Product warehouse-in and warehouse-out control method and system

Abstract

The invention provides a product warehouse-in and warehouse-out control method and system, and the method comprises the steps: determining a commodity identification code of a product, and determining a core specification attribute of the product according to the commodity identification code to form a multi-dimensional attribute; correspondingly storing the trademark identification code and the multi-dimensional attribute; generating a composite two-dimensional code from the commodity identification code and the in-library code of the commodity placement location, and attaching the composite two-dimensional code to the location of the corresponding product; during delivery management, determining product core specification attributes of to-be-delivered commodities, analyzing the product core specification attributes through the NLP technology module, matching the product core specification attributes with the multi-dimensional attribute data model, determining the to-be-delivered commodities in the stock according to the basic attribute dimensions and the functional attribute dimensions, obtaining the current positions, and storing the to-be-delivered commodities in the stock according to commodity placement locations in the composite two-dimensional codes. Path planning is carried out between the current position and the commodity placement position; and after the commodity is arrived at the commodity placement position according to the planned path and the commodities are picked, the warehouse-out unit is subtracted according to the inventory unit to update the inventory record, so that the abstract multi-specification problem is solved, human errors in the working process are avoided, and the working efficiency of warehouse management personnel is improved.

Description

Technical Field

[0001] This invention relates to the field of warehouse management technology, specifically to a product entry and exit control method and system. Background Technology

[0002] Hardware products are widely used in daily life and industrial production. They come in a wide variety of types and have complex specifications. In the warehousing management process, hardware products often exhibit the characteristic of "multi-specification". That is, the same product may have multiple specific specifications due to differences in attributes such as size, material, and process. In addition, the transaction and circulation process often involves the mixed use of multiple units of measurement such as "piece", "box", "carton" and "kilogram". This places high demands on the refinement and intelligence of its warehousing management.

[0003] Currently, common practices for warehouse management of hardware products include manual recording, spreadsheet statistics, or the use of traditional warehouse management systems. For example, Chinese patent application number CN202211472830.5 discloses a QR code-based warehouse management system. The system includes: a first statistical module for calculating the real-time quantity of goods based on acquired inbound and outbound information; an inspection robot for patrolling the warehouse according to a strategy, including calculating the remaining quantity of a particular type of goods after a change in quantity and sending the results to a second statistical module; a second statistical module, independently set up, for real-time updates based on the data collected by the inspection robot; QR code labels placed at corresponding locations on the shelves for each type of goods to display the real-time data collected by the second statistical module; and a processing module for collecting and analyzing data from both the first and second statistical modules in real time, and providing alerts on the warehouse status based on the analysis results.

[0004] However, the above-mentioned documents only use QR codes for inventory statistics, and the inventory corresponding to the same QR code remains fixed. If the inventory changes, the QR code will not change accordingly, and manual modification of the inventory quantity is required. Furthermore, when issuing goods, it is impossible to find the corresponding product or replace the product after scanning the product information. The product must be selected and confirmed manually, resulting in low issuing efficiency. When facing the complex management needs of hardware products, these existing technical solutions generally have the characteristics of low product information standardization and high dependence on human experience and judgment in the inbound and outbound operations, resulting in low efficiency. Summary of the Invention

[0005] This invention provides a product inbound and outbound control method and system that can adapt to the characteristics of multiple product specifications and multiple units, and has a high degree of intelligence in the management process.

[0006] To achieve the above objectives, the technical solution of the present invention is: a product inbound / outbound control method, comprising the following steps: S1. Determine the product identification code, and based on the product identification code, determine the core specifications and attributes of the product to form multi-dimensional attributes. The multi-dimensional attributes include: basic attribute dimension, physical attribute dimension, functional attribute dimension and management attribute dimension. S2. Store the trademark identification code and multi-dimensional attributes accordingly. During the storage process, convert the units in the physical attribute dimension according to the storage unit to complete the product information entry. S3. Generate a composite QR code by combining the product identification code and the warehouse code of the product placement location, and attach the composite QR code to the corresponding product's storage location. When managing outbound shipments using S4, the core specifications and attributes of the goods to be shipped are determined. These are then analyzed using NLP technology and matched with a multidimensional attribute data model. Based on basic and functional attribute dimensions, the goods to be shipped from inventory are identified. Next, the most suitable inventory specifications are searched based on physical and management attribute dimensions, and the conversion from outbound unit to inventory unit is completed. Furthermore, when inventory is insufficient, alternative specifications are recommended based on substitute rules. Simultaneously, the current location is obtained, and path planning is performed between the current location and the product placement location based on the product placement location indicated by the composite QR code. S5. Arrive at the product placement location according to the planned route, pick the products, and update the inventory record by subtracting the outbound unit from the inventory unit.

