Method and device for planning a special piece type rack

Abstract

This invention discloses a planning method and apparatus for special component type shelving, relating to the field of warehouse management technology. One specific implementation of the method includes: acquiring attribute data, order data, and inventory data of goods in the warehouse, and calculating the average daily outbound volume, average daily inventory volume, and best-selling category for each good-selling item; inputting the attribute data of each good-selling item into a pre-trained large language model for special component type identification to obtain the component type category to which each good-selling item belongs; and solving a pre-established mixed integer programming model based on the average daily outbound volume, average daily inventory volume, best-selling category, and component type category to obtain a warehouse shelving selection scheme. This implementation can accurately identify the geometric shape and physical attributes of special automotive components, avoiding the limitations of traditional methods that rely on only a single dimension for classification, achieving differentiated shelving configuration and comprehensive selection, and improving the accuracy and rationality of warehouse shelving selection.

Description

Technical Field

[0001] This invention relates to the field of warehouse management technology, and in particular to a planning method and apparatus for special type shelving. Background Technology

[0002] Automotive warehouse racking planning is a core aspect of automotive parts supply chain management. Appropriate racking selection directly impacts warehousing efficiency, space utilization, and operating costs. Automotive warehouses store a large number of parts of various shapes and sizes. Specialized parts such as doors and hoods, bumpers, side panels, glass, and tires, due to their unique geometry, dimensions, and weight characteristics, require specialized storage racking to ensure storage safety and efficient space utilization.

[0003] In existing technologies, the planning of racking in automotive warehouses mainly faces the following challenges and shortcomings: First, traditional methods usually rely solely on product names or simple size rules for judgment, which are limited in their means of identifying special parts and lack sufficient accuracy. This leads to a large number of special parts being incorrectly planned for regular shelves, resulting in storage risks or wasted space. Second, existing research focuses on the optimization of storage locations or picking routes in conventional warehouses, or adopts empirical and uniform shelf configuration standards. There are few solutions that can make refined classification and shelf selection based on the characteristics of product type (special / regular) and turnover characteristics (fast-selling / slow-moving). There is a lack of comprehensive decision-making methods that integrate and differentiate the selection of special and regular product types. Third, most existing models only consider the purchase or depreciation costs of the racking itself and warehouse rental costs, while ignoring the replenishment operation costs caused by unreasonable racking selection (such as forklift driver labor costs and forklift equipment costs). The cost optimization models for racking selection do not fully characterize the actual operational constraints.

[0004] Therefore, how to construct an intelligent, accurate, and integrated method for selecting automotive warehouse racking, taking into account the special storage needs of special parts, the turnover efficiency of regular parts, size matching constraints, and the operating costs of the entire chain (equipment, personnel, warehouse rent, and racking), is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a planning method and apparatus for special-type shelving, which can accurately identify the geometric shape and physical attributes of special automotive parts, avoiding the limitations of traditional methods that rely on only a single dimension for classification, and providing a precise data foundation for subsequent dedicated shelving matching. Simultaneously, by combining warehouse storage equipment, personnel, shelving, and warehouse rent costs, the total warehouse operating cost is calculated. With the goal of minimizing the total warehouse operating cost, a mixed-integer programming model based on large-scale special-type part identification is established. The model constraints control the correspondence between picking location size and product size. Shelving selection is based on the special-type part identification results and picking location turnover days for shelving planning. This allows for the selection of the optimal shelving combination from the full range of shelving while meeting the requirements of dedicated storage for special-type parts and safety stock for regular-type picking locations. This achieves differentiated shelving configuration and comprehensive selection, and also fully and comprehensively characterizes warehouse storage costs, improving the accuracy and rationality of warehouse shelving selection.

[0006] To achieve the above objectives, according to one aspect of the present invention, a method for planning special-type shelving is provided, comprising: Obtain attribute data, order data, and inventory data of goods in the warehouse, and calculate the average daily outbound volume and average daily inventory volume of each goods based on the order data and inventory data. Determine the best-selling category of each goods based on the order data and the best-selling classification rules. The attribute data of each product is input into a pre-trained large language model for special part type recognition to obtain the part type category to which each product belongs. The part type category includes at least one part type category corresponding to the special part type and a regular part type category. The attribute data includes name, size and weight. Based on the average daily outbound volume, average daily inventory volume, best-selling category, and product type of each product, a pre-established mixed integer programming model is solved to obtain a warehouse racking selection scheme. The scheme includes the racking type and required quantity allocated to each product. The mixed integer programming model aims to minimize the total operating cost, which includes racking floor space cost, racking depreciation cost, replenishment personnel cost, and forklift depreciation cost. The constraints of the mixed integer programming model include at least: turnover days constraint, storage volume constraint, storage location capacity constraint, size matching constraint, and total demand constraint.

[0007] Optionally, the turnover days constraint is used to associate the picking location turnover days of a product with the preset turnover days of its best-selling category; the storage volume constraint is used to determine the required picking location volume and replenishment location volume for each product based on the average daily outbound volume, average daily inventory volume, and picking location turnover days; the storage location capacity constraint is used to ensure that the total volume of picking locations and replenishment locations allocated to a product is not less than its required picking location volume and replenishment location volume, respectively; the size matching constraint is used to restrict the size of the storage locations on the shelves allocated to the product to be greater than or equal to the size of the product; and the total demand constraint is used to calculate the total number of various shelves required to meet the storage needs of all products.

[0008] Optionally, the attribute data of each product is input into a pre-trained large language model for special part type recognition to obtain the part type category to which each product belongs. This includes: inputting the attribute data of each product into a pre-trained large language model to perform special part type recognition based on preset prompt words to obtain the part type category to which each product belongs, the recognition judgment basis, and the confidence level; wherein, the prompt words include predefined name features, size features, and weight features for different special part types.

[0009] Optionally, special component type identification is performed based on preset prompt words to obtain the component type category to which each product belongs, including: analyzing whether the product name contains keywords of the name feature for name matching; in response to successful name matching, determining a first identification result based on the matched name feature, determining whether the size of the product conforms to the size range corresponding to the size feature of the first identification result to obtain a second identification result; determining whether the weight of the product is within the weight range corresponding to the weight feature of the first identification result to obtain a third identification result; and obtaining the component type category to which each product belongs based on the second identification result and the third identification result.

[0010] Optionally, obtaining the component type category of each product based on the second identification result and the third identification result includes: determining the component type category of each product based on the intersection of the second identification result and the third identification result if there is no conflict between the second identification result and the third identification result; and determining the component type category of each product based on the matching degree between the product and the second identification result and the third identification result if there is a conflict between the second identification result and the third identification result.

