AI intelligent analysis method and system suitable for service data

Abstract

The invention relates to an AI intelligent analysis method and system suitable for business data, and the method comprises the steps: obtaining a historical node set, obtaining an initial product type set, obtaining a plurality of product node sets based on the historical node set and a target product type set, obtaining a plurality of region value sets based on the product node sets, and obtaining the monitoring time in real time, acquiring a prediction time and a verification time, when the monitoring time reaches the prediction time, acquiring a prediction value set based on the plurality of region value sets, acquiring an initial value set based on the initial value set, and confirming a calling-in value node set and a calling-out value node set based on the prediction value set and the initial value set; and obtaining a scheduled product system based on the scheduling instruction and the initialized product system, and confirming the scheduled product system as a target product system. According to the invention, the flexibility and efficiency of intelligent analysis of the business data can be improved.

Description

Technical Field

[0001] The present invention relates to the fields of data science and big data processing, and in particular to an AI intelligent analysis method and system applicable to business data. Background Art

[0002] The growth of the e-commerce industry has necessitated the analysis of vast quantities of product-related data to facilitate proactive product scheduling. In today's digital age, vast amounts of business data hold a wealth of information and potential value. For example, e-commerce platforms accumulate vast amounts of product sales data and supply chain information. In-depth analysis of this data enables precision marketing, inventory optimization, and demand forecasting. Furthermore, intelligent analysis can predict product demand, optimize inventory management, and reduce operating costs, thereby creating greater commercial value for enterprises.

[0003] Currently, the e-commerce industry primarily relies on traditional statistical analysis and simple machine learning models for business data analysis. While these methods can provide some data support, they also have significant limitations. For example, traditional methods are inefficient and prone to missing important information. Furthermore, they lack real-time analysis and dynamic adjustment capabilities, making them difficult to adapt to the rapidly changing business environment of the e-commerce industry. Consequently, analytical results lack timeliness and accuracy, failing to meet the industry's demand for precise decision-making.

[0004] While traditional methods can generally achieve intelligent analysis of business data, they lack flexibility. To achieve intelligent analysis of product data during scheduling, the present invention combines historical data with real-time monitoring data to provide an efficient and accurate intelligent analysis method. Furthermore, the present invention possesses a high degree of flexibility and adaptability, enabling flexible selection of scheduling solutions based on different business scenarios, improving product response speed during supply, and enhancing market competitiveness. Therefore, the flexibility and efficiency of intelligent analysis of business data needs to be improved. Summary of the Invention

[0005] The present invention provides an AI intelligent analysis method and system applicable to business data, the main purpose of which is to improve the flexibility and efficiency of intelligent analysis of business data.

[0006] To achieve the above objectives, the present invention provides an AI intelligent analysis method applicable to business data, comprising:

[0007] Receiving an AI intelligent analysis instruction, and confirming to initialize a product system based on the AI ​​intelligent analysis instruction;

[0008] Obtain a historical node set, wherein the historical node set includes multiple historical nodes, and each historical node includes a historical value, a historical product type, a historical position, a historical year, and a historical month;

[0009] Obtaining an initial product type set, wherein the initial product type set includes multiple initial product types, and obtaining a target product type set based on the historical node set and the initial product type set;

[0010] Based on the historical node set and the target product type set, multiple product node sets are obtained, and the following operations are performed on each of the multiple product node sets:

[0011] Based on the product node set, multiple regional value sets are obtained, wherein each of the multiple regional value sets contains multiple historical nodes, monitoring time is obtained in real time, and prediction time and verification time are obtained. When the monitoring time reaches the prediction time, a prediction value set is obtained based on the multiple regional value sets, wherein the number of prediction values ​​in the prediction value set is the same as the number of regional value sets in the multiple regional value sets;

[0012] Obtain a current value set, obtain an initial value set based on the current value set, confirm an incoming value node set and an outgoing value node set based on the predicted value set and the initial value set, confirm a scheduling instruction based on the incoming value node set, the outgoing value node set, and a pre-built storage center, and obtain a scheduled product system based on the scheduling instruction and the initialized product system, where the scheduling instruction is 0 or 1;

[0013] When the monitoring time reaches the verification time, the actual value set is obtained, where the actual value set contains multiple actual values. It is judged whether the predicted value set and the actual value set meet the pre-built evaluation conditions. After confirming that the predicted value set corresponding to each product node set in the multiple product node sets meets the evaluation conditions, the scheduled product system is confirmed as the target product system to realize intelligent analysis of business data.

[0014] On the other hand, the present invention also provides an AI intelligent analysis system applicable to business data, comprising:

[0015] An initialization module is configured to receive an AI intelligent analysis instruction and determine an initialization product system based on the AI ​​intelligent analysis instruction;

[0016] A historical node processing module is used to obtain a historical node set, wherein the historical node set includes multiple historical nodes, and each historical node includes a historical value, a historical product type, a historical position, a historical year, and a historical month;

[0017] Obtaining an initial product type set, wherein the initial product type set includes multiple initial product types, and obtaining a target product type set based on the historical node set and the initial product type set;

[0018] The prediction and scheduling module is used to obtain multiple product node sets based on the historical node set and the target product type set, and perform the following operations on each of the multiple product node sets:

[0019] Based on the product node set, multiple regional value sets are obtained, wherein each of the multiple regional value sets contains multiple historical nodes, monitoring time is obtained in real time, and prediction time and verification time are obtained. When the monitoring time reaches the prediction time, a prediction value set is obtained based on the multiple regional value sets, wherein the number of prediction values ​​in the prediction value set is the same as the number of regional value sets in the multiple regional value sets;

[0020] Obtain a current value set, obtain an initial value set based on the current value set, confirm an incoming value node set and an outgoing value node set based on the predicted value set and the initial value set, confirm a scheduling instruction based on the incoming value node set, the outgoing value node set, and a pre-built storage center, and obtain a scheduled product system based on the scheduling instruction and the initialized product system, where the scheduling instruction is 0 or 1;

[0021] The prediction result verification module is used to obtain the actual value set when the monitoring time reaches the verification time, where the actual value set contains multiple actual values, judge whether the predicted value set and the actual value set meet the pre-built evaluation conditions, and confirm that the predicted value set corresponding to each product node set in multiple product node sets meets the evaluation conditions. After that, the scheduled product system is confirmed as the target product system to realize intelligent analysis of business data.

[0022] The beneficial effects of the present invention are as follows:

[0023] The present invention receives AI intelligent analysis instructions, confirms the initialization product system based on the AI ​​intelligent analysis instructions, and obtains a historical node set. Therefore, by counting historical product types and retrieving target product types from the initial product type set, the present invention can accurately extract product types related to historical nodes, form a target product type set, and improve the pertinence and accuracy of subsequent analysis. At the same time, the present invention can effectively classify and summarize historical nodes according to target product types, facilitating subsequent targeted analysis of data of different product types and improving analysis efficiency.

