Intelligent purchase packaging method and system based on purchase history data

Abstract

The application provides an intelligent purchase packaging method and system based on purchase history data, wherein the method comprises the following steps: obtaining purchase data from a purchase order table and material master data of a business system, generating and storing purchase history indexes to a data warehouse; executing preset index calculation rules based on the purchase history indexes to generate purchase index data; obtaining a purchase packaging request containing a list of materials to be purchased, grouping the list of materials to be purchased according to preset multi-level processing logic, generating candidate material packages, matching a corresponding purchase strategy for each candidate material package, and forming a final purchase material package; and generating an inquiry order and outputting the inquiry order to the business system. By constructing a purchase history index data warehouse and introducing a rule engine, the application realizes automatic packaging of purchase requirements and intelligent matching of purchase strategies, improves purchase efficiency, reduces the burden on the business system, and guarantees the uniformity and scientificity of purchase decisions.

Description

Technical Field

[0001] This invention relates to the field of enterprise procurement management technology, specifically to an intelligent procurement packaging method and system based on historical procurement data. Background Technology

[0002] In the field of corporate procurement management, procurement bundling refers to the process of combining multiple procurement needs into a single procurement package to facilitate unified inquiries or tenders from suppliers. As companies grow and their procurement needs diversify, the complexity and workload of procurement bundling continue to increase.

[0003] Currently, companies typically use manual methods in the procurement and packaging process: procurement staff need to collect purchase requisitions submitted by different departments at different times; for each material, they need to manually query historical purchase orders in the business system to obtain historical purchase prices and quantities; based on personal experience, they need to determine which materials can be combined into a package to achieve economies of scale; based on factors such as purchase amount and material characteristics, they need to determine whether each procurement package should use inquiry procurement, bidding procurement, or other procurement methods; finally, they need to manually create inquiry forms or bidding forms in the business system and enter material information item by item.

[0004] Traditional manual operation methods have the following technical problems: First, purchasing staff need to spend a lot of time on repetitive data querying and entry. Taking a medium-sized enterprise as an example, it may need to process hundreds of purchase requisitions every month, involving thousands of materials. Purchasing staff need to query historical data one by one, make group judgments, and prepare purchase documents. The whole process usually takes several days, becoming an efficiency bottleneck in the purchasing process.

[0005] Secondly, each packaging operation requires real-time queries to the purchase order table and historical data in the business system. When multiple purchasing staff perform packaging operations simultaneously, or during peak purchasing periods, frequent data queries can put heavy concurrent access pressure on the business system, causing system response to slow down and even affecting the normal operation of other core business processes.

[0006] Furthermore, the choice of packaging methods and procurement strategies is highly dependent on the personal experience and subjective judgment of the procurement staff. Different procurement staff may have different packaging methods for the same batch of materials, and their selection criteria for procurement strategies may not be consistent, leading to unstable procurement results and making it difficult to ensure the standardization and consistency of procurement work.

[0007] Furthermore, when a company adjusts its procurement strategy, it needs to modify the program code and redeploy the system. The entire process involves multiple stages, including requirements communication, code development, testing and verification, and deployment, resulting in a long response cycle and an inability to quickly adapt to market changes and enterprise management needs.

[0008] A patent search revealed an invention patent with publication number CN118365248A, which discloses a data analysis-based intelligent purchase order generation and decision-making system. This system includes a data acquisition module, a weight allocation module, a current purchase quantity determination module, a supply quantity determination module, and a purchase order generation module. The data acquisition module divides historical time into multiple historical action time periods and acquires historical data. The weight allocation module calculates the correlation between the current action time period and historical action time periods and assigns weights to these periods. The current purchase quantity determination module inputs historical data and weights from multiple historical action time periods into a purchase quantity determination model, which outputs the current purchase quantity. The supply quantity determination module uses a genetic algorithm to allocate the current purchase quantity to multiple suppliers to obtain a target supply plan. The purchase order generation module generates a purchase order for the current action time period based on the target supply plan. However, this patent only focuses on calculating purchase and supply quantities and does not address the packaging and grouping of purchased materials, thus failing to achieve centralized material procurement and lacking full-process automation support for procurement.

[0009] In summary, given the problems of the existing technologies, researching an intelligent procurement packaging method and system based on historical procurement data has become a critical task that urgently needs to be addressed. Summary of the Invention

[0010] To address the shortcomings of existing technologies, the purpose of this invention is to provide an intelligent procurement packaging method and system based on historical procurement data.

[0011] According to the present invention, an intelligent procurement packaging method based on procurement history data includes the following steps: Data preparation steps: Obtain procurement data from the purchase order table and material master data of the business system, generate and store historical procurement indicators based on the procurement data to the data warehouse, and the historical procurement indicators should at least include historical price data of materials; Indicator generation steps: Based on historical procurement indicators, execute preset indicator calculation rules to generate procurement indicator data to guide procurement packaging. The procurement indicator data must include at least the benchmark price of the material. Intelligent packaging steps: Obtain a purchase packaging request containing a list of materials to be purchased; based on the purchase packaging request, read the corresponding historical purchase indicators and purchase indicator data from the data warehouse; group the list of materials to be purchased according to the preset multi-level processing logic to generate at least one candidate material package; match the corresponding purchase strategy for each candidate material package to form the final purchase material package. Request for Quotation (RFQ) generation steps: Based on procurement indicator data, the final procurement material package and its corresponding procurement strategy, generate an RFQ and output the RFQ to the business system.

[0012] Preferably, in the data preparation step, newly added or changed procurement data is extracted from the business system according to a preset cycle, and the procurement data is cleaned. The cleaning includes removing test data, standardizing the format, and filling missing values ​​with preset default values. The cleaned procurement data is then subjected to integrity verification. After the verification is passed, historical procurement indicators are calculated and generated according to the materials. The historical procurement indicators also include historical data on the procurement quantity of materials. The historical procurement indicators are stored in the data warehouse according to the material dimension.

