Intelligent cargo warehouse management system and management method

Abstract

The invention provides an intelligent cargo warehouse management system and management method. The intelligent cargo warehouse management system comprises a data acquisition module, a cargo pick-and-place module, a data transmission and storage module and an intelligent control module. The data acquisition module is used for reading cargo information; the goods taking and placing module is used for taking and placing goods on the goods shelf; and the intelligent control module is used for constructing a warehouse three-dimensional virtual scene, and carrying out optimal path goods taking on the goods in combination with the goods taking and placing module. High automation of the goods warehousing and ex-warehouse process is achieved, and the requirement for manual intervention is reduced. The intelligent control module is deployed in the local server, and the goods storage and taking operation can be automatically executed only by inputting a goods taking order number or issuing a goods storage command, so that the operation process is simplified, and the working efficiency is improved. The system constructs a three-dimensional virtual scene by using a panoramic jigsaw algorithm, and displays the inventory distribution condition in combination with a manifold visualization technology, thereby providing a visual decision support tool for managers, and enhancing the visual management level of the storage state.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent goods warehouse, in particular to an intelligent goods warehouse management system and a management method. BACKGROUND

[0002] With the rapid development of the logistics industry and the continuous improvement of the intelligent level, the traditional warehouse management mode has been difficult to meet the demand of modern warehouse for efficient, safe and accurate management. At present, most warehouses still rely on manual operation combined with local automation, which has low efficiency, high error rate and information opacity, etc. There are obvious shortcomings in the aspects of goods access path planning, inventory visualization, data security, etc. In order to solve the above problems, an intelligent goods warehouse management system with high integration, strong automation and data traceability and the corresponding management method are urgently needed.

[0003] In recent years, technologies such as Internet of Things, big data and artificial intelligence have been gradually applied in the warehouse field, promoting the development of intelligent warehouse systems. For example, radio frequency identification (RFID) technology has been widely used in goods information collection, improving the efficiency of warehouse entry and exit; the automatic scheduling system of the hoist vehicle combines the path optimization algorithm to realize the preliminary automation of the handling process; three-dimensional modeling and visualization technology enhances the perception ability of the warehouse environment and improves the space utilization rate. However, the existing system still has deficiencies in data whole-process management, information security guarantee, multi-device collaborative control, etc. Especially when dealing with high-frequency and large-scale goods flow, the stability, expansibility and security of the system are challenged. SUMMARY

[0004] The main purpose of the present application is to provide an intelligent goods warehouse management system and a management method, which solves the problem that most warehouses still rely on manual operation combined with local automation, which has low efficiency, high error rate and information opacity, etc.

[0005] To solve the above technical problems, the technical solution adopted by the present application is: an intelligent goods warehouse management system, comprising a data acquisition module, a goods taking and placing module, a data transmission and storage module, and an intelligent control module. The data acquisition module is used to read goods information. The goods taking and placing module is used to take and place goods on the shelves. The intelligent control module is used to build a three-dimensional virtual scene of the warehouse, and to take goods by the goods taking and placing module according to the optimal path.

[0006] In the preferred scheme, the goods taking and placing module comprises a hoist vehicle, the hoist vehicle is provided with a plurality of automatically synchronized telescopic fork rods, vertical rail frames are arranged between adjacent shelves, and the hoist vehicle is in sliding connection with the vertical rail frames. The vertical rail frame is slidably connected with the ground rail and the overhead rail at the upper and lower ends respectively, and a gear rack is used to drive the vertical rail frame to slide on the ground rail and the overhead rail. The overhead rail is provided with a winch connected with the hoist. The hoist is further provided with a camera.

[0007] In the preferred embodiment, the plurality of shelves are further provided with a general conveying belt on one side, and a plurality of goods receiving conveying belts are provided on one side of the general conveying belt, and the goods receiving conveying belts are located on one side of the ground rail.

[0008] The method comprises: S1, the data acquisition module configures a radio frequency identification tag containing a two-dimensional code for each warehouse-incoming goods, and acquires goods information by scanning the radio frequency identification tag through a code scanning device; S2, the data collected by the data acquisition module is transmitted to the intelligent control module through a data transmission storage module, and the intelligent control module generates goods taking number and storage location data according to the data collected by the data acquisition module; S3, the intelligent control module further constructs a three-dimensional scene of the warehouse through a panoramic jigsaw algorithm, and combines manifold visualization to display the inventory distribution; S4, according to the three-dimensional scene of the visualization display of the inventory distribution, the hoist is dispatched to adopt a group intelligence algorithm and a time-varying graph network to optimize the carrying path, and the goods are recognized through the vehicle-mounted camera; S5, after recognizing the goods, the intelligent control module drives the hoist to take the goods, the hoist carries the goods to the goods receiving conveying belt, the goods receiving conveying belt carries the goods to the general conveying belt, the data acquisition module collects the goods for delivery, and the taking of the goods is completed; S6, when storing goods, the goods are placed on the goods receiving conveying belt, the goods information is acquired through the data acquisition module, and the storage information is generated, the intelligent control module dispatches the hoist to adopt a group intelligence algorithm and a time-varying graph network to optimize the carrying path, and transports the goods to the storage location and places the goods on the shelf.

[0009] In the preferred embodiment, in step S1, a control instruction is generated according to the goods information; The goods are delivered to the designated shelf position by driving the goods taking and placing module through the control instruction; The goods number and storage location data are acquired in real time; The goods information collected is transmitted to the intelligent control module through the data transmission storage module; The goods information is protected by an encrypted transmission method; The goods information is stored through a distributed storage structure to support fast query; The goods information is used to dynamically update the warehouse inventory status.

[0010] In the preferred solution, in step S2, wireless network or local area network is used for relay transmission, time convolution network is used to predict data delay, and historical trajectory interpolation algorithm is used to correct if the transmission data is inconsistent. The transmission data is encrypted using an elliptic curve encryption algorithm, and the cloud storage supports ciphertext calculation using homomorphic encryption technology. The real-time data collected by the data acquisition module is stored in the key-value forest database, supporting millisecond-level range queries. The historical data collected by the data acquisition module is stored in the distributed database through a hierarchical index tree structure, and the time-space index block is generated according to the time granularity.