[0024] Furthermore, the present invention obtains predicted values ​​by obtaining the current month and monitoring node set, providing a scientific basis for subsequent scheduling decisions, improving the accuracy and reliability of the prediction, and judging whether the predicted value set and the actual value set meet the pre-built evaluation conditions. After confirming that the predicted value set corresponding to each product node set in multiple product node sets meets the evaluation conditions, the scheduled product system is confirmed as the target product system to realize intelligent analysis of business data. It can be seen that the present invention can generate the first scheduling plan and the second scheduling plan respectively according to different scheduling instructions, and can flexibly select the scheduling plan according to different situations to ensure the adaptability and effectiveness of the scheduling plan. At the same time, by updating the initialized product system, accurate data support is provided for subsequent business operations, which can improve the flexibility and efficiency of intelligent analysis of business data. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A flowchart of an AI intelligent analysis method for business data provided by one embodiment of the present invention;

[0026] Figure 2 This is a functional module diagram of an AI intelligent analysis system for business data provided by one embodiment of the present invention;

[0027] Figure 3 A schematic diagram of the structure of an electronic device for implementing the AI ​​intelligent analysis method applicable to business data provided by one embodiment of the present invention.

[0028] Description of reference numerals:

[0029] 1. Electronic device; 10. Processor; 11. Storage; 12. Bus.

[0030] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0031] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0032] Example 1:

[0033] This embodiment provides an AI intelligent analysis method for business data. The execution subject of the AI ​​intelligent analysis method for business data includes, but is not limited to, at least one of electronic devices such as a server and a terminal that can be configured to execute the method provided in the embodiment of this application. In other words, the AI ​​intelligent analysis method for business data can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster.

[0034] Reference Figure 1 FIG. 1 is a flow chart of an AI intelligent analysis method for business data provided by an embodiment of the present invention. In this embodiment, the AI ​​intelligent analysis method for business data includes:

[0035] S1. Receive an AI intelligent analysis instruction, and confirm an initialization product system based on the AI ​​intelligent analysis instruction.

[0036] Among them, the AI ​​intelligent analysis instruction is an instruction issued by a person who performs intelligent analysis on business data. Due to the development of the e-commerce industry, a large amount of product-related data needs to be analyzed in order to facilitate the scheduling of products in advance. Exemplarily, Xiao Zhang, as the person responsible for analyzing business data, issues the AI ​​intelligent analysis instruction in order to realize the intelligent analysis of business data of products before scheduling, thereby improving the response rate of products during supply. Initializing the product system refers to a basic product system framework constructed according to the initial configuration before performing business data analysis and scheduling, which is used to receive subsequent scheduling instructions and perform corresponding update operations. Here, the initial product system is a system for storing and updating information, and can give scheduling suggestions in combination with historical data. Optionally, the initial product system is an APP or a mini program.

[0037] S2. Obtain a historical node set, wherein the historical node set includes multiple historical nodes, and each historical node includes a historical value, a historical product type, a historical position, a historical year, and a historical month.

[0038] For example, a certain brand of electronic products on the market usually launches multiple versions according to different positioning, functions and target users. The multiple versions include but are not limited to standard version, youth version and student version. In order to improve the inventory supply speed of multiple versions of electronic products, the sales records of the electronic products over the years are collected. For example: taking the standard version of the electronic product as an example, the sales of the standard version, youth version and student version in each month from 2021 to 2024 are collected to obtain a historical node set. Here, only the data of January of each year is taken as an example: {(100-standard version-A-2021-01), (95-standard version-A-2022-01), (93-standard version-A-2023-01), (91-standard version-A- 2024-01), (20-Youth Edition-A-2021-01), (15-Youth Edition-A-2022-01), (18-Youth Edition-A-2023-01), (17-Youth Edition-A-2024-01), (3-Youth Edition-A-2021-01), (4-Youth Edition-A-2022-01), (0-Youth Edition-A-2023-01), (3-Youth Edition-A-2024-01), (12-Student Edition-A-2021-01), (11-Student Edition-A-2022-01), (11-Student Edition-A-2023-01), (10-Student Edition-A-2024-01), (68-Standard Edition -B-2021-01), (67-Standard Edition-B-2022-01), (70-Standard Edition-B-2023-01), (71-Standard Edition-B-2024-01), (60-Youth Edition-B-2021-01), (63-Youth Edition-B-2022-01), (63-Youth Edition-B-2023-01), (65-Youth Edition-B-2024-01), (10-Youth Edition-B-2021-01), (9-Youth Edition-B-2022-01), (9-Youth Edition-B-2023-01), (12-Youth Edition-B-2024-01), (20-Student Edition-B-2021-01), (2 3-Student Edition-B-2022-01), (19-Student Edition-B-2023-01), (22-Student Edition-B-2024-01)}, here we only take the historical node (100-Standard Edition-A-2021-01) as an example, where (100-Standard Edition-A-2021-01) means that the sales volume of the standard version of the electronic product in store A in January 2021 was 100 pieces, where 100, Standard Edition, A, 2021 and 01 respectively represent the historical data, historical product type, historical location, historical year and historical month in the historical node. The remaining historical nodes can achieve the same effects as (100-Standard Edition-A-2021-01), which will not be repeated here.

[0039] S3. Obtain an initial product type set, wherein the initial product type set includes multiple initial product types, and obtain a target product type set based on the historical node set and the initial product type set.

[0040] Specifically, obtaining the target product type set based on the historical node set and the initial product type set includes:

[0041] Counting historical product types in the historical node set to obtain multiple historical product types, sequentially extracting historical product types from the multiple historical product types, searching the initial product type set using the historical product types, and if the initial product type is retrieved from the initial product type set, confirming the initial product type as the target product type, wherein the historical product type is the same as the initial product type;

[0042] Summarize the target product types to obtain a target product type set.

[0043] For example, assume that the historical node set is {(100-Standard Edition-A-2021-01), (95-Standard Edition-A-2022-01), (93-Standard Edition-A-2023-01), (91-Standard Edition-A-2024-01), (20-Youth Edition-A-2021-01), (15-Youth Edition-A-2022-01), (18-Youth Edition-A-2023-01), (17-Youth Edition-A-2024- 01), (3-Youth Edition-A-2021-01), (4-Youth Edition-A-2022-01), (0-Youth Edition-A-2023-01), (3-Youth Edition-A-2024-01), (12-Student Edition-A-2021-01), (11-Student Edition-A-2022-01), (11-Student Edition-A-2023-01), (10-Student Edition-A-2024-01), (68-Standard Edition-B-2 021-01), (67-Standard Edition-B-2022-01), (70-Standard Edition-B-2023-01), (71-Standard Edition-B-2024-01), (60-Youth Edition-B-2021-01), (63-Youth Edition-B-2022-01), (63-Youth Edition-B-2023-01), (65-Youth Edition-B-2024-01), (10-Youth Edition-B-2021-01), (9- Youth Edition-B-2022-01), (9-Youth Edition-B-2023-01), (12-Youth Edition-B-2024-01), (20-Student Edition-B-2021-01), (23-Student Edition-B-2022-01), (19-Student Edition-B-2023-01), (22-Student Edition-B-2024-01)}, based on the historical node set in the example, multiple historical product types are obtained, namely standard edition, youth edition and student edition. The initial product type set is a pre-set set of initial product types. If the initial product type set is {Standard Edition, Youth Edition, Student Edition, Premium Edition}, based on the historical product type "Standard Edition", the corresponding initial product type "Standard Edition" can be retrieved from the initial product type set {Standard Edition, Youth Edition, Student Edition, Premium Edition}, and "Standard Edition" is used as the target product type. Similarly, the historical product types "Youth Edition" and "Student Edition" can also retrieve the corresponding initial product types in the initial product type set, so the target product type set is {Standard Edition, Youth Edition, Student Edition}. The embodiment of the present invention can accurately extract product types related to historical nodes by counting historical product types and retrieving target product types from the initial product type set, forming a target product type set, thereby improving the pertinence and accuracy of subsequent analysis.

