A method and apparatus for generating a purchase order

Abstract

The application provides a procurement order generation method and device, relates to the technical field of enterprise procurement management, and comprises the following steps: step 1, receiving a demand plan containing material information, analyzing a procurement strategy identifier associated with the material information, determining a procurement strategy type, and the procurement strategy type comprises a main supplier strategy, a lowest price strategy, a performance priority strategy and a procurement cycle priority strategy; step 2, based on the procurement strategy type, obtaining associated supplier data sets, converting each supplier into a point cloud, and each point cloud containing three-dimensional attributes, including a quotation, a performance score and a procurement cycle. The application realizes intelligent matching and checking of suppliers by analyzing the procurement strategy identifier, constructing a three-dimensional point cloud model, dynamically segmenting a strategy decision virtual surface and generating an adjustment value to correct parameters, so that procurement orders can be automatically generated and subsequent processes can be completed, and the procurement efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of enterprise procurement management, in particular to a procurement order generation method and device. BACKGROUND

[0002] In the traditional enterprise procurement process, the generation of procurement orders often requires human intervention. Specifically, the staff needs to manually complete supplier selection and price determination (or through a sourcing method) according to the demand plan. In addition, although some procurement management systems on the market attempt to introduce automatic processing functions, most of these systems have the problem of single function, and lack flexibility. It is difficult to intelligently match according to the actual procurement strategy of the enterprise, which is specifically reflected in the following aspects:

[0003] Most existing systems cannot completely realize the automatic matching of suppliers and prices based on procurement strategies, and there is room for improvement in the automation degree of strategy execution. Due to the high proportion of manual operation, the efficiency of the procurement process is easily affected by human factors, and in large-scale procurement scenarios, efficiency may lag. In the procurement decision-making process, human factors may affect the results, making it difficult to fully guarantee the objectivity of the decision in some scenarios. SUMMARY

[0004] The technical problem to be solved by the present application is to provide a procurement order generation method and device to realize multi-dimensional intelligent matching and verification of suppliers.

[0005] To solve the above technical problems, the technical solution of the present application is as follows:

[0006] In a first aspect, a procurement order generation method is provided, which comprises:

[0007] Step 1: receiving a demand plan containing material information, analyzing the procurement strategy identifier associated with the material information, determining the procurement strategy type, and the procurement strategy type includes the main supplier strategy, the lowest price strategy, the performance priority strategy and the procurement cycle priority strategy;

[0008] Step 2: based on the procurement strategy type, obtaining the associated supplier dataset, converting each supplier into a point cloud, and each point cloud contains three-dimensional attributes including bid, performance score and procurement cycle;

[0009] Step 3: based on each point cloud, constructing a strategy decision virtual surface with bid as the horizontal axis, performance score as the vertical axis, and procurement cycle as the vertical axis. On the virtual surface, divide the structure area according to the procurement strategy type, and dynamically segment the area according to the point cloud distribution density to obtain the segmentation result;

[0010] Step 4: based on the segmentation result, mapping the supplier point cloud to the corresponding area, and generating a strategy adjustment value according to the point cloud distribution characteristics in each area;

[0011] Step 5, modify the procurement strategy parameters by adjusting the strategy value, execute supplier matching based on the modified strategy parameters, and perform a verification operation on the matched suppliers, including supplier qualification validity check, price validity period verification, and supplier blacklist screening;

[0012] Step 6, based on the verified suppliers and prices, generate a purchase order, and send a confirmation notice to the supplier according to the purchase order, update the procurement ledger, and freeze the budget occupancy quota.

[0013] Further, receive a demand plan containing material information, parse the procurement strategy identifier associated with the material information, determine the procurement strategy type, which includes the main supplier strategy, the lowest price strategy, the performance priority strategy, and the procurement cycle priority strategy, including:

[0014] Parse the material code in the demand plan and extract the associated procurement strategy identifier;

[0015] Based on the procurement strategy identifier, query the preset strategy mapping table to determine the procurement strategy type, wherein identifier PS1 corresponds to the main supplier strategy; identifier PS2 corresponds to the lowest price strategy; identifier PS3 corresponds to the performance priority strategy; and identifier PS4 corresponds to the procurement cycle priority strategy.

[0016] Further, based on the procurement strategy type, obtain the associated supplier dataset, and convert each supplier into a point cloud, each point cloud containing three-dimensional attributes including bid, performance score, and procurement cycle, including:

[0017] Based on the procurement strategy type, obtain the associated supplier dataset, if it is the main supplier strategy, access the source list database according to the material information, extract the preset main supplier number and the bid within the valid period; if it is the lowest price strategy, query the price master data warehouse according to the material information to obtain all valid bid lists; if it is the performance priority strategy, query the price master data warehouse according to the material information to obtain all valid bid lists; if it is the procurement cycle priority strategy, query the price master data warehouse according to the material information to obtain all valid bid lists;

[0018] Convert the supplier dataset into a point cloud set, each supplier corresponding to a point cloud, and the point cloud containing three-dimensional attributes.

[0019] Further, based on each point cloud, construct a strategy decision virtual surface with bid as the horizontal axis, performance score as the vertical axis, and procurement cycle as the vertical axis, on the virtual surface, divide the structure area according to the procurement strategy type, and dynamically segment the area according to the point cloud distribution density to obtain the segmentation result, including:

[0020] The virtual surface for strategy decision is constructed based on the point cloud set, a three-dimensional coordinate system is established with the bid as the horizontal axis, the performance score as the vertical axis and the procurement cycle as the vertical axis, and a convex hull surface covering all the point clouds is generated as the virtual surface for strategy decision through convex hull calculation of all the supplier point clouds in the coordinate system;

[0021] Based on the virtual surface for strategy decision, a structure-shaped region division rule is determined according to the procurement strategy type, and the structure-shaped region division rule includes:

[0022] When the procurement strategy is the main supplier strategy, a preset bid interval is divided as a rectangular region on the virtual surface;

[0023] When the procurement strategy is the lowest price strategy, a preset bid threshold interval is divided as a trapezoidal region at the bottom of the virtual surface;

[0024] When the procurement strategy is the performance priority strategy, a region with a performance score not lower than a preset score threshold is divided as a polygonal region at the top of the virtual surface;

[0025] When the procurement strategy is the procurement cycle priority strategy, a region with a procurement cycle not exceeding a preset cycle threshold is divided as a columnar region at the front end of the virtual surface;

[0026] For the structure-shaped region, the density distribution value of the point cloud in the region is calculated, and based on the density distribution value, the region is dynamically segmented through a density gradient algorithm, the region with a density distribution value greater than or equal to a threshold value is marked as a high-density core region, and the region with a density distribution value less than the threshold value is marked as a low-density edge region, to obtain a segmentation result.

