Intelligent optimization method and system for store inventory

Abstract

The invention belongs to the technical field of retail supply chains, and particularly relates to an intelligent optimization method and system for store inventory, and the method comprises the steps: the standard daily average sales volume, the standard demand fluctuation rate, the replenishment time and the customer satisfaction rate of a target commodity are included; calculating a predicted corrected sales volume; obtaining a target inventory direction; calculating a dynamic customer satisfaction rate according to a boundary constraint relationship between the minimum customer satisfaction rate and the maximum customer satisfaction rate in the customer satisfaction rates and the target inventory direction and the customer satisfaction rates; obtaining a dynamic safety inventory; obtaining a dynamic ordering point; based on the predicted corrected sales volume of the target commodity and the dynamic ordering point, combining the adjustment effect of the sales volume correction factor on the dynamic safe inventory, adding the adjustment effect with the total predicted corrected sales volume in the replenishment auditing period, and calculating a target inventory; calculating the number of replenishment orders; and generating a replenishment order number of each commodity and starting a purchase allocation process. The method solves the problem that prediction and decision are separated and the future situation cannot be responded in the prior art.

Description

Technical Field

[0001] This invention relates to the field of retail supply chain technology. More specifically, this invention relates to a method and system for intelligent optimization of store inventory. Background Technology

[0002] Store inventory optimization aims to balance customer satisfaction and business costs through dynamic management of merchandise inventory. Traditional methods often rely on moving averages or historical averages for forecasting, but this approach assumes that the future is a simple repetition of the past and cannot cope with the trends and seasonal variations that are common in merchandise sales. For example, using the annual average for stocking merchandise with obvious seasonality can easily lead to stockouts during peak seasons and overstocking during off-seasons, resulting in serious sales losses and wasted funds.

[0003] To address the aforementioned issues, related technologies disclose time series decomposition methods, such as using a seasonal-trend decomposition method to process historical sales data. This method can decompose complex sales series into long-term trend terms, seasonal cycle terms, and residual terms. By predicting and superimposing the trend and seasonality, a benchmark sales forecast that eliminates random noise and better reflects the inherent sales patterns of goods can be obtained.

[0004] However, existing technologies generally suffer from a fundamental problem of separation between forecasting and decision-making when using this benchmark sales volume for inventory decisions. Specifically, after the system generates sales forecasts, subsequent key decision parameters such as safety stock and reorder points still rely on manually set, fixed service level targets. This mechanism cannot effectively integrate known future business scenarios, such as brand promotion days, peak holiday traffic, or competitor discounts, into the dynamic adjustment of inventory strategies. This leads to a situation where even if the forecasting module knows there will be a major promotion next week, the decision-making module may still calculate a conservative replenishment quantity because the service level target remains unchanged, ultimately resulting in stockouts of key products during the promotion period, missing sales opportunities, and preventing true intelligent inventory optimization. Summary of the Invention

[0005] To address the technical problem of existing technologies being disconnected from prediction and decision-making, and thus unable to respond to future scenarios, this invention provides solutions in several aspects.

[0006] In a first aspect, the present invention provides an intelligent optimization method for store inventory, comprising: For each type of product in the store inventory, a dynamic decision-making profile is constructed, extracting the inherent behavioral attributes of the product under normal conditions, including the benchmark daily average sales, benchmark demand volatility, replenishment time, and customer satisfaction rate of the target product; future scenario events related to the target product are integrated, and the predicted adjusted sales are calculated based on the multiplication and superposition rule and the benchmark daily average sales; the target inventory direction is obtained according to the preset rule base; the dynamic customer satisfaction rate is calculated based on the boundary constraint relationship between the target inventory direction and the customer satisfaction rate of the minimum and maximum customer satisfaction rates, and adjusted by the logarithmic response function; the dynamic safety stock is obtained by multiplying the dynamic customer satisfaction rate, benchmark demand volatility, and replenishment time of the target product; the predicted adjusted sales during the replenishment time are added to the dynamic safety stock to obtain the dynamic reorder point; the target inventory is calculated based on the adjustment effect of the predicted adjusted sales on the dynamic safety stock; the replenishment order quantity is calculated based on the instability of inventory during transportation, combined with the dynamic reorder point, current inventory, and target inventory; the replenishment order quantity for each product is generated and the procurement and allocation process is initiated.

[0007] This invention departs from fixed historical average sales volume, obtaining a realistic inventory baseline by breaking down sales data, thus avoiding stockouts during peak seasons and overstocking during off-seasons. It integrates future events, such as promotions and competitive activities, to adjust estimated sales, resolving the disconnect between forecasts and actual demand. Customer satisfaction rate, safety stock, dynamic reorder point, and target inventory can all be flexibly adjusted, no longer using fixed values, avoiding inventory-demand mismatches. When calculating replenishment quantities, it considers in-transit inventory and supply chain risks, avoiding duplicate or insufficient replenishment. This invention aligns with actual demand, reduces stockouts and inventory backlogs, makes inventory management more precise, and solves the problems of traditional methods—the disconnect between forecasting and decision-making, and their inability to respond to future scenarios.

