A digital inventory estimation system
By collecting and cleaning multi-source data, and combining geofencing and behavior compensation models, we have achieved real-time and accurate estimation of terminal inventory and risk warning. This solves the problems of data instability and reliance on manual labor in existing technologies, and improves the automation of inventory management and the efficiency of supply chain collaboration.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- LUZHOU LAOJIAO CO LTD
- Filing Date
- 2026-04-29
- Publication Date
- 2026-07-31
AI Technical Summary
Existing terminal inventory management systems rely on non-standard outbound barcode scanning operations and low frequency of manual inventory checks, resulting in unstable data that cannot reflect consumer behavior in real time, affecting the accuracy of inventory calculations and the efficiency of supply chain decision-making.
By collecting data from terminal warehousing, consumer bottle opening and scanning, and salesperson visits, the system performs data cleaning and multi-layered calculation models. Combined with geofencing filtering and behavioral data compensation, it enables real-time inventory estimation and early warning decision-making.
It improves the authenticity and credibility of inventory data, realizes the automation and real-time operation of inventory management, enhances the adaptability and accuracy of inventory estimation, supports proactive risk control and decision-making, and improves supply chain collaboration efficiency.
Smart Images
Figure CN122492075A_ABST
Abstract
Description
Technical Field
[0001] This invention relates primarily to the field of supply chain management technology, and in particular to a digital inventory estimation system. Background Technology
[0002] With the expansion of distribution channels and diversification of terminal formats in the fast-moving consumer goods (FMCG) industry (such as alcoholic beverages, food, and soft drinks), inventory management at various terminals, including convenience stores, kiosks, and specialty stores, has become a key link in improving supply chain efficiency. From a technological evolution perspective, terminal inventory management has gradually transitioned from early manual inventory counting and bookkeeping to an information-based and systematic management stage. This relies on information technology to achieve digital data entry and automated calculation, with the core objective of accurately grasping the real-time inventory status at the terminal to support upstream replenishment scheduling, inventory optimization, and other supply chain decisions. Currently, the industry has formed a technological system centered on standardized inventory management systems, with inventory accounting models based on the "inbound minus outbound" logic becoming the mainstream technical solution, widely used in the supply chain management of FMCG companies of all sizes.
[0003] Currently, widely used terminal inventory management products on the market are represented by SAP systems and Oracle inventory modules. Their application areas cover terminal inventory accounting across the entire FMCG industry, adapting to various inventory management scenarios from large chain terminals to small individual stores. They are core tools for enterprises to achieve digital management of terminal inventory. The core operating principle of these products is based on the "inbound minus outbound" inventory accounting model. The key technologies supporting this mainstream model mainly include the following three categories: First, using barcode scanners for scanning and manual data entry, this technology is responsible for converting the physical inbound and outbound behaviors of the terminal into digital data, serving as the basic data source for inventory accounting. The core function of this technology is to link physical flow with data records, but it relies on standardized operation by terminal staff. Second, relying on the database storage function of enterprise-level sales systems and inventory management modules, this technology performs structured storage of collected inbound and outbound data and executes inventory accounting formulas in real time or periodically through built-in simple arithmetic logic to output inventory results. The core of this technology is to ensure the stability of data storage and the accuracy of calculation, but the calculation logic is fixed and simple, only capable of basic difference calculations. Third, through basic data entry interfaces (such as the hardware interface between the barcode scanner and the sales system, and the software operation interface for manual data entry), one-way or two-way transmission of terminal field data and the back-end management system can be realized to ensure that the data entering and leaving the warehouse can be synchronized to the computing module in a timely manner. The core of this technology is to open up the data channel between the terminal and the system, but the interaction method is relatively basic and lacks the ability to integrate multi-source data.
