Intelligent Warehouse Management Method Based on Multimodal Feature Fusion

Abstract

This application provides an intelligent warehouse management method based on multimodal feature fusion. The method acquires image and weight features of goods and performs multimodal fusion processing to determine the quality deterioration level of the goods; it calls the three-dimensional topology model of a digital twin warehouse system to retrieve target storage areas matching the goods' quality level and determines the occupancy status of storage locations; if an area is saturated, it retrieves adjacent alternative vacant storage locations based on the topology center and batch modifies their logical attribute labels to achieve dynamic expansion of storage areas and continuity of spatial layout; finally, it controls the handling equipment to enter the warehouse and synchronously updates the virtual space status. This invention breaks the rigid zoning limitations of physical shelves, achieves adaptive and flexible management of storage space, and optimizes the continuity of automated handling paths.

Description

Technical Field

[0001] This disclosure relates to the field of intelligent warehousing technology, and in particular to an intelligent warehouse management method based on multimodal feature fusion. Background Technology

[0002] In modern logistics and supply chain management, warehousing efficiency directly impacts the turnover rate of goods and the operating costs of enterprises. With the increasing level of automation, how to scientifically plan warehouse space and achieve accurate warehousing and efficient scheduling of goods has become a key research focus in the field of smart warehousing.

[0003] Intelligent warehouse management is a core component of warehousing operations, but existing technologies have limitations. For example, Chinese patent CN110599067A discloses a warehouse allocation method. This method processes historical goods inbound and outbound data and goods planning information to obtain warehouse storage rules. Based on these rules, the goods information to be received, and warehouse location information, an inbound plan is generated, which then controls the goods handling equipment to execute storage tasks and update the location information. This technical solution improves the flexibility of location allocation by dynamically updating rules. However, this solution still has limitations: its scheduling decisions are mainly based on historical data and preset rules at the business level, lacking a deep understanding of the real-time physical attributes of individual goods to be received, and the warehouse zoning has strong rigidity in the physical dimension. When a specific storage area approaches saturation, the system lacks the ability to dynamically reconstruct the spatial topology, and cannot achieve adaptive expansion of storage capacity while ensuring the continuity of storage space, resulting in limited space utilization and a tendency for fragmented operation paths. Summary of the Invention

[0004] To achieve the above-mentioned objectives and other advantages of the present invention, a first objective of the present invention is to provide an intelligent warehouse management method based on multimodal feature fusion, characterized by comprising the following steps:

[0005] Obtain the status information of the target cargo; the status information includes image features and weight features;

[0006] Multimodal feature fusion processing is performed on the status information to determine the cargo grade that characterizes the quality deterioration of the target cargo;

[0007] Acquire a pre-built digital twin warehouse system for the smart warehouse;

[0008] The storage location mapping table and its three-dimensional topology model in the digital twin warehouse system are invoked so that the three-dimensional topology model can carry the attribute labels and real-time occupancy status of each storage location; wherein, the real-time occupancy status is constructed from the feedback data of historical inbound goods;

[0009] Based on the digital twin warehouse system, the target storage area that matches the target cargo level is retrieved in the three-dimensional topology model, and the existence of vacant storage space in the target storage area is determined based on the real-time occupancy status.

[0010] If no vacant storage space exists, the digital twin warehouse system searches for candidate vacant storage spaces that meet the spatial adjacency conditions in the three-dimensional topology model based on the topology center of the target storage area, and modifies the logical attribute labels of the candidate vacant storage spaces in batches to ensure that the expanded target storage area meets the spatial layout continuity, and then returns to the step of determining whether there is a vacant storage space in the target storage area.

[0011] If an available storage space exists, the automated handling equipment will be controlled to transport the target goods to the available storage space in the target storage area, and a location occupancy command will be sent back to the digital twin warehouse system to update the real-time occupancy status.

[0012] In a preferred embodiment, the step of performing multimodal feature fusion processing on the state information to determine the cargo grade characterizing the quality deterioration of the target cargo specifically includes the following steps:

[0013] Area feature data characterizing surface liquid diffusion is extracted from image features, and weight loss data characterizing material loss is extracted from weight features.

[0014] Input the area feature data and weight loss data into the preset quality correlation model, and calculate the deviation feature value of the area feature data relative to the weight loss data through the quality correlation model;

[0015] The cargo grade, which characterizes the degree of quality deterioration of the target cargo, is determined based on the deviation from the characteristic value.

[0016] In a preferred embodiment, the step of determining the cargo grade characterizing the degree of quality deterioration of the target cargo based on the deviation from the characteristic value specifically includes the following steps:

[0017] Based on the category information of the target goods, retrieve the preset deterioration threshold from the storage location logical mapping table;

[0018] The quality deterioration degree of the target goods is numerically compared with a quality threshold to determine the goods grade that characterizes the quality deterioration degree of the target goods; wherein, the warehouse location logical mapping table includes at least one deterioration threshold; and the goods grade includes at least two grades.

[0019] In a preferred embodiment, the steps of retrieving a target storage area matching the target cargo level in a three-dimensional topology model based on a digital twin warehouse system, and determining whether there are vacant storage spaces in the target storage area based on real-time occupancy status, specifically include the following steps:

[0020] Identify the geometric dimensions of the target cargo and map them into a 3D topology model to generate a virtual entity mirror image that is proportional to the target cargo.

[0021] Using a digital twin warehouse system, access attribute tags corresponding to cargo levels are retrieved from a 3D topology model to determine the target storage area consisting of multiple storage location nodes;

[0022] Retrieve the real-time occupancy status of each storage location node within the target storage area, extract the coordinates of storage location nodes not occupied by historically inbound goods, and construct a candidate set of free coordinates;

[0023] Using a digital twin warehouse system, the virtual entity image is compared with the physical envelope space corresponding to the candidate free coordinate set to determine whether there is a storage location in the target storage area that can accommodate the virtual entity image.

[0024] If the coordinates of the storage location are found through interference comparison, it is determined that there is an available storage location in the target storage area that can be used for the target goods to be stored.

[0025] If the coordinates of the storage location cannot be obtained through interference comparison, it is determined that there is no free storage location in the target storage area.

[0026] In a preferred embodiment, the step of "using a digital twin warehouse system, based on the topological center of the target storage area, retrieving candidate vacant storage locations that meet the spatial adjacency conditions in the three-dimensional topological model, and batch modifying the logical attribute labels of the candidate vacant storage locations to ensure that the expanded target storage area meets the spatial layout continuity, and returning to the step of determining whether there are vacant storage locations in the target storage area" specifically includes the following steps:

[0027] Retrieve the coordinate set of the target storage area in the three-dimensional topology model, identify the storage location nodes at the edge of the coordinate set, and determine the spatial topological outer edge of the target storage area;

[0028] Using a digital twin warehouse system, adjacent storage location nodes that satisfy shared boundary or vertex constraints are retrieved from the outer edge of the spatial topology outwards, and storage location nodes whose real-time occupancy status is unoccupied are identified as candidate vacant storage locations.

[0029] Generate partition reconstruction instructions for candidate vacant storage locations, and batch overwrite the attribute labels of candidate vacant storage locations in the 3D topology model with identifiers that match the current target cargo level;

[0030] Synchronously update the attribute association data in the storage location mapping table, and return to the step of determining whether there is a free storage location in the target storage area.

[0031] In a preferred embodiment, if no idle storage location node is found within a preset neighborhood on the outer edge of the spatial topology, the following steps are further included:

[0032] Using a digital twin warehouse system, identify storage location nodes located on the outer periphery of the spatial topology and occupied by non-target level goods, and designate them as obstacle nodes;

[0033] Simulate the migration of goods from obstructing nodes to other free areas in a 3D topology model, and calculate the energy cost and spatial continuity benefits required to perform the migration task.

[0034] In response to the fact that the benefits of spatial continuity outweigh the energy costs, a cargo handling instruction is generated for the obstructing node, and the automated handling equipment is controlled to clear the obstructing node and convert it into an alternative vacant storage location.

[0035] In a preferred embodiment, if no idle storage location node is found within a preset neighborhood on the outer edge of the spatial topology, the following steps are further included:

[0036] Using a digital twin warehouse system, non-adjacent free storage locations are retrieved in the three-dimensional topology model in ascending order of topological distance, and these are used as candidate free storage locations.

[0037] Modify the attribute labels of the candidate free storage locations and establish a data mapping relationship between the candidate free storage locations and the target storage area;

[0038] The alternative idle storage locations are used as logical overflow shards of the target storage area for unified scheduling, so that the target goods are preferentially allocated to the logical overflow shards when they are put into storage.