[0008] Preferably, the data includes the benchmark daily average sales volume of the target product, benchmark demand volatility, replenishment time, and customer satisfaction rate, including: Using any type of product in the store's inventory as the target product, obtain the historical daily sales sequence of the target product within a preset time period from the store's sales data system, and decompose it using a seasonal-trend decomposition method to obtain the daily sales trend sequence, seasonal cycle sequence, and residual sequence. The future decision-making day is designated as the decision day, and the daily sales trend sequence corresponding to the decision day is superimposed with the daily sales seasonal cycle sequence to obtain the benchmark daily average sales, denoted as the benchmark daily average sales. The standard deviation of the residual sequence is calculated and denoted as the benchmark demand volatility. The replenishment time is extracted from the store's product data management system. The company's standard customer satisfaction rate target for this product is extracted and denoted as the customer satisfaction rate.

[0009] Preferably, calculating the revised sales forecast includes: We collect all future events related to the target product from data sources such as the enterprise marketing calendar and supply chain management system, and organize them into a list of event tuples, denoted as the future scenario sequence. Each scenario group contains three influencing factors: event type, event start and end dates, and expected impact intensity. We iterate through the future scenario sequence to find all events whose impact time range covers a future replenishment time or replenishment review cycle, denoted as active events. Based on the influencing factors of all active events, we calculate the comprehensive scenario sales correction factor for the target product using a multiplicative summation rule, denoted as the sales correction factor. We multiply the baseline daily average sales by the sales correction factor to obtain the scenario-corrected predicted sales, denoted as the predicted corrected sales.

[0010] Preferably, the dynamic customer satisfaction rate satisfies the following expression: ; In the formula, This indicates the dynamic customer satisfaction rate of the target product. This indicates the customer satisfaction rate of the target product; This represents the sales adjustment factor, which is a real number greater than 0. , This represents the minimum and maximum customer satisfaction rates in the system. Represents the natural logarithm function; Represents the maximum value function; This represents the minimum value function.

[0011] The customer satisfaction rate in this invention is not fixed but flexibly adjusted based on future events. For example, during promotional activities, the customer satisfaction rate is appropriately increased to ensure more customers can purchase the goods. Simultaneously, the adjustment range is flexibly controlled to prevent the customer satisfaction rate from exceeding a reasonable minimum and maximum range. This avoids problems caused by a fixed customer satisfaction rate, such as customers wanting to buy but finding the goods unavailable during promotions, or excessively high satisfaction rates leading to product overstocking. It allows the degree of customer satisfaction to change with actual circumstances, ensuring customers can purchase the goods without wasting costs.

[0012] Preferably, obtaining a dynamic safety stock includes: Obtain the dynamic customer satisfaction rate of the target product, and calculate the standard normal distribution quantile of the dynamic customer satisfaction rate using a statistical algorithm, denoted as the satisfaction rate quantile; calculate the dynamic safety stock of the target product based on the product of the dynamic customer satisfaction rate, the satisfaction rate quantile, and the replenishment time, denoted as the dynamic safety stock.

[0013] Preferably, obtaining dynamic reorder points includes: Obtain the replenishment time of the target product, calculate the sum of the daily forecasted revised sales within the replenishment time of the target product, and record it as the total forecasted revised sales; calculate the dynamic reorder point of the target product based on the sum of the total forecasted revised sales and the dynamic safety stock, and record it as the dynamic reorder point.

[0014] This invention calculates the estimated sales volume during the replenishment period when determining the replenishment timing, and adds a flexibly adjustable safety stock, unlike the previous approach which only considered past average sales. For example, if replenishment requires 3 days, the reorder point is set based on the expected sales volume over those 3 days plus reserve stock. This avoids situations where replenishing too early leads to stockpiling or replenishing too late results in no stock to sell, ensuring that the replenishment timing perfectly matches actual demand, guaranteeing that the store has stock to sell while not occupying excessive inventory space.

[0015] Preferably, the target inventory satisfies the following expression: ; In the formula, Indicates the target inventory of the target product; Indicates the replenishment approval period for the target product; This represents the predicted revised sales volume on day t within the replenishment review period for the target product. This indicates the sales volume adjustment factor for the target product; This represents the dynamic safety stock of the target product.

[0016] When setting the ideal inventory level, this invention considers a future period, such as the estimated total sales volume during the replenishment review cycle, and adds a safety stock adjusted according to the actual situation. This avoids either insufficient inventory or inventory buildup during this period, ensuring that the target inventory just meets the demand during this period, while also dealing with potential risks, reducing inventory waste and stockouts, and making inventory management more in line with actual operations.

[0017] Preferably, the quantity of the replenishment order satisfies the following expression: ; In the formula, This indicates the quantity of replenishment orders for the target product; Indicates the target inventory of the target product; Indicates current inventory; This indicates the sales volume adjustment factor for the target product; Indicates inventory in transit; This represents the maximum value function.

[0018] This invention considers target inventory, current stock levels, and goods en route when calculating replenishment order quantities. It also assesses the reliability of goods in transit based on the impact of future events. For example, during promotional events, it doesn't fully rely on in-transit goods and replenishes slightly more to mitigate risk; otherwise, it calculates normally. This avoids problems caused by neglecting in-transit inventory, such as overstocking or stockouts due to insufficient replenishment orders. It ensures replenishment quantities precisely meet demand, addresses potential supply chain issues, and reduces replenishment errors.

[0019] Preferably, generating replenishment order quantities for each product and initiating the procurement and allocation process includes: Generate replenishment order quantities for each item in the store's inventory, automatically format the replenishment order quantities, and push them to the ERP or supply chain management system of the corresponding enterprise to initiate subsequent procurement or allocation processes, achieving end-to-end automation.

[0020] Secondly, the present invention provides an intelligent optimization system for store inventory, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the aforementioned intelligent optimization method for store inventory is implemented.