[0011] In the preferred solution, in step S3, multiple cameras are arranged inside the warehouse, and 3D modeling is performed on the shelves (5), cranes (1), and ground rails (2) using three-dimensional software, and the model is exported in OBJ format. The OBJ format model and multiple camera visual data construct a three-dimensional virtual scene, and the method is as follows: The intelligent control module constructs a three-dimensional virtual scene of the warehouse using panoramic puzzle algorithm, and the intrinsic matrix of multiple cameras inside the warehouse is obtained through Zhang Zhengyou calibration method and extrinsic matrix RANSAC algorithm is used to remove matching error points, and the image stitching is completed through homography matrix The shelf and crane device model built by the three-dimensional software is aligned and registered with the camera visual data through ICP algorithm, and real-time rendering is performed using WebGL technology. The manifold visualization technology is used to dynamically display the inventory distribution, the high-dimensional inventory data is mapped to a two-dimensional manifold space and color-coded through t-SNE algorithm, and the warehouse heat map is generated based on Gaussian kernel density estimation.

[0012] In the preferred solution, in step S4, a group intelligence algorithm is used to schedule the cranes in the goods taking and placing module. The carrying path is optimized through the time-varying graph network. According to the carrying path, the goods taking and storing operation is performed. The camera on the crane is used to identify the position and category of the goods to confirm the operation accuracy.

[0013] In the preferred solution, the intelligent control module schedules the cranes in the goods taking and placing module through a group intelligence algorithm, and the ant colony optimization algorithm is used to assign tasks to multiple cranes. The ant colony optimization algorithm is used to calculate the task allocation probability, and the artificial potential field algorithm is combined to prevent collision. The formula of the ant colony optimization algorithm is as follows: ; Wherein, represents the path at time t The pheromone concentration on the path, the higher the pheromone concentration, the greater the probability of path selection; As heuristic information, it is usually inversely proportional to the distance between two points, indicating the expected degree of selecting the path; And is a weight coefficient, respectively controlling the influence degree of pheromone and heuristic information on the selection probability; The artificial potential field algorithm formula is: ; Wherein, is the repulsive force coefficient, represents the distance vector between the crane and the crane ; Combined with the time-varying graph network to optimize the carrying path, a time-varying graph model is constructed to represent the edge weight, and the TD-Dijkstra algorithm is used to solve the optimal path and through the cubic spline interpolation smoothing processing; Wherein, represents the weight of the edge between node and at time . Among them, is the actual distance between two points, is a dynamic adjustment factor that changes with time, which is used to reflect the influence of real-time conditions on the path weight; According to the optimal path to pick up and store goods, in the picking process, the camera on the crane identifies the position and category of the goods through the YOLOv5 algorithm, and executes the picking when the information is consistent with the goods to be picked. When it is inconsistent, sound and light alarm is triggered, and the crane is remotely operated by manual according to historical data through tracing the historical position of the goods, and the alarm is triggered again when the goods are not detected.

[0014] In the preferred scheme, in steps S5-S6, during the process of storing and picking the goods, the intelligent control module receives the data collection module of the goods and storage data; The intelligent control module uses the directed acyclic graph technology of block chain to record the whole process information of the goods data.

[0015] The present application provides a kind of wisdom goods warehouse management system and management method, wisdom goods warehouse management system and management method bring many beneficial effects, first, the combination application of radio frequency identification tag and code scanning equipment greatly improves the efficiency and accuracy of goods information acquisition, realizes the high automation of goods warehousing and delivery process, reduces the need for manual intervention.Intelligent control module is deployed in local server, only need to input the goods receipt number or issue storage command can automatically execute storage and retrieval operation, simplify the operation process, improve work efficiency.The system constructs three-dimensional virtual scene using panoramic puzzle algorithm, and combines with manifold visualization technology to show inventory distribution, provides intuitive decision support tool for managers, enhances the visualization management level of warehouse state.

[0016] Overall, the system not only significantly improves the intelligent level of warehouse management and reduces operating costs, but also meets the needs of high-security industries by improving data security and transparency, laying a solid foundation for the development of future intelligent logistics systems.At the same time, the remaining manual monitoring mechanism also ensures that unexpected situations can be handled in addition to automated operations, ensuring stable operation of the system. BRIEF DESCRIPTION OF DRAWINGS

[0017] The present application will be further described below in conjunction with the drawings and examples: Fig. 1 is the warehouse rack hoist car step schematic diagram of the present application; Fig. 2 is the warehouse overhead layout schematic diagram of the present application.

[0018] In the figure: hoist car 1; ground rail 2; fork 3; vertical rail frame 4; rack 5; overhead rail 6; winch 7; goods receiving conveyor belt 8; total conveyor belt 9. DETAILED DESCRIPTION

[0019] Example 1 As shown in Figs. 1-2 , a kind of wisdom goods warehouse management system, including data acquisition module, goods taking and placing module, data transmission storage module, intelligent control module; Data acquisition module is used to read goods information; Goods taking and placing module is used to take and place goods on rack (5); Intelligent control module is used to construct warehouse three-dimensional virtual scene, and combine goods taking and placing module to take goods on optimal path.

