-dimensional displacement vector generated by the offset of the material's center pixel in the previous image frame to the next adjacent image frame. Simultaneously, the diagram exemplarily depicts a thick black solid curve running through the center of the material entity in each image frame, representing the material movement trajectory calculated by performing vector integration and accumulation of all two-dimensional displacement vectors along the time axis, thus visually mapping the continuous evolution of the material's physical displacement from a kinematic perspective.

[0088] The system silence confirmation signal triggered when the bottom shelf support is completely stationary is obtained. The three-dimensional geometric coordinates of the last point in the material movement trajectory corresponding to the trigger time of the system silence confirmation signal are extracted and defined as the actual landing point coordinates of the material. The pre-registered geometric center coordinates of the target storage location node are retrieved, and a spatial projection difference calculation operation is performed using the actual landing point coordinates of the material and the geometric center coordinates to obtain the spatial offset vector. Specifically, the corresponding values ​​of the geometric center coordinates on the horizontal axis, horizontal axis, and vertical axis are subtracted from the actual landing point coordinates of the material to obtain the horizontal physical coordinate deviation, vertical physical coordinate deviation, and height physical coordinate deviation, respectively. The physical coordinate deviation values ​​of the above three dimensions are encapsulated into an array to form the spatial offset vector. The spatial offset vector quantifies the absolute three-dimensional spatial offset direction and offset distance when the material is misplaced across boundaries or tentatively returned to its theoretical center position. See also Figure 5 This diagram illustrates the feature extraction of spatial offset and force crosstalk caused by cross-boundary misplacement of materials according to an embodiment of the present invention. The diagram exemplarily shows adjacent target storage location nodes and adjacent storage location nodes. The thin cross lines within the target storage location node on the left and their center intersection represent a pre-set absolute position reference, i.e., geometric center coordinates. When the target object undergoes a tentative return or misplacement action, the material entity placed on the shelf support dividing line, as indicated by the center point of the white rounded rectangle in the diagram, forms the actual landing point coordinates of the material. An exemplary black solid line with a double-headed arrow connects the corresponding positions of the geometric center coordinates and the actual landing point coordinates of the material. This double-headed arrow indicates and quantifies the absolute geometric distance between them, i.e., the offset distance. Simultaneously, the diagram exemplarily reflects that this cross-boundary overlap action causes the material weight to deviate from the target area, resulting in abnormal mechanical stress transmission to the adjacent weighing sensor on the right.

[0089] Step S10 addresses the technical challenge of traditional independent weighing environments, which rely solely on passive weight threshold triggering and lack multi-dimensional physical spatial perception and temporal isolation mechanisms, making them prone to misjudging unmotivated touching of target objects, vibration crosstalk between adjacent storage locations, or cross-boundary misplacement as genuine outbound actions. This is achieved by constructing an observation window, a regional mass gradient matrix, and a spatial offset vector. Specifically, the observation window transforms discrete physical handling actions into a digitally isolated space with strict temporal start and end boundaries and an absolute three-dimensional spatial addressing range, fundamentally eliminating external unmotivated temporal fluctuations and spatially unrelated interference. The regional mass gradient matrix accurately maps the local physical shelf stress deformation coupling state and the nonlinear topological distribution law of bottom-layer crosstalk weight loss in a two-dimensional mathematical space, constructing a global mass-conserving topology of physical stress coupling. The spatial offset vector precisely quantifies the absolute three-dimensional spatial offset direction and distance of materials deviating from the theoretical center position when cross-boundary misplacement or tentative return occurs from a kinematic perspective, providing external confirmation evidence independent of the bottom-layer weight for the target object's behavior.

[0090] Step S20: Input the spatial offset vector into the regional quality gradient matrix to perform spatial direction feature addressing and coupling comparison operations to obtain the topology offset diagnostic identifier. Extract the coupling offset feature from the topology offset diagnostic identifier to interrupt and suspend the tare zeroing process and generate the basic zero-point baseline.

[0091] Further, step S20 includes:

[0092] Step S21: Perform direction cosine decomposition on the spatial offset vector to obtain the spatial pointing angle. Based on the spatial pointing angle, perform spatial direction feature addressing operation and numerical comparison in the regional quality gradient matrix to output the topology offset diagnostic identifier.

[0093] After obtaining the spatial offset vector and the regional mass gradient matrix, a feature mapping matching mechanism is established at the edge control node to align the spatial offset vector calculated from consecutive image frames with the regional mass gradient matrix representing the crosstalk of mechanical stress at the bottom layer of the shelf support across dimensions. This aims to verify whether the cross-boundary offset behavior of materials captured by the visual perception component truly causes neighborhood stress fluctuations at the underlying physical gravity level.

[0094] Specifically, a direction cosine decomposition operation is performed on the spatial offset vector to obtain the spatial pointing angle and offset distance. Specifically, the lateral and longitudinal physical coordinate deviations of the spatial offset vector are extracted; the physical deflection angle of the two-dimensional plane vector formed by the lateral and longitudinal physical coordinate deviations relative to the reference coordinate axes is calculated using inverse trigonometric functions, defined as the spatial pointing angle; the sum of squares and the square root of the lateral, longitudinal, and height physical coordinate deviations are calculated using the three-dimensional Euclidean distance formula to obtain the three-dimensional geometric modulus, defined as the offset distance.