[0021] By adopting the above technical solution, a computer program for the intelligent optimization method of store inventory is generated and stored in a memory so that it can be loaded and executed by a processor. This allows for the creation of a terminal device based on the memory and processor, making it convenient to use.

[0022] The beneficial effects of this invention are as follows: It significantly improves the operational efficiency and competitiveness of retail stores and enterprises. For customers, stockouts are reduced, making it easier to purchase the goods they need and enhancing the customer experience. For enterprises, it reduces the cost waste caused by inventory backlog and lowers operating costs. Simultaneously, the automatic order push function saves manual time, reduces human error, and makes the procurement and allocation processes smoother, improving overall operational efficiency. Furthermore, it allows for flexible responses to market changes, such as promotions and competitive activities, making inventory management no longer passive but a support for sales, helping enterprises better balance customer demand and cost control. In the long run, it enhances the market adaptability of stores and provides strong support for the stable operation and development of enterprises. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating an intelligent optimization method for store inventory according to the present invention. Detailed Implementation

[0024] This invention discloses an intelligent optimization method for store inventory, referring to... Figure 1This includes steps S1-S4: S1: Build a dynamic decision-making profile for each type of product in the store's inventory, and extract the inherent behavioral attributes of the product under normal conditions, including the benchmark daily sales, benchmark demand volatility, replenishment time, and customer satisfaction rate of the target product.

[0025] It's important to note that in common retail scenarios such as supermarkets, convenience stores, and chain clothing stores, traditional inventory management methods that rely on average daily sales over the past year can lead to inaccurate stocking. For example, supermarket bottled beverage sales surge in summer due to high temperatures and plummet in winter, resulting in frequent stockouts in summer and inventory buildup in winter. Similarly, in chain clothing stores, spring shirts show a gradual upward trend from February onwards as temperatures rise, followed by a gradual decline after May with the arrival of summer clothing. Relying on averages not only fails to pinpoint the need for increased stocking from February to May and ignores the cyclical fluctuations in sales, such as slightly higher sales on weekends, but also cannot handle temporary drops in shirt sales caused by sudden temperature drops. These phenomena directly lead to the system failing to capture trends, cycles, and random fluctuations in product inventory, causing prediction bias, resulting in lost customers and increased inventory costs for stores. This invention addresses these pain points by designing an intelligent optimization method and system for store inventory. Taking bottled beverages in supermarkets as an example, it first uses the STL method to break down the daily sales of beverages over the past year into a sequence of trend items reflecting long-term changes, such as a slight annual increase of 5% in overall beverage sales due to changes in consumer spending habits, and a seasonal cycle reflecting cyclical fluctuations, such as sales from June to August being twice that of December to February, or 20 fewer bottles sold on a certain day due to heavy rain, and 15 more bottles sold on a certain day due to the weekend. Then, it determines the standard stocking benchmark for the following Monday in July, which falls within the summer season. The trend item data corresponding to the following Monday is then overlaid with the seasonal cycle data for summer to obtain a result unaffected by sudden weather events. The benchmark daily sales volume, which accurately matches the actual demand during the summer, is used to determine the impact of temporary promotions. The standard deviation of residual terms is also calculated; for example, the normal fluctuation range is ±15 bottles. The degree of random fluctuation in daily beverage demand is calculated as the benchmark demand volatility. This helps stores determine how much extra stock they need to prepare to cope with random changes. Combined with the 3-day delivery time from suppliers and the 95% customer satisfaction rate when customers can purchase beverages, these four parameters together constitute a stock preparation benchmark that reflects the actual sales patterns of beverages. Subsequent adjustments can be made dynamically based on this precise benchmark, whether it's due to supermarket anniversary promotions requiring additional stock or temporary supplier shortages necessitating adjustments to safety stock.

[0026] Specifically, any type of product in the store's inventory is selected as the target product. The historical daily sales volume time series of the target product within a preset time period is obtained from the store's sales data system and recorded as the historical daily sales series. The Seasonal-Trend Decomposition Method (STL) is used to decompose the historical daily sales series to obtain the daily sales trend series, the seasonal cycle series, and the residual series. The future decision-making day is designated as the decision day. The daily sales trend series corresponding to the decision day is superimposed with the daily sales seasonal cycle series to obtain the benchmark daily average sales volume of the product under the condition that it is not affected by any abnormal events in the future, recorded as the benchmark daily average sales volume. The standard deviation of the residual series is calculated and recorded as the benchmark demand volatility. The average number of days required for the target product from the placement of a replenishment order to the arrival of the goods in the store is extracted and recorded from the store's merchandise data management system and recorded as the replenishment time. The regular customer satisfaction rate target set by the enterprise for this product is extracted and recorded as the customer satisfaction rate.

[0027] It should be noted that stores, as the front end that directly reaches consumers, undertake the function of data collection to provide a basis for the company's inventory decisions. The replenishment needs of stores, i.e. the quantity of replenishment orders, are ultimately pushed to the company's ERP or supply chain management system, and the company completes centralized procurement and cross-store allocation in a unified manner. The company then integrates and analyzes the data through the system to generate accurate optimization strategies. Therefore, the relevant data of the company will be used multiple times in the subsequent invention.

[0028] At this point, the baseline daily average sales volume, baseline demand volatility, replenishment time, and customer satisfaction rate of the target product were obtained.

[0029] S2: Integrate future contextual events related to the target product, and calculate the predicted revised sales based on the multiplication rule and the baseline daily average sales.