[0027] Further, based on the segmentation result, the supplier point cloud is mapped to the corresponding region, and a strategy adjustment value is generated according to the point cloud distribution characteristics in each region, including:

[0028] Based on the segmentation result, each supplier point cloud is mapped to the corresponding sub-region, including the point cloud in the core region being marked as a core cloud cluster and the point cloud in the edge region being marked as an edge cloud cluster;

[0029] For all the mapped core cloud cluster point clouds, the average dispersion degree of the distance between the point clouds is calculated as a core distribution parameter, and for all the mapped edge cloud cluster point clouds, the average degree of the position deviation of the point clouds from the center position is calculated as an edge distribution parameter;

[0030] Based on the core distribution parameter and the edge distribution parameter, a strategy adjustment value is generated.

[0031] Further, the procurement strategy parameters are corrected through the strategy adjustment value, the supplier matching is performed based on the corrected strategy parameters, and a verification operation is performed on the matched suppliers, including supplier qualification validity check, price validity period verification and supplier blacklist screening, including:

[0032] Based on the strategy adjustment value, the parameters of the current procurement strategy are dynamically revised, and the dynamic revision includes:

[0033] When the strategy type is the main supplier strategy, the main supplier quotation floating threshold is revised;

[0034] When the strategy type is the lowest price strategy, the lowest quotation acceptance deviation value is revised;

[0035] When the strategy type is the performance priority strategy, the performance score weight ratio is revised;

[0036] When the strategy type is the procurement cycle priority strategy, the procurement cycle priority coefficient is revised;

[0037] According to the revised strategy parameters, the supplier dataset is re-screened to generate a supplier matching sequence optimized according to the strategy;

[0038] For the first supplier in the matching sequence, a verification operation is performed, including supplier qualification validity check, price validity period verification, and supplier blacklist screening.

[0039] Further, based on the verified supplier and price, a procurement order is generated, and according to the procurement order, a confirmation notice is sent to the supplier, the procurement ledger is updated, and the budget occupancy quota is frozen, including:

[0040] Based on the verified supplier and price, a structured procurement order is generated, including supplier code, transaction price, and expected arrival date;

[0041] According to the procurement order, an order confirmation notice is sent to the supplier, the procurement ledger is updated, and the budget occupancy quota is frozen.

[0042] In a second aspect, a procurement order generation device includes:

[0043] A receiving and parsing module is configured to receive a demand plan containing material information, parse the procurement strategy identifier associated with the material information, and determine the procurement strategy type;

[0044] A point cloud generation module is configured to obtain an associated supplier dataset based on the procurement strategy type, and convert each supplier into a point cloud, each point cloud containing three-dimensional attributes including quotation, performance score as the vertical axis, and procurement cycle as the vertical axis;

[0045] A segmentation module is configured to construct a strategy decision virtual surface based on each point cloud, with quotation as the horizontal axis, performance score as the vertical axis, and procurement cycle as the vertical axis, divide the structure shape area on the virtual surface according to the procurement strategy type, and dynamically segment the area according to the point cloud distribution density to obtain a segmentation result;

[0046] The mapping adjustment module is configured to map the supplier point cloud to a corresponding area based on the segmentation result, and generate a strategy adjustment value according to the point cloud distribution characteristics in each area.

[0047] The checking module is configured to correct the procurement strategy parameters by the strategy adjustment value, perform supplier matching based on the corrected strategy parameters, and perform a checking operation on the matched suppliers, including supplier qualification validity checking, price validity period verification and supplier blacklist screening.

[0048] The order processing module is configured to generate a procurement order based on the suppliers and prices that pass the checking, and send a confirmation notice to the suppliers, update the procurement ledger and freeze the budget occupancy quota according to the procurement order.

[0049] In a third aspect, a computer-readable storage medium stores a program, which, when executed by a processor, implements the method.

[0050] The above-mentioned scheme of the present application at least has the following beneficial effects:

[0051] By analyzing the procurement strategy identifier (such as PS1-PS4) in the demand plan, the main supplier, the lowest price, the performance priority or the procurement cycle priority strategy can be automatically matched, without manual intervention in strategy selection, solving the problem of low efficiency of manual decision-making in traditional processes. The supplier data is converted into a three-dimensional point cloud containing bid, performance score and procurement cycle, and multi-dimensional data visualization analysis is realized through the strategy decision virtual surface, compared with the traditional single dimension (such as only according to price or performance) screening of suppliers. The virtual surface is divided according to different strategy types (such as rectangular, trapezoidal and polygonal areas), and the core area and the edge area are dynamically segmented according to the point cloud density, so that the system can automatically adjust the strategy parameters (such as bid threshold, performance weight) according to the distribution characteristics of the supplier data, and adapt to the demand changes in different procurement scenarios.

[0052] The strategy adjustment value is generated according to the point cloud distribution characteristics, and the strategy parameters (such as main supplier bid floating threshold and lowest price deviation value) are dynamically corrected, avoiding the limitations of traditional fixed parameter strategy and improving the accuracy of supplier matching. The matched suppliers are subjected to three checks of qualification validity, price validity period and blacklist screening, reducing the decision bias caused by human factors and being more objective and reliable than traditional manual screening. The strategy adjustment value is generated based on the distribution characteristics (such as dispersion degree and deviation from center degree) of the three-dimensional point cloud, so that the procurement decision is based on quantitative data rather than subjective experience, improving the scientific nature of the decision. From demand analysis, supplier matching to order generation, ledger update and budget freezing, end-to-end automation is realized, the procurement cycle is shortened, and the problem of low efficiency of traditional manual process is solved. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1is a flowchart of a purchase order generation method provided by an embodiment of the present application.

[0054] Figure 2 is a schematic diagram of a purchase order generation device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0055] Exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it is understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly understood, and the scope of the present disclosure can be accurately conveyed to those skilled in the art.

[0056] As Figure 1 shown, an embodiment of the present application proposes a purchase order generation method, which comprises the following steps:

[0057] Step 1, receiving a demand plan containing material information, parsing the procurement strategy identifier associated with the material information, determining the procurement strategy type, the procurement strategy type including the main supplier strategy, the lowest price strategy, the performance priority strategy and the procurement cycle priority strategy;

[0058] Step 2, based on the procurement strategy type, obtaining the associated supplier dataset, converting each supplier into a point cloud, each point cloud containing three-dimensional attributes including bid, performance score and procurement cycle;

[0059] Step 3, based on each point cloud, constructing a strategy decision virtual surface with bid as the horizontal axis, performance score as the vertical axis and procurement cycle as the vertical axis, on the virtual surface, dividing the structure shape area according to the procurement strategy type, and dynamically segmenting the area according to the point cloud distribution density to obtain the segmentation result;

[0060] Step 4, based on the segmentation result, mapping the supplier point cloud to the corresponding area, and generating a strategy adjustment value according to the point cloud distribution characteristics in each area;

[0061] Step 5, correcting the procurement strategy parameters through the strategy adjustment value, executing supplier matching based on the corrected strategy parameters, and performing verification operations on the matched suppliers, including supplier qualification validity check, price validity period verification and supplier blacklist screening;

[0062] Step 6, based on the suppliers and prices that pass the verification, generating a purchase order, and sending a confirmation notice to the suppliers according to the purchase order, updating the procurement account and freezing the budget occupancy quota.

