Finished product label printing method and finished product warehousing method and system
By collecting multi-source data and using deep neural network models to generate intelligent label templates, combined with multispectral imagers and ant colony algorithms to optimize storage location allocation, the problems of manual dependence and data fragmentation in the finished product label printing and warehousing process are solved, and efficient and accurate label printing and storage location optimization are achieved, meeting the company's needs for efficient collaboration and quality traceability.
Patent Information
- Application Number
- CN202510905847.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-02
- Publication Date
- 2025-10-17
AI Technical Summary
The existing finished product label printing and warehousing process relies on manual experience in label template design and cannot dynamically adapt to corporate visual specifications, resulting in low printing efficiency, high error rate, insufficient optimization of warehousing paths, low AGV transportation efficiency, low storage capacity utilization, and data fragmentation leading to delays in quality traceability.
Smart label templates are generated through multi-source data collection and deep neural network models, printing parameters are adjusted in real time, defects are detected in combination with multispectral imagers, ant colony algorithms are used to optimize storage location allocation, and full-process data association and correction are achieved through RFID and digital twin technology.
It achieves the accuracy and standardization of label printing, improves printing efficiency and storage utilization, reduces manual intervention, ensures the real-time and traceability of data, and meets the needs of efficient collaboration and quality traceability.
Smart Images

Figure CN120805959A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of information technology, in particular to a method for product label printing, a product warehousing method and system. BACKGROUND
[0002] Information technology is a general term for various technologies used to manage and process information. It mainly applies computer science and communication technology to design, develop, install and implement information systems and application software. It is also commonly known as information and communication technology, mainly including sensing technology, computer and intelligent technology, communication technology and control technology.
[0003] A method for product label printing, a product warehousing method and system are one of the core technologies for intelligent upgrading of manufacturing industry, aiming to realize full-process automation and precise control from product off-line to warehouse management. The system integrates Internet of Things sensing, intelligent label design, automated printing and quality inspection, and intelligent warehouse scheduling to ensure the accuracy of product information identification and the efficiency of warehousing, supporting efficient collaboration and quality traceability requirements of enterprise supply chain.
[0004] Currently, due to the involvement of multiple links in the process of product label printing and warehousing, the existing technology has defects. The label template design relies on manual experience and cannot dynamically adapt to enterprise visual specifications, resulting in delayed label compliance verification, the need to repeatedly adjust template parameters, and a decrease in printing efficiency and an increase in error rate. The printing process is disconnected from the production line conditions, and dynamic parameters such as environmental temperature and humidity, equipment vibration spectrum, etc. are not fused in real time, resulting in defects such as ink bleeding and label adhesion during high-precision inkjet printing. Manual secondary quality inspection and reprinting are required, increasing costs and time loss. The warehousing process relies on manual decision-making, and the allocation of storage locations lacks intelligent optimization algorithms, making it impossible to generate optimal paths based on product three-dimensional size, warehouse frequency, etc. in real time, resulting in low AGV transportation efficiency and insufficient storage capacity utilization. Manual checking can also cause discrepancies between physical objects and system records. In addition, label and warehousing data are fragmented, and printing records are not automatically synchronized with inventory systems, making it difficult to meet real-time requirements when tracing quality.
[0005] Therefore, the present application provides a method for product label printing, a product warehousing method and system to solve the above problems. SUMMARY
[0006] (I) Technical problems solved
[0007] In view of the deficiencies of the prior art, the present application provides a method for product label printing, a product warehousing method and system to solve the problems raised in the background art.
