A purchase order generation method, device and equipment

Abstract

The present application provides a purchase order generation method, device and equipment, the method comprising: if the material to be purchased is a non-infinite supply material, then the delivery mode of the material to be purchased is obtained; if the delivery mode is a delivery date mode, then the first reference feature is obtained based on the material to be purchased and the target supplier, and the predicted lead time is determined based on the first reference feature; the predicted available date is determined based on the predicted lead time and the negotiated available date; if the delivery mode is a wait notification mode, then the logistics days are obtained based on the material to be purchased and the target supplier, and the logistics days are the days required for the material to be delivered from the target supplier to the demander; the predicted available date is determined based on the delivery notification date and the logistics days; and a target purchase order for the material to be purchased is generated, and the target purchase order includes the material to be purchased, the target supplier, the predicted available date and the negotiated available date. Through the present application scheme, the gap between the target purchase order and the real scenario is reduced, and the uncertainty of the material supply capacity is reduced.

Description

Technical Field

[0001] The present application relates to the field of material procurement, and in particular to a method, device and equipment for generating a purchase order. Background Art

[0002] The APS (Advanced Planning and Scheduling) system is a system that uses advanced algorithms and optimization technologies to efficiently plan and schedule production, logistics and other processes. The APS system can comprehensively consider various resource constraints, order requirements, process routes and other factors to achieve the optimal scheduling solution and improve production efficiency and resource utilization. However, when the APS system obtains a purchase order, the purchase order may include the delivery date negotiated between the demander and the supplier, which is not accurate, resulting in the demander being unable to know the exact delivery date and unable to plan in advance based on the accurate delivery date. Summary of the invention

[0003] The present application provides a method for generating a purchase order, the method comprising:

[0004] If the material to be purchased is a material with non-infinite supply, the delivery mode of the material to be purchased is obtained, and the delivery mode is a delivery date mode or a wait notification mode; wherein the delivery date mode indicates that the material is delivered according to the negotiated available date, and the negotiated available date is the delivery date negotiated between the demander and the supplier; the wait notification mode indicates that the material is delivered according to the delivery notification date, and the delivery notification date is the date when the demander sends the delivery notification to the supplier, and the delivery notification date is earlier than the negotiated available date;

[0005] If the delivery mode is the delivery date mode, a first reference feature is obtained based on the material to be purchased and the target supplier, and the forecast lead time is determined based on the first reference feature; the forecast available date is determined based on the forecast lead time and the negotiated available date; if the delivery mode is the wait notification mode, the logistics days are obtained based on the material to be purchased and the target supplier, and the logistics days are the days required for the material to be delivered from the target supplier to the demander; the forecast available date is determined based on the delivery notification date and the logistics days; wherein the forecast available date is the predicted delivery date;

[0006] A target purchase order for the material to be purchased is generated, wherein the target purchase order includes the material to be purchased, the target supplier, the predicted availability date, and the negotiated availability date.

[0007] The application provides a purchase order generating device, the device comprising:

[0008] an acquisition module, for acquiring a delivery mode of the material to be purchased if the material to be purchased is a non-infinite supply material, wherein the delivery mode is a delivery date mode or a wait notification mode; wherein the delivery date mode indicates that the material is delivered according to a negotiated availability date, which is a delivery date negotiated between the demander and the supplier; and the wait notification mode indicates that the material is delivered according to a delivery notification date, which is a date when the demander sends a delivery notification to the supplier, and the delivery notification date is earlier than the negotiated availability date;

[0009] A determination module, for, if the delivery mode is the delivery date mode, obtaining a first reference feature based on the material to be purchased and the target supplier, determining the forecast lead time based on the first reference feature; determining the forecast availability date based on the forecast lead time and the negotiated availability date; if the delivery mode is the wait notification mode, obtaining logistics days based on the material to be purchased and the target supplier, the logistics days being the days required for the material to be delivered from the target supplier to the demander; determining the forecast availability date based on the delivery notification date and the logistics days; wherein the forecast availability date is the forecasted delivery date;

[0010] A generation module is used to generate a target purchase order for the material to be purchased, wherein the target purchase order includes the material to be purchased, the target supplier, the predicted availability date and the negotiated availability date.

