Replenishment amount determination method, replenishment amount determination device, electronic equipment and storage medium

Abstract

The invention provides a replenishment amount determination method, a replenishment amount determination device, electronic equipment and a storage medium. The method is applied to the technical field of intelligent warehousing. The method comprises the steps that under the condition that the current stock of a target commodity is smaller than a preset stock, a first model obtains a first replenishment amount of the target commodity based on the current stock of the target commodity and the daily average sales volume; obtaining the position of the target commodity in a front warehouse, and packaging the current stock, the daily average sales, and the weather data and holiday data corresponding to the position of the front warehouse to obtain commodity parameters; the second model obtains a third replenishment amount and a second replenishment amount based on the commodity parameters and the target large model, and verifies the third replenishment amount based on the first replenishment amount and the second replenishment amount; and determining the third replenishment amount as the replenishment amount of the target commodity under the condition that the verification is passed. The method can improve the accuracy of replenishment quantity prediction.

Description

Technical Field

[0001] This application relates to the field of smart warehousing technology, and more specifically, to a replenishment quantity determination method, a replenishment quantity determination device, an electronic device, and a storage medium. Background Technology

[0002] With the rapid development of the retail industry, the level of intelligence in supply chain management directly impacts operational efficiency and customer satisfaction. The core of supply chain management lies in the accurate forecasting of market demand.

[0003] In related technologies, fixed formulas are typically used to calculate replenishment quantities based on historical product efficiency and inventory levels. However, when faced with unforeseen circumstances (such as extreme weather), the replenishment quantities calculated using these formulas may deviate significantly from actual product demand, making it difficult to meet merchants' actual needs. Therefore, improving the accuracy of replenishment quantity forecasting has become an urgent problem to be solved. Summary of the Invention

[0004] This application provides a replenishment quantity determination method, a replenishment quantity determination device, an electronic device, and a storage medium. The method verifies the third replenishment quantity by using a first replenishment quantity and a second replenishment quantity. This effectively constrains the uncertainty of the target large model, avoids the problem of inaccurate replenishment quantity caused by illusion risk, ensures the accuracy of the replenishment quantity of the target product, and thus improves the user experience for merchants.

[0005] Firstly, this application provides a method for determining replenishment quantity. The method includes: when the current inventory of a target product is less than a preset inventory, a first model obtains a first replenishment quantity for the target product based on the current inventory and average daily sales; obtaining the location of the forward warehouse where the target product is located, and encapsulating the current inventory, average daily sales, and weather and holiday data corresponding to the location of the forward warehouse to obtain product parameters; sending the product parameters and the first replenishment quantity to a second model; the second model inputs the product parameters into a target large model, which obtains a second replenishment quantity for the target product based on the current inventory and average daily sales; obtaining a third replenishment quantity based on the second replenishment quantity, holiday data, and weather data, and verifying the third replenishment quantity based on the first and second replenishment quantities; if the verification passes, determining the third replenishment quantity as the replenishment quantity for the target product.

[0006] The above technical solution involves a first model that uses the current inventory and average daily sales of the target product to obtain a reliable first replenishment quantity as a benchmark. This first replenishment quantity, along with the packaged product parameters, is sent to a second model. The second model then uses the target large model to process the complex product parameters, resulting in a third replenishment quantity adjusted for external factors. Based on this, the third replenishment quantity is validated using both the first and second replenishment quantities. This effectively constrains the uncertainty of the target large model, avoids inaccurate replenishment quantities caused by illusionary risks, ensures the accuracy of the target product's replenishment quantity, and thus improves the merchant's user experience.

[0007] In one possible implementation, the third replenishment quantity is verified based on the first replenishment quantity and the second replenishment quantity, including: if the first replenishment quantity and the second replenishment quantity are the same, it is determined that the verification of the third replenishment quantity has passed; if the first replenishment quantity and the second replenishment quantity are different, it is determined that the verification of the third replenishment quantity has failed.

[0008] The above technical solution determines that the verification of the third replenishment quantity passes when the first and second replenishment quantities are the same, and fails when they differ. By comparing the theoretically identical first and second replenishment quantities, the accuracy of the second replenishment quantity calculated by the target large-scale model is ensured. Given the accuracy of the second replenishment quantity calculated by the target large-scale model, the model adjusts the second replenishment quantity based on weather and holiday data to obtain the third replenishment quantity, thus ensuring the reliability and accuracy of the third replenishment quantity.

[0009] In one possible implementation, the method further includes: obtaining the adjustment quantity of the target product and the corresponding text information based on the target large model and product parameters; wherein the adjustment quantity is the difference between the third replenishment quantity and the second replenishment quantity, and the text information is used to indicate the degree of influence of product parameters on the adjustment quantity.

[0010] The above technical solution, based on the target large model and product parameters, obtains the adjustment quantity of the target product and the corresponding text information. The adjustment quantity is the difference between the third replenishment quantity and the second replenishment quantity, and the text information is used to indicate the degree of influence of product parameters on the adjustment quantity. By performing numerical calculations on the third replenishment quantity and the second replenishment quantity and determining the text information corresponding to the adjustment quantity, the reliability of the target large model output can be ensured through interactive verification of the text information and the adjustment quantity.

[0011] In one possible implementation, the method further includes: in response to a user's operation to view the replenishment quantity of the target product, displaying the replenishment quantity, adjustment quantity, and text information of the target product on the user interface.

[0012] The above technical solution allows users to clearly understand the replenishment and adjustment quantities of target products by viewing their replenishment quantities, and to understand the impact of external factors on the replenishment quantities of target products through text information. This improves the interpretability of the data displayed on the interface, thereby enhancing the user experience.

[0013] One possible implementation involves inputting product parameters into a target large model, including: assembling current inventory, average daily sales, holiday data, and weather data to obtain prompt words; and inputting the prompt words into the target large model.

[0014] The above technical solution assembles current inventory, average daily sales, holiday data, and weather data to obtain prompt words, which are then input into the target large model. Through flexible configuration of prompt words, influencing factors can be added or adjusted as the data changes, meeting the actual needs of merchants.

[0015] In one possible implementation, before obtaining the location of the forward warehouse where the target product is located, the method further includes: obtaining the historical replenishment quantity of the target product, where the historical replenishment quantity is used to indicate the first replenishment quantity of the previous cycle in the current cycle; determining whether the historical replenishment quantity is the same as the first replenishment quantity; and obtaining the location of the forward warehouse where the target product is located if the historical replenishment quantity is different from the first replenishment quantity and the first replenishment quantity is greater than the preset replenishment quantity.

[0016] The above technical solution, when the first replenishment quantity is greater than the preset replenishment quantity and the first replenishment quantity is different from the historical replenishment quantity, obtains the third replenishment quantity based on the target large model and product parameters; by comparing different replenishment quantities, it is determined whether the sales volume of the target product has changed, and then it is determined whether the replenishment quantity of the target product needs to be predicted, so as to reduce the consumption of resources on the target large model while ensuring the accuracy of the replenishment quantity of the target product.

[0017] In one possible implementation, before obtaining the location of the forward warehouse where the target product is located, the method further includes: obtaining the location of the forward warehouse where the target product is located if the target product was not sold during the period from the previous cycle to the current cycle, and the replenishment volume of the target product in the previous cycle is based on the replenishment volume obtained from the second model.

[0018] The above technical solution determines whether the target large model needs to be called by calculating the sales of the target product during the period from the previous cycle to the current cycle and the replenishment quantity of the product in the previous cycle. The target large model is only called when the replenishment quantity needs to be updated, thereby reducing the consumption of resources on the target large model.

[0019] Secondly, this application provides a replenishment quantity determination device, which includes: a processing module, used to: when the current inventory of a target product is less than a preset inventory, a first model obtains a first replenishment quantity of the target product based on the current inventory and average daily sales of the target product; obtains the location of the front warehouse where the target product is located, and encapsulates the current inventory, average daily sales, weather data and holiday data corresponding to the location of the front warehouse to obtain product parameters; and sends the product parameters and the first replenishment quantity to a second model; the processing module is further used to: input the product parameters into a target large model, and the target large model obtains a second replenishment quantity of the target product based on the current inventory and average daily sales; obtain a third replenishment quantity based on the second replenishment quantity, holiday data and weather data, and verify the third replenishment quantity based on the first replenishment quantity and the second replenishment quantity; and, if the verification passes, determine the third replenishment quantity as the replenishment quantity of the target product.