[0008] (II) Technical solutions
[0009] To achieve the above object, the present application provides the following technical solutions to achieve: a method for finished product label printing, a finished product warehousing method and system, the method comprising the following steps:
[0010] S1, obtaining finished product basic information, production line real-time parameters and historical label template library data through a multi-source data acquisition terminal, generating a label element original data set;
[0011] S2, performing feature fusion processing on the label element original data set based on a deep neural network model, extracting a label content core feature vector and a layout feature vector, and generating a label intelligent design parameter set;
[0012] S3, matching degree calculation of the label intelligent design parameter set and a preset enterprise visual identification specification, triggering a dynamic optimization module when the matching degree is lower than a threshold value, and generating a compliant label template data;
[0013] S4, adjusting the working mode of the printing equipment according to the production line real-time working condition parameters, outputting the compliant label template data to the surface of the finished product through a high-precision inkjet printing system, and generating a label traceability code containing a space-time stamp;
[0014] S5, collecting the label image after printing by using a multi-spectral imager, performing image defect detection by using a convolutional neural network, outputting a finished product label completion signal when the detection is qualified, otherwise triggering a label reprint instruction.
[0015] Preferably, the S1 comprises the following steps:
[0016] S11, collecting finished product material parameters, geometric size data and batch number through an Internet of Things sensor network to form a finished product basic information subset;
[0017] S12, real-time acquisition of production line environment temperature and humidity, equipment vibration spectrum and transmission belt speed parameters to generate a working condition dynamic parameter subset;
[0018] S13, calling historical label template library data from a distributed storage system, and extracting a template structure feature matrix and a content element distribution heat map.
[0019] Preferably, the S2 comprises the following steps:
[0020] S21, constructing a double-channel feature extraction network, the first channel adopts an LSTM network to process text content features, and the second channel adopts a CNN network to process layout structure features;
[0021] S22, fusing double-channel output features through an attention mechanism to generate a feature tensor containing content priority weight;
[0022] S23, enhancing the feature tensor using the generative adversarial network, and outputting a set of intelligent design parameters of the label
[0023]
[0024] wherein C content is a content feature matrix, L layout is a layout feature vector, and a and β are dynamic adjustment coefficients.
[0025] Preferably, the S3 comprises the following steps:
[0026] S31, establishing a three-dimensional evaluation space of the enterprise visual identification specification:
[0027]
[0028] wherein the color specification vector is an RGB / CMYK color value vector, the layout compliance matrix is a layout rule matrix, and the information density threshold is the maximum information quantity per unit area;
[0029] S32, calculating the Manhattan distance between the design parameter set and the specification space:
[0030]
[0031] wherein, is the i-th dimensional feature of the design parameter set, is the i-th dimensional standard value of the specification;
[0032] S33, when D m > δ, starting the dynamic optimization module to output the compliant label template data T compliant by a constraint optimization algorithm.
[0033] Preferably, the method comprises the following steps:
[0034] P1, receiving a finished product label completion signal from claim 1, activating an RFID read-write device to write product traceability information;
[0035] P2, collecting finished product outer contour point cloud data through a 3D vision system to generate a three-dimensional size feature vector;
[0036] P3, calculating an optimal storage location allocation scheme based on an ant colony algorithm to generate storage location navigation path data;
[0037] P4, driving an AGV transportation system to transport the finished product to the target storage location along the navigation path, and updating the inventory topology map in real time;
[0038] P5, when the finished product reaches the specified storage location, triggering a multi-modal verification system to check the physical location and digital mapping relationship.
[0039] Preferably, the P3 comprises the following steps:
[0040] P31. Establish the objective function of storage location optimization:
[0041]
[0042] Among them D k is the transportation distance cost, is the estimated retrieval time, ω1 is the weight coefficient of transportation distance, and ω2 is the weight coefficient of retrieval time;
[0043] P32. Initialize the pheromone matrix τ ij (0) = τ0, set the ant colony size N ant ;
[0044] P33, using dynamic volatile factors to update pheromones:
[0045] τ ij (t+1)=(1-ρ(t))·τ ij (t)+Δτ ij
[0046] ρ(t)=ρ min +(ρ max -ρ min )·e -λt
[0047] Among them, λ is the adaptive adjustment coefficient, τ ij (t) is the pheromone of path (i, j) at time t, ρ(t) is the adaptive volatility factor, Δτ ij The amount of new pheromone added.