[0011] The present application provides an electronic device, comprising: a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores machine-executable instructions that can be executed by the processor; the processor is used to execute the machine-executable instructions to implement the above-mentioned purchase order generation method.

[0012] The present application provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the above-mentioned purchase order generation method is implemented.

[0013] The present application provides a machine-readable storage medium, which stores machine-executable instructions that can be executed by a processor; wherein the processor is used to execute the machine-executable instructions, and when the machine-executable instructions are executed, the above-mentioned purchase order generation method is implemented.

[0014] It can be seen from the above technical solutions that in the embodiment of the present application, on the basis that the target purchase order includes a negotiated available date (the delivery date negotiated between the demander and the supplier), a predicted available date can be determined. The predicted available date is the predicted delivery date, which is a more accurate delivery date, and the target purchase order includes the predicted available date, so that the demander can know the accurate delivery date based on the target purchase order and can plan in advance based on the accurate delivery date. For example, the demander can respond to user orders in a timely manner, which improves the order fulfillment rate. By predicting the material supply capacity (such as the predicted available date), the gap between the target purchase order and the actual scenario is reduced, that is, the predicted available date is closer to the actual delivery date, the uncertainty of the material supply capacity is reduced, and the time for the entire link of urgent orders is effectively reduced, which improves the work efficiency of the planner. It can provide the material plan with a material supply capacity forecast report for a period of time in the future, and provide issuance suggestions and assistance when the material plan performs actual material extraction operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flowchart of a method for generating a purchase order in one embodiment of the present application;

[0016] Figure 2 It is a flowchart of a method for generating a purchase order in one embodiment of the present application;

[0017] Figure 3 It is a flowchart of a method for generating a purchase order in one embodiment of the present application;

[0018] Figure 4 It is a flow chart of a material supply capacity prediction method in one embodiment of the present application;

[0019] Figure 5 It is a structural schematic diagram of a purchase order generating device in one embodiment of the present application;

[0020] Figure 6 It is a hardware structure diagram of an electronic device in one embodiment of the present application. DETAILED DESCRIPTION

[0021] In the embodiment of the present application, a purchase order generation method is proposed. The method can be applied to electronic equipment. Figure 1 FIG. 1 is a flow chart of the purchase order generation method, which may include:

[0022] Step 101: If the material to be purchased is a non-infinite supply material, the delivery mode of the material to be purchased is obtained, and the delivery mode can be a delivery date mode or a wait notification mode. The delivery date mode can indicate that the delivery is made according to the negotiated available date, and the negotiated available date is the delivery date negotiated between the demander and the supplier; the wait notification mode can indicate that the delivery is made according to the delivery notification date, and the delivery notification date is the date when the demander sends the delivery notification to the supplier, and the delivery notification date is earlier than the negotiated available date.

[0023] Step 102: If the delivery mode is the delivery date mode, a first reference feature is obtained based on the material to be purchased and the target supplier, and the forecast lead time is determined based on the first reference feature; the forecast available date is determined based on the forecast lead time and the negotiated available date; if the delivery mode is the wait notification mode, the logistics days are obtained based on the material to be purchased and the target supplier, and the logistics days may be the days required for the material to be delivered from the target supplier to the demander; the forecast available date is determined based on the delivery notification date and the logistics days. The forecast available date may be the predicted delivery date.

[0024] Step 103: Generate a target purchase order for the materials to be purchased, wherein the target purchase order may include the materials to be purchased, the target supplier, the predicted availability date, and the negotiated availability date.

[0025] Exemplarily, obtaining a first reference feature based on the material to be purchased and the target supplier, and determining the forecast lead time based on the first reference feature may include but is not limited to: constructing a bipartite graph of materials and suppliers, the bipartite graph includes all material nodes and all supplier nodes, and the edge weight of the bipartite graph is the number of deliveries of the material by the supplier in historical time; querying the material node corresponding to the material to be purchased and the supplier node corresponding to the target supplier from the bipartite graph; based on the bipartite graph, using graph representation to determine the first representation vector of the material node and the second representation vector of the supplier node; wherein, the closer the connection between the material to be purchased and the target supplier, the more similar the first representation vector and the second representation vector are; obtaining a first reference feature, the first reference feature includes a first representation vector and a second representation vector; inputting the first reference feature into a trained first prediction model, and the first prediction model outputs the forecast lead time based on the first reference feature.