[0020] First, configure radio frequency identification tag containing two-dimensional code for each warehousing goods by data acquisition module, and scan the label using code scanning equipment to obtain goods information; Subsequently, the data acquisition module transmits the collected data to the intelligent control module through the data transmission storage module, and the intelligent control module generates the goods taking number and storage location data according to the data; Meanwhile, the intelligent control module constructs a three-dimensional virtual scene of the warehouse through panoramic puzzle algorithm and displays the inventory distribution by combining with manifold visualization technology. Based on the visualized inventory distribution information in the three-dimensional scene, the system dispatches the forklift to optimize the carrying path by using swarm intelligence algorithm and time-varying graph network, and identifies the target goods through the vehicle-mounted camera. When the goods are identified, the intelligent control module drives the forklift to take out the goods and transport them to the receiving transmission belt, and then the receiving transmission belt transports the goods to the general transmission belt. At this time, the data acquisition module collects the goods for delivery, completing the entire goods taking process. When storing goods, the goods are first placed on the receiving transmission belt, and the data acquisition module scans them to obtain the goods information and generate the storage information. Then the intelligent control module dispatches the forklift again, optimizes the carrying path by using the same swarm intelligence algorithm and time-varying graph network, and transports the goods to the designated shelf location for storage.

[0021] Through the combination of the data acquisition module and the radio frequency identification tag, efficient reading of goods information is realized, and the information collection efficiency in the warehousing and delivery process is improved. The three-dimensional virtual scene of the warehouse constructed by the intelligent control module can intuitively display the inventory distribution state, which helps to improve the visualization level of warehouse management. Through the swarm intelligence algorithm and time-varying graph network optimization technology, the carrying path of the forklift is effectively planned, the time cost in the transportation process is reduced, and the overall logistics efficiency is improved. In addition, the application of the vehicle-mounted camera enhances the accuracy of goods identification and reduces the need for manual intervention, further promoting the intelligentization process of warehouse management.

[0022] In the preferred scheme, the goods taking and placing module includes a forklift 1, a plurality of automatic synchronous telescopic goods forks 3 are arranged on the forklift 1, vertical track frames 4 are arranged between adjacent shelves 5, and the forklift 1 is slidingly connected with the vertical track frames 4. The upper and lower ends of the vertical track frames 4 are slidingly connected with the ground rails 2 and the overhead rails 6 respectively, and a gear and rack is used to drive the vertical track frames 4 to slide on the ground rails 2 and the overhead rails 6. An elevator 7 is arranged on the overhead rails 6, and the elevator 7 is connected with the forklift 1. A camera is further arranged on one side of the forklift 1.

[0023] In the preferred scheme, a general transmission belt 9 is further arranged on one side of the plurality of shelves 5, a plurality of receiving transmission belts 8 are arranged on one side of the general transmission belt 9, and the receiving transmission belts 8 are located on one side of the ground rails 2.

[0024] When in use, the winch 7 can lift the trolley 1 to a suitable height through the connection, the vertical rail frame 4 slides on the ground rail 2 and the overhead rail 6 by means of the rack and pinion drive, and drives the trolley 1 to move to the designated position between adjacent shelves 5. The camera on the trolley 1 assists in positioning the goods, and multiple automatically synchronized telescopic fork rods 3 are extended to insert into the bottom of the goods, completing the picking action; when goods need to be put down, the fork rods 3 are retracted after transporting the goods to the designated position. When the goods need to be transported out of the warehouse, the trolley 1 can place the goods on the goods receiving conveyor 8, and then the goods are uniformly transported to the subsequent link through the general conveyor 9.

[0025] The cooperation of the trolley 1 with the vertical rail frame 4, the ground rail 2 and the overhead rail 6 realizes the flexible movement of the trolley 1 in the three-dimensional space, greatly expands the operation range, and can efficiently cover all shelves 5 in the warehouse. The automatically synchronized telescopic fork rods 3 can simultaneously operate multiple goods, improve the number of goods picked or put down at a time, and improve the operation efficiency, and the synchronous telescopic function ensures the stability and accuracy of the picking and putting down process. The rack and pinion drive sliding of the vertical rail frame 4 is stable, reliable in power transmission, accurate in positioning, and can avoid the influence of the movement deviation of the trolley 1 on the operation. The winch 7 is connected with the trolley 1, which is convenient for lifting the trolley 1 and meets the operation needs of different height shelves 5. The camera on one side of the trolley 1 assists the operator or the intelligent system to accurately identify the position and state of the goods, reducing the operation mistakes. The setting of the general conveyor 9 and the goods receiving conveyor 8 constructs an efficient goods conveying channel, realizes the rapid transportation of goods from the shelves 5 to the warehouse link, further optimizes the warehouse operation process, reduces the manual carrying cost, and improves the overall warehouse logistics operation efficiency.

[0026] Embodiment 2 S1, the data acquisition module configures a radio frequency identification tag containing a two-dimensional code for each warehousing goods, and acquires the goods information by scanning the radio frequency identification tag through a code scanning device; S2, the data collected by the data acquisition module is transmitted to the intelligent control module through the data transmission and storage module, and the intelligent control module generates the picking goods number and the storage position data according to the data collected by the data acquisition module; S3, the intelligent control module also constructs a three-dimensional scene of the warehouse through a panoramic jigsaw algorithm, and combines the manifold visualization to display the inventory distribution; S4, according to the three-dimensional scene of the visualization display of the inventory distribution, the trolley adopts a group intelligent algorithm and a time-varying graph network to optimize the carrying path, and identifies the goods through the vehicle-mounted camera; S5, after identifying the goods, the intelligent control module drives the trolley 1 to pick the goods, the trolley 1 transports the goods to the goods receiving conveyor 8, the goods receiving conveyor 8 transports the goods to the general conveyor 9, the data acquisition module collects the goods for delivery, and the picking is completed; S6, when storing goods, the goods are placed on the receiving conveyor belt 8, the goods information is acquired by scanning through the data acquisition module, and the storage information is generated, and the intelligent control module schedules the forklift to use the group intelligence algorithm and the time-varying graph network to optimize the carrying path again, and transports to the storage location, and places the goods on the shelf 5.

