A logistics distribution site monitoring method and system based on real-time data fusion

Abstract

The present application relates to the technical field of logistics monitoring, in particular to a logistics distribution site monitoring method and system based on real-time data fusion, comprising: acquiring multi-source basic data of a logistics distribution site, performing cargo area classification modeling operation on the logistics distribution site according to the multi-source basic data to obtain a cargo area classification model of the logistics distribution site; wherein the multi-source basic data comprises site static topology data and real-time initial perception data; determining parking space semantic labeling parameters corresponding to the cargo area classification model, performing semantic labeling operation on parking spaces of the logistics distribution site according to the parking space semantic labeling parameters to obtain a parking space semantic model of the logistics distribution site. The present application can realize dynamic monitoring of cargo areas and parking spaces by acquiring multi-source data of a logistics distribution site in real time, and this real-time data fusion technology enables timely discovery of abnormalities in site use, thereby helping managers to respond and adjust more efficiently.

Description

Technical Field

[0001] This invention relates to the field of logistics monitoring technology, specifically to a method and system for monitoring logistics distribution sites based on real-time data fusion. Background Technology

[0002] Logistics networks encompass multiple links such as cross-border transportation, warehousing, and distribution. Traditional manual management and single data source methods are proving inadequate. Production, inventory, and distribution links around the world must be connected quickly and accurately. This requires logistics systems to monitor and respond to market changes in real time to avoid resource waste and transportation delays.

[0003] Currently, traditional methods typically rely on manual inspections or periodic data collection, lacking real-time update capabilities. As a result, managers cannot be aware of any anomalies in the site in a timely manner, which can easily lead to delays in the discovery and handling of problems. Furthermore, they often depend on a single data source and lack the ability to integrate multi-source data, making it difficult for information to complement and verify each other, thus affecting the comprehensiveness and accuracy of the data. This is especially true in complex logistics scenarios, where managers cannot fully grasp the actual situation of the site, and their decisions lack real-time data support.

[0004] Furthermore, traditional cargo area planning and parking space allocation are usually based on fixed rules or preset patterns, lacking the flexibility for dynamic adjustment. With changes in factors such as cargo type, mobility, and seasonality, traditional methods often cannot efficiently cope with demand fluctuations, resulting in wasted space and inefficient operations. Moreover, they often rely on floor plans or two-dimensional maps for site planning and monitoring, which cannot present the detailed relationships and real-time status in complex spaces. The lack of three-dimensional modeling means that managers have difficulty intuitively grasping the spatial relationships between elements such as cargo areas, parking spaces, and equipment, and also find it difficult to effectively cope with changing operating environments. Summary of the Invention

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for monitoring logistics distribution sites based on real-time data fusion, comprising:

[0006] Multi-source basic data of the logistics distribution site is acquired, and a cargo area classification modeling operation is performed on the logistics distribution site based on the multi-source basic data to obtain the cargo area classification model of the logistics distribution site; wherein, the multi-source basic data includes site static topology data and real-time initial sensing data.

[0007] The semantic annotation parameters of the parking spaces corresponding to the cargo area classification model are determined. Based on the semantic annotation parameters of the parking spaces, the parking spaces of the logistics distribution site are semantically annotated to obtain the semantic model of the parking spaces of the logistics distribution site. The semantic annotation parameters of the parking spaces include coordinate system mapping parameters, semantic label configuration parameters, parking space-cargo area association parameters, and spatial accuracy calibration parameters.

[0008] Based on the cargo area classification model and the parking space semantic model, a three-dimensional model generation operation is performed on the logistics distribution site to obtain a three-dimensional distribution site model of the logistics distribution site.

[0009] Determine the dynamic monitoring parameters corresponding to the three-dimensional distribution site model, and perform dynamic monitoring operations on the logistics distribution site based on the dynamic monitoring parameters and real-time operating data.

[0010] Preferably, a cargo area classification modeling operation is performed on the logistics distribution site to obtain a cargo area classification model for the logistics distribution site, including:

[0011] Acquire static topology data and real-time initial perception data of the logistics distribution site; wherein, the static topology data includes site physical size data, building load-bearing structure data and initial cargo area planning data, and the real-time initial perception data includes lidar point cloud data, visual camera image data and RFID tag initial reading data.

[0012] Based on the static topology data of the site, the planarable cargo area range and cargo area classification dimensions of the logistics distribution site are determined; the cargo area classification dimensions include cargo turnover rate, cargo weight class, cargo timeliness requirements, and cargo storage temperature zone.

[0013] Based on the real-time initial perception data and the cargo area classification dimension, cargo area classification feature parameters within the planarable cargo area range are determined; wherein, the cargo area classification feature parameters include turnover rate distribution parameters, weight load adaptation parameters, timeliness response threshold parameters, and temperature zone demand distribution parameters;

[0014] Based on the cargo area classification feature parameters, the planned cargo area range is divided into cargo areas and its attributes are modeled to obtain the cargo area classification model of the logistics distribution site.

[0015] Preferably, based on the parking space semantic annotation parameters, semantic annotation operations are performed on the parking spaces in the logistics distribution area to obtain a parking space semantic model of the logistics distribution area, including:

[0016] Based on the cargo area classification model, the parking space functional requirements and parking space associations of the logistics distribution site are determined; the parking space functional requirements include unloading parking space requirements, temporary storage parking space requirements, and outgoing parking space requirements; the parking space associations include the spatial association between parking spaces and cargo areas and the adaptation association between parking spaces and equipment.

[0017] Based on the functional requirements of the parking spaces and the association relationships between the parking spaces, the semantic annotation parameters of the parking spaces are determined;

[0018] Based on the parking space semantic annotation parameters, the parking spaces in the logistics distribution site are spatially located and labeled to obtain a parking space semantic model that includes parking space semantic information.

[0019] Preferably, based on the cargo area classification model and the parking space semantic model, a three-dimensional model generation operation is performed on the logistics distribution site to obtain a three-dimensional distribution site model of the logistics distribution site, including:

[0020] Acquire real-time detailed perception data of the logistics distribution site; wherein, the real-time detailed perception data includes three-dimensional structural data of the shelves, location data of sorting equipment, and aisle identification data;

[0021] The model fusion benchmark is determined based on the spatial boundary parameters of the cargo area classification model and the coordinate parameters of the parking space semantic model.

[0022] The real-time detail perception data is fused and geometrically reconstructed with the cargo area classification model and the parking space semantic model according to the model fusion benchmark to obtain a three-dimensional distribution site model of the logistics distribution site.

[0023] Preferably, determining the dynamic monitoring parameters corresponding to the three-dimensional distribution site model includes:

[0024] Obtain the cargo area operation feature parameters of the cargo area classification model and the parking space usage feature parameters of the parking space semantic model; wherein, the cargo area operation feature parameters include cargo area busy period parameters and cargo throughput parameters, and the parking space usage feature parameters include parking space occupancy rate parameters and parking space turnover efficiency parameters;

[0025] Based on the cargo area operation characteristic parameters and the parking space usage characteristic parameters, the operation sensitive area and the monitoring key area of ​​the logistics distribution site are determined, and the first area attribute parameters of the operation sensitive area and the second area attribute parameters of the monitoring key area are determined.

[0026] Based on the attribute parameters of the first region and the attribute parameters of the second region, the dynamic monitoring parameters corresponding to the three-dimensional distribution site model are determined; wherein, the dynamic monitoring parameters include the sensor deployment parameters, data acquisition frequency parameters, anomaly judgment threshold parameters, and model update cycle parameters.

[0027] Preferably, the planned cargo area is divided and its attributes are modeled based on the cargo area classification feature parameters to obtain the cargo area classification model of the logistics distribution site, including:

[0028] The real-time initial sensing data is preprocessed; the lidar point cloud data is denoised and ground point filtered out to obtain the three-dimensional contour data of the site; the visual camera image data is feature extracted and distortion corrected to obtain cargo area identification and equipment location data; the initial reading data of the RFID tags is deduplicated and coordinate matched to obtain the initial cargo distribution data.

[0029] The preprocessed site 3D contour data, cargo area identification, equipment location data, and initial cargo distribution data are mapped according to the cargo area classification dimension to generate cargo area classification feature parameters.

[0030] Hierarchical clustering algorithm is used to perform cluster analysis on the classification feature parameters of the cargo area, resulting in multiple cargo area clusters;

[0031] Based on the characteristic attributes of the cargo area clusters and the spatial constraints of the planarable cargo area range, the boundaries of each cargo area cluster are divided and attribute values ​​are assigned to construct the cargo area classification model.