[0044] S4. Acquire multiple product node sets based on the historical node set and the target product type set, including:

[0045] Extract target product types from the target product type set in sequence, and perform the following operations on the extracted target product types:

[0046] Retrieving multiple historical nodes from the historical node set based on the target product type, wherein the historical product types corresponding to the multiple historical nodes are the target product types;

[0047] Aggregate multiple historical nodes to obtain a product node set;

[0048] The product node sets are aggregated to obtain multiple product node sets.

[0049] Among them, based on the historical node set and the target product type set in the above example, take the target product type standard version in the target product type set as an example: extract the target product type standard version from the target product type set, and use the target product type standard version as the search condition to retrieve multiple historical nodes corresponding to the standard version in the historical node set, and obtain the product node set corresponding to the target product type standard version as {(100-standard version-A-2021-01), (95-standard version-A-2022-01), (93-standard version-A-2023-01), (91-Standard Edition-A-2024-01), (68-Standard Edition-B-2021-01), (67-Standard Edition-B-2022-01), (70-Standard Edition-B-2023-01), (71-Standard Edition-B-2024-01)}, Similarly, the product node set corresponding to the target product type "Youth Edition" and the product node set corresponding to the target product type "Student Edition" are obtained in the same way as the product node set corresponding to the target product type "Standard Edition" and can achieve the same effect, which will not be repeated here. The embodiment of the present invention retrieves the corresponding historical nodes from the historical node set based on the target product type to form multiple product node sets. This method can effectively classify and summarize historical nodes according to the target product type, facilitate subsequent targeted analysis of data of different product types, and improve analysis efficiency.

[0050] S5. Obtain multiple regional value sets based on the product node set, wherein each regional value set in the multiple regional value sets contains multiple historical nodes, obtain the monitoring time in real time, obtain the prediction time and the verification time, and when the monitoring time reaches the prediction time, obtain the prediction value set based on the multiple regional value sets, wherein the number of predicted values ​​in the prediction value set is the same as the number of regional value sets in the multiple regional value sets.

[0051] For example, assuming that the product node set corresponding to the target product type standard version is {(100-Standard Edition-A-2021-01), (95-Standard Edition-A-2022-01), (93-Standard Edition-A-2023-01), (91-Standard Edition-A-2024-01), (68-Standard Edition-B-2021-01), (67-Standard Edition-B-2022-01), (70-Standard Edition-B-2023-01), (71-Standard Edition-B-2024-01)}, search the product node set with store A as the search condition. By taking the historical nodes corresponding to store A, we get the regional dataset corresponding to store A as {(100-standard version-A-2021-01), (95-standard version-A-2022-01), (93-standard version-A-2023-01), (91-standard version-A-2024-01)}. Similarly, we get the regional dataset corresponding to store B as (68-standard version-B-2021-01), (67-standard version-B-2022-01), (70-standard version-B-2023-01), (71-standard version-B-2024-01)}.

[0052] Furthermore, obtaining a predicted value set based on multiple regional value sets includes:

[0053] The following operations are performed on each of the multiple region value sets:

[0054] Get the current month, and identify the monitoring node set based on the current month and the regional value set, where the historical month corresponding to the historical node in the monitoring node set is the same as the current month;

[0055] Sort the historical nodes in the monitoring node set in order from the beginning to the end of the historical years to obtain a monitoring node sequence, and extract from the monitoring node sequence to obtain a monitoring value sequence, wherein the monitoring value sequence includes multiple monitoring historical values;

[0056] A prediction value acquisition scheme is constructed based on the monitoring value sequence, and the prediction value is acquired using the prediction value acquisition scheme, wherein the prediction value acquisition scheme is as follows:

[0057]

[0058] Among them, P k+1 Indicates the predicted values, P1, P2, P i-1 、P i 、P k They represent the first, second, i-1th, i-th and kth monitoring historical values ​​in the monitoring value sequence respectively, and k represents that there are k monitoring historical values ​​in the monitoring value sequence. Indicates rounding down, max{} indicates taking the maximum value;

[0059] The predicted values ​​are aggregated to obtain a predicted value set.

[0060] For example, in order to predict the sales volume of standard edition electronic products of the target product type in store A in February 2024, when the monitoring time reaches the 5 days before February 1, 2024, the regional data set corresponding to store A {(100-Standard Edition-A-2021-01), (95-Standard Edition-A-2022-01), (93-Standard Edition-A-2023-01), (91-Standard Edition-A-2024-01)} is used to obtain the forecast data. The forecast time here is January 26, 2024, and the corresponding verification time is March 1, 2024. Assuming that the current month is January 2024, when the monitoring time reaches the prediction time January 26, based on the current month January 2024, the monitoring node set {(100-standard version-A-2021-01), (95-standard version-A-2022-01), (93-standard version-A-2023-01), (91-standard version-A-2024-01)} is obtained from the regional data set corresponding to store A. -Standard Edition-A-2021-01), (95-Standard Edition-A-2022-01), (93-Standard Edition-A-2023-01), (91-Standard Edition-A-2024-01)}, sort the historical nodes in the monitoring node set in order from front to back of the historical year, and obtain a monitoring value sequence of {100, 95, 93, 91}, where 100, 95, 93 and 91 all represent monitoring historical values. It should be understood that in the embodiment of the present invention, only the historical nodes corresponding to January of each year are listed, and usually the historical node set will include historical nodes corresponding to each month. Optionally, the prediction time can be set manually, and the monitoring time refers to the current time. The method for obtaining the verification time is the same as the method for obtaining the prediction time, which will not be repeated here. The purpose of setting the prediction time and the verification time is to dynamically update the information in the initialized product system so as to improve the accuracy of prediction using the initialized product system.