[0027] When using the intelligent goods warehouse management system, first in the goods warehousing stage S1, each goods will be configured with a radio frequency identification tag containing a two-dimensional code, the staff scans the tag to obtain the goods information through the code scanning device, and transmits the data to the remote terminal control module, which controls the forklift 1 in the goods taking and placing module to complete the goods carrying task, and the forklift accurately places the goods on the shelf 5 through the automatically telescopic fork lever 3. Subsequently in the data processing stage S2, the data acquisition module transmits all the goods information to the remote terminal control module for centralized management through the data transmission and storage module, the intelligent control module constructs a three-dimensional virtual scene of the warehouse based on the panoramic puzzle algorithm, and dynamically displays the inventory distribution in combination with the manifold visualization technology, to provide intuitive decision support for the management personnel; at the same time, the forklift 1 schedules the optimal path using the group intelligence algorithm and the time-varying graph network, to realize efficient storage and retrieval operation. In the process of taking goods S4-S5, the forklift 1 reaches the target shelf 5 according to the optimal path, identifies the goods position and information through the vehicle-mounted camera, executes the taking task, and transports the goods to the receiving conveyor belt 8, which is then transported to the general conveyor belt 9, and the data acquisition module again collects the goods for the outgoing process, to complete the entire taking process. In the process of storing goods (S6), the goods are first placed on the receiving conveyor belt 8, the data acquisition module scans the goods information to generate a storage instruction, and the intelligent control module again schedules the forklift 1 to transport and place the goods to the shelf 5 according to the optimal path. In addition, the system also introduces a full life cycle management module, extracts goods data from the remote terminal control module, records the whole process information of the goods using the directed acyclic graph of the block chain, and guarantees data privacy and traceability through the zero-knowledge proof algorithm.

[0028] Through the cooperation of the radio frequency identification tag and the code scanning device, the efficiency and accuracy of goods information acquisition are improved; the remote terminal control module realizes precise control of the forklift 1 and the fork lever 3, improving the automation level of goods carrying; the three-dimensional virtual scene and the manifold visualization technology enhance the visual management capability of the storage state, facilitating real-time grasp of the inventory situation; the group intelligence algorithm and the time-varying graph network optimization technology effectively reduce the travel path time of the forklift 1, improving the carrying efficiency; the application of the vehicle-mounted camera enhances the reliability of goods identification, reducing the need for manual intervention; the outgoing and storage processes are automatically completed by the system, improving the overall logistics automation level; the introduction of the block chain and the zero-knowledge proof technology strengthens the data security and traceability, meeting the data management needs in high-security scenarios.

[0029] Regarding the use of the intelligent control module: In the monitoring room, a dedicated person is responsible for manual monitoring to ensure the stable operation of the system and the timely handling of abnormal situations. The intelligent control module is deployed in the local server, and the operator only needs to input the pick-up order number in the system interface to start the automatic picking process: the data acquisition module reads the goods information through the scanning equipment, the intelligent control module determines the storage location according to the goods number, and combines the three-dimensional virtual scene and manifold visualization display of inventory distribution constructed by the panoramic puzzle algorithm to optimize the path of the crane 1 using the swarm intelligence algorithm and time-varying graph network to complete the picking task at the goods shelf 5; After the crane 1 picks up the goods, it transports them to the receiving conveyor belt 8, which then transports them to the general conveyor belt 9, and the data acquisition module scans the goods again to complete the warehouse collection. During the storage process, the operator only needs to issue a storage command, and the system will automatically complete the goods information scanning, generate the storage location, dispatch the crane 1 to transport and place it on the goods shelf 5 according to the optimal path. Throughout the process, the vehicle-mounted camera is used to identify the goods location and information, improving the accuracy of identification, while the full life cycle management module uses the directed acyclic graph technology of blockchain to record the full process data, ensuring data security and traceability.

[0030] The intelligent control module is deployed in the local server, with fast response speed, stable operation, and simple operation, only needing to input the pick-up order number or issue a storage command to realize automatic operation, greatly reducing the complexity of manual operation; the three-dimensional virtual scene and manifold visualization technology improves the visual management level of warehouse state, making it easy to grasp the inventory dynamics; the swarm intelligence algorithm and time-varying graph network optimization effectively improve the path planning efficiency, reduce the handling time, and improve the overall logistics efficiency; the coordinated action of the crane 1 and the fork lever 3 realizes precise storage and retrieval; the vehicle-mounted camera enhances the reliability of goods identification; the cooperation of the data acquisition module and the transmission storage module ensures efficient information flow; the application of blockchain and zero-knowledge proof technology ensures the security and privacy protection capability of data; at the same time, the system retains the manual monitoring mechanism, balancing the automation level, safety, and emergency response capability.

[0031] In the preferred scheme, in step S1, control instructions are generated according to the goods information; The goods picking and placing module is driven by the control instructions to transport the goods to the designated shelf location; Real-time data acquisition is used to obtain goods number and storage location data; The data transmission and storage module is used to transport the collected goods information to the intelligent control module; An encrypted transmission method is used to protect the safety of the goods information; The distributed storage structure is used to store the goods information to support fast query; The warehouse inventory status is dynamically updated according to the goods information.

[0032] The data acquisition module obtains the goods information, and generates a control instruction according to the goods information, which is used to guide the subsequent operation process. Then, the intelligent control module drives the hoist 1 and the fork 3 in the goods taking and placing module through the control instruction, and automatically transports the goods to the designated shelf 5 position to complete the storage. In the goods transportation process, the system adopts a real-time data acquisition mode to obtain the goods number and its current storage position data, ensuring accurate synchronization of information; the collected goods information is transmitted to the intelligent control module through the data transmission and storage module, and an encryption transmission mode is adopted to ensure the safety of the goods information during transmission; in order to improve the data management efficiency, the system adopts a distributed storage structure to store the goods information, supporting fast query and efficient processing; in addition, the system also dynamically updates the warehouse inventory status according to the latest goods information, and realizes the visualization display of the inventory in combination with the three-dimensional virtual scene.

