CN-121979623-A - Intelligent sales terminal real-time transaction data processing system based on memory computing framework

CN121979623ACN 121979623 ACN121979623 ACN 121979623ACN-121979623-A

Abstract

The invention provides an intelligent sales terminal real-time transaction data processing system based on a memory computing framework, which relates to the technical field of intelligent terminals of the Internet of things, and comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for capturing an original transaction data stream in real time through an acquisition array formed by a code scanning module, a payment module and a member sensing module which are deployed at each intelligent sales terminal; and the calculation module is used for constructing a memory data association structure corresponding to the physical position of the acquisition array in the memory area based on the memory calculation frame, executing parallel transaction processing on the original transaction data stream carried by the memory data association structure, acquiring the time sequence data of the transaction frequency of each intelligent sales terminal in real time in the parallel transaction processing process, fitting the transaction frequency in the current time window through three Bezier curves, and calculating the inflection point of the Bezier curve. The invention improves the transaction processing efficiency of the intelligent sales terminal and the refinement level of retail operation.

Inventors

ZHU YUNFENG
PAN ZHILONG
CHEN FENGSHENG

Assignees

厦门市华美云海科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260407

Claims (10)

1. The intelligent sales terminal real-time transaction data processing system based on the memory computing framework is characterized by comprising: the acquisition module is used for capturing the original transaction data stream in real time through an acquisition array formed by the code scanning module, the payment module and the member sensing module which are deployed at each intelligent sales terminal; The calculation module is used for constructing a memory data association structure corresponding to the physical position of the acquisition array in a memory area based on a memory calculation frame, executing parallel transaction processing on an original transaction data stream carried by the memory data association structure, acquiring transaction frequency time sequence data of each intelligent sales terminal in real time in the parallel transaction processing process, fitting the transaction frequency in a current time window through a three-time Bezier curve, and calculating inflection points of the Bezier curve; The pushing module is used for synchronizing the processed transaction data to the online applet inventory service and the store background inventory service in real time, constructing a distributed inventory consistency state map based on memory mirror images, and pushing the transaction log to the store management service cluster in real time; The processing module is used for dynamically constructing a sales data stress tensor field in the memory based on the received transaction log by the store management service cluster, carrying out main direction analysis and main value calculation on the stress tensor field to obtain a compensation factor representing a main mode of sales fluctuation, wherein the compensation factor is used for correcting an inventory threshold value and a promotion triggering condition of the next period.
2. The memory computing framework-based intelligent sales terminal real-time transaction data processing system according to claim 1, wherein capturing the original transaction data stream in real time through an acquisition array of a code scanning module, a payment module and a member sensing module deployed at each intelligent sales terminal comprises: the code scanning module scans the commodity bar code and generates commodity identification data and transaction triggering signals, the payment module collects payment modes, payment accounts and transaction amount data, and the member sensing module reads member card information and generates member identification data and member account state data; And carrying out multi-source heterogeneous data fusion on commodity identification data, payment mode, payment account data, transaction amount data, member identification data and point account state data according to the time stamp of transaction occurrence to form original transaction data with a transaction unique identification.
3. The intelligent sales terminal real-time transaction data processing system based on memory computing framework of claim 2, wherein based on memory computing framework, a memory data association structure corresponding to the physical position of the acquisition array is constructed in the memory area, parallel transaction processing is executed on the original transaction data stream carried by the memory data association structure, during the parallel transaction processing, the transaction frequency time sequence data of each intelligent sales terminal are acquired in real time, the transaction frequency in the current time window is fitted through three Bezier curves, and the inflection point of the Bezier curve is calculated, comprising: According to the physical position identification of each intelligent sales terminal, corresponding first data objects, second data objects and third data objects are respectively distributed to the code scanning module, the payment module and the member sensing module of each terminal in a memory area, and three data objects generated in the same transaction are mutually linked through a data blood relationship to form a memory data association structure corresponding to the physical position of the acquisition array; Loading an original transaction data stream into the memory data association structure, simultaneously executing transaction processing on a plurality of transaction data objects by adopting a parallel computing framework, and recording the transaction occurrence times of each intelligent sales terminal in a continuous sliding time window in real time in the transaction processing process to generate transaction frequency time sequence data; and selecting four control points to construct a cubic Bezier curve for the time sequence data of the transaction frequency in the current time window to fit the change trend of the transaction frequency, and obtaining the inflection point of the curve by solving the point that the second derivative of the Bezier curve is zero.
4. The memory computing framework-based intelligent sales terminal real-time transaction data processing system of claim 3, wherein dynamically adjusting the point of sale magnification and the inventory locking priority according to the occurrence position and the direction of the inflection point to obtain the processed transaction data of the locked inventory identification, the point of sale variable and the settlement amount comprises: Analyzing inflection point information of a cubic Bezier curve in a current time window, wherein the inflection point information comprises the position of a time period where an inflection point is located in the time window and the concave-convex direction of the inflection point; according to a preset adjustment strategy, if the inflection point is positioned at the front half section of the time window and the inflection point direction is downwards convex, increasing the integration multiplying power of the subsequent transaction and improving the inventory locking priority, if the inflection point is positioned at the rear half section of the time window and the inflection point direction is upwards convex, reducing the integration multiplying power of the transaction and reducing the inventory locking priority, and if the inflection point is positioned at the front half section of the time window and the direction is upwards convex, the inflection point is positioned at the rear half section of the time window and the direction is downwards convex, and the inflection point is positioned at the middle position of the time window, maintaining the current integration multiplying power and the inventory locking priority unchanged, and obtaining the adjusted integration multiplying power; And calculating the integral variable of the current transaction based on the adjusted integral multiplying power, executing inventory locking operation according to the adjusted inventory locking priority, generating a locked inventory identifier, generating a settlement amount by combining the transaction amount, and finally outputting processed transaction data containing the locked inventory identifier, the integral variable and the settlement amount.