[0032] Preferably, the parking spaces in the logistics distribution area are spatially located and labeled according to the parking space semantic annotation parameters to obtain a parking space semantic model including parking space semantic information, including:

[0033] The semantic annotation parameters of the parking space are parsed to generate a corresponding semantic tag set and spatial mapping rules; wherein, the semantic tag set includes parking space number tag, function type tag, carrying capacity limit tag and associated cargo area ID tag, and the spatial mapping rules include conversion rules between parking space coordinates and cargo area coordinates and binding rules between tags and the geometric center of the parking space;

[0034] The precise three-dimensional coordinates of each parking space are determined by constructing a three-dimensional point cloud cluster based on LiDAR point cloud data and combining it with the coordinate transformation rules in the spatial mapping rules.

[0035] The tags in the semantic tag set are associated and mapped with the corresponding precise three-dimensional coordinates of the parking space according to the binding rules, and the association relationship between the tags and the coordinates is stored to form the parking space semantic model.

[0036] Preferably, the dynamic monitoring operation of the logistics distribution site is performed based on the dynamic monitoring parameters and real-time operating data, including:

[0037] Based on the data acquisition frequency parameter in the dynamic monitoring parameters, real-time operational data of the logistics distribution site is collected through a preset cluster of sensing devices; wherein, the real-time operational data includes cargo inventory data in the cargo area, vehicle parking data in parking spaces, equipment operating status data, and personnel flow data;

[0038] Anomaly detection is performed on the real-time running data; the real-time running data is compared with the anomaly judgment threshold parameter in the dynamic monitoring parameters. If a certain data exceeds the corresponding threshold, it is judged as abnormal data, and the corresponding abnormal location and anomaly type are marked.

[0039] The three-dimensional distribution site model is dynamically updated based on the real-time operating data and the abnormal data.

[0040] Preferably, the dynamic monitoring operation of the logistics distribution site based on the dynamic monitoring parameters and real-time operating data further includes:

[0041] According to the model update cycle parameter in the dynamic monitoring parameters, the cargo inventory data of the cargo area and the vehicle parking data of the parking space are synchronized to the corresponding positions of the three-dimensional distribution site model, and the abnormal positions and abnormal types are visually marked in the three-dimensional distribution site model.

[0042] Output the updated 3D distribution site model and abnormal early warning information.

[0043] A logistics distribution site monitoring system based on real-time data fusion, applicable to the aforementioned logistics distribution site monitoring method based on real-time data fusion, includes:

[0044] The cargo area classification unit is used to acquire multi-source basic data of the logistics distribution site, and to perform cargo area classification modeling operation on the logistics distribution site based on the multi-source basic data to obtain the cargo area classification model of the logistics distribution site; wherein, the multi-source basic data includes site static topology data and real-time initial sensing data.

[0045] A semantic annotation unit is used to determine the semantic annotation parameters of parking spaces corresponding to the cargo area classification model. Based on the semantic annotation parameters, semantic annotation operations are performed on the parking spaces of the logistics distribution site to obtain the semantic model of the parking spaces of the logistics distribution site. The semantic annotation parameters of the parking spaces include coordinate system mapping parameters, semantic label configuration parameters, parking space-cargo area association parameters, and spatial accuracy calibration parameters.

[0046] The modeling unit is used to perform a three-dimensional model generation operation on the logistics distribution site based on the cargo area classification model and the parking space semantic model, so as to obtain a three-dimensional distribution site model of the logistics distribution site.

[0047] The site monitoring unit is used to determine the dynamic monitoring parameters corresponding to the three-dimensional distribution site model, and to perform dynamic monitoring operations on the logistics distribution site based on the dynamic monitoring parameters and real-time operating data.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] (1) This invention enables dynamic monitoring of cargo areas and parking spaces by acquiring multi-source data from logistics distribution sites in real time. This real-time data fusion technology allows for timely detection of anomalies in site usage, thereby helping managers respond and adjust more efficiently. By classifying and modeling cargo areas, scientific planning can be carried out based on multiple dimensions such as cargo turnover rate, weight class, and timeliness requirements, thereby optimizing cargo storage and parking space allocation. This not only improves space utilization but also increases operational efficiency and reduces unnecessary resource waste.

[0050] (2) By generating a three-dimensional distribution site model, the present invention can intuitively display the layout and operation status of the entire site. This kind of three-dimensional modeling helps managers to view the spatial relationship between cargo areas, parking spaces and equipment in real time, and enhances their understanding and ability to cope with complex logistics scenarios. By detecting anomalies in real-time operating data and setting dynamic monitoring parameters, timely alarms can be set for data exceeding preset thresholds, helping managers to quickly identify problems and take measures to reduce potential safety hazards or operational bottlenecks.

[0051] (3) This invention can obtain the usage status of cargo areas and parking spaces in real time, and provide data support for key indicators such as cargo throughput and parking space turnover rate, thereby providing a scientific basis for logistics scheduling and personnel allocation. Through these data analysis, logistics operations can more accurately schedule vehicles and personnel, improve overall operational efficiency, and ensure the accuracy and real-time performance of the model by combining data from different sources (static topology data and real-time perception data). This intelligent update enables the system to adapt to changing environments and operating conditions, providing more accurate and stable monitoring results. Attached Figure Description

[0052] Figure 1 This is a schematic flowchart of the overall method in one embodiment of the present invention;

[0053] Figure 2 This is a schematic diagram of the overall system architecture in one embodiment of the present invention.

[0054] In the diagram: 1. Goods area classification unit; 2. Semantic annotation unit; 3. Model building unit; 4. Site monitoring unit. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1, please refer to Figure 1 This invention provides a technical solution: a method for monitoring logistics distribution sites based on real-time data fusion, comprising:

[0057] S1. Obtain multi-source basic data of the logistics distribution site. Based on the multi-source basic data, perform cargo area classification modeling operation on the logistics distribution site to obtain the cargo area classification model of the logistics distribution site. Among them, the multi-source basic data includes site static topology data and real-time initial sensing data.

[0058] S2. Determine the semantic annotation parameters of the parking spaces corresponding to the cargo area classification model. Based on the semantic annotation parameters of the parking spaces, perform semantic annotation operations on the parking spaces in the logistics distribution area to obtain the semantic model of the parking spaces in the logistics distribution area. Among them, the semantic annotation parameters of the parking spaces include coordinate system mapping parameters, semantic label configuration parameters, parking space-cargo area association parameters, and spatial accuracy calibration parameters.

[0059] S3. Based on the cargo area classification model and the parking space semantic model, perform a three-dimensional model generation operation on the logistics distribution site to obtain a three-dimensional distribution site model of the logistics distribution site.

[0060] S4. Determine the dynamic monitoring parameters corresponding to the three-dimensional distribution site model, and perform dynamic monitoring operations on the logistics distribution site based on the dynamic monitoring parameters and real-time operation data.

[0061] It should be noted that basic information about logistics distribution centers is obtained through data from different sources. This multi-source basic data includes: static topology data describing the physical structure, layout, parking spaces, and cargo areas of the center; and real-time initial sensing data, which is real-time data acquired through sensors (such as RFID, cameras, and ground sensors), including information on cargo location, transport vehicle status, and cargo area status. For example, suppose a logistics distribution center has multiple cargo areas (e.g., Area A, Area B, and Area C), each with different cargo categories. Using the static topology data, a cargo area classification model can be established to distinguish between different cargo areas. For instance, Area A is for storing electronic products, Area B for clothing, and Area C for heavy machinery. Real-time sensing data, on the other hand, uses sensors to track the quantity and location of goods within each cargo area in real time.

[0062] Based on the cargo area classification model from step 1, determine the semantic labeling parameters for each parking space. These parameters include: coordinate system mapping parameters: mapping the actual location of the parking space to its coordinates in the model; semantic label configuration parameters: assigning appropriate labels to each parking space to help identify its purpose, cargo area, and other information; parking space-cargo area association parameters: determining the association between the parking space and the cargo area, for example, parking spaces in area A are specifically used for storing electronic products; and spatial accuracy calibration parameters: ensuring that the positional accuracy of each parking space meets the requirements and avoiding inaccurate positioning due to sensor errors. For example, in area A of a logistics site, there are multiple parking spaces used for storing electronic products. The coordinate system mapping parameters determine the precise location of these parking spaces; the semantic label configuration parameters mark these parking spaces as electronic product parking spaces; the parking space-cargo area association parameters further associate them with cargo area A; and the spatial accuracy calibration ensures that the positions of these parking spaces are not deviated.

[0063] Based on the aforementioned cargo area classification model and parking space semantic model, a 3D model of the site is generated. This 3D model will intuitively present the layout of the entire logistics distribution site, including cargo areas, parking spaces, and the spatial distribution of goods. For example, by integrating all cargo area and parking space models, a 3D virtual logistics distribution site model can be generated. This model can be displayed on the screen, and users can view the layout of the site through virtual reality technology and further optimize and adjust it.