[0007] The beneficial effects of this invention are as follows: When products are received into the warehouse, the trademark identification code is determined, and the core specifications of the product are further determined to form multi-dimensional attributes. The core specifications of the product are layered from multiple aspects such as function, physical, basic, and management. This facilitates subsequent searching based on units and functions. Then, the product identification code and multi-dimensional attributes are stored to achieve warehouse management. During the receiving process, the units in the multi-dimensional attributes are converted to the inventory units, and the corresponding products are placed in the preset storage locations, thus forming a composite QR code. In the subsequent export process, it is only necessary to first determine the core specifications of the product to be exported, then perform semantic parsing to determine the basic and functional attributes to determine the products to be exported from the inventory, and determine the export unit conversion based on the physical and management attributes. When the inventory is insufficient, the export is carried out using the difference of substitutes according to the substitute rule. After determining the storage location corresponding to the product to be exported, the path planning is determined, which facilitates navigation to the corresponding storage location for export management. After export, the inventory units are updated, thus enabling the entire receiving and export process to be automated without manual unit conversion and substitute selection, making it convenient to implement.

[0008] Furthermore, the basic attribute dimensions include material, type, specifications, and brand information; the physical dimension includes length, diameter, weight, and color information; the functional attribute dimensions include strength level, corrosion resistance level, and applicable environment information; and the management attribute dimensions include basic units, packaging units, conversion relationships, safety stock, and supplier information.

[0009] The above settings allow for hierarchical management of core product specifications based on different attributes, facilitating subsequent searches.

[0010] Furthermore, in step S1, "determining the product identification code of the product packaging" includes scanning the barcode or supplier standard code on the product packaging with a scanning camera, or finding the product identification code by searching for images in the cloud after taking a picture of the product, or obtaining the corresponding multi-dimensional attributes by analyzing the product information through voice or manual input using NLP technology.

[0011] The above settings can determine the product identification code in different ways.

[0012] Furthermore, step S3 also includes: during inventory counting, product information is displayed by scanning a composite QR code, and inventory personnel enter the actual inventory quantity to complete the inventory count; at the same time, dynamic safety stock is set for each product specification, and when the inventory is lower than the threshold, a stockout reminder is triggered, and alternative product specifications in the inventory are recommended based on the alternative product rule.

[0013] The above settings facilitate inventory checks, provide out-of-stock alerts when inventory is low, and recommend alternative products.

[0014] Furthermore, step S3 also includes: during inventory counting, after the mobile terminal scans the composite QR code, it displays product images and core product specifications and attributes, compares them with the actual inventory items, and promptly updates the inventory information corresponding to the composite QR code after confirming the inventory data.

[0015] The above settings allow for comparison of product images and core specifications during inventory checks, thus ensuring the reliability of the inventory process.

[0016] Furthermore, step S4 includes: using A * The algorithm determines the optimal picking path between the current location and the product placement location in the composite QR code; after confirming the location, it displays the product image and core specifications, and after picking, it scans the composite QR code to complete the final verification.

[0017] The above settings, through the determination of the optimal picking route, make route planning more reliable.

[0018] Furthermore, step S4, the NLP technology module parsing and matching with multi-dimensional attribute data, includes: acquiring product input information; if the product input information is product description information, identifying key attribute information from the product description information; based on the weight values ​​in the preset key attribute information and the corresponding first similarity, returning matching products according to the similarity ranking; if the product input information is a product image, preprocessing the image, using a convolutional neural network to extract visual feature vectors, comparing the extracted feature vectors with the feature modules in the database, calculating the second similarity, and returning matching products based on the similarity; combining the first and second similarities with preset first and second confidence levels to form a comprehensive matching similarity, and determining the final matching product.