5. The intelligent sales terminal real-time transaction data processing system based on a memory computing framework of claim 4, wherein synchronizing processed transaction data in real-time to an online applet inventory service and a store backend inventory service, constructing a memory mirror based distributed inventory consistency state map while pushing transaction logs in real-time to a store management service cluster, comprises: the processed transaction data containing the locked inventory identification, the integral variable and the settlement amount are respectively pushed to a memory area of an online applet inventory service and a memory area of a store background inventory service through a data pipeline provided by a memory computing framework; The online applet inventory service and the store background inventory service execute atomic update on the corresponding commodity inventory quantity according to the locked inventory identification based on the processed transaction data pushed to the respective memory area, so as to form a memory mirror inventory state which is decoupled from the physical inventory and logically consistent with the physical inventory in the respective memory area; Establishing a bidirectional checking mechanism according to the comparison result, and triggering tracing correction operation based on the blood-edge relationship of memory data to construct distributed inventory consistency state mapping when the inconsistency of the states of the two is detected; based on the pushed processed transaction data, the complete execution process is packaged into a transaction log, and the transaction log is pushed to a store management service cluster in parallel through a real-time message bus of a memory computing framework.
6. The intelligent sales terminal real-time transaction data processing system based on a memory computing framework of claim 5, wherein the online applet inventory service and the store backend inventory service perform an atomic update on corresponding commodity inventory amounts based on the locked inventory identification based on the processed transaction data pushed to the respective memory regions, thereby forming a memory mirror inventory state in the respective memory regions that is decoupled from but logically consistent with the physical inventory, comprising: Analyzing the locked inventory identification in the processed transaction data, and respectively positioning corresponding commodity inventory records according to the memory area of the online applet inventory service and the memory area of the store background inventory service; Based on the located commodity inventory records, performing deduction on inventory quantity by adopting memory atom operation, and simultaneously updating version stamps of the inventory records to ensure data consistency under concurrent updating, so as to obtain updated inventory quantity and version stamps; And obtaining memory mirror image inventory states containing the current inventory snapshot according to the updated inventory quantity and version stamps, and taking the memory mirror image inventory states as the memory mirror image inventory states of the online applet inventory service and the store background inventory service.
7. The intelligent sales terminal real-time transaction data processing system based on a memory computing framework of claim 6, wherein the store management service cluster dynamically constructs a sales data stress tensor field in the memory based on the received transaction log, performs main direction analysis and main value calculation on the stress tensor field to obtain a compensation factor representing a main mode of sales fluctuation, and the compensation factor is used for correcting an inventory threshold and a promotion triggering condition of the next period, and comprises: The store management service cluster analyzes commodity identifications, transaction time stamps, settlement amounts and integral variables contained in each transaction log based on the pushed and received transaction logs, and builds a sales data stress tensor field with commodities as dimensions and time in a memory; analyzing the main direction of the stress tensor field of the sales data, and extracting the main direction characteristics of the fluctuation of the sales data by calculating the gradient and the divergence of the stress tensor field to obtain the main direction of the stress tensor field; Based on the obtained main direction, carrying out main value calculation on the stress tensor field, obtaining a compensation factor representing the main mode of sales fluctuation by solving the characteristic value of the stress tensor field, and applying the obtained compensation factor to correction of the inventory threshold value and the promotion triggering condition of the next period.
8. The memory computing framework-based intelligent sales terminal real-time transaction data processing system of claim 7, wherein the main value calculation is performed on the stress tensor field based on the obtained main direction, and the compensation factor representing the main mode of sales fluctuation is obtained by solving the characteristic value of the stress tensor field, and the obtained compensation factor is applied to the correction of the inventory threshold and the promotion triggering condition of the next period, and the method comprises the following steps: based on the obtained main direction, carrying out main value calculation on the stress tensor field, and obtaining an initial compensation factor representing a main sales fluctuation mode by solving the characteristic value of the stress tensor field; analyzing the value and positive and negative directions of the initial compensation factors, and generating an inventory correction coefficient for adjusting an inventory threshold value and a sales promotion correction coefficient for adjusting a sales promotion trigger condition according to the analysis result; The sales promotion correction coefficient is acted on the sales promotion trigger condition of the current period to obtain the sales promotion trigger condition of the next period after correction; And outputting the corrected inventory threshold value of the next period and the corrected sales promotion triggering condition of the next period to inventory management service and sales promotion management service for guiding the operation of the business of the next period.
9. A computing device, comprising: One or more processors; storage means for storing one or more programs that when executed by the one or more processors cause the one or more processors to perform the system of any of claims 1-8.
10. A computer readable storage medium, characterized in that the computer readable storage medium has stored therein a program which, when executed by a processor, performs the system according to any of claims 1 to 8.