[0026] Exemplarily, the first reference feature may also include, but is not limited to, at least one of the following features: purchase order basic features, the purchase order basic features include at least one of the following: the number of days between the generation date of the target purchase order and the current date, the number of days between the demander's expected delivery date and the current date, the number of days between the supplier's expected delivery date and the current date, and the purchase quantity of the material to be purchased; historical delivery features, the historical delivery features include a first statistical feature of a target matching pair, the target matching pair includes the material to be purchased and a target supplier, the first statistical feature is obtained by counting the number of lead times and / or the total number of days required for purchase of the historical purchase orders of the target matching pair; supplier historical delivery features, the supplier historical delivery features include a second statistical feature of the target supplier, the second statistical feature is obtained by counting the number of lead times and / or the total number of days required for purchase of the historical purchase orders of the target supplier; material historical delivery features, the material historical delivery features include a third statistical feature of the material to be purchased, the third statistical feature is obtained by counting the number of lead times and / or the total number of days required for purchase of the historical purchase orders of the material to be purchased.

[0027] Exemplarily, obtaining logistics days based on the materials to be purchased and the target supplier may include, but are not limited to: querying the first mapping table through the target matching pair to obtain the logistics days corresponding to the target matching pair; wherein the target matching pair may include the materials to be purchased and the target supplier, and the first mapping table may include the correspondence between the matching pair and the logistics days. The maintenance process of the first mapping table may include, but is not limited to: obtaining a sample matching pair and a historical logistics time series corresponding to the sample matching pair based on historical purchase orders, wherein the sample matching pair may include a sample material and a sample supplier, and the historical logistics time series may include multiple historical logistics days, and each historical logistics day represents the days required for the sample material to be delivered from the sample supplier to the demander; predicting the next logistics day of the sample matching pair based on multiple historical logistics days; and recording the correspondence between the sample matching pair and the next logistics day in the first mapping table.

[0028] Exemplarily, if the material to be purchased is an unlimited supply material, the predicted availability date can be determined based on the current date. The process of determining unlimited supply materials and non-infinite supply materials may include but is not limited to: for each material, statistics on the material's warehousing data in historical time, the warehousing data may include but is not limited to the first ratio value and / or the second ratio value; the first ratio value may be the ratio of the material's warehousing days to the historical time, and the second ratio value may be the ratio of the candidate quantity to the historical time; for each day in the historical time, if the actual material warehousing quantity in the K days after that day is greater than the expected material warehousing quantity, then the candidate quantity is increased by 1, otherwise, the candidate quantity is kept unchanged. Based on the storage data of each material, an abnormal data analysis method is used to analyze all materials to obtain an abnormal data candidate set, and the abnormal data candidate set includes multiple abnormal materials; for each abnormal material in the abnormal data candidate set, if the storage data of the abnormal material meets a preset condition, the abnormal material is determined to be an unlimited supply material; wherein, if the storage data includes a first ratio value and a second ratio value, when the first ratio value is greater than a first threshold value, and the second ratio value is greater than a second threshold value, the storage data of the abnormal material meets the preset condition; and the remaining materials except the unlimited supply materials in all materials are determined to be non-infinite supply materials.

[0029] Exemplarily, the process of obtaining the negotiated available date may include, but is not limited to: querying the second mapping table through the target matching pair to obtain the short-term time interval corresponding to the target matching pair; determining the initial available date based on the short-term time interval and the current date, and determining the initial available date as the negotiated available date, or adjusting the initial available date to obtain the negotiated available date; wherein the target matching pair includes the material to be purchased and the target supplier, and the second mapping table includes the corresponding relationship between the matching pair and the short-term time interval;

[0030] Among them, the maintenance process of the second mapping table may include but is not limited to: obtaining sample matching pairs and the purchasing day sequence corresponding to the sample matching pairs based on historical purchase orders, the sample matching pairs include sample materials and sample suppliers, the purchasing day sequence includes multiple total purchasing days, and the total purchasing days represent the difference between the actual warehousing date of the sample material and the purchase application date of the sample material; clustering the multiple total purchasing days to obtain at least one cluster cluster; selecting candidate cluster clusters from all cluster clusters, and the number of total purchasing days in the candidate cluster clusters is greater than a threshold; calculating the average value of all total purchasing days in each candidate cluster cluster, selecting the minimum average value from the average values ​​corresponding to all candidate cluster clusters, and determining the short-term time interval based on the minimum average value; recording the correspondence between the sample matching pairs and the short-term time interval in the second mapping table.