[0033] By automatically generating control instructions and accurately driving the goods taking and placing module, the automation level and operation accuracy of the warehouse operation are improved; the real-time data acquisition mechanism ensures the timeliness and reliability of the inventory information; the encryption transmission mode effectively prevents the goods information from being illegally stolen or tampered with during transmission, enhancing the information security protection capability; the distributed storage structure not only improves the data processing efficiency of the system, but also enhances the scalability and stability of the system; the dynamic updating function of the inventory status helps the management personnel to grasp the warehouse operation situation at any time, optimizes the warehousing scheduling strategy, and thus improves the intelligent, safe and efficient level of warehouse management as a whole.

[0034] In the preferred scheme, in step S2, wireless network or local area network relay transmission is adopted, time convolution network is used to predict data delay, and if the transmission data is inconsistent, historical trajectory interpolation algorithm is used for correction; The transmission data is encrypted using an elliptic curve encryption algorithm, and the cloud storage uses homomorphic encryption technology to support ciphertext calculation; The real-time data collected by the data acquisition module is stored in the key-value pair forest database, supporting millisecond-level range query; The historical data collected by the data acquisition module is stored in the distributed database through the hierarchical index tree structure, and the time-space index block is generated according to the time granularity.

[0035] Step S2: data transmission and storage algorithm step.

[0036] Ensure efficient and accurate data transmission, and deal with network delay and data loss problems. The time convolution network adopts a causal convolution design, ensuring that the model can only see historical data and avoiding future information leakage. The dilated convolution expands the receptive field by introducing a dilation factor without increasing the number of parameters.

[0037] ; The dilation factor controls the spacing of the convolution kernels; The kernel size; Given the length of the input sequence. Captures the time dependencies of long sequences and predicts future data transmission delays.

[0038] Loss function design: Combining Huber loss and attention mechanism, higher weight is given to abnormal latency values.

[0039] ; Huber loss More robust to outliers, defined as: ; Time attention weight Make the model pay more attention to the prediction error closer to the current time: ; Historical trajectory interpolation correction using the Kalman filter algorithm: Based on the system state-space model, it integrates historical trajectories and current observations to achieve optimal estimation.

[0040] ; in: :time Optimal state estimation; :time Observed values; Kalman gain, which balances the weights of predictions and observations; and : Covariance matrix of system noise and observation noise.

[0041] Data security encryption mechanisms ensure the security and privacy of data transmission and storage. Private and public key generation: Based on the Curve 25519 curve, the private key... It's a random number, public key Through the base point Scalar multiplication generation.

[0042] ; Elliptic curve addition: Given two points on the curve and Calculate their sum : ; ; This achieves efficient asymmetric encryption, where the private key is used for signing and the public key is used for verification.

[0043] Homomorphic encryption supports ciphertext computation, BGV scheme: supports homomorphic operations of addition and multiplication, allowing direct computation on ciphertext.

[0044] ; ; Parameter interpretation: : Encryption function; : plaintext message; : plaintext modulus, determines the range of values that can be represented.

[0045] Key-value forest database (KV-Forest), millisecond-level real-time data query and storage. Time series data is divided into different time windows, and each window corresponds to an index tree.

[0046] ; The root node maintains the mapping of time range to sub-tree: ; Quickly locate the sub-tree where the data is located, reduce the query range.

[0047] Range query optimization, skip list acceleration, use skip list structure in each sub-tree, support fast range scan.

[0048] ; Time complexity: ; : Total data volume; : query result set size.

[0049] Hierarchical index tree and spatiotemporal index block, efficient storage of historical data, support multi-dimensional spatiotemporal query. Build hierarchical index according to year, month, day, hour, each node maintains statistical information.

[0050] ; Node structure: ; Through layer-by-layer filtering, quickly locate the time range where the target data is located.

[0051] Spatiotemporal index block generation, map multi-dimensional space coordinates (such as shelf location, time) to one-dimensional space, maintain spatial locality.

[0052] ; Time granularity adaptation: automatically adjust the index granularity according to the data time span, balance query efficiency and storage cost.

[0053] ; Network transmission layer optimization, QUIC protocol dynamic window adjustment, dynamically adjust the transmission window size according to the predicted delay, avoid network congestion.

[0054] ; : initial window size; : time constant, control the rate of window adjustment.

[0055] Database performance test, under high concurrent query, the response time of KV-Forest database conforms to the three-order Erlang distribution.

[0056] ; : rate parameter fitted by actual test data.

[0057] Encryption algorithm efficiency, ECC encryption throughput decreases with the increase of key length, which conforms to the power law relationship.

[0058] ; 、 、 : parameters determined by benchmark test.

[0059] Embodiment 3 In step S3, multiple cameras are arranged inside the warehouse, 3D software is used to model the shelf (5), the lifting vehicle (1) and the ground rail (2) equipment, and the OBJ format model is exported; The OBJ format model and the multiple camera visual data construct a three-dimensional virtual scene, the method is: The intelligent control module uses panoramic puzzle algorithm to construct the three-dimensional virtual scene of the warehouse, and the intrinsic matrix of the multiple cameras in the warehouse is obtained by Zhang Zhengyou calibration method and the extrinsic matrix , after removing the matching error points by RANSAC algorithm, the image stitching is completed through the homography matrix , and the shelf and lifting vehicle equipment model built by the three-dimensional software is aligned and registered with the camera visual data through ICP algorithm, and real-time rendering is realized by using WebGL technology; The manifold visualization technology is used to dynamically display the inventory distribution, the high-dimensional inventory data is mapped to the two-dimensional manifold space by t-SNE algorithm and color coded, and the heat map of the warehouse area is generated based on the Gaussian kernel density estimation.

[0060] In the preferred scheme, in step S4, the group intelligence algorithm is used to schedule the lifting vehicle in the goods taking and placing module; Optimize the carrying path through the time-varying graph network; Perform pickup and storage operations according to the described handling path; The location and type of goods are identified by cameras on the crane to confirm the accuracy of the operation.