[0063] In the embodiment of the present application, by analyzing the strategy identification (such as PS1-PS4) in the demand plan, the main supplier, the lowest price and other strategies are automatically matched, without manual judgment, reducing the time consumption of the process. Covering four strategies such as main supplier and performance priority, adapting to different material procurement scenarios (such as strategic materials giving priority to main suppliers and general materials giving priority to low prices), avoiding the limitations of traditional single strategy. The supplier is converted into a three-dimensional point cloud containing quotation, performance and cycle, breaking the one-sidedness of traditional single index (such as only looking at price), and realizing comprehensive strength evaluation. A decision virtual surface is constructed through a three-dimensional coordinate system, and the area is divided according to the strategy type (such as the bottom quotation threshold area corresponding to the lowest price strategy), so that the supplier screening logic is visualized and quantified. According to the distribution density (core area / edge area) of the point cloud on the virtual surface, the strategy parameters (such as the main supplier quotation floating threshold) are automatically adjusted to adapt to market fluctuations and supplier data changes.

[0064] The strategy adjustment value is generated based on the dispersion degree of the point cloud in the area, avoiding the matching deviation caused by fixed parameters (such as dynamic correction of score weight in the performance priority strategy). The qualification, price validity period and blacklist screening of the matched supplier are performed to reduce compliance risks and make the manual review more comprehensive and objective. From order generation to notification of suppliers, updating the account book and freezing the budget, it is automatically completed, reducing human operation errors and ensuring real-time data synchronization. In addition to the existing four strategies, new strategies can be quickly integrated by adding virtual surface area division rules (such as adding supplier credit level dimension).

[0065] In a preferred embodiment of the present application, the above step 1, receiving a demand plan containing material information, analyzing the procurement strategy identifier associated with the material information, determining the procurement strategy type, the procurement strategy type includes main supplier strategy, lowest price strategy, performance priority strategy and procurement cycle priority strategy, which can include:

[0066] Step 100, analyzing the material code in the demand plan and extracting the associated procurement strategy identifier;

[0067] Step 101, based on the procurement strategy identifier, querying the preset strategy mapping table to determine the procurement strategy type, wherein identifier PS1 corresponds to the main supplier strategy; identifier PS2 corresponds to the lowest price strategy; identifier PS3 corresponds to the performance priority strategy; and identifier PS4 corresponds to the procurement cycle priority strategy.

[0068] In the embodiments of the present application, first, a demand plan containing material information is received through an API interface or a file import mode (such as JSON, XML format data or database records), for example, the received demand plan can contain fields such as material_code (material code), quantity (quantity), delivery_date (delivery date) and the like. The demand plan is structurally parsed, and the material_code field (for example, the material code is "M001") is located, and the code is the unique identifier of the associated procurement strategy. The system takes the material_code as a query condition, and accesses an enterprise material master database or a configuration table (such as a Material_Strategy_Mapping table). In the database, each material record corresponds to a strategy_identifier field (that is, a procurement strategy identifier), for example:

[0069] The material code "M001" corresponds to the strategy_identifier "PS3";

[0070] The material code "M002" corresponds to the strategy_identifier "PS2".

[0071] The record of the material code is quickly located through indexing, and the strategy_identifier value (such as "PS3") is extracted.

[0072] Step 101, the preset strategy mapping table (such as Strategy_Mapping_Table) is a static mapping relationship table configured when the system is initialized, and stores the corresponding relationship between the identifier and the strategy type in the form of key-value pairs, for example:

[0073] The identifier PS1 corresponds to the strategy type of the main supplier strategy;

[0074] The identifier PS2 corresponds to the strategy type of the lowest price strategy;

[0075] The identifier PS3 corresponds to the strategy type of the performance priority strategy;

[0076] The identifier PS4 corresponds to the strategy type of the procurement cycle priority strategy.

[0077] The table is stored in a system configuration file or a database, and can be manually maintained and updated by an administrator. The strategy_identifier (such as "PS3") extracted from the material code is taken as a query key value. In the Strategy_Mapping_Table, the matching key value is searched through iteration or hash:

[0078] If the identifier is "PS1", the "main supplier strategy" is matched;

[0079] If "PS3", match "Performance Priority Strategy".

[0080] After matching, return the corresponding strategy type string as the decision-making basis for the current material procurement.

[0081] In a preferred embodiment of the present application, step 2, obtaining the associated supplier dataset, converting each supplier into a point cloud, each point cloud containing three-dimensional attributes including quotation, performance score and procurement cycle, can include:

[0082] Step 200, based on the procurement strategy type, obtain the associated supplier dataset, if it is a primary supplier strategy, access the source list database according to the material information to extract the preset primary supplier number and the quotation within the valid period; if it is a lowest price strategy, query the price main data warehouse according to the material information to obtain all valid quotation lists; if it is a performance priority strategy, query the price main data warehouse according to the material information to obtain all valid quotation lists; if it is a procurement cycle priority strategy, query the price main data warehouse according to the material information to obtain all valid quotation lists;

[0083] Step 201, convert the supplier dataset into a point cloud set, each supplier corresponds to a point cloud, and the point cloud contains three-dimensional attributes.

[0084] In an embodiment of the present application, the material code (such as "M001") in the demand plan is received, the code is used as a key field, and the enterprise source list database (Source_List_DB) is accessed. The database stores the mapping relationship between the material and the primary supplier, as well as the quota proportion, price validity period and other information of the supplier.

[0085] Primary supplier screening logic:

[0086] Query the record corresponding to material_code from Source_List_DB, extract the primary_supplier_id (primary supplier number) and price_validity_period (price validity period) fields. Compare the current date of the system (such as 2025-06-23) with the start / end date of price_validity_period, and only keep the quotation data within the valid period. For example, if the primary supplier quotation is valid from 2025-06-01 to 2025-07-31, the quotation is valid. Return a structured record containing supplier_id, material_price, validity_start, and validity_end.

[0087] Data acquisition process of lowest price strategy / performance priority strategy / procurement cycle priority strategy:

[0088] Access the Price_Master_DB, which stores all valid quotes from suppliers for the material and related attributes, based on the material code. Filter the Supplier_Qualification_DB to exclude suppliers with expired licenses or violations. Extract quote records from Price_Master_DB with a price_validity_date greater than or equal to the current system date. Interface with the Risk_Control_System to exclude suppliers on the blacklist.

[0089] Lowest Price Strategy: Extract all valid quotes, retaining the supplier_id, material_price, performance_score, and lead_time fields.

[0090] Performance First / Lead Time First Strategy: Extract the performance_score or lead_time field in addition to the filtered records.