[0039] A second objective of this invention is to provide an intelligent warehouse management system based on multimodal feature fusion, comprising:

[0040] The state awareness module is used to acquire the state information of the target cargo; the state information includes image features and weight features.

[0041] The quality grade assessment module is used to perform multimodal feature fusion processing on the status information to determine the cargo grade that characterizes the degree of quality deterioration of the target cargo.

[0042] The digital twin warehouse system is used to call a preset storage location mapping table and the corresponding three-dimensional topology model of the physical warehouse, and to carry the attribute labels and real-time occupancy status of each storage location in the three-dimensional topology model; the real-time occupancy status is constructed from the feedback data of historical inbound goods.

[0043] The spatial retrieval decision module is used to retrieve target storage areas that match the target cargo level in the three-dimensional topology model based on the digital twin warehouse system, and to determine whether there are any available storage locations in the target storage area that can be used for warehousing based on the real-time occupancy status.

[0044] The dynamic topology reconstruction module is used to search for candidate free storage locations that meet the spatial adjacency conditions in the three-dimensional topology model based on the topology center of the target storage area when there are no free storage locations in the target storage area, and to batch modify the attribute labels of the candidate free storage locations in order to realize the dynamic expansion of the target storage area in the virtual space.

[0045] The handling task control module is used to control and drive the automated handling equipment to transport the target goods to the designated storage location when there is an available storage location, and to send the location occupancy instruction back to the digital twin warehouse system to synchronously update the real-time occupancy status.

[0046] A third objective of this invention is to provide an electronic device, including a memory and a processor; wherein the memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement an intelligent warehouse management method based on multimodal feature fusion.

[0047] A fourth objective of this invention is to provide a computer storage medium storing computer instructions thereon; wherein, when the computer instructions are executed by a processor, they implement an intelligent warehouse management method based on multimodal feature fusion.

[0048] Compared with the prior art, the beneficial effects of this application are:

[0049] First, this application acquires image and weight features through a state-aware module and uses a quality correlation model to calculate deviation feature values ​​to determine the cargo grade. This mechanism establishes a logical correlation between visual changes in images and physical quality loss, eliminating the judgment errors that may arise from single-dimensional sensor data and providing a definite decision-making basis for differentiated storage.

[0050] Secondly, this application utilizes a dynamic topology reconstruction module to retrieve alternative storage locations that meet the adjacency conditions in the 3D topology model based on the topology center when the target storage area is saturated, and then overwrites the attribute tags. This processing logic breaks the rigid partitioning limitations of physical shelving, enabling dynamic expansion of the storage area in virtual space and resolving the contradiction between local storage overflow and local space redundancy caused by fixed partitions.

[0051] Furthermore, this application adheres to the principle of spatial layout continuity during the expansion process, ensuring that the newly added storage locations remain connected to the original areas in three-dimensional space. This topology-based reconstruction mechanism reduces path fragmentation caused by automated handling equipment operating between discrete storage locations, lowers idle running losses during cross-regional scheduling, and improves the overall operational efficiency of the warehousing system. Attached Figure Description

[0052] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0053] Figure 1 A flowchart illustrating an intelligent warehouse management method based on multimodal feature fusion, as provided in Example 1;

[0054] Figure 2 This is a flowchart of the steps in Example 1 to perform multimodal feature fusion processing on state information to determine the cargo grade that characterizes the quality deterioration of the target cargo.

[0055] Figure 3 This is a flowchart illustrating the steps in Example 1 to call the storage location mapping table and its three-dimensional topology model in the digital twin warehouse system so that the three-dimensional topology model can carry the attribute labels and real-time occupancy status of each storage location.

[0056] Figure 4 This is a schematic diagram of the virtual warehouse space and three-dimensional topology model provided in Example 1;

[0057] Figure 5 In Example 1, the digital twin warehouse system retrieves candidate vacant storage locations that meet spatial adjacency conditions in a three-dimensional topology model based on the topology center of the target storage area, and modifies the logical attribute labels of the candidate vacant storage locations in batches so that the expanded target storage area meets the spatial layout continuity, and returns a flowchart of the step of determining whether there are vacant storage locations in the target storage area.

[0058] Figure 6 This is a topological logic diagram of the dynamic expansion process of the target storage area provided in Example 1;

[0059] Figure 7 This is a structural block diagram of the intelligent warehouse management system based on multimodal feature fusion provided in Example 2;

[0060] Figure 8 A structural block diagram of an electronic device is provided according to Embodiment 3 of the present invention;

[0061] Figure 9 This is a schematic diagram of a computer storage medium provided according to Embodiment 4 of the present invention. Detailed Implementation

[0062] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0063] It should be noted that, without conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0064] It should be noted that the terms "first," "second," etc., in the specification, claims, and drawings of this invention are used to distinguish different objects, rather than to limit a specific order.

[0065] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0066] Example 1

[0067] According to a first aspect of the embodiments of this application, an intelligent warehouse management method based on multimodal feature fusion is provided, such as... Figure 1 As shown, it includes the following steps:

[0068] S1. Obtain the status information of the target cargo; whereby the status information includes image features and weight features;

[0069] S2. Perform multimodal feature fusion processing on the status information to determine the cargo grade that characterizes the quality deterioration of the target cargo;

[0070] S3. Obtain the pre-established digital twin warehouse system of the smart warehouse;

[0071] S4. Call the storage location mapping table and its three-dimensional topology model in the digital twin warehouse system so that the three-dimensional topology model can carry the attribute labels and real-time occupancy status of each storage location; wherein, the real-time occupancy status is constructed from the feedback data of historical inbound goods;

[0072] S5. Based on the digital twin warehouse system, retrieve the target storage area that matches the target cargo level in the three-dimensional topology model, and determine whether there is a free storage space in the target storage area based on the real-time occupancy status; if there is no free storage space, proceed to step S6; if there is a free storage space, proceed to step S7.

[0073] S6. Using the digital twin warehouse system, based on the topology center of the target storage area, search for candidate empty storage locations that meet the spatial adjacency conditions in the three-dimensional topology model, and batch modify the logical attribute labels of the candidate empty storage locations so that the expanded target storage area meets the spatial layout continuity, and return to the step of determining whether there are empty storage locations in the target storage area.

[0074] S7 controls and drives the automated handling equipment to transport the target goods to an available storage location in the target storage area, and sends a location occupancy command back to the digital twin warehouse system to update the real-time occupancy status.

[0075] It should be noted that the aforementioned target goods refer to physical items awaiting storage in the smart warehouse. The status information reflects the physical attributes and external characteristics of the goods. Specifically, image features can be acquired through visual sensors deployed in the warehouse's receiving area to capture visual changes in the surface of the goods; weight features are acquired through weighing sensors to detect subtle changes in the quality of the goods. It should be noted that this dual-dimensional data of image and weight provides a primary reference for accurately determining the degree of subsequent quality deterioration.

[0076] Optionally, in step S1 above, obtaining the status information of the target cargo specifically involves using a sensing terminal to collect the physical parameters of the target cargo.

[0077] Specifically, the system utilizes industrial cameras positioned around the perimeter of the inbound conveyor to capture multi-view visual data of the target goods in order to extract image features. It should be noted that these image features include the surface contour, texture distribution, and geometric dimensions of the target goods. Furthermore, the system uses weighing sensors located at the bottom of the inbound conveyor line to collect real-time load data of the target goods in order to extract weight features. It should be noted that these weight features reflect the immediate mass of the target goods.

[0078] Optionally, the system ensures that image features and weight features are acquired at the same time reference by synchronous trigger pulses, and encapsulates the acquired raw data into a status information message with a unique cargo identifier and timestamp. Furthermore, the status information is transmitted in real time to the digital twin warehouse system via an industrial communication network. It should be noted that the placement of the aforementioned industrial cameras and weighing sensors should be based on the ability to completely cover the visible surface and supporting stress points of the target cargo, aiming to provide accurate multi-dimensional feature data for subsequent quality grading.

[0079] It should be noted that the steps for acquiring status information are not limited to the combination of cameras and pressure sensors mentioned above. Depending on the material and physicochemical properties of the target cargo, the status information may optionally include temperature distribution characteristics obtained through an infrared thermal imager, or odor volatility concentration characteristics obtained through a chemical sensor. Specifically, these supplementary features, together with image features and weight features, constitute a comprehensive status profile of the target cargo. Furthermore, after acquiring the status information, the system automatically compares it with historical records. If it detects that previous status data already exists for the cargo's serial number, it will associate the newly acquired status information as incremental data to support subsequent dynamic assessment of quality deterioration trends.