[0061] Furthermore, the implementation process of the predicted value acquisition scheme is: if the monitored historical values ​​show a regular change trend (monotonically increasing or monotonically decreasing) in historical years, then the predicted value is obtained by calculating the average value of the relative change rate of the monitored historical values; otherwise, it indicates that the monitored historical values ​​in the monitored value sequence do not have a regular change trend, and the largest monitored historical value in the monitored value sequence is used as the predicted value. In the example, the monitored value sequence {100, 95, 93, 91} corresponding to the standard version of store A shows a monotonically decreasing trend. Therefore, the predicted data can be obtained by calculating the average value of the relative change rate of the historical data, which is 88. The predicted value of the standard version of store A in February 2024 is 88. Similarly, the predicted value of the standard version of store B in February 2024 is 71. Among them, the process of obtaining the predicted value of store B is the same as that of store A, and will not be repeated here. The embodiment of the present invention obtains the current month and the monitoring node set, analyzes the historical values ​​in the monitoring node sequence, calculates the predicted values ​​using the predicted value acquisition scheme, and obtains the predicted values ​​based on the time series characteristics of the monitored historical values, thereby providing a scientific basis for subsequent scheduling decisions and improving the accuracy and reliability of the prediction.

[0062] S6. Obtain the current value set, obtain the initial value set based on the current value set, confirm the incoming value node set and the outgoing value node set based on the predicted value set and the initial value set, confirm the scheduling instruction based on the incoming value node set, the outgoing value node set and the pre-built storage center, and obtain the scheduled product system based on the scheduling instruction and the initialized product system, where the scheduling instruction is 0 or 1.

[0063] Specifically, the step of determining the incoming value node set and the outgoing value node set based on the predicted value set and the initial value set includes:

[0064] Obtaining a value node set based on the predicted value set and the initial value set, wherein the value node set includes a plurality of value nodes, and each value node includes a predicted value and an initial value, wherein a historical position corresponding to the predicted value is the same as a historical position corresponding to the initial value;

[0065] The following operations are performed on each value node in the value node set:

[0066] Compare the initial value and the predicted value in the numerical node;

[0067] If the initial value is less than the predicted value, the value node is used as the incoming value node, and the incoming value nodes are summarized to obtain an incoming value node set;

[0068] Gets a callout value node-set based on the callout value node-set and the value node-set.

[0069] For example, based on the above example, the predicted data for the standard version of store A in February 2024 is 88, and the predicted data for the standard version of store B in February 2024 is 71. In order to predict the sales volume of the target product type standard version electronic products of store A in February 2024, when the prediction time arrives, the number of remaining standard version electronic products of store A and store B is obtained, which is the current value. Assuming that the current values ​​of store A and store B as of January 26, 2024 are 52 and 78 respectively, the current value set is {52, 78}.

[0070] If the current values ​​52 and 78 are used directly to calculate the forecast value, the forecast result will be inaccurate. The reason is that when the forecast is made on January 26, 2024, there are still 5 days before February. During these 5 days, the standard version of electronic products will continue to be sold. Therefore, the number of electronic products sold between January 26 and February should be subtracted. Therefore, using (current values ) is the number of standard electronic products sold between January 26 and February, which is the initial value. It improves the accuracy of the prediction to a certain extent. Here, t1 represents the cumulative number of days from the first day of the current month to the detection time, which is 26 days from January 1 to January 26, 2024. T2 represents the cumulative number of days from the detection time to the last day of the current month, which is 5 days from January 27 to January 31, 2024. Using the initial value calculate Take 42 as the initial value. Similarly, the initial value corresponding to the current value 78 is 63.

[0071] For example, assuming that the numerical nodes corresponding to stores A, B, C, D, E, and F are (88, 52), (71, 78), (30, 24), (50, 72), (10, 23), and (15, 28), respectively, the input numerical node set obtained based on multiple numerical node sets is {(88, 52), (30, 24)}, and the output numerical node set is {(71, 78), (50, 72), (10, 23), (15, 28)}. The acquisition process of the input numerical node set and the output numerical node set is as follows: Here, only the input numerical node (88, 52) and the output numerical node (71, 78) are used as examples: Among them, (88, 52) represents (predicted value 88, initial value 52), indicating that the predicted sales of store A is 88 pieces. The current inventory of store A is 52 pieces, indicating that the electronic products of store A are insufficient. In order to meet the sales demand of store A, the missing 36 pieces need to be transferred from other stores or storage centers. Therefore, (88, 52) is used as the transfer-in value node. Similarly, (71, 78) represents (predicted value 71, initial value 78), indicating that the predicted sales of store B is 71 pieces, while the current inventory of store B is 78 pieces. The supply of store B is sufficient. In order to increase the sales of other stores, the redundant electronic products of store B can be dispatched to other stores. Therefore, (71, 78) is used as the transfer-out value node. The remaining transfer-in value nodes and transfer-out value nodes can achieve the same effect as the transfer-in value nodes (88, 52) and the transfer-out value nodes (71, 78).

[0072] Therefore, this embodiment can compare the predicted value with the initial value, determine the set of value nodes to be called in and the set of value nodes to be called out, and can confirm the value nodes that need to be scheduled based on the difference between the predicted value and the initial value, thereby ensuring the scientificity and rationality of the scheduling decision.

[0073] Furthermore, the step of confirming the scheduling instruction based on the incoming value node set, the outgoing value node set, and the pre-built storage center includes:

[0074] Identify a storage center node based on the storage center;

[0075] The following operations are performed for each imported value node in the imported value node set:

[0076] Extracting outgoing value nodes from the outgoing value node set in sequence, and obtaining the area distance based on the incoming value nodes and the outgoing value nodes;

[0077] Summarize the regional distances to obtain a regional distance set, wherein the number of regional distances in the regional distance set is the same as the number of called-out value nodes in the called-out value node set;

[0078] Obtaining a center distance based on the input value node and the storage center;

[0079] The following operations are performed for each regional distance in the regional distance set:

[0080] Compare the size of the regional distance and the center distance;

[0081] If the regional distance is less than the center distance, the value node corresponding to the regional distance is taken as the priority value node, and the priority value nodes are summarized to obtain the priority value node set;

[0082] A scheduling instruction acquisition plan is constructed based on the set of incoming value nodes and the set of priority outgoing value nodes. The scheduling instruction acquisition plan is used to confirm the scheduling instruction. The scheduling instruction acquisition plan is as follows:

[0083]

[0084] Among them, n means that there are n priority value nodes in the priority value node set. Indicates the initial value of the jth priority value node, p jc ' indicates the current value corresponding to the initial value in the jth priority value node, P jc represents the predicted value in the jth priority value node, p represents the current value, t1 represents the cumulative number of days from the first day of the current month to the detection time, and t2 represents the cumulative number of days from the detection time to the last day of the current month. Indicates the initial value in the value node, P r 'The current value corresponding to the initial value in the value node, P r represents the predicted value transferred into the value node, and α represents the scheduling instruction.

[0085] The storage center refers to a central warehouse for storing and managing various types of electronic products, and can supply various types of electronic products to various stores according to scheduling instructions to meet the sales needs of the stores. When the stores cannot schedule each other, the storage center serves as a second choice for scheduling to ensure the timely and accurate supply of electronic products to each store. Exemplarily, the storage center node can be represented as {(Standard Edition-1000), (Youth Edition-650), (Student Edition-300), (Premium Edition-500)}, where only: Standard Edition-1000) is used as an example, where (Standard Edition-1000) means that the storage center has 1,000 pieces of standard version electronic products in stock, and (Youth Edition-650), (Student Edition-300), (Premium Edition-500) and (Standard Edition-1000) can achieve the same effect.