[0011] Optionally, the size matching constraint is achieved by pre-generating a set of optional shelves for each product: traversing all optional shelves, and adding shelves whose length, width, and height are all greater than or equal to the length, width, and height of the product to the set of optional shelves for the product.

[0012] Optionally, before solving the pre-established mixed integer programming model, the method further includes: responding to the fact that the part type category to which the product belongs is the part type category corresponding to a special part type, matching the corresponding special shelf type according to the part type category to which the product belongs, and selecting the optional shelf set of the product from the shelves of the special shelf type; wherein, when the part type category to which the product belongs is two covers, the optional shelf set of the product is limited to door racks or special cover racks; when the part type category to which the product belongs is bumpers, the optional shelf set of the product is limited to bumper-specific racks; when the part type category to which the product belongs is glass, the optional shelf set of the product is limited to pallet picking positions or special wooden racks.

[0013] Optionally, the best-selling product classification rule includes: classifying products into best-selling, ordinary, and slow-moving products based on the average daily number of outbound order lines in the order data; and the constraint conditions also include a slow-moving product constraint, which sets the picking location volume and picking location turnover days of slow-moving products to zero, and does not allocate picking locations to slow-moving products.

[0014] Optionally, the objective function of the mixed-integer programming model is: Min L = warehouse rent cost + rack depreciation cost + replenishment personnel cost + forklift depreciation cost; where, warehouse rent cost = unit area warehouse rent cost × (total projected area of ​​all picking racks and replenishment racks); rack depreciation cost = depreciation cost of all picking racks + depreciation cost of all replenishment racks; replenishment personnel cost = total monthly replenishment task volume / (replenishment efficiency × daily working hours × monthly attendance days) × number of days in the month; forklift depreciation cost = number of forklifts required × unit depreciation cost of forklifts.

[0015] According to another aspect of the present invention, a planning device for a special type of shelving is provided, comprising: The data acquisition and processing module is used to acquire attribute data, order data and inventory data of goods in the warehouse, and calculate the average daily outbound volume and average daily inventory volume of each goods based on the order data and the inventory data, and determine the best-selling category of each goods based on the order data and the best-selling classification rules. The product part type recognition module is used to input the attribute data of each product into a pre-trained large language model to identify special part types and obtain the part type category to which each product belongs. The part type category includes at least one part type category corresponding to special part types and regular part type categories. The attribute data includes name, size and weight. The selection scheme generation module is used to solve a pre-established mixed integer programming model based on the average daily outbound volume, average daily inventory volume, best-selling category, and component type of each product to obtain a warehouse racking selection scheme. The scheme includes the racking type and required quantity allocated to each product. The mixed integer programming model aims to minimize the total operating cost, which includes racking floor space cost, racking depreciation cost, replenishment personnel cost, and forklift depreciation cost. The constraints of the mixed integer programming model include at least: turnover days constraint, storage volume constraint, storage location capacity constraint, size matching constraint, and total demand constraint.

[0016] According to another aspect of the present invention, an electronic device is provided, comprising: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the special component type shelf planning method provided in the embodiments of the present invention.

[0017] According to another aspect of the present invention, a computer-readable medium is provided having a computer program stored thereon, which, when executed by a processor, implements the special-type shelving planning method provided in the embodiments of the present invention.

[0018] According to another aspect of the present invention, a computer program product is provided, including a computer program that, when executed by a processor, implements the special-type shelving planning method provided in the embodiments of the present invention.

[0019] One embodiment of the above invention has the following advantages or beneficial effects: By acquiring attribute data, order data, and inventory data of goods in the warehouse, and calculating the average daily outbound volume and average daily inventory volume of each goods based on the order data and inventory data, the best-selling category of each goods is determined according to the order data and best-selling classification rules; the attribute data of each goods is input into a pre-trained large language model for special part type recognition to obtain the part type category to which each goods belong; based on the average daily outbound volume, average daily inventory volume, best-selling category, and part type category of each goods, a pre-established mixed integer programming model is solved to obtain the technical solution for warehouse racking selection. The large language model can intelligently identify special part types of goods such as automotive parts by using multiple dimensions of attributes such as name, size (length, width, height), and weight. It can accurately identify the geometric shape and physical attributes of special automotive parts, avoiding the limitations of traditional methods that rely on only a single dimension for classification. The multimodal recognition capability of the large model can effectively distinguish parts with similar shapes but different uses, providing a precise data foundation for subsequent specialized racking matching. Simultaneously, the total warehouse operating cost is calculated by combining the costs of equipment, personnel, shelving, and warehouse rent. With the goal of minimizing the total warehouse operating cost, a mixed integer programming model based on special item type identification is established. The model constraints control the correspondence between picking location size and product size. Shelving selection is carried out based on the special item type identification results and picking location turnover days. Under the premise of meeting the requirements of dedicated storage for special item types and safety stock for picking locations of regular item types, the optimal shelving combination is selected from the full range of shelving. This not only realizes differentiated shelving configuration and comprehensive selection, but also fully and comprehensively describes the warehouse storage cost, improving the accuracy and rationality of warehouse shelving selection.

[0020] The further effects of the aforementioned unconventional alternative methods will be explained below in conjunction with specific implementation methods. Attached Figure Description

[0021] The accompanying drawings are provided to better understand the invention and are not intended to unduly limit the scope of the invention. Wherein: Figure 1 This is a schematic diagram of the main steps of a special component type shelf planning method according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the planning principle of a special-type shelving unit according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the planning process for a special type of shelving according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the main modules of the planning device for a special type of shelving according to an embodiment of the present invention; Figure 5This is an exemplary system architecture diagram in which embodiments of the present invention can be applied; Figure 6 This is a schematic diagram of the structure of a computer system suitable for implementing terminal devices or servers of the present invention. Detailed Implementation

[0022] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0023] It should be noted that the technical solutions disclosed in this invention, regarding the collection, updating, analysis, processing, use, transmission, and storage of user personal information, all comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken to prevent unauthorized access to user personal information data and to safeguard user personal information security, network security, and national security.

