A vending machine dynamic sales analysis and low-efficiency point identification method and system

Abstract

This invention discloses a method and system for dynamic sales analysis and inefficient location identification of unmanned vending machines. It addresses the problems of poor timeliness and weak adaptability in traditional fixed-period offline sales calculations, and the reliance on manual verification and lack of logical management for inefficient locations. This invention employs FlinkCDC to capture and broadcast changes to the business database configuration in real time, combined with Spark's offline batch retrieval of business transaction data. It calculates the cumulative dynamic sales of locations based on dynamic time window parameters and sales adjustment weights. Based on preset entry and exit rules and a whitelist mechanism, it achieves automated logical flow between the normal location pool and the inefficient location pool. This invention balances rule response timeliness with massive data processing capabilities, improves the accuracy of sales assessment, reduces manual maintenance, and supports refined operation of unmanned vending machines.

Description

Technical Field

[0001] This invention relates to the field of data processing and intelligent operation technology, specifically to a method and system for dynamic sales analysis and inefficient location identification of unmanned vending machines. Background Technology

[0002] In the unmanned retail sector, sales monitoring and efficiency evaluation of vending machine locations are central to refined operations, directly determining location optimization strategies and overall operational efficiency. Currently, the industry generally adopts a fixed-cycle offline batch calculation model, typically using natural weeks or months to calculate total sales at each location to assess operational performance. This traditional model has significant shortcomings: First, it lacks timeliness; changes to business rules, sales thresholds, and other configurations take effect only after the next offline cycle, failing to respond to business adjustments in real time. Second, it lacks flexibility; fixed cycles are difficult to adapt to dynamic scenarios such as holidays, seasonal fluctuations, and marketing activities, making sales evaluation results easily disconnected from actual operational conditions. Third, it lacks logical management; inefficient location identification relies on manual verification, lacking automated entry and exit rules and whitelist management mechanisms, resulting in cumbersome and inefficient location optimization processes that cannot support efficient and refined operations of unmanned vending machine locations. Summary of the Invention

[0003] To address these issues, this invention provides a method and system for dynamic sales analysis and inefficient location identification of unmanned vending machines. This system solves the technical problems of poor timeliness and weak adaptability of traditional fixed-period offline sales calculation, reliance on manual verification of inefficient locations, lack of automated entry and exit rules and whitelist logic management, and difficulty in adapting to dynamic business needs.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for dynamic sales analysis and inefficient location identification of unmanned vending machines, comprising:

[0005] S1. Use FlinkCDC technology to capture configuration change information in the business database in real time, and broadcast the configuration change information to the Spark distributed computing engine;

[0006] S2. The Spark distributed computing engine pulls daily real sales data from each location of the unmanned vending machine in batches from the business pipeline database.

[0007] S3. Based on the dynamic time window parameters of broadcast, calculate the cumulative dynamic sales of each location within the corresponding dynamic time window;

[0008] S4. Based on the calculated cumulative dynamic sales and the preset entry and exit rules, identify inefficient locations and implement the transfer management between the normal operation location pool and the inefficient location pool.

[0009] S5. Store the calculated cumulative dynamic sales data, inefficient location identification results, and location flow logs into the dynamic sales and inefficient location profile database.

[0010] As a preferred solution for dynamic sales analysis and inefficient location identification of unmanned vending machines, in S1, the configuration change information includes the dynamic time window parameter, inefficient sales threshold, high-quality sales threshold, observation period parameter, and whitelist information.

[0011] As a preferred solution for dynamic sales analysis and inefficient location identification of unmanned vending machines, in S2, the Spark distributed computing engine performs offline T+1 / hour-level aggregation of business transaction data.

[0012] As a preferred method for dynamic sales analysis and inefficient location identification of unmanned vending machines, in S3, the formula for calculating the cumulative dynamic sales is:

[0013]

[0014] In the formula, Let T represent the cumulative dynamic sales volume at point i, and T be the current calculation date. For dynamic time window parameters, Let i be the actual daily sales volume of point i on day t. Adjust the weight for sales volume on day t.