[0013] Preferably, for non-new materials with historical purchase records in the business system, the corresponding calculation rule is selected from the preset calculation rule library according to the material category to which it belongs, and the historical average purchase price and historical purchase quantity of the material are calculated. The calculation rule library includes arithmetic mean method, weighted moving average method, and median method. For new materials without historical purchase records, the estimated usage is read from the material master data as the historical purchase quantity, and the estimated usage is marked. The estimated usage is determined according to the material's production plan requirements or the usage of similar materials in history.

[0014] Preferably, the indicator generation step includes: after new historical procurement indicators are written into the data warehouse, a data update event is sent through a message queue to trigger the rule engine to start; the rule engine has a pre-set digital business rule set, which includes at least indicator calculation rules, strategy matching rules, and quotation generation rules; the rule engine is implemented using Drools, EasyRules, or a self-developed rule engine; the rule engine uses the new historical procurement indicators as input facts, executes the indicator calculation rules, and generates procurement indicator data.

[0015] Preferably, the indicator calculation rules include a benchmark purchase price calculation rule, which includes: extracting the most recent purchase transaction price and the historical average purchase price from historical price data; obtaining the corresponding weight coefficient configuration from the rule engine according to the material category, with the weight coefficient being a value between 0 and 1, and the sum of the weight coefficients of the most recent purchase transaction price and the historical average purchase price being 1; and calculating the benchmark purchase price according to a weighted summation formula. ,in, To set a benchmark price for procurement, The price is the most recent purchase transaction price. The historical average purchase price The weighting coefficient is the most recent purchase transaction price. The weighting coefficients for different material categories are dynamically adjusted in the rule engine through the configuration interface.

[0016] Preferably, the multi-level processing logic in the intelligent packaging step includes the following sub-steps: Step a, Material Filtering: According to the preset filtering conditions, remove materials that do not need to participate in the subsequent packaging from the list of materials to be purchased, and obtain the set of materials to be packaged. The filtering conditions include that the material type is one or more of office supplies, labor protection supplies, and fixed assets, or the estimated purchase amount of a single material is lower than the preset threshold. Step b, Account Set Grouping: Based on the purchasing account set information, the collection of materials to be packaged is grouped, and materials belonging to the same purchasing account set are grouped together. Each group is an account set-level material collection. Step c, Brand Grouping: For each set of materials at the account level, determine the number of brands contained in each set of materials at the account level. If it contains multiple brands, group the materials of the same brand together to form a brand-level material set and proceed to step d; if it contains only a single brand, treat the set of materials at the account level as a candidate material package, package it by account level, and terminate the subsequent grouping process. Step d, Source Grouping: Determine the number of sourcing sources included in each brand-level material collection. If it includes multiple sourcing sources, group the materials from the same sourcing source together to form a source-level material collection and proceed to step e. If it only includes a single sourcing source, treat the brand-level material collection as a candidate material package, package it by brand, and terminate the subsequent grouping process. Step e, Attribute Grouping: Determine the number of material attributes contained in each source region-level material set. If it contains multiple material attributes, group materials with the same attribute together to form an attribute-level material set, and proceed to step f. If it contains only a single material attribute, treat the source region-level material set as a candidate material package, package it by source region, and terminate the subsequent grouping process. Step f, candidate package generation: Define each attribute-level material set as a candidate material package, and package it by category.

[0017] Preferably, in the intelligent packaging step, a feature vector for quantitatively describing the procurement characteristics of each candidate material package is extracted or calculated from the data of the materials within each candidate material package. The feature vector includes at least one or more of the following: packaging method, total budget price, benchmark total price, and number of historical suppliers. The packaging method is determined based on the candidate material package during the multi-level grouping process. The total budget price is calculated by summing the budget unit price and required quantity of each material in the procurement packaging request for all materials in the candidate material package. The calculation formula is: Total Budget Price = ∑(Budget Unit Price of Material × Required Quantity of Material). The benchmark total price is calculated by summing the benchmark price and required quantity of each material in the procurement indicator data for all materials in the candidate material package. The calculation formula is: Benchmark Total Price = ∑(Benchmark Price of Material × Required Quantity of Material). The number of historical suppliers is calculated based on the historical procurement indicators, counting the number of different suppliers that all materials in the candidate material package have cooperated with in the past preset time period. If multiple materials have cooperated with the same supplier, the supplier is only counted once. The feature vector is used as input fact to call the rule engine, which executes the preset strategy matching rules. The rule engine matches the corresponding procurement strategy for each candidate material package through pattern matching. The procurement strategy includes one or more of the following: inquiry, bidding, competitive negotiation, and single source. The candidate material package that matches the procurement strategy is output as the final procurement material package.

[0018] Preferably, it also includes a query quantum step, performed after generating the candidate material package and before constructing the feature vector: Based on the type of the procurement package request, the query volume calculation rules pre-bound to the type are obtained from the rule engine. The types of procurement package requests include annual framework agreement procurement, one-time project procurement, and emergency replenishment procurement. The query volume calculation rules are executed, and a suggested query volume is calculated for each candidate material package based on historical procurement indicators. The suggested query volume is calculated by multiplying the historical monthly average usage of the material by the number of months in the procurement cycle, or by multiplying the total project demand by a preset coefficient. When the historical procurement metrics used to calculate the suggested inquiry volume meet the preset abnormal trigger conditions, the adjustment rules corresponding to the abnormality are obtained from the rule engine to correct the suggested inquiry volume. Abnormal trigger conditions include historical usage fluctuations exceeding preset thresholds and materials being new products with no historical data.