[0020] In one possible implementation, the processing module is used to determine that the verification of the third replenishment quantity has passed when the first replenishment quantity is the same as the second replenishment quantity; and to determine that the verification of the third replenishment quantity has failed when the first replenishment quantity is different from the second replenishment quantity.

[0021] In one possible implementation, the processing module is used to obtain the change in replenishment quantity based on the first replenishment quantity and the third replenishment quantity; when the change in replenishment quantity is less than a preset change, it determines that the verification of the third replenishment quantity has passed; when the change in replenishment quantity is greater than or equal to the preset change, it determines that the verification of the third replenishment quantity has failed.

[0022] In one possible implementation, the processing module is used to obtain the adjustment quantity of the target product and the corresponding text information based on the target large model and product parameters; wherein, the adjustment quantity is the difference between the third replenishment quantity and the second replenishment quantity, and the text information is used to indicate the degree of influence of product parameters on the adjustment quantity.

[0023] In one possible implementation, the processing module is used to respond to the user's operation of viewing the replenishment quantity of the target product, and to display the replenishment quantity, adjustment quantity and text information of the target product on the user interface.

[0024] In one possible implementation, the processing module assembles current inventory, average daily sales, holiday data, and weather data to obtain prompt words; the prompt words are then input into the target large model.

[0025] In one possible implementation, the processing module is used to obtain the historical replenishment quantity of the target product, which is used to indicate the first replenishment quantity of the previous period in the current period; determine whether the historical replenishment quantity is the same as the first replenishment quantity; if the historical replenishment quantity is different from the first replenishment quantity and the first replenishment quantity is greater than the preset replenishment quantity, obtain the location of the front warehouse where the target product is located.

[0026] In one possible implementation, the processing module is used to obtain the location of the front warehouse where the target product is located, provided that the target product was not sold during the period from the previous cycle to the current cycle, and the replenishment quantity of the target product in the previous cycle is based on the replenishment quantity obtained from the second model.

[0027] Thirdly, this application provides an electronic device, including a memory and a processor, wherein the memory is used to store executable program code; and the processor is used to call and run the executable program code from the memory, causing the electronic device to execute the replenishment quantity determination method in the first aspect or any possible implementation of the first aspect.

[0028] Fourthly, a computer-readable storage medium is provided that stores computer program code, which, when executed on a computer, causes the computer to perform the replenishment quantity determination method in the first aspect or any possible implementation thereof.

[0029] Fifthly, a computer program product is provided, comprising: computer program code, which, when run on a computer, causes the computer to execute the replenishment quantity determination method in the first aspect or any possible implementation thereof.

[0030] The above technical solution involves a first model that uses the current inventory and average daily sales of the target product to obtain a reliable first replenishment quantity as a benchmark. This first replenishment quantity, along with the packaged product parameters, is sent to a second model. The second model then uses the target large model to process the complex product parameters, resulting in a third replenishment quantity adjusted for external factors. Based on this, the third replenishment quantity is validated using both the first and second replenishment quantities. This effectively constrains the uncertainty of the target large model, avoids inaccurate replenishment quantities caused by illusionary risks, ensures the accuracy of the target product's replenishment quantity, and thus improves the merchant's user experience.

[0031] The technical solution of this invention can be applied to the transaction and delivery services of instant e-commerce platforms, such as Taobao Flash Sale, Taoxianda, Ele.me takeaway and retail. Attached Figure Description

[0032] Figure 1 This is a schematic diagram illustrating a scenario of a method for determining replenishment quantity provided in an embodiment of this application; Figure 2 This is a schematic flowchart illustrating a method for determining replenishment quantity provided in an embodiment of this application; Figure 3 This is a schematic flowchart illustrating another method for determining replenishment quantity provided in an embodiment of this application; Figure 4 This is a schematic diagram of a replenishment quantity determination system provided in an embodiment of this application; Figure 5 This is a schematic diagram of another replenishment quantity determination device provided in an embodiment of this application; Figure 6 This is a schematic diagram of a replenishment quantity determination device provided in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0033] The technical solutions in this application will be clearly and thoroughly described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. "And / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more than two.

[0034] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0035] Before introducing the solutions of the embodiments of this application, the technical terms that may be involved in the embodiments of this application will be explained.

[0036] Large Language Model (LLM): A deep learning-based artificial intelligence system. It is trained on large-scale text datasets to generate and understand natural language.

[0037] Pre-training: Training a model (e.g., an LLM model) on domain-specific data, enabling the pre-trained model to understand domain knowledge and output corresponding data according to actual needs.

[0038] Replenishment Recommendations: Based on a product's historical sales volume, current inventory, and preset replenishment rules, this feature automatically calculates and generates guidance data on whether a specific product should be replenished and, if so, how much. For example, it provides merchants with recommendations on whether to replenish stock and, if so, how much to replenish, based on given formulas and relevant data.

[0039] Model distillation: This technique involves transferring knowledge from a large model to a smaller model (which typically has more parameters) through knowledge transfer. The aim is to reduce the number of parameters and computational requirements, thereby improving model performance and efficiency. Model distillation enables smaller models to achieve faster inference speeds, lower computational costs, and reproduce performance close to that of larger models, thus addressing issues such as response latency faced by large models in large-scale applications.

[0040] With the rapid development of the retail industry, the level of intelligence in supply chain management directly impacts operational efficiency and customer satisfaction. The core of supply chain management lies in the accurate forecasting of market demand. Typically, a fixed formula is used to calculate replenishment quantities based on historical product efficiency and inventory levels. However, in the event of unforeseen circumstances (such as extreme weather), the replenishment quantities calculated using this formula may deviate significantly from actual product demand, making it difficult to meet the actual needs of merchants.

[0041] In practical applications, replenishment recommendations are supply chain tools provided by Aoxiang SaaS to merchants. Their main function is to calculate whether an item is out of stock and how much should be replenished based on sales volume and parameters configured by the merchant, using a formula provided by the product. Replenishment recommendations rely on fixed formulas to calculate replenishment quantities. During peak sales periods, holidays, or extreme weather events, the calculated quantities often fail to reflect the dynamic influence of real-world factors, making the recommendations merely "for reference only." With the current booming development of LLM (Local Management Model), intelligent replenishment recommendations based on large models with thought chains (such as Deepseek-r1), which analyze weather data (including PM2.5, wind speed, and basic weather information), holiday data, and product categories, will provide merchants with more robust replenishment data. It can be understood that a large model with a thought chain is a large model with a cot (considered factors), capable of analyzing and reasoning about complex, unstructured information (such as weather and holidays) like a human expert.

[0042] Previously, industry-standard replenishment recommendations for retailers, based on sales volume or inventory capacity, had limited reliability. In the event of unforeseen circumstances such as holidays or extreme weather, the recommended replenishment quantities often failed to reflect real-world conditions. With the rapid development of AI technology, using artificial intelligence to address this pain point in replenishment recommendations can make them more reliable, fully considering real-world situations, and empowering merchants to replenish their inventory more effectively and efficiently.

[0043] In view of this, this application provides a method, device, electronic device and storage medium for determining replenishment quantity, which, combined with the text generation capabilities of a large model with cot, implements AI replenishment suggestions based on real-world factors (weather information, holidays, product categories, etc.). Based on the replenishment suggestions, it brings merchants more intelligent, more visual and more human-centered replenishment suggestions.

[0044] The following will combine Figure 1 The application scenarios of the method for determining replenishment quantity are introduced.

[0045] Figure 1 This is a schematic diagram of a scenario for a method for determining replenishment quantity provided in an embodiment of this application.

[0046] For example, Figure 1 The diagram illustrates a graphical user interface (GUI) 100 for a user to view the replenishment quantity of spicy snacks (i.e., the target product). Optionally, the type of user terminal includes, but is not limited to, smartphones, smart tablets, laptops, desktop computers, etc. This application embodiment does not limit the form of the user terminal.

[0047] User interface 100 displays product information, replenishment status, and replenishment suggestions. Product information includes the target product's forward warehouse and product type; users learn about the product's basic information through this information. Replenishment status includes the number of days the product has been out of stock and the stockout loss; when a product is out of stock, the corresponding values ​​for the number of days out of stock and the stockout loss will be displayed, with the stockout loss determined based on the number of days out of stock.