[0015] As a preferred solution for dynamic sales analysis and inefficient location identification of unmanned vending machines, the sales adjustment weight The time decay weight / holiday adjustment factor is used under normal circumstances. When FlinkCDC detects a special event rule configured in the background, it adjusts the dynamic time window parameters in real time. The value of .

[0016] As a preferred solution for dynamic sales analysis and inefficient location identification of unmanned vending machines, in S4, the jump-in rule is: if a location is not in the whitelist, and its dynamic sales for N consecutive days are all lower than the inefficient sales threshold. If this is triggered, the jump rule will be activated, and the condition function will be activated. The expression is:

[0017]

[0018] In the formula, For point In the Cumulative dynamic sales over the days For inefficient observation of consecutive days, For point The whitelist status indicator; when When this happens, the location is automatically transferred from the pool of normally operating locations to the pool of inefficient locations.

[0019] As a preferred solution for dynamic sales analysis and inefficient location identification of unmanned vending machines, in S4, the exit rule is: if a location is already in the inefficient location pool, and its dynamic sales for M consecutive days are not lower than the high-quality sales threshold. If this location is manually added to the whitelist, the aforementioned bounce rule will be triggered. Triggering conditions: The expression is:

[0020]

[0021] In the formula, For point In the Cumulative dynamic sales over the days To ensure a high-quality number of consecutive observation days, For point The whitelist status indicator; when When this happens, the location is automatically transferred from the inefficient location pool back to the normal operating location pool.

[0022] As a preferred solution for dynamic sales analysis and inefficient location identification of unmanned vending machines, FlinkCDC technology captures the configuration change information with millisecond-level latency.

[0023] As a preferred solution for dynamic sales analysis and inefficient location identification of unmanned vending machines, the dynamic time window parameter is scaled and adjusted in real time according to holidays, seasonal fluctuations / marketing activity scenarios.

[0024] This invention also provides a dynamic sales analysis and inefficient location identification system for unmanned vending machines, employing the aforementioned method for dynamic sales analysis and inefficient location identification for unmanned vending machines, comprising:

[0025] Real-time configuration capture unit: used to capture configuration change information in the business database in real time using FlinkCDC technology, and broadcast the configuration change information to the Spark distributed computing engine;

[0026] Offline data retrieval unit: used by the Spark distributed computing engine to retrieve daily real sales data of each location of the unmanned vending machine in batches from the business pipeline database;

[0027] Dynamic sales calculation unit: used to calculate the cumulative dynamic sales of each location within the corresponding dynamic time window based on the dynamic time window parameters of the broadcast;

[0028] Location pool management unit: used to identify inefficient locations and manage the flow between the normal operation location pool and the inefficient location pool based on the calculated cumulative dynamic sales and preset entry and exit rules.

[0029] Data storage unit: used to store the calculated cumulative dynamic sales data, inefficient location identification results and location flow logs into the dynamic sales and inefficient location profile library.

[0030] The present invention has the following advantages:

[0031] First, this invention adopts a real-time and offline dual architecture of FlinkCDC and Spark, which captures configuration changes in milliseconds and makes them effective in real time. At the same time, it efficiently processes massive historical streaming data, taking into account both the timeliness of rule response and the throughput of data computing.

[0032] Secondly, this invention abandons the fixed statistical period and calculates sales through dynamic time windows and sales adjustment weights, which can accurately adapt to dynamic business scenarios such as holidays, seasonal fluctuations and marketing activities, and improve the accuracy of sales assessment.

[0033] Third, this invention establishes an automated entry / exit rule and whitelist logic management mechanism to achieve automatic identification and transfer of inefficient locations, reduce the workload of manual review, improve the efficiency of location operation optimization, and support the refined operation of unmanned vending machines. Attached Figure Description

[0034] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0035] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0036] Figure 1 This is a schematic diagram of a method for dynamic sales analysis and inefficient location identification of unmanned vending machines provided in Embodiment 1 of the present invention;

[0037] Figure 2This is a technical roadmap for a method for dynamic sales analysis and inefficient location identification of unmanned vending machines provided in Embodiment 1 of the present invention;

[0038] Figure 3 The diagram shows the flow state of the inefficient location jump-in / jump-out rules in the dynamic sales analysis and inefficient location identification method for unmanned vending machines provided in Embodiment 1 of the present invention.