[0086] For example, assume that the input value node set is {(88, 52), (30, 24)}, corresponding to store A and store C respectively, and the output value node set is {(71, 78), (50, 72), (10, 23), (15, 28)}, corresponding to store B, store D, store E and store F respectively. Taking the incoming value node (88, 52) corresponding to store A as an example, the regional distances between store A and all stores corresponding to the outgoing value nodes are obtained. Assuming that the regional distances are {49km, 20km, 70km, 30km}, the center distance between store A and the storage center is 50km, and the outgoing value nodes with regional distances less than 50km are identified as priority outgoing value nodes, namely {(71, 78), (50, 72), (15, 28)}. The scheduling instruction acquisition scheme is used to obtain the scheduling instruction. The implementation process of the scheduling instruction acquisition scheme is as follows: If the total number of standard electronic products that can be called out after compensation from the priority outgoing value node set {(71, 78), (50, 72), (15, 28)} is greater than the number of incoming demands of the incoming value node (88, 52), then the stores corresponding to the priority outgoing value node set are used to dispatch to the store corresponding to the incoming value node. At this time, the scheduling instruction is 1. Otherwise, it is necessary to dispatch from the storage center to the store corresponding to the incoming value node, and execute the scheduling instruction 0.

[0087] The dispatch instruction is a decision signal obtained based on the dispatch instruction acquisition scheme, and is used to guide the dispatch of electronic products between stores and storage centers, or between stores, to optimize inventory management and meet sales demand. This embodiment comprehensively considers the relationship between the value node set, the call-out value node set, and the storage center, and calculates the regional distance and center distance based on the call-in value node set, the call-out value node set, and the storage center to generate the dispatch instruction, ensuring the rationality and effectiveness of the dispatch instruction, thereby improving dispatch efficiency and resource utilization.

[0088] Furthermore, the step of obtaining a scheduled product system based on the scheduling instruction and the initialization product system includes:

[0089] When the scheduling instruction is 1, the first scheduling solution is determined by using the set of incoming value nodes and the set of priority outgoing value nodes. The first scheduling solution is obtained as follows:

[0090] The following operations are performed on each priority value-calling node in the priority value-calling node set:

[0091] Calculate the absolute difference between the initial value and the predicted value corresponding to the priority value node to obtain the scheduling value, sort the priority value nodes according to the regional distance corresponding to the priority value node set to obtain a scheduling sequence, summarize the scheduling values ​​to obtain a scheduling value set, and obtain a scheduling value sequence based on the scheduling value set and the scheduling sequence;

[0092] The following operations are performed on each imported value node in the imported value node set:

[0093] Calculate the absolute difference between the predicted value and the initial value in the transferred numerical node to obtain a first absolute difference, use the first absolute difference as a first quantity, use the first quantity and the scheduling value sequence to obtain a first scheduling instruction, send the first scheduling instruction to the AI ​​intelligent analysis instruction initiator, confirm receipt of the target scheduled instruction, parse the target scheduled instruction, and obtain a first scheduled node;

[0094] Summarize the first scheduled nodes to obtain a first scheduled node set;

[0095] Using the first scheduled node set to update the initialized product system to obtain a scheduled product system;

[0096] When the scheduling instruction is 0, a second scheduling plan is confirmed using the set of incoming value nodes, the set of outgoing value nodes and the storage center node;

[0097] A scheduled product system is obtained based on the first scheduling solution or the second scheduling solution.

[0098] Specifically, when the scheduling instruction is 1, the first scheduling scheme is used to perform scheduling, that is, the scheduling of electronic products between stores. Taking store A corresponding to the input numerical node (88, 52) as an example, the priority value nodes (71, 78), (50, 72), and (15, 28) are respectively corresponding to store B, store D, and store F. The absolute difference between the initial value and the predicted value corresponding to the priority value node is calculated, and the scheduling values ​​are 7, 22, and 13 respectively. The regional distance corresponding to the priority value node set is {49km, 20km, 30km}. The priority value nodes are sorted to obtain a scheduling sequence of {store D-(50, 72), store F-(15, 28), store B-(71, 78)}. The scheduling values ​​are summarized to obtain a scheduling value set of {store B-7, store D-22, store F-13}. The scheduling value sequence obtained based on the scheduling value set and the scheduling sequence is {(store D-22), (store F-13), (store B-7). The absolute difference between the transferred numerical nodes (88, 52) is calculated to be 36, and the first quantity corresponding to store A is 36. The first scheduling instruction is obtained using the first quantity 36 and the scheduling numerical sequence {(store D-22), (store F-13), (store B-7): Scheduling according to the distance between the regions: It is recommended that the administrator first dispatch 22 electronic products from store D, which is closest to store A, to store A, and then dispatch 13 electronic products from store F to store A, and then dispatch 1 electronic product from store B to store A. The first scheduling instruction is sent to the AI ​​intelligent analysis instruction initiator, and the administrator executes the scheduling according to the first scheduling instruction. After judgment, the administrator believes that it is meaningless to dispatch only one electronic product from store B to store A. Therefore, only 35 products are dispatched by the dispatcher. The target scheduled instruction is obtained from the 35 products and the information corresponding to the products, and the target scheduled instruction is confirmed to be received. Here, the target scheduling instruction is that 22 standard electronic products have been scheduled from store D to store A and 13 standard electronic products have been scheduled from store F to store A. The target scheduled instruction is parsed to obtain the first scheduled node (22 standard electronic products have been scheduled from store D to store A and 13 standard electronic products have been scheduled from store F to store A). Taking the first scheduled node as an example, the initialized product system is updated using the first scheduled node to obtain the scheduled product system. The updating process is that the standard electronic products in store D in the initialized product system are reduced by 22, the standard electronic products in store F are reduced by 13, and the standard electronic products in store A are increased by 35.

[0099] Furthermore, the step of using the input value node set, the output value node set, and the storage center node to determine a second scheduling solution includes:

[0100] The second scheduling scheme is obtained by using the input value node set, the output value node set and the storage center node, wherein the second scheduling scheme is obtained in the following manner:

[0101] The following operations are performed on each imported value node in the imported value node set:

[0102] Extracting the input value nodes from the input value node set in sequence and calculating the absolute difference between the predicted value and the initial value in the input value node to obtain an absolute difference value set, wherein the absolute difference value set includes multiple absolute differences;

[0103] Sort the absolute differences in the absolute difference set in descending order of absolute difference to obtain an absolute difference sequence, and sort the imported numerical nodes in the imported numerical node set using the order of the absolute differences in the absolute difference sequence to obtain an imported node sequence;

[0104] Extracting incoming value nodes from the incoming node sequence in sequence, taking the absolute differences corresponding to the incoming value nodes as a second quantity, and acquiring a second scheduled node set based on the second quantity and the storage center node;

[0105] Summarizing the second scheduled nodes to obtain a second scheduled node set;

[0106] The following operations are performed on each node in the node set:

[0107] Calculating an absolute difference between a predicted value and an initial value in the retrieved value node to obtain a third quantity, and obtaining a third scheduled node using the third quantity and the storage center node;

[0108] Summarizing the third scheduled nodes to obtain a third scheduled node set;

[0109] The initialized product system is updated using the second scheduled node set and the third scheduled node set to obtain a scheduled product system.