[0080] In step S2 above, multimodal feature fusion processing refers to the computational logic of correlating heterogeneous data from different sensors. Specifically, the system uses the collected image features and weight features as input parameters and calculates the cargo grade representing the quality degradation degree of the target cargo using a preset correlation model. It should be noted that quality degradation degree is a quantitative indicator used to measure the degree of quality decline of cargo compared to its standard state. The cargo grade is divided according to the numerical range of degradation degree, such as into Class I, Class II, etc., with different grades corresponding to different storage environment requirements. Furthermore, this judgment logic eliminates potential misjudgments caused by a single feature, improving the accuracy of the classification.

[0081] In step S5, the target storage area refers to the set of storage locations selected in the 3D topology model whose attribute labels match the cargo level. Specifically, the system first locks the corresponding functional partition based on the cargo level, and then retrieves the real-time occupancy status of each storage location within that partition to identify whether there are any unmarked vacant storage locations.

[0082] Regarding step S6, when there are no available empty storage spaces in the preset target storage area, the system will initiate dynamic reconstruction processing logic. It should be noted that the topology center is a reference point determined by geometric calculation of the coordinates of existing nodes within the target storage area. Specifically, using this center as a reference, the system detects physically adjacent storage space nodes that are currently idle in the 3D topology model. Furthermore, by batch modifying the logical attribute labels of these candidate storage spaces (e.g., overwriting the "General Area" label with "Specific Grade Goods Area"), the storage area is dynamically expanded. It should be noted that the expansion process follows the principle of spatial layout continuity, ensuring that the newly added storage spaces remain spatially connected to the original area, thereby optimizing the operation paths of subsequent automated handling equipment.

[0083] In step S7, the automated handling equipment includes, but is not limited to, automated guided vehicles (AGVs) or stacker cranes. Specifically, the system generates handling control instructions and sends them to the automated handling equipment, instructing it to deliver the target goods to the designated vacant storage location according to the planned path. It should be noted that after the goods are placed, the handling equipment sends a location occupancy instruction back to the digital twin warehouse system. Furthermore, the digital twin warehouse system updates the real-time occupancy status of the corresponding coordinates in the 3D topology model based on this instruction, ensuring the spatiotemporal consistency between the virtual environment and the physical entity.

[0084] It should be noted that the method provided in this embodiment combines cargo quality grading with dynamic reconstruction of storage space using digital twin technology, solving the problems of rigid zoning and low space utilization in traditional warehouses. The above process runs automatically under the control of a computer system, requiring no manual intervention, thus realizing the intelligent transformation of warehouse management.

[0085] In an optional embodiment, step S2 involves performing multimodal feature fusion processing on the state information to determine the cargo grade characterizing the quality deterioration of the target cargo, as follows: Figure 2 As shown, the specific steps include:

[0086] S21. Extract area feature data representing surface liquid diffusion from image features, and extract weight loss data representing material loss from weight features.

[0087] S24. Input the area feature data and weight loss data into the preset quality correlation model, and calculate the deviation feature value of the area feature data relative to the weight loss data through the quality correlation model.

[0088] S25. Determine the cargo grade that characterizes the degree of quality deterioration of the target cargo based on the deviation characteristic value.

[0089] In step S21 above, specifically, this embodiment takes a cold chain logistics warehousing scenario as an example, where the target goods are frozen fresh products or liquid chemical raw materials. In a cold chain environment, due to fluctuations in ambient temperature or failure of packaging sealing, the target goods may thaw and leak liquid or their contents may leak, resulting in the diffusion of liquid substances and a decrease in weight. Specifically, the system utilizes the image features acquired by the aforementioned industrial camera, and through a preset color threshold segmentation algorithm or edge extraction algorithm, identifies liquid seepage areas on the outer packaging surface or the carrying pallet of the target goods. The system counts the number of pixels occupied by the liquid seepage areas and, combined with the camera's calibration parameters, converts the pixel area into an actual physical area, thereby obtaining area feature data characterizing surface liquid diffusion. It should be noted that, in a cold chain scenario, this area feature data... This reflects the extent of the external damage to the goods.

[0090] Meanwhile, the system uses the aforementioned pressure sensor to obtain the real-time mass value of the target cargo. And retrieve the standard initial weight of the batch of goods from the location logical mapping table of the digital twin warehouse system. Specifically, the system calculates the standard initial weight. With the real-time quality value The absolute value of the difference between them yields the weight loss data characterizing the material loss. The specific calculation formula is as follows:

[0091]

[0092] It should be noted that the weight reduction data mentioned The physical mass loss of goods due to thawing loss or evaporation was quantified. Specifically, area characteristic data... With weight loss data They remain synchronized over time, forming a heterogeneous feature data pair for subsequent evaluation of quality degradation.

[0093] In step S24 above, specifically, the area feature data extracted above is... With weight loss data Input into the preset quality correlation model.

[0094] It should be noted that in cold chain scenarios, the leakage and weight loss vary in different proportions for different types of goods (such as frozen meat and chemical reagents). Therefore, the quality correlation model includes preset correlation weight coefficients to balance the dimensional differences between visual pixel data and physical quality data. Specifically, the system calculates the area feature data by performing weighted subtraction operations. Relative to the weight reduction data deviation eigenvalues The specific calculation formula is as follows:

[0095]

[0096] It should be noted that, in order to ensure the equality of the formulas and the dimensionality, the aforementioned correlation weight coefficients... Defined as a unit of measurement having the ratio of area to weight (e.g., ), used to reduce weight loss data in the physical mass dimension The mapping is transformed into area feature data in the visual pixel dimension. Equivalent values ​​with the same dimensions cause the calculated deviation from the eigenvalue. The numerical deviation under a unified dimension (or the dimensionless value after normalization).

[0097] It should be noted that the deviation from the eigenvalue This is used to characterize the discrepancy between the external visual appearance of goods and internal quality loss. Specifically, a significant increase in the deviation characteristic value indicates that the goods have experienced severe deterioration beyond the normal thawing range or serious external leakage interference.

[0098] Furthermore, step S25, which determines the cargo grade characterizing the degree of quality deterioration of the target cargo based on the deviation eigenvalue, specifically includes the following steps:

[0099] S251. Based on the category information of the target goods, retrieve the preset deterioration threshold from the storage location logical mapping table;

[0100] S252. Compare the quality deterioration degree of the target goods with the quality threshold to determine the goods grade that characterizes the quality deterioration degree of the target goods; wherein, the warehouse location logical mapping table includes at least one deterioration threshold; and the goods grade includes at least two grades.

[0101] In step S251 above, specifically, the system obtains the category information of the target goods by parsing the identification carrier of the target goods, and uses this category information as an index to retrieve the corresponding storage location logical mapping table in the database of the digital twin warehouse system. It should be noted that the storage location logical mapping table stores a set of deterioration thresholds corresponding to different physicochemical properties of goods categories. Specifically, the set of deterioration thresholds includes at least a first quality threshold. .

[0102] In step S252 above, specifically, the system uses the aforementioned deviation eigenvalues. Defined as the degree of quality degradation characterizing the target goods. The system compares this quality degradation degree with a retrieved first quality threshold. Perform a numerical comparison. Optionally, if the condition is met... If the condition is met, then the target goods are classified as Class 1; If so, the cargo is classified as Class II.

[0103] It should be noted that setting multiple cargo levels is used to address the uneven utilization of storage space and spatial fragmentation issues caused by fixed storage areas in automated warehouses. Specifically, cargo level serves as a selection criterion for matching target cargo with target storage areas in a three-dimensional topology model. Since the demand for inbound cargo of different qualities varies across different operational cycles, setting multiple levels allows the digital twin warehouse system to define the spatial distribution boundaries of cargo of different qualities.

[0104] It should be noted that when the target storage area corresponding to a specific grade of goods is saturated, the system triggers the execution of the subsequent neighbor detection step S6 in the three-dimensional topology model based on the goods grade. Specifically, by batch modifying the attribute labels of the storage location nodes adjacent to the outer edge of the target storage area to be consistent with the currently determined goods grade, the expanded area maintains spatial continuity and allows goods of that specific grade to be stored. It should be noted that this dynamic reconstruction mechanism based on multi-level quality classification eliminates the limitations of physical partitioning on storage capacity allocation, resolves the contradiction between local storage overflow and local space redundancy caused by rigid partitioning, and thus optimizes the access efficiency of automated handling equipment. This embodiment introduces multi-level judgment based on deviation feature values ​​in sub-steps S251 and S252, providing a quantitative basis for the attribute reconstruction of the three-dimensional topology model. It should be noted that the above threshold retrieval and comparison logic is automatically executed by the processor of the electronic device, ensuring the determinism of the grade classification. Specifically, the deep coupling of goods grade and attribute labels realizes the adaptive and flexible management of storage space in the intelligent warehouse.