[0031] Exemplarily, the target purchase order may also include a negotiated available quantity, which is the delivery quantity negotiated between the demander and the supplier. The process of obtaining the negotiated available quantity may include but is not limited to: obtaining a second reference feature based on the material to be purchased and the target supplier, the second reference feature including a first characterization vector and a second characterization vector; wherein, based on the constructed bipartite graph of materials and suppliers, a graph representation is used to determine the first characterization vector of the material node corresponding to the material to be purchased, and the second characterization vector of the supplier node corresponding to the target supplier is determined; the second reference feature is input into the trained second prediction model so that the second prediction model outputs an initial available quantity based on the second reference feature; the initial available quantity is determined as the negotiated available quantity, or the initial available quantity is adjusted to obtain the negotiated available quantity.

[0032] Exemplarily, the second reference feature may also include, but is not limited to, at least one of the following features: purchase order basic features, the purchase order basic features include at least one of the following: the short-term time interval corresponding to the target purchase order, the configured standard time interval corresponding to the target purchase order, the ratio of the short-term time interval to the standard time interval, and the delivery mode of the materials to be purchased; supplier inventory features, the supplier inventory features include at least one of the following: the target supplier's latest feedback inventory, the target supplier's latest feedback inventory after supply and demand offset, the target supplier's expected delivery quantity in the future, and the number of days between the target supplier's latest feedback date and the purchase application date; supplier supply capacity features, the supplier supply capacity features are determined based on all historical purchase orders; supplier supply capacity features include at least one of the following: the target supplier's average daily delivery quantity in the past period of time, the target supplier's average delivery day interval, the target supplier's minimum total procurement days, the maximum total procurement days of the target supplier, the average total procurement days of the target supplier, the minimum delivery quantity of the target supplier, the maximum delivery quantity of the target supplier, and the average delivery quantity of the target supplier; the supplier's short-term capability characteristics, which are determined based on the candidate historical procurement orders among all historical procurement orders, and the short-term time intervals corresponding to the candidate historical procurement orders are less than the threshold; the supplier's short-term capability characteristics include at least one of the following: the average daily delivery quantity of the target supplier in the past period of time, the average delivery days interval of the target supplier, the minimum total procurement days of the target supplier, the maximum total procurement days of the target supplier, the average total procurement days of the target supplier, the minimum delivery quantity of the target supplier, the maximum delivery quantity of the target supplier, and the average delivery quantity of the target supplier; material supply and demand characteristics, which include at least one of the following: the demand quantity of the material to be purchased in the future period of time, and the supply quantity of the material to be purchased in the future period of time.

[0033] It can be seen from the above technical solutions that in the embodiment of the present application, on the basis that the target purchase order includes a negotiated available date (the delivery date negotiated between the demander and the supplier), a predicted available date can be determined. The predicted available date is the predicted delivery date, which is a more accurate delivery date, and the target purchase order includes the predicted available date, so that the demander can know the accurate delivery date based on the target purchase order and can plan in advance based on the accurate delivery date. For example, the demander can respond to user orders in a timely manner, which improves the order fulfillment rate. By predicting the material supply capacity (such as the predicted available date), the gap between the target purchase order and the actual scenario is reduced, that is, the predicted available date is closer to the actual delivery date, the uncertainty of the material supply capacity is reduced, and the time for the entire link of urgent orders is effectively reduced, which improves the work efficiency of the planner. It can provide the material plan with a material supply capacity forecast report for a period of time in the future, and provide issuance suggestions and assistance when the material plan performs actual material extraction operations.

[0034] The following describes the purchase order generation method of the embodiment of the present application in combination with a specific application scenario.