[0024] To address at least one technical problem in existing technologies, this invention provides a planning method for special-type shelving. By using a large language model (hereinafter referred to as "large model"), it intelligently identifies multiple attributes of automotive parts and other goods, such as product name, dimensions (length, width, height), and weight. This accurately identifies special-type parts like doors and hoods, bumpers, side panels, glass, and tires, and precisely identifies their geometric shape and physical properties, avoiding the limitations of traditional methods that rely on only a single dimension for classification. The multimodal recognition capability of the large model effectively distinguishes parts with similar shapes but different uses, providing a precise data foundation for subsequent matching of specialized shelving. Simultaneously, the total warehouse operating cost is calculated by combining the costs of equipment, personnel, shelving, and warehouse rent. With the goal of minimizing the total warehouse operating cost, a mixed integer programming model based on special item type identification is established. The model constraints control the correspondence between picking location size and product size. Shelving selection is carried out based on the special item type identification results and picking location turnover days. Under the premise of meeting the requirements of dedicated storage for special item types and safety stock for picking locations of regular item types, the optimal shelving combination is selected from the full range of shelving. This not only realizes differentiated shelving configuration and comprehensive selection, but also fully and comprehensively describes the warehouse storage cost, improving the accuracy and rationality of warehouse shelving selection.

[0025] Figure 1 This is a schematic diagram illustrating the main steps of a special-type shelving planning method according to an embodiment of the present invention. Figure 1 As shown, the planning method for special component type shelves in this embodiment of the invention mainly includes the following steps S101 to S103.

[0026] Step S101: Obtain the attribute data, order data, and inventory data of the goods in the warehouse, and calculate the average daily outbound volume and average daily inventory volume of each goods based on the order data and inventory data. Determine the best-selling category of each goods based on the order data and the best-selling classification rules.

[0027] According to embodiments of the present invention, in addition to acquiring product attribute data, order data, and inventory data, preset best-selling ranking rules, optional shelf information, and unit cost data for equipment, personnel, and shelves can also be acquired during data acquisition. Inventory data is the data source for the large model to identify special part types, while inventory data and order data are the data sources for the average daily inventory volume and average daily outbound volume of processed goods. Product inventory data mainly includes: product name, length, width, height, weight, and inventory quantity. Product order data mainly includes: order number, product name, product quantity, and outbound time. By aggregating inventory data and order data over a certain period, the average daily inventory volume and average daily outbound volume of the goods can be calculated. Optional shelving information includes the shelving number, location, length, width, height, and number of layers for both standard and special-type shelving (such as beam shelving commonly found in automotive warehouses, D700 shelving, floor stacks, and some customized shelving for special parts). It also includes the number of storage locations on each layer and the length, width, and height of each storage location. Based on this information, the projected area of ​​the shelving can be calculated, as well as the volume of goods that can be stored in each storage location. The matching relationship between goods and shelving / storage locations can be processed, and it can be determined whether a good can be placed in a specific shelf location. Specifically, a good can only be placed in a shelf location if its length, width, and height are all smaller than the length, width, and height of the storage location. The best-selling ranking rules, combined with order data, can label goods based on their sales volume. Generally, goods can be divided into best-selling, average, and slow-moving items.

[0028] In one embodiment of the present invention, determining the best-selling category of each product may specifically include: classifying products based on the absolute value of the average daily number of outbound order lines: if the average daily number of outbound order lines is greater than a first threshold, it is classified as a best-selling product; if the average daily number of outbound order lines is 0, it is classified as a slow-moving product; if the average daily number of outbound order lines is between 0 and the first threshold, it is classified as a regular product. Specifically, since the types of goods in the automotive warehouse vary greatly, outbound order lines (the sub-order data corresponding to a product type included in an outbound order is a single outbound order line) often better represent the outbound behavior of the goods. Therefore, the best-selling status of a product can be determined by combining the absolute value of the average daily number of outbound order lines. For example, if it is greater than a certain number of lines, it is a best-selling product; if it is 0, it is a slow-moving product; and if it is between 0 and the certain number of lines, it is a regular product.

[0029] Step S102: Input the attribute data of each product into a pre-trained large language model for special part type recognition to obtain the part type category of each product. The part type category includes at least one part type category corresponding to the special part type and a regular part type category. The attribute data includes name, size, and weight. Among them, size refers to the length, width, and height data of the product.

[0030] In one embodiment, the component categories corresponding to special component types include, for example, tires, covers, engines, side panels, glass, seats, doors, bumpers, and batteries. For special component types, the corresponding goods can be stored in their respective dedicated shelf locations; for regular component types, the corresponding goods can be stored in regular shelf locations.

[0031] Large-scale models can comprehensively analyze input product attribute data using deep learning algorithms to establish a mapping relationship between product features and corresponding product categories. Specifically, for four-door, two-hatch products, the large-scale model focuses on analyzing their length, width, and height proportions, weight range, and keyword features in their names; for bumper products, it mainly identifies them through the geometric proportions of length, width, and height, weight density, and name identifiers; side panel products are judged based on their large-size flat panel features, length, width, and height proportions, corresponding weight, and name information; glass products are classified based on their specific length, width, and height dimensions, relatively light weight characteristics, and the glass identifier in their names; and tire products are identified through their approximately square length and width dimensions, standardized height, rubber material weight characteristics, and name keywords.

[0032] According to one embodiment of the present invention, the attribute data of each product is input into a pre-trained large language model for special part type recognition to obtain the part type category to which each product belongs. This includes: inputting the attribute data of each product into a pre-trained large language model to perform special part type recognition based on preset prompt words to obtain the part type category to which each product belongs, the recognition judgment basis, and the confidence level; wherein, the prompt words include predefined name features, size features, and weight features for different special part types.

[0033] The recognition performance of the large model is highly dependent on the writing of the prompts. Expert experience can be leveraged to extract key attributes for each specific part type, preparing separate name, size, and weight features for each part type. For example, for the tire part type, the name features would include keywords such as "tire," "tire," and "fetal," the size features would typically be circular with a diameter of 400-800mm and a width of 150-300mm, and the weight would be 5-25kg. Furthermore, the prompts should specify the process for determining the part type category.

[0034] According to one embodiment of the present invention, special part type identification is performed based on preset prompt words to obtain the part type category to which each product belongs. Specifically, this may include: analyzing whether the name of the product contains keywords with name features for name matching; in response to successful name matching, determining a first identification result based on the matched name features, determining whether the size of the product conforms to the size range corresponding to the size feature of the first identification result to obtain a second identification result; determining whether the weight of the product is within the weight range corresponding to the weight feature of the first identification result to obtain a third identification result; and obtaining the part type category to which each product belongs based on the second identification result and the third identification result.