[0061] In the preferred scheme, the intelligent control module uses a swarm intelligence algorithm to schedule the cranes in the cargo handling module, and employs an ant colony optimization algorithm to assign tasks to multiple cranes. The ant colony optimization algorithm is used to calculate the task assignment probability, combined with an artificial potential field algorithm to prevent collisions. The formula for the ant colony optimization algorithm is as follows: ; in, express Time Path The higher the pheromone concentration, the greater the probability that the path will be selected. As heuristic information, it is usually inversely proportional to the distance between two points, indicating the likelihood of choosing that path; and These are weighting coefficients, which control the degree of influence of pheromones and heuristic information on the selection probability, respectively. The formula for the artificial potential field algorithm is: ; in, The repulsive force coefficient, Indicates crane With crane The distance vector between them; Then, by combining time-varying graph networks to optimize the transport path, a time-varying graph model is constructed to achieve this. The edge weights are represented, and the optimal path is solved using the TD-Dijkstra algorithm and smoothed by cubic spline interpolation. in, Indicates at time ,node and The weights of the edges between them. Wherein, The actual distance between the two points. This is a dynamic adjustment factor that changes over time, used to reflect the impact of real-time conditions on the path weights. Picking and storing goods are carried out according to the optimal route. During the picking process, the camera on the crane identifies the location and category of the goods through the YOLOv5 algorithm. If the information matches the goods to be picked, the picking is performed. If they do not match, an audible and visual alarm is triggered. The crane is operated remotely by the operator to trace the historical location of the goods based on historical data. If they still do not match after the second check, an alarm is triggered that the goods are not in the warehouse.

[0062] The specific steps are as follows: 1. Formulas related to panoramic mosaic algorithm Zhang Zhengyou's calibration method parameters: intrinsic parameter matrix It includes internal parameters such as the camera's focal length and principal point coordinates, used to describe the geometric characteristics of camera imaging; extrinsic parameter matrix. middle, Let be a rotation matrix. These are translation vectors, which together describe the position and orientation of the camera in the world coordinate system.

[0063] Homography matrix: Used to describe the projection transformation relationship between two planes, in image stitching, the images are mapped to the same plane by calculating the homography matrix between images from different cameras, thus achieving stitching.

[0064] ICP algorithm: Iterative Closest Point algorithm, is used to align and register the device model modeled in 3D software with camera vision data. By continuously iterating to find corresponding points, it minimizes the distance error between points, so that the virtual model matches the actual scene.

[0065] 2. Formulas related to manifold visualization technology t-SNE algorithm: t-distributed random neighborhood embedding algorithm, which maps high-dimensional inventory data (such as multi-dimensional information including goods type, quantity, storage location, etc.) to a two-dimensional manifold space, making it easier to intuitively display the data distribution characteristics. Different color codes are used to distinguish different types or states of inventory data.

[0066] Gaussian kernel density estimation: By calculating the kernel density at each location, a heatmap of the storage area is generated, which visually reflects the density of inventory distribution. Areas with higher kernel density are darker in color on the heatmap, indicating that the area has a larger amount of stored goods or frequent inbound and outbound transactions.

[0067] 3. Swarm intelligence algorithm scheduling Ant colony optimization algorithm: Used to calculate at time 1 ,Ant From node Transfer to node The probability of . Among them, express Time Path The higher the pheromone concentration, the greater the probability that the path will be selected. As heuristic information, it is usually inversely proportional to the distance between two points, indicating the likelihood of choosing that path; and These are weighting coefficients, which control the degree of influence of pheromones and heuristic information on the selection probability, respectively.

[0068] Artificial potential field method: For calculating the repulsion force between the forklifts to avoid collision. Among them, is the repulsion force coefficient, represents the distance vector between the forklifts , the formula shows that the closer the distance, the greater the repulsion force, prompting the forklift to maintain a safe distance.

[0069] 4. Time-varying graph network optimization related formulas Time-varying graph model edge weight: represents the weight of the edge between node and at time . Among them, is the actual distance between the two points, is a dynamic adjustment factor that changes over time, used to reflect the influence of real-time conditions on the path weight (such as congestion, equipment occupation, etc.).

[0070] TD-Dijkstra algorithm: a time-dependent Dijkstra algorithm for solving the optimal path from the starting point to the ending point in the time-varying graph, considering the change of edge weight with time, finding the shortest path starting at the current time.

[0071] Cubic spline interpolation: smooth the optimal path obtained, make the driving path of the forklift more smooth, reduce frequent turning and acceleration and deceleration, improve the running efficiency and stability.

[0072] 5. Goods identification and traceability related formulas Goods location traceability: Through the goods ID and time range, query the historical location information of the goods, when the forklift camera recognizes the goods information and the to-be-taken goods are inconsistent, it is used to trace the possible location of the goods, and assist the manual remote operation.

[0073] Actual use: 1. Three-dimensional virtual scene construction Device modeling: use professional three-dimensional modeling software (such as 3ds Max, Maya) to accurately model the shelves (5), forklifts (1), ground rails (2) and other equipment according to the actual size of the warehouse and the specifications of the equipment. After completion, export the model in OBJ format.

[0074] Camera calibration: install multiple cameras in the warehouse to cover the main operating area. Use Zhang Zhengyou calibration method to calculate the intrinsic matrix and extrinsic matrix of each camera by shooting images of calibration boards with checkerboard patterns, record and save the parameters.

[0075] ​Image stitching and model alignment: Start the remote terminal control module, and the camera collects warehouse image data in real time. Use the RANSAC algorithm to match feature points in different camera images, eliminate incorrect matching points, and calculate the homography matrix Complete image stitching to form a panoramic image. Align and register the OBJ format device model with the stitched visual data using the ICP algorithm to ensure accurate correspondence between the virtual model and the actual scene.

[0076] Real-time rendering: Use WebGL technology to render the registered three-dimensional virtual scene in real time on a remote terminal (such as a computer or tablet), allowing management personnel to view the real-time status of the warehouse through the terminal.

[0077] 2. Inventory distribution visualization Data preparation: Extract high-dimensional inventory data containing information such as goods types, quantities, and storage locations from the warehouse management system.