[0019] Preferably, in the Request for Quotation (RFQ) generation step, based on the procurement strategy corresponding to each final procurement material package, the relevant procurement indicator data is read from the data warehouse and the rule engine is invoked. The rule engine executes the preset RFQ generation rules to create the RFQ. The RFQ generation rules specify the RFQ templates, required fields, approval processes, and potential supplier screening criteria under different procurement strategies. The RFQ templates include RFQ templates, tender templates, competitive negotiation templates, and single-source templates, with different formats and field requirements for each template. The required fields are determined according to the procurement strategy and include one or more of the following: technical parameter confirmation letter, deadline for multiple rounds of quotations, and qualification certificates. The potential supplier screening criteria are determined based on the material category and historical transaction records, including suppliers that match the material category and have a historical cooperation count greater than a preset threshold, or high-quality suppliers selected through supplier profile scoring. The generated RFQ is stored in the business system via an API interface. The rules engine provides a management interface for digital business rule sets. The management interface allows for adding, modifying, deleting, and querying rules through a RESTful API or web page. It also supports inputting simulated factual data, executing modified rules, and outputting the rule execution results for users to compare and verify. The verification process does not affect the online operation of the rules. After the rules are verified, the rules engine's hot deployment function reloads the rule file and makes it effective without restarting the system. Subsequent requests will use the new rules after they take effect.

[0020] This invention also provides an intelligent procurement and packaging system based on historical procurement data, comprising: Data preparation module: Obtain procurement data from the purchase order table and material master data of the business system, generate and store historical procurement indicators to the data warehouse based on the procurement data, and the historical procurement indicators must include at least the historical price data of the materials; Indicator generation module: Based on historical procurement indicators, execute preset indicator calculation rules to generate procurement indicator data to guide procurement packaging. The procurement indicator data must include at least the benchmark price of the material. The intelligent packaging module obtains a purchase packaging request containing a list of materials to be purchased, reads the corresponding historical purchase indicators and purchase indicator data from the data warehouse based on the purchase packaging request, groups the list of materials to be purchased according to a preset multi-level processing logic, generates at least one candidate material package, matches the corresponding purchase strategy to each candidate material package, and forms the final purchase material package. Inquiry generation module: Based on procurement indicator data, final procurement material packages and their corresponding procurement strategies, it generates inquiry forms and outputs them to the business system.

[0021] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention pre-generates and stores historical procurement indicators in a data warehouse. Upon receiving a procurement packaging request, it eliminates the need to query the purchase order table in the business system in real time; instead, it directly reads the required data in batches from the data warehouse. Simultaneously, it automatically groups materials through multi-level processing logic and automatically matches procurement strategies and generates quotations through a rule engine. This reduces the time required for procurement staff to complete tasks such as material searching, manual packaging, strategy judgment, and document creation from several days to just minutes, significantly improving the processing efficiency of the procurement packaging process.

[0022] 2. This invention adopts an architecture that separates the data warehouse from the business system. New or changed procurement data is extracted from the business system according to a preset cycle, generating historical procurement indicators and storing them in the data warehouse. When a procurement packaging request is initiated, the system reads data from the data warehouse, avoiding the need to query the business system in real time for each packaging. This removes the high-frequency operation of querying historical data from the core business system, reducing the concurrent query pressure on the business system and ensuring its stable operation.

[0023] 3. This invention uses a rule engine to pre-configure a set of digital business rules, including indicator calculation rules, strategy matching rules, and quotation generation rules. In the intelligent packaging process, the procurement strategy matching for all candidate material packages is based on the same rule conditions for pattern matching, eliminating inconsistencies in decision-making caused by differences in individual experience among different procurement personnel, making the procurement packaging results more stable and objective.

[0024] 4. This invention provides a visual rule management interface through a rule engine, supporting the addition, modification, deletion, and querying of rules via a RESTful API or web page. When procurement strategies are adjusted, no program code needs to be modified; simply update the corresponding strategy matching rules through the rule management interface, and utilize the rule engine's hot deployment function to make the new rules take effect immediately. Attached Figure Description

[0025] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a flowchart of an intelligent procurement and packaging method based on historical procurement data, as described in an embodiment of the present invention. Detailed Implementation

[0026] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0027] Glossary: Business systems: These are the core systems for the daily operation of an enterprise, such as ERP (Enterprise Resource Planning) systems, which store original purchase orders, receiving slips, invoices, and other data.

[0028] A data warehouse is a standalone, subject-oriented database used to store cleaned, transformed, and computed historical purchasing metrics, providing a unified data source for analytical decision-making.

[0029] The rules engine is a standalone software component used to load, manage, and execute sets of digital business rules. It decouples complex business logic from application code, enabling dynamic configuration and management of rules.

[0030] This invention provides an intelligent procurement packaging method and system based on historical procurement data. The method includes: a data preparation step, which involves obtaining procurement data from the purchase order table and material master data of the business system, generating and storing historical procurement indicators in a data warehouse; an indicator generation step, which generates procurement indicator data based on the historical procurement indicators by executing preset indicator calculation rules, wherein the procurement indicator data includes at least the benchmark procurement price of the materials; an intelligent packaging step, which involves obtaining a procurement packaging request containing a list of materials to be procured, reading the corresponding historical procurement indicators and procurement indicator data from the data warehouse, grouping the list of materials to be procured according to preset multi-level processing logic, generating candidate material packages, and matching a corresponding procurement strategy to each candidate material package to form a final procurement material package; and a request for quotation generation step, which generates a request for quotation based on the procurement indicator data, the final procurement material package and its corresponding procurement strategy, and outputs it to the business system. This invention, by constructing a historical procurement indicator data warehouse and introducing a rule engine, realizes automatic packaging of procurement needs and intelligent matching of procurement strategies, improving procurement efficiency, reducing the burden on the business system, and ensuring the consistency and scientific nature of procurement decisions.