[0064] Based on the 3D distribution site model and real-time operational data (such as cargo location, vehicle entry and exit, parking space usage, etc.), dynamic monitoring operations are carried out. Dynamic monitoring parameters include: real-time operational status of the monitoring site, such as changes in inventory in the cargo area and parking space usage. Real-time data and monitoring models are combined to promptly identify problems and take corresponding adjustment measures.

[0065] Example: Suppose a large number of goods suddenly appear in Zone A of a logistics distribution center and need to be moved quickly. The monitoring system will detect the pressure on the goods storage area of ​​Zone A in real time, automatically dispatch the nearest available parking space, and adjust the distribution plan. For example, if there are empty parking spaces in Zone B, the system will automatically guide vehicles to Zone B to move electronic products, avoiding overcrowding in Zone A.

[0066] In one optional embodiment, a cargo area classification modeling operation is performed on the logistics distribution site to obtain a cargo area classification model for the logistics distribution site, including:

[0067] Acquire static topology data and real-time initial perception data of the logistics distribution site; wherein, the static topology data includes site physical size data, building load-bearing structure data and initial cargo area planning data, and the real-time initial perception data includes LiDAR point cloud data, visual camera image data and RFID tag initial reading data.

[0068] Based on the static topology data of the site, the planarable cargo area range and cargo area classification dimensions of the logistics distribution site are determined; the cargo area classification dimensions include cargo turnover rate, cargo weight class, cargo timeliness requirements, and cargo storage temperature zone.

[0069] Based on real-time initial perception data and cargo area classification dimensions, cargo area classification characteristic parameters within the planarable cargo area range are determined; among them, cargo area classification characteristic parameters include turnover rate distribution parameters, weight load adaptation parameters, timeliness response threshold parameters, and temperature zone demand distribution parameters.

[0070] Based on the cargo area classification characteristic parameters, the scope of the planarable cargo area is divided into cargo areas and its attributes are modeled to obtain the cargo area classification model of the logistics distribution site.

[0071] It should be noted that the physical dimensions of the site describe the size of the logistics distribution site, such as the total area of ​​the site and the length, width and height of the cargo area; the load-bearing structure data describes the load-bearing capacity of each area within the site, especially the ground load-bearing capacity, the load-bearing capacity of the shelves, etc., to ensure that the goods storage will not be overloaded; the initial cargo area planning data is the cargo area division data during the initial design of the site, including the purpose and storage capacity of each cargo area.

[0072] LiDAR point cloud data: By scanning the three-dimensional space of the site with LiDAR, point cloud data is generated to accurately obtain the spatial layout of the cargo area, the location of obstacles, etc.; Visual camera image data: Image data is acquired by cameras installed in the site, which can detect the location of goods and the status of the cargo area in real time; Initial RFID tag reading data: RFID technology helps to read the tag information of goods in the logistics site to determine the type, quantity, location, etc. of the goods; Example: Suppose the total area of ​​a logistics distribution center is 10,000 square meters. The initial cargo area planning data can tell that 10 areas are planned for storing electronic products, home appliances, food, etc.; and by using LiDAR and camera data, the status of goods in these areas can be monitored in real time.

[0073] Based on static topology data, the first step is to determine the areas within the logistics distribution center where cargo zones can be planned, i.e., the scope of plannable cargo zones. Then, based on different management needs, the dimensions for cargo zone classification are determined. Common classification dimensions include: Cargo turnover rate: cargo zones are classified according to the frequency of goods entering and leaving within a certain period; goods with high turnover rates may need to be allocated to areas near entrances and exits. Cargo weight level: cargo zones are classified according to weight; heavy items need to be allocated to areas with strong ground bearing capacity. Cargo timeliness requirements: cargo zones are classified according to timeliness requirements; some goods requiring rapid outbound processing may need to be placed closer to the exit. Cargo storage temperature zone: cargo zones are classified according to storage temperature requirements; for example, refrigerated or frozen goods need to be stored in temperature-controlled zones. Example: Suppose that in a distribution center, electronic products have high turnover, are lightweight, and have high timeliness requirements; they can be located near entrances and exits and close to the picking area. Perishable foods, on the other hand, require a temperature-controlled zone, and this zone should not be too far from the cold chain transportation channel.

[0074] Based on real-time sensing data and established classification dimensions, the classification characteristic parameters of the shipping area are analyzed. Each shipping area will have the following characteristic parameters: Turnover rate distribution parameter: a parameter reflecting the speed of goods turnover; high-turnover shipping areas need to have better storage and retrieval efficiency; Weight load capacity adaptation parameter: based on the weight class of the goods, the load capacity of each area is calculated and adapted; Timeliness response threshold parameter: to determine the timeliness requirements for goods processing, and set response thresholds for different timeliness requirements, such as triggering an alarm for goods that have not been processed within the time limit; Temperature zone demand distribution parameter: for temperature-controlled shipping areas, the distribution of goods with different needs such as refrigeration and freezing is analyzed to ensure that the configuration of temperature-controlled areas meets all needs; Example: if the turnover rate of a certain shipping area is high, the frequency of goods flow in that area can be monitored in real time through RFID, thereby determining that the area belongs to a high-turnover area; while for heavy mechanical equipment, it is necessary to allocate it to areas with strong load-bearing capacity according to the weight load capacity adaptation parameter;

[0075] Based on the extracted cargo area classification feature parameters, the planarable cargo area range is divided and its attributes are modeled. Each cargo area is assigned one or more attributes for more precise management and scheduling. These attributes include: the type of each cargo area (e.g., high-turnover cargo area, refrigerated cargo area, etc.); the capacity of each cargo area (e.g., weight capacity, cargo storage capacity); and the timeliness requirements of each cargo area (e.g., whether fast delivery is required). For example, in an e-commerce logistics center, after classification modeling, the following types of cargo areas may be formed: High-turnover cargo area: close to the entrance / exit, goods need to move quickly, and the turnover rate is high; Low-turnover cargo area: far from the entrance / exit, goods are stored for a longer time; Temperature-controlled cargo area: suitable for frozen food or medicine, with a temperature control system; Heavy cargo area: located in an area with a large ground load-bearing capacity, suitable for storing heavy equipment.

[0076] In an optional embodiment, semantic annotation operations are performed on the parking spaces in the logistics distribution site according to the parking space semantic annotation parameters to obtain a parking space semantic model of the logistics distribution site, including:

[0077] Based on the cargo area classification model, determine the functional requirements of parking spaces and the relationships between parking spaces in the logistics distribution area. The functional requirements of parking spaces include the requirements for unloading parking spaces, temporary storage parking spaces, and outgoing parking spaces. The relationships between parking spaces include the spatial relationship between parking spaces and cargo areas and the compatibility relationship between parking spaces and equipment.

[0078] Based on the functional requirements of parking spaces and the relationships between parking spaces, determine the semantic annotation parameters for parking spaces;

[0079] Based on the semantic annotation parameters of parking spaces, the parking spaces in the logistics distribution site are spatially located and labeled, resulting in a parking space semantic model that includes parking space semantic information.

[0080] It should be noted that, firstly, based on the previously established cargo area classification model, the functional requirements of parking spaces and the relationships between them are determined. The functional requirements of parking spaces can be derived from operations such as cargo storage, unloading, and shipping, specifically including the following requirements: Unloading parking space requirements: A certain number of parking spaces need to be set up for transport vehicles to unload goods; these parking spaces are usually close to goods that need to be processed quickly (such as high-turnover goods); Temporary storage parking space requirements: Used for temporarily storing unsorted or unprepared goods; these parking spaces are usually close to the unloading area but not directly used for high-frequency operations; Shipping parking space requirements: Parking spaces used to transfer goods from the warehouse to transport vehicles; these parking spaces are usually close to the shipping area or loading area; Spatial relationship between parking spaces and cargo areas: The distribution of parking spaces should match the layout of the cargo areas to ensure convenient access for transport vehicles. For example, unloading spaces are typically located near high-turnover areas or receiving areas. Parking spaces should also be compatible with equipment in the distribution area (such as forklifts, conveyors, and automated robots) to ensure efficient transfer of goods from the parking spaces to the appropriate cargo areas. For example, in an e-commerce logistics center, unloading spaces are usually located near receiving areas, such as near the electronics section; delivery spaces are located relatively close to the distribution center to ensure goods can be quickly transferred to transport vehicles; and temporary storage spaces are located between the receiving area and other cargo areas to temporarily store goods that need processing or are awaiting sorting.