[0030] It's important to note that traditional store inventory management often fails to factor in the impact of various product-related events on demand when calculating revised sales forecasts. These events include supermarket anniversary promotions, beverage brand exclusive sales promotion days, convenience store holiday peaks, and limited-time discounts from competing products. Relying solely on fixed historical sales benchmarks for prediction can easily lead to a disconnect between revised sales forecasts and actual demand. For example, if a supermarket stocking up on a particular dairy product without considering the upcoming member-only sale a week later and the "buy two get one free" promotion from competing products, and instead stocks based on daily sales benchmarks, it may experience stockouts due to surging demand during the promotion or inventory buildup due to competition from other products. This invention integrates the impact of future events. Taking a supermarket dairy product as an example, it first obtains information from data sources such as the company's marketing calendar (member day promotion), supply chain management system (confirming inventory scheduling during the promotion period), and external business intelligence platform (competitive product activity information). All future events related to this dairy product are collected and organized into a list of event tuples. For example, for the member day promotion, event type A: store promotion, event start and end dates July 15th-July 17th, expected impact strength 1.7; event type B: competitor product promotion, event start and end dates July 16th-July 18th, expected impact strength 0.8; then... By iterating through this sequence of future scenarios, we identify events whose impact timeframe covers a future replenishment period, such as 4 days, or a replenishment review cycle, confirming that both events are active events. Then, according to the rule of multiplication and superposition, we multiply the expected impact strength of the two active events, for example, 1.7 × 0.8 = 1.36, to calculate the comprehensive scenario sales correction factor for the dairy product. Finally, we multiply the baseline daily average sales by the comprehensive scenario sales correction factor to obtain the scenario-corrected predicted sales, ensuring that the inventory can meet the increased demand on member days while avoiding the risk of demand diversion caused by competing product activities, thus avoiding the bias of traditional forecasting methods.

[0031] Specifically, all future events related to the target product are collected from data sources such as the enterprise marketing calendar and supply chain management system, and organized into a list of event tuples, denoted as the future scenario sequence. Each scenario group contains three influencing factors: event type, event start and end dates, and expected impact intensity. The future scenario sequence is traversed to identify all events whose impact time range covers a future replenishment time or replenishment review cycle, denoted as active events. Based on the influencing factors of all active events, the comprehensive scenario sales correction factor for the target product is calculated using a multiplicative summation rule, denoted as the sales correction factor. The baseline daily average sales are multiplied by the sales correction factor to obtain the scenario-corrected predicted sales, denoted as the predicted corrected sales.

[0032] S3: Obtain the target inventory direction based on the preset rule base; calculate the dynamic customer satisfaction rate by adjusting the minimum and maximum customer satisfaction rates on the boundary constraints between the target inventory direction and the customer satisfaction rate through a logarithmic response function; multiply the dynamic customer satisfaction rate of the target product, the baseline demand volatility, and the replenishment time to obtain the dynamic safety stock; add the predicted adjusted sales volume within the replenishment time to the dynamic safety stock to obtain the dynamic reorder point; calculate the target inventory based on the adjustment effect of the predicted adjusted sales volume on the dynamic safety stock.

[0033] It's important to note that traditional inventory management often relies on fixed, empirical coefficients when adjusting customer satisfaction rates, failing to consider dynamic characteristics such as future product-related events like campus promotions for snacks, early warnings of fresh food logistics delays, and new product launches by competitors. This can easily lead to a mismatch between customer satisfaction rates and actual demand. For example, if a snack brand prepares inventory and ignores the risks of campus promotions a week later and concurrent logistics delays, allocating inventory based on the conventional satisfaction rate will either cause a sharp drop in satisfaction rate due to a surge in promotional demand, or result in redundant inventory and an artificially high satisfaction rate due to logistics delays. This invention, however, dynamically adjusts based on future active events. Taking this snack as an example, it first acquires active events such as campus promotions and logistics delays, determines the direction of inventory adjustment using a matching rule base, and then calculates a satisfaction rate adjustment coefficient using a logarithmic response function. This dynamically scales the customer satisfaction rate, ensuring it aligns with changes in promotional demand while mitigating the impact of supply risks, thus overcoming the limitations of traditional fixed coefficients.

[0034] It should be noted that the dynamic customer satisfaction rate is built upon the classic dynamic service level adjustment model and demand elasticity response theory in the field of retail supply chain management. It is a mature and improved method for balancing customer service experience and inventory holding costs in chain store inventory optimization. For the inventory management of multi-category goods in chain retail stores, future demand fluctuations brought about by business scenarios will directly change the risk of stockouts and the probability of inventory backlog. Based on the benchmark customer satisfaction rate, logarithmic smoothing is applied using a sales adjustment factor. This method can adapt to complex retail scenarios such as promotional activities, peak holiday traffic, and competitor diversion, eliminating the supply-demand mismatch problem caused by fixed service level targets. In retail scenarios, the impact of normal demand fluctuations and extreme demand changes on inventory strategies is fundamentally different. This invention, through the adjustment of the sales adjustment factor using a natural logarithmic function, achieves a smooth adaptation of moderately increasing service levels when demand rises and reasonably decreasing service levels when demand falls. This aligns with the operational rules of supply-demand matching in the retail industry and avoids drastic inventory fluctuations and uncontrolled operating costs caused by sudden changes in service levels. Store inventory operations inherently involve compliance and cost boundaries for service levels. This invention utilizes a range constraint mechanism constructed through maximum and minimum value functions to strictly limit the dynamically adjusted customer satisfaction rate within a pre-defined reasonable range. This ensures that the basic service baseline is not breached while avoiding excessive pursuit of high service levels that leads to inventory backlog and capital tied up, effectively balancing customer experience and business operating costs. The sales characteristics and demand fluctuation patterns of chain retail stores differ significantly. The dynamic customer satisfaction rate, through a dual design of logarithmic smoothing adjustment and range boundary constraints, confines service level anomalies caused by regular random fluctuations within a reasonable range. Service level adjustments are only made for predictable changes in effective demand, fully adapting to the daily inventory operations and promotional stocking decisions required by chain retail across multiple categories and scenarios.