[0031] Example 1: Figure 1 This is a flowchart of an intelligent procurement and packaging method based on historical procurement data, as described in an embodiment of the present invention.

[0032] like Figure 1 As shown, this embodiment provides an intelligent procurement packaging method based on historical procurement data, including the following sub-steps: Data preparation steps: Obtain procurement data from the purchase order table and material master data of the business system, generate and store historical procurement indicators in the data warehouse based on the procurement data, and the historical procurement indicators should at least include historical price data of materials.

[0033] Specifically, new or changed procurement data is extracted from the business system according to a preset cycle. The procurement data is cleaned, including removing test data, standardizing the format, and filling missing values ​​with preset default values. The cleaned procurement data is then verified for completeness. After the verification is passed, historical procurement indicators are calculated and generated according to the materials. The historical procurement indicators also include historical data on the procurement quantity of materials. The historical procurement indicators are stored in the data warehouse according to the material dimension.

[0034] In this embodiment, an ETL task is initiated every night at 2:00 AM according to a preset cycle. New or status-changed purchase data for the day is extracted from the purchase order table of the business system. The extracted purchase data is then processed, including: Remove test data: Delete records whose purchase order numbers begin with TEST; Standardize the format: unify the date format to YYYY-MM-DD, and unify the unit of amount to yuan; Missing value imputation: For missing supplier IDs, fill in the default value UNKNOWN; for missing material descriptions, complete them from the material master data.

[0035] Perform an integrity check on the cleaned procurement data, checking whether all required fields (material code, procurement quantity, procurement unit price, supplier ID) have values. The pass rate of the check must reach 100%; otherwise, return to the previous step and reprocess.

[0036] For non-new materials with historical purchase records in the business system, the corresponding calculation rules are selected from the preset calculation rule library according to the material category to calculate the historical average purchase price and historical purchase quantity of the material. The calculation rule library includes arithmetic mean method, weighted moving average method, and median method. For new materials without historical purchase records, the estimated usage is read from the material master data as the historical purchase quantity, and the estimated usage is marked. The estimated usage is determined based on the material's production plan requirements or the usage of similar materials in the past.

[0037] In this embodiment, the calculation of historical procurement indicators includes: for material type A chips, which have appeared in historical purchase orders, based on their material classification: electronic components, the weighted moving average method rule is called from the calculation rule library to calculate their historical average purchase price as 2.5 yuan / piece. Simultaneously, the historical purchase frequency of this material is calculated to be 15 times, with 3 suppliers involved, and the historical average purchase quantity is 5000 pieces / order.

[0038] For Material B, a new type of display screen, no record was found in the historical orders, indicating it was a new material. Therefore, its estimated usage was retrieved from the material master data. This estimated usage was determined to be 2000 units / month based on the production department's production plan for the next quarter. This value was used as its historical purchase quantity, and the data source field of this record was marked as estimated in the data warehouse.

[0039] The calculated historical procurement metrics are stored in the historical procurement metrics table of the data warehouse, using the material code as the primary key.

[0040] Indicator generation steps: Based on historical procurement indicators, execute preset indicator calculation rules to generate procurement indicator data to guide procurement packaging. The procurement indicator data must include at least the benchmark price of the materials.

[0041] Specifically, once new historical procurement metrics have been written to the data warehouse, a data update event is sent via a message queue, triggering the start of the rules engine. The rule engine has a pre-set set of digital business rules, which includes at least indicator calculation rules, strategy matching rules, and quotation generation rules. The rule engine is implemented using Drools, EasyRules, or a self-developed rule engine. The rules engine uses new historical procurement metrics as input facts, executes the metric calculation rules, and generates procurement metric data.

[0042] In this embodiment, after the data warehouse completes the writing of new data, it sends a procurement history indicator update event through a message queue. This event triggers the start of the rules engine. This embodiment uses the Drools rules engine, which has a pre-set set of digital business rules.

[0043] Furthermore, the rules for calculating the indicators include the rules for calculating the benchmark purchase price, which include: Extract the most recent purchase transaction price and the historical average purchase price from historical price data; Based on the category of the material, the corresponding weight coefficient configuration is obtained from the rule engine. The weight coefficient is a value between 0 and 1, and the sum of the weight coefficients of the most recent purchase transaction price and the historical average purchase price is 1. Calculate the benchmark purchase price using the weighted summation formula: ,in, To set a benchmark price for procurement, The price is the most recent purchase transaction price. The historical average purchase price The weighting coefficient is the most recent purchase transaction price. The weighting coefficients for different material categories are dynamically adjusted in the rule engine through the configuration interface.

[0044] In this embodiment, the rule engine uses new historical procurement indicators as input facts to execute indicator calculation rules. The calculation rules for the benchmark procurement price are specified as follows: For electronic component materials, the weighting coefficient is configured as follows: weight of the most recent purchase transaction price w_most_recent = 0.7w Recent value = 0.7, historical average purchase price weight 1 w most recently = 0.31 w recently = 0.3.

[0045] The rules engine reads the historical price data of the A-type chip: the most recent purchase price was 2.6 yuan, and the historical average purchase price was 2.5 yuan. Substituting these into the weighted summation formula: P_benchmark = 2.6 × 0.7 + 2.5 × 0.3 = 2.57 yuan / piece.

[0046] The benchmark price for procurement is stored back into the procurement indicator data table in the data warehouse as new procurement indicator data.

[0047] Intelligent packaging steps: Obtain a purchase packaging request containing a list of materials to be purchased; based on the purchase packaging request, read the corresponding historical purchase indicators and purchase indicator data from the data warehouse; group the list of materials to be purchased according to the preset multi-level processing logic to generate at least one candidate material package; match the corresponding purchase strategy for each candidate material package to form the final purchase material package.