[0081] After determining the functional requirements and relationships of parking spaces, the next step is to determine the semantic annotation parameters of the parking spaces based on this information. These parameters include the function, purpose, equipment compatibility requirements, and spatial distribution of the parking spaces. Specific annotation parameters may include: Parking space function: for example, whether it is a unloading, temporary storage, or delivery parking space; Parking space capacity: how many vehicles or goods each parking space can accommodate; Equipment compatibility: whether the parking space is suitable for the operation of large forklifts, automated handling equipment, etc.; Spatial configuration: the specific location of the parking space, such as whether it is close to the main aisle, unloading area, or cargo area. For example, for delivery parking spaces in an e-commerce logistics center, the semantic annotation parameters may include: Function: delivery parking space; Capacity: each parking space is suitable for parking one large transport truck; Equipment compatibility: suitable for automated loading equipment; Spatial configuration: close to the delivery area for convenient loading and delivery. Once the semantic annotation parameters of the parking spaces are determined, a semantic model of the parking spaces can be generated through spatial positioning and label mapping. This means... By mapping the specific location of each parking space to its labeled parameters, a digital and operational semantic model of the parking space is formed for management and scheduling. This model includes information such as the specific spatial location, function, capacity, and equipment compatibility of each parking space, enabling the logistics center's management system to understand the status and function of each parking space in real time. For example, suppose a logistics distribution site has 10 unloading parking spaces, 5 temporary storage parking spaces, and 8 delivery parking spaces. Through spatial positioning, the specific location of each parking space can be marked (e.g., near the receiving area, far from the exit, etc.), and the semantic labels of each parking space can be mapped, such as: Parking space 1: Unloading parking space, suitable for receiving bulk goods, with forklift operating space; Parking space 2: Temporary storage parking space, with moderate capacity, suitable for short-term parking; Parking space 3: Delivery parking space, suitable for large transport vehicles, equipped with automated loading equipment. In the management system, each parking space has a corresponding label and description, allowing managers to understand the status of each parking space at any time through the system and schedule parking spaces according to demand.

[0082] Through the spatial positioning and label mapping described above, a complete parking space semantic model is obtained. This model integrates various attributes of parking spaces (such as function, capacity, location, equipment compatibility, etc.) and provides them to the logistics center's management system to help optimize parking space scheduling and utilization efficiency. For example, for a parking space management system in an e-commerce logistics center, the parking space semantic model might look like this: Parking Space 1: Function - Unloading; Location - Receiving area; Compatible Equipment - Forklift, conveyor belt; Capacity - 1 large truck; Parking Space 2: Function - Temporary storage; Location - Middle area; Compatible Equipment - None; Capacity - 1 small truck; Parking Space 3: Function - Shipping; Location - Shipping area; Compatible Equipment - Automated handling robot; Capacity - 2 small trucks. In this way, managers can clearly understand the specific function and operational requirements of each parking space, thereby achieving efficient parking space scheduling and utilization.

[0083] In an optional embodiment, a three-dimensional model generation operation is performed on the logistics distribution site based on the cargo area classification model and the parking space semantic model to obtain a three-dimensional distribution site model, including:

[0084] Acquire real-time detailed perception data of the logistics distribution site; the real-time detailed perception data includes three-dimensional structure data of the shelves, location data of sorting equipment, and aisle identification data;

[0085] The model fusion benchmark is determined based on the spatial boundary parameters of the cargo area classification model and the coordinate parameters of the parking space semantic model.

[0086] The real-time detailed perception data is fused and geometrically reconstructed with the cargo area classification model and parking space semantic model according to the model fusion benchmark to obtain a three-dimensional distribution site model of the logistics distribution site.

[0087] It's important to note that before generating the 3D model, real-time information about the distribution area needs to be obtained through real-time detail perception data. This data is typically collected in real-time using devices such as sensors, scanners, and cameras, and includes the following: 3D shelf structure data: recording the spatial layout and height of the shelves; this data reflects the structure of goods storage in the warehouse and helps construct the 3D model of the shelves; Sorting equipment location data: recording the precise location of sorting equipment (such as automated sorting machines, conveyor belts, etc.); these devices play a crucial role in the logistics distribution process, and location data helps determine their function within the entire distribution system. Used in relation to cargo areas and parking spaces; Aisle identification data: records the layout and markings of various aisles (such as forklift aisles, manual walkways, etc.) in the logistics distribution area to help confirm the passage routes within the area; Example: Suppose in an e-commerce logistics warehouse, real-time detailed perception data may include: three-dimensional structural data of shelves: the height of the shelves from the ground to the top, the gap between shelves, the arrangement of shelves, etc.; Sorting equipment location data: the precise coordinates of the automated sorting robots in the warehouse, the running path of the automated conveyor belt, etc.; Aisle identification data: marking employee passage areas, forklift passage areas, their widths, and aisle directions;

[0088] The cargo area classification model and parking space semantic model provide functional and spatial distribution information for the distribution area. The spatial boundary parameters of the cargo area classification model and the coordinate parameters of the parking space semantic model serve as the basis for model fusion. Specifically: the spatial boundary parameters of the cargo area classification model define the spatial boundaries of each cargo area, such as the physical boundaries (walls, pillars, aisles, etc.) of the receiving area, sorting area, and storage area in the warehouse; the coordinate parameters of the parking space semantic model define the specific location, size, and functional information of parking spaces with different functions (such as unloading spaces, temporary storage spaces, and delivery spaces). These data will help determine the spatial relationships between various areas and parking spaces within the site and provide a basis for subsequent 3D modeling fusion. For example, in an e-commerce warehouse, the spatial boundaries of the cargo area classification model might be defined as follows: Receiving area (near the warehouse entrance): from the entrance to the designated shelf area, with a spatial dimension of 20 meters × 10 meters; Sorting area (located in the center of the warehouse): an open space for placing sorting equipment, with a spatial dimension of 30 meters × 25 meters. The parking space semantic model may include parking space coordinate parameters, such as: unloading parking space (near the receiving area): parking space number 1, coordinate (5, 3), capacity suitable for parking one truck; temporary storage parking space (near the sorting area): parking space number 3, coordinate (10, 12), capacity suitable for temporarily storing small trucks.

[0089] Once the model fusion baseline is established, the next step is to fuse real-time detail-aware data with the cargo area classification model and the parking space semantic model. This includes: comparing real-time 3D shelving data with the spatial boundaries of the cargo area classification model to ensure that shelving locations match warehouse space; comparing real-time sorting equipment data with the semantic information of parking spaces to ensure that sorting equipment can work efficiently near specific parking spaces; and comparing aisle signage data with the aisle layout of the site to ensure that logistics aisles are clear and that their relationship with parking spaces and cargo areas is reasonable. By fusing and geometrically reconstructing this data, a 3D distribution site model can be obtained, which can accurately reflect... This reflects the spatial location and functional requirements of various elements within the warehouse, such as shelves, equipment, and parking spaces. For example, in an e-commerce logistics warehouse, real-time data collected by sensors and equipment may include: the three-dimensional structure of shelves: the height, width, and depth of the shelf arrangement, as well as the gaps between shelves; the location of sorting equipment: the real-time location and operating path of sorting robots and conveyor belts in the warehouse; and aisle markings: the location, direction, and width of each aisle. This data will be integrated with the warehouse's cargo area classification model and parking space semantic model to generate a complete three-dimensional distribution site model, helping managers understand the precise layout of various areas and equipment within the site.

[0090] Finally, after data fusion and geometric reconstruction, the resulting 3D model will include: spatial layout of the site: including the spatial distribution of various cargo areas (such as receiving area, sorting area, and storage area) and parking spaces (such as unloading parking spaces, temporary storage parking spaces, and delivery parking spaces); real-time status data: including real-time monitoring of shelf and equipment locations, aisle information, etc., the model can reflect the current status of the distribution site in real time; dynamic adjustment capability: based on the 3D model, the layout of parking spaces or equipment can be dynamically adjusted according to needs to optimize the distribution process; Example: Suppose that in the 3D model of an e-commerce logistics warehouse, there are functional areas such as receiving area, sorting area, and delivery area, and the specific locations of shelves, sorting equipment, and parking spaces are clearly marked in the model; For example: Receiving area: a large area that can accommodate multiple unloading parking spaces and shelves; Sorting area: equipped with automated sorting equipment, the locations of sorting machines and conveyor belts are already shown in the model; Delivery area: contains multiple delivery parking spaces, suitable for large transport vehicles; Through this 3D model, managers can easily view the layout of the entire warehouse, the current operating status, and how to optimize the relationship between parking spaces and cargo areas.

[0091] In an optional embodiment, determining the dynamic monitoring parameters corresponding to the three-dimensional distribution site model includes:

[0092] Obtain the cargo area operation feature parameters of the cargo area classification model and the parking space usage feature parameters of the parking space semantic model; among them, the cargo area operation feature parameters include the cargo area busy period parameters and cargo throughput parameters, and the parking space usage feature parameters include the parking space occupancy rate parameters and the parking space turnover efficiency parameters.

[0093] Based on the operational characteristic parameters of the cargo area and the usage characteristic parameters of the parking spaces, the operationally sensitive areas and key monitoring areas of the logistics distribution site are determined, and the first area attribute parameters of the operationally sensitive areas and the second area attribute parameters of the key monitoring areas are determined.