[0004] Although the "inbound minus outbound" model has become the industry mainstream, its inherent flaws in practical applications are difficult to avoid due to limitations in its technical architecture and design logic, resulting in inventory management effects falling short of expectations. Firstly, the core data foundation of this model relies on the completeness and accuracy of outbound records. However, in many small and medium-sized terminals, such as individual convenience stores and small liquor stores, staff often engage in non-standard operations to save time, such as batch entry of "inbound as outbound" and skipping the scanning step during outbound processing. This leads to missing outbound data or discrepancies with actual sales, causing inventory calculations to deviate significantly from physical inventory. Secondly, the model relies on regular on-site inventory checks by business personnel to calibrate inventory data. However, the inventory check cycle varies significantly due to factors such as labor costs and regional distribution, with some areas checking weekly and others monthly. This prevents high-frequency updates of inventory data, resulting in a 1-7 day delay between the system's presented inventory status and the actual terminal inventory, making it difficult to support real-time supply chain decisions. Finally, the model only focuses on the "inbound-outbound" data at the terminal, failing to incorporate actual consumer behavior data, such as the bottle opening and scanning records of alcoholic beverages, into the inventory accounting logic. In the fast-moving consumer goods (FMCG) industry, the outflow of goods from the terminal does not equate to immediate consumption by consumers. Some products may be temporarily stored at the terminal or used later after purchase. Existing models cannot track actual inventory consumption through consumer data, resulting in a lack of key basis for estimating the remaining inventory at the terminal. This is especially true in scenarios such as alcoholic beverages where consumption needs to be confirmed by opening the bottle, which significantly limits the accuracy of inventory accounting.
[0005] The above reflects several issues: First, the heavy reliance on outbound barcode scanning operations is disconnected from the actual execution scenarios at the terminal, resulting in an unstable data foundation. Second, the data sources are limited to the terminal channels and do not extend to the consumer end, lacking core dimensions that reflect actual consumption. Furthermore, the manual inventory calibration model is inefficient and cannot meet the high-frequency, real-time inventory management needs. These technical deficiencies collectively restrict the accuracy and effectiveness of terminal inventory management, leading to passive supply chain decision-making and resource waste in the industry, requiring more advanced technical solutions. Summary of the Invention
[0006] This invention provides a digital inventory estimation system, which aims to solve the problems of unreliable data, high dependence on manual operations, and low model accuracy in existing terminal inventory management technologies.
[0007] The technical solution adopted by the present invention to solve the above-mentioned technical problems A digital inventory estimation system, comprising: The data acquisition module is used to collect terminal inventory data, consumer bottle opening and scanning data, and salesperson visit data. The data cleaning module is used to deduplicatize the data collected by the data acquisition module, filter invalid barcode scans based on geofencing, and bind the barcode scan data to the box code data to which the bottle code belongs. The inventory estimation module performs the following steps on the data processed by the data cleaning module: S1: Based on the terminal warehousing data statistics and after deducting return and transfer records, obtain the net number of full cases received; S2: Calculate the number of boxes opened based on the correlation between bottle code and box code according to the consumer's bottle opening and scanning data, and compensate and correct the number of boxes opened based on the historical scanning rate; S3: Obtain the theoretical inventory by subtracting the compensated and corrected number of boxes from the net number of boxes received. S4: By analyzing the historical order and sales fluctuations of the terminal, fit the loss aversion coefficient, and adjust the theoretical inventory number according to the loss aversion coefficient to make a safety boundary, and output the real-time estimated inventory. The early warning and decision-making module is used to classify inventory status and judge risks based on preset safety stock thresholds, number of days without restocking and number of days without business visits, combined with real-time inventory calculations, and automatically generate replenishment suggestions, visit tasks or approval instructions.
[0008] Furthermore, the system also includes a visual user interface that outputs inventory trends and decision results based on a dashboard or API interface.
[0009] Furthermore, the terminal warehousing data includes product code, terminal code, warehousing time, and box code information; the consumer bottle opening and scanning data includes bottle code, scanning time, and geographical location information; and the salesperson visit data includes terminal code and check-in time.