[0039] Figure 4 This is a schematic diagram of the architecture of a dynamic sales analysis and inefficient location identification system for unmanned vending machines provided in Embodiment 1 of the present invention. Detailed Implementation

[0040] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] Example 1

[0042] See Figure 1 , Figure 2 and Figure 3 This invention provides a method for dynamic sales analysis and inefficient location identification of unmanned vending machines, comprising:

[0043] S1. Use FlinkCDC technology to capture configuration change information in the business database in real time, and broadcast the configuration change information to the Spark distributed computing engine;

[0044] Specifically, FlinkCDC (Change Data Capture) subscribes to the binlog of the business database to capture data change operations such as INSERT, UPDATE, and DELETE in real time, eliminating the need for database polling and avoiding performance pressure on the business database. The captured configuration change information is distributed to all Spark compute nodes via Flink's broadcast streaming mechanism, ensuring that all compute nodes use the latest and consistent configuration, thus resolving configuration consistency issues in a distributed environment.

[0045] S2. The Spark distributed computing engine pulls daily real sales data from each location of the unmanned vending machine in batches from the business pipeline database.

[0046] Specifically, the business transaction database stores the transaction details of all vending machines, with the data volume increasing linearly with the number of locations and the number of transactions. The Spark distributed computing engine distributes the massive transaction data to multiple computing nodes for parallel retrieval and processing through a data sharding mechanism, significantly improving data processing throughput compared to single-machine computing and supporting batch data processing for more than 100,000 locations.

[0047] S3. Based on the dynamic time window parameters of broadcast, calculate the cumulative dynamic sales of each location within the corresponding dynamic time window;

[0048] Specifically, the dynamic time window abandons the limitations of the traditional fixed period (natural week / month), and the window size is dynamically determined by the parameters broadcast in real time. The calculation starts from the current date and accumulates the sales data for a specified number of days backward. It can flexibly adjust the evaluation period according to business needs and better fit the sales fluctuation patterns of actual operating scenarios.

[0049] S4. Based on the calculated cumulative dynamic sales and the preset entry and exit rules, identify inefficient locations and implement the transfer management between the normal operation location pool and the inefficient location pool.

[0050] Specifically, the automatic transition of site status is achieved through clearly defined mathematical triggering conditions, without the need for manual intervention. The normal operation site pool and the inefficient site pool are two logical data sets. The transfer of sites between the two sets is achieved by modifying the status tags of the sites, forming an automated site lifecycle management process.

[0051] S5. Store the calculated cumulative dynamic sales data, inefficient location identification results, and location flow logs into the dynamic sales and inefficient location profile database.

[0052] Specifically, the dynamic sales and inefficient location profile database is a dedicated database for storing calculation results and operational data, employing a columnar storage format to optimize query performance. The stored workflow logs fully record the status change time, triggering conditions, and operator for each location, providing data support for subsequent operational performance tracking and rule optimization.

[0053] In one possible embodiment, in S1, the configuration change information includes the dynamic time window parameter, inefficient sales threshold, high-quality sales threshold, observation period parameter, and whitelist information.

[0054] Specifically, the dynamic time window parameter specifies the number of days to backtrack for sales calculation and is a key parameter for dynamic sales calculation; the inefficient sales threshold is the minimum sales standard for determining whether a location enters the inefficient pool; locations below this threshold may have operational issues; the high-quality sales threshold is the minimum sales standard for determining whether an inefficient location has returned to normal; reaching this threshold indicates an improvement in the location's operational status; the observation period parameter includes an inefficient observation period N and a high-quality observation period M, used to avoid misjudgments caused by daily sales fluctuations and ensure the stability of status changes; whitelist information stores identifiers for special locations. These locations will not be judged as inefficient even if their sales are below the threshold, applicable to strategically located locations, newly launched locations, and other special cases. All configuration information is uniformly stored in the business database for centralized management and modification.

[0055] In one possible embodiment, in S2, the Spark distributed computing engine performs offline T+1 / hourly aggregation on the business transaction data.