[0094] Based on the attribute parameters of the first region and the attribute parameters of the second region, the dynamic monitoring parameters corresponding to the three-dimensional distribution site model are determined; among them, the dynamic monitoring parameters include the sensor deployment parameters, data acquisition frequency parameters, anomaly judgment threshold parameters, and model update cycle parameters.

[0095] It should be noted that the cargo area classification model and parking space semantic model provide key information about site operation, including: Cargo area busy period parameters: describing the usage frequency and workload of certain areas during different time periods; for example, the receiving area may become particularly busy during certain periods due to a large influx of goods; Cargo throughput parameters: referring to the capacity of each cargo area in the warehouse to handle goods; for example, the number of items processed per hour, or the loading and unloading capacity of each shelf; Parking space occupancy rate parameters: indicating the utilization of parking spaces in the warehouse, especially during unloading or loading; excessively high occupancy rates may indicate insufficient parking spaces, affecting sorting efficiency; Parking space turnover efficiency parameters: referring to the time required for a vehicle to leave from entering the unloading area, reflecting the utilization efficiency of parking spaces and the smoothness of flow within the site; Example: Suppose that in an e-commerce warehouse, the busy period for the receiving area typically occurs between 8 pm and 10 pm, with a cargo throughput of 200 orders per hour; the parking space occupancy rate typically reaches 80% during the morning peak period, while the turnover efficiency is low (e.g., it takes an average of 1 hour for each vehicle to enter and exit the unloading area).

[0096] Based on the above characteristic parameters, we can determine which areas are most critical to the efficiency of logistics operations. These areas are called operationally sensitive areas and key monitoring areas. Operationally sensitive areas: These areas have a significant impact on the normal operation of the distribution center and are usually key locations in the cargo area or parking spaces. For example, if an area has a high throughput or high parking space occupancy rate, it may be an operationally sensitive area. Key monitoring areas: These areas require more frequent monitoring because they may experience efficiency bottlenecks or potential operational problems. These areas may require denser equipment deployment and more accurate data collection. Examples: Operationally sensitive area: In the receiving area, during the busy period from 8 pm to 10 pm, due to increased throughput, it may become a sensitive area. If the sorting and warehousing speed is slow at this time, it will affect the overall logistics efficiency. Key monitoring area: The area around the unloading parking spaces may be a key monitoring area because of high parking space occupancy and low turnover efficiency. Monitoring these parking spaces can help identify bottlenecks in a timely manner and improve parking space utilization.

[0097] Based on the characteristics of the sensitive operational area (Area 1) and the key monitoring area (Area 2), dynamic monitoring parameters are determined. These parameters will help with real-time monitoring and respond to potential problems. Sensing device deployment parameters: determine the number and location of sensors, cameras, RFID devices, etc., placed in different areas; for example, deploying more sensors in the sensitive operational area and key monitoring area ensures accurate capture of dynamic information within the site. Data acquisition frequency parameters: set the data acquisition frequency, determining the interval at which the system acquires data from the site; for example, parking space occupancy and cargo throughput data may need to be collected every minute for real-time understanding; while other less sensitive parameters (such as ambient temperature) may only need to be collected once per hour. Anomaly detection threshold parameters: set thresholds to determine when the system should issue an alarm or make adjustments; for example, when parking space occupancy exceeds 90%, the system can automatically trigger an alarm to remind staff to make adjustments. Model update cycle parameters: determine how to update the 3D site model based on real-time data, ensuring that changes in site layout and equipment are reflected in the model in a timely manner; for example, updating parking space and cargo area usage data daily, or adjusting in real-time as needed. The system updates parameters hourly. Examples include: Sensor deployment parameters: Installing more cameras and sensors in key locations of the receiving and sorting areas to ensure real-time monitoring of goods flow; placing RFID sensors around parking spaces to monitor occupancy; Data collection frequency parameters: Collecting data on cargo throughput and parking space occupancy rate every minute, as these areas are crucial to logistics speed; collecting other data such as environmental monitoring every hour; Anomaly detection threshold parameters: Setting a parking space occupancy rate threshold of 90%; once this value is exceeded, the system will automatically alert and suggest rescheduling; setting a cargo throughput threshold of 500 pieces / hour; when throughput falls below this value, the system will remind managers to intervene manually; Model update cycle parameters: The warehouse management system updates the warehouse's 3D model daily, adjusting it based on daily usage data (such as parking space occupancy rate and cargo throughput); real-time adjustments are performed hourly to ensure the site layout adapts to actual needs. Through these dynamic monitoring parameters, the entire logistics distribution site can be monitored and managed in real-time; managers can make decisions based on monitoring data at any time and respond quickly to anomalies, thereby improving logistics distribution efficiency and reducing potential bottlenecks.

[0098] In an optional embodiment, the planarable cargo area is divided and its attributes are modeled based on cargo area classification feature parameters to obtain a cargo area classification model for the logistics distribution site, including:

[0099] Preprocessing is performed on real-time initial sensing data; noise reduction and ground point filtering are performed on lidar point cloud data to obtain three-dimensional contour data of the site; feature extraction and distortion correction are performed on visual camera image data to obtain cargo area identification and equipment location data; deduplication and coordinate matching are performed on the initial reading data of RFID tags to obtain initial cargo distribution data.

[0100] The preprocessed site 3D contour data, cargo area identification, equipment location data, and initial cargo distribution data are mapped according to the cargo area classification dimension to generate cargo area classification feature parameters.

[0101] Hierarchical clustering algorithm was used to perform cluster analysis on the classification feature parameters of cargo areas, resulting in multiple cargo area clusters;

[0102] Based on the characteristic attributes of cargo area clusters and the spatial constraints of the planarable cargo area range, the boundaries of each cargo area cluster are divided and attribute values ​​are assigned to construct a cargo area classification model.

[0103] It's important to note that real-time sensing data comes from various sensors and devices, including LiDAR, visual cameras, and RFID tags. This data requires preprocessing before it can be effectively used for analysis and modeling. LiDAR creates point cloud data in three-dimensional space by emitting lasers and receiving reflected signals. Preprocessing operations include: Noise reduction: LiDAR point cloud data may be affected by environmental noise (such as weather, obstacles, etc.); noise reduction algorithms (such as statistical filtering, mean filtering, etc.) remove unnecessary noise points; Ground point filtering: LiDAR point cloud data may contain a large number of ground points, which are useless for building a 3D model of the warehouse area; ground point filtering algorithms remove these irrelevant points, retaining only the valid site contour data. Example: Suppose a warehouse area scanned by LiDAR generates some additional noise points (such as reflected light from chandeliers at a high place). The noise reduction algorithm removes these noise points and also removes ground points (such as debris on the ground); the final 3D contour data more accurately reflects the structure of the warehouse area.

[0104] Visual cameras can acquire image data from a warehouse to identify storage area markers and equipment locations. Preprocessing operations include: Feature extraction: Using computer vision algorithms (such as edge detection and object recognition), key features such as storage area markers, shelf locations, and equipment locations are extracted from the image; Distortion correction: Due to camera lens distortion, the positions of objects in the image may be inaccurate. Distortion correction algorithms can correct these distortions, restoring the true positions of objects. Example: An image acquired by a camera shows partial distortion of shelf locations (due to the wide-angle effect of the lens). After distortion correction, the position of each shelf is accurately identified. Location and marking of shipping areas; RFID (Radio Frequency Identification) technology is used to track the location and distribution of goods; the initial RFID data may contain duplicate tag readings; preprocessing operations include: deduplication: removing duplicate tag readings to ensure that the RFID tag data for each item is unique; coordinate matching: matching the coordinates of the RFID tag with the 3D model of the site to obtain the actual location of each item; example: some items' RFID tags are read by multiple readers simultaneously, resulting in duplicate data; after deduplication, the system accurately obtains the unique coordinates of each item and maps them to the 3D model of the warehouse;

[0105] After preprocessing, feature mapping can begin, generating cargo area classification feature parameters based on different classification dimensions. The purpose of this step is to categorize and map all data (site 3D contours, equipment locations, cargo area identifiers, etc.) according to different dimensions. Feature mapping involves merging and mapping the preprocessed data, generating corresponding feature parameters according to cargo area classification dimensions (e.g., cargo area purpose, turnover frequency, cargo type, etc.). For example, within a cargo area, some areas are for receiving goods, while others are for storing goods. Through feature mapping, the purpose and turnover frequency of each area are identified; for instance, the receiving area is a high-frequency turnover area, and the storage area is a low-frequency turnover area.

[0106] Hierarchical clustering algorithms are used to cluster cargo areas based on feature parameters, generating multiple cargo area clusters. Hierarchical clustering helps to classify cargo areas into different classes based on their similarity, with each class containing cargo areas with similar characteristics. Hierarchical clustering: Based on the feature parameters of cargo areas (such as turnover frequency, purpose, and cargo type), similar cargo areas are grouped into one class. Example: Using hierarchical clustering algorithms, the system separates frequently used cargo areas (such as receiving areas and sorting areas) from less frequently used areas (such as storage areas). Receiving areas may be clustered into one cluster, while sorting areas and storage areas may be assigned to different clusters.