[0010] Furthermore, the data acquisition module acquires data in the following way: it asynchronously streams terminal data, consumer bottle opening and scanning data, and salesperson visit data based on a message queue.
[0011] Furthermore, the cleaning module's invalid barcode scanning filtering based on geofence specifically includes: calculating the distance between the consumer's geographical location information when scanning the barcode and the corresponding terminal registration address; if the distance exceeds the preset geofence radius, the barcode scanning data is marked as invalid and isolated, and is not included in subsequent inventory calculations.
[0012] Furthermore, the scanning rate is a statistical ratio based on historical bottle opening scanning data and actual consumption data at the regional or terminal level.
[0013] Furthermore, the early warning and decision-making module is also used for: decision-making logic based on rule engine and threshold management, triggering an emergency replenishment suggestion when the real-time calculated inventory value is lower than the safety stock limit and the number of days without restocking exceeds the first threshold; and triggering a salesperson visit task when the number of days without visits exceeds the second threshold.
[0014] Furthermore, the visual user interface is used to generate interactive dashboards, trend charts, and terminal map views, and provides inventory trend data and early warning information to external systems in a standardized JSON format via API.
[0015] Furthermore, the system communicates and connects with the terminal sales system, consumer scanning platform, and enterprise ERP or CRM system via API or IoT interface.
[0016] Furthermore, the data cleaning module is also used to perform deduplication verification based on the unique bottle code within a set time window to ensure that the same barcode is not counted repeatedly.
[0017] The beneficial effects of this invention are: 1. It improves the authenticity and credibility of inventory data, abandons the traditional single data source model that relies on terminal outbound scanning, and fundamentally avoids inventory deviations caused by problems such as missing or mis-recorded outbound records. This makes the inventory data more consistent with the actual inventory situation at the terminal, providing a solid foundation for subsequent decision-making.
[0018] 2. It has achieved automated and real-time operation of inventory management. The entire process, from streaming collection of multi-source data and automatic cleaning and structuring to dynamic inventory calculation, does not require manual intervention for inventory counting or data entry. It completely eliminates the dependence on manual operation, reduces the impact of human error, and breaks the dilemma of inventory data lag caused by the traditional inventory counting cycle. This allows managers to grasp the dynamics of terminal inventory in a timely manner and improve management response efficiency.
[0019] 3. The behavioral data compensation sub-model enhances the adaptability and accuracy of the inventory estimation model, effectively addressing the issue of insufficient consumer scanning rates in different scenarios. It avoids reducing estimation accuracy due to incomplete scanning coverage and adapts to various terminal operation scenarios. At the same time, the behavioral decision correction sub-model fully considers the ordering preferences and operational characteristics of different terminals, dynamically adjusting the inventory safety boundary to make the inventory estimation results more in line with the individual operational needs of the terminal.
[0020] 4. It strengthens the proactive management and accurate prediction of inventory risks. The system integrates multi-dimensional auxiliary information such as the number of days without receiving goods and the number of days without visits. Through the rule engine, it realizes the automatic identification and graded early warning of risks such as stockouts and backlogs. Instead of passively dealing with inventory problems, it can predict potential risks in advance and trigger targeted decisions, helping the supply chain to achieve proactive intervention, optimize inventory turnover efficiency, and reduce resource waste and loss of sales opportunities.