[0048] In this embodiment, the procurement package request includes: a list of materials to be procured, containing 30 items such as 50,000 Type A chips and 2,000 Type B new display screens, with a budgeted unit price set for each material (2.6 yuan and 180 yuan respectively). The request type is a one-time project procurement. Historical procurement indicators (such as the number of historical suppliers) and procurement indicator data (such as benchmark prices) for the materials in the material list are read in batches from the data warehouse.

[0049] Furthermore, the multi-level processing logic includes the following sub-steps: Step a, Material Filtering: According to the preset filtering conditions, remove materials that do not need to participate in the subsequent packaging from the list of materials to be purchased, and obtain the set of materials to be packaged. The filtering conditions include that the material type is one or more of office supplies, labor protection supplies, and fixed assets, or the estimated purchase amount of a single material is lower than the preset threshold. In this embodiment, five items classified as office supplies and labor protection supplies were removed, as these items were procured through a simplified offline process. The remaining 25 items were then added to the packing list.

[0050] Step b, Account Set Grouping: Based on the purchasing account set information, the collection of materials to be packaged is grouped, and materials belonging to the same purchasing account set are grouped together. Each group is an account set-level material collection. In this embodiment, the 25 materials were grouped according to the purchasing account set, and it was found that they belonged to two account sets: the Beijing factory set and the Shanghai factory set. Therefore, we obtained two account set-level material sets: the [Beijing Factory Set] and the [Shanghai Factory Set].

[0051] Step c, Brand Grouping: For each set of materials at the account level, determine the number of brands contained in each set of materials at the account level. If it contains multiple brands, group the materials of the same brand together to form a brand-level material set and proceed to step d; if it contains only a single brand, treat the set of materials at the account level as a candidate material package, package it by account level, and terminate the subsequent grouping process. In this embodiment, the [Beijing Factory Set] is processed: Upon inspection, this set contains three brands: Samsung, Sharp, and BOE. Therefore, it is split into three brand-level material sets: [Beijing Factory - Samsung], [Beijing Factory - Sharp], and [Beijing Factory - BOE].

[0052] Processing the "Shanghai Factory Collection": Upon inspection, all materials within this collection are from the Oriental brand. According to the rules, meaningful brand splitting is not possible. Therefore, the "Shanghai Factory Collection" is directly treated as a candidate material package, packaged by account set, and the subsequent grouping process is terminated.

[0053] Step d, Source Grouping: Determine the number of sourcing sources included in each brand-level material collection. If it includes multiple sourcing sources, group the materials from the same sourcing source together to form a source-level material collection and proceed to step e. If it only includes a single sourcing source, treat the brand-level material collection as a candidate material package, package it by brand, and terminate the subsequent grouping process. In this embodiment, the set

Beijing Factory - Samsung

Beijing Factory - Samsung - Domestic

Beijing Factory - Samsung - Import

[0054] Processing the "Beijing Factory - Sharp" group: All materials in this group are imported. According to the rules, meaningful source separation is not possible. Therefore, the "Beijing Factory - Sharp" group is directly treated as a candidate material package, packaged by brand, and the subsequent grouping process is terminated.

[0055] Processing the "Beijing Factory - BOE" group: All materials in this group are from domestic sources. Therefore, meaningful source separation is not possible. The "Beijing Factory - BOE" group is directly treated as a candidate material package, packaged by brand, and subsequent grouping processes are terminated.

[0056] Step e, Attribute Grouping: Determine the number of material attributes contained in each source region-level material set. If it contains multiple material attributes, group materials with the same attribute together to form an attribute-level material set, and proceed to step f. If it contains only a single material attribute, treat the source region-level material set as a candidate material package, package it by source region, and terminate the subsequent grouping process. In this embodiment, the set

Beijing Factory-Samsung-Domestic

Beijing Factory-Samsung-Domestic-Displays

Beijing Factory-Samsung-Domestic-Power Cords

[0057] The "Beijing Factory - Samsung - Imported" set is processed as follows: This set contains materials with two attributes: displays and driver chips. Therefore, it is split into two attribute-level material sets: "Beijing Factory - Samsung - Imported - Displays" and "Beijing Factory - Samsung - Imported - Driver Chips".

[0058] Step f, candidate package generation: Define each attribute-level material set as a candidate material package, and package it by category.

[0059] In this embodiment, the final candidate material package is: [Beijing Factory - Samsung - Domestic - Display Screen] Package, packaged by product category.

[0060] [Beijing Factory - Samsung - Domestic - Power Cord] Package, packaged by product category.

[0061] [Beijing Factory - Samsung - Import - Display Screen] Package, packaged by product category.

[0062] The package, labeled "[Beijing Factory - Samsung - Imported - Driver Chip]", is packaged by product category.

[0063] [Shanghai Factory] Packages are packed according to account sets.

[0064] [Beijing Factory - Sharp] Bags are packaged by brand.

[0065] [Beijing Factory - BOE] Packages are packaged by brand.

[0066] Ultimately, we obtained a total of 7 candidate material packages with clearly defined packaging methods.

[0067] The query quantum step is performed after generating the candidate material package and before constructing the feature vector: Based on the type of the procurement package request, the query volume calculation rules pre-bound to the type are obtained from the rule engine. The types of procurement package requests include annual framework agreement procurement, one-time project procurement, and emergency replenishment procurement. The query volume calculation rules are executed, and a suggested query volume is calculated for each candidate material package based on historical procurement indicators. The suggested query volume is calculated by multiplying the historical monthly average usage of the material by the number of months in the procurement cycle, or by multiplying the total project demand by a preset coefficient. When the historical procurement metrics used to calculate the suggested inquiry volume meet the preset abnormal trigger conditions, the adjustment rules corresponding to the abnormality are obtained from the rule engine to correct the suggested inquiry volume. Abnormal trigger conditions include historical usage fluctuations exceeding preset thresholds and materials being new products with no historical data.