[0056] Specifically, T+1 aggregation: Retrieves all transaction data from the previous day in batches every morning at midnight, calculating the daily sales and cumulative dynamic sales for the previous day. This is suitable for regular operational scenarios, balancing computational costs and data timeliness. Hourly aggregation: Retrieves transaction data from the most recent hour every hour, updating the real-time sales and cumulative dynamic sales for the current day. This is suitable for scenarios requiring high-frequency sales monitoring, such as major promotional events. Spark achieves data fault tolerance and efficient computation through the Resilient Distributed Dataset (RDD) mechanism, ensuring the stable operation of batch aggregation tasks.

[0057] In one possible embodiment, in S3, the formula for calculating the cumulative dynamic sales volume is:

[0058]

[0059] In the formula, Let T represent the cumulative dynamic sales volume at point i, and T be the current calculation date. For dynamic time window parameters, Let i be the actual daily sales volume of point i on day t. Adjust the weight for sales volume on day t.

[0060] Specifically, the formula uses a weighted summation method to calculate the cumulative sales within a dynamic time window, which can reflect the degree of impact of sales on the current evaluation results on different dates; Indicates the start date of the dynamic time window, with a window length of [value missing]. sky; It is the actual number of transactions at point i on day t, aggregated from the business flow database, excluding invalid transactions such as canceled orders and refunds; It is a sales adjustment weight, used to correct for sales differences on different days, so that the calculation results can better reflect the actual operational capabilities of the location.

[0061] In one possible embodiment, the sales adjustment weight The time decay weight / holiday adjustment factor is used under normal circumstances. When FlinkCDC detects a special event rule configured in the background, it adjusts the dynamic time window parameters in real time. The value of .

[0062] Specifically, time decay weighting: sales data closer to the current date has a higher weight, while data from further back has a lower weight, reflecting sales trends and avoiding excessive influence of historical data on current evaluation results; Holiday adjustment factor: Sales during holidays usually differ significantly from weekdays. Setting different adjustment factors can eliminate the impact of holidays on sales evaluation, for example, multiplying holiday sales by a factor of 0.8 to make them comparable to weekday sales; Under normal circumstances... This means that sales volume has the same weight across all dates. When special event rules are configured in the backend (such as major promotional events, epidemic control measures, etc.), FlinkCDC captures the configuration change in real time and broadcasts it. The Spark computing engine immediately uses the new weight parameters for calculation without restarting the computing task.

[0063] In one possible embodiment, in S4, the jump-in rule is: if a certain location is not in the whitelist, and its dynamic sales for N consecutive days are all lower than the inefficient sales threshold. If this is triggered, the jump rule will be activated, and the condition function will be activated. The expression is:

[0064]

[0065] In the formula, For point In the Cumulative dynamic sales over the days For inefficient observation of consecutive days, For point The whitelist status indicator; when When this happens, the location is automatically transferred from the pool of normally operating locations to the pool of inefficient locations.

[0066] Specifically, the jump rule adopts the judgment logic of "N consecutive days", which can effectively filter out misjudgments caused by abnormal fluctuations in daily sales (such as equipment failure, temporary stockouts, etc.). This indicates that the dynamic sales volume for the consecutive N days preceding the current date T is checked, and the threshold is only applied if the sales volume for all N days is below a certain threshold. The jump is only triggered at certain times; This indicates that locations within the whitelist should be excluded to avoid misjudging specific locations; trigger condition function. It is a binary function; an output of 1 indicates that the jump-in condition is met, and an output of 0 indicates that it is not met. When When this happens, the system automatically changes the status label of the location to "inefficient" and adds it to the inefficient location pool.

[0067] In one possible embodiment, in S4, the exit rule is: if a certain location is already in the inefficient location pool, and its dynamic sales for M consecutive days are not lower than the high-quality sales threshold. If this location is manually added to the whitelist, the aforementioned bounce rule will be triggered. Triggering conditions: The expression is:

[0068]

[0069] In the formula, For point In the Cumulative dynamic sales over the days To ensure a high-quality number of consecutive observation days, For point The whitelist status indicator; when When this happens, the location is automatically transferred from the inefficient location pool back to the normal operating location pool.