[0021] 5. Improved supply chain collaboration efficiency and management convenience: The visual user interface transforms complex inventory data and risk information into intuitive and easy-to-understand charts and maps, enabling managers to have a global grasp of the overall inventory health status and the distribution of high-risk terminals, reducing the cost of information interpretation; at the same time, through standardized interfaces, it seamlessly connects with external ERP, CRM and other systems to achieve data exchange and rapid flow of decision-making instructions, building a complete collaborative closed loop from terminal inventory monitoring to upstream supply chain response, improving the operational efficiency and collaboration capabilities of the entire supply chain. Attached Figure Description
[0022] Figure 1 This is a schematic diagram of the framework of a digital inventory estimation system according to the present invention; Figure 2 This is a schematic diagram of the inventory estimation logic for the inventory estimation module. Detailed Implementation
[0023] Existing terminal inventory management technologies suffer from unreliable data, high reliance on manual operations, and low model accuracy. The core of this invention's technical solution is to use a data acquisition module to collect terminal inbound data, consumer bottle opening and scanning data, and salesperson visit data in parallel. After data cleaning, deduplication, geofencing filtering, and related structured processing, the inventory estimation module first calculates the net inbound quantity of full cases based on terminal inbound data, and then calculates the number of cases opened based on consumer bottle opening and scanning data. Next, it uses dynamically calculated consumer scanning rates to compensate and correct the number of cases opened to obtain the theoretical inventory. Finally, it dynamically adjusts the theoretical inventory to a safety boundary by fitting the terminal's loss aversion coefficient, outputting the real-time estimated inventory. Subsequently, the early warning and decision-making module automatically generates replenishment suggestions or visit tasks based on the safety inventory threshold, the number of days without restocking, and the number of days without visits, thus eliminating reliance on traditional outbound scanning data and achieving accurate real-time estimation of terminal inventory.
[0024] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.
[0025] like Figure 1As shown, the digital inventory estimation system of this invention includes a data source layer, a core processing layer, a data storage layer, and an application output layer at the data level. The core processing layer mainly includes a data processing server, in which data acquisition, data cleaning, inventory estimation, and early warning and decision-making modules are deployed. The data storage layer mainly includes a distributed database, and the application output layer mainly includes a visual user interface. The data processing server communicates and connects with the terminal sales system, consumer QR code scanning platform, and enterprise ERP / CRM system via API or IoT interface.
[0026] The task execution logic of each module in this system is explained below: The data acquisition module receives data from terminal scanning devices, consumer bottle opening and scanning data, and salesperson visit data. It employs message queue-based streaming data access and buffering technology to receive and sort raw data records in real time before transmitting them to the data cleaning module.
[0027] The data cleaning module is used to deduplicatize the data collected by the data acquisition module, filter invalid barcodes based on geofencing, and perform structured processing based on the association between terminals and products. The clean and standardized dataset is then output and transmitted to the inventory calculation module.
[0028] The inventory estimation module receives standard data processed by the data cleaning module and uses a multi-layer calculation model to estimate inventory.
[0029] Specifically, the inventory estimation module performs the following inventory estimation process: S1: Based on the terminal warehousing data and after deducting return / transfer records, obtain the net number of boxes received; S2: Based on the consumer's bottle opening and scanning data, calculate the observed number of boxes opened according to the correlation between bottle code and box code, and dynamically calculate the consumer scanning rate based on historical data. The number of boxes opened is then compensated and corrected based on the consumer scanning rate. Specifically, the corrected number of boxes opened is equal to the observed number of boxes opened divided by the scanning rate. S3: The theoretical inventory is obtained based on the net number of boxes received and the number of boxes opened after compensation and correction. Finally, the loss aversion coefficient is fitted by analyzing the historical order and sales fluctuations of the terminal, and the theoretical inventory is adjusted to a safety boundary based on the loss aversion coefficient to output the real-time estimated inventory.
[0030] The early warning and decision-making module first automatically classifies inventory status into normal, warning, and emergency based on preset safety stock upper and lower limits. Second, it runs spatiotemporal correlation analysis rules, combining "number of days without restocking" and "absolute inventory value" to determine stockout risk, or combining "number of days without visits" and "data volatility" to assess data credibility. Finally, it executes a decision triggering mechanism, automatically generating structured instructions when the spatiotemporal correlation analysis rules are met: generating a replenishment suggestion order, triggering a salesperson visit task, or marking it as requiring manual review.