[0068] In this embodiment, based on the request type: one-time project procurement, the corresponding query volume calculation rule is obtained from the rule engine: suggested query volume = total project demand × 1.2 (with a 20% margin).

[0069] For the Beijing factory - Samsung - imported - display package, which contains multiple displays totaling 100 pieces, the suggested inquiry quantity is 100 × 1.2 = 120 pieces.

[0070] This material has stable historical data, with usage fluctuations of less than 20%, requiring no correction.

[0071] Furthermore, feature vectors are extracted or calculated from the data of materials within each candidate material package to quantify their procurement characteristics. These feature vectors include at least one or more of the following: packaging method, total budget price, benchmark price, and number of historical suppliers. The packaging method is determined during the multi-level grouping process of the candidate material package. The total budget price is calculated by summing the budget unit price and required quantity of each material in the procurement package request across all materials within the candidate material package. The formula is: Total Budget Price = ∑(Budget Unit Price of Material × Required Quantity of Material). The benchmark price is calculated by summing the benchmark price and required quantity of each material in the procurement indicator data across all materials within the candidate material package. The formula is: Benchmark Price = ∑(Benchmark Price of Material × Required Quantity of Material). The number of historical suppliers is calculated based on historical procurement indicators, counting the number of different suppliers that have cooperated with all materials within the candidate material package in the past preset time period. If multiple materials have cooperated with the same supplier, the supplier is counted only once. The feature vector is used as input fact to call the rule engine, which executes the preset strategy matching rules. The rule engine compares the input fact with the condition part of the rule through pattern matching. When all conditions are met, the action part of the rule is triggered to match the corresponding procurement strategy for each candidate material package. The procurement strategy includes one or more of the following: inquiry, bidding, competitive negotiation, and single source. The candidate material package that matches the procurement strategy is output as the final procurement material package.

[0072] In this embodiment, for the candidate material package: Beijing factory - Samsung - import - display screen, the process of constructing and calculating the feature vector is as follows: Packaging method: Since this package will eventually enter step f in the grouping process, the packaging method is to package it by product category; Total budget price: The unit price of the display screen is 195 yuan, and the required quantity is 100 pieces. The total budget price = 195 × 100 = 19500 yuan; Benchmark total price: The benchmark price for display screen procurement is 190 yuan, obtained from the procurement indicator data. Benchmark total price = 190 × 100 = 19000 yuan; Number of historical suppliers: Based on procurement history indicators, statistics were compiled on the different suppliers that have cooperated with all materials in the package (i.e. all displays) in the past year. It was found that the company had cooperated with Shenzhen XX Technology and Hong Kong YY Electronics, with a characteristic value of 2.

[0073] Ultimately, the feature vector of this candidate material package is: Packaging method = "Packaging by category", Total budget price = 19500, Benchmark price = 19000, Number of historical suppliers = 2.

[0074] The feature vector is used as the fact input to the rule engine. The pre-defined strategy matching rule in the rule engine is: if the packaging method is equal to packaging by category or packaging by brand, and the total budget price is greater than or equal to 10,000, and the number of historical suppliers is greater than or equal to 2, then the procurement strategy attribute of the candidate material package is assigned the value of competitive negotiation.

[0075] In this embodiment, when the rule engine is executed, it matches the input facts with the above rule conditions: since all conditions are met, the candidate material package is labeled with a competitive negotiation strategy.

[0076] Following this process, procurement strategies are matched one by one for all candidate material packages, generating eight final procurement material packages with procurement strategy tags. Procurement strategies include inquiry, bidding, competitive negotiation, single source, etc.

[0077] Request for Quotation (RFQ) generation steps: Based on procurement indicator data, the final procurement material package and its corresponding procurement strategy, generate an RFQ and output the RFQ to the business system.

[0078] Specifically, based on the procurement strategy corresponding to each final procurement material package, the relevant procurement indicator data is read from the data warehouse and the rule engine is invoked. The rule engine executes the preset inquiry form generation rules to create an inquiry form. The inquiry form generation rules specify the inquiry form template, required fields, approval process, and potential supplier screening criteria under different procurement strategies. The inquiry form templates include inquiry templates, bidding templates, competitive negotiation templates, and single-source templates, with different formats and field requirements for different templates. The required fields are determined according to the procurement strategy and include one or more of the following: technical parameter confirmation letter, deadline for multiple rounds of quotations, and qualification certificates. The potential supplier screening criteria are determined based on the material category and historical transaction records, including suppliers that match the material category and have a historical cooperation count greater than a preset threshold, or high-quality suppliers screened through supplier profile scoring. The generated inquiry forms are stored in the business system via API interface.

[0079] In this embodiment, the reading strategy for the Beijing factory-Samsung-import-display package is as follows: the procurement strategy is competitive negotiation. The rule engine is invoked to execute the Request for Quotation (RFQ) generation rules. The rules stipulate: Template: Use a competitive negotiation template, including technical terms, commercial terms, and a quotation form format; Required fields: Technical parameter confirmation letter, deadline for multiple rounds of quotations, and qualification certificates; Approval process: Three-level approval from the purchasing manager → legal department → finance department; Potential supplier screening: Select suppliers with matching material categories and ≥2 cooperation times from historical suppliers, resulting in Shenzhen XX Technology and Hong Kong YY Electronics, confirmed through supplier profile scoring (≥80 points). RFQ generation: The rule engine returns structured data, creating an RFQ, filling in the bill of materials, a benchmark total price of 19,000 yuan as a negotiation reference price, a link to the technical parameter confirmation letter attachment, and information such as the approver. The RFQ is written back to the business system via API. The user logs into the business system, sees the RFQ in the to-do list, clicks confirm and send, and begins the formal RFQ process.