[0035] In practical implementation, the keywords in the product name are first analyzed to determine if they match keywords in the name features of a specific part type. If they do not match, the product is determined not to be a specific part type, and its part type category is the regular part type category (which can be divided according to business needs). If they match, the first identification result (i.e., the part type category of the initially identified specific part type) is determined based on the matched name features, and the product's size and weight are further determined to match the first identification result. Specifically, the second identification result is obtained by determining whether the product's size falls within the size range corresponding to the size features of the first identification result, and the third identification result is obtained by determining whether the product's weight falls within the weight range corresponding to the weight features of the first identification result. Finally, a comprehensive judgment is made based on the second and third identification results to determine the part type category of the most matching specific part type.

[0036] According to one embodiment of the present invention, the component type category of each product is obtained based on the second identification result and the third identification result. Specifically, this may include: in response to the absence of conflict between the second and third identification results, determining the component type category of each product based on the intersection of the second and third identification results; and in response to a conflict between the second and third identification results, determining the component type category of each product based on the matching degree between the product and the second and third identification results. Wherein, when two identification results conflict, the component type category of the specific component type corresponding to the identification result with the higher matching degree can be selected as the component type category of the product.

[0037] Step S103: Based on the average daily outbound volume, average daily inventory volume, best-selling category, and item type of each product, solve the pre-established mixed integer programming model to obtain a warehouse racking selection scheme. The scheme includes the racking type and required quantity allocated to each product. The mixed integer programming model aims to minimize the total operating cost, which includes racking floor space cost, racking depreciation cost, replenishment personnel cost, and forklift depreciation cost. The constraints of the mixed integer programming model include at least: turnover days constraint, storage volume constraint, storage location capacity constraint, size matching constraint, and total demand constraint.

[0038] Step S103 solves the pre-established mixed-integer programming model based on the outputs of the previous two steps S101 and S102 to obtain the shelf types allocated to the goods and the required number of shelf locations. Shelf types include, for example, picking location shelves and replenishment location shelves. Picking location shelves refer to shelves located in the picking area, while replenishment location shelves refer to shelves within the warehouse used for replenishing the picking location shelves.

[0039] According to one embodiment of the present invention, the objective function of the mixed integer programming model is: Min L = warehouse rent cost + rack depreciation cost + replenishment personnel cost + forklift depreciation cost, that is: the sum L of warehouse rent cost, rack depreciation cost, replenishment personnel cost and forklift depreciation cost is minimized. Wherein, warehouse rent cost = unit area warehouse rent cost × (total projected area of ​​all picking racks and replenishment racks); rack depreciation cost = depreciation cost of all picking racks + depreciation cost of all replenishment racks; replenishment personnel cost = total monthly replenishment task volume / (replenishment efficiency × daily working hours × monthly attendance days) × number of days in the month; forklift depreciation cost = number of forklifts required × unit depreciation cost of forklifts.

[0040] According to one embodiment of the present invention, the following constraints are provided: a turnover days constraint, used to associate the picking location turnover days of a product with the preset turnover days of its best-selling category; a storage volume constraint, used to determine the required picking location volume and replenishment location volume for each product based on the average daily outbound volume, average daily inventory volume, and picking location turnover days; a storage location capacity constraint, used to ensure that the total volume of picking locations and replenishment locations allocated to a product is not less than its required picking location volume and replenishment location volume, respectively; a size matching constraint, used to restrict the size of the storage locations on the shelves allocated to the product to be greater than or equal to the size of the product; and a total demand constraint, used to calculate the total number of various shelves required to meet the storage needs of all products.

[0041] In the specific implementation process, the picking location volume is equal to the average daily outbound volume of the goods multiplied by the picking location turnover days of the best-selling category of the goods; the replenishment location volume is at least equal to the average daily inventory volume of the goods minus the picking location volume; based on whether the goods have both picking and replenishment locations, they are determined to be goods that need replenishment, and a replenishment task is generated cumulatively; the replenishment personnel cost and forklift depreciation cost are calculated based on the replenishment task and the preset replenishment efficiency parameters.

[0042] According to one embodiment of the present invention, the best-selling item classification rule includes classifying products into best-selling, ordinary, and slow-moving items based on the average daily number of outbound order lines in the order data. Furthermore, the constraints also include a slow-moving item constraint, which sets the picking location volume and picking location turnover days of slow-moving items to zero, and does not assign picking locations to slow-moving items.

[0043] According to one embodiment of the present invention, the size matching constraint is implemented by pre-generating a set of optional shelves for each product: traversing all optional shelves, adding shelves whose length, width, and height of the storage location are all greater than or equal to the length, width, and height of the product to the set of optional shelves for the product. In the model, the size matching constraint controls the matching relationship between products and storage locations, avoiding storage safety hazards and space waste caused by size mismatch, and ensuring that each product can obtain a suitable storage location while satisfying the size matching constraint.

[0044] The mixed-integer programming model of the present invention is described below with reference to specific embodiments. First, the sets, parameters, and decision variables related to the model are given. Table 1 below shows the set list of the mixed-integer programming model of the present invention; Table 2 below shows the parameter list of the mixed-integer programming model of the present invention; and Table 3 below shows the decision variable list of the mixed-integer programming model of the present invention.

[0045] Table 1

[0046] Table 2

[0047] Table 3

[0048] Based on Tables 1 to 3 above, the objective function of the mixed-integer programming model established in this invention is: Min ; in, It is the cost of the floor space occupied by the shelving, the total projected area of ​​picking location shelving and replenishment location shelving and the unit area warehouse rental cost; It is the depreciation cost of the shelves, including the depreciation cost of the picking location shelves and the depreciation cost of the replenishment location shelves; It's the cost of replenishment personnel. This is the depreciation cost of the forklift.

[0049] The main constraints are as follows:

[0050] Constraint (1) restricts the relationship between the picking location turnover days of a product and the picking location turnover days corresponding to the best-selling category.

[0051]

[0052] Constraint (2) is the picking location volume constraint, where the volume of the picking location is equal to the average daily outbound volume multiplied by the set picking location turnover days.

[0053]

[0054] Constraint (3) is the storage location volume constraint. The volume of the storage location should be greater than or equal to the average daily inventory volume of the goods minus the picking location volume.

[0055]

[0056] Constraint (4) restricts each product to only one picking location. For all products, at most one picking location can be assigned.

[0057]

[0058] Constraint (5) restricts each product to only one storage location. For all products, at most one storage location can be assigned.

[0059]

[0060] Constraint (6) restricts the selected picking location to have a volume greater than the volume of the product picking location.

[0061]

[0062] Constraint (7) restricts the selected storage location volume to be greater than the storage location volume of the goods.

[0063]

[0064] Constraint (8) restricts the correspondence between shelves and goods. When the length, width and height of a goods are not less than the length, width and height of the shelf, that is, when the goods cannot be placed on the shelf, the goods cannot be assigned to that type of shelf.