[0078] Data mapping and visualization: Input high-dimensional inventory data into the manifold visualization module, map it to a two-dimensional manifold space using the t-SNE algorithm, and color code it based on goods types or inventory status. Use Gaussian kernel density estimation to generate a warehouse heat map based on the mapped data, and superimpose it in the three-dimensional virtual scene to visually present the inventory distribution.

[0079] 3. Crane scheduling and path optimization Task allocation: When there is a goods picking or placing task, the remote terminal control module starts the ant colony optimization algorithm. Set parameters such as pheromone weight , heuristic information weight , etc., calculate the probability of each crane executing different task paths according to the formula, and complete task allocation.

[0080] Collision avoidance: During the execution of the task by the crane, calculate the repulsive force between them using the artificial potential field method, and adjust the driving direction in real time to avoid collisions. According to the formula, dynamically calculate the repulsive force based on the distance between the cranes to ensure safe operation.

[0081] Path optimization: Construct a time-varying graph model, determine the dynamic adjustment factor in the edge weight based on the real-time state of the warehouse (such as other device path occupation, traffic congestion), use the TD-Dijkstra algorithm to solve the optimal path in the time-varying graph, and perform smoothing processing on the path using cubic spline interpolation, and send the optimized path to the crane for execution.

[0082] 4. Goods picking and placing operation Cargo identification: when the forklift is performing the task of picking up goods, the camera on the forklift uses the YOLOv5 algorithm to identify the goods in real time, obtains the position and category information of the goods, and compares it with the information of the goods to be picked up.

[0083] Abnormal handling: if the identified goods information is inconsistent with the goods to be picked up, the forklift triggers an audible and light alarm to prompt an anomaly. At the same time, the remote terminal control module traces the historical position of the goods, and manually remotely operates the forklift to go to the position to detect again. If it is still inconsistent, it is judged that the warehouse is out of stock, an out-of-stock alarm is issued, and relevant information is recorded for subsequent processing.

[0084] Normal picking: when the identified goods information is consistent with the goods to be picked up, the forklift performs the picking operation according to the predetermined process and delivers the goods to the designated position to complete the storage task.

[0085] In the preferred scheme, in steps S5-S6, during the storage and picking of the goods, the intelligent control module receives the out-of-stock and in-stock data of the data acquisition module; The intelligent control module records the whole process information of the goods data using the blockchain directed acyclic graph technology.

[0086] In the process of storing and picking the goods (S5-S6), when the intelligent control module receives the out-of-stock and in-stock data from the data acquisition module, the system automatically starts the blockchain directed acyclic graph (DAG) recording process. The specific algorithm steps are as follows: first, the data acquisition module scans the information of the goods to obtain the out-of-stock or in-stock state of the current goods, and uploads the data to the intelligent control module; then, the intelligent control module encapsulates the goods data as a data node, and establishes a link with the previous node through a hash pointer according to the time sequence and operation type, forming a directed acyclic graph structure; every time a new operation record is added, a new data node is generated and logically connected with the associated historical node, ensuring that the whole process information is traceable; during the whole process, all operation records are stored in the block node in an encrypted form to prevent data tampering and illegal access.

[0087] When using this scheme, whenever the goods are picked up or stored in the position of the goods shelf 5 by the forklift 1, the data acquisition module immediately collects the goods information and uploads it to the intelligent control module through the data transmission storage module, which calls the blockchain DAG technology to record the whole process data of the goods; at the same time, the whole life cycle management module combines the zero-knowledge proof algorithm to realize the traceable and credible data under the premise of protecting privacy. Management personnel can view the complete operation history of any goods through the visual interface, including timestamp, operator, device state, path planning, etc., further improving the transparency and security of warehouse management.

[0088] ​The blockchain directed acyclic graph technology is adopted, the tamper-proof and traceability of the whole-process data of goods are effectively realized, and the data security level of the warehouse system is improved; compared with the traditional chain block chain structure, the DAG has higher concurrent processing capability and scalability, and is suitable for high-frequency data writing requirements in a large-scale warehouse environment; combined with the automatic recording mechanism of the intelligent control module, manual intervention is reduced, and the accuracy and real-time performance of data entry are improved; in addition, through the support of privacy protection algorithms such as zero-knowledge proof, the safety of sensitive information is ensured, and the compliance requirements are met, and it is especially suitable for industries such as medicine, military industry, finance and other industries with high requirements for data integrity.

[0089] The above embodiments are only preferred technical solutions of the present application, and should not be regarded as a limitation of the present application. The protection scope of the present application should be based on the technical solutions recited in the claims, including equivalent replacement solutions of the technical features recited in the claims. That is, equivalent replacement improvements within this scope are also within the protection scope of the present application.

Claims

1. A smart cargo warehouse management system, characterized by: It includes a data acquisition module, a cargo handling module, a data transmission and storage module, and an intelligent control module; The data acquisition module is used to read cargo information; The goods picking and placing module is used to pick up and place goods on the shelf (5); The intelligent control module is used to construct a three-dimensional virtual scene of the warehouse, and combined with the goods retrieval module, it retrieves goods along the optimal path.

2. The intelligent cargo warehouse management system according to claim 1, characterized in that: The cargo handling module includes a crane (1), which is equipped with multiple automatically synchronously telescopic forklifts (3), and vertical rail frames (4) are provided between adjacent shelves (5). The crane (1) and the vertical rail frames (4) are slidably connected. The vertical track frame (4) is slidably connected to the ground track (2) and the overhead track (6) at its upper and lower ends, respectively, and is driven by a gear and rack to slide on the ground track (2) and the overhead track (6); A winch (7) is installed on the overhead rail (6), and the winch (7) is connected to the crane (1); A camera is also installed on one side of the crane (1).

3. The intelligent cargo warehouse management system according to claim 2, characterized in that: Multiple shelves (5) are also provided with a main conveyor belt (9) on one side, and multiple receiving conveyor belts (8) are provided on one side of the main conveyor belt (9), with the receiving conveyor belts (8) located on one side of the ground rail (2).