[0091] Step 201: Create a three-dimensional point cloud object for each supplier, defined as PointCloud=(x,y,z), where:

[0092] x-axis corresponds to the quote (material_price) in Chinese Yuan;

[0093] y-axis corresponds to the performance score (performance_score) ranging from 0 to 100 points;

[0094] z-axis corresponds to the lead time (lead_time) in natural days.

[0095] Point cloud object additional attributes: supplier_id, material_code, data_source.

[0096] Convert string-type prices (e.g., "100.00") to floating-point numbers, performance scores (e.g., "95") to integers, and lead times (e.g., "7 days") to integers by extracting the numerical part.

[0097] Outlier handling:

[0098] If the quote is 0 or negative, fill in the historical average price for the material by default.

[0099] If the performance score is missing, it is marked as "to be evaluated" and assigned a temporary value (e.g. 60 points);

[0100] If the procurement cycle exceeds the industry's maximum value (e.g. 180 days), it is considered invalid data and excluded.

[0101] Traverse the supplier dataset and generate a point cloud object for each record. For example, the supplier "S001" has a quote of 100 yuan, a performance score of 95 points, and a procurement cycle of 7 days. The corresponding point cloud coordinates are (100, 95, 7), stored in the list = [PointCloud1, PointCloud2,...], and the set is stored in the memory cache or temporary data table.

[0102] For different procurement strategies (main supplier / lowest price, etc.), automatically call the corresponding data source (inventory / price master data) to avoid manual switching of data query paths and improve process efficiency. For example, the main supplier strategy can directly obtain the preset supplier, reducing the screening time. Convert the supplier key indicators (quote, performance, cycle) into a three-dimensional point cloud to break through the limitations of traditional two-dimensional or single-dimensional evaluation. For example, the performance priority strategy can intuitively filter high-score suppliers through the y-axis coordinates of the point cloud, while combining the x-axis price for secondary verification. Embed the qualification, price validity period, and other verification mechanisms during data acquisition to ensure the validity of the point cloud data; synchronize the price master data of the ERP system in real time to avoid decision-making deviations caused by outdated information. Point cloud attributes (such as performance score, procurement cycle) can be updated in real time with dynamic supplier data, allowing the system to automatically adjust strategy parameters based on market changes. For example, when a supplier's performance decreases, its point cloud position in the virtual plane will dynamically migrate, triggering a strategy re-evaluation.

[0103] In a preferred embodiment of the present application, step 3 above, based on each point cloud, constructs a strategy decision virtual plane with the quote as the horizontal axis, the performance score as the vertical axis, and the procurement cycle as the vertical axis. On the virtual plane, according to the procurement strategy type, divide the structure shape area, and dynamically segment the area according to the point cloud distribution density to obtain the segmentation result, which can include:

[0104] Step 300, based on the point cloud set, constructs a strategy decision virtual plane with the quote as the horizontal axis, the performance score as the vertical axis, and the procurement cycle as the vertical axis to establish a three-dimensional coordinate system. In the coordinate system, perform convex hull surface calculation on all supplier point clouds to generate a convex hull surface that covers all point clouds as the strategy decision virtual plane;

[0105] Step 301, based on the strategy decision virtual plane, determine the structure shape area division rule according to the procurement strategy type, which includes:

[0106] When the procurement strategy is the main supplier strategy, define the preset quote interval as a rectangular area on the virtual plane;

[0107] When the procurement strategy is the lowest price strategy, a preset bid threshold interval is demarcated as a trapezoidal area at the bottom of the virtual surface;

[0108] When the procurement strategy is the performance priority strategy, an area with a performance score not lower than a preset score threshold is demarcated as a polygonal area at the top of the virtual surface;

[0109] When the procurement strategy is the procurement cycle priority strategy, an area with a procurement cycle not exceeding a preset cycle threshold is demarcated as a columnar area at the front end of the virtual surface;

[0110] In step 302, for the structure-shaped area, the point cloud density distribution value in the area is calculated, and based on the density distribution value, the area is dynamically segmented by a density gradient algorithm, the area with a density distribution value greater than or equal to a threshold value is marked as a high-density core area, and the area with a density distribution value less than the threshold value is marked as a low-density edge area, to obtain a segmentation result.

[0111] In the embodiment of the application, a three-dimensional Cartesian coordinate system is established with bid (unit: yuan) as the x-axis, performance score (0-100 points) as the y-axis, and procurement cycle (unit: days) as the z-axis. The point cloud set generated in step 201 (such as Supplier_PointClouds) is mapped into the coordinate system, and each point cloud corresponds to a three-dimensional coordinate point (x, y, z). For example, the supplier point cloud (100, 95, 7) represents a bid of 100 yuan, a performance score of 95 points, and a cycle of 7 days.

[0112] Convex hull surface calculation process:

[0113] The convex hull surface is the smallest convex polyhedron that wraps all point clouds, ensuring that all point clouds are located inside or on the surface of the convex hull surface. The three-dimensional coordinates of all point clouds are extracted to form a point set P={p1, p2,..., pn}. A three-dimensional convex hull algorithm (such as the GiftWrapping algorithm or the QuickHull algorithm) is used to traverse the point set, find all vertices and faces that constitute the convex hull surface, generate a convex hull surface mesh, each face is determined by three vertices, and finally form a closed three-dimensional surface as a virtual surface for strategy decision-making. The subdivision accuracy (such as the number of mesh faces) of the convex hull surface can be set to balance the calculation efficiency and visualization effect. The convex hull surface is rendered as a visualization model for administrators to view the supplier distribution situation, and the geometric parameters (such as vertex coordinates, face equations) of the convex hull surface are stored.

[0114] Step 301, get the valid offer P0 of the main supplier from the source list database, and set the offer interval as [P0×(1-δ), P0×(1+δ)], where δ is the set floating threshold (e.g. δ=10%). In the three-dimensional coordinate system, take the x-axis (offer) as the dimension, and set the interval of x∈[P0×0.9, P0×1.1], and the y-axis and z-axis cover the full range of the convex hull surface, forming a rectangular area perpendicular to the x-axis. Check whether the main supplier point cloud is included in the rectangular area. If not, automatically expand the δ value to 15%.

[0115] Calculate the x-axis (offer) minimum value P_min of all point clouds, and set the threshold as P_min×(1+ε), where ε is the allowed price floating ratio (e.g. ε=15%). At the bottom of the virtual surface (x-axis smaller value area), set the region of x≤P_min×1.15, and form a trapezoidal cross-section in the y-axis and z-axis directions according to the convex hull surface boundary (bottom surface x≤P_min×1.15, top surface x≤P_min×1.15 and y-z plane projection intersection with convex hull). Set the performance score threshold S_threshold (e.g. 80 points), and set the region of y≥S_threshold at the top of the virtual surface (y-axis higher value area). The projection of this region in the y-z plane is the intersection of the convex hull surface and y≥80, forming a polygonal cross-section, and the x-axis direction covers the full range. Set the procurement cycle threshold T_threshold (e.g. 10 days), and set the region of z≤T_threshold at the front end of the virtual surface (z-axis smaller value area). The projection of this region in the x-y plane is the intersection of the convex hull surface and z≤10, and it extends along the z-axis direction to form a cylindrical region.