[0035] In the embodiment of the present application, a purchase order generation method is proposed, which reduces the gap between the target purchase order and the actual scenario by predicting the material supply capacity (such as predicting the availability date), that is, the predicted availability date is closer to the actual delivery date, reducing the uncertainty of the material supply capacity. The purchase order generation method can be used in conjunction with the APS system, that is, during the processing of the APS system, the purchase order generation method is used to generate the target purchase order. Alternatively, the purchase order generation method can be used alone, that is, during the processing of any system, the purchase order generation method is used to generate the target purchase order.

[0036] For example, when a demander needs to purchase a certain material from a supplier, the material is called the material to be purchased, and the supplier is called the target supplier. The target supplier provides the material to be purchased to the demander.

[0037] The purchase of materials to be purchased usually involves the following processes: purchase application process, purchase process, logistics process and warehousing process. Regarding the purchase application process, the purchase application process is an internal process of the demander and has nothing to do with the target supplier. The material planner submits the purchase application to determine the expected delivery date of the demander. In the purchase application process, a purchase application form needs to be generated. The purchase application form includes purchase application related information. See Table 1 for an example of a purchase application form. The purchase application form may include materials to be purchased, target suppliers, standard LT (standard interval time), expected delivery date, purchase quantity, etc.

[0038] Table 1

[0039]

[0040] In Table 1, Order1 and the like may represent the serial number of the purchase application, Matnr1 and the like may represent the unique identifier of the material to be purchased, and Vender1 and the like may represent the unique identifier of the target supplier. Optionally, the purchase application table may also include the supplier inventory quantity (not shown in Table 1), which may represent the number of materials remaining to be purchased by the supplier, and the supplier may provide the materials to be purchased in this quantity. The purchase quantity may represent that the demander needs to purchase the materials to be purchased in this quantity. The standard LT represents the time required from the start to the completion of the order. For example, if the target supplier takes 15 days from the receipt of the order to the completion of the material production, the standard LT is 15 days. The standard LT is the capacity provided by the target supplier to the demander. The expected delivery date represents the expected delivery date (expected delivery date) of the demander, that is, the demander receives the materials to be purchased on the expected delivery date. Considering the time required for the target supplier to complete the order from the start to the completion, the expected delivery date may be greater than or equal to the sum of the standard LT and the current date.

[0041] For the procurement process, after the material planner submits the purchase application, the demander can internally review and approve the purchase application. If the purchase application is not approved, the process ends and the subsequent procurement process is no longer executed. If the purchase application is approved, the subsequent procurement process continues. The procurement process is the process of reaching an agreement between the demander and the target supplier, that is, the procurement process is related to the target supplier, and the supplier information and the negotiated available date need to be determined. The negotiated available date is the delivery date negotiated between the demander and the supplier, that is, the estimated delivery time. In the procurement process, a purchase order table needs to be generated, which includes purchase order related information. See Table 2 for an example of a purchase order table. The purchase order table may include materials to be purchased, target suppliers, delivery modes, negotiated available quantities, negotiated available dates, etc.

[0042] Table 2

[0043]

[0044] In Table 2, Order1 etc. may represent the serial number of the purchase order, Matnr1 etc. may represent the unique identifier of the material to be purchased, and Vender1 etc. may represent the unique identifier of the target supplier.

[0045] The negotiated available quantity in Table 2 may be different from the purchase quantity in Table 1. For the sake of distinction, the purchase quantity in Table 1 is called the initial available quantity, and the purchase quantity in Table 2 is called the negotiated available quantity. The initial available quantity indicates that the demander needs to purchase the material to be purchased in this quantity, that is, the demander's expected value. However, the target supplier may agree to the initial available quantity or may refuse the initial available quantity. Therefore, the purchase quantity needs to be negotiated between the demander and the target supplier, and the negotiated purchase quantity is the negotiated available quantity, that is, the negotiated available quantity is the delivery quantity negotiated between the demander and the supplier.

[0046] Obviously, if the target supplier agrees to the initial available quantity, the initial available quantity can be determined as the negotiated available quantity, or, if the target supplier rejects the initial available quantity, the initial available quantity can be adjusted to obtain the negotiated available quantity, which is greater than or less than the initial available quantity.