[0019] The above settings determine the first similarity matching product based on the product input information, then determine the second similarity matching product based on the photograph taken, and finally combine the product input information and the image recognition to determine the final matching product, thus ensuring the reliability of the matching.

[0020] Furthermore, step S4, "based on substitute rules," includes: determining the content matching degree based on the basic attribute dimension and management attribute dimension in the multi-dimensional attributes; then determining the collaborative filtering weight based on the frequency of occurrence of the corresponding substitute within a preset time period and the user acceptance rate of the product to be shipped; the substitute rule algorithm determines the comprehensive substitution score, which is calculated as: comprehensive substitution score = content matching degree × 0.6 + collaborative filtering score × 0.4. When the comprehensive substitution score > preset value, it is listed as a valid substitute.

[0021] The above settings further improve the reliability of replacement products based on historical alternatives after matching the product.

[0022] Furthermore, NLP techniques include: segmenting the input text, part-of-speech tagging, named entity recognition, identifying user query intent and key attributes, searching for matching entities in the product knowledge graph, generating matching results, and sorting them by similarity.

[0023] The above settings can be easily parsed and matched using NLP technology.

[0024] This invention also provides a product inbound / outbound control system, including an intelligent inbound / outbound management module, a cloud-based product database, a data module, and a device module. The intelligent inbound / outbound management module includes determining product labels and inbound / outbound product information. The cloud-based product database stores basic data such as standard product information, specifications, and supplier information. The data module includes a multi-dimensional attribute data model and an intelligent algorithm engine. The intelligent algorithm engine includes an NLP technology module, an image recognition module, a path planning module, and a substitute-based rule module. The device module includes a mobile terminal and a QR code printer. The cloud-based product database is communicatively connected to the intelligent inbound / outbound management module, the intelligent inbound / outbound management module is communicatively connected to the data module, and the data module is communicatively connected to the device module. Attached Figure Description

[0025] Figure 1 This is a flowchart illustrating the product warehousing process of the present invention.

[0026] Figure 2 This is a flowchart illustrating the product shipment process of this invention.

[0027] Figure 3 This is a block diagram of the product inbound and outbound control system in this invention. Detailed Implementation

[0028] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0029] like Figure 3 As shown, a product inbound / outbound control system includes an intelligent inbound / outbound management module, a cloud-based product database, a data module, and a device module. The cloud-based product database is used to search for the corresponding product identification code based on an image when no corresponding product identification code is found. The intelligent inbound / outbound management module includes an inbound and outbound rule engine. The cloud-based product database stores basic data such as standard product information, specifications, and supplier information. The data module includes a multi-dimensional attribute data model and an intelligent algorithm engine. The intelligent algorithm engine includes an NLP (Natural Language Processing) technology module, an image recognition module, a path planning module, and a substitute-based rule module. The device module includes a mobile terminal and a QR code printer. The cloud-based product database is communicatively connected to the intelligent inbound / outbound management module, the intelligent inbound / outbound management module is communicatively connected to the data module, and the data module is communicatively connected to the device module.

[0030] like Figure 1 and Figure 2 As shown, a product inbound / outbound control method includes the following steps: S1. Determine the product's commodity identification code. Based on the commodity identification code, determine the product's core specifications and attributes to form multi-dimensional attributes. These multi-dimensional attributes include: basic attribute dimension, physical attribute dimension, functional attribute dimension, and management attribute dimension. The basic attribute dimension includes material, type, specifications, and brand information. The physical attribute dimension includes length, diameter, weight, and color information. The functional attribute dimension includes strength grade, corrosion resistance grade, and applicable environment information. The management attribute dimension includes basic units, packaging units, conversion relationships, safety stock, and supplier information. For hardware products, the product's trademark identification code is the product ID. Basic attribute dimensions: Material: "Stainless Steel / Copper / Iron / Aluminum", Type: "Screw / Nut / Washer / Bolt", Specification: "M6 / M8 / M10 / M12", Brand: "Brand Name"; Physical attribute dimensions: Length: "Value + Unit", Diameter: "Value + Unit", Weight: "Value + Unit", Color: "Silver / Black / Gold"; Functional attribute dimension: Strength Grade: "8.8 / 10.9 / 12.9". Corrosion resistance rating: "Grade A / Grade B / Grade C", Applicable environment: "Indoor / Outdoor / High temperature / Corrosive environment", Management attribute dimensions: Basic unit: "piece / kilogram / meter", Packaging unit: "box / carton / bag", Conversion relationship: "1 box = 100 pieces", Safety stock: "Quantity threshold", Supplier: "Supplier information".