[0105] In an optional embodiment, the status information further includes transportation environment monitoring data. Step S2, which involves performing multimodal feature fusion processing on the status information to determine the cargo grade characterizing the deterioration degree of the target cargo quality, also includes the following steps:

[0106] S22. Based on the transportation environment monitoring data, extract the heat accumulation parameters of the target cargo during the transportation process;

[0107] S23. Based on the thermal accumulation parameter, the correlation weights in the quality correlation model are offset and corrected to eliminate the surface condensation interference characteristics of the target goods caused by the environmental temperature difference.

[0108] It should be noted that steps S22 and S23 are preferably executed before step S24, which calculates the deviation eigenvalues. Specifically, steps S22 and S23 can be executed in parallel with the aforementioned step S21, or after step S21 and before step S24, so that the quality correlation model called in step S24 can pre-complete parameter compensation for environmental factors.

[0109] In step S22 above, specifically, the system retrieves transportation environment monitoring data recorded by environmental sensors during the transportation of the target goods. It should be noted that the transportation environment monitoring data includes time-series values ​​of the ambient temperature changing over time during the transportation period. Specifically, the system performs cumulative calculations on the ambient temperature to extract a heat accumulation parameter characterizing the total amount of heat absorbed. Specifically, the calculation formula is as follows:

[0110]

[0111] in, For the first The ambient temperature value for each sampling period This refers to the standard storage temperature value for this category of goods. is the sampling time step, i is the sampling sequence number, and m is the total number of samples.

[0112] It should be noted that the thermal accumulation parameter The thermal effects of the target cargo due to deviation from the standard temperature control environment are quantified. This parameter is used to determine whether the surface of the target cargo meets the temperature difference conditions for the generation of physical condensation.

[0113] In step S23 above, specifically, the system assigns correlation weights in the quality correlation model based on the extracted heat accumulation parameters. Offset correction is performed. It should be noted that when condensation occurs on the surface of the target cargo due to environmental temperature differences, the resulting water mist or droplets will be captured by the industrial camera and included in the image features, causing the area feature data extracted in step S21 to be affected. This introduces deviations not caused by deterioration factors. Specifically, the system generates corrected correlation weights by calculating environmental compensation factors. Specifically, the calculation formula is as follows:

[0114]

[0115] in, This is the preset environmental compensation coefficient. It should be noted that the corrected correlation weights... This is used to adjust the proportion of weight loss data in the correlation model to offset condensation noise in the area feature data.

[0116] It should be noted that, in order to ensure the consistency of dimensions for each term within the parentheses in the formula, the above environmental compensation coefficient... The units are preset to the above-mentioned heat accumulation parameters. The reciprocal of the unit (e.g.) Specifically, through and The multiplication operation eliminates the time and temperature dimensions of the heat accumulation parameter, making it possible to... The entire factor becomes a dimensionless correction factor, thus ensuring the corrected correlation weights. Compared with the initial association weight They are completely equivalent in dimensions.

[0117] In step S24 above, specifically, the system calls the corrected association weights generated in step S23. And combined with the area feature data extracted in step S21 and weight loss data The deviation eigenvalue is calculated. Specifically, the logic for calculating the deviation between the gray values ​​in the gray value sequence and the preset standard value of the detection area can refer to the calculation method of gray value deviation in metal plate defect detection. In this embodiment, the calculation formula is as follows:

[0118]

[0119] It should be noted that by introducing a thermal accumulation parameter to correct the weights, interference components caused by the environment in the image features can be eliminated. Specifically, in step S25 above, the system calculates the deviation feature values... As a basis for judgment, the value is compared with a preset quality threshold to determine the cargo grade. Specifically, by performing positional or logical layering analysis on multi-directional features, it is ensured that the final determined cargo grade reflects the true quality status of the target cargo. It should be noted that the above processing flow is automatically executed by the processor of the electronic device, ensuring the accuracy of cargo quality determination in complex transportation environments.

[0120] In an optional embodiment, step S4 involves calling the storage location mapping table and its three-dimensional topology model in the digital twin warehouse system to enable the three-dimensional topology model to carry the attribute labels and real-time occupancy status of each storage location, such as... Figure 3 As shown, the specific steps include:

[0121] S41. Call the three-dimensional topology model to construct a virtual warehouse space, and map the spatial coordinates of the physical warehouse locations to the virtual warehouse space based on the warehouse location mapping table to generate virtual warehouse location nodes with topological relationships;

[0122] S42. Parse the logical rules in the storage location mapping table, and use the corresponding deterioration access level, physical isolation requirements and outbound priority as attribute tags to dynamically associate them with the corresponding virtual storage location nodes in the virtual warehouse space.

[0123] S43. Retrieve the historical inbound location data of goods and perform occupancy marking processing on the coordinate area corresponding to the inbound data in the three-dimensional topology model to form a real-time occupancy status that represents the current availability of the storage location.

[0124] In step S41 above, specifically, the system constructs the virtual warehouse space by retrieving a preset 3D mesh model. This virtual warehouse space is a proportional reconstruction of the physical warehouse environment in digital space. It should be noted that the system, based on the physical index in the storage location mapping table, uses a spatial coordinate transformation algorithm to map the absolute or relative coordinates of the physical storage locations to the coordinate system of the virtual warehouse space. Specifically, the virtual storage location nodes generated during the mapping process not only contain three-dimensional spatial coordinates but also topological data characterizing the connectivity and proximity between nodes. It should be noted that this topological association provides underlying data support for subsequent path planning and spatial neighborhood detection.

[0125] Specifically, such as Figure 4 As shown in (a), the processor of the digital twin warehouse system reconstructs and generates a virtual warehouse space 200 in the digital space by retrieving a preset 3D mesh model. It should be noted that the virtual warehouse space 200 constitutes a proportionally mapped entity of the physical warehouse in the digital dimension, serving as the basic support environment (or execution carrier) for all subsequent simulation scheduling tasks. Specifically, the processor reads the storage location mapping table to obtain the spatial location parameters of the physical storage locations. Within the virtual warehouse space 200, the system delineates a target storage area 201 according to management logic.

[0126] To visually demonstrate the detailed distribution of storage locations within target storage area 201, Figure 4 (b) provides a view of a magnified portion of region a201 within target storage region 201.

[0127] The processor reads the storage location mapping table, obtains the three-dimensional coordinates of the center point of the physical shelf compartment, and maps these physical coordinates to the corresponding positions within the target storage area 201 by executing a spatial coordinate transformation algorithm, thereby generating multiple virtual storage location nodes 202 with spatial coordinate vectors. It should be noted that the virtual storage location nodes 202 are distributed throughout the entire target storage area 201; the enlarged area a201 is only used as an example to illustrate the distribution and mapping logic of the nodes in the three-dimensional topology model.

[0128] Virtual storage location node 202 encapsulates topological association data in its data structure, representing the adjacency relationships and connectivity weights between nodes. This topological association data defines the physical accessibility and topological length between storage location nodes by recording the adjacency matrix or logical edge connection information between each node. This established topological association provides underlying data support for the subsequent execution of the transport path planning algorithm and for retrieving alternative empty storage locations that meet the spatial adjacency conditions based on the topological center during step S6.

[0129] In step S42 above, specifically, the system uses a preset logic parser to read the structured description fields in the storage location mapping table and extracts the logical rules regarding storage location management. Further, the system encapsulates the extracted management parameters, such as the degradation access level, the physical isolation requirements, and the outbound priority, into attribute tags, and uses a data association operator to dynamically bind the attribute tags to the corresponding virtual storage location nodes.

[0130] It should be noted that the aforementioned deterioration access level corresponds to the cargo level determined in step S2 above, and is used to limit the storage range of goods of specific quality. Specifically, physical isolation requirements are used to identify prohibited areas around dangerous or sensitive goods, and outbound priority is used to optimize inventory turnover strategies. Through this dynamic association, the three-dimensional topology model is transformed from a simple geometric model into a functional model with management attributes.