[0095] Based on the spatial pointing angle, a spatial orientation feature addressing operation is performed in the regional mass gradient matrix to extract the normalized neighborhood stress value as the first comparison item. Specifically, the process of performing the spatial orientation feature addressing operation is as follows: the center point index coordinate position of the regional mass gradient matrix is ​​regarded as the origin of a two-dimensional polar coordinate system, and the spatial pointing angle is mapped to this two-dimensional polar coordinate system; the relative offset index coordinate positions located in the quadrant pointed to by the spatial pointing angle are selected; the normalized neighborhood stress value corresponding to the relative offset index coordinate position is extracted from the regional mass gradient matrix as the first comparison item; simultaneously, the central mass fluctuation is extracted from the center point index coordinate position of the regional mass gradient matrix as the second comparison item. The values ​​of the first comparison item and the second comparison item are algebraically added together to obtain the stress closed-loop residual. When cross-boundary misplacement occurs, the target storage location node exhibits a decrease in gravity (i.e., a negative central mass fluctuation), while adjacent storage location nodes exhibit an increase in force (i.e., a positive normalized neighborhood stress value). Under the ideal physical law of energy conservation, their algebraic sum approaches zero. A pre-set bottom-level background noise threshold is obtained. This bottom-level background noise threshold refers to the inherent zero-point drift extreme value of the weighing sensor generated by the intelligent storage location unit under minor environmental vibration interference in an unloaded state. The absolute value of the stress closed-loop residual is numerically compared with the bottom-level background noise threshold, and the offset distance is simultaneously compared with zero. When the absolute value of the stress closed-loop residual is less than the bottom background noise threshold and the offset distance is greater than zero, it indicates that the offset distance in the physical space and the weight loss at the bottom layer can achieve logical closed-loop. The edge control node outputs a topology offset diagnostic label representing the logical truth value status label indicating that a cross-boundary offset or trial error replacement has occurred. Conversely, if the absolute value of the stress closed-loop residual is greater than or equal to the bottom background noise threshold, or the offset distance is equal to zero, the edge control node outputs a topology offset diagnostic label representing the logical false value status label.

[0096] Step S22: Based on the topology offset diagnostic identifier, trigger the abnormal interception logic to perform feature stripping calculation to obtain the coupling offset feature.

[0097] After obtaining the topology offset diagnostic identifier, in order to solve the blind spot problem that single-point weighing sensors cannot distinguish between real material leaving the site and tentative cross-boundary return, the edge control node performs parameter decoupling operation on the underlying data that is judged as an erroneous action.

[0098] Specifically, the topology offset diagnostic identifier is extracted. When the topology offset diagnostic identifier is a logical truth status label, the edge control node triggers anomaly interception logic, performs feature stripping calculation on the center mass fluctuation and the normalized neighborhood stress value as the first comparison item, and obtains the coupled offset feature. Specifically, the absolute value of the center mass fluctuation is defined as the target weight reduction; the absolute value of the normalized neighborhood stress value as the first comparison item is mathematically defined as the neighborhood weight increase; the stress decoupling transformation equation is constructed through the offset distance, and algebraic mapping calculation is performed on the target weight reduction and the neighborhood weight increase to obtain the compensation feature parameter C. The compensation feature parameter is used to quantify the purely physical numerical scalar of the mechanical stress crosstalk error caused by the cross-boundary shift of the material's physical center of gravity. It represents the real mass compensation benchmark that needs to be added back or subtracted from the registers of the underlying physical hardware. The analytical expression of the stress decoupling transformation equation is: in, This indicates the amount of the target weight reduction; This indicates the increase in weight within the neighborhood; V represents the offset distance. The spatial deviation multiplier factor refers to a mathematical proportionality constant used to convert pure geometric displacement scale into weight tolerance penalty. Its value is set based on static statistical fitting of the eccentric moment distribution data of the historical trial return action of the target object. The mechanical structure compensation coefficient refers to a physical constant used to mitigate the local force amplification effect caused by the inconsistent stiffness of different levels of the rack support structure. Its value is determined by calibration using the static elastic deformation ratio measured with standard weights applied to the unloaded intelligent storage unit. The stress decoupling transformation equation is constructed based on the principle of harmonic response. The target weight reduction and the neighboring weight increase are algebraically multiplied, and the resulting product is divided by the algebraic sum of the target weight reduction and the neighboring weight increase. Using this harmonic mean calculation method, a stable central common physical loss can be extracted when the bottom force values ​​of the target storage node and adjacent storage nodes are close. Simultaneously, the offset distance multiplied by the spatial deviation multiplier is added as a geometric penalty term to the aforementioned central common physical loss, and the mechanical structure compensation coefficient is used for overall algebraic amplification and mitigation, thus rigorously mapping the true mass loss parameter after excluding lateral contact interference from the weighing sensor. The target weight reduction, neighborhood weight increase, and compensation feature parameters are encapsulated in a structured array to generate a coupling offset feature. This coupling offset feature is a digital fingerprint set that comprehensively quantifies the degree of underlying quality loss and stress compensation direction caused by a single trial-and-error placement or cross-boundary misplacement action. If the topology offset diagnostic flag is a logical false value, the action is determined to be within the normal physical load-bearing range, and the abnormal interception logic is not triggered.

[0099] Step S23: Use the coupling offset feature to interrupt and suspend the tare zeroing process to generate the basic zero baseline.

[0100] After extracting the coupling offset features, in order to eliminate the underlying calculation error caused by the cross-boundary offset of the physical center of gravity of the target material, the edge control node intervenes in the data processing loop of the weighing sensor and performs dynamic calibration of the underlying physical benchmark based on the coupling offset features, thereby achieving adaptive repair of the underlying hardware without downtime.