[0048] The replenishment suggestions include AI-suggested quantities, and when a user clicks on the suggested quantity for a product, the replenishment quantity, adjustment quantity, and text information are displayed. Additionally, replenishment parameters are also displayed. Optionally, in response to a user clicking on the AI-suggested quantity for the spicy snack, the user interface 100 displays the replenishment quantity, adjustment quantity, and text information for the spicy snack. The corresponding replenishment quantity for the spicy snack is 32 bags, and the adjustment quantity is 3 bags. The text information includes the reason for the adjustment and the calculation method for the replenishment quantity. It is understood that during the replenishment data calculation process, the replenishment quantity is rounded up to the nearest integer.

[0049] For example, when the current inventory of spicy snacks is less than the preset inventory (or when the available inventory is less than the warning inventory), a basic suggested quantity (or second replenishment quantity) is obtained based on the current inventory of spicy snacks (29) and the average daily sales. An AI-calculated suggested quantity (32) is then obtained based on predictions made by the target large model regarding the current inventory, average daily sales, weather data, and holiday data. If the first and second replenishment quantities calculated by the formula are successfully verified against the third replenishment quantity (i.e., the target large model passes the illusion detection), the AI-calculated suggested quantity 32 is the replenishment quantity for the spicy snacks.

[0050] Optionally, in response to the user's viewing of the AI-suggested quantity of spicy snacks, the user interface 100 displays the AI-calculated suggested quantity (or replenishment quantity of the target product) of spicy snacks: 32 bags, AI adjustment quantity: 3 bags, and text information: Moderate rain and strong winds (with wind speeds up to 48.64 m / s) are expected in the next five days, which may stimulate demand for home consumption. At the same time, the approaching Dragon Boat Festival may bring an increase in demand for snacks before the festival.

[0051] Optionally, the warning inventory is determined by the number of days the goods arrive, the safe inventory days, and the average daily sales. For example, the warning inventory can be determined by multiplying the cumulative number of days the goods arrive and the safe inventory days by the average daily sales.

[0052] Furthermore, when the user stays on the user interface 100 for an extended period, they can update the replenishment suggestion data by clicking the update control. Optionally, in response to the user's update operation on the replenishment quantity of the user interface 100, the replenishment data of the goods displayed on the user interface 100 is updated.

[0053] The above technical solution, combined with the text generation capabilities of large models, has implemented AI replenishment suggestions based on weather information, holidays, and product categories. On top of these suggestions, it provides merchants with more intelligent, more visual, and more human-centered replenishment recommendations.

[0054] Figure 2 This is a schematic flowchart illustrating a method for determining replenishment quantity provided in an embodiment of this application.

[0055] For example, such as Figure 2 As shown, the replenishment quantity determination method 200 includes S210 to S240.

[0056] S210, when the current inventory of the target product is less than the preset inventory, the first model obtains the first replenishment quantity of the target product based on the current inventory and average daily sales of the target product; obtains the location of the front warehouse where the target product is located, encapsulates the current inventory, average daily sales, and weather and holiday data corresponding to the location of the front warehouse to obtain product parameters; and sends the product parameters and the first replenishment quantity to the second model.

[0057] It should be noted that the first model is a model that calculates the replenishment quantity of goods using a fixed formula, and the first model can be a system.

[0058] For example, the target product refers to any retail item; for instance, the target product could be spicy snacks, bottled water, umbrellas, raincoats, etc. The current inventory level of the target product indicates the existing quantity of the product in the forward warehouse where it is located. The preset inventory level of the target product indicates the warning inventory level for the target product; that is, if the current inventory level of the target product is less than the preset inventory level, replenishment needs to be triggered, i.e., confirming whether replenishment of the target product is necessary. The preset inventory level can be pre-configured for the product or determined based on historical sales volume; no specific limitation is made here.

[0059] Optionally, if the current inventory of the target product is less than the preset inventory, the first replenishment quantity of the target product is determined based on the current inventory and average daily sales, i.e., the first replenishment quantity of the target product in the forward warehouse. If the current inventory of the target product is greater than or equal to the preset inventory, the current inventory of the target product continues to be monitored. It is understandable that as the location of the forward warehouse varies, the current inventory and average daily sales of the target product in the forward warehouse will differ, resulting in different replenishment quantities for the target product.

[0060] For example, the first replenishment quantity of the target product is obtained based on its current inventory and average daily sales. Specifically, the cumulative number of days is determined by summing the replenishment cycle of the target product, the number of days it takes to arrive, and the number of days it has a safety stock. The difference between the product of the cumulative number of days and the average daily sales and the available inventory of the target product is then used to determine the first replenishment quantity of the target product.

[0061] Safety days of inventory are a parameter configured by merchants to mitigate uncertainties. Safety days of inventory are the number of days of sales that extra safety stock can sustain, in addition to normal replenishment cycles and delivery days, to cope with risks such as fluctuations in demand (e.g., a sudden increase in sales) or supply delays (e.g., late deliveries from suppliers).

[0062] The initial replenishment quantity for the target product is calculated using a formula. For example, the expression for the initial replenishment quantity can be as follows: ; in, Indicates the first replenishment quantity of the target product; Indicates the replenishment cycle of the target product. Indicates the number of days until the target product arrives. This indicates the number of days the target product is safe to keep in stock. This indicates the average daily sales volume of the target product; This indicates the available inventory of the target product, which includes the current inventory and the inventory in transit. This indicates rounding up. For example, the target product in front warehouse A is snacks. The replenishment cycle for snacks is 7 days, the replenishment period is 5 days, the safety stock days are 3 days, the average daily sales are 3.83 bags, and the available inventory is 29, of which the current inventory is 29 and the in-transit inventory is 0. The result calculated according to the formula is 28.45. Rounding 28.45 up gives the first replenishment quantity of 29.

[0063] In one example, when the current inventory of the target product is less than the preset inventory, the first replenishment quantity of the target product is obtained based on the current inventory and average daily sales, and the second and third replenishment quantities are obtained based on the target product model and product parameters.

[0064] Predicting the replenishment quantity of a target product by calling the target large model requires computational power from the target large model. Therefore, it can be determined whether to call the target large model based on the changes in the first replenishment quantity.

[0065] In another example, the historical replenishment quantity of the target product is obtained, which indicates the first replenishment quantity of the previous period in the current period; it is determined whether the historical replenishment quantity is the same as the first replenishment quantity; if the historical replenishment quantity is different from the first replenishment quantity, and the first replenishment quantity is greater than the preset replenishment quantity, the location of the front warehouse where the target product is located is obtained.

[0066] Optionally, the preset replenishment quantity can be 0, 1, etc. The preset replenishment quantity can be determined according to the actual situation, and no specific limit is made here.

[0067] Historical replenishment quantity indicates the first replenishment quantity of the previous period compared to the current period. The current period and the previous period can be determined based on the data update interval; for example, if the current inventory of the target product in the previous period was less than the preset inventory, the first replenishment quantity of the previous period is calculated based on the current inventory of the target product and the average daily sales. If no replenishment was performed, and the current inventory of the target product in the current period is less than the preset inventory, the first replenishment quantity of the current period is calculated based on the current inventory of the target product and the average daily sales. The current period and the previous period can indicate the current calculation and the last calculation.

[0068] For example, if the initial replenishment quantity for umbrellas calculated in the previous period was 2, and the replenishment quantity calculated in the current period is also 2, this indicates that the umbrella inventory has hardly changed. In this case, there is no need to call the target model to determine the historical replenishment quantity of the target product calculated in the previous period as the replenishment quantity for the target product in the current period. However, if the replenishment quantity calculated in the current period is 5, this indicates a change in the umbrella inventory. Therefore, the target model should be called to predict the replenishment data for umbrellas, and the replenishment quantity should be determined accordingly.

[0069] The above technical solution, when the first replenishment quantity is greater than the preset replenishment quantity and the first replenishment quantity is different from the historical replenishment quantity, obtains the third replenishment quantity based on the target large model and product parameters; by comparing different replenishment quantities, it is determined whether the sales volume of the target product has changed, and then it is determined whether the replenishment quantity of the target product needs to be predicted, so as to reduce the consumption of resources on the target large model while ensuring the accuracy of the replenishment quantity of the target product.

[0070] In another example, if the target product was not sold during the period from the previous cycle to the current cycle, and the replenishment quantity of the target product in the previous cycle was based on the replenishment quantity obtained from the second model, the location of the front warehouse where the target product is located is obtained.