[0065]

[0066] Constraints (9) and (10) restrict the relationship between the number of merchandise locations and the number of shelves.

[0067]

[0068] Constraints (11) and (12) restrict the number of shelf groups to be greater than or equal to the total demand for all goods.

[0069]

[0070] Constraints (13) and (14) determine whether each item needs replenishment. If both the picking location and the storage location have inventory, then the item needs replenishment. 。

[0071]

[0072] Constraint (15) is used to determine the number of items that need to be replenished per day for each best-selling category. For any best-selling category, the product of the number of items that need to be replenished per day and the picking location turnover days is greater than the number of items that need to be replenished for each best-selling category.

[0073]

[0074] Constraint (16) is used to determine the demand for forklift equipment and the number of forklifts. Replenishment efficiency The average daily effective working hours are greater than the total number of goods that need to be restocked daily for each best-selling category.

[0075]

[0076] Constraint (17) is based on the consideration of safety stock at picking locations, and replenishment capacity for each best-selling item. The turnover days are greater than the total number of items that need to be replenished, and there is sufficient replenishment capacity to ensure that the picking positions are not out of stock.

[0077] Special business constraints: No picking locations are provided for slow-moving products.

[0078] Constraint (18) is a special business constraint. All slow-moving goods are placed in the storage location and no longer have shelves in the picking area.

[0079]

[0080] Constraint (19) is a special business constraint: all slow-moving goods are placed in the inventory location, and the picking location turnover days are forcibly set to 0. Other special business constraints can be easily added based on existing decision variables.

[0081] The final decision variables in the above model are the picking location turnover days for each item and the required picking location shelf type and storage location shelf type. When solving the mixed-integer programming model, for example, in a Python environment, gurobi, cplex, or similar solvers can be used to solve the above model and obtain the final picking location turnover days and specific shelf requirements.

[0082] According to one embodiment of the present invention, before solving the pre-established mixed integer programming model, the method may further include: responding to the component type category to which the product belongs being the component type category corresponding to a special component type, matching the corresponding dedicated shelf type according to the component type category to which the product belongs, and selecting an optional shelf set for the product from the shelves of the dedicated shelf type; wherein, when the component type category to which the product belongs is two covers, the optional shelf set for the product is limited to door racks or dedicated cover racks; when the component type category to which the product belongs is bumpers, the optional shelf set for the product is limited to bumper dedicated shelves; when the component type category to which the product belongs is glass, the optional shelf set for the product is limited to pallet picking locations or dedicated wooden racks. By setting dedicated shelf types for special component types, warehouse utilization can be improved.

[0083] The following example illustrates the application effect of the special-item racking planning method of this invention. Assume a selected automobile warehouse stores approximately 4800 types of goods, with an average daily inventory volume of around 5000 cubic meters and a daily outbound volume of approximately 160 cubic meters. The warehouse contains approximately 200 best-selling items, 2500 ordinary items, and 2100 slow-moving items.

[0084] Table 4 below shows the racking selection results for a certain car warehouse.

[0085] Table 4

[0086] Based on Table 4 above, we analyze the impact of different storage types on total warehouse operating costs. Without separate racks for special-type items, the required total storage space is 8128 cubic meters. With separate racks for special-type items, the required total storage space is 5151 cubic meters. The total floor space occupied by the racks also decreases from 6446 square meters to 5987 square meters, a reduction of 7.1%. The total cost decreases by approximately 10.9%. The results show that identifying and storing special-type items separately can reduce total warehouse operating costs. Table 4 details the demand for various types of racks. According to the results in Table 4, differentiating storage types for special-type items significantly reduces the use of large beam racks. The increase in the number of rack groups for special-type items is less than the decrease in the number of regular rack groups. This indicates that in automotive warehouses, differentiating storage types for special-type items significantly improves storage efficiency. This can be understood as special racks enabling the stacking and storage of irregularly shaped items within the warehouse.

[0087] The following analysis examines the impact of storage location capacity constraints on the results. The consideration of matching storage location size with product dimensions is based on the significant differences in product types within the automotive warehouse. Storage location size should be greater than or equal to the product size to ensure products can be placed in the appropriate location. The following compares the impact of considering size matching constraints on racking selection. Without considering size matching constraints, the total warehouse operating cost is 122,173; with consideration, the total warehouse operating cost is 145,446, an increase of approximately 16%. This reduces 1,685 types of racking that cannot be placed in picking locations and 2,977 types that cannot be placed in storage locations. The cost increase is due to ensuring the matching relationship between products and racking. If product size is not considered, the number of high-beam racking units with lower average cost per square meter would increase significantly, while high-cost racking units such as large pallet racking units would not be selected. Without considering the size matching relationship between the storage location and the product, the proportion of products whose assigned picking and storage locations are smaller than the product size increases significantly. Under each sales volume, a large proportion of products are placed in storage locations that do not match their size. This proves that in a car warehouse with products of varying sizes, the size of the storage location is one of the factors that must be considered during the shelf selection process.

[0088] The following analysis examines the warehousing performance after using a large model for marking special part types. In a data sample with clearly defined special part types, the recognition performance was explored using human experience, the DeepSeek non-deep thinking model, and the DeepSeek deep thinking model. The marking accuracy is shown in Table 5 below. The large model significantly improves the accuracy compared to rule recognition based on human experience. The deep thinking model has the highest accuracy but lower time efficiency. Based on the recognition accuracy results for various special part types, it can be seen that for some common part types, human experience and the large model method can achieve similar accuracy. However, for some special part types, such as "two covers," the rule recognition effect of human experience is very poor, often confusing them with car doors. But thanks to the language understanding of the large model, it can distinguish them better.

[0089] Table 5

[0090] Figure 2 This is a schematic diagram illustrating the planning principle of a special-type shelving unit according to an embodiment of the present invention. Figure 2 This illustrates the implementation logic of applying a special-item racking planning method to the selection scenario of automotive warehouse equipment racking. For example... Figure 2 As shown, in this embodiment, the method for selecting automotive warehouse equipment racks mainly uses a pre-built mixed-integer programming model, combined with input data and constraints, to plan the racks. The input data primarily includes available rack information, product attributes, warehouse order data, inventory data, unit cost data (mainly including the unit costs of equipment, personnel, racks, etc.), and special product types identified through the large model's product attribute analysis. Constraints mainly include turnover days constraints, storage volume constraints, storage location capacity constraints, size matching constraints, total demand constraints, and so on.