[0131] In step S43 above, specifically, the system accesses the operational history database of the digital twin warehouse system and retrieves operational feedback data transmitted back by automated handling equipment or warehouse sensor nodes. It should be noted that the transmitted data records the actual placement coordinates of historically received goods in the physical warehouse. Specifically, the system searches for virtual storage location nodes in the three-dimensional topology model that match the coordinates of the transmitted data and changes the occupancy status of these nodes from idle to occupied. Further, through statistical processing of all active transmitted data, the system forms a distribution map representing the real-time availability of physical storage locations in the virtual space, i.e., the real-time occupancy status. It should be noted that this method of marking occupancy based on location transmitted data effectively avoids inconsistencies between virtual and real states caused by communication delays or system crashes, ensuring the reliability of spatial retrieval and decision-making.

[0132] In an optional embodiment, step S5, based on the digital twin warehouse system, involves retrieving a target storage area matching the target cargo level in a three-dimensional topology model and determining whether there are any vacant storage spaces in the target storage area based on the real-time occupancy status. This step specifically includes the following steps:

[0133] S51. Identify the geometric dimensions of the target cargo and map them into a three-dimensional topology model to generate a virtual entity mirror image that is proportional to the target cargo.

[0134] S52. Using a digital twin warehouse system, retrieve the access attribute tags corresponding to the cargo level in the three-dimensional topology model to determine the target storage area composed of multiple storage location nodes;

[0135] S53. Retrieve the real-time occupancy status of each storage location node within the target storage area, extract the coordinates of storage location nodes not occupied by historically inbound goods, and construct a candidate set of free coordinates.

[0136] S54. Using a digital twin warehouse system, perform an interference comparison between the virtual entity image and the physical envelope space corresponding to the candidate free coordinate set to determine whether there is a storage location in the target storage area that can accommodate the virtual entity image.

[0137] If the coordinates of the storage location are found through interference comparison, it is determined that there is an available storage location in the target storage area that can be used for the target goods to be stored.

[0138] If the coordinates of the storage location cannot be obtained through interference comparison, it is determined that there is no free storage location in the target storage area.

[0139] In step S51 above, specifically, the system uses the aforementioned sensing terminal to acquire the geometric dimension data of the target cargo, including the length of the cargo. ,width and height The system uses these geometric dimensions as parameters to map and generate a proportionally scaled 3D envelope in the 3D topological model, serving as a virtual entity mirror. It should be noted that this virtual entity mirror is used to simulate the actual physical volume occupied by the target goods within the virtual warehouse space.

[0140] In step S52 above, specifically, the system utilizes a digital twin warehouse system to retrieve the cargo level determined in the preceding steps and searches the three-dimensional topology model for access attribute tags that have a preset mapping relationship with the cargo level. The system logically categorizes all virtual storage location nodes carrying the access attribute tags to determine the target storage area composed of these virtual storage location nodes. It should be noted that this target storage area defines the access space boundary that conforms to the current cargo quality management rules.

[0141] In step S53 above, specifically, the system retrieves the real-time occupancy status of each storage location node within the target storage area. The system identifies and filters nodes marked as occupied, extracting the coordinates of free nodes not occupied by historically inbound goods. Further, the system aggregates these free node coordinates to construct a candidate free coordinate set. It should be noted that this candidate free coordinate set provides a preliminary distribution of potential locations in the physical space that meet the inbound conditions in terms of occupancy.

[0142] In step S54 above, specifically, the system utilizes the spatial topology analysis function of the digital twin warehouse system to perform an interference comparison between the virtual entity image and the physical envelope space corresponding to the candidate idle coordinate set. Specifically, the system sequentially places the virtual entity image at each storage location coordinate in the candidate idle coordinate set and detects whether there is spatial overlap between the virtual entity image and the surrounding shelf supports, adjacent goods images, or auxiliary equipment images at that storage location coordinate. The system calculates the volume value of the overlapping area. To determine the interference situation. It should be noted that if the determination meets the conditions... If the physical envelope space corresponding to the storage location coordinates is sufficient to accommodate the virtual entity image, then the system determines that there is a free storage location in the target storage area that can accommodate the target goods. Furthermore, if there is a storage location coordinate that passes the interference comparison and has a volume value of zero, it is determined that there is a free storage location in the target storage area for the target goods to be stored; if there is no storage location coordinate that passes the interference comparison, it is determined that there is no free storage location in the target storage area. It should be noted that the above-mentioned interference comparison process using the virtual entity image ensures that, in complex physical environments, the system can pre-identify storage obstacles caused by mismatched goods dimensions or insufficient storage location clearance height. Specifically, when it is determined that there is no free storage location in the target storage area, the system will automatically trigger subsequent dynamic reconstruction logic for the storage area.

[0143] In an optional embodiment, step S6, "using a digital twin warehouse system, based on the topological center of the target storage area, retrieves candidate vacant storage locations that meet the spatial adjacency conditions in the three-dimensional topological model, and batch modifies the logical attribute labels of the candidate vacant storage locations to ensure that the expanded target storage area meets the spatial layout continuity, and returns to the step of determining whether there are vacant storage locations in the target storage area," is as follows: Figure 5 As shown, the specific steps include:

[0144] S61. Retrieve the coordinate set of the target storage area in the three-dimensional topology model, identify the storage location nodes at the edge of the coordinate set, and determine the spatial topology outer edge of the target storage area.

[0145] Specifically, the processor retrieves the set of virtual storage location node coordinates corresponding to the target storage area 101, which is currently in a saturated state, from the spatial database of the digital twin warehouse system, denoted as... It should be noted that the virtual storage location nodes have a preset adjacency relationship in the three-dimensional topological mesh. Specifically, the processor traverses the coordinate set, identifies nodes with a connectivity degree less than a preset full connectivity threshold, and defines them as edge nodes. Further, as... Figure 6 As shown in (a), the red closed frame formed by the edge nodes in the three-dimensional spatial coordinate system is identified as the spatial topological outer edge 102. It should be noted that by identifying the spatial topological outer edge 102, the system establishes the permeable boundary between the current logical partition and the surrounding physical free space at the digital level. This step utilizes topological connectivity analysis to transform the discrete layout of the physical shelving into continuous geometric features, providing a data foundation for subsequently breaking down rigid physical partitions.

[0146] S62. Using a digital twin warehouse system, search for adjacent storage location nodes that satisfy shared boundary or vertex constraints from the outer edge of the spatial topology outwards, and determine the storage location nodes whose real-time occupancy status is unoccupied as candidate vacant storage locations.

[0147] Specifically, the processor uses the coordinates of the edge nodes on the outer edge 102 of the spatial topology as the starting point and performs a scan search outward using a probe vector. The search range covers the adjacent shelves, the shelves behind, and the shelves opposite the current shelf. It should be noted that to ensure the expanded storage area maintains strict physical continuity, the processor uses Euclidean distance constraints to limit spatial adjacency conditions. Specifically, the magnitude of the probe vector... The following formula should be satisfied:

[0148]

[0149] in, , , These are the preset mesh step sizes for the three-dimensional topological model in the horizontal, depth, and height dimensions, respectively. At the physical level, such as... Figure 6 As shown in (a), the step lengths correspond to the width, depth, and height dimensions of the physical shelf compartments. Specifically, the storage location nodes that satisfy this formula are the direct neighbors of the original area in three-dimensional geometric logic, sharing faces, edges, or vertices. The processor retrieves real-time sensor feedback data from these nodes, extracts the occupancy status bits, and identifies nodes not occupied by goods as candidate vacant storage locations 103. It should be noted that by introducing a quantized distance constraint based on the physical step length, the system transforms fuzzy spatial adjacency relationships into deterministic mathematical judgment logic, thereby ensuring that the expanded area remains continuous in physical space and directly solving the problem of fragmented operation paths of automated handling equipment caused by partitioning.

[0150] S63. Generate partition reconstruction instructions for the candidate vacant storage locations, and batch overwrite the attribute tags of the candidate vacant storage locations in the three-dimensional topology model with identifiers that match the current target cargo level.

[0151] Specifically, the processor of the digital twin warehouse system generates partition reconstruction control instructions based on the unique hardware address of the identified candidate idle storage location 103. It should be noted that these instructions dynamically overwrite attribute tags by performing write operations on the logical attribute fields of the corresponding nodes in the database. Specifically, the admission level tag of the candidate idle storage location 103 is modified from a general idle state to a specific level identifier consistent with the target storage area 101. For example... Figure 6As shown in Figure (b), after executing S63, the alternative idle storage location 103, which was originally outside the original red closed frame, is logically incorporated into the target storage area 101, causing the outer edge 102 of the spatial topology to dynamically shift, forming a closed boundary after expansion. This logic-based overwrite mechanism achieves flexible reconstruction of physical storage space without changing the physical hardware structure, improving the adaptive scheduling capability of storage resources in response to sudden logistics peaks.