[0031] In this embodiment, step S1, "determining the product identification code of the product packaging", includes scanning the barcode or supplier standard code on the product packaging with a scanning camera, or for products without barcodes or with unclear barcodes, taking pictures of the products and then searching for the product identification code in the cloud image, or obtaining the corresponding multi-dimensional attributes by analyzing the product information through voice or manual input and NLP technology.

[0032] S2. Store the trademark identification code and multi-dimensional attributes accordingly. During the storage process, the units in the physical attribute dimension are converted according to the storage unit to complete the product information entry. The conversion process is to perform corresponding calculations based on the conversion relationship. For example, if there are currently 50 items and the unit in the inventory unit is a box, then according to the conversion relationship of 1 box = 100 items, it is determined to be 1 / 2 box.

[0033] S3. Generate a composite QR code by combining the product identification code and the warehouse code of the product placement location, and attach the composite QR code to the corresponding product's storage location.

[0034] During inventory checks, product information is displayed by scanning composite QR codes, including product images and core specifications, which are then compared with the actual inventory. After confirming the inventory data, the corresponding inventory information for the composite QR codes is updated promptly. Inventory personnel then input the actual inventory quantity to complete the inventory check. Simultaneously, dynamic safety stock is set for each product specification. When the inventory falls below the threshold, a stockout alert is triggered, and alternative products from the inventory are recommended based on substitute product rules. During S4 outbound management, the core specifications and attributes of the goods to be shipped are determined. The NLP technology module parses and matches these attributes with a multidimensional attribute data model. Based on the basic attribute dimensions and functional attribute dimensions, the goods to be shipped in the inventory are identified. Then, based on the physical attribute dimensions and management attribute dimensions, the most suitable inventory specifications are searched and the conversion from outbound unit to inventory unit is completed. When the inventory quantity is insufficient, alternative specifications are recommended based on the substitute product rule. At the same time, the current location is obtained, and the path planning between the current location and the product placement location is performed based on the product placement location in the composite QR code.

[0035] S5. Arrive at the product placement location according to the planned route, pick the products, and update the inventory record by subtracting the outbound unit from the inventory unit.

[0036] In this embodiment, path planning uses A. * The algorithm determines the optimal picking path between the current location and the product placement location in the composite QR code; after confirming the location, it displays the product image and core specifications, and after picking, it scans the composite QR code to complete the final verification.

[0037] Step S4, the NLP technology module parsing and matching with multidimensional attribute data, includes: acquiring product input information; if the product input information is product description information, identifying key attribute information from the product description information, and returning matching products based on the weight values ​​in the preset key attribute information and the corresponding first similarity, sorting them according to similarity; if the product input information is a product image, preprocessing the image, extracting visual feature vectors using a convolutional neural network, comparing the extracted feature vectors with feature modules in the database, calculating the second similarity using cosine similarity or Euclidean distance, and returning matching products based on the similarity; combining the first and second similarities with preset first and second confidence levels to form a comprehensive matching similarity, and determining the final matching product.

[0038] In this embodiment, the first similarity calculation process is as follows: receiving a vague description input by the user, such as "20 mm stainless steel screw"; entity extraction: identifying key attribute information (length = 20 mm, material = stainless steel, type = screw); weight calculation: assigning matching weights to each attribute, with length weight 0.4, material weight 0.3, and type weight 0.3; similarity calculation: performing a first similarity calculation with all products in the database; result sorting: sorting by the first similarity score and returning the most matching product.