[0080] Furthermore, the rules engine provides a management interface for digital business rule sets. The management interface allows for adding, modifying, deleting, and querying rules through a RESTful API or web page. It also supports inputting simulated factual data, executing modified rules, and outputting the rule execution results for users to compare and verify. The verification process does not affect the online operation of the rules. After the rules are verified, the rules engine's hot deployment function reloads the rule file and makes it effective without restarting the system. Subsequent requests will use the new rules after they take effect.

[0081] Example 2: The present invention also provides an intelligent procurement packaging system based on procurement history data. The intelligent procurement packaging system based on procurement history data can be implemented by executing the process steps of the intelligent procurement packaging method based on procurement history data. That is, those skilled in the art can understand the intelligent procurement packaging method based on procurement history data as a preferred embodiment of the intelligent procurement packaging system based on procurement history data.

[0082] Specifically, the intelligent procurement and packaging system based on historical procurement data includes: Data preparation module: Obtain procurement data from the purchase order table and material master data of the business system, generate and store historical procurement indicators based on the procurement data, and the historical procurement indicators shall at least include historical price data of materials; Indicator generation module: Based on historical procurement indicators, execute preset indicator calculation rules to generate procurement indicator data to guide procurement packaging. The procurement indicator data must include at least the benchmark price of the material. The intelligent packaging module obtains a purchase packaging request containing a list of materials to be purchased, reads the corresponding historical purchase indicators and purchase indicator data from the data warehouse based on the purchase packaging request, groups the list of materials to be purchased according to a preset multi-level processing logic, generates at least one candidate material package, matches the corresponding purchase strategy to each candidate material package, and forms the final purchase material package. Inquiry generation module: Based on procurement indicator data, final procurement material packages and their corresponding procurement strategies, it generates inquiry forms and outputs them to the business system.

[0083] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0084] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A smart procurement packaging method based on historical procurement data, characterized in that, Includes the following steps: Data preparation steps: Obtain purchase data from the purchase order table and material master data of the business system, generate and store historical purchase indicators to the data warehouse based on the purchase data, and the historical purchase indicators shall at least include historical price data of materials; Indicator generation steps: Based on the historical procurement indicators, execute the preset indicator calculation rules to generate procurement indicator data to guide procurement packaging. The procurement indicator data shall at least include the benchmark price of the materials. Intelligent packaging steps: Obtain a purchase packaging request containing a list of materials to be purchased; based on the purchase packaging request, read the corresponding historical purchase indicators and purchase indicator data from the data warehouse; group the list of materials to be purchased according to a preset multi-level processing logic to generate at least one candidate material package; match the corresponding purchase strategy to each candidate material package to form the final purchase material package. Inquiry Form Generation Steps: Based on the procurement indicator data, the final procurement material package and its corresponding procurement strategy, generate an inquiry form and output the inquiry form to the business system.

2. The intelligent procurement packaging method based on historical procurement data according to claim 1, characterized in that, In the data preparation step, newly added or changed procurement data is extracted from the business system according to a preset period. The procurement data is cleaned, including removing test data, standardizing the format, and filling missing values ​​with preset default values. The cleaned procurement data is then subjected to integrity verification. After the verification is passed, historical procurement indicators are calculated and generated according to the materials. The historical procurement indicators also include historical data on the procurement quantity of materials. The historical procurement indicators are stored in the data warehouse according to the material dimension.

3. The intelligent procurement packaging method based on historical procurement data according to claim 2, characterized in that, For non-new materials with historical purchase records in the business system, the corresponding calculation rules are selected from the preset calculation rule library according to the material category to which the material belongs, and the historical average purchase price and historical purchase quantity of the material are calculated. The calculation rule library includes arithmetic mean method, weighted moving average method, and median method. For new materials without historical purchase records, the estimated usage is read from the material master data as the historical purchase quantity, and the estimated usage is marked. The estimated usage is determined based on the material's production plan requirements or the usage of similar materials in the past.

4. The intelligent procurement packaging method based on historical procurement data according to claim 1, characterized in that, The indicator generation steps include: after a new historical procurement indicator is written into the data warehouse, a data update event is sent through a message queue to trigger the start of the rule engine; the rule engine has a preset digital business rule set, which includes at least indicator calculation rules, strategy matching rules, and quotation generation rules; the rule engine is implemented using Drools, EasyRules, or a self-developed rule engine; the rule engine uses the new historical procurement indicator as input fact, executes the indicator calculation rules, and generates the procurement indicator data.

5. The intelligent procurement packaging method based on historical procurement data according to claim 4, characterized in that, The indicator calculation rules include a benchmark purchase price calculation rule, which includes: extracting the most recent purchase transaction price and the historical average purchase price from the historical price data; obtaining the corresponding weight coefficient configuration from the rule engine according to the material category, wherein the weight coefficient is a value between 0 and 1, and the sum of the weight coefficients of the most recent purchase transaction price and the historical average purchase price is 1; and calculating the benchmark purchase price according to the weighted summation formula. ,in, To set a benchmark price for procurement, The price is the most recent purchase transaction price. The historical average purchase price The weighting coefficient is the most recent purchase transaction price. The weighting coefficients for different material categories are dynamically adjusted in the rule engine through the configuration interface.