[0110] When the scheduling instruction is 0, the second scheduling scheme is used for scheduling, that is, the scheduling of electronic products between stores and storage centers. The implementation process of the second scheduling scheme is as follows: Assume that the set of transferred-in numerical nodes is {(88, 66), (90, 88), (60, 50)}, which correspond to stores G, H and I respectively. Calculate the absolute difference between the predicted value and the initial value in the transferred-in numerical node, and obtain the absolute difference set of {22, 2, 10}. Sort the absolute differences in the absolute difference set in descending order, and obtain the absolute difference sequence of {22, 10, 2}. Use the order of the absolute differences in the absolute difference sequence to transfer the transferred-in numerical nodes into the set. The numerical nodes are sorted to obtain the transferred-in node sequence of {store G-22, store I-10, store H-2}. The transferred-in numerical nodes are extracted from the transferred-in node sequence in sequence. If the storage center node is {(standard version-1000), (youth version-650), (student version-300), (premium version-500)}, taking the first-order transferred-in numerical node store G-22 as an example, the second quantity 22 and the storage center node are used to perform scheduling: 22 standard version electronic products are dispatched from the storage center node to store G, and the second scheduled node is: the standard version electronic products in the storage center node are reduced by 22, and the standard version electronic products in store G are increased by 22. Similarly, assuming that the set of transferred-out value nodes is {(20, 52), (15, 20)}, corresponding to store J and store K respectively, the absolute difference between the predicted value and the initial value in the transferred-in value node is calculated, and the absolute difference set is {32, 5}. The absolute differences in the absolute difference set are sorted in descending order, and the absolute difference sequence is obtained {32, 5}. The order of the absolute differences in the absolute difference sequence is used to sort the transferred-in value nodes in the transferred-in value node set, and the transferred-in node sequence is {store J- 32, store K-10}, if the storage center nodes at this time are {(Standard Edition-1000), (Youth Edition-650), (Student Edition-300), (Premium Edition-500)}, taking the first-order called-out value node store J-32 as an example, use the third quantity 32 and the storage center node to perform scheduling: dispatch 32 standard edition electronic products from store J to the storage center node, and the third scheduled node is: the standard edition electronic products in the storage center node increase by 32, and the standard edition electronic products in store J decrease by 32.

[0111] Furthermore, the second scheduled node and the third scheduled node are used to update the initialized product system to obtain a scheduled product system, wherein the updating process is the same as that of the updating according to the first scheduling scheme and can achieve the same effect, and will not be repeated here. The embodiment of the present invention generates a first scheduling scheme and a second scheduling scheme respectively according to different scheduling instructions, and can flexibly select a scheduling scheme according to different situations, ensuring the adaptability and effectiveness of the scheduling scheme, and at the same time providing accurate data support for subsequent business operations by updating the initialized product system.

[0112] S7. When the monitoring time reaches the verification time, obtain the actual value set, where the actual value set contains multiple actual values, judge whether the predicted value set and the actual value set meet the pre-built evaluation conditions, and after confirming that the predicted value set corresponding to each product node set in the multiple product node sets meets the evaluation conditions, confirm the scheduled product system as the target product system to realize intelligent analysis of business data.

[0113] Specifically, the determination of whether the predicted value set and the actual value set meet the pre-established evaluation conditions includes:

[0114] Obtaining a verification node set based on the predicted value set and the actual value set, wherein the verification node set includes multiple verification nodes, and each verification node includes a predicted value and an actual value;

[0115] Determine whether the verification node set meets the evaluation conditions, where the evaluation conditions are as follows:

[0116]

[0117] Among them, m means there are m verification nodes in the verification node set, P' lc Represents the predicted value in the lth verification node, P ls represents the actual value in the lth verification node, and μ represents the preset difference threshold.

[0118] In summary, this embodiment can accurately extract product types related to historical nodes by counting historical product types and retrieving target product types from the initial product type set, forming a target product type set, thereby improving the pertinence and accuracy of subsequent analysis. At the same time, historical nodes can be effectively classified and summarized according to target product types, facilitating subsequent targeted analysis of data of different product types and improving analysis efficiency.

[0119] Furthermore, by obtaining the predicted values ​​for the current month and the monitoring node set, a scientific basis is provided for subsequent scheduling decisions, the accuracy and reliability of the prediction are improved, and it is determined whether the predicted value set and the actual value set meet the pre-built evaluation conditions. After confirming that the predicted value set corresponding to each product node set in multiple product node sets meets the evaluation conditions, the scheduled product system is confirmed as the target product system to realize intelligent analysis of business data, and different scheduling plans are generated according to the scheduling instructions to ensure the adaptability and effectiveness of the scheduling plan. At the same time, accurate data support is provided for subsequent business operations by updating the initialized product system.

[0120] Example 2:

[0121] like Figure 2 As shown, this embodiment provides an AI intelligent analysis system suitable for business data, which can implement the AI ​​intelligent analysis system described in Example 1.

[0122] The AI ​​intelligent analysis system 100 can be installed in an electronic device. According to the functions implemented, the AI ​​intelligent analysis system 100 applicable to business data may include an initialization module 101, a historical node processing module 102, a prediction and scheduling module 103, and a prediction result verification module 104;

[0123] The initialization module 101 is configured to receive an AI intelligent analysis instruction and determine an initialization product system based on the AI ​​intelligent analysis instruction.

[0124] The historical node processing module 102 is used to obtain a historical node set, wherein the historical node set includes multiple historical nodes, and each historical node includes a historical value, a historical product type, a historical position, a historical year, and a historical month;

[0125] Obtaining an initial product type set, wherein the initial product type set includes multiple initial product types, and obtaining a target product type set based on the historical node set and the initial product type set;

[0126] The prediction and scheduling module 103 is used to obtain multiple product node sets based on the historical node set and the target product type set, and perform the following operations on each of the multiple product node sets:

[0127] Based on the product node set, multiple regional value sets are obtained, wherein each of the multiple regional value sets contains multiple historical nodes, monitoring time is obtained in real time, and prediction time and verification time are obtained. When the monitoring time reaches the prediction time, a prediction value set is obtained based on the multiple regional value sets, wherein the number of prediction values ​​in the prediction value set is the same as the number of regional value sets in the multiple regional value sets;

[0128] Obtain a current value set, obtain an initial value set based on the current value set, confirm an incoming value node set and an outgoing value node set based on the predicted value set and the initial value set, confirm a scheduling instruction based on the incoming value node set, the outgoing value node set, and a pre-built storage center, and obtain a scheduled product system based on the scheduling instruction and the initialized product system, where the scheduling instruction is 0 or 1;

[0129] The prediction result verification module 104 is used to obtain an actual value set when the monitoring time reaches the verification time, wherein the actual value set includes multiple actual values, and judge whether the predicted value set and the actual value set meet the pre-built evaluation conditions. After confirming that the predicted value set corresponding to each product node set in the multiple product node sets meets the evaluation conditions, the scheduled product system is confirmed as the target product system to realize intelligent analysis of business data.