[0031] Preferably, the system also includes a visual user interface, which receives decision results and real-time estimated inventory values, and constructs inventory trend data from these estimated values. The visual user interface uses a business intelligence chart engine and RESTful API service encapsulation technology: on the one hand, it transforms decision results and inventory trend data into interactive dashboards, trend charts, and terminal map views, intuitively displaying global inventory health, high-risk terminal distribution, and early warning statistics; on the other hand, it provides the inventory trend data and early warning information from the decision results to external systems in standardized JSON format via API, and supports on-demand data export services. The processed visual graphical interface and API data stream are ultimately output to terminal managers or upstream business systems, completing the decision support closed loop of the entire tool.
[0032] Example: The data acquisition module is responsible for receiving inbound data from liquor sales terminals, consumer bottle opening and scanning data, and salesperson visit data. For example, a liquor store's terminal code is A2311502226. When a case of 38-degree Guojiao liquor enters the store, its inventory management system scans the entire case. After scanning, a record containing the product code (corresponding to 38-degree Guojiao), terminal code (A2311502226), and inbound time (2023-11-18) is pushed to the inventory calculation module in real time. Simultaneously, when a consumer purchases the product at the liquor store and scans the QR code inside the bottle cap using WeChat to participate in brand activities, this bottle opening and scanning record, along with the scanning time and the phone's GPS location information, is also sent to this system by the brand's mini-program server. Furthermore, the liquor store's salesperson completed this week's in-store visits using the enterprise mobile CRM, and their check-in time was 2024-05-21, which was also recorded and uploaded.
[0033] The data acquisition module uses streaming data access technology based on Apache Kafka message queues to process this data. The system establishes independent Kafka topics for "whole case warehousing," "consumer bottle opening and barcode scanning," and "business visits." All external data is published to its respective topic in JSON format via the corresponding RESTful API interface. The data consumption service within the data acquisition module continuously monitors these topics to ensure that warehousing records from a wine store, bottle opening records from consumers, and visit records from sales personnel are received and buffered in real time in sequence, forming an ordered raw data stream that is then transmitted to the data cleaning module.
[0034] After receiving the raw data stream, the data cleaning module begins to perform the cleaning and standardization process: First, it performs uniqueness verification and deduplication. The system checks the consumer's bottle opening and scanning records, and compares them with the unique bottle code scanned within the most recent 24-hour window to ensure that the same code will not be counted repeatedly due to network retransmission.
[0035] Next, invalid QR code scans are filtered based on geofencing. The system calls the built-in map service to calculate the distance between the consumer's mobile phone location coordinates when scanning the QR code and the registered address coordinates of the liquor store. Assuming the calculated distance is 15 kilometers, which exceeds the 5-kilometer core consumer geofence radius preset by the system for this liquor store, the QR code record will be marked as "suspected out-of-area consumption" and temporarily isolated. It will not participate in the liquor store's real-time inventory calculation, but may be used for other analyses.
[0036] Finally, multi-source data association and structuring are performed: the system associates the case-by-case warehousing records of a liquor store with all valid bottle-opening and barcode scanning records verified by geofencing. By querying a pre-set "bottle code-case code" association table, each individual bottle code is assigned to its original case code at the time of manufacture. For example, a consumer scanning a bottle code may correspond to a case code. The system automatically aggregates all data according to the terminal and product dimensions of "liquor store - 38-degree Guojiao" and converts it into a standard format with a unified timestamp. The cleaned data is written to a standardized log table in a distributed database and triggers an event notification to the inventory calculation module.
[0037] The logic of the inventory estimation module is as follows: Figure 2 As shown: By reading all standardized data on "38-degree Guojiao" from a liquor store over the past month, the net number of cases received was first calculated. According to the formula "Net number of cases received = Number of cases received - Number of cases returned / transferred", and considering the actual situation of 2 cases received and no returns at this terminal, the net number of cases received was determined to be 2. Next, the number of cases opened was calculated. After deduplication by merging case codes, the total number of valid barcode scans and bottle return barcodes corresponded to 1 case opened. Finally, according to the formula "Terminal estimated inventory = Net number of cases received - Number of cases opened", the basic inventory value was obtained as 2 cases - 1 case = 1 case.