[0071] The above technical solution determines whether the target large model needs to be called by calculating the sales of the target product during the period from the previous cycle to the current cycle and the replenishment quantity of the product in the previous cycle. The target large model is only called when the replenishment quantity needs to be updated, thereby reducing the consumption of resources on the target large model.

[0072] Of course, if the first replenishment quantity is greater than the preset replenishment quantity, the difference between the first replenishment quantity and the current replenishment quantity can be determined. It is then determined whether the replenishment quantity difference is greater than the preset difference; if the difference is greater than the preset difference, a third replenishment quantity is obtained based on the target model and product parameters; if the difference is less than or equal to the preset difference, the current replenishment quantity is determined as the replenishment quantity for the target product. The current replenishment quantity indicates the current replenishment quantity for the target product, that is, the replenishment quantity of the target product that has already been stored.

[0073] To update the replenishment quantity of goods, the current inventory level of goods is periodically checked and the replenishment data is updated accordingly. For example, the current replenishment quantity of raincoats is 3 (already calculated). When verifying the current inventory level of raincoats, if the current inventory level is less than the preset inventory level, the first replenishment quantity of raincoats is obtained based on the current inventory level and average daily sales. If the difference between the first replenishment quantity and the current replenishment quantity is greater than the preset difference, the third replenishment quantity of the target product is obtained based on the target model and product parameters. Furthermore, if the third replenishment quantity passes verification, it is determined as the replenishment quantity of the target product. It can be understood that when predicting the replenishment quantity of the target product again, the replenishment quantity of the target product is the current replenishment quantity.

[0074] If the current inventory of the target product is less than the preset inventory, the first replenishment quantity of the target product is obtained based on the current inventory and average daily sales. The location of the forward warehouse where the target product is located is also obtained. The current inventory, average daily sales, and weather and holiday data corresponding to the location of the forward warehouse are encapsulated to obtain product parameters. The product parameters and the first replenishment quantity are then sent to the second model.

[0075] It is understandable that the weather data for different forward warehouses will vary, so it is necessary to locate the forward warehouse where the target product is located and obtain the weather data.

[0076] For example, data encapsulation is a process of encapsulating data according to a fixed data format. The first model decouples from the second model by sending data to the second model without requiring the second model to perform data calculations.

[0077] It should be noted that the second model is configured with a target large model, and the second model can also be understood as a system.

[0078] S220, the second model inputs the product parameters into the target large model. The target large model obtains the second replenishment quantity of the target product based on the current inventory and average daily sales. Based on the second replenishment quantity, holiday data and weather data, the third replenishment quantity is obtained, and the third replenishment quantity is verified based on the first replenishment quantity and the second replenishment quantity.

[0079] Understandably, the target large model will first calculate the second replenishment quantity of the target product according to a fixed formula. Theoretically, the second replenishment quantity should be equal to the first replenishment quantity. However, due to the possibility of "illusion" in the calculation process of the large model—that is, errors in the calculation data of the large model—the second replenishment quantity may not be equal to the first replenishment quantity. Therefore, numerical verification by comparing the second replenishment quantity with the first replenishment quantity can verify the validity of the third replenishment quantity, i.e., detect the "illusion" of the target large model.

[0080] In practical applications, a post-processing step that does not affect the main process is added to the existing replenishment suggestion system (i.e., calculating replenishment quantity via formula). Specifically, it determines whether the first replenishment quantity in the existing replenishment process meets the AI ​​replenishment suggestion requirements: the replenishment method is sales-driven procurement (in production and operation, sales plans are determined first based on market demand, and then procurement plans are formulated); the first replenishment quantity is greater than 0; and there is a change in the first replenishment quantity compared to the previous first replenishment quantity. If the first replenishment quantity meets the conditions, the replenishment suggestion system will send additional information, including the weather data for the past 15 days (such as PM2.5, weather level, temperature, rainfall, etc.) and holiday data for the past 30 days for the front-end warehouse where the product corresponding to the replenishment suggestion is located. This information is then used by the target large model to predict the third replenishment quantity for the target product.

[0081] The above technical solution inputs product parameters into the target big model. Based on the current inventory and average daily sales, the target big model obtains the second replenishment quantity of the target product, and obtains the third replenishment quantity based on the second replenishment quantity and holiday data. By progressively calculating and analyzing the product parameters, the target big model ensures the interpretability and verifiability of the output data, thus ensuring the accuracy of the target big model output.

[0082] Understandably, the product parameters are assembled during the input of the product parameters into the target large model. The assembly method of the product parameters is flexibly designed based on the prompt words, allowing for the addition or adjustment of influencing factors as the product develops, thus better meeting actual needs.

[0083] For example, current inventory, average daily sales, holiday data, and weather data are assembled to obtain prompt words, which are then input into the target large model.

[0084] By leveraging the capabilities of AI large-scale models (or target large-scale models), corresponding prompts are designed and combined with weather and / or holiday data to provide intelligent replenishment suggestions for goods. Within the AI ​​replenishment suggestion framework, the original main replenishment suggestion chain remains unchanged; instead, a post-processing step is added. The filtered replenishment suggestion data is assembled, incorporating 15 days of weather data and the most recent 30 days of holiday data for the location of the forward warehouse, and then sent via a message chain. On the AI ​​replenishment suggestion chain, the received message is processed, and the AI ​​large-scale model is invoked through the Bailian platform. The returned data undergoes AI illusion processing (i.e., validity checking of the third replenishment quantity), and qualified data is added to the AI ​​replenishment suggestion details table. Subsequently, when merchants initiate replenishment suggestion queries on the page, qualified AI replenishment suggestions are populated into the original replenishment suggestion data, providing merchants with intelligent replenishment advice. By implementing the capabilities of AI large-scale models in real-world application scenarios, the matching accuracy of replenishment suggestions with actual conditions can be improved.

[0085] Assemble current inventory levels, average daily sales, holiday data, and weather data to obtain a prompt. The prompt can be expressed as: You are a supply chain management expert specializing in the retail industry, possessing rich data analysis and forecasting capabilities. You specialize in providing precise replenishment strategies for stores primarily engaged in online sales based on specific formulas and market changes. \n\n"+"##Skills\n"+"1. **Replenishment Quantity Calculation**: Calculate the suggested replenishment quantity for the store based on the formula "Replenishment Cycle + Arrival Days + Inventory Safety Days" * Average Daily Sales - Available Inventory, as input for the next step.\n"+"2. Round up the suggested replenishment quantity calculated in the previous step.\n"+"3. **Market Trend Analysis**: Assess the impact of holidays, seasonal changes, and weather conditions on product sales, and conduct a comprehensive analysis to adjust the basic replenishment quantity.\n"+"4. **Data Sensitivity**: Highly sensitive to key indicators such as store sales data, inventory status, and supplier delivery time to ensure the accuracy and timeliness of replenishment decisions. "+"##Restrictions"+"1. **Focus on Online Channels**: Replenishment quantity calculations are only performed based on sales data generated by online sales platforms such as Ele.me and Meituan. "+"2. **Parameter Dependency**: Replenishment decisions are strictly calculated based on the provided parameters (replenishment cycle, arrival days, safe inventory days, average daily sales, available inventory, etc.), and cannot be changed or assumed independently. "+"3. **Output Standard**: Output replenishment recommendations must follow the preset JSON format, without ```json content, and no additional parameters are allowed: "{\\"baseReplenish\\": base replenishment quantity, \\"otherFactors\\": detailed analysis of the reasons for influencing factors such as weather and holidays, \\"finalReplenish\\": final recommended replenishment quantity}". "+"##Task Instructions\n"+"Considering replenishment parameters (replenishment cycle, arrival days, safe inventory days, average daily sales, available inventory), today's date, nearby holidays & solar terms, "+"weather and temperature trends for the next five days, and the product, please comprehensively consider all factors, including but not limited to holiday sales fluctuations and the impact of weather on consumer behavior, and "+"provide an adjusted replenishment quantity suggestion. The output format is JSON, which includes the basic replenishment quantity, adjustment factor analysis, and the final suggested replenishment quantity. No additional information is required; simply generate the JSON result.

[0086] The above technical solution assembles current inventory, average daily sales, holiday data, and weather data to obtain prompt words, which are then input into the target large model. Through flexible configuration of prompt words, influencing factors can be added or adjusted as the data changes, meeting the actual needs of merchants.