[0035] Preferably, the event type of the active events of the target product in the future scenario sequence is obtained and denoted as the target event type; the corresponding inventory adjustment direction is determined by matching the target event type with a preset rule base and denoted as the target inventory direction; based on the target inventory direction, a dimensionless multiplier coefficient for dynamically scaling the customer satisfaction rate of the target product is obtained through a logarithmic response function and denoted as the satisfaction rate adjustment coefficient; the dynamic customer satisfaction rate is obtained by multiplying the satisfaction rate adjustment coefficient and the customer satisfaction rate, including: Dynamic customer satisfaction rate satisfies the following expression: ; In the formula, This indicates the dynamic customer satisfaction rate of the target product. This indicates the customer satisfaction rate of the target product; This represents the sales adjustment factor, which is a real number greater than 0. , This represents the minimum and maximum customer satisfaction rates in the system. Represents the natural logarithm function; Represents the maximum value function; This represents the minimum value function.

[0036] In the formula, This represents the satisfaction rate adjustment coefficient, used to achieve smooth and controllable adjustment of the dynamic customer satisfaction rate by the sales correction factor. The natural logarithmic function has the characteristics of monotonically increasing but with a gradually slowing growth rate, which can avoid drastic fluctuations in the adjustment strength caused by sudden changes in sales volume, prevent jumps in customer satisfaction rate, and adapt to the gradual changes in demand brought about by events such as promotional activities, peak customer traffic during holidays, and diversion of customers by competitors in retail scenarios; This means that the customer satisfaction rate is dynamically adjusted based on the benchmark customer satisfaction rate of the target product and combined with the satisfaction rate adjustment coefficient to obtain the preliminary adjusted customer satisfaction rate. This realizes the core logic of adapting the customer satisfaction rate to changes in sales caused by future situational events. This expression uses nested constraints of maximum and minimum value functions to strictly limit the initially adjusted customer satisfaction rate within the system's preset range of minimum and maximum customer satisfaction rates. When the initial adjustment value is lower than the minimum customer satisfaction rate, the minimum customer satisfaction rate is output to ensure the basic service baseline set by the enterprise. When the initial adjustment value is higher than the maximum customer satisfaction rate, the maximum customer satisfaction rate is output to avoid inventory backlog and capital occupation caused by excessive pursuit of high service levels. When the initial adjustment value is within the range, the corrected calculated value is directly output. This expression can dynamically map the impact of future active events on customer satisfaction rate. Different target event types, such as store promotions, peak holiday traffic, competitor diversion, and new product launches, will dynamically adjust the customer satisfaction rate through sales adjustment factors. This makes the customer satisfaction rate no longer a fixed value set manually, but can achieve intelligent and smooth dynamic adjustment based on the real-time market situation and demand changes of the product, while always constraining the results within a reasonable range acceptable to the enterprise's operations. This effectively solves the core problems of supply and demand mismatch and the disconnect between forecasting and decision-making caused by traditional fixed service levels.

[0037] It should be noted that the target inventory is constructed based on the classic periodic inventory counting model and the dynamic adjustment theory of safety stock in the field of inventory control. It is a mature and improved paradigm for determining the optimal inventory level in retail supply chain replenishment management. For multi-cycle replenishment scenarios of chain stores, the total demand within the replenishment review cycle directly determines the basic scale of inventory replenishment. Using the cumulative value of the daily predicted and corrected sales within the review cycle as the basic inventory level, it can adapt to various products with different replenishment cycles and different sales frequencies, eliminating the problem of inventory and demand disconnect caused by static inventory preparation in fixed cycles. In retail scenarios, the uncertainty of regular demand is fundamentally different from the uncertainty of demand brought about by events such as promotions. This formula introduces the sales correction factor into the dynamic adjustment term of safety stock, realizing a risk adaptation mechanism where the greater the demand fluctuation, the stronger the safety stock buffer. This aligns with the positive correlation between demand uncertainty and safety stock in the retail industry, effectively addressing the risk of stockouts caused by future events. Store replenishment management inherently involves demand forecasting bias. This regular bias is fundamentally different from the forecasting bias caused by extreme events. Target inventory, through a combination of cyclical demand accumulation and dynamic safety stock design, constrains the impact of regular forecasting bias within a safe buffer zone, amplifying inventory buffer capacity only for extreme demand changes, and adapting to the replenishment decision-making needs of chain retail stores across multiple categories and scenarios.

[0038] Preferably, the dynamic customer satisfaction rate of the target product is obtained, and the standard normal distribution quantile of the dynamic customer satisfaction rate is calculated using a statistical algorithm, denoted as the satisfaction rate quantile; based on the dynamic customer satisfaction rate, satisfaction rate quantile, and replenishment time of the target product, the dynamic safety stock of the target product is calculated, denoted as the dynamic safety stock, including: Dynamic safety stock satisfies the following expression: ; In the formula, This indicates the dynamic safety stock of the target product; Indicates the satisfaction rate quantile of the target product; This represents the benchmark demand volatility for the target commodity. Indicates the replenishment time for the target product; This represents the square root function.