[0152] S64. Synchronously update the attribute association data in the storage location mapping table, and return to step S5 to determine whether there is a free storage location in the target storage area.

[0153] Specifically, the processor synchronizes the updated virtual node attribute data to the underlying physical storage location management mapping table in real time, ensuring strong consistency between the digital twin environment and the physical warehouse management system. It should be noted that after completing the underlying data synchronization, the system automatically triggers a return to step S5. Since the target storage area has undergone dynamic expansion based on mathematical neighborhood definitions, the system can assign entry coordinates that meet physical continuity requirements to the target goods when performing spatial interference comparison again. Specifically, this synchronization mechanism eliminates management barriers between physical partitions through digital modeling, significantly enhancing the comprehensive utilization rate of limited space in the automated warehousing system.

[0154] In an optional embodiment, for the aforementioned step S62, if no idle storage location node is found within the preset neighborhood range around the outer edge 102 of the spatial topology, the following step is further included:

[0155] S6211. Using a digital twin warehouse system, identify storage location nodes located on the outer perimeter of the spatial topology and occupied by non-target level goods, and use them as obstacle nodes.

[0156] Specifically, the processing module accesses the real-time attribute database of the digital twin warehouse system and polls the status of storage location nodes adjacent to the outer edge of the spatial topology 102. It should be noted that the processing module identifies these storage location nodes as occupied, and that the cargo level tags they carry are inconsistent with the target cargo level to be stored. Specifically, the processor marks these storage location nodes that physically block the outward penetration of the target storage area 101 as obstruction nodes in the three-dimensional topology model. Through this identification mechanism, the system can locate physical obstruction points that cause spatial layout discretization, providing operational targets for subsequent storage optimization.

[0157] S6212. Simulate the migration of goods from obstructing nodes to other free areas in a three-dimensional topology model, and calculate the energy cost and spatial continuity benefits required to perform the migration task.

[0158] Specifically, the digital twin warehouse system performs goods transfer simulations in a virtual simulation environment. The processor retrieves the current coordinates of the obstructing node and the coordinates of the preset alternative storage areas, calculates the estimated power consumption required for the automated handling equipment to execute the migration path, and interprets this as energy cost. Simultaneously, the processor calculates the corresponding spatial continuity gain based on the percentage increase in connectivity of the target storage region 101 in the three-dimensional topology after the obstructing nodes are cleared. It should be noted that the aforementioned spatial continuity benefits quantify the value of automated handling equipment path optimization and management efficiency improvement resulting from centralized cargo storage. Specifically, the processor provides data support for the migration decisions of physical entities by establishing a quantitative evaluation model of energy consumption and benefits.

[0159] S6213. In response to the fact that the benefits of spatial continuity outweigh the energy costs, a cargo handling instruction is generated for the obstructing node, and the automated handling equipment is controlled to clear the obstructing node and convert it into an alternative vacant storage location.

[0160] Specifically, in response to satisfying Based on the judgment conditions, the logic control layer of the digital twin warehouse system sends hardware control messages to the execution unit of the automated handling equipment. According to the coordinate information in the instructions, the automated handling equipment physically moves non-target level goods from the obstructing node to other designated storage locations. It should be noted that after the goods are physically removed, the processing module synchronously resets the occupancy status bit of the node in the 3D topology model to unoccupied status through sensor feedback data, thereby converting the original obstructing node into a candidate free storage location 103. Specifically, this step, through the dynamic rearrangement of physical entities, eliminates spatial barriers caused by improper historical storage, ensuring that the target storage area 101 can achieve continuous physical expansion, fundamentally solving the space utilization bottleneck problem in automated warehousing environments.

[0161] This embodiment achieves self-healing management of physical warehouse space through the automated scheduling logic of S6211 to S6213 described above, effectively suppressing the impact of space fragmentation on the efficiency of automated operations.

[0162] In an optional embodiment, for the aforementioned step S62, if no idle storage location node is found within a preset neighborhood on the outer edge of the spatial topology, the following step is further included:

[0163] S6221. Using a digital twin warehouse system, search for non-adjacent free storage locations in the three-dimensional topology model in ascending order of topological distance, and use them as candidate free storage locations.

[0164] Specifically, when the processor determines that all physically adjacent nodes around the outer edge of the spatial topology 102 are either occupied or locked by transport commands, the digital twin warehouse system initiates a global topology search program. Specifically, the processor traverses all candidate storage location nodes in the 3D topology model except for the target storage area 101, and calculates the topological distance between each available storage location node to be detected and the geometric center of the target storage area 101. Optionally, the topological distance The calculation formula is as follows:

[0165]

[0166] in, The spatial coordinates of the idle storage location node to be detected. The reference center coordinates are for the target storage area 101. Specifically, the processor sorts the calculated topological distances in ascending order of value and extracts a preset number of storage location nodes at the top of the sorting as candidate idle storage locations. It should be noted that this step, through quantified distance weighting, ensures that, in the absence of physical continuous expansion, the idle resources with the highest logical correlation can be selected, thereby reducing the path length of subsequent automated handling equipment between non-adjacent storage locations.

[0167] S6222. Modify the attribute labels of the candidate free storage locations and establish a data mapping association between the candidate free storage locations and the target storage area.

[0168] Specifically, the logical control layer of the digital twin warehouse system issues an attribute change instruction to the database based on the candidate idle storage location identifiers retrieved in step S6221. It should be noted that the attribute change instruction batch overwrites the logical attribute labels of the candidate idle storage locations from "idle" or "general" to identifiers matching the current target goods level. Specifically, the processor establishes a one-to-many data mapping index in the storage location mapping table, and mounts the hardware IDs of the candidate idle storage locations as overflow fragments under the logical master node of the target storage area 101. It should be noted that by establishing the data mapping association, the system achieves semantic alignment between non-adjacent spaces and original partitions at the digital level, eliminating management gaps caused by physical space discontinuity.

[0169] S6223. The alternative idle storage locations are used as logical overflow fragments of the target storage area for unified scheduling, so that the target goods are preferentially allocated to the logical overflow fragments when they enter the warehouse.

[0170] Specifically, the scheduling module of the digital twin warehouse system defines the established associated candidate idle storage locations as logical overflow fragments. It should be noted that when a target item triggers an inbound request, the scheduling module, by retrieving the mapping association, prioritizes directing the item's placement to the logical overflow fragment, rather than searching for a new random storage area. Specifically, the processor, through a unified scheduling instruction set, treats the logical overflow fragment and the target storage area 101 as a single logical unit to perform inventory statistics and task distribution.

[0171] Through steps S6221 to S6223, when the target area is completely blocked by physical shelf supports, aisles, or other levels of goods, the system automatically finds the nearest alternative resource by sorting by topological distance, avoiding system downtime due to spatial deadlock. By prioritizing allocation to logical overflow fragments, the system ensures the relative concentration of goods of the same level in three-dimensional space, effectively reducing the ineffective empty runs of automated handling equipment (such as AGVs) during cross-area operations. By using data mapping association to integrate discrete storage locations into a unified logical view, the system solves the technical problem of the exponential increase in the complexity of inventory counting and route planning caused by storage fragmentation.

[0172] In an optional embodiment, step S7 is executed through the control layer of the digital twin warehouse system to synchronize the physical warehousing operations of goods with the state of the virtual model. Specifically, step S7 includes the following steps:

[0173] S71, Control and drive the automatic handling equipment to transport the target goods to an available storage location in the target storage area;

[0174] Specifically, the processor of the digital twin warehouse system, in response to the available storage location determined in step S5 or the alternative available storage location generated by the reconstruction in step S6, retrieves the spatial coordinates of the storage location in the physical coordinate system.

[0175] It should be noted that the processor performs path calculations based on the expanded regional topology. Specifically, the processor starts at the inlet coordinates of the target goods to be put into storage and ends at the center of the determined available storage location, using a three-dimensional topological mesh to perform path search, and calculates the integral length of the target trajectory. Construct the following objective function:

[0176]

[0177] in, The start time, The time to reach the target storage location. This refers to the real-time position coordinates of the automated handling equipment in three-dimensional space.

[0178] Specifically, the processor generates control commands containing start and end coordinates and a sequence of actions based on the calculated minimum trajectory length, and sends these commands to the onboard controller of the automated handling equipment (AWTA) via an industrial wireless communication module. The onboard controller drives the power unit of the AWTA to perform physical displacement. During transport, the AWTA monitors its real-time pose in physical space using an onboard LiDAR scanner. This step transforms the digital topology path into the mechanical motion trajectory of a physical entity, ensuring the target goods are delivered to the pre-locked logical storage location node.