[0116] Step 302, calculate the number of point clouds in a unit volume using the formula: density = number of point clouds in the region ÷ volume of the region. Divide the structure-shaped region into a number of small cubes (voxels), and set the edge length to be the product of the offer precision (e.g. 1 yuan), performance score precision (e.g. 5 points), and cycle precision (e.g. 1 day). Calculate the number of point clouds in each voxel, and divide the voxel volume by the density value. For each voxel, calculate the density difference with the adjacent voxels to form a density gradient field. Set the density threshold D_threshold (e.g. 1.5 times the average density), and mark the voxels with density≥D_threshold as "high-density core area", indicating that the suppliers are densely distributed in this area, which is the preferred selection range. Mark the voxels with density<D_threshold as "low-density edge area", indicating that the suppliers are sparsely distributed, which is the alternative range. Generate a three-dimensional grid model of the core area and the edge area, and store the area label (core / edge) of each point cloud.

[0117] The three-dimensional virtual surface converts abstract procurement strategies into visual spatial areas, such as a trapezoidal area of the lowest price strategy to intuitively display an acceptable low price range, avoiding artificial subjective judgment. The density gradient segmentation can automatically adjust the core area and the edge area according to fluctuations in supplier data, such as when the market price fluctuates, the core area of the lowest price strategy will dynamically migrate to a new low price dense area. The virtual surface simultaneously considers the three-dimensional attributes of the bid, performance, and cycle, such as the performance priority strategy in a high-score area (polygon), and can also combine the bid dimension to filter cost-effective suppliers, avoiding single-index deviation. After the area is segmented, the supplier matching can be directly performed in the core area, reducing the data processing amount, such as the rectangular core area of the main supplier strategy only containing suppliers within the preset bid range, accelerating the matching speed.

[0118] In a preferred embodiment of the present application, step 4, mapping the supplier point cloud to the corresponding area, generating a strategy adjustment value according to the point cloud distribution characteristics in each area, can include:

[0119] Step 400, based on the segmentation result, mapping each supplier point cloud to the corresponding partition, including marking the point cloud located in the core area as a core cloud cluster; marking the point cloud located in the edge area as an edge cloud cluster;

[0120] Step 401, for all mapped core cloud cluster point clouds, calculating the average dispersion degree of the distance between the point clouds as the core distribution parameter; for all mapped edge cloud cluster point clouds, calculating the average degree of the position deviation from the center position as the edge distribution parameter;

[0121] Step 402, generating a strategy adjustment value based on the core distribution parameter and the edge distribution parameter.

[0122] In an embodiment of the present application, the region segmentation result is obtained from the output of step 302, which contains the region label (core area / edge area) and the corresponding geometric coordinate range of each three-dimensional voxel. For example, the coordinate range of the core area voxel is x∈[90,110], y∈[85,100], z∈[5,8], and the label is "core". Traverse the supplier point cloud set generated in step 201 (such as Supplier_PointClouds), and perform the following operations for each point cloud (x, y, z):

[0123] Check whether the point coordinates are located within the coordinate range of any core area voxel. For example, whether the point cloud (100, 95, 7) is within the x∈[90,110], y∈[85,100], z∈[5,8] range of the core area voxel.

[0124] If matched, mark the point cloud as "core cloud cluster"; otherwise, mark it as "edge cloud cluster".

[0125] Add a cluster_label field to each point cloud, with values of "core" or "edge", forming a labeled point cloud set Labeled_PointClouds; if the point cloud coordinates exceed the convex hull surface range (in theory, this should not occur, as the convex hull has wrapped all point clouds), the default label is marked as "edge cloud cluster", and an exception log is generated.

[0126] Step 401, extract all point clouds labeled as "core cloud clusters" to form a subset Core_Clouds, for each pair of point clouds (x1, y1, z1) and (x2, y2, z2) in Core_Clouds, calculate the three-dimensional Euclidean distance, accumulate all inter-point cloud distances, and divide by the number of distance groups to obtain the average value. Extract all "edge cloud cluster" point clouds to form a subset Edge_Clouds, and calculate the center point coordinates (x_center, y_center, z_center):

[0127] x_center = ∑xi / n, y_center = ∑yi / n, z_center = ∑zi / n (n is the number of edge point clouds).

[0128] For each edge point cloud (xi, yi, zi), calculate its three-dimensional Euclidean distance to the center point, accumulate the deviation distances of all edge point clouds, and divide by the number of point clouds n to obtain the average value.

[0129] Step 402, since the core distribution parameters and edge distribution parameters have the same dimension (both in distance units), numerical comparison can be directly performed, if the dimensions are different (such as subsequent expansion of other dimensions), the parameters need to be scaled to the [0, 1] interval through min-max normalization. The ratio of "core dispersion degree" is used as the basis for strategy adjustment value:

[0130] Strategy adjustment value = core distribution parameter ÷ (edge distribution parameter + ε), where ε is a very small value (such as 0.01) to avoid division by zero.

[0131] According to different procurement strategy types, the strategy adjustment value corresponds to different parameter correction directions:

[0132] Main supplier strategy: the strategy adjustment value is positively related to the bid price floating threshold (the larger the strategy adjustment value, the higher the allowed bid price increase ratio);

[0133] Lowest price strategy: the strategy adjustment value is negatively related to the lowest bid acceptance deviation (the larger the strategy adjustment value, the smaller the allowed price deviation);

[0134] Performance / cycle strategy: the strategy adjustment value is positively related to the scoring weight / cycle priority (the larger the strategy adjustment value, the higher the strategy weight of core area suppliers).

[0135] The dynamic division of the core / edge area and the calculation of the distribution parameters enable the system to automatically adjust the strategy parameters according to the real-time fluctuations of the supplier data (such as the addition of new suppliers and price changes), thereby avoiding the matching failure caused by fixed parameters. The adjustment value is generated based on the distribution characteristics (discrete degree and deviation degree) of the point cloud, so that the strategy parameters are more in line with the distribution characteristics of the actual supplier group. For example, if the core area supplier distribution is concentrated (low discrete degree), it indicates that the market quotation is stable, and the price floating threshold can be tightened. When the number of core area suppliers is small (high edge area proportion), the strategy conditions are automatically relaxed (such as expanding the quotation interval) to ensure that there are enough suppliers to choose from; when the core area distribution is dispersed, it is suggested that the strategy may need to be refined (such as increasing the performance score weight). The adjustment value is generated based on the quantified point cloud distribution data, avoiding the subjective bias caused by artificial experience, and making the procurement decision more scientific. For example, the adjustment value of the lowest price strategy can directly reflect the intensity of market price competition. The whole process from point cloud mapping to adjustment value generation is automated, forming a closed loop with the upstream virtual face segmentation and downstream strategy parameter modification, reducing the manual intervention link and improving the procurement process efficiency.