[0091] In practical implementation, this mixed-integer programming model aims to minimize the total operating cost, which mainly includes the cost of shelving floor space, shelving depreciation, replenishment personnel costs, and forklift depreciation. Thus, by combining the input data and constraints, the warehouse shelving selection scheme that minimizes the total warehouse operating cost can be obtained by solving this mixed-integer programming model.

[0092] Figure 3 This is a schematic diagram illustrating the planning process of a special-type shelving unit according to an embodiment of the present invention. Figure 3The diagram illustrates the implementation logic of applying a special-part type shelving planning method to the selection of equipment shelving in an automotive warehouse. In an embodiment of the invention, the planning process for this special-part type shelving is as follows: First, raw data is read, including attribute data of goods in the warehouse, order data, inventory data, preset best-selling category rules, optional shelving information, and unit cost data for equipment, personnel, and shelving. Then, the raw data is processed to obtain the average daily outbound volume, average daily inventory volume, and best-selling category for each product. For example, the average daily outbound volume and average daily inventory volume for each product can be calculated based on order data and inventory data, and the best-selling category for each product can be determined based on order data and best-selling category rules, etc. Next, the attribute data of each product is input into a large model to obtain the part type category to which each product belongs. Then, combining the previously obtained average daily outbound volume, average daily inventory volume, and best-selling category for each product, as well as the part type category to which each product belongs, the pre-established mixed integer programming model is solved, and a warehouse shelving selection scheme is generated based on the model's solution.

[0093] Figure 4 This is a schematic diagram of the main modules of a special-type shelving planning device according to an embodiment of the present invention. Figure 4 As shown, the special component type shelf planning device 400 of this embodiment mainly includes a data acquisition and processing module 401, a component type identification module 402, and a selection scheme generation module 403.

[0094] The data acquisition and processing module 401 is used to acquire attribute data, order data and inventory data of goods in the warehouse, calculate the average daily outbound volume and average daily inventory volume of each goods based on the order data and inventory data, and determine the best-selling category of each goods based on the order data and best-selling classification rules. The product type recognition module 402 is used to input the attribute data of each product into a pre-trained large language model to identify special product types and obtain the product type category to which each product belongs. The product type category includes at least one product type category corresponding to the special product type and a regular product type category. The attribute data includes name, size and weight. The selection scheme generation module 403 is used to solve a pre-established mixed integer programming model based on the average daily outbound volume, average daily inventory volume, best-selling category, and product type of each product to obtain a warehouse racking selection scheme. The scheme includes the racking type and required quantity allocated to each product. The mixed integer programming model aims to minimize the total operating cost, which includes racking floor space cost, racking depreciation cost, replenishment personnel cost, and forklift depreciation cost. The constraints of the mixed integer programming model include at least the following: turnover days constraint, storage volume constraint, storage location capacity constraint, size matching constraint, and total demand constraint.

[0095] According to one embodiment of the present invention, the following constraints are provided: a turnover days constraint, used to associate the picking location turnover days of a product with the preset turnover days of its best-selling category; a storage volume constraint, used to determine the required picking location volume and replenishment location volume for each product based on the average daily outbound volume, average daily inventory volume, and picking location turnover days; a storage location capacity constraint, used to ensure that the total volume of picking locations and replenishment locations allocated to a product is not less than its required picking location volume and replenishment location volume, respectively; a size matching constraint, used to restrict the size of the storage locations on the shelves allocated to the product to be greater than or equal to the size of the product; and a total demand constraint, used to calculate the total number of various shelves required to meet the storage needs of all products.

[0096] According to one embodiment of the present invention, the commodity part type recognition module 402 can be specifically used to: input the attribute data of each commodity into a pre-trained large language model, so as to perform special part type recognition according to preset prompt words, and obtain the part type category, recognition judgment basis and confidence level of each commodity; wherein, the prompt words include predefined name features, size features and weight features for different special part types.

[0097] According to one embodiment of the present invention, the commodity part type identification module 402 can be specifically used to: analyze whether the name of the commodity includes keywords of name features for name matching; in response to successful name matching, determine a first identification result based on the matched name features, determine whether the size of the commodity conforms to the size range corresponding to the size feature of the first identification result to obtain a second identification result; determine whether the weight of the commodity is within the weight range corresponding to the weight feature of the first identification result to obtain a third identification result; and obtain the part type category to which each commodity belongs based on the second identification result and the third identification result.

[0098] According to one embodiment of the present invention, the commodity type identification module 402 can be specifically used to: in response to the absence of conflict between the second identification result and the third identification result, determine the type category of each commodity based on the intersection of the second identification result and the third identification result; in response to the presence of conflict between the second identification result and the third identification result, determine the type category of each commodity based on the matching degree between the commodity and the second identification result and the third identification result.

[0099] According to one embodiment of the present invention, the size matching constraint is achieved by pre-generating a set of optional shelves for each product: traversing all optional shelves, adding shelves whose length, width, and height are all greater than or equal to the length, width, and height of the product to the set of optional shelves for the product.

[0100] According to one embodiment of the present invention, the special part type shelf planning device 400 may further include a dedicated shelf matching module (not shown in the figure), used for: before solving the pre-established mixed integer programming model, in response to the part type category to which the product belongs being the part type category corresponding to the special part type, matching the corresponding dedicated shelf type according to the part type category to which the product belongs, and selecting the optional shelf set of the product from the shelves of the dedicated shelf type; wherein, when the part type category to which the product belongs is two covers, the optional shelf set of the product is limited to door racks or dedicated cover racks; when the part type category to which the product belongs is bumpers, the optional shelf set of the product is limited to bumper dedicated shelves; when the part type category to which the product belongs is glass, the optional shelf set of the product is limited to pallet picking positions or dedicated wooden racks.

[0101] According to one embodiment of the present invention, the best-selling product classification rule includes: classifying products into best-selling products, ordinary products, and slow-moving products based on the average daily number of outbound order lines in the order data; and the constraint conditions also include a slow-moving product constraint, which sets the picking location volume and picking location turnover days of slow-moving products to zero, and does not allocate picking locations to slow-moving products.

[0102] According to an embodiment of the present invention, the objective function of the mixed integer programming model is: Min L = warehouse rent cost + rack depreciation cost + replenishment personnel cost + forklift depreciation cost; where, warehouse rent cost = unit area warehouse rent cost × (total projected area of ​​all picking racks and replenishment racks); rack depreciation cost = depreciation cost of all picking racks + depreciation cost of all replenishment racks; replenishment personnel cost = total monthly replenishment task volume / (replenishment efficiency × daily working hours × monthly attendance days) × number of days in the month; forklift depreciation cost = number of forklifts required × unit depreciation cost of forklifts.