[0107] Based on the characteristic attributes of the cargo area clusters and the spatial constraints of the planarable cargo area range, each cargo area cluster is delineated and its attributes are assigned. This step is crucial for constructing the cargo area classification model. Boundary delineation: Based on the clustering results and the spatial layout of the site, the boundaries of different cargo areas are delineated. Attribute assignment: Based on the characteristics of each cargo area, different attributes are assigned, such as storage capacity, processing capacity, and cargo type. Example: In a warehouse, the clusters in the receiving area may require larger spatial boundaries and high throughput, while the clusters in the storage area may have smaller spatial boundaries and low throughput. Through boundary delineation and attribute assignment, specific attribute models for different areas such as the receiving area, sorting area, and storage area are finally obtained.

[0108] In an optional embodiment, parking spaces in the logistics distribution area are spatially located and labeled according to parking space semantic annotation parameters to obtain a parking space semantic model including parking space semantic information, including:

[0109] Parse the semantic annotation parameters of parking spaces to generate corresponding semantic tag sets and spatial mapping rules; the semantic tag set includes parking space number tags, function type tags, carrying capacity limit tags and associated cargo area ID tags, and the spatial mapping rules include conversion rules between parking space coordinates and cargo area coordinates as well as binding rules between tags and the geometric center of parking spaces;

[0110] The precise three-dimensional coordinates of each parking space are determined by constructing a three-dimensional point cloud cluster based on LiDAR point cloud data and combining the coordinate transformation rules in the spatial mapping rules.

[0111] The tags in the semantic tag set are associated with the precise three-dimensional coordinates of the corresponding parking spaces according to the binding rules, and the association relationship between the tags and the coordinates is stored to form a parking space semantic model.

[0112] It should be noted that the semantic tag set is a collection describing various attributes of a parking space, typically including the following tags: Parking space number tag: Each parking space has a unique number for easy identification and management; Function type tag: The purpose or function of the parking space, such as loading area parking space, unloading area parking space, storage parking space, etc.; Maximum load capacity tag: Describes the maximum weight or volume limit that the parking space can bear; Associated cargo area ID tag: Identifies the association between the parking space and a specific cargo area, indicating which goods will use the parking space;

[0113] Spatial mapping rules are used to associate the physical location of parking spaces with semantic labels. They mainly include: conversion rules between parking space coordinates and cargo area coordinates: parking space coordinates and cargo area coordinates may not be in the same coordinate system, so a conversion rule needs to be established to accurately map the parking space coordinates to the cargo area coordinate system; and binding rules between labels and the geometric center of the parking space: each parking space has a geometric center, and there is a one-to-one correspondence between labels and geometric centers; therefore, it is necessary to clarify the binding rules between each label and the geometric center of the parking space. Example: Suppose there is a logistics site with parking space number 101, a function type label of unloading area parking space, a maximum load capacity of 5 tons, and this parking space is associated with cargo area A; the spatial mapping rules include how to correspond parking space number 101 with its position in the cargo area coordinate system, ensuring that the three-dimensional location of the parking space can be accurately found through this rule.

[0114] LiDAR can be used to scan logistics sites and generate 3D point cloud data of parking spaces. Point cloud data consists of spatial data points generated by the LiDAR device emitting laser beams and receiving reflected signals. Parking space 3D point cloud clusters: Parking space point cloud clusters generated by LiDAR scanning represent the physical location and shape of parking spaces, and are usually an aggregation of multiple points in three-dimensional space.

[0115] Since the parking space coordinates and cargo area coordinates provided by LiDAR may belong to different coordinate systems, spatial mapping rules are needed to transform the point cloud data obtained by LiDAR scanning into the cargo area coordinate system. For example, the LiDAR scanning data is based on a certain origin, but this point cloud data needs to be mapped to the cargo area coordinate system in order to interface with the cargo area management system. Example: Suppose the parking space point cloud data is generated based on a certain local coordinate system, while the cargo area coordinate system is different. Through coordinate transformation rules, the parking space point cloud data can be converted from the local coordinate system to the cargo area coordinate system, so that the actual location of the parking space can be correctly displayed on the cargo area map.

[0116] The geometric center of a parking space is usually the centroid of its shape and is the representative point of the parking space in three-dimensional space. By binding semantic tags to the geometric center of the parking space, accurate identification of the parking space can be achieved. Association operation: For each parking space, the precise three-dimensional coordinates of the parking space are obtained through spatial mapping rules and LiDAR point cloud data. Then, the semantic tags of the parking space (such as parking space number, function type, carrying capacity limit, associated cargo area ID, etc.) are bound to its geometric center.

[0117] The mapping relationship between semantic tags and parking space coordinates is stored in the database to form a parking space semantic model. This model can help the management system obtain specific information about each parking space in real time, including location, function, and carrying capacity. For example, through binding rules, the semantic tag of parking space number 101 will be associated with the three-dimensional coordinates of the parking space, such as (x=15.3, y=8.7, z=0.5). This means that when querying parking space number 101, the system can simultaneously provide information such as the coordinates, function (unloading area parking space), and carrying capacity limit (5 tons) of the parking space.

[0118] In an optional embodiment, dynamic monitoring of the logistics distribution site is performed based on dynamic monitoring parameters and real-time operational data, including:

[0119] Based on the data acquisition frequency parameters in the dynamic monitoring parameters, real-time operational data of the logistics distribution site is collected through a pre-set cluster of sensing devices; among which, real-time operational data includes cargo inventory data in the cargo area, vehicle parking data in parking spaces, equipment operating status data, and personnel flow data.

[0120] Perform anomaly detection on real-time running data; compare the real-time running data with the anomaly judgment threshold parameters in the dynamic monitoring parameters; if a certain data exceeds the corresponding threshold, it is judged as abnormal data, and the corresponding abnormal location and anomaly type are marked.

[0121] The 3D distribution site model is dynamically updated based on real-time operational data and abnormal data.

[0122] It should be noted that, based on the data acquisition frequency parameters in the dynamic monitoring parameters, the preset sensing device cluster (such as sensors, cameras, RFID devices, etc.) will collect real-time operating data of the logistics distribution site at set time intervals (such as per second, per minute, etc.); the sensing device cluster can cover various key areas of the site, such as cargo areas, parking spaces, equipment, and personnel activity areas.

[0123] Cargo inventory data for each cargo area: This data monitors the quantity, type, and storage location of goods in each cargo area, helping to track the distribution of goods in real time; Parking data for each parking space: This monitors the parking status of vehicles in each parking space, including whether it is occupied, the type of vehicle parked, and the parking time; Equipment operation status data: This monitors the operation status of equipment related to the site (such as conveyors, forklifts, stacker cranes, etc.) to ensure normal operation of the equipment; Personnel flow data: By monitoring personnel activities, this ensures that the movement and working status of personnel in the site comply with regulations and procedures, avoiding personnel lingering or unsafe behavior; Example: Suppose a cluster of sensing devices is set up with a data collection frequency of once every 5 seconds. The devices will automatically record whether each parking space in the site is occupied, record the cargo inventory of a certain cargo area (e.g., cargo area A: 30 boxes of electronic products), and collect the operation status of equipment (e.g., stacker cranes are operating normally) and personnel flow (e.g., personnel X in cargo area B) every 5 seconds;

[0124] After data collection, the system performs real-time anomaly detection. This detection process is based on preset anomaly threshold parameters in the dynamic monitoring parameters. These threshold parameters are set according to normal operational standards; once the actual data exceeds the threshold range, an anomaly alarm is triggered. Anomaly threshold parameters: For example, the inventory of goods in a certain cargo area should normally be between 50-100 boxes. If the inventory of goods in cargo area A is found to be below 50 boxes at a certain moment, the system will consider this an anomaly and require further processing. Anomaly types: Anomaly data can be of various types, such as inventory anomalies, vehicle parking anomalies, equipment malfunctions, or personnel anomalies. Example: Suppose the set inventory anomaly threshold is below 30 boxes. If the inventory monitoring data for a certain cargo area shows that cargo area A has only 28 boxes, the system will automatically mark this data as anomaly and label it as an inventory anomaly. If the system monitors that the parking time of a vehicle in a parking space exceeds a set threshold (e.g., the maximum parking time for each parking space is 2 hours), and the parking space is still not released, the system will mark this data as a vehicle parking anomaly.