[0136] In a preferred embodiment of the present application, the step 5 of modifying the procurement strategy parameters through the strategy adjustment value, performing supplier matching based on the modified strategy parameters, and performing verification operations on the matched suppliers, including supplier qualification validity check, price validity period verification and supplier blacklist screening, can include:

[0137] Step 500, based on the strategy adjustment value, dynamically modifying the parameters of the current procurement strategy, wherein the dynamic modification includes:

[0138] When the strategy type is the main supplier strategy, the main supplier quotation floating threshold is modified;

[0139] When the strategy type is the lowest price strategy, the lowest quotation acceptance deviation value is modified;

[0140] When the strategy type is the performance priority strategy, the performance score weight ratio is modified;

[0141] When the strategy type is the procurement cycle priority strategy, the procurement cycle priority coefficient is modified;

[0142] Step 501, reselecting the supplier data set according to the modified strategy parameters, and generating a supplier matching sequence optimized according to the strategy;

[0143] Step 502, performing verification operations on the first supplier in the matching sequence, including supplier qualification validity check, price validity period verification and supplier blacklist screening.

[0144] In this embodiment of the present invention, a policy adjustment value (e.g., 0.57) is read from the output of step 402. This value is calculated by the ratio of the core distribution parameter to the edge distribution parameter and reflects the degree of concentration of the supplier point cloud within the virtual surface. The current procurement policy type (e.g., primary supplier policy, lowest price policy, etc.) is identified, and the corresponding parameters to be adjusted are determined.

[0145] Main supplier strategy - quote floating threshold correction:

[0146] The original quote floating threshold δ0 (e.g. 10%) is modified as follows:

[0147] δ1=δ0×(1+k1×strategy adjustment value), where k1 is the correction coefficient (preset to 0.5).

[0148] Lowest Price Strategy - Lowest Bid Acceptance Deviation Value Correction:

[0149] The original deviation value ε0 (such as 15%), the correction formula is:

[0150] ε1=ε0×(1-k2×strategy adjustment value), k2=0.3.

[0151] Performance Priority Strategy - Performance Score Weight Ratio Modification:

[0152] The original weight w0 (such as 60%), the correction formula is:

[0153] w1=w0+k3×strategy adjustment value×100%, k3=0.2.

[0154] Procurement cycle priority strategy-cycle priority coefficient correction:

[0155] The original coefficient c0 (such as 0.7) is corrected as follows:

[0156] c1=c0+k4×strategy adjustment value, k4=0.3.

[0157] The revised parameters must meet the preset boundary conditions (such as the quotation floating threshold ≤ 20%). If they exceed the boundary value, the revised parameters will be stored in the temporary strategy configuration table (Temp_Strategy_Params).

[0158] Step 501: Based on the supplier dataset obtained in step 200 (such as the preset suppliers under the master supplier strategy and all suppliers with valid quotations under the lowest price strategy), perform screening in combination with the revised parameters:

[0159] Main supplier strategy: Filter suppliers whose quotations are within the range of [main supplier base price × (1-δ1), main supplier base price × (1+δ1)].

[0160] Lowest Price Strategy: Filter out suppliers with price ≤ lowest price × (1 + ε1) ;

[0161] Performance / Cycle Strategy: Score and rank suppliers by modified weight ÷ coefficient;

[0162] Primary Supplier Strategy: Only keep pre-set primary suppliers, if its price is within the modified range, rank first, otherwise rank by price closeness;

[0163] Lowest Price Strategy: Sort filtered suppliers by price in ascending order, if price is the same, sort by performance score in descending order;

[0164] Performance First Strategy: Calculate comprehensive score = performance score × w1 + price × (1 - w1), rank by score in descending order, if score is the same, rank by price in ascending order;

[0165] Cycle First Strategy: Calculate comprehensive score = cycle × c1 + price × (1 - c1), rank by score in ascending order, if score is the same, rank by price in ascending order.

[0166] Generate ranked supplier list (e.g. [Supplier A, Supplier B, Supplier C]) and store in memory cache, marked as “to-be-verified sequence”.

[0167] Step 502, access the supplier qualification database (Supplier_Qual_DB) to query the business license number and validity period of the first supplier in the matching sequence, compare whether the current date (e.g. 2025-06-23) is within the validity period of the business license, if it is expired, mark it as “invalid qualification”, if the material involves special industries (e.g. medical devices), verify whether the supplier has relevant certification (e.g. ISO13485), if not, mark it as invalid. Get the supplier's price record for the current material from the price master database (Price_Master_DB), extract price_validity_end_date (price validity end date), if the current date ≤ validity end date, the price is valid; otherwise, trigger the price update process and re-match from the alternative suppliers. Connect to the enterprise risk control system (Risk_Control_System) and input the supplier number to query the blacklist status, if it exists in the blacklist (e.g. historical default record), mark it as “prohibited cooperation” and skip this supplier and verify the next one.

[0168] The policy adjustment value is calculated in real time based on the supplier point cloud distribution, so that the parameter correction can adapt to market fluctuations. For example, when the core area supplier quotation is highly dispersed, the main supplier policy automatically relaxes the quotation threshold to ensure that the supplier can be matched. The corrected parameters are more in line with the actual supplier distribution. For example, the minimum price strategy reduces the deviation value to ensure price advantage while avoiding the risk of extremely low price; the performance strategy increases the weight to preferentially select high-score suppliers and reduce cooperation risk. The three verification mechanisms (qualification, price, and blacklist) ensure that the supplier's qualification is legal, the price is effective, and there is no credit risk, which is more comprehensive than traditional manual review and reduces compliance loopholes. Automatic parameter correction and supplier sorting reduce manual intervention. For example, the performance strategy is automatically sorted according to the corrected weight, without the need for manual adjustment of the scoring rules; when the verification fails, the next supplier is automatically switched to improve process continuity. From point cloud distribution to policy adjustment to supplier matching, a complete data-driven chain is formed, which enables the procurement strategy to be continuously optimized with market changes. For example, when the supplier delivery cycle is generally extended, the cycle priority strategy automatically increases the cycle priority coefficient to ensure timely delivery.

[0169] In a preferred embodiment of the present application, the above step 6, based on the verified supplier and price, generates a purchase order, and sends a confirmation notice to the supplier according to the purchase order, updates the purchase ledger, and freezes the budget occupancy quota, which can include:

[0170] Step 600, based on the verified supplier and price, generates a structured purchase order containing supplier code, transaction price, and expected arrival date;

[0171] Step 601, according to the purchase order, sends an order confirmation notice to the supplier, updates the purchase ledger, and freezes the budget occupancy quota.

[0172] In an embodiment of the present application, the supplier information that passes the verification is obtained from the verification result of step 502, including:

[0173] supplier_code (supplier code, such as "S001"), negotiated_price (transaction price, such as 100.00 yuan), lead_time (procurement cycle, such as 7 days).