[0130] In detail, the modules in the AI ​​intelligent analysis system 100 for business data in the embodiment of the present invention adopt the same Figure 1 The same technical means as the AI ​​intelligent analysis method applicable to business data described in the previous section can produce the same technical effects, so I will not go into details here.

[0131] like Figure 3 , which is a schematic diagram of the structure of an electronic device for implementing an AI intelligent analysis method applicable to business data provided by one embodiment of the present invention.

[0132] The electronic device 1 may include a processor 10, a memory 11 and a bus 12, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as the AI ​​intelligent analysis method program applicable to business data described in Example 1.

[0133] The memory 11 includes at least one type of readable storage medium, including flash memory, mobile hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), magnetic memory, disk, optical disk, etc.

[0134] In some embodiments, the processor 10 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and a combination of various control chips. The processor 10 is the control core (Control Unit) of the electronic device, which connects the various components of the entire electronic device using various interfaces and lines, and executes the programs or modules stored in the memory 11 (such as AI intelligent analysis method programs for business data, etc.), and calls the data stored in the memory 11 to perform various functions of the electronic device 1 and process data.

[0135] The bus 12 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus.

[0136] Figure 3 Only the electronic device with components is shown, and it can be understood by those skilled in the art that Figure 3 The structure shown does not constitute a limitation on the electronic device 1 , and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0137] Furthermore, the electronic device 1 may also include a network interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device 1 and other electronic devices.

[0138] The AI ​​intelligent analysis method program for business data stored in the memory 11 of the electronic device 1 is a combination of multiple instructions. When running in the processor 10, the AI ​​intelligent analysis method for business data described in Example 1 can be implemented. Specifically, the specific implementation method of the above instructions by the processor 10 can refer to Figures 1 to 3 The description of the relevant steps in the corresponding embodiments will not be repeated here.

[0139] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor of an electronic device, it can implement the AI ​​intelligent analysis method for business data described in Example 1.

[0140] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An AI intelligent analysis method applicable to business data, characterized in that: The method comprises: Receiving an AI intelligent analysis instruction, and confirming to initialize a product system based on the AI ​​intelligent analysis instruction; Obtain a historical node set, wherein the historical node set includes multiple historical nodes, and each historical node includes a historical value, a historical product type, a historical position, a historical year, and a historical month; Obtaining an initial product type set, wherein the initial product type set includes multiple initial product types, and obtaining a target product type set based on the historical node set and the initial product type set; Based on the historical node set and the target product type set, multiple product node sets are obtained, and the following operations are performed on each of the multiple product node sets: Based on the product node set, multiple regional value sets are obtained, wherein each of the multiple regional value sets contains multiple historical nodes, monitoring time is obtained in real time, and prediction time and verification time are obtained. When the monitoring time reaches the prediction time, a prediction value set is obtained based on the multiple regional value sets, wherein the number of prediction values ​​in the prediction value set is the same as the number of regional value sets in the multiple regional value sets; Obtain a current value set, obtain an initial value set based on the current value set, confirm an incoming value node set and an outgoing value node set based on the predicted value set and the initial value set, confirm a scheduling instruction based on the incoming value node set, the outgoing value node set, and a pre-built storage center, and obtain a scheduled product system based on the scheduling instruction and the initialized product system, where the scheduling instruction is 0 or 1; When the monitoring time reaches the verification time, the actual value set is obtained, where the actual value set contains multiple actual values. It is judged whether the predicted value set and the actual value set meet the pre-built evaluation conditions. After confirming that the predicted value set corresponding to each product node set in the multiple product node sets meets the evaluation conditions, the scheduled product system is confirmed as the target product system to realize intelligent analysis of business data.

2. The AI ​​intelligent analysis method according to claim 1, characterized in that: The step of obtaining the target product type set based on the historical node set and the initial product type set includes: Counting historical product types in the historical node set to obtain multiple historical product types, sequentially extracting historical product types from the multiple historical product types, searching the initial product type set using the historical product types, and if the initial product type is retrieved from the initial product type set, confirming the initial product type as the target product type, wherein the historical product type is the same as the initial product type; Summarize the target product types to obtain a target product type set.

3. The AI ​​intelligent analysis method according to claim 2, characterized in that: The acquiring of multiple product node sets based on the historical node set and the target product type set includes: Extract target product types from the target product type set in sequence, and perform the following operations on the extracted target product types: Retrieving multiple historical nodes from the historical node set based on the target product type, wherein the historical product types corresponding to the multiple historical nodes are the target product types; Aggregate multiple historical nodes to obtain a product node set; The product node sets are aggregated to obtain multiple product node sets.

4. The AI ​​intelligent analysis method according to claim 3, characterized in that: The obtaining of a predicted value set based on multiple regional value sets includes: The following operations are performed on each of the multiple region value sets: Get the current month, and identify the monitoring node set based on the current month and the regional value set, where the historical month corresponding to the historical node in the monitoring node set is the same as the current month; Sort the historical nodes in the monitoring node set in order from the beginning to the end of the historical years to obtain a monitoring node sequence, and extract from the monitoring node sequence to obtain a monitoring value sequence, wherein the monitoring value sequence includes multiple monitoring historical values; A prediction value acquisition scheme is constructed based on the monitoring value sequence, and the prediction value is acquired using the prediction value acquisition scheme, wherein the prediction value acquisition scheme is as follows: if else,P k+1 =max{P1,P2,…,P k } Among them, P k+1 Indicates the predicted values, P1, P2, P i-1 、P i 、P k They represent the first, second, i-1th, i-th and kth monitoring historical values ​​in the monitoring value sequence respectively, and k represents that there are k monitoring historical values ​​in the monitoring value sequence. Indicates rounding down, max{} indicates taking the maximum value; The predicted values ​​are aggregated to obtain a predicted value set.

5. The AI ​​intelligent analysis method according to claim 4, characterized in that: The step of determining the input value node set and the output value node set based on the predicted value set and the initial value set includes: Obtaining a value node set based on the predicted value set and the initial value set, wherein the value node set includes a plurality of value nodes, and each value node includes a predicted value and an initial value, wherein a historical position corresponding to the predicted value is the same as a historical position corresponding to the initial value; The following operations are performed on each value node in the value node set: Compare the initial value and the predicted value in the numerical node; If the initial value is less than the predicted value, the value node is used as the incoming value node, and the incoming value nodes are summarized to obtain an incoming value node set; Gets a callout value node-set based on the callout value node-set and the value node-set.