[0048] Preferably, the P5 comprises the following steps:
[0049] P51, scan the actual coordinates of the storage location through the laser radar to generate the spatial coordinate vector L real ;
[0050] P52. Retrieve the theoretical location coordinates L from the digital twin system virtual ;
[0051] P53. Calculate coordinate alignment error:
[0052] E align =||L real -L virtual ||2
[0053] Among them, L real is the actual coordinate of the laser radar scan, L virtual is the theoretical coordinate in the digital twin system;
[0054] P54、when E align <ε is updated to in-stock, otherwise trigger position correction instructions.
[0055] Preferably, comprising:
[0056] A label intelligent generation module, a multi-modal printing control module, and an intelligent warehouse scheduling module.
[0057] The label intelligent generation module comprises:
[0058] A multi-source data fusion unit that integrates real-time data and historical template features of a production line;
[0059] A deep learning design unit equipped with a dual-channel neural network architecture;
[0060] A compliance verification unit with a built-in dynamic optimization algorithm engine;
[0061] The multi-modal printing control module comprises:
[0062] An adaptive printing unit supporting UV inkjet / laser engraving dual-mode switching;
[0063] An online quality inspection unit integrating a spectrum analyzer and a defect detection model;
[0064] The intelligent warehouse scheduling module comprises:
[0065] A three-dimensional visual perception unit equipped with a ToF depth camera array;
[0066] A storage location optimization unit that realizes hardware acceleration of the ant colony algorithm;
[0067] A digital twin mapping unit that constructs a real-time inventory topology model.
[0068] Preferably, the multi-source data fusion unit is configured with a space-time alignment processor, and its data synchronization equation is:
[0069]
[0070] wherein, is the change rate of the i-th type of sensor data, is the change rate of the historical template data.
[0071] Preferably, the storage location optimization unit includes an FPGA accelerator, and its hardware implementation architecture comprises:
[0072] A pheromone update pipeline that processes N path path calculations per cycle;
[0073] A parallel cost calculation unit with an 8x8 matrix operator built-in;
[0074] Dynamic volatile factor controller, supports real-time reconstruction of p(t).
[0075] (III) Beneficial Effects
[0076] Compared with the prior art, the present application provides a method, a finished product storage method and a system for finished product label printing, which have the following beneficial effects:
[0077] 1. In the present application, by setting an intelligent design unit, when printing the finished product label, an automatic compliance verification mechanism is established to match the enterprise visual identification specification in real time, effectively solving the format deviation problem caused by manual design, ensuring the standardized output of labels for different specifications of products, dynamically optimizing the label design parameters by deeply integrating the production line working condition parameters and historical template characteristics, ensuring the consistency and compliance of label generation in complex production environment, and improving the accuracy and standardization of label printing.
[0078] 2. In the present application, by setting a working condition adaptive unit, real-time detection of dynamic parameters such as environmental temperature and humidity, equipment vibration, etc. is performed in the printing execution link, the printing mode is automatically switched and the inkjet pressure is adjusted, when the ink adhesion risk and positioning deviation are detected, the parameter calibration mechanism is triggered immediately, ensuring that the printing effect is not affected by the fluctuation of production line working conditions, avoiding the defects such as label blur and falling caused by environmental changes in traditional technology, and ensuring the persistent readability of the label.
[0079] 3. In the present application, by setting an intelligent scheduling unit, the three-dimensional size of the product and the frequency characteristics of the warehouse in and out are analyzed in real time during the finished product storage process, the optimal storage location allocation strategy is automatically planned, when the deviation between the actual location of the storage location and the system mapping is detected, the coordinate alignment mechanism is used for automatic correction, solving the problems of path detour and storage capacity waste caused by manual scheduling, and realizing the coordinated improvement of warehouse space utilization rate and in and out efficiency.