[0047] The negotiated availability date in Table 2 may be different from the expected delivery date in Table 1. For the sake of distinction, the expected delivery date in Table 1 is referred to as the initial availability date, and the expected delivery date in Table 2 is referred to as the negotiated availability date. For example, the initial availability date indicates the delivery date expected by the demander, that is, the demander's expectation for the delivery date. However, the target supplier may agree to the initial availability date or may reject the initial availability date. Therefore, the delivery date needs to be negotiated between the demander and the target supplier, and the negotiated delivery date is the negotiated availability date, that is, the negotiated availability date is the delivery date negotiated between the demander and the target supplier. Obviously, if the target supplier agrees to the initial availability date, the initial availability date is determined as the negotiated availability date, or if the target supplier rejects the initial availability date, the initial availability date is adjusted to obtain the negotiated availability date, and the negotiated availability date is earlier or later than the initial availability date.

[0048] The delivery mode of the materials to be purchased (i.e. the material delivery mode) can be the waiting notification mode (abbreviated as waiting notification), the delivery date mode (abbreviated as delivery date) and the unlimited material mode (abbreviated as unlimited material).

[0049] For the materials to be purchased, you can determine whether the materials to be purchased are unlimited supply materials or non-unlimited supply materials. The determination method can be found in the subsequent process. If the materials to be purchased are unlimited supply materials, the delivery mode of the materials to be purchased is unlimited supply mode. Unlimited supply mode means that the target supplier has a large number of materials to be purchased, and the demander does not need to wait. The target supplier can provide the materials to be purchased at any time.

[0050] If the material to be purchased is a non-infinite supply material, the delivery mode of the material to be purchased is the waiting notification mode or the delivery date mode, that is, the delivery mode of the material to be purchased can be pre-configured as the waiting notification mode, or the delivery mode of the material to be purchased can be pre-configured as the delivery date mode, and there is no restriction on this.

[0051] The delivery date mode means that the goods are delivered according to the negotiated availability date, which is the delivery date negotiated between the demander and the supplier. That is to say, after the target supplier negotiates the delivery date with the demander, it no longer waits for the demander's delivery notice and can deliver the goods according to the negotiated availability date. For example, when the negotiated availability date is 2024-09-20, the target supplier needs to deliver the materials to be purchased to the demander before 2024-09-20. In order to deliver the materials to be purchased to the demander before the negotiated availability date, considering the logistics time, the target supplier usually needs to send the materials to be purchased a few days in advance so that they can be delivered before the negotiated availability date.

[0052] The waiting notification mode means that the goods are delivered according to the delivery notification date. The delivery notification date is the date when the demander sends the delivery notification to the target supplier. In other words, the target supplier needs to wait for the delivery notification from the demander. Only after receiving the delivery notification will the target supplier send the materials to be purchased to the demander.

[0053] For example, after the demander negotiates the delivery date with the target supplier, the target supplier is usually given a period of time during which the target supplier produces the materials to be purchased. After a period of time, the demander sends a delivery notice to the target supplier. The date of sending the delivery notice can be called the delivery notice date. Considering the logistics time, in order to deliver the materials to be purchased to the demander before the negotiated availability date, the delivery notice date is earlier than the negotiated availability date. After the target supplier receives the delivery notice, the target supplier sends the materials to be purchased to the demander and delivers the materials to be purchased to the demander before the negotiated availability date.

[0054] With regard to the logistics process, after obtaining the purchase order form, the target supplier can issue the materials to be purchased to the demander based on the purchase order form. When issuing the materials to be purchased, the logistics process is involved, that is, the materials to be purchased are delivered to the demander through logistics. The target supplier can send an internal delivery note (also called an inbound delivery note form) to the demander. The internal delivery note includes logistics information (i.e., supplier delivery record information), such as materials, suppliers, and expected delivery date. This embodiment does not limit this logistics process.

[0055] Regarding the warehousing process, after the materials to be purchased arrive at the demander, the demander performs the warehousing operation on the materials to be purchased. After the warehousing of the materials to be purchased is completed, a warehousing order (also known as a material movement voucher table) can be recorded. The warehousing order includes the supplier's delivery and warehousing information (i.e., the real warehousing record), such as materials, suppliers, actual warehousing date, etc. This embodiment does not limit this warehousing process.