[0087] It is understood that the prompts shown above represent one possible approach in practical applications and do not limit the content of the assembled prompts in actual applications. The specific content of the assembled prompts can be determined based on the actual situation and is not limited here. Furthermore, when the prompts shown above are used as input to the target large model, the target large model also outputs adjustment amounts and related information.

[0088] For example, based on the target large model and product parameters, the adjustment quantity of the target product and the corresponding text information are obtained; the adjustment quantity is the difference between the third replenishment quantity and the second replenishment quantity, and the text information is used to indicate the degree of influence of product parameters on the adjustment quantity.

[0089] The adjustment amount is the difference between the third replenishment quantity and the second replenishment quantity. When the target large model adjusts the second replenishment quantity in a positive direction to obtain the second replenishment quantity, the difference between the third replenishment quantity and the second replenishment quantity is a positive value; when the target large model adjusts the second replenishment quantity in a negative direction to obtain the second replenishment quantity, the difference between the third replenishment quantity and the second replenishment quantity is a negative value.

[0090] The text information corresponding to the adjustment quantity indicates the degree of influence of the product parameters on the adjustment quantity, that is, the reason why the product parameters affect the adjustment quantity. For example, if the target product is zongzi (sticky rice dumplings), the third replenishment quantity is 20 bags, and the second replenishment quantity is 5 bags, the target model will make a positive adjustment to the second replenishment quantity because the sales volume of zongzi will increase as the Dragon Boat Festival approaches. The final adjustment quantity for the target product is 15 bags, and the corresponding text information is: The approaching Dragon Boat Festival may lead to an increase in the demand for zongzi during the festival.

[0091] The above technical solution, based on the target large model and product parameters, obtains the adjustment quantity of the target product and the corresponding text information. The adjustment quantity is the difference between the third replenishment quantity and the second replenishment quantity, and the text information is used to indicate the degree of influence of product parameters on the adjustment quantity. By performing numerical calculations on the third replenishment quantity and the second replenishment quantity and determining the text information corresponding to the adjustment quantity, the reliability of the target large model output can be ensured through interactive verification of the text information and the adjustment quantity.

[0092] For example, the target large model is an existing large model; or, the target large model is a large model obtained by model distillation of an existing large model, that is, the target large model is a large model obtained by model distillation.

[0093] In practical applications, there are already solutions to handle the increase in subsequent actual requests. Currently, the DeepSeek-R1 model is directly invoked. This model has a thought chain and a large number of parameters. Although the calculation is relatively accurate, the return speed is slow, meaning that the output of the large model and the return of replenishment suggestion data are slow. Therefore, by continuously collecting the thought chain and answers returned by the large model during data processing, and forming a dataset with the questions, this dataset is used for subsequent model distillation to obtain smaller models for replenishment prediction in specific vertical domains.

[0094] During model distillation, the capabilities of the dataset and existing large models are leveraged to perform model distillation, resulting in models specific to the vertical domain. This model distillation addresses issues such as slow data return from large models, laying a solid foundation for product development.

[0095] The above technical solution allows for flexible and rapid switching between large models, regardless of whether the target large model is a vertical domain model or a superior large model, by calling the interface, thus improving the convenience and reliability of calling the target large model.

[0096] Furthermore, after determining the second and third replenishment quantities of the target product, the third replenishment quantity is verified using the first and second replenishment quantities to ensure the validity of the third replenishment quantity.

[0097] For example, product parameters are input into the target model. Based on the current inventory level and average daily sales, the target model obtains the second replenishment quantity for the target product. The target model then uses the second replenishment quantity and holiday data to obtain the third replenishment quantity. Furthermore, the third replenishment quantity is validated based on the first and second replenishment quantities.

[0098] The second replenishment quantity is determined by the target model based on the current inventory and average daily sales of the target product. In other words, the method for determining the second replenishment quantity is the same as that for the first. The difference is that the first replenishment quantity is calculated using a traditional formula, while the second replenishment quantity is calculated by the target model using a formula. Understandably, the accuracy of the first replenishment quantity calculated directly using the formula is guaranteed, but the reliability of the second and third replenishment quantities calculated by the target model is difficult to guarantee. Because the target model is prone to data illusions—that is, errors in data substitution or calculation during the calculation process—the data foundation for subsequent analysis may be flawed, making it difficult to guarantee the accuracy of the third replenishment quantity. Validating the third replenishment quantity using the first and second replenishment quantities allows for timely detection of calculation errors in the target model, preventing incorrect replenishment quantities.

[0099] The above technical solution verifies the third replenishment quantity based on the first and second replenishment quantities. Since the second replenishment quantity is calculated using data from the target large model, the accuracy of the second replenishment quantity is ensured by verifying the values ​​of the first and second replenishment quantities. Based on this, the target large model obtains the third replenishment quantity by adjusting the second replenishment quantity, thus ensuring the accuracy of the third replenishment quantity.

[0100] The third replenishment quantity is verified based on the first and second replenishment quantities of the target product. This includes: if the first and second replenishment quantities are the same, the verification of the third replenishment quantity is deemed successful; if the first and second replenishment quantities are different, the verification of the third replenishment quantity is deemed unsuccessful. Since the first and second replenishment quantities should theoretically be the same, comparing the values ​​of the first and second replenishment quantities can avoid the problem of the target large model being misled without the user's knowledge.

[0101] Determine whether the first replenishment quantity and the second replenishment quantity of the target product are the same. If the first replenishment quantity and the second replenishment quantity are the same, the verification of the third replenishment quantity is deemed to have passed; if the first replenishment quantity and the second replenishment quantity are different, the verification of the third replenishment quantity is deemed to have failed.

[0102] The above technical solution determines that the verification of the third replenishment quantity passes when the first and second replenishment quantities are the same, and fails when they differ. By comparing the theoretically identical first and second replenishment quantities, the accuracy of the second replenishment quantity calculated by the target large-scale model is ensured. Given the accuracy of the second replenishment quantity calculated by the target large-scale model, the model adjusts the second replenishment quantity based on weather and holiday data to obtain the third replenishment quantity, thus ensuring the reliability and accuracy of the third replenishment quantity.

[0103] In practical applications, after receiving a message (i.e., the output data of the target large model), the AI ​​replenishment suggestion system assembles the data and template within the message body (ensuring the regularity of the data output) and calls the AI ​​large model text generation interface of the Bailian platform. After the interface returns the data, it first checks whether the calculated base value (i.e., the second replenishment quantity) in the JSON returned by the AI ​​large model is incorrect. That is, it compares the second replenishment quantity calculated by the target large model with the first replenishment quantity calculated by the formula (i.e., illusion detection). If there is an error, the data is discarded and a new prediction is made. If the verification is successful, the AI ​​replenishment suggestion data is stored in the database, that is, the third replenishment quantity and the second replenishment quantity output by the target large model are retained.

[0104] S230, if the verification passes, the third replenishment quantity is determined as the replenishment quantity of the target product.

[0105] For example, if the verification of the third replenishment quantity based on the first replenishment quantity of the target product passes, the third replenishment quantity is determined as the replenishment quantity of the target product.

[0106] For example, in response to a user's action of viewing the replenishment quantity of a target product, the replenishment quantity, adjustment quantity, and text information of the target product are displayed on the user interface. For instance, such as... Figure 1 The content displayed in the user interface shown.

[0107] The above technical solution allows users to clearly understand the replenishment and adjustment quantities of target products by viewing their replenishment quantities, and to understand the impact of external factors on the replenishment quantities of target products through text information. This improves the interpretability of the data displayed on the interface, thereby enhancing the user experience.

[0108] In practical applications, after the AI ​​replenishment suggestion calculation and data storage are completed, the reliable AI replenishment suggestion results are structured and stored in a dedicated table in the database. This provides a data foundation for subsequent queries, display, and analysis. When merchants initiate a viewing operation on the replenishment suggestion interface, if a corresponding AI replenishment suggestion is available, the relevant information will be populated, revealing the AI ​​suggestion quantity and the reasons for its impact to the user on the page; date, holidays, weather, product attributes, etc. Replenishment suggestions include the replenishment quantity of the target product, the second replenishment quantity, the adjustment quantity, and text information, etc. Furthermore, based on the intelligent replenishment suggestion capability of the AI ​​large model, the replenishment suggestion recommendation value and recommendation reasons are significantly better than other replenishment suggestion systems.