[0101] Specifically, when the topology offset diagnostic flag is a logical truth status label, the tare zeroing process is extracted from the target storage location node and the affected adjacent storage location nodes, which are in a scheduling waiting state in the underlying communication bus. The affected adjacent storage location nodes refer to specific adjacent storage location nodes whose physical spatial orientation corresponds to the normalized neighborhood stress value extracted as the first comparison item in the regional quality gradient matrix. The tare zeroing process refers to the firmware program code executed by the microcontroller unit after detecting that the absolute value of the algebraic difference between the currently output sampled digital value and the initial empty weight value is less than a preset zero-point tracking threshold and remains stable for a preset period of time, forcibly overwriting the current residual sampled digital value to zero. The initial empty weight value refers to the base scalar value output by the weighing sensor when the intelligent storage location unit is not carrying any target material and is in a static state. The numerical values ​​are obtained by continuously reading the steady-state sampled digital values ​​output by the corresponding microcontroller unit through the underlying communication bus during the initialization and commissioning phase of the warehouse racking system or after the completion of the last legal material emptying action and physical settling. The zero-point tracking threshold refers to the maximum allowable inherent zero-point drift error range determined by the hardware characteristics of the weighing sensor. The time period is set based on the mechanical oscillation convergence time of the intelligent storage unit after being subjected to a typical physical impact, to ensure that the tare calculation can only be performed after the rack support deformation has completely stopped. The microcontroller unit refers to the micro-computing processing chip integrated at the bottom of the intelligent storage unit, which is responsible for controlling the weighing sensor. The edge control node generates a kernel-level interrupt control signal and sends the kernel-level interrupt control signal to the microcontroller units of the target storage location node and the affected adjacent storage location nodes. The kernel-level interrupt control signaling refers to dedicated machine instruction code used to trigger the underlying hardware register state latching mechanism. In response to the kernel-level interrupt control signaling, the microcontroller performs an interrupt suspension operation on the tare zeroing process, thereby forcibly terminating the physical automatic tare action that the microcontroller is about to perform. After completing the interrupt suspension operation and confirming that the rack support has completely returned to a mechanically static state, the edge control node reads the sampled digital values ​​output by the target storage location node and the affected adjacent storage location nodes in the current steady state through the underlying communication bus, defining them as the static weight register values. The compensation feature parameters contained in the coupling offset features are extracted. An algebraic cancellation calculation operation is performed using the static weight register values ​​and the compensation feature parameters to obtain the updated values.Specifically, the process of performing the algebraic cancellation calculation is as follows: For the target storage location node, the corresponding static weight register value is added to the compensation feature parameter to fill in the weight value lost due to the physical center of gravity shift, thus obtaining the updated value of the target storage location node; for the affected adjacent storage location nodes, the corresponding static weight register value is subtracted from the compensation feature parameter to deduct the weight value artificially increased due to the unexpected contact with the target material, thus obtaining the updated value of the adjacent storage location node. The updated values ​​of the target storage location node and the adjacent storage location node are forcibly burned back into the reference configuration memory sector of the corresponding microcontroller unit through the underlying communication bus, completing the underlying physical calibration replacement operation. The reference configuration memory sector refers to the physical address range in the non-volatile memory inside the microcontroller unit, specifically used to persistently store the physical zero-point calibration parameters of the weighing sensor. The value in this range determines the base deduction standard when the microcontroller unit outputs the net weight. The updated value output after the burning and replacement is uniformly defined as the basic zero-point baseline. The aforementioned basic zero-point baseline refers to a new calibration constant that, after excluding cross-boundary physical stress crosstalk deviations, can truly reflect the current steady-state unloaded force of each node's intelligent storage unit. If the topology offset diagnostic flag is a logical false value status label, i.e., no abnormal interception logic is triggered, the edge control node directly reads the original calibration constant in the corresponding microcontroller's reference configuration memory sector as the basic zero-point baseline via the underlying communication bus. Therefore, in the above steps of this embodiment, through the deep cooperative coupling of the calculated spatial offset vector and the constructed regional quality gradient matrix, the edge control node overcomes the traditional technical bias in this field that "each storage location weighing sensor must be independently tare-zeroed to cut off adjacent interference." Specifically, random erroneous actions such as "trial placement" or "cross-boundary misplacement" occurring in the target object, which would originally interfere with the underlying tare logic, are transformed into detection pulses for the coupling parameters of adjacent nodes of the shelf support. Edge control nodes utilize the local weight redistribution data caused by these erroneous actions—i.e., coupling offset features—to silently reconstruct the entire racking stress deformation model and calibrate the underlying hardware's baseline zero point without interrupting any normal inbound / outbound business processes. This logical derivation demonstrates that as the physical usage time of warehouse racking increases, it not only avoids accumulating the zeroing error inherent in traditional independent tare algorithms, but also leverages each human error to achieve fatigue-resistant self-healing and downtime-free calibration of the hardware architecture. This fundamentally eliminates global baseline drift and accumulated errors after prolonged high-concurrency operation, achieving unexpected technical results.

[0102] Step S20 solves the blind spot problem of single-point weighing sensors being unable to distinguish between actual material departure and tentative cross-boundary placement by outputting topology offset diagnostic markers, extracting coupling offset features, and generating a basic zero-point baseline. It also addresses the technical challenges of global zero-point baseline drift and accumulated calculation errors caused by traditional distributed node independent tare algorithms. This achieves fatigue-resistant self-healing of the hardware architecture and adaptive repair of the underlying hardware in a downtime-free state. Specifically, the topology offset diagnostic marker performs cross-dimensional alignment and comparison between the visually captured three-dimensional absolute displacement and the underlying metal stress crosstalk, achieving a logical closed-loop verification of physical spatial absolute offset and underlying weight loss. The coupling offset features comprehensively quantify the degree of mass loss and stress compensation direction caused by tentative actions or cross-boundary misplacement, cleverly transforming random erroneous actions that would otherwise interfere with the system into detection pulses for the coupling parameters of adjacent nodes of the shelf support. The kernel-level interrupt suspension mechanism of the underlying microcontroller unit of the basic zero-point baseline uses decoupling parameters to forcibly overwrite the static weight register value, completely eliminating errors caused by physical cross-boundary stress at the source and restoring the absolutely steady-state no-load calibration constant.

[0103] Step S30: Obtain the steady-state sampled digital value and perform algebraic difference conversion with the basic zero-point baseline to obtain the actual material change. Perform data fusion based on the actual material change to obtain the inbound and outbound registration data. Perform encryption processing calculation on the inbound and outbound registration data to obtain a high-reliability traceability block. Write the high-reliability traceability block into the traceability block storage chain for solution and output the global inventory total.