[0039] The second similarity calculation process involves: denoising, enhancing, and standardizing the captured product images; feature extraction: extracting visual feature vectors using a convolutional neural network; feature comparison: comparing the extracted feature vectors with feature templates in the database; similarity evaluation: calculating cosine similarity or Euclidean distance; result recommendation: returning the top N product options with the highest similarity.

[0040] Fusion matching process: Information fusion: Combining the first similarity of text descriptions and the second similarity of image features; Weight allocation: Dynamically adjusting the weights of each modality based on confidence levels (first and second confidence levels are preset parameters); Comprehensive scoring: Calculating a comprehensive matching score; Intelligent recommendation: Providing the most likely matching result.

[0041] Step S4, "based on substitute rules," includes: determining the content matching degree based on the basic attribute dimension and management attribute dimension in the multi-dimensional attributes; then determining the collaborative filtering weight based on the frequency of occurrence of the corresponding substitute within a preset time period and the user acceptance rate of the product to be shipped; the substitute rule algorithm determines the comprehensive substitution score, where the comprehensive substitution score = content matching degree × 0.6 + collaborative filtering score × 0.4. When the comprehensive substitution score > a preset value, it is listed as a valid substitute. In this embodiment, if it is found that a production line uses 'M8×25' to replace 'M8×20' for three consecutive months without any quality problems, then the collaborative filtering score is determined to be the highest score.

[0042] NLP technology includes: word segmentation of input text, part-of-speech tagging, named entity recognition, identification of user query intent and key attributes, finding matching entities in the product knowledge graph, generating matching results and sorting them by similarity.

[0043] The working principle of this invention is as follows: When products are received into the warehouse, the trademark identification code is determined, and the core specifications and attributes of the product are further determined to form multi-dimensional attributes. The core specifications and attributes of the product are layered from multiple aspects such as function, physical, basic, and management. This facilitates subsequent retrieval based on units and functions. Then, the product identification code and multi-dimensional attributes are stored to achieve warehouse management. During the receiving process, the units in the multi-dimensional attributes are converted to the inventory units. Then, the corresponding products are placed in the preset storage locations, thus forming a composite QR code. In the subsequent export process, it is only necessary to first determine the core specifications and attributes of the product to be exported, then perform semantic parsing to determine the basic and functional attributes to determine the product to be exported from the inventory, and determine the export unit conversion based on the physical and management attributes. If the inventory is insufficient, the product is exported using a substitute substitution rule. After determining the storage location of the product to be exported, path planning is determined, which facilitates navigation to the corresponding storage location for export management. After export, the inventory units are updated, thus automating the entire receiving and export process without the need for manual unit conversion and substitute selection, making it convenient to implement.

Claims

1. A method for controlling product entry and exit from a warehouse, characterized in that: Includes the following steps: S1. Determine the product identification code, and based on the product identification code, determine the core specifications and attributes of the product to form multi-dimensional attributes. The multi-dimensional attributes include: basic attribute dimension, physical attribute dimension, functional attribute dimension and management attribute dimension. S2. Store the trademark identification code and multi-dimensional attributes accordingly. During the storage process, convert the units in the physical attribute dimension according to the storage unit to complete the product information entry. S3. Generate a composite QR code by combining the product identification code and the warehouse code of the product placement location, and attach the composite QR code to the corresponding product's storage location. When managing outbound shipments using S4, the core specifications and attributes of the goods to be shipped are determined. These are then analyzed using NLP technology and matched with a multidimensional attribute data model. Based on basic and functional attribute dimensions, the goods to be shipped from inventory are identified. Next, the most suitable inventory specifications are searched based on physical and management attribute dimensions, and the conversion from outbound unit to inventory unit is completed. Furthermore, when inventory is insufficient, alternative specifications are recommended based on substitute rules. Simultaneously, the current location is obtained, and path planning is performed between the current location and the product placement location based on the product placement location indicated by the composite QR code. S5. Arrive at the product placement location according to the planned route, pick the products, and update the inventory record by subtracting the outbound unit from the inventory unit.