[0125] Based on real-time collected operational data and anomaly detection results, the system dynamically updates the 3D distribution site model. This 3D model is a digital representation of the site, containing the spatial layout of various elements such as cargo areas, parking spaces, and equipment, as well as the real-time status of these elements. Dynamic updates mean that when the system detects anomalies, the model will automatically reflect these changes. For example, when the inventory of goods in a cargo area falls below a threshold, the model will mark that cargo area as an anomaly in the digital display. The model updates can also reflect changes in the actual physical environment, such as the parking situation in parking spaces, the operating status of equipment, and the location of personnel. This helps managers obtain accurate site information in real time and take timely countermeasures. Example: Suppose the system monitors that the inventory of goods in cargo area A drops below 30 boxes, and the vehicle parking status is abnormal (parking time in parking space C exceeds 2 hours); at this time, the system will automatically mark it in the 3D distribution site model: the inventory area of ​​cargo area A is marked in red, indicating abnormal inventory; parking space C is displayed in an abnormal state, possibly marked with a different color (such as yellow or red), indicating that the parking space has been used for too long; when the management personnel view the model, they will immediately know that there are problems with cargo area A and parking space C, and can take corresponding actions, such as adjusting the replenishment of goods or arranging parking space clearing.

[0126] In an optional embodiment, the dynamic monitoring operation of the logistics distribution site based on dynamic monitoring parameters and real-time operating data further includes:

[0127] According to the model update cycle parameters in the dynamic monitoring parameters, the cargo inventory data of the cargo area and the vehicle parking data of the parking space are synchronized to the corresponding positions in the three-dimensional distribution site model, and the abnormal locations and abnormal types are visually marked in the three-dimensional distribution site model.

[0128] Output the updated 3D distribution site model and anomaly warning information.

[0129] It should be noted that the model update cycle parameter refers to how often the system automatically updates the 3D distribution site model. This parameter controls the frequency of data synchronization and model updates. Assuming a model update cycle of 30 minutes is set, the system will synchronize current cargo inventory data, parking space usage, equipment status, and other data to the 3D model at the end of each cycle. At the end of each model update cycle, the system will synchronize this data to the corresponding locations in the 3D distribution site model based on real-time collected data. Specifically, this includes cargo inventory data for each cargo area: for example, if the inventory in cargo area A decreases from 100 boxes to 80 boxes, the system will update this change to the corresponding cargo area location in the 3D model. The number at that location will be displayed as 80 boxes; Parking data: For example, parking space 1 currently has a truck parked, while parking space 2 is vacant; the system will mark parking space 1 as parked and parking space 2 as vacant; Specific example: Assuming the model update cycle is every 30 minutes, at the end of an update cycle, the real-time data report shows: the inventory of goods in cargo area A has decreased from 100 boxes to 80 boxes; there is a truck parked in parking space 1, and parking space 2 is vacant; at this time, the system will synchronize these changes to the 3D distribution site model: the corresponding location in cargo area A (which may be a three-dimensional shelf or cargo area in the 3D view) will display the updated 80 boxes; the corresponding location in parking space 1 will be displayed as parked, and parking space 2 will be displayed as vacant;

[0130] When the system detects certain anomalies (such as abnormal inventory levels or parking space usage), these anomalies will be visually marked in the 3D model. Specifically: For inventory anomalies: when the inventory level of a certain cargo area falls below a set threshold (e.g., below 50 boxes), that cargo area will be marked red in the 3D model, indicating an inventory anomaly. For vehicle parking anomalies: for example, if a vehicle parked in a parking space exceeds the allowed parking time (e.g., 2 hours), that parking space will be marked yellow in the 3D model to alert management personnel. For example: suppose the system detects the following anomalies: the inventory level in cargo area B drops to 40 boxes, below the set threshold of 50 boxes; the parking time in parking space 3 exceeds 2 hours; in the 3D distribution site model: cargo area B will turn red, with an "Inventory Anomaly: 40 Boxes" label next to it; parking space 3 will turn yellow, with a "Parking Time Exceeded: Over 2 Hours" label. This marking method helps management personnel quickly locate and handle anomalies.

[0131] At the end of each update cycle, the system outputs an updated 3D distribution site model and anomaly warning information. This information helps managers understand the site's operational status and potential problems in real time, supporting subsequent decision-making and optimization. The updated 3D distribution site model is an updated digital model displaying the inventory of cargo areas, parking space usage, equipment status, and personnel distribution. All real-time data and anomaly information are synchronized to the model. Anomaly warning information: When the system detects an anomaly, it automatically generates a warning report listing the anomaly's type, location, severity, and other detailed information, and pushes it to relevant managers or operators. For example, suppose that after the update cycle ends, the system outputs the following information: The updated 3D model shows the status of cargo areas A and B and parking space 1. Cargo area B is marked in red, and parking space 3 is marked in yellow. Anomaly warning information includes: Cargo area B inventory anomaly: inventory is below 50 boxes, current inventory is 40 boxes; Parking space 3 parking anomaly: parking time exceeds 2 hours. This information will be compiled into a report and sent to relevant personnel, such as warehouse managers and dispatchers, to help them take timely measures.

[0132] Example 2, please refer to Figure 2 This invention provides a technical solution: a logistics distribution site monitoring system based on real-time data fusion, applicable to the aforementioned logistics distribution site monitoring method based on real-time data fusion, comprising:

[0133] Cargo area classification unit 1 is used to acquire multi-source basic data of the logistics distribution site. Based on the multi-source basic data, cargo area classification modeling operation is performed on the logistics distribution site to obtain the cargo area classification model of the logistics distribution site. Among them, the multi-source basic data includes site static topology data and real-time initial sensing data.

[0134] Semantic annotation unit 2 is used to determine the semantic annotation parameters of parking spaces corresponding to the cargo area classification model. Based on the semantic annotation parameters of parking spaces, semantic annotation operations are performed on the parking spaces in the logistics distribution area to obtain the semantic model of parking spaces in the logistics distribution area. Among them, the semantic annotation parameters of parking spaces include coordinate system mapping parameters, semantic label configuration parameters, parking space-cargo area association parameters, and spatial accuracy calibration parameters.

[0135] Modeling unit 3 is used to generate a three-dimensional model of the logistics distribution site based on the cargo area classification model and the parking space semantic model, so as to obtain a three-dimensional distribution site model of the logistics distribution site.

[0136] Site monitoring unit 4 is used to determine the dynamic monitoring parameters corresponding to the three-dimensional distribution site model, and to perform dynamic monitoring operations on the logistics distribution site based on the dynamic monitoring parameters and real-time operation data.

[0137] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A method for monitoring logistics distribution sites based on real-time data fusion, characterized in that, include: Multi-source basic data of the logistics distribution site is acquired, and a cargo area classification modeling operation is performed on the logistics distribution site based on the multi-source basic data to obtain the cargo area classification model of the logistics distribution site; wherein, the multi-source basic data includes site static topology data and real-time initial sensing data. The semantic annotation parameters of the parking spaces corresponding to the cargo area classification model are determined. Based on the semantic annotation parameters of the parking spaces, the parking spaces of the logistics distribution site are semantically annotated to obtain the semantic model of the parking spaces of the logistics distribution site. The semantic annotation parameters of the parking spaces include coordinate system mapping parameters, semantic label configuration parameters, parking space-cargo area association parameters, and spatial accuracy calibration parameters. Based on the cargo area classification model and the parking space semantic model, a three-dimensional model generation operation is performed on the logistics distribution site to obtain a three-dimensional distribution site model of the logistics distribution site. Determine the dynamic monitoring parameters corresponding to the three-dimensional distribution site model, and perform dynamic monitoring operations on the logistics distribution site based on the dynamic monitoring parameters and real-time operating data. Specifically, based on the cargo area classification model and the parking space semantic model, a three-dimensional model generation operation is performed on the logistics distribution site to obtain a three-dimensional distribution site model, including: Acquire real-time detailed perception data of the logistics distribution site; wherein, the real-time detailed perception data includes three-dimensional structural data of the shelves, location data of sorting equipment, and aisle identification data; The model fusion benchmark is determined based on the spatial boundary parameters of the cargo area classification model and the coordinate parameters of the parking space semantic model. The real-time detail perception data is fused and geometrically reconstructed with the cargo area classification model and the parking space semantic model according to the model fusion benchmark to obtain a three-dimensional distribution site model of the logistics distribution site. The determination of the dynamic monitoring parameters corresponding to the three-dimensional distribution site model includes: Obtain the cargo area operation feature parameters of the cargo area classification model and the parking space usage feature parameters of the parking space semantic model; wherein, the cargo area operation feature parameters include cargo area busy period parameters and cargo throughput parameters, and the parking space usage feature parameters include parking space occupancy rate parameters and parking space turnover efficiency parameters; Based on the cargo area operation characteristic parameters and the parking space usage characteristic parameters, the operation sensitive area and the monitoring key area of ​​the logistics distribution site are determined, and the first area attribute parameters of the operation sensitive area and the second area attribute parameters of the monitoring key area are determined. Based on the attribute parameters of the first region and the attribute parameters of the second region, the dynamic monitoring parameters corresponding to the three-dimensional distribution site model are determined; wherein, the dynamic monitoring parameters include the sensor deployment parameters, data acquisition frequency parameters, anomaly judgment threshold parameters, and model update cycle parameters.