[0174] Extract material-related information from demand planning:

[0175] material_code (material code, such as "M001"), required_quantity (quantity required, such as 100 pieces), department_code (demand department code, such as "DPT001").

[0176] Estimated arrival date = current system date + procurement cycle (natural day, you can configure the working day rule to exclude holidays).

[0177] Specific implementation:

[0178] Get the current system date (such as 2025-06-23) and convert it to a date object.

[0179] The arrival date is calculated based on the supplier's lead time (e.g., 7 days): 2025-06-23 + 7 days = 2025-06-30. If the procurement lead time is 0 or a negative number, the default calculation is based on the industry standard period (e.g., 3 days), and a warning log is generated. Structured data is generated in JSON or XML format and converted to a PDF document for download.

[0180] Step 601: Send an order confirmation letter to the supplier via the corporate email system or electronic data interchange (EDI) platform, including:

[0181] Order number, material information, purchase quantity and unit price;

[0182] Estimated arrival date, delivery address and contact information;

[0183] A link or contact information for the supplier's confirmation receipt.

[0184] Paper channel (optional): The system automatically generates a PDF order with an electronic official seal and mails it to the supplier's registered address via the logistics service integrated with the Enterprise Resource Planning (ERP) system.

[0185] The system records the notification sending time, channel, and supplier receipt status in real time. If the supplier does not confirm the order within 24 hours, the system will automatically trigger a second reminder process.

[0186] Procurement Ledger Update Process

[0187] The purchase ledger is stored in a database table (Purchase_Ledger) and contains the following fields: order number, material code, supplier code, order amount, order date, and order status. The order number, amount, date, and other information generated in step 600 are inserted into the purchase ledger table. When the order status subsequently changes (such as shipment or acceptance), the ledger record is updated in real time to ensure data consistency. Updating the ledger and generating the order are included in the same database transaction to avoid data inconsistencies caused by partial operation failures. Access the budget management system and query the corresponding material procurement budget account (such as "Raw Material Procurement - Plastic Parts") based on the demand department code to obtain the remaining available budget and frozen amount for that account.

[0188] The frozen amount is equal to the total order amount. For example, if the total order amount is 10000 yuan, the budget frozen quota is increased by 10000 yuan.

[0189] A database update statement is executed, for example:

[0190] UPDATE budget account SET frozen quota = frozen quota + 10000, remaining available budget = remaining available budget - 10000 WHERE department code = 'DPT001' AND material code = 'M001'.

[0191] If the remaining available budget is less than the total order amount, the system will refuse to generate the order and return an "insufficient budget" error prompt; send a warning notice to the budget administrator to apply for adjusting the budget quota.

[0192] From order generation to notification, account updating, and budget freezing, the whole process is automated, reducing manual operation links (such as manually sending emails and entering account books). Order data is synchronized in real time with the account book and budget system, avoiding decision-making deviations caused by information lag. For example, after budget freezing, other departments can view the remaining quota in real time when applying for the same material procurement. The budget freezing mechanism ensures that procurement expenditure does not exceed the budget, meeting the requirements of enterprise financial control; the account record is traceable, supporting audit verification. Budget shortage, supplier unconfirmed, and other abnormal scenarios automatically trigger early warning, avoiding process stagnation and improving the agility of procurement management.

[0193] As shown in FIG. Figure 2 The embodiment of the present application also provides a procurement order generation device, which comprises:

[0194] A receiving and analyzing module is configured to receive a demand plan containing material information, analyze the procurement strategy identifier associated with the material information, and determine the procurement strategy type;

[0195] A point cloud generation module is configured to obtain a set of associated supplier data based on the procurement strategy type, and convert each supplier into a point cloud, each point cloud containing three-dimensional attributes including a bid price, a performance score as a horizontal axis, and a procurement cycle as a vertical axis;

[0196] A segmentation module is configured to construct a strategy decision virtual surface based on each point cloud, with the bid price as the horizontal axis, the performance score as the vertical axis, and the procurement cycle as the vertical axis, divide the structure into regions on the virtual surface according to the procurement strategy type, and dynamically segment the regions according to the point cloud distribution density to obtain a segmentation result;

[0197] A mapping and adjusting module is configured to map the supplier point cloud to the corresponding region based on the segmentation result, and generate a strategy adjustment value according to the point cloud distribution characteristics in each region;

[0198] The checking module is configured to correct the procurement strategy parameters by the strategy adjustment value, perform supplier matching based on the corrected strategy parameters, and perform a checking operation on the matched suppliers, including supplier qualification validity checking, price validity period verification, and supplier blacklist screening;

[0199] The order processing module is configured to generate a procurement order based on the checked suppliers and prices, and send a confirmation notice to the suppliers, update a procurement account, and freeze a budget occupancy quota according to the procurement order.

[0200] It should be noted that the device is a device corresponding to the above method, and all the implementation manners in the above method embodiment are applicable to this embodiment and can also achieve the same technical effects.

[0201] Embodiments of the present application also provide a computer-readable storage medium storing instructions which, when executed on a computer, cause the computer to perform the method described above. All the implementation manners in the above method embodiment are applicable to this embodiment and can also achieve the same technical effects.

[0202] The above is the preferred embodiment of the present application. It should be noted that for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can be made, which should also be considered within the scope of protection of the present application.

Claims

1. A purchase order generation method characterized by, The method comprises: Step 1, receiving a demand plan containing material information, parsing the procurement strategy identifier associated with the material information, determining the procurement strategy type, and the procurement strategy type includes the main supplier strategy, the lowest price strategy, the performance priority strategy and the procurement cycle priority strategy; Step 2, based on the procurement strategy type, obtaining the associated supplier dataset, converting each supplier into a point cloud, and each point cloud contains three-dimensional attributes including bid, performance score and procurement cycle; Step 3, based on each point cloud, constructing a strategy decision virtual surface with bid as the horizontal axis, performance score as the vertical axis and procurement cycle as the vertical axis, and according to the procurement strategy type, dividing the structure area on the virtual surface, and dynamically segmenting the area according to the point cloud distribution density to obtain the segmentation result; Step 4, based on the segmentation result, mapping each supplier point cloud to the corresponding partition, including marking the point cloud located in the core area as the core cloud cluster; marking the point cloud located in the edge area as the edge cloud cluster; calculating the average dispersion degree of the distance between each point cloud as the core distribution parameter for all mapped core cloud cluster point clouds; calculating the average degree of the position deviation from the center position as the edge distribution parameter for all mapped edge cloud cluster point clouds; generating a strategy adjustment value based on the core distribution parameter and the edge distribution parameter, the strategy adjustment value = core distribution parameter ÷ (edge distribution parameter + ε), wherein ε is a minimum value; Step 5, correcting the procurement strategy parameters through the strategy adjustment value, executing the supplier matching based on the corrected strategy parameters, and performing the verification operation on the matched suppliers, including the supplier qualification validity check, the price validity period verification and the supplier blacklist screening; Step 6, based on the verified suppliers and prices, generating a procurement order, and sending a confirmation notice to the suppliers according to the procurement order, updating the procurement account and freezing the budget occupancy quota.