[0104] Further, step S30 includes:

[0105] Step S31: Obtain the steady-state sampled digital value, perform algebraic difference conversion operation using the steady-state sampled digital value and the baseline zero point to obtain the net weight fluctuation value, and perform integer division and down-rounding mapping calculation on the net weight fluctuation value to obtain the actual material change.

[0106] After interrupting and suspending the tare and zeroing process using coupling offset features and generating a baseline zero point, the edge control node enters the application layer's data settlement phase. This aims to map the net weight fluctuation value, which excludes cross-boundary physical stress crosstalk deviations, to the actual material change representing material access actions.

[0107] Specifically, the edge control node acquires the system silence confirmation signal through the underlying data link. This system silence confirmation signal refers to a high-level logic control signal sent to the edge control node after the observation window reaches its end timestamp. This signal indicates that the rack support has fully recovered its mechanical static state. When the microcontroller unit corresponding to the target storage location node detects that the absolute value of the first-order time derivative of the continuously output sampled digital quantity from the weighing sensor is less than the underlying background noise threshold, the edge control node sends this signal. In response to the system silence confirmation signal, the edge control node reads the single valid sampled data output by the target storage location node in its current mechanically static state through the underlying communication bus, defining it as the steady-state sampled digital quantity value. An algebraic difference conversion operation is performed between the steady-state sampled digital quantity value and the baseline zero point to obtain the net weight fluctuation value. Specifically, the baseline zero point is subtracted from the steady-state sampled digital quantity value to obtain the net physical difference value with positive and negative mathematical signs. The net weight fluctuation value filters out the influence of zero-point baseline drift caused by trial return actions or adjacent storage location nodes, and quantifies the absolute amount of increase or decrease in the underlying physical gravity of the target storage location node before and after a single material storage and retrieval action due to the spatial transfer of the actual target material.

[0108] Obtain the unit baseline weight. The unit baseline weight refers to the standard physical unit mass quantity, uniquely bound to the material type stored at the target storage location node, and pre-entered into the warehousing database. Its value is set based on: before the material is officially put into storage, it is obtained by batch weighing and sampling using a standardized high-precision electronic balance, and the arithmetic mean is calculated. Using the net weight fluctuation value and the unit baseline weight, a discrete quantity conversion equation is constructed, and integer division and rounding are performed to obtain the actual material change Q. The actual material change Q refers to the exact increase or decrease in the number of discrete standard material entities that undergo real physical spatial transfer during a single material storage and retrieval operation at the target storage location node. The analytical expression of the discrete quantity conversion equation is: ,in, This indicates the net weight fluctuation value; This indicates the standard weight of the individual item; This represents the floor function operator; The physical tolerance compensation factor refers to a scalar quantity used to compensate for slight quality fluctuations in the outer packaging of materials or weight gain errors caused by changes in humidity in the storage environment. Its value is set at 3% of the base weight of a single item. The discrete quantity conversion equation is constructed based on the linear accumulation law of ideal discrete items' mass. By introducing the physical tolerance compensation factor into the numerator, it statically accommodates environmental and packaging physical tolerances. Furthermore, it uses a floor function to force the continuous distribution of simulated physical quality fluctuations into discrete integer-level increases and decreases in item quantity that strictly conform to business settlement logic. This avoids data misalignment caused by rounding or truncation of floating-point numbers. The calculated actual material change is an integer scalar with positive and negative mathematical signs. A positive sign indicates that the material has been put into storage, while a negative sign indicates that the material has been retrieved from storage.

[0109] Step S32: Perform data fusion based on the actual material change to obtain inbound and outbound registration data. Calculate a dynamic salting key based on the inbound and outbound registration data and use the dynamic salting key to perform encryption processing on the inbound and outbound registration data to obtain a highly reliable traceability block.

[0110] After obtaining the actual material changes, the edge control node needs to cryptographically bind the increase or decrease in material quantity to the personnel identity information of the target object. This aims to prevent personnel with advanced database access from illegally tampering with historical business accounts in the application layer information system, ensuring that the data generated by the underlying physical actions is irreversible in both time and space.

[0111] Specifically, identity binding information and actual material changes are extracted. At the edge control node, the identity binding information and actual material changes are fused to obtain inbound / outbound registration data. Specifically, the identity binding information is used as the header attribute identifier of the underlying network data packet; the actual material changes are used as the core business payload of this underlying network data packet; the physical identifier code of the target storage location node pre-registered in the warehousing database is extracted and used as additional location metadata; the header attribute identifier, core business payload, and additional location metadata are concatenated and combined according to a specified format to generate a plain text message entity, defined as inbound / outbound registration data. This inbound / outbound registration data completely records the participants in a single material storage / retrieval event, the location of the physical node where the event occurred, and the pure business logic representing the actual material changes in quantity. To endow this inbound / outbound registration data with an immutable underlying traceability attribute, the inbound / outbound registration data is encrypted and calculated at the edge control node to obtain a highly reliable traceability block. Specifically, the system time at the moment the inbound / outbound registration data is generated is obtained and defined as the global physical timestamp. Simultaneously, the power supply voltage characteristics and signal-to-noise ratio (SNR) characteristics of the weighing sensor at the target storage location node are extracted. The power supply voltage characteristic refers to the transient floating-point voltage sample value generated by the microcontroller's power supply pin at a specific physical microsecond due to internal hardware load fluctuations. The SNR characteristic refers to the real-time physical ratio of the effective power of the analog signal of the weighing sensor under the current mechanically static state to the background electromagnetic noise power. A bitwise concatenation operation is performed between the binary sequence of the power supply voltage characteristic and the binary sequence of the SNR characteristic, and the resulting concatenated data is encapsulated and defined as the underlying sensor SNR characteristic. The underlying sensor SNR characteristic is a random number seed characterizing the transient electromagnetic environment physical fluctuations of the underlying hardware.