[0080] 4. In the present application, by setting a full-chain traceability unit, a space-time stamp code is embedded in the label generation stage, and the printing record is automatically associated with the inventory topology model, when quality traceability is needed, the system automatically locates the product life cycle data, avoids the traceability delay problem caused by cross-system data fragmentation, and realizes the whole-process closed-loop management and control from label printing to warehouse scheduling. BRIEF DESCRIPTION OF DRAWINGS
[0081] Figure 1 A flowchart of a method for printing a finished product label according to the present application;
[0082] Figure 2 A flowchart of a method for storing a finished product according to the present application;
[0083] Figure 3 A framework diagram of a system for printing and storing a finished product label according to the present application. DETAILED DESCRIPTION
[0084] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.
[0085] Please refer to Figure 1 The method, the product storage method and the system for product label printing comprise the following steps:
[0086] S1, acquiring product basic information, real-time parameters of production lines and historical label template library data through a multi-source data acquisition terminal to generate a label element original data set;
[0087] S2, performing feature fusion processing on the label element original data set based on a deep neural network model to extract a label content core feature vector and a layout feature vector and generate a label intelligent design parameter set;
[0088] S3, performing matching degree calculation on the label intelligent design parameter set and a preset enterprise visual identification specification, triggering a dynamic optimization module when the matching degree is lower than a threshold value to generate compliant label template data;
[0089] S4, adjusting the working mode of a printing device according to real-time working condition parameters of production lines, outputting the compliant label template data to the surface of a product through a high-precision inkjet printing system, and generating a label traceability code containing a space-time stamp;
[0090] S5, collecting a label image after printing through a multispectral imager and performing image defect detection through a convolutional neural network, outputting a product label completion signal when the detection is qualified, or triggering a label reprint instruction otherwise;
[0091] S1 comprises the following steps:
[0092] S11, collecting product material parameters, geometric size data and batch numbers through an Internet of Things sensor network to form a product basic information subset;
[0093] S12, acquiring real-time production line environment temperature and humidity, device vibration spectrum and transmission belt speed parameters to generate a working condition dynamic parameter subset;
[0094] S13, calling historical label template library data from a distributed storage system to extract a template structure feature matrix and a content element distribution heat map;
[0095] S2 comprises the following steps:
[0096] S21, a double-channel feature extraction network is constructed, the first channel adopts an LSTM network to process text content features, and the second channel adopts a CNN network to process layout structure features;
[0097] S22, the output features of the two channels are fused through an attention mechanism to generate a feature tensor containing content priority weights;
[0098] S23, using an adversarial generative network to enhance the feature tensor, outputting a label intelligent design parameter set
[0099]
[0100] where C content is the content feature matrix, L layout is the layout feature vector, and α, β are dynamic adjustment coefficients;
[0101] S3 includes the following steps:
[0102] S31, a three-dimensional evaluation space of enterprise visual identification specifications is established:
[0103]
[0104] wherein the color specification vector is an RGB / CMYK color value vector, the layout compliance matrix is a layout rule matrix, and the information density threshold is the maximum information quantity per unit area;
[0105] S32, calculate the Manhattan distance between the design parameter set and the specification space:
[0106]
[0107] wherein, is the i-th dimensional feature of the design parameter set, is the i-th dimensional standard value of the specification;
[0108] S33, when D m >δ, start the dynamic optimization module, and output the compliance label template data T compliant through a constraint optimization algorithm;
[0109] The method includes the following steps:
[0110] P1, receiving a finished label completion signal from claim 1, activating an RFID read-write device to write product traceability information;
[0111] P2, collecting finished product contour point cloud data through a 3D vision system to generate a three-dimensional size feature vector;
[0112] P3, calculating an optimal storage location allocation scheme based on an ant colony algorithm to generate storage location navigation path data;
[0113] P4, driving the AGV transportation system to transport the finished product to the target storage location along the navigation path, and updating the inventory topology map in real time;