[0109] The above technical solution involves a first model that uses the current inventory and average daily sales of the target product to obtain a reliable first replenishment quantity as a benchmark. This first replenishment quantity, along with the packaged product parameters, is sent to a second model. The second model then uses the target large model to process the complex product parameters, resulting in a third replenishment quantity adjusted for external factors. Based on this, the third replenishment quantity is validated using both the first and second replenishment quantities. This effectively constrains the uncertainty of the target large model, avoids inaccurate replenishment quantities caused by illusionary risks, ensures the accuracy of the target product's replenishment quantity, and thus improves the merchant's user experience.

[0110] Figure 3 This is a schematic flowchart illustrating another method for determining replenishment quantity provided in the embodiments of this application.

[0111] For example, such as Figure 3 As shown, the replenishment quantity determination method 300 includes S310 to S370.

[0112] S310: If the current inventory of the target product is less than the preset inventory, the first replenishment quantity of the target product is obtained based on the current inventory and average daily sales of the target product.

[0113] For example, the target product refers to any retail item; for instance, the target product could be spicy snacks, bottled water, umbrellas, raincoats, etc. The current inventory of the target product indicates the quantity of the product already in the forward warehouse where the target product is located.

[0114] Optionally, if the current inventory of the target product is less than the preset inventory, the first replenishment quantity of the target product is determined based on the current inventory and average daily sales, i.e., the first replenishment quantity of the target product in the forward warehouse. Specifically, the sum of the replenishment cycle of the target product, the arrival days of the target product, and the safe inventory days is determined as the cumulative number of days. The difference between the product of the cumulative number of days and the average daily sales and the available inventory of the target product is determined as the first replenishment quantity of the target product.

[0115] If the current inventory of the target product is greater than or equal to the preset inventory level, continue monitoring the current inventory of the target product. It's understandable that the current inventory and average daily sales of the target product will differ depending on the location of the forward warehouse, thus requiring different replenishment quantities for the target product.

[0116] S320, when the first replenishment quantity is greater than the preset replenishment quantity and the first replenishment quantity is different from the historical replenishment quantity, assembles the current inventory, average daily sales, holiday data and weather data to obtain prompt words.

[0117] For example, if the first replenishment quantity is greater than the preset replenishment quantity and differs from the historical replenishment quantity, a third replenishment quantity is obtained based on the target large model and product parameters. The historical replenishment quantity indicates the first replenishment quantity of the previous period in the current period. Specifically, if the first replenishment quantity is greater than the preset replenishment quantity and differs from the historical replenishment quantity, the current inventory, average daily sales, holiday data, and weather data are assembled to obtain a prompt word. This prompt word is then input into the target large model to output the third replenishment quantity and other relevant data.

[0118] Optionally, the preset replenishment quantity can be 0, 1, etc. The preset replenishment quantity can be determined according to the actual situation, and no specific limit is made here.

[0119] Historical replenishment quantity indicates the first replenishment quantity of the previous period compared to the current period. The current period and the previous period can be determined based on the data update interval; for example, if the current inventory of the target product in the previous period was less than the preset inventory, the first replenishment quantity of the previous period is calculated based on the current inventory of the target product and the average daily sales. If no replenishment was performed, and the current inventory of the target product in the current period is less than the preset inventory, the first replenishment quantity of the current period is calculated based on the current inventory of the target product and the average daily sales. The current period and the previous period can indicate the current calculation and the last calculation.

[0120] S330: Input the prompt words into the target model. Based on the current inventory and average daily sales, the target model obtains the second replenishment quantity of the target product. Based on the second replenishment quantity, holiday data, and weather data, it obtains the third replenishment quantity, adjustment quantity, and the corresponding text information of the adjustment quantity.

[0121] For example, the prompt words are input into the target big model. Based on the current inventory and average daily sales, the target big model obtains the second replenishment quantity of the target product. Then, based on the second replenishment quantity, holiday data, and weather data, the target big model obtains the third replenishment quantity, the adjustment quantity, and the corresponding text information. The adjustment quantity is the difference between the third replenishment quantity and the second replenishment quantity. The text information is used to indicate the degree of influence of product parameters on the adjustment quantity.

[0122] The adjustment amount is the difference between the third replenishment quantity and the second replenishment quantity. When the target large model adjusts the second replenishment quantity in a positive direction to obtain the second replenishment quantity, the difference between the third replenishment quantity and the second replenishment quantity is a positive value; when the target large model adjusts the second replenishment quantity in a negative direction to obtain the second replenishment quantity, the difference between the third replenishment quantity and the second replenishment quantity is a negative value.

[0123] The text information corresponding to the adjustment quantity indicates the degree of influence of the product parameters on the adjustment quantity, that is, the reason why the product parameters affect the adjustment quantity. For example, if the target product is zongzi (sticky rice dumplings), the third replenishment quantity is 20 bags, and the second replenishment quantity is 5 bags, the target model will make a positive adjustment to the second replenishment quantity because the sales volume of zongzi will increase as the Dragon Boat Festival approaches. The final adjustment quantity for the target product is 15 bags, and the corresponding text information is: The approaching Dragon Boat Festival may lead to an increase in the demand for zongzi during the festival.

[0124] S340, determine whether the third replenishment quantity is the same as the second replenishment quantity; if not, proceed to S350; if yes, proceed to S360.

[0125] For example, the second replenishment quantity is determined by the target model based on the current inventory and average daily sales of the target product. That is, the method for determining the second replenishment quantity is the same as that for the first replenishment quantity. The difference is that the first replenishment quantity is calculated using a traditional formula, while the second replenishment quantity is calculated by the target model using a formula. Understandably, the accuracy of the first replenishment quantity calculated directly using the formula is guaranteed, but the reliability of the second and third replenishment quantities calculated by the target model is difficult to guarantee. Because the target model is prone to data illusions—that is, errors in data substitution or calculation during the calculation process—the accuracy of the third replenishment quantity is difficult to guarantee due to incorrect data foundation for subsequent analysis. Verifying the third replenishment quantity using the first and second replenishment quantities allows for timely detection of calculation errors in the target model, preventing incorrect replenishment quantities.

[0126] S350, it has been determined that the verification of the third replenishment quantity failed.

[0127] For example, if the first replenishment quantity differs from the second replenishment quantity, it is determined that the verification of the third replenishment quantity has failed. Further, if the verification of the third replenishment quantity fails, the third replenishment quantity will be recalculated and verified using the current inventory level of the target product and the average daily sales volume.

[0128] S360, confirm that the verification of the third replenishment quantity has passed, and determine the third replenishment quantity as the replenishment quantity of the target product.

[0129] For example, if the first replenishment quantity is the same as the second replenishment quantity, the verification of the third replenishment quantity is confirmed to be successful. Since the first and second replenishment quantities should theoretically be the same, comparing the first and second replenishment quantities numerically can avoid the problem of the target large model being unknowingly delusional. Furthermore, if the verification of the third replenishment quantity is successful, the third replenishment quantity is determined as the replenishment quantity of the target product.

[0130] S370, in response to the user's operation of viewing the replenishment quantity of the target product, displays the replenishment quantity, adjustment quantity and text information of the target product on the user interface.

[0131] For example, in response to a user's action of viewing the replenishment quantity of a target product, the replenishment quantity, adjustment quantity, and text information of the target product are displayed on the user interface. For instance, such as... Figure 1 The content displayed in the user interface shown.

[0132] The aforementioned technical solution, based on the current inventory and average daily sales of the target product, obtains the first replenishment quantity. Based on the target product's large-scale model and product parameters, it obtains the third replenishment quantity. The third replenishment quantity is then validated using the first replenishment quantity. If the validation passes, the third replenishment quantity is determined as the target product's replenishment quantity. By combining the target product's current inventory and average daily sales to obtain a reliable first replenishment quantity as a benchmark, and utilizing the target product's large-scale model to process complex product parameters, the third replenishment quantity is obtained after adjustments for external factors. On this basis, validating the third replenishment quantity using the first replenishment quantity effectively constrains the uncertainty of the target product's large-scale model, avoiding inaccurate replenishment quantities caused by illusionary risks, ensuring the accuracy of the target product's replenishment quantity, and thus improving the merchant's user experience. Furthermore, the introduction of large-scale model capabilities into the replenishment data prediction of the target product, by assembling weather and holiday factors into the system, intelligently introduces large-scale model capabilities into the replenishment recommendations for retailers, making the replenishment recommendations more professional and intelligent.