[0112] The data entering and leaving the warehouse undergoes binary serialization, converting text-formatted data into a one-dimensional binary byte stream, defined as the one-dimensional basic byte stream. The global physical timestamp is converted into binary format data, defined as a binary time series; the underlying sensor signal-to-noise ratio characteristics are converted into binary format data, defined as a binary physical feature sequence. A bitwise XOR operation is performed between the binary time series and the binary physical feature sequence to obtain a dynamically salted key. This dynamically salted key is a binary cryptographic encryption string that integrates absolute time attributes and hardware spatial physical environment fluctuations. Because it is generated from two fixed-length binary format data, the dynamically salted key has a fixed inherent byte length. Data block cross-interference and shifting operations are performed between the one-dimensional basic byte stream and the dynamically salted key to obtain a scrambled data stream. Specifically, the inherent byte length of the dynamically salted key is obtained. The one-dimensional basic byte stream is then sliced ​​into equal-length segments according to this inherent byte length, with any insufficient length at the end padded with zero bytes to obtain multiple basic data blocks. The dynamically salted key is inserted between each adjacent basic data block for string concatenation, resulting in an expanded data sequence. Each byte in the expanded data sequence undergoes a cyclic left shift mathematical operation according to a preset constant, resulting in a scrambled data stream with rearranged positions. The constant is set based on the cryptographic principle of deviating from the default half-byte (4-bit) alignment boundary of the underlying computer architecture to enhance the bitwise operation scrambling effect and maximize data tamper resistance. A prime scalar greater than zero and not a multiple of 4 is selected; for example, it is set to the value 3. The cyclic left shift mathematical operation refers to the bitwise operation process of shifting out the leftmost bit of a binary byte and refilling it to the rightmost least significant bit. The scrambled data stream refers to the long binary data stream in its ciphertext state generated after the above slicing, key insertion, and cyclic left shift.

[0113] The SHA-256 standard hash algorithm is invoked to perform hash calculations on the scrambled data stream, yielding a tamper-proof state hash value. This hash calculation is a unique process that uses the SHA-256 algorithm to map input data of arbitrary length to a fixed 256-bit output. The tamper-proof state hash value refers to the fixed-length hexadecimal string characteristic of the calculated output. The basis of this encryption process is that even for inbound and outbound registration data with identical actual material changes occurring at different times, the global physical timestamp and the signal-to-noise ratio (SNR) characteristics of the underlying sensors reflecting fluctuations in the underlying electromagnetic environment cannot be precisely reproduced in absolute physical spatiotemporal coordinates. Therefore, the dynamically salted key generated by bitwise XOR logic operations will inevitably be different, triggering a cryptographic avalanche effect in the SHA-256 standard hash algorithm. This results in a significant out-of-order change in the final calculated tamper-proof state hash value, thus preventing application-layer replay attacks and forgery / tampering of historical data. The tamper-proof state hash value, the underlying sensor SNR characteristics, the global physical timestamp, and the inbound / outbound registration data are then mapped to the same directory and logically encapsulated to obtain a highly reliable traceability block. Specifically, using the tamper-proof status hash value as the data header signature, and combining the underlying sensor signal-to-noise ratio characteristics, global physical timestamps, and the order of entry and exit registration data as the core payload, a high-reliability traceability block is constructed. This high-reliability traceability block is a closed and independent data structure unit with self-falsifying hash signature attributes.

[0114] Step S33: Write the high-reliability traceability block into the traceability block storage chain. Perform algebraic accumulation and iterative calculation on the actual material change based on the traceability block storage chain to obtain the latest inventory quantity at a single point. Perform global traversal and summation aggregation calculation on the latest inventory quantity at a single point to obtain the global inventory total.

[0115] After obtaining the highly reliable traceability block, in order to support the real-time display of application data and replenishment decisions, the edge control node writes the highly reliable traceability block into the warehouse database and restores the latest material reserve status in the current physical warehouse space.

[0116] Specifically, the process involves extracting the warehouse database, which contains an independently partitioned memory block with a chained append-only read-only logical structure, defined as the traceability block storage chain. This traceability block storage chain strictly prohibits physical data overwriting or deletion operations. A memory logical pointer is obtained; this pointer is a low-level database index marker used to indicate the relative offset of the physical storage address and the unidirectional successor association with the data node. The most recent historical traceability block is extracted; this block refers to the high-reliability traceability block entity that was last successfully written to the traceability block storage chain before the current material access operation. From the data header signature of this most recent historical traceability block, its corresponding tamper-proof status hash value is extracted, and the characteristic bits of the tamper-proof status hash value are obtained. These characteristic bits are specific consecutive byte segments in the hexadecimal string of the tamper-proof status hash value, specifically used to identify the connection verification of the data block's predecessor node. A high-reliability traceability block is written to the end of the storage queue of the traceability block storage chain. Using the memory logical pointer, this high-reliability traceability block is unidirectionally linked to the characteristic bit of the tamper-proof state hash value of the most recent preceding historical traceability block, resulting in an irreversible, permanently fixed chained storage record. This permanently fixed chained storage record ensures that physical interaction actions leave an absolutely time-ordered tracking record at the data logic layer. The exact number of remaining discrete standard material entities cached in the warehouse database before the start timestamp of the target storage location node is extracted and defined as historical inventory data. The actual material change is extracted, and an algebraic cumulative iterative calculation operation is performed using the historical inventory data and the actual material change to obtain the latest inventory level at a single point. Specifically, the process involves extracting integer scalars from historical inventory data; extracting integer scalars with positive or negative mathematical signs from actual material changes; performing signed algebraic addition directly on the integer scalars of historical inventory data and actual material changes; if the integer scalar of actual material changes is negative, indicating a valid outbound retrieval action, the cumulative calculation shows a decrease in the historical inventory data of that target storage location node; if the integer scalar of actual material changes is positive, indicating a valid inbound action, the cumulative calculation shows an increase in the historical inventory data of that target storage location node; and outputting the final cumulative calculation result as the single-point latest inventory quantity. A global traversal and summation aggregation calculation is performed on the single-point latest inventory quantities corresponding to all registered smart storage units in the warehouse database at the current moment to obtain the global inventory sum. Specifically, the single-point latest inventory quantities of all smart storage units in the warehouse database are extracted and arithmetic summation is performed on all extracted single-point latest inventory quantities to obtain the global inventory sum.The total global inventory is a high-confidence final business indicator that absolutely reflects the precise quantity of materials currently stored in the physical warehouse space after undergoing multiple logical processes, including dynamic calibration of the underlying physical zero point, filtering of probing action features, and cryptographic anti-tampering encryption calculation.