[0038] Considering the insufficient scanning rate of consumers after opening bottles in the industry, the system calls the preset consumer scanning rate for this area (valued at 0.9, derived from regional scanning probability statistics), activates the compensation algorithm to optimize the results, and the correction formula is "estimated inventory = net number of cases received - (number of cases opened / consumer scanning rate)". Substituting the data, the corrected inventory value is calculated to be 2 items - (1 item / 0.9) ≈ 0.89 items, which makes up for the estimation deviation caused by unscanned records.
[0039] Two key auxiliary parameters were calculated simultaneously: first, the time without receiving goods = current date - latest receiving date = 2025-04-03 - 2024-11-18 = 317 days; second, the time without visits = current date - last visit date = 2025-04-03 - 2024-05-21 = 317 days. The system combined these two parameters to assess data reliability. Because the time without visits far exceeded the 30-day threshold, the data confidence level was marked as "Medium," and the influencing factors were noted in the final results.
[0040] Finally, the system analyzed the historical order records of a liquor store and found that in the past six months, its pre-holiday order volume was significantly lower than the suggested volume given by the system based on sales trends, and it often experienced temporary stockouts after the holidays. The system determined that this retailer had a significant loss aversion characteristic, calculating its loss aversion coefficient to be 1.4, which means that the liquor store has a strong fear of accumulating inventory. Based on this, in subsequent decision-making, the system will dynamically adjust its safety stock boundary, for example, by raising its replenishment trigger line by 40% from the standard, to offset its subjective conservative tendency and guide it to maintain a safer inventory level.
[0041] The module's final output includes: terminal code A2311502226, terminal name a liquor store, product 38-degree Guojiao, real-time calculated inventory after correction of 0.89 units, no restocking time of 317 days, no visit time of 317 days, and data confidence level of "medium". This result is stored in the database and sent to the early warning and decision-making module.
[0042] After receiving the above results, the early warning and decision-making module loads the preset business rules for automatic judgment. First, it classifies the inventory status: it compares the inventory value of 0.89 units with the system's preset safety stock limit of 1 case for 38-degree Guojiao liquor, and determines that the safety stock status of the product is "critical".
[0043] Subsequently, the spatiotemporal correlation analysis rules are run. The spatiotemporal correlation analysis rule engine makes a judgment based on two dimensions: first, the absolute value of inventory is lower than the safety line; second, it combines auxiliary parameters to meet the high-risk rule of "inventory is lower than the safety stock and no purchase time exceeds 7 days and no visit time exceeds 30 days", triggering a double warning.
[0044] The system automatically generates two structured instructions for decision-making: First, it generates an emergency replenishment suggestion form, suggesting the product (38-degree Guojiao), quantity (2 cases), and corresponding supply warehouse, and sends it directly to the winery's order management system via API interface; Second, it creates a high-priority visit task in the CRM system, assigning it to the corresponding salesperson, with the task description marked "38-degree Guojiao inventory is critically low, no purchases and no visits for more than 300 days, on-site verification of inventory and promotion of replenishment are required."
[0045] The visual user interface outputs the above information in two ways. On the one hand, using a business intelligence charting engine, the winery's icon turns red and flashes on the map on the supply chain manager's monitoring screen. Clicking on its details shows that the inventory trend chart shows the inventory has continued to decline to 0.38 cases, while the "Warning Information" section clearly lists the statuses of "Inventory Shortage" and "Replenishment Order Generated".