[0133] The replenishment quantity determination system will be introduced below.

[0134] Figure 4 This is a schematic diagram of a replenishment quantity determination system provided in an embodiment of this application.

[0135] For example, such as Figure 4 As shown, the replenishment quantity determination system 400 includes a first model 410 and a second model 420.

[0136] The first model 410 is used to obtain a first replenishment quantity of the target product based on its current inventory and average daily sales when the current inventory of the target product is less than a preset inventory. It then obtains the location of the forward warehouse where the target product is located, encapsulates the current inventory, average daily sales, and weather and holiday data corresponding to the location of the forward warehouse to obtain product parameters, and sends the product parameters and the first replenishment quantity to the second model. The second model 420 is used to input the product parameters into a target large model, which obtains a second replenishment quantity of the target product based on the current inventory and average daily sales. Based on the second replenishment quantity, the holiday data, and the weather data, it obtains a third replenishment quantity and verifies the third replenishment quantity based on the first and second replenishment quantities. If the verification passes, the second model 420 further determines the third replenishment quantity as the replenishment quantity of the target product.

[0137] In one possible implementation, the second model 420 is used to determine that the verification of the third replenishment quantity has passed when the first replenishment quantity is the same as the second replenishment quantity; and to determine that the verification of the third replenishment quantity has failed when the first replenishment quantity is different from the second replenishment quantity.

[0138] In one possible implementation, the second model 420 is also used to obtain the adjustment quantity of the target product and the corresponding text information based on the target large model and product parameters; wherein, the adjustment quantity is the difference between the third replenishment quantity and the second replenishment quantity, and the text information is used to indicate the degree of influence of product parameters on the adjustment quantity.

[0139] In one possible implementation, the second model 420 is also used to assemble current inventory, average daily sales, holiday data and weather data to obtain prompt words; the prompt words are then input into the target large model.

[0140] By adding post-processing that does not affect the main chain to the existing replenishment suggestion system link (i.e., the first model), the matching between replenishment suggestions / replenishment quantities and the actual situation is improved, thereby increasing the accuracy of replenishment suggestions / replenishment quantities. Specifically, the first replenishment quantity is calculated through the first model (i.e., the existing replenishment suggestion system link), and it is determined whether the first replenishment quantity meets the AI ​​replenishment suggestion screening conditions (i.e., the target large model is called to determine the replenishment quantity). If the first replenishment quantity meets the AI ​​replenishment suggestion screening conditions, the target large model is called to predict the third replenishment quantity and the second replenishment quantity. The third replenishment quantity is verified by the first and second replenishment quantities to achieve illusion detection of the target large model. If the verification of the third replenishment quantity by the first and second replenishment quantities passes, that is, the illusion detection of the target large model passes, then the third replenishment quantity output by the target large model is accurate.

[0141] The above technical solution, by adding post-processing to the existing system links, introduces the intelligent analysis and decision-making capabilities of the target large model without affecting the stability and reliability of the original system's business logic. This achieves an upgrade from automation to intelligence in replenishment quantity generation at a relatively low cost. Furthermore, by decoupling the existing system links from the AI ​​replenishment system, it ensures that the two do not interfere with each other, improving the stability of the replenishment quantity determination system.

[0142] Figure 5 This is a schematic diagram of another replenishment quantity determination system provided in the embodiments of this application.

[0143] For example, such as Figure 5 As shown, the replenishment quantity determination system 500 includes a first model 510 and a second model 520. The first model 510 and... Figure 4 The first model 410 is the same model, and the second model 520 is the same as... Figure 4 The second model 420 is the same model.

[0144] The first model 510 includes a replenishment suggestion module 511, a judgment module 512, a data acquisition module 513, and a communication module 514. The replenishment suggestion module 511 is an existing module in the system chain, used to determine the first replenishment quantity of the target product based on the current inventory and average daily sales. The judgment module 512 judges whether the first replenishment quantity meets the AI ​​replenishment suggestion screening conditions, i.e., whether the first replenishment quantity is greater than 0 and whether there is a change compared to the first replenishment quantity of the previous period. If the first replenishment quantity does not meet the AI ​​replenishment suggestion screening conditions, the current replenishment quantity determination process is terminated, and the replenishment quantity of the target product in the previous period is used as the final replenishment quantity.

[0145] If the first replenishment quantity meets the AI ​​replenishment suggestion screening conditions, the data acquisition module 513 acquires the weather data and holiday data of the front warehouse where the target product is located (different front warehouses have different corresponding weather data, so it is necessary to locate the front warehouse where the target product is located), and sends it to the first processing module 521 in the second model 520 through the communication module 514.

[0146] The second model 520 includes a first processing module 521, a large model module 522, a detection module 423, and a second processing module 524. When the first processing module 521 receives data from the communication module 514, it first determines whether the current target large model has completed processing the data from the previous period. If not, it discards the current data or waits for the target large model to complete processing before forwarding the data to the large model module 522. Upon receiving the data, the large model module 522 assembles the data based on the prompt word template, the current inventory of the target product, average daily sales, weather data, and holiday data to generate prompt words. The prompt words are then input into the target large model, and the module outputs the third replenishment quantity, the second replenishment quantity, the adjustment quantity, and text information. The adjustment quantity is the difference between the third and second replenishment quantities, and the text information explains the impact of weather data and / or holiday data on the adjustment quantity.

[0147] Furthermore, the detection module 523 performs illusion detection on the target large model based on the second replenishment quantity and the first replenishment quantity to verify the validity of the third replenishment quantity. If the illusion detection of the target large model fails, i.e., the verification of the third replenishment quantity fails, the current product replenishment quantity determination process is terminated. If the illusion detection of the target large model passes, i.e., the verification of the third replenishment quantity passes, the second processing module 524 saves the third replenishment quantity, adjustment quantity, and text information to the replenishment suggestion library.

[0148] Optionally, in response to a user's action of viewing the replenishment quantity of a target product, the replenishment quantity, adjustment quantity, and text information of the target product are displayed on the user interface.

[0149] It should be understood that the above examples are provided to help those skilled in the art understand the embodiments of this application, and are not intended to limit the embodiments of this application to the specific values ​​or scenarios illustrated. Those skilled in the art can obviously make various equivalent modifications or changes based on the above examples, and such modifications or changes also fall within the scope of the embodiments of this application.

[0150] The above text combined Figures 1 to 5 The method for determining replenishment quantity provided in the embodiments of this application is described in detail below; the following will be combined with Figure 6 and Figure 7 This application provides a detailed description of embodiments of the replenishment quantity determination device. It should be understood that the replenishment quantity determination device in this application can execute various methods described in the foregoing embodiments of this application; that is, the specific working processes of the various products described below can be referred to the corresponding processes in the foregoing method embodiments.

[0151] Figure 6 This is a schematic diagram of a replenishment quantity determination device provided in an embodiment of this application.

[0152] For example, such as Figure 6 As shown, the replenishment quantity determination device 600 includes: The processing module 610 is used to: Firstly, when the current inventory of the target product is less than the preset inventory, obtain the first replenishment quantity of the target product based on the current inventory and average daily sales; obtain the location of the forward warehouse where the target product is located, encapsulate the current inventory, average daily sales, and weather and holiday data corresponding to the location of the forward warehouse to obtain product parameters; send the product parameters and the first replenishment quantity to the second model; The processing module 610 is also used to: The second model inputs the product parameters into the target large model, which obtains the second replenishment quantity of the target product based on the current inventory and average daily sales; Based on the second replenishment quantity, holiday data, and weather data, obtain the third replenishment quantity, and verify the third replenishment quantity based on the first and second replenishment quantities; If the verification passes, the processing module 610 determines the third replenishment quantity as the replenishment quantity of the target product.

[0153] In one possible implementation, the processing module 610 is used to determine that the verification of the third replenishment quantity has passed when the first replenishment quantity is the same as the second replenishment quantity; and to determine that the verification of the third replenishment quantity has failed when the first replenishment quantity is different from the second replenishment quantity.

[0154] In one possible implementation, the processing module 610 is used to obtain the adjustment quantity of the target product and the corresponding text information based on the target large model and product parameters; wherein, the adjustment quantity is the difference between the third replenishment quantity and the second replenishment quantity, and the text information is used to indicate the degree of influence of product parameters on the adjustment quantity.