[0103] According to the technical solution of this invention, by acquiring attribute data, order data, and inventory data of goods in a warehouse, and calculating the average daily outbound volume and average daily inventory volume of each goods based on the order data and inventory data, and determining the best-selling category of each goods based on the order data and best-selling classification rules, the attribute data of each goods is input into a pre-trained large language model for special part type recognition to obtain the part type category to which each goods belong. Based on the average daily outbound volume, average daily inventory volume, best-selling category, and part type category of each goods, a pre-established mixed integer programming model is solved to obtain the technical solution for warehouse racking selection. This method can intelligently identify special part types of goods such as automotive parts by using a large language model to identify multiple dimensions of attributes such as name, size (length, width, height), and weight. It can accurately identify the geometric shape and physical attributes of special automotive parts, avoiding the limitations of traditional methods that rely on only a single dimension for classification. The multimodal recognition capability of the large model can effectively distinguish parts with similar shapes but different uses, providing a precise data foundation for subsequent specialized racking matching. Simultaneously, the total warehouse operating cost is calculated by combining the costs of equipment, personnel, shelving, and warehouse rent. With the goal of minimizing the total warehouse operating cost, a mixed integer programming model based on special item type identification is established. The model constraints control the correspondence between picking location size and product size. Shelving selection is carried out based on the special item type identification results and picking location turnover days. Under the premise of meeting the requirements of dedicated storage for special item types and safety stock for picking locations of regular item types, the optimal shelving combination is selected from the full range of shelving. This not only realizes differentiated shelving configuration and comprehensive selection, but also fully and comprehensively describes the warehouse storage cost, improving the accuracy and rationality of warehouse shelving selection.

[0104] Figure 5 An exemplary system architecture 500 is shown, which can be applied to a special-item rack planning method or a special-item rack planning device according to embodiments of the present invention.

[0105] like Figure 5 As shown, system architecture 500 may include terminal devices 501, 502, and 503, a network 504, and a server 505. Network 504 serves as the medium for providing communication links between terminal devices 501, 502, and 503 and server 505. Network 504 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0106] Users can use terminal devices 501, 502, and 503 to interact with server 505 via network 504 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 501, 502, and 503, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0107] Terminal devices 501, 502, and 503 can be various electronic devices with displays that support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0108] Server 505 can be a server that provides various services, such as a backend management server that supports the website browsed by the user using terminal devices 501, 502, and 503 (for example only). The backend management server can perform data acquisition and processing on received data such as special item type shelf planning requests, item type identification, mixed integer programming model solving, and warehouse shelf selection scheme generation, and feed back the processing results (such as warehouse shelf selection scheme - for example only) to the terminal devices.

[0109] It should be noted that the special component type shelf planning method provided in the embodiments of the present invention is generally executed by the server 505, and correspondingly, the special component type shelf planning device is generally set in the server 505.

[0110] It should be understood that Figure 5 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0111] The following is for reference. Figure 6 It shows a schematic diagram of the structure of a computer system 600 suitable for implementing terminal devices or servers of the present invention. Figure 6 The terminal device or server shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0112] like Figure 6 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 602 or programs loaded from storage section 608 into random access memory (RAM) 603. The RAM 603 also stores various programs and data required for the operation of the system 600. The CPU 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0113] The following components are connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 610 as needed so that computer programs read from it can be installed into storage section 608 as needed.

[0114] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 609, and / or installed from removable medium 611. When the computer program is executed by central processing unit (CPU) 601, it performs the functions defined above in the system of this invention.

[0115] It should be noted that the computer-readable medium shown in this invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0116] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0117] The units or modules described in the embodiments of the present invention can be implemented in software or hardware. The described units or modules can also be housed in a processor; for example, a processor can be described as including a data acquisition and processing module, a product type identification module, and a selection scheme generation module. The names of these units or modules do not necessarily limit the specific unit or module itself. For example, the selection scheme generation module can also be described as "a module used to solve a pre-established mixed integer programming model based on the average daily outbound volume, average daily inventory volume, best-selling category, and product type category of each product to obtain a warehouse shelf selection scheme."

[0118] In another aspect, the present invention also provides a computer-readable medium, which may be included in the device described in the above embodiments; or it may exist independently and not assembled into the device. The computer-readable medium carries one or more programs, which, when executed by the device, cause the device to: acquire attribute data, order data, and inventory data of goods in a warehouse; calculate the average daily outbound volume and average daily inventory volume of each goods based on the order data and inventory data; determine the best-selling category of each goods based on the order data and best-selling classification rules; input the attribute data of each goods into a pre-trained large language model for special part type recognition, obtaining the part type category to which each goods belong, wherein the part type category includes at least one part type category corresponding to a special part type and a regular part type category; and the attribute data... The system includes name, size, and weight. Based on the average daily outbound volume, average daily inventory volume, best-selling category, and product type of each product, a pre-established mixed integer programming model is solved to obtain a warehouse racking selection scheme. The scheme includes the racking type and required quantity allocated to each product. The mixed integer programming model aims to minimize the total operating cost, which includes racking floor space cost, racking depreciation cost, replenishment personnel cost, and forklift depreciation cost. The constraints of the mixed integer programming model include at least the following: turnover days constraint, storage volume constraint, storage location capacity constraint, size matching constraint, and total demand constraint.

[0119] According to the technical solution of this invention, by acquiring attribute data, order data, and inventory data of goods in a warehouse, and calculating the average daily outbound volume and average daily inventory volume of each goods based on the order data and inventory data, and determining the best-selling category of each goods based on the order data and best-selling classification rules, the attribute data of each goods is input into a pre-trained large language model for special part type recognition to obtain the part type category to which each goods belong. Based on the average daily outbound volume, average daily inventory volume, best-selling category, and part type category of each goods, a pre-established mixed integer programming model is solved to obtain the technical solution for warehouse racking selection. This method can intelligently identify special part types of goods such as automotive parts by using a large language model to identify multiple dimensions of attributes such as name, size (length, width, height), and weight. It can accurately identify the geometric shape and physical attributes of special automotive parts, avoiding the limitations of traditional methods that rely on only a single dimension for classification. The multimodal recognition capability of the large model can effectively distinguish parts with similar shapes but different uses, providing a precise data foundation for subsequent specialized racking matching. Simultaneously, the total warehouse operating cost is calculated by combining the costs of equipment, personnel, shelving, and warehouse rent. With the goal of minimizing the total warehouse operating cost, a mixed integer programming model based on special item type identification is established. The model constraints control the correspondence between picking location size and product size. Shelving selection is carried out based on the special item type identification results and picking location turnover days. Under the premise of meeting the requirements of dedicated storage for special item types and safety stock for picking locations of regular item types, the optimal shelving combination is selected from the full range of shelving. This not only realizes differentiated shelving configuration and comprehensive selection, but also fully and comprehensively describes the warehouse storage cost, improving the accuracy and rationality of warehouse shelving selection.