[0070] Specifically, the exit rules are set with two parallel trigger conditions, and meeting either one will allow the user to exit the inefficient pool; the first condition is that the dynamic sales volume is not lower than the quality threshold for M consecutive days. The first condition is that the sales volume of the location has returned to normal levels after the operational adjustments; the second condition is that the location has been manually added to the whitelist, which applies to strategic locations that need to be retained or temporary adjustments. This represents a logical OR operation, where either of the two conditions must be met; it triggers a conditional function. An output of 1 indicates that the exit condition is met. At this time, the system will automatically change the status label of the point to "normal" and remove it from the inefficient point pool.

[0071] In one possible implementation, FlinkCDC technology captures the configuration change information with a latency of milliseconds.

[0072] Specifically, FlinkCDC captures changes based on the database's transaction logs. When configuration data in the business database changes, the database immediately generates binlog logs. FlinkCDC reads the binlog logs in real time via streaming, parses the changed content, and converts it into structured data. The latency of the entire process is typically between tens and hundreds of milliseconds. Compared to the traditional polling database method (which usually has a latency of minutes), FlinkCDC can achieve near real-time effectiveness of configuration changes, ensuring timely adjustments to business rules.

[0073] In one possible embodiment, the dynamic time window parameter is scaled and adjusted in real time according to holidays, seasonal fluctuations / marketing campaign scenarios.

[0074] Specifically, the sales fluctuation cycle varies significantly across different business scenarios: sales fluctuate greatly during holidays, making a shorter time window (e.g., 3 days) suitable for evaluation; seasonal fluctuations (e.g., increased beverage sales in summer) have longer cycles, making a longer time window (e.g., 14 days) suitable for evaluation; during marketing campaigns, sales may experience short-term spikes, requiring adjustments to the time window to accurately assess campaign effectiveness; operations personnel can modify dynamic time window parameters through the backend configuration system, and FlinkCDC captures these changes in real time and broadcasts them to all computing nodes, allowing subsequent sales calculations to immediately use the new window parameters without waiting for the fixed period to end.

[0075] The application scenarios of this invention are as follows:

[0076] Daily operation monitoring scenarios for routine vending machine locations: Applicable to the daily operation and management of vending machines in locations such as office buildings, industrial parks, residential communities, university campuses, and transportation hubs; compatible with various vending machine products such as food and beverages, daily necessities, and snacks; The system uses FlinkCDC to synchronize operation configuration in real time, calculates dynamic sales at each location in batches through Spark, and automatically identifies inefficient locations based on jump-in / jump-out rules for continuous observation periods, synchronously updating the location status to the dynamic sales and inefficient location profile database, meeting the needs of daily sales monitoring, real-time perception of operational status, and automatic discovery of inefficient locations for large-scale vending machine networks.

[0077] Dynamic business fluctuation adaptation scenarios for unmanned vending machines: Suitable for operational scenarios with large sales fluctuations, such as peak customer traffic during holidays, seasonal transitions, large-scale marketing activities, and sudden public events; adaptable to various products whose sales fluctuate due to external factors, such as seasonal products, promotional products, and emergency supplies; the system adjusts dynamic time window parameters and sales adjustment weights in real time, dynamically scales the sales evaluation cycle, corrects sales deviations during holidays and event periods, and automatically updates inefficiency judgment thresholds to meet the needs of accurate assessment of the actual operational capabilities of locations and avoid misjudging inefficient locations in scenarios with fluctuating sales.

[0078] Rapid evaluation of new vending machine locations: Applicable to the potential assessment and decision-making process of newly deployed and trial-operation vending machine locations; adaptable to various newly laid out areas, new business districts, and new scenarios; the system flexibly configures short-cycle dynamic evaluation windows, tracks the sales growth trend of new locations in real time, dynamically adjusts inefficient judgment criteria, and automatically generates location operation evaluation reports, meeting the needs of rapid trial and error, accurate potential assessment, shortened decision-making cycle, and reduced trial and error costs for new locations.