2. The purchase order generation method of claim 1, wherein, Receiving a demand plan containing material information, parsing the procurement strategy identifier associated with the material information, determining the procurement strategy type, and the procurement strategy type includes the main supplier strategy, the lowest price strategy, the performance priority strategy and the procurement cycle priority strategy, comprising: Parsing the material code in the demand plan and extracting the associated procurement strategy identifier; Based on the procurement strategy identifier, query the preset strategy mapping table to determine the procurement strategy type, wherein identifier PS1 corresponds to the main supplier strategy; identifier PS2 corresponds to the lowest price strategy; identifier PS3 corresponds to the performance priority strategy; and identifier PS4 corresponds to the procurement cycle priority strategy.

3. The purchase order generation method of claim 2, wherein, Based on the procurement strategy type, obtaining the associated supplier dataset, converting each supplier into a point cloud, and each point cloud contains three-dimensional attributes including bid, performance score and procurement cycle, comprising: Based on the procurement strategy type, the associated supplier dataset is obtained, if it is a main supplier strategy, the price source list database is accessed according to the material information to extract the preset main supplier number and the quotation within the valid period; if it is a lowest price strategy, the price main data warehouse is queried according to the material information to obtain all valid quotation lists; if it is a performance priority strategy, the price main data warehouse is queried according to the material information to obtain all valid quotation lists; if it is a procurement cycle priority strategy, the price main data warehouse is queried according to the material information to obtain all valid quotation lists; The supplier dataset is converted into a point cloud set, each supplier corresponds to a point cloud, and the point cloud contains three-dimensional attributes.

4. The purchase order generation method of claim 3, wherein, Based on each point cloud, a strategy decision virtual surface is constructed with the quotation as the horizontal axis, the performance score as the vertical axis, and the procurement cycle as the vertical axis. On the virtual surface, the structure shape area is divided according to the procurement strategy type, and the area is dynamically segmented according to the point cloud distribution density to obtain the segmentation result, including: Based on the point cloud set, a strategy decision virtual surface is constructed, a three-dimensional coordinate system is established with the quotation as the horizontal axis, the performance score as the vertical axis, and the procurement cycle as the vertical axis, and a convex hull surface is calculated for all supplier point clouds in the coordinate system to generate a convex hull surface that covers all point clouds as the strategy decision virtual surface; Based on the strategy decision virtual surface, the structure shape area division rule is determined according to the procurement strategy type, and the structure shape area division rule includes: When the procurement strategy is a main supplier strategy, a preset quotation interval is drawn on the virtual surface as a rectangular area; When the procurement strategy is a lowest price strategy, a preset quotation threshold interval is drawn at the bottom of the virtual surface as a trapezoidal area; When the procurement strategy is a performance priority strategy, an area with a performance score not lower than a preset score threshold is drawn at the top of the virtual surface as a polygonal area; When the procurement strategy is a procurement cycle priority strategy, an area with a procurement cycle not exceeding a preset cycle threshold is drawn at the front end of the virtual surface as a columnar area; For the structure shape area, the point cloud density distribution value in the area is calculated, and based on the density distribution value, the area is dynamically segmented by a density gradient algorithm. The area with a density distribution value greater than or equal to a threshold value is marked as a high-density core area, and the area with a density distribution value less than a threshold value is marked as a low-density edge area to obtain the segmentation result.

5. The purchase order generation method of claim 4, wherein, The procurement strategy parameters are corrected by the strategy adjustment value, the supplier matching is performed based on the corrected strategy parameters, and the matched suppliers are subjected to a verification operation, including supplier qualification validity check, price validity period verification, and supplier blacklist screening, including: Based on the strategy adjustment value, the parameters of the current procurement strategy are dynamically corrected, and the dynamic correction includes: When the strategy type is a main supplier strategy, the main supplier quotation floating threshold is corrected; When the strategy type is a lowest price strategy, the lowest quotation acceptance bias value is corrected; When the strategy type is a performance priority strategy, the performance score weight ratio is corrected; When the strategy type is a procurement cycle priority strategy, the procurement cycle priority coefficient is corrected; The supplier dataset is re-screened according to the corrected strategy parameters to generate a supplier matching sequence sorted according to the strategy optimization; The first supplier in the matching sequence is subjected to a verification operation, including supplier qualification validity check, price validity period verification, and supplier blacklist screening.

6. The purchase order generation method of claim 5, wherein, Based on the passed supplier and price, a purchase order is generated, and an order confirmation notice is sent to the supplier, the purchase ledger is updated, and the budget occupancy quota is frozen according to the purchase order, including: Based on the passed supplier and price, a structured purchase order is generated, including the supplier code, the transaction price, and the expected arrival date; According to the purchase order, an order confirmation notice is sent to the supplier, the purchase ledger is updated, and the budget occupancy quota is frozen.

7. A purchase order generation device, which implements the method according to any one of claims 1 to 6, characterized in that, It includes: A receiving and analyzing module is used to receive a demand plan containing material information, analyze the procurement strategy identifier associated with the material information, and determine the procurement strategy type; A point cloud generation module is used to obtain the associated supplier data set based on the procurement strategy type, and convert each supplier into a point cloud, each point cloud containing three-dimensional attributes including the bid price, performance score, and procurement cycle; A segmentation module is used to construct a strategy decision virtual surface based on each point cloud, with the bid price as the horizontal axis, the performance score as the vertical axis, and the procurement cycle as the vertical axis, divide the structure shape area on the virtual surface according to the procurement strategy type, and dynamically segment the area according to the point cloud distribution density to obtain the segmentation result; A mapping and adjusting module is used to map each supplier point cloud to the corresponding partition based on the segmentation result, including marking the point cloud located in the core area as a core cloud cluster; marking the point cloud located in the edge area as an edge cloud cluster; calculating the average dispersion degree of the distance between each point cloud as the core distribution parameter for all mapped core cloud cluster point clouds; calculating the average degree of deviation of the position of each point cloud from the center position as the edge distribution parameter for all mapped edge cloud cluster point clouds; generating a strategy adjustment value based on the core distribution parameter and the edge distribution parameter, the strategy adjustment value = core distribution parameter ÷ (edge distribution parameter + ε), where ε is a very small value; A checking module is used to correct the procurement strategy parameters through the strategy adjustment value, execute supplier matching based on the corrected strategy parameters, and perform a checking operation on the matched suppliers, including supplier qualification validity check, price validity period verification, and supplier blacklist screening; An order processing module is used to generate a purchase order based on the passed supplier and price, and send an order confirmation notice to the supplier, update the purchase ledger, and freeze the budget occupancy quota according to the purchase order.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a program which is executed by the processor to implement the method of any one of claims 1-6. The computer readable storage medium stores a program which is executed by the processor to implement the method of any one of claims 1-6.

Images