Description

Intelligent sales terminal real-time transaction data processing system based on memory computing framework Technical Field The invention relates to the technical field of intelligent terminals of the Internet of things, in particular to an intelligent sales terminal real-time transaction data processing system based on a memory computing framework. Background Under the background of rapid development of retail informatization and intelligent terminals of the Internet of things, intelligent sales terminals are widely deployed in off-line retail stores, automatic processing of transaction transactions is realized by means of modules such as code scanning, payment and member sensing, most of retail enterprises currently adopt a traditional disk storage type transaction processing mode, high-efficiency memory computing support is lacked, more than 20 intelligent sales terminals are deployed in a chain fresh supermarket for example, commodity code scanning data, payment information and member point data acquired by the terminals are required to be written into the disk for storage first, transaction analysis and processing are carried out, delay is brought to on-line applet inventory and store background inventory, member point can be calculated according to fixed point multiplying power, inventory is locked according to unified priority, and cannot be flexibly adjusted according to store real-time fluctuation. Disclosure of Invention The invention provides an intelligent sales terminal real-time transaction data processing system based on a memory computing framework, which improves the transaction processing efficiency of the intelligent sales terminal and the refinement level of retail operation. In order to solve the technical problems, the technical scheme of the invention is as follows: in a first aspect, an intelligent sales terminal real-time transaction data processing system based on a memory computing framework includes: the acquisition module is used for capturing the original transaction data stream in real time through an acquisition array formed by the code scanning module, the payment module and the member sensing module which are deployed at each intelligent sales terminal; The calculation module is used for constructing a memory data association structure corresponding to the physical position of the acquisition array in a memory area based on a memory calculation frame, executing parallel transaction processing on an original transaction data stream carried by the memory data association structure, acquiring transaction frequency time sequence data of each intelligent sales terminal in real time in the parallel transaction processing process, fitting the transaction frequency in a current time window through a three-time Bezier curve, and calculating inflection points of the Bezier curve; The pushing module is used for synchronizing the processed transaction data to the online applet inventory service and the store background inventory service in real time, constructing a distributed inventory consistency state map based on memory mirror images, and pushing the transaction log to the store management service cluster in real time; The processing module is used for dynamically constructing a sales data stress tensor field in the memory based on the received transaction log by the store management service cluster, carrying out main direction analysis and main value calculation on the stress tensor field to obtain a compensation factor representing a main mode of sales fluctuation, wherein the compensation factor is used for correcting an inventory threshold value and a promotion triggering condition of the next period. Further, through disposing in the collection array that each intelligent sales terminal sweep sign indicating number module, payment module and member response module constitute, the real-time original transaction data stream of catching includes: the code scanning module scans the commodity bar code and generates commodity identification data and transaction triggering signals, the payment module collects payment modes, payment accounts and transaction amount data, and the member sensing module reads member card information and generates member identification data and member account state data; And carrying out multi-source heterogeneous data fusion on commodity identification data, payment mode, payment account data, transaction amount data, member identification data and point account state data according to the time stamp of transaction occurrence to form original transaction data with a transaction unique identification. Further, based on a memory computing frame, a memory data association structure corresponding to the physical position of the acquisition array is constructed in a memory area, parallel transaction processing is executed on an original transaction data stream carried by the memory data association structure, in the parallel transaction processing process, tra