[0117] Step S30 solves the problem of business data logic misalignment caused by rounding and truncation of the last digit in the underlying simulated physical quality fluctuation calculation by calculating the actual material change, encapsulating the high-reliability traceability block, and outputting the global inventory total. It also addresses the security risk of personnel with advanced database access illegally tampering with historical accounts in the application layer information system. This achieves seamless and accurate mapping from pure physical weight to discrete commodity quantity and high-security iterative traceability of accounts that are tamper-proof at the underlying level. Among them, the actual material change, combined with the physical tolerance compensation factor and the down-rounding mapping operation, forces the continuously distributed physical weight difference to be filtered into discrete integer-level increase and decrease scalars that strictly conform to the business settlement logic; the high-reliability traceability block cryptographically interweaves and splices personnel identity, actual material change, and irreproducible absolute timestamps and physical fluctuations of the underlying sensor electromagnetic environment, triggering an avalanche effect of hash feature calculation, generating a closed data base with self-falsifying hash signature attributes; after the global inventory summation is solidified by blockchain-style appending, it aggregates the latest single-point inventory of all nodes in the warehousing database after noise reduction, decoupling compensation and anti-tampering calculation, providing a high-confidence final business indicator that absolutely reflects the current material reserve status.

[0118] Example 2

[0119] This embodiment, based on Embodiment 1, provides an automated inbound and outbound registration system that integrates weighing and barcode scanning in warehouses, such as... Figure 6 As shown, it includes:

[0120] Feature reconstruction module: used to acquire multimodal trigger event signals, construct observation windows to collect weight fluctuation time series data, perform spatial feature reconstruction calculation on weight fluctuation time series data to obtain regional mass gradient matrix, extract continuous image frames based on multimodal trigger event signals, and perform feature displacement extraction and spatial projection difference calculation on continuous image frames to obtain spatial offset vector;

[0121] Offset calibration module: It is used to input the spatial offset vector into the regional quality gradient matrix to perform spatial orientation feature addressing and coupling comparison operations, obtain the topology offset diagnostic identifier, extract the coupling offset feature from the topology offset diagnostic identifier, perform an interruption and suspension operation on the tare zeroing process, and generate the basic zero-point baseline.

[0122] Trusted Traceability Module: This module is used to acquire steady-state sampled digital values ​​and perform algebraic difference conversion with the baseline zero point to obtain the actual material change. Based on the actual material change, it performs data fusion to obtain inbound and outbound registration data. It performs encryption processing on the inbound and outbound registration data to obtain a high-trust traceability block. The high-trust traceability block is written into the traceability block storage chain for calculation and outputs the global inventory total.

[0123] In the feature reconstruction module, the process involves acquiring multimodal trigger event signals, constructing an observation window to collect weight fluctuation time-series data, performing spatial feature reconstruction calculations on the weight fluctuation time-series data to obtain the regional mass gradient matrix, extracting continuous image frames based on the multimodal trigger event signals, and performing feature displacement extraction and spatial projection difference calculations on the continuous image frames to obtain a spatial offset vector, including:

[0124] Step S11: Obtain multimodal trigger event signals and construct an observation window based on the multimodal trigger event signals;

[0125] Step S12: Activate the underlying analog-to-digital conversion channel based on the observation window to generate weight fluctuation time series data, and call the stress extraction algorithm to perform spatial feature reconstruction calculation on the weight fluctuation time series data to obtain the regional mass gradient matrix;

[0126] Step S13: Extract continuous image frames based on multimodal trigger event signals, perform feature displacement extraction operation on the continuous image frames to obtain the material movement trajectory, and perform spatial projection difference calculation operation on the material movement trajectory to obtain the spatial offset vector.

[0127] In the offset calibration module, the process of inputting the spatial offset vector into the regional quality gradient matrix to perform spatial orientation feature addressing and coupling comparison operations to obtain a topological offset diagnostic identifier, extracting coupling offset features from the topological offset diagnostic identifier to interrupt and suspend the tare zeroing process, and generating a base zero-point baseline includes:

[0128] Step S21: Perform direction cosine decomposition on the spatial offset vector to obtain the spatial pointing angle. Based on the spatial pointing angle, perform spatial direction feature addressing operation and numerical comparison in the region quality gradient matrix to output the topology offset diagnostic identifier.

[0129] Step S22: Based on the topology offset diagnostic identifier, trigger the anomaly interception logic to perform feature stripping calculation to obtain the coupling offset feature;

[0130] Step S23: Use the coupling offset feature to interrupt and suspend the tare zeroing process to generate the basic zero baseline.

[0131] In the trusted traceability module, the steady-state sampled digital value is acquired and converted using algebraic difference with the baseline zero point to obtain the actual material change. Data fusion is then performed based on the actual material change to obtain inbound / outbound registration data. This data is then encrypted to obtain a high-trustworthiness traceability block. This high-trustworthiness traceability block is written into the traceability block storage chain for processing, and the total global inventory is output, including:

[0132] Step S31: Obtain the steady-state sampled digital value, perform algebraic difference conversion operation with the steady-state sampled digital value and the baseline zero point to obtain the net weight fluctuation value, and perform integer division and down rounding mapping calculation on the net weight fluctuation value to obtain the actual material change.

[0133] Step S32: Perform data fusion based on the actual material change to obtain inbound and outbound registration data. Calculate a dynamic salting key based on the inbound and outbound registration data and use the dynamic salting key to perform encryption processing on the inbound and outbound registration data to obtain a highly reliable traceability block.