[0039] In the formula, This means that the dynamic customer satisfaction rate is converted into the inventory risk resistance strength, denoted as the risk coefficient. That is, when the dynamic customer satisfaction rate increases due to events such as promotions, the satisfaction rate quantile will increase non-linearly and rapidly, indicating that a stronger inventory buffer is needed. This indicates the magnitude of random fluctuations in demand during the replenishment period. This means multiplying the risk coefficient by the magnitude of demand fluctuations to obtain the uncertainty buffer inventory required to achieve dynamic customer satisfaction. By combining customer satisfaction targets, demand fluctuations, and replenishment times, a precise buffer inventory is calculated as dynamic safety stock, making inventory management more flexible and aligned with actual business needs.

[0040] It's important to note that traditional inventory management often relies on fixed historical average daily sales to extrapolate demand within the replenishment period when calculating dynamic reorder points. This fails to incorporate dynamically adjusted sales forecasts and often uses static safety stock, easily leading to a disconnect between the reorder point and actual demand. For example, if a supermarket's dairy product has a 3-day replenishment period, traditional methods calculate the total demand for those 3 days based on average daily sales. This approach neither considers the increased sales forecasts due to member-day promotions nor utilizes dynamic safety stock. Either the surge in promotional demand causes existing inventory to fall below the dynamic reorder point prematurely, resulting in stockouts, or the excessively high demand forecast leads to an artificially inflated dynamic reorder point, resulting in inventory buildup after replenishment. This invention first obtains the replenishment time for the target product, then calculates the sum of the daily adjusted sales forecasts within that replenishment period, and finally adds the total adjusted sales forecasts to the dynamic safety stock to obtain a dynamic reorder point that accurately reflects the actual demand and risks during the replenishment period. This ensures precise timing of replenishment actions and reasonable inventory levels.

[0041] Specifically, the replenishment time of the target product is obtained, and the sum of the daily forecasted revised sales volume within the replenishment time of the target product is calculated and recorded as the total forecasted revised sales volume; based on the sum of the total forecasted revised sales volume and the dynamic safety stock, the dynamic reorder point of the target product is calculated and recorded as the dynamic reorder point.

[0042] It's important to note that using historical sales averages over a fixed period without incorporating dynamic sales forecasts within the replenishment approval cycle, and calculating target inventory using a fixed safety stock value, often leads to a disconnect between target inventory and actual demand. For example, if a convenience store sells snacks and sets a 5-day replenishment approval cycle, the traditional approach calculates total demand based on the average daily sales over the past 5 days. This doesn't account for the increased sales expected from next week's snack promotions, nor does it utilize dynamic safety stock. Either the promotional demand exceeds expectations, resulting in insufficient target inventory and rapid stockouts after replenishment; or the demand forecast is too high, leading to inflated target inventory and inventory buildup.

[0043] It should be noted that the target inventory is constructed based on the classic periodic inventory counting model and the dynamic adjustment theory of safety stock in the field of inventory control. It is a mature and improved method for determining the optimal inventory level in retail supply chain replenishment management. For multi-cycle replenishment scenarios of chain stores, the total demand within the replenishment review cycle directly determines the basic scale of inventory replenishment. Using the cumulative value of the daily predicted and corrected sales within the review cycle as the basic inventory level, it can adapt to various products with different replenishment cycles and different sales frequencies, eliminating the problem of inventory and demand disconnect caused by static inventory preparation in fixed cycles. In retail scenarios, the uncertainty of regular demand is fundamentally different from the uncertainty of demand brought about by events such as promotions. This invention introduces the sales correction factor into the dynamic adjustment item of safety stock, realizing a risk adaptation mechanism where the greater the demand fluctuation, the stronger the safety stock buffer. This aligns with the positive correlation between demand uncertainty and safety stock in the retail industry, effectively addressing the risk of stockouts caused by future event scenarios. Store replenishment management inherently involves demand forecasting bias. This regular bias is fundamentally different from the forecasting bias caused by extreme events. Target inventory, through a combination of cyclical demand accumulation and dynamic safety stock design, constrains the impact of regular forecasting bias within a safe buffer zone, amplifying inventory buffer capacity only for extreme demand changes, and adapting to the replenishment decision-making needs of chain retail stores across multiple categories and scenarios.

[0044] Preferably, the target inventory is obtained by adjusting the sum of the total future forecast sales and the dynamic safety stock within the replenishment review period, including: Obtain the predicted revised sales volume of the target product for each day in the future, and record it as the future predicted revised sales volume; obtain the replenishment approval cycle of the target product from the product data management system, obtain the dynamic safety stock of the target product, and calculate the adjusted dynamic safety stock in combination with the sales volume correction factor of the target product.

[0045] The target inventory satisfies the following expression: ; In the formula, Indicates the target inventory of the target product; Indicates the replenishment approval period for the target product; This represents the predicted revised sales volume on day t within the replenishment review period for the target product. This indicates the sales volume adjustment factor for the target product; This represents the dynamic safety stock of the target product.

[0046] In the formula, This represents the total revised sales forecast within the replenishment review period for the target product. It is the total expected sales before the next inventory check, i.e., the basic demand that replenishment must meet. It is a safety buffer that should still be retained after basic needs have been met; express cThe larger the value, the greater the actual uncertainty surrounding the predicted sales volume, and therefore a risk buffer that is amplified by an equal amount is needed. Conversely, when an event causes a contraction in sales, c < 1, the risk is reduced accordingly, and the required buffer should also be reduced. and The sum is used to set the ideal maximum inventory level for this replenishment, which is recorded as the target inventory, to avoid the problem of over- or under-replenishing after experience-based replenishment.