[0114] P5, when the finished product reaches the designated storage location, triggering the multi-modal verification system to check the physical location and digital mapping relationship;
[0115] P3 includes the following steps:
[0116] P31, establishing a storage location optimization objective function:
[0117]
[0118] where D k is the transportation distance cost, is the estimated retrieval time, ω1 is the weight coefficient of transportation distance, and ω2 is the weight coefficient of retrieval time;
[0119] P32, initializing the pheromone matrix τ ij (0) = τ0, setting the ant colony size N ant ;
[0120] P33, updating the pheromone using a dynamic evaporation factor:
[0121] τ ij (t+1) = (1-ρ(t))τ ij (t) + Δτ ij
[0122] ρ(t) = ρ min + (ρ max - ρ min )·e -λt
[0123] where λ is an adaptive adjustment coefficient, τ ij (t) is the pheromone of path (i,j) at time t, ρ(t) is the adaptive evaporation factor, and Δτ ij is the amount of new pheromone;
[0124] P5 includes the following steps:
[0125] P51, scanning the actual coordinates of the storage location by laser radar to generate a spatial coordinate vector L real ;
[0126] P52, calling the theoretical storage location coordinates L virtual from the digital twin system;
[0127] P53, calculating the coordinate alignment error:
[0128] E align = ||L real -Lvirtual ||2
[0129] wherein, L real is the actual coordinate of the laser radar scanning, L virtual is the theoretical coordinate in the digital twin system;
[0130] P54, when E align <epsilon> updates the inventory status as in the warehouse, otherwise triggers the position correction instruction;
[0131] It comprises:
[0132] a label intelligent generation module, a multi-modal printing control module, and an intelligent warehouse scheduling module.
[0133] The label intelligent generation module comprises:
[0134] a multi-source data fusion unit that integrates real-time data of the production line and historical template features;
[0135] a deep learning design unit equipped with a dual-channel neural network architecture;
[0136] a compliance verification unit with a built-in dynamic optimization algorithm engine;
[0137] The multi-modal printing control module comprises:
[0138] an adaptive printing unit that supports UV inkjet / laser printing dual-mode switching;
[0139] an online quality inspection unit that integrates a spectrum analyzer and a defect detection model;
[0140] The intelligent warehouse scheduling module comprises:
[0141] a three-dimensional visual perception unit equipped with a ToF depth camera array;
[0142] a storage location optimization unit that realizes hardware acceleration of the ant colony algorithm;
[0143] a digital twin mapping unit that constructs a real-time inventory topology model;
[0144] The multi-source data fusion unit is configured with a space-time alignment processor, and its data synchronization equation is:
[0145]
[0146] wherein, is the change rate of the i-th type of sensor data, is the historical template data change rate; The storage location optimization unit contains an FPGA accelerator, and its hardware implementation architecture comprises:
[0147] a pheromone update pipeline that processes N path path calculations per cycle;
[0148] Parallel cost calculation unit with built-in 8×8 matrix operator;
[0149] Dynamic volatility factor controller supports real-time reconstruction of ρ(t).
[0150] Example 1: Label printing and warehouse scheduling for auto parts manufacturing companies
[0151] Implementation process of the finished product label printing method: When the production line completes the processing of the crankshaft finished product, the system starts the label printing process, collects multi-source data, obtains the production batch number, material parameters and geometric dimensions of the crankshaft through the PLC controller, and at the same time, calls the customer's historical label template library from the SQL Server database, extracts the required QR code position specifications and warning sign standards, and generates intelligent templates. In the dual-channel architecture of the deep neural network, the LSTM network analyzes the technical parameters of the crankshaft, and the CNN network identifies the layout constraints. When the output parameter is close to the enterprise VI specification, the dynamic optimization module automatically adjusts the text line spacing to 1.5 times and changes the safety warning icon Move 3mm to the left to generate a label template that complies with the SAEJ2716 standard. It can be printed adaptively under working conditions. When the humidity in the spray workshop is detected to be rising, the print control module switches to anti-diffusion mode, increases the ink injection pressure from 35kPa to 42kPa, and enables the pre-drying device at the same time. After the printing is completed, a traceability code containing a timestamp and spatial coordinates is generated, and a QR code is embedded in the Base64 format. Multi-spectral quality inspection uses 950nm infrared spectrum imaging to detect the ink penetration depth. When it is detected that the penetration difference in a local area is greater than 15μm, the system automatically marks the label as invalid and triggers the reprinting instruction for the same batch. At the same time, the defect image is stored in the quality traceability database.