[0079] Differentiated management scenarios for special strategic locations of unmanned vending machines: Applicable to the operation and management of special locations such as remote pilot locations, brand display locations, public service locations, and strategic layout locations; adaptable to various unmanned vending machine locations that undertake non-profit functions or have long-term strategic value; the system exempts inefficiency judgments through a whitelist mechanism, monitors the operation data of special locations in real time, configures evaluation rules and thresholds separately, retains the ability to trace the operation data of locations, and meets the needs of balancing operational efficiency and strategic layout, differentiated management of special locations, and avoiding the accidental removal of strategic locations.

[0080] Large-scale network optimization decision-making scenarios for unmanned vending machines: Applicable to the site optimization, equipment scheduling, and resource allocation of operating companies with a network of more than a thousand unmanned vending machines; adaptable to unmanned vending machine site networks across all regions, categories, and scenarios; the system automatically collects all inefficient sites and generates an optimization list, statistically analyzes the operational efficiency of sites in each region, compares revenue data of different site types, and assists in formulating replenishment, relocation, and withdrawal strategies to meet the needs of efficient optimization of large-scale site networks, precise resource allocation, and improved overall operational efficiency.

[0081] It should be noted that the method of this embodiment can be executed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method of this embodiment, and these multiple devices will interact with each other to complete the described method for dynamic sales analysis and inefficient location identification of unmanned vending machines.

[0082] It should be noted that the above description describes some embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0083] Example 2

[0084] See Figure 4 Embodiment 2 of the present invention also provides a dynamic sales analysis and inefficient location identification system for unmanned vending machines, which adopts the above-mentioned dynamic sales analysis and inefficient location identification method for unmanned vending machines, including:

[0085] Real-time configuration capture unit 100: used to capture configuration change information in the business database in real time using FlinkCDC technology, and broadcast the configuration change information to the Spark distributed computing engine;

[0086] Offline data retrieval unit 200: used by the Spark distributed computing engine to retrieve daily real sales data of each location of the unmanned vending machine in batches from the business pipeline database;

[0087] Dynamic sales calculation unit 300: used to calculate the cumulative dynamic sales of each location within the corresponding dynamic time window based on the dynamic time window parameters of broadcast;

[0088] Location pool management unit 400: Used to identify inefficient locations and manage the flow between the normal operation location pool and the inefficient location pool based on the calculated cumulative dynamic sales and preset entry and exit rules.

[0089] Data storage unit 500: used to store the calculated cumulative dynamic sales data, inefficient location identification results and location flow logs into the dynamic sales and inefficient location profile library.

[0090] It should be noted that the information interaction and execution process between the modules of the above-mentioned device are based on the same concept as the method embodiment in Embodiment 1 of this application, and the resulting technical effects are the same as those in the method embodiment of this application. For details, please refer to the description in the method embodiment shown above in this application, and it will not be repeated here.

[0091] Example 3

[0092] Embodiment 3 of the present invention provides a non-transitory computer-readable storage medium storing program code for a method of dynamic sales analysis and inefficient location identification of unmanned vending machines. The program code includes instructions for executing the method of dynamic sales analysis and inefficient location identification of unmanned vending machines according to Embodiment 1 or any possible implementation thereof.

[0093] Computer-readable storage media can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

[0094] Example 4

[0095] Embodiment 4 of the present invention provides an electronic device, including: a memory and a processor;

[0096] The processor and the memory communicate with each other via a bus; the memory stores program instructions that can be executed by the processor, and the processor can call the program instructions to execute a method for dynamic sales analysis and inefficient location identification of unmanned vending machines according to Embodiment 1 or any possible implementation thereof.

[0097] Specifically, a processor can be implemented in hardware or software. When implemented in hardware, the processor can be a logic circuit, an integrated circuit, etc. When implemented in software, the processor can be a general-purpose processor that reads software code stored in memory. This memory can be integrated into the processor or located outside the processor and exist independently.

[0098] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0099] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0100] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A method for dynamic sales analysis and inefficient location identification of unmanned vending machines, characterized in that, include: S1. Use FlinkCDC technology to capture configuration change information in the business database in real time, and broadcast the configuration change information to the Spark distributed computing engine; S2. The Spark distributed computing engine pulls daily real sales data from each location of the unmanned vending machine in batches from the business pipeline database. S3. Based on the dynamic time window parameters of broadcast, calculate the cumulative dynamic sales of each location within the corresponding dynamic time window; S4. Based on the calculated cumulative dynamic sales and the preset entry and exit rules, identify inefficient locations and implement the transfer management between the normal operation location pool and the inefficient location pool. S5. Store the calculated cumulative dynamic sales data, inefficient location identification results, and location flow logs into the dynamic sales and inefficient location profile database.