[0047] S4: Based on the instability of inventory during transportation, combined with dynamic reorder points, current inventory, and target inventory, calculate the quantity of replenishment orders; generate the quantity of replenishment orders for each type of product and initiate the procurement and allocation process.

[0048] It's important to note that simply comparing the target inventory with the current actual inventory ignores in-transit inventory that has already been ordered but is still en route, which can easily lead to a mismatch between replenishment quantities and actual demand. For example, if a supermarket is replenishing beverages with a target inventory of 100 units, a current actual inventory of 30 units, and 50 units in transit, the traditional method of subtracting 30 from 100 would result in a need for 70 units, leading to over-replenishment and inventory buildup because the in-transit inventory was not included. Conversely, if replenishment is urgently needed but the in-transit inventory is not calculated, misjudging the shortage could lead to stockouts. This invention simultaneously acquires the target inventory, current inventory, and in-transit inventory to calculate the replenishment order quantity, avoiding duplicate replenishment of in-transit inventory.

[0049] It should be noted that the replenishment order quantity is constructed based on the classic in-transit inventory hedging model and replenishment gap dynamic accounting theory in the field of supply chain management. It is a mature and improved paradigm for accurately calculating the replenishment scale in retail store replenishment execution. For chain stores' cross-cycle replenishment scenarios, the fulfillment reliability of in-transit inventory directly affects the calculation result of the actual inventory gap. By deducting the current effective inventory from the target inventory as the benchmark level, it can adapt to different replenishment lead times and different supply chain fulfillment capabilities, eliminating the problems of duplicate replenishment and inventory backlog caused by ignoring in-transit inventory. In retail scenarios, the reliability of in-transit inventory under normal stable demand is fundamentally different from that under high fulfillment pressure during promotional periods. The target inventory expression introduces the sales adjustment factor into the reduction item of in-transit inventory, realizing a risk hedging mechanism where the stronger the demand and the higher the supply chain fulfillment risk, the stronger the reduction of in-transit inventory. This aligns with the correlation between supply chain fulfillment risk and demand intensity in the retail industry, effectively avoiding the stockout risk caused by delayed delivery of in-transit inventory during promotional periods. The execution of store replenishment inherently involves supply chain fulfillment uncertainties. These regular fulfillment fluctuations are fundamentally different from the fulfillment risks under extreme demand. The replenishment order quantity incorporates the impact of regular fulfillment fluctuations into the inventory gap calculation through a dual design of effective inventory dynamic accounting and non-negative constraints. It strengthens the reduction of in-transit inventory only for high-risk scenarios, adapting to the full-scenario execution needs of chain retail stores for daily replenishment and promotional stock preparation.

[0050] Specifically, the target inventory of the target product is obtained; the current inventory of the target product is obtained through a real-time inventory monitoring system; it is determined whether the current inventory is less than or equal to the dynamic reorder point; if so, the quantity of replenishment orders is calculated based on the target inventory, current inventory, and in-transit inventory, including: The quantity of replenishment orders satisfies the following expression: ; In the formula, This indicates the quantity of replenishment orders for the target product; Indicates the target inventory of the target product; Indicates current inventory; This indicates the sales volume adjustment factor for the target product; Indicates inventory in transit; This represents the maximum value function.

[0051] In the formula, This indicates that a sales adjustment factor is used as an evaluation indicator of supply chain stress. The reliability of in-transit inventory is adjusted in real-time. When there is a large promotional event (c>1), the system automatically divides the effective value of the in-transit inventory by the sales adjustment factor of the target product to apply a discount. This implies that the system does not fully trust that the in-transit inventory will arrive on time, thus calculating a larger inventory gap. More replenishment orders are then placed in advance to hedge against supply chain risks. When the market is stable or demand contracts, c≤1. If the value is 1, the system fully trusts the inventory in transit; This indicates a future event that will trigger a surge in demand. A higher c-value will also put enormous pressure on the entire supply chain's production and logistics, thereby significantly increasing in-transit inventory. Delivery risks, such as delays and shortages; This means that by subtracting the effective inventory after situational stress risk assessment from the target inventory, a true inventory gap can be obtained that can intelligently predict and hedge against supply chain fulfillment risks. This represents the final replenishment order quantity calculated after a supply chain stress risk assessment, ensuring a non-negative result.

[0052] It should be noted that manually entering data into the enterprise system is not only prone to errors due to inconsistent formats, making it difficult for the system to quickly recognize the information, but also susceptible to human input errors such as incorrect quantity entries or confused product codes, which can delay procurement or allocation processes and even cause a break in the inventory connection between the enterprise and its stores. This invention solves this problem through an automated process. For each product, the system first automatically calculates the accurate replenishment order quantity, then automatically formats the quantity according to the enterprise system's requirements, such as matching product code rules and order data field formats, and directly pushes it to the corresponding enterprise's ERP or supply chain management system.

[0053] Preferably, a replenishment order quantity is generated for each item in the store's inventory, and the replenishment order quantity is automatically formatted and pushed to the ERP or supply chain management system of the corresponding enterprise of the store to initiate the subsequent procurement or allocation process, thereby achieving end-to-end automation.

[0054] This invention also discloses an intelligent optimization system for store inventory, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement an intelligent optimization method for store inventory according to the present invention.