4. A management method for an intelligent cargo warehouse management system, characterized by: Applied to the system as described in any one of claims 1-3, the method is characterized in that: the method comprises: S1. The data acquisition module configures an RFID tag with a QR code for each inbound item, and obtains the item information by scanning the RFID tag with a barcode scanning device; S2. The data collected by the data acquisition module is transmitted to the intelligent control module through the data transmission and storage module. The intelligent control module generates the goods number and storage location data based on the data collected by the data acquisition module. The S3 intelligent control module also uses a panoramic mosaic algorithm to construct a 3D warehouse scene and combines manifold visualization to display the inventory distribution. S4. Based on the visualized 3D scene of inventory distribution, the crane is dispatched to optimize the handling path using swarm intelligence algorithm and time-varying graph network, and the goods are identified through the on-board camera. S5. After identifying the goods, the intelligent control module drives the crane (1) to pick up the goods. The crane (1) transports the goods to the receiving conveyor belt (8). After the receiving conveyor belt (8) transports the goods to the main conveyor belt (9), the data acquisition module collects the goods for shipment and completes the picking up. S6. When storing goods, the goods are placed on the receiving conveyor belt (8). The data acquisition module scans and obtains the goods information and generates the storage information. The intelligent control module schedules the crane to use the swarm intelligence algorithm and time-varying graph network to optimize the handling path and transport the goods to the storage location and place the goods on the shelf (5).

5. The management method of the intelligent cargo warehouse management system according to claim 4, characterized in that: In step S1, control instructions are generated based on the cargo information; The control commands drive the goods picking and placing module to transport the goods to the designated shelf location. The cargo number and storage location data are acquired using real-time data collection methods; The collected cargo information is transmitted to the intelligent control module via the data transmission and storage module; Encrypted transmission methods are used to protect the security of the cargo information; The cargo information is stored using a distributed storage structure to support fast querying; The warehouse inventory status is dynamically updated based on the cargo information.

6. The management method of the intelligent cargo warehouse management system according to claim 4, characterized in that: In step S2, wireless network or local area network is used for relay transmission, and data delay is predicted by temporal convolutional network. If the transmitted data is inconsistent, historical trajectory interpolation algorithm is used for correction. Elliptic curve cryptography is used to encrypt transmitted data, and homomorphic encryption technology is used in cloud storage to support ciphertext computation. The real-time data collected by the data acquisition module is stored in a key-value pair forest database, which supports millisecond-level range queries; Historical data collected by the data acquisition module is stored in a distributed database through a hierarchical index tree structure, and spatiotemporal index blocks are generated according to time granularity.

7. The management method of the intelligent cargo warehouse management system according to claim 4, characterized in that: In step S3, multiple cameras are installed inside the warehouse. 3D software is used to create 3D models of the shelving (5), crane (1), and ground rail (2) equipment, and the models are exported as OBJ format models. The method for constructing a 3D virtual scene using OBJ format models and visual data from multiple cameras is as follows: The intelligent control module uses a panoramic mosaic algorithm to construct a 3D virtual scene of the warehouse and obtains the intrinsic parameter matrix of multiple cameras in the warehouse using the Zhang Zhengyou calibration method. With extrinsic matrix After removing mismatched points using the RANSAC algorithm, the homography matrix is ​​used... After completing the image stitching, the 3D model of the shelving and crane equipment and the camera visual data are aligned and registered using the ICP algorithm, and then rendered in real time using WebGL technology. Manifold visualization technology is used to dynamically display inventory distribution. The high-dimensional inventory data is mapped to a two-dimensional manifold space and color-coded using the t-SNE algorithm. A heat map of the storage area is generated based on Gaussian kernel density estimation.

8. The management method of the intelligent cargo warehouse management system according to claim 7, characterized in that: Step S4 In this process, a swarm intelligence algorithm is used to schedule the cranes in the cargo handling module; Optimize transport paths using time-varying graph networks; Perform pickup and storage operations according to the described handling path; The location and type of goods are identified by cameras on the crane to confirm the accuracy of the operation.

9. The management method of the intelligent cargo warehouse management system according to claim 8, characterized in that: In the intelligent control module, the cranes in the cargo handling module are scheduled using a swarm intelligence algorithm. An ant colony optimization algorithm is used to assign tasks to multiple cranes. The ant colony optimization algorithm calculates the task assignment probability, and an artificial potential field algorithm is combined to prevent collisions. The formula for the ant colony optimization algorithm is: ； in, express Time Path The higher the pheromone concentration, the greater the probability that the path will be selected. As heuristic information, it is usually inversely proportional to the distance between two points, indicating the likelihood of choosing that path; and These are weighting coefficients, which control the degree of influence of pheromones and heuristic information on the selection probability, respectively. The formula for the artificial potential field algorithm is: ； in, The repulsive force coefficient, Indicates crane With crane The distance vector between them; Then, by combining time-varying graph networks to optimize the transport path, a time-varying graph model is constructed to achieve this. The edge weights are represented, and the optimal path is solved using the TD-Dijkstra algorithm and smoothed by cubic spline interpolation. in, Indicates at time ,node and The weights of the edges between them. Wherein, The actual distance between the two points. This is a dynamic adjustment factor that changes over time, used to reflect the impact of real-time conditions on the path weights. Picking and storing goods are carried out according to the optimal route. During the picking process, the camera on the crane identifies the location and category of the goods through the YOLOv5 algorithm. If the information matches the goods to be picked, the picking is performed. If they do not match, an audible and visual alarm is triggered. The crane is operated remotely by the operator to trace the historical location of the goods based on historical data. If they still do not match after the second check, an alarm is triggered that the goods are not in the warehouse.

10. The management method of the intelligent cargo warehouse management system according to claim 4, characterized in that: In steps S5-S6, during the process of storing and retrieving goods, the intelligent control module receives the shipment and inventory data from the data acquisition module. The intelligent control module uses blockchain directed acyclic graph technology to record the entire process information of the cargo data.

Images