[0152] Implementation process of the finished product warehousing method: The finished crankshaft products with completed labels enter the warehousing process, information binding and dimension collection, the RFID reader writes the traceability code into the UHF chip of the finished product label, and at the same time the 3D vision system generates the crankshaft outer contour point cloud data, calculates the minimum circumscribed rectangular size, and dynamically allocates storage locations. The ant colony algorithm generates a path plan based on three optimization objectives: transportation distance weight, retrieval time weight, and space utilization weight. It is finally allocated to the B2-15-07 storage location in the heavy-duty shelf area, and an AGV obstacle avoidance path is generated. Digital twin verification, after the AGV is transported to the location, the lidar scans the actual storage location coordinates, compares them with the theoretical coordinates in the digital twin system, measures the Euclidean distance, and the system automatically updates the inventory status to "in storage" and marks the storage capacity occupancy rate as 82% on the topological map.
[0153] System synergy instance: In the concurrent processing of gearbox housings, the system exhibits the following synergy mechanisms: label-warehouse data linkage, when the quality inspection module detects that the batch number of the housing label does not match the purchase order in the ERP system, the finished product warehouse-in process is immediately frozen, and an alarm code E007 is sent to the MES system. After manual confirmation that the data acquisition terminal transmission is delayed, the system automatically corrects the batch number and reactivates the warehouse-in task. Emergency order priority processing: when receiving an urgent order from a VIP customer, the scheduling module dynamically adjusts the evaporation factor of the ant colony algorithm, compresses the number of warehouse location search iterations from 50 to 20, and reduces the AGV transportation path planning time from 8.2 seconds to 3.5 seconds, ensuring that the finished product of the urgent order completes the whole process from label printing to warehouse-in within 15 minutes.
[0154] Example two, smart home appliance manufacturing enterprise label printing and warehouse-in task
[0155] Raw material batch label printing implementation: when the supplier delivers electronic components, the warehouse administrator scans the delivery order QR code through the PDA, triggering the label generation process. The system automatically extracts the material code, supplier batch number, and expiration date, generates a material label template that meets the IEC60417 standard through a double-channel neural network model, and dynamically activates the anti-fading mode when the warehouse environment humidity reaches 75%. The UV ink jet pressure is increased from the standard value of 38kPa to 45kPa, and the preheating plate is turned on to preheat the substrate to 60°C. After printing, the multispectral imager automatically detects the ink adhesion, and when the local area penetration depth difference is greater than 12μm, the system automatically marks the batch label and triggers the reprint instruction, and locks the supplier batch for quality traceability.
[0156] Production process flow label implementation: after the SMT patching process is completed, the production line sensor triggers the process label printing, and the MES system pushes the PCBA board serial number, process parameters, and defect detection results in real time. The deep learning design unit automatically matches the enterprise VI specification, and when the safety warning icon size deviation exceeds ±0.3mm, the dynamic optimization module moves the icon 2.1mm to the right and enlarges it to the standard proportion. When the AGV transports the carrier carrying the PCBA to the next station, the reader automatically scans the RFID chip in the label to check the process completion status. When a board card that has not completed the burning process is mistakenly flowed into the test station, the system immediately freezes the board card and triggers an audible and light alarm, and synchronizes the abnormal data to the quality board.
[0157] Product packaging label implementation: The air conditioner controller that has passed final inspection enters the packaging link, the 3D vision system collects product outline data, and matches it with the outer box specifications required by the order. The label generation module performs triple verification, parses customer special requirements through the LSTM network, optimizes the graphic layout based on the CNN network, dynamically calculates the information density value to avoid content overload, and during the printing process, the vibration sensor detects abnormal vibration on the production line. The system immediately switches to the anti-offset mode, reduces the inkjet frequency from 20kHz to 12kHz, and extends the drying time to 1.3 times the standard value. The generated label is embedded with a four-dimensional traceability code.