[0046] On the other hand, through RESTful API services and a business intelligence charting engine, the icon of a liquor store on the monitoring screen of supply chain managers turned red and flashed. Clicking on the details page, the BI view clearly displayed core data such as the terminal's 38-degree Guojiao warehousing records, number of cases opened, estimated inventory, time without restocking, and time without visit. The inventory trend chart intuitively showed the continuous decline in inventory, and the "Warning Information" column clearly marked "Inventory Crisis, Double High Risk" and "Replenishment Order Generated". Through the API service, inventory crisis information and replenishment suggestions were pushed to the enterprise's ERP system and delivery scheduling system in real time in standardized JSON format, and the distribution center prioritized the preparation and delivery of goods. At the same time, the salesperson's mobile CRM App received a task push notification, guiding them to go to the terminal immediately to verify the inventory and promote replenishment.
Claims
1. A digital inventory estimation system, characterized in that, include: The data acquisition module is used to collect terminal inventory data, consumer bottle opening and scanning data, and salesperson visit data. The data cleaning module is used to deduplicatize the data collected by the data acquisition module, filter invalid barcode scans based on geofencing, and bind the barcode scan data to the box code data to which the bottle code belongs. The inventory estimation module performs the following steps on the data processed by the data cleaning module: S1: Based on the terminal warehousing data statistics and after deducting return and transfer records, obtain the net number of full cases received; S2: Calculate the number of boxes opened based on the correlation between bottle code and box code according to the consumer's bottle opening and scanning data, and compensate and correct the number of boxes opened based on the historical scanning rate; S3: Obtain the theoretical inventory by subtracting the compensated and corrected number of boxes from the net number of boxes received. S4: By analyzing the historical order and sales fluctuations of the terminal, fit the loss aversion coefficient, and adjust the theoretical inventory number according to the loss aversion coefficient to make a safety boundary, and output the real-time estimated inventory. The early warning and decision-making module is used to classify inventory status and judge risks based on preset safety stock thresholds, number of days without restocking and number of days without business visits, combined with real-time inventory calculations, and automatically generate replenishment suggestions, visit tasks or approval instructions.
2. The digital inventory estimation system according to claim 1, characterized in that, The system also includes a visual user interface that outputs inventory trends and decision results based on dashboards or API interfaces.
3. The digital inventory estimation system according to claim 1, characterized in that, The terminal warehousing data includes product code, terminal code, warehousing time, and box code information; the consumer bottle opening and scanning data includes bottle code, scanning time, and geographical location information; the salesperson visit data includes terminal code and check-in time.
4. The digital inventory estimation system according to claim 1, characterized in that, The data acquisition module collects data in the following way: it asynchronously streams terminal data, consumer bottle opening and scanning data, and salesperson visit data based on message queues.
5. A digital inventory estimation system according to claim 1, characterized in that, The cleaning module's invalid barcode scanning filtering based on geofencing specifically includes: calculating the distance between the consumer's geographical location information when scanning the barcode and the corresponding terminal registration address; if the distance exceeds the preset geofence radius, the barcode scanning data is marked as invalid and isolated, and will not be included in subsequent inventory calculations.
6. The digital inventory estimation system according to claim 1, characterized in that, The scanning rate is a statistical ratio based on historical bottle opening scanning data and actual consumption data at the regional or terminal level.
7. A digital inventory estimation system according to claim 1, characterized in that, When the real-time estimated inventory value is lower than the safety stock limit and the number of days without restocking exceeds the set first threshold, the early warning and decision-making module triggers an emergency replenishment suggestion; when the number of days without visits exceeds the set second threshold, the early warning and decision-making module triggers a salesperson visit task.
8. A digital inventory estimation system according to claim 2, characterized in that, The visual user interface is used to generate interactive dashboards, trend charts, and terminal map views, and provides inventory trend data and early warning information to external systems in a standardized JSON format via API.
9. A digital inventory estimation system according to claim 1, characterized in that, The system communicates and connects with terminal sales systems, consumer QR code scanning platforms, and enterprise ERP or CRM systems via API or IoT interfaces.
10. A digital inventory estimation system according to claim 1, characterized in that, The data cleaning module is also used to perform deduplication verification based on the unique bottle code within a set time window to ensure that the same barcode is not counted repeatedly.