[0055] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0056] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A smart optimization method for store inventory, characterized in that, include: Build a dynamic decision-making profile for each type of product in the store inventory, and extract the inherent behavioral attributes of the product under normal conditions, including the benchmark daily sales, benchmark demand volatility, replenishment time, and customer satisfaction rate of the target product. By integrating future scenario events related to the target product, and based on the multiplication rule, combined with the baseline daily average sales, the predicted revised sales are calculated. Based on a pre-defined rule base, obtain the target inventory direction; Based on the boundary constraints between the minimum and maximum customer satisfaction rates and the target inventory direction and customer satisfaction rates, the dynamic customer satisfaction rate is calculated after adjustment by the logarithmic response function. The dynamic safety stock is obtained by multiplying the dynamic customer satisfaction rate of the target product, the baseline demand volatility, and the replenishment time. The dynamic reorder point is obtained by adding the predicted adjusted sales volume within the replenishment time to the dynamic safety stock. The target inventory is calculated based on the adjustment effect of the predicted adjusted sales volume on the dynamic safety stock. Based on the instability of inventory levels during transportation, and combined with dynamic reorder points, current inventory, and target inventory, the quantity of replenishment orders is calculated. Generate replenishment order quantities for each product and initiate the procurement and allocation process.

2. The intelligent optimization method for store inventory according to claim 1, characterized in that, The data includes the benchmark daily average sales volume, benchmark demand volatility, replenishment time, and customer satisfaction rate for the target product, including: Using any type of product in the store's inventory as the target product, obtain the historical daily sales sequence of the target product within a preset time period from the store's sales data system, and decompose it using a seasonal-trend decomposition method to obtain the daily sales trend sequence, seasonal cycle sequence, and residual sequence. The future decision-making day is designated as the decision day, and the daily sales trend sequence corresponding to the decision day is superimposed with the daily sales seasonal cycle sequence to obtain the benchmark daily average sales, denoted as the benchmark daily average sales. The standard deviation of the residual sequence is calculated and denoted as the benchmark demand volatility. The replenishment time is extracted from the store's product data management system. The company's standard customer satisfaction rate target for this product is extracted and denoted as the customer satisfaction rate.

3. The intelligent optimization method for store inventory according to claim 1, characterized in that, The calculation of the revised sales forecast includes: We collect all future events related to the target product from data sources such as the enterprise marketing calendar and supply chain management system, and organize them into a list of event tuples, denoted as the future scenario sequence. Each scenario group contains three influencing factors: event type, event start and end dates, and expected impact intensity. We iterate through the future scenario sequence to find all events whose impact time range covers a future replenishment time or replenishment review cycle, denoted as active events. Based on the influencing factors of all active events, we calculate the comprehensive scenario sales correction factor for the target product using a multiplicative summation rule, denoted as the sales correction factor. We multiply the baseline daily average sales by the sales correction factor to obtain the scenario-corrected predicted sales, denoted as the predicted corrected sales.

4. The intelligent optimization method for store inventory according to claim 1, characterized in that, The dynamic customer satisfaction rate satisfies the following expression: ； In the formula, This indicates the dynamic customer satisfaction rate of the target product. This indicates the customer satisfaction rate of the target product; This represents the sales adjustment factor, which is a real number greater than 0. , This represents the minimum and maximum customer satisfaction rates in the system. Represents the natural logarithm function; Represents the maximum value function; This represents the minimum value function.

5. The intelligent optimization method for store inventory according to claim 1, characterized in that, The process of obtaining dynamic safety stock includes: Obtain the dynamic customer satisfaction rate of the target product, and calculate the standard normal distribution quantile of the dynamic customer satisfaction rate using a statistical algorithm. This quantile is denoted as the satisfaction rate quantile. The product of the dynamic customer satisfaction rate, the satisfaction rate quantile, and the replenishment time of the target product is used as the dynamic safety stock of the target product.

6. The intelligent optimization method for store inventory according to claim 1, characterized in that, The process of obtaining dynamic order points includes: Obtain the replenishment time of the target product, calculate the sum of the daily forecasted revised sales within the replenishment time of the target product, and record it as the total forecasted revised sales; calculate the dynamic reorder point of the target product based on the sum of the total forecasted revised sales and the dynamic safety stock, and record it as the dynamic reorder point.

7. The intelligent optimization method for store inventory according to claim 1, characterized in that, The target inventory satisfies the following expression: ； In the formula, Indicates the target inventory of the target product; Indicates the replenishment approval period for the target product; This represents the predicted revised sales volume on day t within the replenishment review period for the target product. This indicates the sales volume adjustment factor for the target product; This represents the dynamic safety stock of the target product.

8. The intelligent optimization method for store inventory according to claim 1, characterized in that, The quantity of the replenishment order satisfies the following expression: ； In the formula, This indicates the quantity of replenishment orders for the target product; Indicates the target inventory of the target product; Indicates current inventory; This indicates the sales volume adjustment factor for the target product; Indicates inventory in transit; This represents the maximum value function.

9. The intelligent optimization method for store inventory according to claim 1, characterized in that, The process of generating replenishment order quantities for each product and initiating the procurement and allocation process includes: Generate replenishment order quantities for each item in the store's inventory, automatically format the replenishment order quantities, and push them to the ERP or supply chain management system of the corresponding enterprise to initiate subsequent procurement or allocation processes, achieving end-to-end automation.

10. An intelligent optimization system for store inventory, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a smart optimization method for store inventory according to any one of claims 1-9.

Images