[0134] Step S33: Write the high-reliability traceability block into the traceability block storage chain. Perform algebraic accumulation and iterative calculation on the actual material change based on the traceability block storage chain to obtain the latest inventory quantity at a single point. Perform global traversal and summation aggregation calculation on the latest inventory quantity at a single point to obtain the global inventory total.

[0135] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0136] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely preferred embodiments of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An automated warehouse entry and exit registration method integrating weighing and barcode scanning, characterized in that, The method includes: Acquire multimodal trigger event signals and construct an observation window to collect weight fluctuation time series data. Perform spatial feature reconstruction calculation on the weight fluctuation time series data to obtain the regional mass gradient matrix. Extract continuous image frames based on multimodal trigger event signals. Perform feature displacement extraction and spatial projection difference calculation on the continuous image frames to obtain the spatial offset vector. The spatial offset vector is input into the regional quality gradient matrix to perform spatial orientation feature addressing and coupling comparison operations to obtain the topology offset diagnostic identifier. The coupling offset feature is extracted from the topology offset diagnostic identifier to interrupt and suspend the tare zeroing process and generate the basic zero-point baseline. Obtain the steady-state sampled digital value and perform algebraic difference conversion with the basic zero-point baseline to obtain the actual material change. Perform data fusion based on the actual material change to obtain the inbound and outbound registration data. Perform encryption processing calculation on the inbound and outbound registration data to obtain a high-reliability traceability block. Write the high-reliability traceability block into the traceability block storage chain for solution and output the global inventory summation.

2. The automatic inbound and outbound registration method integrating weighing and barcode scanning in warehouses according to claim 1, characterized in that, The method for constructing the observation window includes: Extract identity binding information and operation location coordinates from multimodal trigger event signals; Obtain the warehouse database, perform spatial absolute addressing in the warehouse database based on the operation location coordinates to determine the target storage location node, and retrieve adjacent storage location nodes based on the network association physical pointers pre-registered for the target storage location node; Extract the start timestamp from the multimodal trigger event signal, obtain the action duration, and add it to the start timestamp to get the end timestamp; Extract the bounding box boundary parameters of the target storage location node and adjacent storage location nodes respectively, and use the bounding box boundary parameters to construct the global coordinate boundary; A multidimensional data buffer is constructed, and the original sampled data is extracted from the multimodal trigger event signal. The original sampled data that falls within the global coordinate boundary and is in the interval from the start time stamp to the end time stamp is filled into the multidimensional data buffer to generate the observation window.

3. The automatic inbound and outbound registration method integrating weighing and barcode scanning in warehouses according to claim 2, characterized in that, The method for acquiring the weight fluctuation time series data includes: Activate the underlying analog-to-digital conversion channels of the target cargo location node and all adjacent cargo location nodes located within the observation window, and collect analog voltage signals at preset equal time intervals to obtain sampled digital values. The sampled digital values ​​continuously output from the start timestamp to the end timestamp are concatenated in chronological order to generate time-series data of weight fluctuations for the target cargo location node and each adjacent cargo location node, and are simultaneously stored in the multidimensional data cache.

4. The automatic inbound and outbound registration method integrating weighing and barcode scanning in warehouses according to claim 3, characterized in that, The method for calculating the regional quality gradient matrix includes: Extract the weight fluctuation time series data of the target cargo location node, calculate the algebraic difference between the sampled digital values ​​corresponding to the end timestamp and the start timestamp to obtain the central mass fluctuation amount, and simultaneously extract the weight fluctuation time series data of each adjacent cargo location node to calculate the algebraic difference to obtain the neighborhood stress fluctuation amount. The stress transmission weight of adjacent cargo location nodes is calculated by obtaining the spatial damping attenuation coefficient and the three-dimensional geometric straight line distance. The stress transmission weight is then multiplied by the neighborhood stress fluctuation to obtain the normalized neighborhood stress value. Perform matrix orthogonalization numerical filling operations on the central mass fluctuation and the normalized neighborhood stress values ​​to obtain the regional mass gradient matrix.

5. The automatic inbound and outbound registration method integrating weighing and barcode scanning in warehouses according to claim 2, characterized in that, The method for calculating the spatial offset vector includes: Optical features are extracted from multimodal trigger event signals, and optical features are continuously captured within the time interval from the start timestamp to the end timestamp to generate continuous image frames. Calculate the two-dimensional displacement vector between adjacent image frames and integrate it along the time axis to obtain the continuous pixel movement trajectory. Obtain the depth mapping parameters and back-project the continuous pixel movement trajectory onto the three-dimensional coordinate system to generate the material movement trajectory. Obtain the system's silent confirmation signal and extract the three-dimensional geometric coordinates of the material's movement trajectory, defining them as the actual landing point coordinates of the material. Obtain the geometric center coordinates of the target storage location node, subtract the geometric center coordinates from the actual landing point coordinates of the material to obtain the lateral physical coordinate deviation, longitudinal physical coordinate deviation, and height physical coordinate deviation, and encapsulate them into a spatial offset vector.

6. The automatic inbound and outbound registration method integrating weighing and barcode scanning in warehouses according to claim 5, characterized in that, The method for outputting the topology offset diagnostic identifier includes: The spatial offset vector is used to perform directional cosine decomposition to obtain the spatial pointing angle. The spatial offset vector is then used to calculate the offset distance. The spatial pointing angle is mapped to a two-dimensional polar coordinate system with the center point index coordinate position of the preset region quality gradient matrix as the origin. The relative offset index coordinate position within the quadrant in which the spatial pointing angle points is selected. The normalized neighborhood stress value is extracted from the relative offset index coordinates as the first comparison item, and the central mass fluctuation is extracted from the center point index coordinates as the second comparison item. The first comparison item and the second comparison item are algebraically added to obtain the stress closed-loop residual. When the absolute value of the stress closed-loop residual is less than the preset bottom background noise threshold and the offset distance is greater than zero, the topology offset diagnosis flag of the logic true value status label is output; otherwise, the topology offset diagnosis flag of the logic false value status label is output.