[0155] In one possible implementation, the processing module 610 is used to respond to the user's operation of viewing the replenishment quantity of the target product and display the replenishment quantity, adjustment quantity and text information of the target product on the user interface.

[0156] In one possible implementation, the processing module 610 is used to assemble the current inventory, average daily sales, holiday data and weather data to obtain prompt words; and input the prompt words into the target large model.

[0157] In one possible implementation, the processing module 610 is used to obtain a third replenishment quantity based on the target large model and product parameters when the first replenishment quantity is greater than the preset replenishment quantity and the first replenishment quantity is different from the historical replenishment quantity. The historical replenishment quantity indicates the first replenishment quantity of the previous period in the current period.

[0158] In one possible implementation, the processing module 610 is used to obtain the location of the front warehouse where the target product is located, provided that the target product has not been sold during the period from the previous cycle to the current cycle, and the replenishment quantity of the target product in the previous cycle is based on the replenishment quantity obtained from the second model.

[0159] It should be noted that the aforementioned replenishment quantity determination device 600 is embodied in the form of a functional unit. The term "module" here can be implemented in software and / or hardware, without specific limitations.

[0160] For example, a "module" can be a software program, a hardware circuit, or a combination of both that implements the above functions. The hardware circuit may include an application-specific integrated circuit (ASIC), electronic circuits, a processor (e.g., a shared processor, a proprietary processor, or a group processor) and memory for executing one or more software or firmware programs, integrated logic circuits, and / or other suitable components that support the described functions.

[0161] Therefore, the units of the various examples described in the embodiments of this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0162] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0163] For example, such as Figure 7As shown, the electronic device 700 includes a memory 710 and a processor 720, wherein the memory 710 stores executable program code 730, and the processor 720 is used to call and execute the executable program code 730 to perform a replenishment quantity determination method.

[0164] For example, the memory 710 can be used to store the relevant program of the replenishment quantity determination method provided in the embodiments of this application; the processor 720 can call the relevant program of the replenishment quantity determination method stored in the memory 710 to execute the replenishment quantity determination method of the embodiments of this application; for example, when the current inventory of the target product is less than the preset inventory, the first model obtains the first replenishment quantity of the target product based on the current inventory and average daily sales of the target product; obtains the location of the front warehouse where the target product is located, encapsulates the current inventory, average daily sales, weather data and holiday data corresponding to the location of the front warehouse to obtain product parameters; sends the product parameters and the first replenishment quantity to the second model; the second model inputs the product parameters to the target large model, and the target large model obtains the second replenishment quantity of the target product based on the current inventory and average daily sales; obtains the third replenishment quantity based on the second replenishment quantity, holiday data and weather data, and verifies the third replenishment quantity based on the first replenishment quantity and the second replenishment quantity; if the verification passes, the third replenishment quantity is determined as the replenishment quantity of the target product.

[0165] This embodiment can divide functional modules according to the above method example. For example, each module can correspond to a separate functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0166] It should be noted that all relevant content of each step involved in the above method embodiments can be referenced from the functional description of the corresponding functional module, and will not be repeated here.

[0167] When using integrated units, the device may include a processing module and a storage module. The processing module may be a processor or a controller that can implement or execute various exemplary logic blocks, modules, and circuits shown in conjunction with the disclosure of this application. The processor may also be a combination of functions that implement computing capabilities, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc., and the storage module may be a memory.

[0168] In addition, the apparatus provided in the embodiments of this application may specifically be a chip, component or module. The chip may include a connected processor and a memory. The memory is used to store instructions. When the processor calls and executes the instructions, the chip can execute a replenishment quantity determination method provided in the above embodiments.

[0169] This application also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement the replenishment quantity determination method provided in the above embodiments. The computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, Digital Video Discs (DVDs), Compact Disc Read-Only Memory (CD-ROMs), microdrives, and magneto-optical disks, read-only memory (ROMs), random access memory (RAMs), erasable programmable read-only memory (EPROMs), electrically erasable programmable read-only memory (EEPROMs), dynamic random access memory (DRAMs), video random access memory (VRAMs), flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), or any type of media or device suitable for storing instructions and / or data.

[0170] This application also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned related steps to implement the replenishment quantity determination method provided in the above embodiments.

[0171] The computer-readable storage medium, computer program product, or chip provided in this application are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0172] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0173] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0174] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for determining replenishment quantity, characterized in that, The method includes: If the current inventory of the target product is less than the preset inventory, the first model obtains the first replenishment quantity of the target product based on the current inventory and average daily sales; obtains the location of the front warehouse where the target product is located; encapsulates the current inventory, the average daily sales, and the weather and holiday data corresponding to the location of the front warehouse to obtain product parameters; and sends the product parameters and the first replenishment quantity to the second model. The second model inputs the product parameters into the target large model. The target large model obtains the second replenishment quantity of the target product based on the current inventory and the average daily sales. Based on the second replenishment quantity, the holiday data, and the weather data, the third replenishment quantity is obtained, and the third replenishment quantity is verified based on the first replenishment quantity and the second replenishment quantity. If the verification passes, the third replenishment quantity is determined as the replenishment quantity for the target product.

2. The method according to claim 1, characterized in that, The verification of the third replenishment quantity based on the first replenishment quantity and the second replenishment quantity includes: If the first replenishment quantity is the same as the second replenishment quantity, then the verification of the third replenishment quantity is deemed successful. If the first replenishment quantity is different from the second replenishment quantity, it is determined that the verification of the third replenishment quantity has failed.

3. The method according to claim 2, characterized in that, The method further includes: Based on the target large model and the product parameters, the adjustment amount of the target product and the text information corresponding to the adjustment amount are obtained; The adjustment amount is the difference between the third replenishment amount and the second replenishment amount, and the text information is used to indicate the degree of influence of the product parameters on the adjustment amount.

4. The method according to claim 3, characterized in that, The method further includes: In response to a user's action of viewing the replenishment quantity of the target product, the replenishment quantity of the target product, the adjustment quantity, and the text information are displayed on the user interface.

5. The method according to claim 1, characterized in that, The second model inputs the product parameters into the target large model, including: The current inventory, average daily sales, holiday data, and weather data are combined to obtain prompt words; Input the prompt words into the target large model.

6. The method according to any one of claims 1 to 5, characterized in that, Before obtaining the location of the forward warehouse where the target product is located, the method further includes: Obtain the historical replenishment quantity of the target product, wherein the historical replenishment quantity is used to indicate the first replenishment quantity of the previous period in the current period; Determine whether the historical replenishment quantity is the same as the first replenishment quantity; If the historical replenishment quantity is different from the first replenishment quantity, and the first replenishment quantity is greater than the preset replenishment quantity, the location of the forward warehouse where the target product is located is obtained.

7. The method according to any one of claims 1 to 5, characterized in that, Before obtaining the location of the forward warehouse where the target product is located, the method further includes: If the target product was not sold during the period from the previous cycle to the current cycle, and the replenishment quantity of the target product in the previous cycle was based on the replenishment quantity obtained from the second model, then the location of the forward warehouse where the target product is located is obtained.

8. A replenishment quantity determination device, characterized in that, The device includes: The processing module is used to: when the current inventory of the target product is less than the preset inventory, the first model obtains the first replenishment quantity of the target product based on the current inventory and average daily sales of the target product; obtain the location of the front warehouse where the target product is located; encapsulate the current inventory, the average daily sales, and the weather and holiday data corresponding to the location of the front warehouse to obtain product parameters; and send the product parameters and the first replenishment quantity to the second model. The processing module is further configured to input the product parameters into the target large model in the second model, and the target large model obtains the second replenishment quantity of the target product based on the current inventory and the average daily sales; obtains the third replenishment quantity based on the second replenishment quantity, the holiday data and the weather data, and verifies the third replenishment quantity based on the first replenishment quantity and the second replenishment quantity; The processing module is further configured to determine the third replenishment quantity as the replenishment quantity of the target product if the verification passes.

9. An electronic device, characterized in that, include: Memory, used to store executable program code; A processor is configured to call and run the executable program code from the memory, causing the electronic device to perform the replenishment quantity determination method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, implements the replenishment quantity determination method as described in any one of claims 1 to 7.

Images