2. The method for dynamic sales analysis and inefficient location identification of unmanned vending machines according to claim 1, characterized in that, In S1, the configuration change information includes the dynamic time window parameter, inefficient sales threshold, high-quality sales threshold, observation period parameter, and whitelist information.

3. The method for dynamic sales analysis and inefficient location identification of unmanned vending machines according to claim 1, characterized in that, In S2, the Spark distributed computing engine performs offline T+1 / hourly aggregation on business transaction data.

4. The method for dynamic sales analysis and inefficient location identification of unmanned vending machines according to claim 1, characterized in that, In S3, the formula for calculating the cumulative dynamic sales volume is: In the formula, Let T represent the cumulative dynamic sales volume at point i, and T be the current calculation date. For dynamic time window parameters, Let i be the actual daily sales volume of point i on day t. Adjust the weight for sales volume on day t.

5. The method for dynamic sales analysis and inefficient location identification of unmanned vending machines according to claim 4, characterized in that, Sales adjustment weight The time decay weight / holiday adjustment factor is used under normal circumstances. When FlinkCDC detects a special event rule configured in the background, it adjusts the dynamic time window parameters in real time. The value of .

6. The method for dynamic sales analysis and inefficient location identification of unmanned vending machines according to claim 1, characterized in that, In S4, the jump-in rule is: if a certain location is not in the whitelist, and its dynamic sales for N consecutive days are all lower than the inefficient sales threshold. If this is triggered, the jump rule will be activated, and the condition function will be activated. The expression is: In the formula, For point In the Cumulative dynamic sales over the days For inefficient observation of consecutive days, For point The whitelist status indicator; when When this happens, the location is automatically transferred from the pool of normally operating locations to the pool of inefficient locations.

7. The method for dynamic sales analysis and inefficient location identification of unmanned vending machines according to claim 1, characterized in that, In S4, the exit rule is: if a certain location is already in the inefficient location pool, and its dynamic sales for M consecutive days are not lower than the high-quality sales threshold. If this location is manually added to the whitelist, the aforementioned bounce rule will be triggered. Triggering conditions: The expression is: In the formula, For point In the Cumulative dynamic sales over the days To ensure a high-quality number of consecutive observation days, For point The whitelist status indicator; when When this happens, the location is automatically transferred from the inefficient location pool back to the normal operating location pool.

8. The method for dynamic sales analysis and inefficient location identification of unmanned vending machines according to claim 1, characterized in that, FlinkCDC technology captures the configuration change information with millisecond-level latency.

9. The method for dynamic sales analysis and inefficient location identification of unmanned vending machines according to claim 1, characterized in that, The dynamic time window parameter is scaled and adjusted in real time according to holidays, seasonal fluctuations, and marketing campaign scenarios.

10. A dynamic sales analysis and inefficient location identification system for unmanned vending machines, employing the dynamic sales analysis and inefficient location identification method for unmanned vending machines as described in any one of claims 1-9, characterized in that, include: Real-time configuration capture unit: used to capture configuration change information in the business database in real time using FlinkCDC technology, and broadcast the configuration change information to the Spark distributed computing engine; Offline data retrieval unit: used by the Spark distributed computing engine to retrieve daily real sales data of each location of the unmanned vending machine in batches from the business pipeline database; Dynamic sales calculation unit: used to calculate the cumulative dynamic sales of each location within the corresponding dynamic time window based on the dynamic time window parameters of the broadcast; Location pool management unit: used to identify inefficient locations and manage the flow between the normal operation location pool and the inefficient location pool based on the calculated cumulative dynamic sales and preset entry and exit rules. Data storage unit: used to store the calculated cumulative dynamic sales data, inefficient location identification results and location flow logs into the dynamic sales and inefficient location profile library.

Images