[0179] S72. Send the location occupancy command back to the digital twin warehouse system;

[0180] Specifically, after the automated handling equipment places the target goods at the center of an empty storage location, it acquires a physical occupancy signal through pressure sensors or photoelectric sensors deployed at the bottom of the shelf compartments. It should be noted that after the automated handling equipment completes the unloading operation, its onboard controller encapsulates a status feedback message containing a storage location hardware identifier, goods feature code, and an operation completion timestamp, and defines this message as a location occupancy command.

[0181] Specifically, the location occupancy command is uploaded to the management terminal of the digital twin warehouse system via a communication base station. This feedback mechanism establishes a data feedback channel from the physical device layer to the digital twin logic layer, enabling the perception of the physical space occupancy status.

[0182] S73, Update real-time occupancy status.

[0183] Specifically, the processor of the digital twin warehouse system parses the location occupancy command and locates the corresponding data node in the storage location mapping table and the 3D topology model. It should be noted that the processor changes the real-time occupancy status bit of the node from 0 to 1, indicating that the physical space has been occupied by the target goods, and simultaneously records the storage start time of the node. Specifically, in the visual presentation layer of the 3D topology model, the topology node corresponding to the grid changes from an empty texture to the rendering color of the corresponding goods level. This step completes the mapping of the digital twin environment to the physical storage results, ensuring consistency between the 3D topology model and the actual physical warehouse situation through transaction updates in the underlying database. This synchronization mechanism provides a storage distribution benchmark for subsequent execution steps S5 or S6 of the system.

[0184] Example 2

[0185] According to a second aspect of the embodiments of this application, such as Figure 7 As shown, an intelligent warehouse management system 60 based on multimodal feature fusion is provided, including:

[0186] The state perception module 61 is used to acquire the state information of the target cargo; wherein, the state information includes image features and weight features;

[0187] Specifically, the state perception module 61 is connected to an industrial camera and a weighing sensor located at the inbound end via an industrial communication interface. The industrial camera is used to collect multi-view visual image data of the target goods to extract image features, and the weighing sensor is used to collect real-time load data of the goods to extract weight features.

[0188] Furthermore, the state perception module 61 adopts a synchronous trigger pulse mechanism to ensure that image features and weight features are acquired under the same time reference and encapsulated into a state information message with a timestamp.

[0189] It should be noted that, depending on the material and storage requirements of the target goods, the status information may optionally include temperature distribution characteristics obtained by an infrared thermal imager, or odor volatility concentration characteristics obtained by a chemical sensor, in order to construct a multi-dimensional status profile of the target goods.

[0190] The quality grade assessment module 62 is used to perform multimodal feature fusion processing on the status information to determine the cargo grade that characterizes the degree of quality deterioration of the target cargo.

[0191] Specifically, the quality grade evaluation module 62 calls the processor to execute the correlation analysis algorithm, and uses heterogeneous image features and weight features as input parameters for fusion calculation.

[0192] Preferably, the system classifies the goods into corresponding grades (such as Class I, Class II, etc.) based on the calculated quantitative value of quality deterioration and in accordance with the preset grading standards.

[0193] It should be noted that the multimodal feature fusion processing eliminates the identification error of single sensor data under specific environments, and improves the accuracy and stability of quality grading.

[0194] The digital twin warehouse system 63 is used to call the preset storage location mapping table and the corresponding three-dimensional topology model of the physical warehouse, and to carry the attribute labels and real-time occupancy status of each storage location in the three-dimensional topology model; wherein, the real-time occupancy status is constructed from the feedback data of historical inbound goods;

[0195] Specifically, the three-dimensional topology model digitally maps physical storage locations using virtual nodes. Each virtual node carries attribute labels (such as environmental requirements and access levels) and real-time occupancy status (occupied or idle).

[0196] It should be noted that the real-time occupancy status is dynamically constructed and continuously updated by the feedback data of historical inbound goods (such as location occupancy instructions), thereby ensuring that the virtual model can reflect the real occupancy status of the physical warehouse in real time.

[0197] The spatial retrieval decision module 64 is used to retrieve the target storage area that matches the target cargo level in the three-dimensional topology model based on the digital twin warehouse system, and to determine whether there are any vacant storage locations available for storage in the target storage area based on the real-time occupancy status.

[0198] Specifically, the spatial retrieval decision module 64 combines cargo grade and storage location attribute tags to perform matching logic in order to locate the target storage area.

[0199] Furthermore, if there are free storage spaces in the target storage area, the module sends an inbound instruction to the handling task control module 66; if there are no free storage spaces, the dynamic topology reconstruction module 65 is triggered to perform expansion.

[0200] The dynamic topology reconstruction module 65 is used to search for candidate free storage locations that meet the spatial adjacency conditions in the three-dimensional topology model based on the topology center of the target storage area when there are no free storage locations in the target storage area, and to batch modify the attribute labels of the candidate free storage locations in order to realize the dynamic expansion of the target storage area in the virtual space.

[0201] Specifically, the dynamic topology reconstruction module 65 determines the topology center by performing geometric calculations on the coordinates of existing nodes within the target storage area.

[0202] Furthermore, the module uses this center as a reference to detect physically adjacent (sharing boundaries, edges, or vertices) and idle storage locations in the surrounding area.

[0203] It should be noted that by batch modifying the logical attribute tags of these candidate storage locations, dynamic expansion of the target storage area within the virtual space can be achieved. The expansion process follows the principle of spatial layout continuity, ensuring that the newly added storage locations are connected to the original areas to avoid creating logical silos.

[0204] The handling task control module 66 is used to control and drive the automatic handling equipment to transport the target goods to the designated storage location when there is an available storage location, and to send the location occupancy instruction back to the digital twin warehouse system to synchronously update the real-time occupancy status.

[0205] Specifically, the handling task control module 66 plans the driving path for the automated guided vehicle (AGV) or stacker crane and monitors its movement status in real time.

[0206] Furthermore, after the goods are successfully placed, the module receives the arrival feedback signal from the device and sends a location occupancy command to the digital twin warehouse system 63.

[0207] It should be noted that the instruction triggers the digital twin warehouse system to update the occupancy status bit of the corresponding coordinates in real time, thereby completing the management closed loop from physical execution to virtual synchronization.

[0208] In some optional embodiments, the intelligent warehouse management system based on multimodal feature fusion provided in this embodiment can be used to execute and implement the intelligent warehouse management method based on multimodal feature fusion provided in the aforementioned embodiment 1. For a detailed description of the method, please refer to the corresponding description in the above method embodiments, which will not be repeated here.

[0209] Example 3

[0210] According to another aspect of the embodiments of the present invention, such as Figure 8 As shown, an electronic device 700 is also provided, including a memory 701 and a processor 702; wherein the memory 701 is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor 702 to implement the intelligent warehouse management method based on multimodal feature fusion as described above. For a detailed description of the intelligent warehouse management method based on multimodal feature fusion, please refer to the corresponding description in the above method embodiments, and it will not be repeated here.

[0211] Example 4

[0212] According to another aspect of the embodiments of the present invention, such as Figure 9 As shown, a storage medium 800 is also provided, on which computer instructions are stored; wherein, when the computer instructions are executed by a processor, the above-described intelligent warehouse management method based on multimodal feature fusion is implemented. For a detailed description of the intelligent warehouse management method based on multimodal feature fusion, please refer to the corresponding description in the above method embodiments, and it will not be repeated here.

[0213] The program instructions are stored in a computer-readable storage medium (such as a CD-ROM, USB flash drive, or portable hard drive) or on a network, and include several computer program instructions to cause a computing device (such as a personal computer, server, or network device) to execute the above-described method according to the embodiments of this application.

[0214] Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention, and other modifications can be easily made by those skilled in the art. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and examples shown and described herein.

[0215] The apparatus, electronic device, and non-volatile computer storage medium and method provided in the embodiments of this specification are corresponding. Therefore, the apparatus, electronic device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the corresponding apparatus, electronic device, and non-volatile computer storage medium will not be repeated here.

[0216] Those skilled in the art will also know that, besides implementing the controller in the form of purely computer-readable program code, the same functions can be achieved by logically programming the method steps, making the controller take the form of logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers (PLCs), and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the devices included within it for implementing various functions can also be considered structures within that hardware component. Alternatively, the devices for implementing various functions can be considered as both software modules implementing the method and structures within a hardware component.

[0217] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0218] For ease of description, the above apparatus is described by dividing it into various functional units. Of course, when implementing one or more embodiments of this specification, the functions of each unit can be implemented in one or more software and / or hardware.