[0158] Automatic warehouse scheduling implementation: The finished product after labeling is transported to the warehouse by AGV, the RFID reader writes the traceability code into the UHF chip, and the warehouse space optimization unit generates a path plan based on three constraints. The space constraint dynamically calculates the shelf load capacity based on the three-dimensional size of the finished product. The efficiency constraint combines historical data to predict the monthly access frequency of B2-07 warehouse. The time constraint automatically promotes the path priority of VIP orders. After the AGV is transported to the target warehouse, the laser radar scans the actual coordinates and compares them with the theoretical coordinates of the digital twin system. The system automatically updates the inventory status and releases AGV resources. When the warehouse capacity utilization reaches 90%, the dynamic evaporation factor of the ant colony algorithm is triggered, reducing the warehouse search iteration from 100 times to 35 times, and the path planning time from 15 seconds to 4.2 seconds.
[0159] System fault tolerance and linkage mechanism: In the typical fault scenario of concurrent processing of export orders, the label data conflicts, the customer address pushed by the ERP system does not match the FCC identification version required by the logistics system, and the compliance verification unit automatically calls the FCC Part 15 standard template in the version library to cover the error data. When the AGV detects a sensor failure in the C area shelf during transportation, the scheduling module immediately enables the redundancy strategy, dynamically allocates the 12 boxes of finished products originally intended for C2-11 to D3-09, and corrects the inventory topology map through the digital twin system.
[0160] It should be noted that, in the present document, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.
[0161] While embodiments of the application have been shown and described, it is to be understood that the application is not limited to the details of the embodiments described, since numerous changes, modifications, substitutions and variations can be made thereto without departing from the spirit and scope of the application as defined by the appended claims and their equivalents.
Claims
1. A method for printing labels on finished products, characterized in that: The method comprises the following steps: S1. Obtain basic information of finished products, real-time parameters of production lines, and historical label template library data through multi-source data acquisition terminals to generate original data sets of label elements; S2. Perform feature fusion processing on the original data set of label elements based on a deep neural network model, extract the core feature vector and layout feature vector of the label content, and generate a label intelligent design parameter set; S3. Calculate the matching degree between the label intelligent design parameter set and the preset enterprise visual recognition specification. When the matching degree is lower than a threshold, a dynamic optimization module is triggered to generate compliant label template data. S4. Adjust the working mode of the printing equipment according to the real-time working parameters of the production line, output the compliant label template data to the surface of the finished product through a high-precision inkjet printing system, and generate a label traceability code containing a time and space stamp; S5. Use a multispectral imager to collect the printed label image, and perform image defect detection through a convolutional neural network. When the detection is qualified, a finished label completion signal is output, otherwise a label reprinting instruction is triggered.
2. The method for printing finished product labels according to claim 1, characterized in that: Said S1 comprises the following steps: S11. Collecting finished product material parameters, geometric dimension data, and batch numbers through the IoT sensor network to form a subset of finished product basic information; S12. Real-time acquisition of production line environmental temperature and humidity, equipment vibration spectrum, and conveyor belt speed parameters to generate a subset of working condition dynamic parameters; S13. Retrieve historical tag template library data from the distributed storage system, and extract the template structure feature matrix and content element distribution heat map.
3. The method for printing finished product labels according to claim 1, wherein: The S2 comprises the following steps: S21. Build a dual-channel feature extraction network. The first channel uses an LSTM network to process text content features, and the second channel uses a CNN network to process layout structure features. S22, fuse the dual-channel output features through the attention mechanism to generate a feature tensor containing the content priority weight; S23. Use the generative adversarial network to enhance the feature tensor and output the label intelligent design parameter set Among them C content is the content feature matrix, L layout is the layout feature vector, and α, β are dynamic adjustment coefficients.