6. The AI ​​intelligent analysis method according to claim 5, characterized in that: The step of confirming the scheduling instruction based on the input value node set, the output value node set, and the pre-built storage center includes: Identify a storage center node based on the storage center; The following operations are performed for each imported value node in the imported value node set: Extracting outgoing value nodes from the outgoing value node set in sequence, and obtaining the area distance based on the incoming value nodes and the outgoing value nodes; Summarize the regional distances to obtain a regional distance set, wherein the number of regional distances in the regional distance set is the same as the number of called-out value nodes in the called-out value node set; Obtaining a center distance based on the input value node and the storage center; The following operations are performed for each regional distance in the regional distance set: Compare the size of the regional distance and the center distance; If the regional distance is less than the center distance, the value node corresponding to the regional distance is taken as the priority value node, and the priority value nodes are summarized to obtain the priority value node set; A scheduling instruction acquisition plan is constructed based on the set of incoming value nodes and the set of priority outgoing value nodes. The scheduling instruction acquisition plan is used to confirm the scheduling instruction. The scheduling instruction acquisition plan is as follows: if a=1 else,α=0 Among them, n means that there are n priority value nodes in the priority value node set. Indicates the initial value of the jth priority value node, p jc ' indicates the current value corresponding to the initial value in the jth priority value node, P jc represents the predicted value in the jth priority value node, p represents the current value, t1 represents the cumulative number of days from the first day of the current month to the detection time, and t2 represents the cumulative number of days from the detection time to the last day of the current month. Indicates the initial value in the value node, P r 'The current value corresponding to the initial value in the value node, P r represents the predicted value transferred into the value node, and α represents the scheduling instruction.

7. The AI ​​intelligent analysis method according to claim 6, characterized in that: The step of obtaining a scheduled product system based on the scheduling instruction and the initialization product system includes: When the scheduling instruction is 1, the first scheduling solution is determined by using the set of incoming value nodes and the set of priority outgoing value nodes. The first scheduling solution is obtained as follows: The following operations are performed on each priority value-calling node in the priority value-calling node set: Calculate the absolute difference between the initial value and the predicted value corresponding to the priority value node to obtain the scheduling value, sort the priority value nodes according to the regional distance corresponding to the priority value node set to obtain a scheduling sequence, summarize the scheduling values ​​to obtain a scheduling value set, and obtain a scheduling value sequence based on the scheduling value set and the scheduling sequence; The following operations are performed on each imported value node in the imported value node set: Calculate the absolute difference between the predicted value and the initial value in the transferred numerical node to obtain a first absolute difference, use the first absolute difference as a first quantity, use the first quantity and the scheduling value sequence to obtain a first scheduling instruction, send the first scheduling instruction to the AI ​​intelligent analysis instruction initiator, confirm receipt of the target scheduled instruction, parse the target scheduled instruction, and obtain a first scheduled node; Summarize the first scheduled nodes to obtain a first scheduled node set; Using the first scheduled node set to update the initialized product system to obtain a scheduled product system; When the scheduling instruction is 0, a second scheduling plan is confirmed using the set of incoming value nodes, the set of outgoing value nodes and the storage center node; A scheduled product system is obtained based on the first scheduling solution or the second scheduling solution.

8. The AI ​​intelligent analysis method according to claim 7, wherein: The step of using the input value node set, the output value node set, and the storage center node to determine the second scheduling solution includes: The second scheduling scheme is obtained by using the input value node set, the output value node set and the storage center node, wherein the second scheduling scheme is obtained in the following manner: The following operations are performed on each imported value node in the imported value node set: Extracting the input value nodes from the input value node set in sequence and calculating the absolute difference between the predicted value and the initial value in the input value node to obtain an absolute difference value set, wherein the absolute difference value set includes multiple absolute differences; Sort the absolute differences in the absolute difference set in descending order of absolute difference to obtain an absolute difference sequence, and sort the imported numerical nodes in the imported numerical node set using the order of the absolute differences in the absolute difference sequence to obtain an imported node sequence; Extracting incoming value nodes from the incoming node sequence in sequence, taking the absolute differences corresponding to the incoming value nodes as a second quantity, and acquiring a second scheduled node set based on the second quantity and the storage center node; Summarizing the second scheduled nodes to obtain a second scheduled node set; The following operations are performed on each node in the node set: Calculating an absolute difference between a predicted value and an initial value in the retrieved value node to obtain a third quantity, and obtaining a third scheduled node using the third quantity and the storage center node; Summarizing the third scheduled nodes to obtain a third scheduled node set; The initialized product system is updated using the second scheduled node set and the third scheduled node set to obtain a scheduled product system.

9. The AI ​​intelligent analysis method according to claim 8, characterized in that: The step of determining whether the predicted value set and the actual value set meet pre-established evaluation conditions includes: Obtaining a verification node set based on the predicted value set and the actual value set, wherein the verification node set includes multiple verification nodes, and each verification node includes a predicted value and an actual value; Determine whether the verification node set meets the evaluation conditions, where the evaluation conditions are as follows: Among them, m means there are m verification nodes in the verification node set, P' lc Represents the predicted value in the lth verification node, P ls represents the actual value in the lth verification node, and μ represents the preset difference threshold.

10. An AI intelligent analysis system suitable for business data, characterized in that: The system comprises: An initialization module is configured to receive an AI intelligent analysis instruction and determine an initialization product system based on the AI ​​intelligent analysis instruction; A historical node processing module is used to obtain a historical node set, wherein the historical node set includes multiple historical nodes, and each historical node includes a historical value, a historical product type, a historical position, a historical year, and a historical month; Obtaining an initial product type set, wherein the initial product type set includes multiple initial product types, and obtaining a target product type set based on the historical node set and the initial product type set; The prediction and scheduling module is used to obtain multiple product node sets based on the historical node set and the target product type set, and perform the following operations on each of the multiple product node sets: Based on the product node set, multiple regional value sets are obtained, wherein each of the multiple regional value sets contains multiple historical nodes, monitoring time is obtained in real time, and prediction time and verification time are obtained. When the monitoring time reaches the prediction time, a prediction value set is obtained based on the multiple regional value sets, wherein the number of prediction values ​​in the prediction value set is the same as the number of regional value sets in the multiple regional value sets; Obtain a current value set, obtain an initial value set based on the current value set, confirm an incoming value node set and an outgoing value node set based on the predicted value set and the initial value set, confirm a scheduling instruction based on the incoming value node set, the outgoing value node set, and a pre-built storage center, and obtain a scheduled product system based on the scheduling instruction and the initialized product system, where the scheduling instruction is 0 or 1; The prediction result verification module is used to obtain the actual value set when the monitoring time reaches the verification time, where the actual value set contains multiple actual values, judge whether the predicted value set and the actual value set meet the pre-built evaluation conditions, and confirm that the predicted value set corresponding to each product node set in multiple product node sets meets the evaluation conditions. After that, the scheduled product system is confirmed as the target product system to realize intelligent analysis of business data.

Images