[0219] Those skilled in the art will understand that the embodiments of this specification can be provided as methods, systems, or computer program products. Therefore, the embodiments of this specification can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects.

[0220] This specification is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this specification. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0221] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0222] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0223] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0224] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0225] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0226] This specification can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This specification can also be practiced in distributed computing environments, where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside on local and remote computer storage media, including storage devices.

[0227] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0228] The above description is merely an embodiment of this specification and is not intended to limit the scope of one or more embodiments of this specification. Various modifications and variations can be made to one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of one or more embodiments of this specification.

Claims

1. A smart warehouse management method based on multimodal feature fusion, characterized in that, Includes the following steps: Obtain the status information of the target cargo; wherein, the status information includes image features and weight features; Multimodal feature fusion processing is performed on the state information to determine the quality degradation degree of the target goods. Item grade; Obtain the pre-established digital twin warehouse system of the smart warehouse; The storage location mapping table and its three-dimensional topology model in the digital twin warehouse system are invoked so that the three-dimensional topology model carries the attribute labels and real-time occupancy status of each storage location; wherein, the real-time occupancy status is constructed from the feedback data of historical inbound goods; Based on the digital twin warehouse system, a target storage area matching the target cargo level is retrieved in the three-dimensional topology model, and it is determined whether there are any vacant storage spaces in the target storage area based on the real-time occupancy status. If no available storage space exists, the digital twin warehouse system searches for candidate available storage spaces that meet the spatial adjacency conditions in the three-dimensional topology model based on the topology center of the target storage area, and modifies the logical attribute labels of the candidate available storage spaces in batches so that the expanded target storage area meets the spatial layout continuity, and returns to the step of determining whether there is an available storage space in the target storage area. If the available storage space exists, the automated handling equipment is controlled to transport the target goods to the available storage space in the target storage area, and a location occupancy command is sent back to the digital twin warehouse system to update the real-time occupancy status. The step of retrieving a target storage area matching the target cargo level in the three-dimensional topology model based on the digital twin warehouse system, and determining whether there are any vacant storage spaces in the target storage area based on the real-time occupancy status, specifically includes the following steps: Identify the geometric dimensions of the target cargo and map them into the three-dimensional topology model to generate a virtual entity mirror image of the target cargo with the same scale. Using the digital twin warehouse system, the access attribute tags corresponding to the cargo level in the three-dimensional topology model are retrieved to determine the target storage area composed of multiple storage location nodes; Retrieve the real-time occupancy status of each storage location node within the target storage area, extract the coordinates of storage location nodes not occupied by historically inbound goods, and construct a candidate set of free coordinates; Using the digital twin warehouse system, the virtual entity image is compared with the physical envelope space corresponding to the candidate free coordinate set to determine whether there is a storage location in the target storage area that can accommodate the virtual entity image. If the storage location coordinates obtained through the interference comparison exist, it is determined that there is a storage area in the target storage region available for the target goods. Available storage locations for goods to be stored; If the coordinates of the storage location cannot be obtained through the interference comparison, it is determined that there is no free storage location in the target storage area.

2. The method according to claim 1, characterized in that, The step of performing multimodal feature fusion processing on the state information to determine the cargo grade characterizing the quality deterioration of the target cargo specifically includes the following steps: Area feature data characterizing surface liquid diffusion is extracted from the image features, and weight loss data characterizing material loss is extracted from the weight features; The area feature data and the weight loss data are input into a preset quality correlation model, and the deviation feature value of the area feature data relative to the weight loss data is calculated through the quality correlation model. The cargo grade, which characterizes the degree of quality deterioration of the target cargo, is determined based on the deviation feature value.

3. The method according to claim 2, characterized in that, The step of determining the cargo grade characterizing the quality deterioration of the target cargo based on the deviation feature value specifically includes the following steps: Based on the category information of the target goods, a preset deterioration threshold is retrieved from the storage location mapping table; The quality deterioration degree of the target goods is numerically compared with a quality threshold to determine the goods grade that characterizes the quality deterioration degree of the target goods; wherein, the storage location mapping table includes at least one deterioration threshold; and the goods grade includes at least two grades.

4. The method according to claim 1, characterized in that, The process involves using the digital twin warehouse system to retrieve candidate vacant storage locations that meet spatial adjacency conditions in the three-dimensional topology model based on the topology center of the target storage area. The logical attribute tags of these candidate vacant storage locations are then modified in batches to ensure spatial layout continuity in the expanded target storage area. Finally, the process returns to the step of determining whether there are vacant storage locations in the target storage area. Specifically, this includes the following steps: The coordinate set of the target storage area in the three-dimensional topology model is retrieved, and the storage location node at the edge position in the coordinate set is identified to determine the spatial topological outer edge of the target storage area; Using the digital twin warehouse system, adjacent storage location nodes that satisfy shared boundary or vertex constraints are retrieved from the outer edge of the spatial topology outwards, and storage location nodes whose real-time occupancy status is unoccupied are determined as the candidate free storage locations; Generate partition reconstruction instructions for the candidate vacant storage locations, and batch overwrite the attribute labels of the candidate vacant storage locations in the three-dimensional topology model with identifiers that match the current target cargo level; The attribute association data in the storage location mapping table is updated synchronously, and the process returns to the step of determining whether there is a free storage location in the target storage area.

5. The method according to claim 4, characterized in that, If no idle storage location node is found within a preset neighborhood on the outer edge of the spatial topology, the following steps are also included: Using the digital twin warehouse system, storage location nodes located on the outer periphery of the spatial topology and occupied by non-target level goods are identified as obstacle nodes; The three-dimensional topology model is used to simulate the migration of goods from the obstructing nodes to other vacant areas, and the energy consumption cost and spatial continuity benefits required to perform the migration task are calculated. In response to the spatial continuity benefit being greater than the energy consumption cost, a cargo handling instruction is generated for the obstructing node, and the automated handling equipment is controlled to clear the obstructing node so as to convert it into the alternative vacant storage location.

6. The method according to claim 4, characterized in that, If no idle storage location node is found within a preset neighborhood on the outer edge of the spatial topology, the following steps are also included: Using the digital twin warehouse system, non-adjacent free storage locations are retrieved in the three-dimensional topology model in ascending order of topological distance, and these are used as candidate free storage locations. Modify the attribute labels of the candidate free storage locations and establish a data mapping association between the candidate free storage locations and the target storage area; The alternative idle storage locations are used as logical overflow fragments of the target storage area for unified scheduling, so that the target goods are preferentially allocated to the logical overflow fragments when they are put into storage.

7. An intelligent warehouse management system based on multimodal feature fusion, characterized in that, include: A state perception module is used to acquire the state information of the target cargo; wherein, the state information includes image features and weight features; The quality grade assessment module is used to perform multimodal feature fusion processing on the status information to determine the cargo grade that characterizes the quality deterioration of the target cargo; The digital twin warehouse system is used to call a preset storage location mapping table and a three-dimensional topology model corresponding to the physical warehouse, and to carry the attribute labels and real-time occupancy status of each storage location in the three-dimensional topology model; wherein, the real-time occupancy status is constructed from the feedback data of historical inbound goods; The spatial retrieval decision module is used to identify the geometric dimension data of the target goods based on the digital twin warehouse system and map it into a virtual entity image in the three-dimensional topology model. It retrieves the access attribute tags corresponding to the goods' level in the three-dimensional topology model to determine the target storage area, retrieves the real-time occupancy status of each storage location node within the target storage area and constructs a candidate idle coordinate set. By interferometrically comparing the virtual entity image with the physical envelope space corresponding to the candidate idle coordinate set, it determines whether there are any available empty storage locations in the target storage area. The dynamic topology reconstruction module is used to, when no empty storage location exists in the target storage area, retrieve candidate empty storage locations that meet spatial adjacency conditions in the three-dimensional topology model based on the topology center of the target storage area, and batch modify the attribute tags of the candidate empty storage locations to achieve dynamic expansion of the target storage area in the virtual space. The handling task control module is used to control and drive the automated handling equipment to transport the target goods to the designated storage location when there is an available storage location, and to send a location occupancy instruction back to the digital twin warehouse system to synchronously update the real-time occupancy status.

8. An electronic device comprising a memory and a processor; wherein, The memory is used to store one or more computer instructions, wherein the one or more computer instructions are executed by the processor to implement the method according to any one of claims 1-6.

9. A storage medium having computer instructions stored thereon; wherein, When executed by a processor, the computer instructions implement the method described in any one of claims 1-6.

Images