4. The method for printing finished product labels, the method for warehousing finished products, and the system according to claim 1, characterized in that: The S3 includes the following steps: S31. Establish a three-dimensional evaluation space for corporate visual identity standards: Among them, the color specification vector is the RGB / CMYK color value vector, the layout compliance matrix is the layout rule matrix, and the information density threshold is the maximum amount of information per unit area; S32. Calculate the Manhattan distance between the design parameter set and the specification space: in, is the i-th dimension feature of the design parameter set, is the standard value of the i-th dimension of the specification; S33, when D m >δ, the dynamic optimization module is started, and the compliant label template data T is output through the constraint optimization algorithm. compliant .
5. A method for warehousing finished products, characterized by: The method comprises the following steps: P1, receiving the finished product label completion signal from claim 1, activating the RFID reader / writer to write product traceability information; P2, collect the finished product outer contour point cloud data through the 3D vision system and generate a three-dimensional dimension feature vector; P3, calculate the optimal storage location allocation plan based on the ant colony algorithm and generate storage location navigation path data; P4 drives the AGV transport system to deliver finished products to the target warehouse along the navigation path, and updates the inventory topology map in real time; P5. When the finished product arrives at the designated storage location, the multimodal verification system is triggered to check the relationship between the physical location and the digital mapping.
6. A method for warehousing finished products according to claim 5, characterized in that: The P3 comprises the following steps: P31. Establish the objective function of storage location optimization: Among them D k is the transportation distance cost, is the estimated retrieval time, ω1 is the weight coefficient of transportation distance, and ω2 is the weight coefficient of retrieval time; P32. Initialize the pheromone matrix τ ij (0) = τ0, set the ant colony size N ant ; P33, using dynamic volatile factors to update pheromones: t ij (t+1)=(1-ρ(t))·τ ij (t)+Δτ ij p(t)=p min +(r max -r min )·e -λt Among them, λ is the adaptive adjustment coefficient, τ ij (t) is the pheromone of path (i, j) at time t, ρ(t) is the adaptive volatility factor, Δτ ij The amount of new pheromone added.
7. The method for warehousing finished products according to claim 5, characterized in that: The P5 comprises the following steps: P51, scan the actual coordinates of the storage location through the laser radar to generate the spatial coordinate vector L real ; P52. Retrieve the theoretical location coordinates L from the digital twin system virtual ; P53. Calculate coordinate alignment error: AND align =||L real -THE virtual ||2 Among them, L real is the actual coordinate of the laser radar scan, L virtual is the theoretical coordinate in the digital twin system; P54, when E align When <ε, the inventory status is updated to "in stock", otherwise the position correction instruction is triggered.
8. A system for printing and warehousing finished product labels, characterized by: include: Label intelligent generation module, multimodal printing control module, intelligent warehousing scheduling module; The label intelligent generation module includes: Multi-source data fusion unit, integrating real-time production line data and historical template features; Deep learning design unit equipped with a dual-channel neural network architecture; Compliance verification unit with built-in dynamic optimization algorithm engine; The multimodal printing control module includes: Adaptive printing unit, supports UV inkjet / laser engraving dual mode switching; Online quality inspection unit, integrating spectrum analyzer and defect detection model; The intelligent warehousing scheduling module includes: 3D visual perception unit, equipped with a ToF depth camera array; The storage location optimization unit implements hardware acceleration of the ant colony algorithm; Digital twin mapping unit to build real-time inventory topology model.
9. The system according to claim 8, characterized in that: The multi-source data fusion unit is equipped with a time-space alignment processor, and its data synchronization equation is: in, is the rate of change of sensor data of type i, is the change rate of historical template data.
10. The system according to claim 8, characterized in that: The storage location optimization unit includes an FPGA accelerator, and its hardware implementation architecture includes: Pheromone update pipeline, processing N per cycle path Path calculation; Parallel cost calculation unit with built-in 8×8 matrix operator; Dynamic volatility factor controller supports real-time reconstruction of ρ(t).
Citation Information
Cited By
Variable two-dimensional code printing control method
CN121448015A