6. The intelligent procurement packaging method based on historical procurement data according to claim 1, characterized in that, In the intelligent packaging step, the multi-level processing logic includes the following sub-steps: Step a, Material Filtering: According to the preset filtering conditions, remove materials that do not need to participate in the subsequent packaging from the list of materials to be purchased, and obtain the set of materials to be packaged. The filtering conditions include that the material type is one or more of office supplies, labor protection supplies, and fixed assets, or the estimated purchase amount of a single material is lower than a preset threshold. Step b, Account Grouping: Based on the purchase account information, the set of materials to be packaged is grouped, and materials belonging to the same purchase account are grouped together. Each group is an account-level material set. Step c, Brand Grouping: For each set of account-level materials, determine the number of brands contained in each set of account-level materials. If it contains multiple brands, group the materials of the same brand together to form a brand-level material set and proceed to step d; if it contains only a single brand, treat the set of account-level materials as a candidate material package, package it by account set, and terminate the subsequent grouping process. Step d, Source Grouping: Determine the number of sourcing sources included in each brand-level material set. If it includes multiple sourcing sources, group the materials from the same sourcing source together to form a source-level material set and proceed to step e. If it includes only a single sourcing source, treat the brand-level material set as a candidate material package, package it by brand, and terminate the subsequent grouping process. Step e, Attribute Grouping: Determine the number of material attributes contained in each source region-level material set. If it contains multiple material attributes, group materials with the same attribute together to form an attribute-level material set, and proceed to step f. If it contains only a single material attribute, treat the source region-level material set as a candidate material package, package it by source region, and terminate the subsequent grouping process. Step f, candidate package generation: Define each attribute-level material set as a candidate material package, and package it by category.

7. The intelligent procurement packaging method based on historical procurement data according to claim 6, characterized in that, In the intelligent packaging step, a corresponding feature vector for quantitatively describing the procurement characteristics is extracted or calculated from the data of materials in each candidate material package. The feature vector includes at least one or more of the following: packaging method, total budget price, benchmark total price, and number of historical suppliers. The packaging method is determined according to the candidate material package during the multi-level grouping process. The total budget price is calculated by summing all materials in the candidate material package based on the budget unit price and required quantity of each material in the procurement packaging request. The calculation formula is: Total Budget Price = ∑(Budget Unit Price of Material × Required Quantity of Material). The benchmark total price is calculated by summing all materials in the candidate material package based on the benchmark price and required quantity of each material in the procurement indicator data. The calculation formula is: Benchmark Total Price = ∑(Benchmark Price of Material × Required Quantity of Material). The number of historical suppliers is calculated based on the procurement history indicators, counting the number of different suppliers that all materials in the candidate material package have cooperated with in the past preset time period. If multiple materials have cooperated with the same supplier, the supplier is counted only once. The feature vector is used as input fact to call the rule engine, which executes preset strategy matching rules. The rule engine matches the corresponding procurement strategy for each candidate material package through pattern matching. The procurement strategy includes one or more of the following: inquiry, bidding, competitive negotiation, and single source. The candidate material package that matches the procurement strategy is output as the final procurement material package.

8. The intelligent procurement packaging method based on historical procurement data according to claim 6, characterized in that, It also includes a query quantum step, performed after the candidate material package is generated and before the feature vector is constructed: Based on the type of the procurement package request, the query volume calculation rules pre-bound to the type are obtained from the rule engine. The types of procurement package requests include annual framework agreement procurement, one-time project procurement, and emergency replenishment procurement. The inquiry volume calculation rule is executed, and a suggested inquiry volume is calculated for each candidate material package based on the historical procurement indicators. The suggested inquiry volume is calculated by multiplying the historical monthly average usage of the material by the number of months in the procurement cycle, or by multiplying the total project demand by a preset coefficient. When the historical procurement indicators used to calculate the suggested inquiry volume meet the preset abnormal triggering conditions, the adjustment rules corresponding to the abnormality are obtained from the rule engine to correct the suggested inquiry volume. The abnormal triggering conditions include historical usage fluctuations exceeding a preset threshold and the material being a new product with no historical data.

9. The intelligent procurement packaging method based on historical procurement data according to claim 1, characterized in that, In the process of generating the inquiry form, based on the procurement strategy corresponding to each final procurement material package, the corresponding procurement indicator data is read from the data warehouse and the rule engine is invoked. The rule engine executes preset query form generation rules to create query forms. These rules specify query form templates, required fields, approval processes, and potential supplier screening criteria for different procurement strategies. The query form templates include query templates, tender templates, competitive negotiation templates, and single-source templates, each with different formats and field requirements. The required fields are determined based on the procurement strategy and include one or more of the following: technical parameter confirmation letter, deadlines for multiple rounds of quotations, and qualification certificates. The potential supplier screening criteria are determined based on material categories and historical transaction records, including suppliers matching the material category and with a historical cooperation count exceeding a preset threshold, or high-quality suppliers selected through supplier profile scoring. The generated quotation form is stored in the business system via the API interface; The rule engine provides a management interface for the digital business rule set. The management interface allows for adding, modifying, deleting, and querying rules via a RESTful API or a web page. It also supports inputting simulated factual data, executing modified rules, and outputting the rule execution results for users to compare and verify. The verification process does not affect the online operation of the rules. After verification, the rule file is reloaded and takes effect without restarting the system through the rule engine's hot deployment function. Subsequent requests will use the new rules after the rules take effect.

10. An intelligent procurement packaging system based on historical procurement data, employing the intelligent procurement packaging method based on historical procurement data as described in any one of claims 1-9, characterized in that, include: Data preparation module: Obtains purchase data from the purchase order table and material master data of the business system, generates and stores historical purchase indicators to the data warehouse based on the purchase data, and the historical purchase indicators include at least the historical price data of the materials; Indicator generation module: Based on the historical procurement indicators, execute preset indicator calculation rules to generate procurement indicator data for guiding procurement packaging. The procurement indicator data includes at least the benchmark price of the materials. The intelligent packaging module obtains a purchase packaging request containing a list of materials to be purchased, reads the corresponding historical purchase indicators and purchase indicator data from the data warehouse based on the purchase packaging request, groups the list of materials to be purchased according to a preset multi-level processing logic, generates at least one candidate material package, and matches a corresponding purchase strategy to each candidate material package to form the final purchase material package. Inquiry Form Generation Module: Based on the procurement indicator data, the final procurement material package and its corresponding procurement strategy, generate an inquiry form and output the inquiry form to the business system.

Images