7. The automatic inbound and outbound registration method integrating weighing and barcode scanning in warehouses according to claim 6, characterized in that, The method for extracting the coupling offset features includes: When the topology offset diagnostic flag is a logical truth status label, the abnormal interception logic is triggered, the absolute value of the center mass fluctuation is defined as the target weight reduction, and the absolute value of the first comparison item is defined as the neighborhood weight increase. Obtain the spatial deviation multiplier factor and mechanical structure compensation coefficient, construct the stress decoupling transformation equation using the offset distance, perform algebraic mapping calculation on the target weight reduction and the neighborhood weight increase to obtain the compensation characteristic parameters; The target weight reduction, neighborhood weight increase, and compensation feature parameters are encapsulated in a structured array to generate coupled offset features.

8. The automatic inbound and outbound registration method integrating weighing and barcode scanning in warehouses according to claim 7, characterized in that, The method for obtaining the basic zero-point baseline includes: Generate kernel-level interrupt control signaling and send it to the target cargo location node and the adjacent cargo location nodes that provide the spatial orientation of the first comparison item in the regional quality gradient matrix, forcing the tare and zeroing process to be interrupted and suspended. After confirming that the mechanical system is at rest, the sampled digital values ​​output by the target storage location node and the affected adjacent storage location nodes under steady state are read and defined as the static weight register values. The compensation feature parameter is extracted from the coupling offset feature. The static weight register value of the target storage location node is added to the compensation feature parameter, and the static weight register value of the affected adjacent storage location node is subtracted from the compensation feature parameter. Algebraic cancellation calculation is then performed to obtain the updated value. The updated values ​​will be burned back to the baseline configuration memory sector of the corresponding microcontroller unit and uniformly defined as the baseline zero point.

9. The automatic inbound and outbound registration method integrating weighing and barcode scanning in warehouses according to claim 8, characterized in that, The method for calculating the actual material change includes: The system acquires a silent confirmation signal and reads the single valid sampled data output by the target cargo location node in a state of absolute mechanical stillness through the underlying communication bus, which is defined as the steady-state sampled digital value. Subtract the baseline zero-point value from the steady-state sampled digital value and perform an algebraic difference conversion operation to obtain the net weight fluctuation value; Obtain the standard weight and physical tolerance compensation factor of the single item, construct a discrete quantity conversion equation using the net weight fluctuation value and the standard weight of the single item, and perform a floor function mapping calculation with the physical tolerance compensation factor to obtain the actual material change with positive and negative mathematical signs.

10. The automatic inbound and outbound registration method integrating weighing and barcode scanning in warehouses according to claim 9, characterized in that, The method for generating the inbound / outbound registration data and the high-reliability traceability block includes: Extract the physical identification code of the target storage location node from the warehouse database, and combine it with the identity binding information and the actual material change to generate inbound and outbound registration data; Obtain the global physical timestamp, extract the power supply voltage characteristics and signal-to-noise ratio characteristics of the weighing sensor, perform bit concatenation to generate the underlying sensor signal-to-noise ratio characteristics, and perform bitwise XOR logic operation on the binary sequence of the global physical timestamp and the underlying sensor signal-to-noise ratio characteristics to obtain the dynamic salting key; The data entry and exit registration data is converted into a one-dimensional basic byte stream, and then subjected to cross-obfuscation and cyclic left shift mathematical operations with a dynamically salted key to obtain a scrambled data stream. The standard hash algorithm is called to perform hash calculation on the scrambled data stream to obtain the tamper-proof status hash value. The tamper-proof status hash value is logically encapsulated with the inbound and outbound registration data to obtain a highly reliable traceability block.

11. The automatic inbound and outbound registration method integrating weighing and barcode scanning in warehouses according to claim 10, characterized in that, The method for outputting the global inventory summation includes: Extract the traceability block storage chain from the warehouse database, obtain the most recent previous historical traceability block and the memory logical pointer used to indicate the data association relationship; Extract the tamper-proof state hash value from the most recent preceding historical trace block, and obtain the characteristic bits of the tamper-proof state hash value; Write the high-reliability traceability block to the end of the storage queue of the traceability block storage chain, and use a memory logical pointer to link it unidirectionally to the feature bit; Extract the historical inventory data cached in the warehouse database before the start timestamp of the target storage location node, and perform a signed algebraic cumulative iterative calculation operation on the historical inventory data and the actual material change to obtain the latest inventory quantity at a single point. Extract all registered smart storage units from the warehouse database, perform global traversal and arithmetic addition on the latest inventory of each smart storage unit, and output the total global inventory.

12. An automated warehouse entry and exit registration system integrating weighing and barcode scanning, used to implement the automated warehouse entry and exit registration method integrating weighing and barcode scanning as described in any one of claims 1-11, characterized in that, The system includes: Feature reconstruction module: used to acquire multimodal trigger event signals, construct observation windows to collect weight fluctuation time series data, perform spatial feature reconstruction calculation on weight fluctuation time series data to obtain regional mass gradient matrix, extract continuous image frames based on multimodal trigger event signals, and perform feature displacement extraction and spatial projection difference calculation on continuous image frames to obtain spatial offset vector; Offset calibration module: It is used to input the spatial offset vector into the regional quality gradient matrix to perform spatial orientation feature addressing and coupling comparison operations, obtain the topology offset diagnostic identifier, extract the coupling offset feature from the topology offset diagnostic identifier, perform an interruption and suspension operation on the tare zeroing process, and generate the basic zero-point baseline. Trusted Traceability Module: This module is used to acquire steady-state sampled digital values ​​and perform algebraic difference conversion with the baseline zero point to obtain the actual material change. Based on the actual material change, it performs data fusion to obtain inbound and outbound registration data. It performs encryption processing on the inbound and outbound registration data to obtain a high-trust traceability block. The high-trust traceability block is written into the traceability block storage chain for calculation and outputs the global inventory total.

Images