2. The product inbound / outbound control method according to claim 1, characterized in that: The basic attribute dimensions include material, type, specifications, and brand information; the physical dimension dimensions include length, diameter, weight, and color information; the functional attribute dimensions include strength grade, corrosion resistance grade, and applicable environment information; and the management attribute dimensions include basic units, packaging units, conversion relationships, safety stock, and supplier information.

3. The product inbound / outbound control method according to claim 1, characterized in that: Step S1, "determining the product identification code on the product packaging," includes scanning the barcode or supplier standard code on the product packaging with a scanning camera, or identifying the product identification code by taking a picture of the product and then searching for it in the cloud image; and obtaining the corresponding multi-dimensional attributes by analyzing product information through voice or manual input using NLP technology.

4. The product inbound / outbound control method according to claim 1, characterized in that: Step S3 also includes: during inventory counting, product information is displayed by scanning a composite QR code, and inventory personnel enter the actual inventory quantity to complete the inventory count; at the same time, dynamic safety stock is set for each product specification, and when the inventory is lower than the threshold, a stockout reminder is triggered, and alternative product specifications in the inventory are recommended based on the alternative product rule.

5. The product inbound / outbound control method according to claim 4, characterized in that: Step S3 also includes: during inventory counting, after the mobile terminal scans the composite QR code, it displays product images and core product specifications and attributes, compares them with the actual inventory items, and promptly updates the inventory information corresponding to the composite QR code after confirming the inventory data.

6. The product inbound / outbound control method according to claim 1, characterized in that: Step S4 includes: using A * The algorithm determines the optimal picking path between the current location and the product placement location in the composite QR code; after confirming the location, it displays the product image and core specifications, and after picking, it scans the composite QR code to complete the final verification.

7. The product inbound / outbound control method according to claim 1, characterized in that: Step S4, the NLP technology module parsing and matching with multidimensional attribute data, includes: acquiring product input information; if the product input information is product description information, identifying key attribute information from the product description information, and returning matching products based on the weight values ​​in the preset key attribute information and the corresponding first similarity, sorting them according to similarity; if the product input information is a product image, preprocessing the image, using a convolutional neural network to extract visual feature vectors, comparing the extracted feature vectors with feature modules in the database, calculating the second similarity, and returning matching products based on the similarity; combining the first and second similarities with preset first and second confidence levels to form a comprehensive matching similarity, and determining the final matching product.

8. The product inbound / outbound control method according to claim 1, characterized in that: Step S4, "based on substitute rules", includes: determining the content matching degree based on the basic attribute dimension and management attribute dimension in the multi-dimensional attributes, and then determining the collaborative filtering weight based on the frequency of occurrence of the corresponding substitute within a preset time and the user acceptance rate of the product to be shipped; the substitute rule algorithm determines the comprehensive substitution score, which is: comprehensive substitution score = content matching degree × 0.6 + collaborative filtering score × 0.

4. When the comprehensive substitution score > preset value, it is listed as a valid substitute.

9. The product inbound / outbound control method according to claim 1, characterized in that: NLP technology includes: word segmentation of input text, part-of-speech tagging, named entity recognition, identification of user query intent and key attributes, finding matching entities in the product knowledge graph, generating matching results and sorting them by similarity.

10. A product inbound / outbound control system, used to implement the product inbound / outbound control method according to any one of claims 1-9, characterized in that: The system includes an intelligent inbound / outbound management module, a cloud-based product database, a data module, and a device module. The intelligent inbound / outbound management module determines product codes and records inbound and outbound product information. The cloud-based product database stores basic data such as standard product information, specifications, and supplier information. The data module includes a multi-dimensional attribute data model and an intelligent algorithm engine. The intelligent algorithm engine includes an NLP technology module, an image recognition module, a path planning module, and a substitute-based rule module. The device module includes a mobile terminal and a QR code printer. The cloud-based product database is communicatively connected to the intelligent inbound / outbound management module, the intelligent inbound / outbound management module is communicatively connected to the data module, and the data module is communicatively connected to the device module.

Images