2. The method for monitoring logistics distribution sites based on real-time data fusion according to claim 1, characterized in that, Perform a cargo area classification modeling operation on the logistics distribution site to obtain the cargo area classification model of the logistics distribution site, including: Acquire static topology data and real-time initial perception data of the logistics distribution site; wherein, the static topology data includes site physical size data, building load-bearing structure data and initial cargo area planning data, and the real-time initial perception data includes lidar point cloud data, visual camera image data and RFID tag initial reading data. Based on the static topology data of the site, the planarable cargo area range and cargo area classification dimensions of the logistics distribution site are determined; the cargo area classification dimensions include cargo turnover rate, cargo weight class, cargo timeliness requirements, and cargo storage temperature zone. Based on the real-time initial sensing data and the cargo area classification dimensions, cargo area classification feature parameters within the planarable cargo area range are determined; wherein, the cargo area classification feature parameters include turnover rate distribution parameters, weight load adaptation parameters, timeliness response threshold parameters, and temperature zone demand distribution parameters; Based on the cargo area classification feature parameters, the planned cargo area range is divided into cargo areas and its attributes are modeled to obtain the cargo area classification model of the logistics distribution site.

3. The method for monitoring logistics distribution sites based on real-time data fusion according to claim 2, characterized in that, Based on the parking space semantic annotation parameters, semantic annotation operations are performed on the parking spaces in the logistics distribution area to obtain a parking space semantic model of the logistics distribution area, including: Based on the cargo area classification model, the parking space functional requirements and parking space associations of the logistics distribution site are determined; the parking space functional requirements include unloading parking space requirements, temporary storage parking space requirements, and outgoing parking space requirements; the parking space associations include the spatial association between parking spaces and cargo areas and the adaptation association between parking spaces and equipment. Based on the functional requirements of the parking spaces and the association relationships between the parking spaces, the semantic annotation parameters of the parking spaces are determined; Based on the parking space semantic annotation parameters, the parking spaces in the logistics distribution site are spatially located and labeled to obtain a parking space semantic model that includes parking space semantic information.

4. The method for monitoring logistics distribution sites based on real-time data fusion according to claim 3, characterized in that, Based on the cargo area classification feature parameters, the planarable cargo area range is divided into cargo areas and its attributes are modeled to obtain the cargo area classification model of the logistics distribution site, including: The real-time initial sensing data is preprocessed; the lidar point cloud data is denoised and ground point filtered out to obtain the three-dimensional contour data of the site; the visual camera image data is feature extracted and distortion corrected to obtain cargo area identification and equipment location data; the initial reading data of the RFID tags is deduplicated and coordinate matched to obtain the initial cargo distribution data. The preprocessed site 3D contour data, cargo area identification, equipment location data, and initial cargo distribution data are mapped according to the cargo area classification dimension to generate cargo area classification feature parameters. Hierarchical clustering algorithm is used to perform cluster analysis on the classification feature parameters of the cargo area, resulting in multiple cargo area clusters; Based on the characteristic attributes of the cargo area clusters and the spatial constraints of the planarable cargo area range, the boundaries of each cargo area cluster are divided and attribute values ​​are assigned to construct the cargo area classification model.

5. A method for monitoring logistics distribution sites based on real-time data fusion according to claim 4, characterized in that, Based on the parking space semantic annotation parameters, the parking spaces in the logistics distribution area are spatially located and labeled to obtain a parking space semantic model that includes parking space semantic information, including: The semantic annotation parameters of the parking space are parsed to generate a corresponding semantic tag set and spatial mapping rules; wherein, the semantic tag set includes parking space number tag, function type tag, carrying capacity limit tag and associated cargo area ID tag, and the spatial mapping rules include conversion rules between parking space coordinates and cargo area coordinates and binding rules between tags and the geometric center of the parking space; The precise three-dimensional coordinates of each parking space are determined by constructing a three-dimensional point cloud cluster based on LiDAR point cloud data and combining it with the coordinate transformation rules in the spatial mapping rules. The tags in the semantic tag set are associated and mapped with the corresponding precise three-dimensional coordinates of the parking space according to the binding rules, and the association relationship between the tags and the coordinates is stored to form the parking space semantic model.

6. A method for monitoring logistics distribution sites based on real-time data fusion according to claim 5, characterized in that, Based on the dynamic monitoring parameters and real-time operational data, dynamic monitoring operations are performed on the logistics distribution site, including: Based on the data acquisition frequency parameter in the dynamic monitoring parameters, real-time operational data of the logistics distribution site is collected through a preset cluster of sensing devices; wherein, the real-time operational data includes cargo inventory data in the cargo area, vehicle parking data in parking spaces, equipment operating status data, and personnel flow data; Anomaly detection is performed on the real-time running data; the real-time running data is compared with the anomaly judgment threshold parameter in the dynamic monitoring parameters. If a certain data exceeds the corresponding threshold, it is judged as abnormal data, and the corresponding abnormal location and anomaly type are marked. The three-dimensional distribution site model is dynamically updated based on the real-time operating data and the abnormal data.

7. A method for monitoring logistics distribution sites based on real-time data fusion according to claim 6, characterized in that, The dynamic monitoring operation of the logistics distribution site based on the dynamic monitoring parameters and real-time operating data also includes: According to the model update cycle parameter in the dynamic monitoring parameters, the cargo inventory data of the cargo area and the vehicle parking data of the parking space are synchronized to the corresponding positions of the three-dimensional distribution site model, and the abnormal positions and abnormal types are visually marked in the three-dimensional distribution site model. Output the updated 3D distribution site model and anomaly warning information.

8. A logistics distribution site monitoring system based on real-time data fusion, applicable to the logistics distribution site monitoring method based on real-time data fusion as described in any one of claims 1-7, characterized in that, include: The cargo area classification unit is used to acquire multi-source basic data of the logistics distribution site, and to perform cargo area classification modeling operation on the logistics distribution site based on the multi-source basic data to obtain the cargo area classification model of the logistics distribution site; wherein, the multi-source basic data includes site static topology data and real-time initial sensing data. A semantic annotation unit is used to determine the semantic annotation parameters of parking spaces corresponding to the cargo area classification model. Based on the semantic annotation parameters, semantic annotation operations are performed on the parking spaces of the logistics distribution site to obtain the semantic model of the parking spaces of the logistics distribution site. The semantic annotation parameters of the parking spaces include coordinate system mapping parameters, semantic label configuration parameters, parking space-cargo area association parameters, and spatial accuracy calibration parameters. The modeling unit is used to perform a three-dimensional model generation operation on the logistics distribution site based on the cargo area classification model and the parking space semantic model, so as to obtain a three-dimensional distribution site model of the logistics distribution site. The site monitoring unit is used to determine the dynamic monitoring parameters corresponding to the three-dimensional distribution site model, and to perform dynamic monitoring operations on the logistics distribution site based on the dynamic monitoring parameters and real-time operating data. Specifically, based on the cargo area classification model and the parking space semantic model, a three-dimensional model generation operation is performed on the logistics distribution site to obtain a three-dimensional distribution site model, including: Acquire real-time detailed perception data of the logistics distribution site; wherein, the real-time detailed perception data includes three-dimensional structural data of the shelves, location data of sorting equipment, and aisle identification data; The model fusion benchmark is determined based on the spatial boundary parameters of the cargo area classification model and the coordinate parameters of the parking space semantic model. The real-time detail perception data is fused and geometrically reconstructed with the cargo area classification model and the parking space semantic model according to the model fusion benchmark to obtain a three-dimensional distribution site model of the logistics distribution site. The determination of the dynamic monitoring parameters corresponding to the three-dimensional distribution site model includes: Obtain the cargo area operation feature parameters of the cargo area classification model and the parking space usage feature parameters of the parking space semantic model; wherein, the cargo area operation feature parameters include cargo area busy period parameters and cargo throughput parameters, and the parking space usage feature parameters include parking space occupancy rate parameters and parking space turnover efficiency parameters; Based on the cargo area operation characteristic parameters and the parking space usage characteristic parameters, the operation sensitive area and the monitoring key area of ​​the logistics distribution site are determined, and the first area attribute parameters of the operation sensitive area and the second area attribute parameters of the monitoring key area are determined. Based on the attribute parameters of the first region and the attribute parameters of the second region, the dynamic monitoring parameters corresponding to the three-dimensional distribution site model are determined; wherein, the dynamic monitoring parameters include the sensor deployment parameters, data acquisition frequency parameters, anomaly judgment threshold parameters, and model update cycle parameters.

Images