[0120] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A planning method for special component shelving, characterized in that, include: Obtain attribute data, order data, and inventory data of goods in the warehouse, and calculate the average daily outbound volume and average daily inventory volume of each goods based on the order data and inventory data. Determine the best-selling category of each goods based on the order data and the best-selling classification rules. The attribute data of each product is input into a pre-trained large language model for special part type recognition to obtain the part type category to which each product belongs. The part type category includes at least one part type category corresponding to the special part type and a regular part type category. The attribute data includes name, size and weight. Based on the average daily outbound volume, average daily inventory volume, best-selling category, and product type of each product, a pre-established mixed integer programming model is solved to obtain a warehouse racking selection scheme. The scheme includes the racking type and required quantity allocated to each product. The mixed integer programming model aims to minimize the total operating cost, which includes racking floor space cost, racking depreciation cost, replenishment personnel cost, and forklift depreciation cost. The constraints of the mixed integer programming model include at least: turnover days constraint, storage volume constraint, storage location capacity constraint, size matching constraint, and total demand constraint.

2. The method according to claim 1, characterized in that, The turnover days constraint is used to associate the picking location turnover days of a product with the preset turnover days of its best-selling category. The storage volume constraint is used to determine the required picking location volume and replenishment location volume for each product based on the average daily outbound volume, average daily inventory volume, and picking location turnover days. The storage location capacity constraint is used to ensure that the total volume of the picking location and replenishment location allocated to the goods is not less than the required picking location volume and replenishment location volume, respectively. The size matching constraint is used to restrict the size of the shelf locations allocated to the goods to be greater than or equal to the size of the goods; The total demand constraint is used to calculate the total number of various types of shelves required to meet the storage needs of all goods.

3. The method according to claim 1, characterized in that, The attribute data of each product is input into a pre-trained large language model for special item type recognition, resulting in the item type category to which each product belongs, including: The attribute data of each product is input into a pre-trained large language model to identify special part types based on preset prompt words, thereby obtaining the part type category, identification criteria, and confidence level of each product; wherein, the prompt words include predefined name features, size features, and weight features for different special part types.

4. The method according to claim 3, characterized in that, Based on preset prompts, special item type identification is performed to obtain the item type category to which each product belongs, including: Analyze whether the product name contains keywords of the aforementioned name characteristics for name matching; In response to a successful name match, a first identification result is determined based on the matched name features; a second identification result is obtained by determining whether the size of the product conforms to the size range corresponding to the size features of the first identification result; and a third identification result is obtained by determining whether the weight of the product is within the weight range corresponding to the weight features of the first identification result. Based on the second and third identification results, the item type category of each product is obtained.

5. The method according to claim 4, characterized in that, Based on the second identification result and the third identification result, the item type category to which each product belongs is obtained, including: In response to the absence of conflict between the second identification result and the third identification result, the item type category of each product is determined based on the intersection of the second identification result and the third identification result; In response to a conflict between the second identification result and the third identification result, the item type category of each item is determined based on the matching degree between the item and the second identification result and the third identification result.

6. The method according to claim 1, characterized in that, The size matching constraint is achieved by pre-generating a set of available shelf options for each product: Iterate through all available shelves and add the shelves whose length, width, and height are all greater than or equal to the length, width, and height of the product to the set of available shelves for the product.

7. The method according to claim 6, characterized in that, Before solving the pre-established mixed-integer programming model, the following steps are also included: In response to the fact that the product belongs to a specific product type category, the corresponding special shelf type is matched according to the product's product type category, and the set of optional shelves for the product is selected from the shelves of the special shelf type. When the product belongs to the category of two covers, the optional shelf set for the product is limited to door racks or special cover racks; When the product belongs to the component category of bumper, the set of optional shelves for the product is limited to bumper-specific shelves; When the product belongs to the glass category, the optional shelf set for the product is limited to pallet picking locations or dedicated wooden racks.

8. The method according to claim 1, characterized in that, The best-selling product classification rules include: classifying products into best-selling, average, and slow-moving products based on the average daily number of outbound order lines in the order data; Furthermore, the constraints also include a slow-moving goods constraint, which sets the picking location volume and picking location turnover days of slow-moving goods to zero, and does not assign picking locations to slow-moving goods.

9. The method according to claim 1, characterized in that, The objective function of the mixed-integer programming model is: Min L = Warehouse rent cost + Shelving depreciation cost + Replenishment personnel cost + Forklift depreciation cost; in, Warehouse rental cost = warehouse rental cost per unit area × (total projected area of ​​all picking and replenishment racks). Depreciation cost of shelves = Depreciation cost of all picking location shelves + Depreciation cost of all replenishment location shelves; Replenishment personnel cost = Total monthly replenishment task / (Replenishment efficiency × Daily working hours × Monthly attendance days) × Number of days in the month; Forklift depreciation cost = Number of forklifts required × Unit depreciation cost of forklift.

10. A planning device for a special type of shelving, characterized in that, include: The data acquisition and processing module is used to acquire attribute data, order data and inventory data of goods in the warehouse, and calculate the average daily outbound volume and average daily inventory volume of each goods based on the order data and the inventory data, and determine the best-selling category of each goods based on the order data and the best-selling classification rules. The product part type recognition module is used to input the attribute data of each product into a pre-trained large language model to identify special part types and obtain the part type category to which each product belongs. The part type category includes at least one part type category corresponding to special part types and regular part type categories. The attribute data includes name, size and weight. The selection scheme generation module is used to solve a pre-established mixed integer programming model based on the average daily outbound volume, average daily inventory volume, best-selling category, and component type of each product to obtain a warehouse racking selection scheme. The scheme includes the racking type and required quantity allocated to each product. The mixed integer programming model aims to minimize the total operating cost, which includes racking floor space cost, racking depreciation cost, replenishment personnel cost, and forklift depreciation cost. The constraints of the mixed integer programming model include at least: turnover days constraint, storage volume constraint, storage location capacity constraint, size matching constraint, and total demand constraint.

11. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-9.

12. A computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-9.

13. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-9.

Images