[0056] In a possible implementation, in the actual operation scenario, for the purchase application process, there is often a situation where a certain material lacks a purchase order or the existing purchase order is insufficient to meet the material gap. According to the standard procurement process, a minimum of the standard LT is required before the purchase can be successful (that is, the expected delivery date can be greater than or equal to the sum of the standard LT and the current date). However, the material planner hopes to know whether the material can be delivered within the short-term LT interval and the quantity that can be delivered. Based on this, the actual supply capacity of the supplier can be predicted, that is, the short-term LT corresponding to the material and the quantity that can be delivered can be predicted. In this way, the short-term LT and the quantity that can be delivered can be provided to the material planner for reference.

[0057] See also Figure 2 FIG. 1 is a flowchart of a method for generating a purchase order, and the process may include:

[0058] Step 201: Obtain the material to be purchased and the target supplier, and the target supplier provides the material to be purchased to the demander. For the convenience of description, the combination of the material to be purchased and the target supplier can be called a target matching pair, that is, the target matching pair can include the material to be purchased and the target supplier.

[0059] Step 202: query the second mapping table through the target matching pair to obtain the short-term time interval corresponding to the target matching pair, which may also be referred to as the short-term LT (Lead Time).

[0060] Exemplarily, a second mapping table may be maintained in advance, and the second mapping table may include a correspondence between matching pairs (i.e., combinations of materials and suppliers) and short-term time intervals. On this basis, the second mapping relationship may be queried through a target matching pair to obtain the short-term time interval corresponding to the target matching pair.

[0061] Exemplarily, the maintenance process for the second mapping table may include the following steps:

[0062] Step 2021: Based on historical purchase orders, obtain sample matching pairs and a purchasing day sequence corresponding to the sample matching pairs. The sample matching pairs may include sample materials and sample suppliers (i.e., the combination of sample materials and sample suppliers is called a sample matching pair), and the purchasing day sequence may include multiple total purchasing days.

[0063] Exemplarily, for each matching pair (material, supplier), the total number of historical purchase orders of the matching pair in the past period of time (such as 6 months) can be counted. If the total number of historical purchase orders is less than a preset threshold (which can be configured based on experience, such as 10, 15, etc., and there is no limit on this), the historical purchase orders of the matching pair are filtered. If the total number of historical purchase orders is greater than or equal to the preset threshold, the matching pair is used as a sample matching pair (i.e., a valid matching pair), the material in the matching pair is called a sample material, and the supplier in the matching pair is called a sample supplier. In addition, each historical purchase order of the sample matching pair corresponds to a total purchasing day, so that the purchasing day sequence corresponding to the sample matching pair can be obtained.

[0064] For example, the sample matching pair corresponds to historical purchase order a1-historical purchase order a6, historical purchase order a1 corresponds to total purchase days b1, and so on, historical purchase order a6 corresponds to total purchase days b6. In this way, the purchase days sequence corresponding to the sample matching pair includes total purchase days b1-total purchase days b6.

[0065] For the total purchase days corresponding to the historical purchase orders, the total purchase days can represent the difference between the actual storage date of the sample material (i.e. the actual delivery date) and the purchase application date of the sample material (i.e. the purchase application creation date). For example, the historical purchase order belongs to the purchase process, and the historical purchase order can be associated with the purchase application process and the storage process. In the purchase application process, a purchase application form can be created, and the creation date of the purchase application form is the purchase application date. In the storage process, the demander performs storage operations on the material and records the storage order, and the storage date of the material is the actual storage date. In this way, the difference between the actual storage date and the purchase application date can be used as the total purchase days of the sample material.

[0066] Step 2022: For each sample matching pair, the purchase days sequence corresponding to the sample matching pair includes multiple total purchase days, and the multiple total purchase days can be clustered to obtain at least one cluster. For example, for each cluster, the cluster can include at least one total purchase day.

[0067] For example, a hierarchical clustering method can be used to cluster multiple total purchase days to obtain cluster clusters, a K-Means clustering method can be used to cluster multiple total purchase days to obtain cluster clusters, and other clustering methods can be used to cluster multiple total purchase days to obtain cluster clusters. There is no restriction on this.

[0068] For example, the purchase days sequence corresponding to the sample matching pair includes 8 total purchase days, which are 7, 9, 10, 11, 14, 15, 16, and 19. The hierarchical clustering method is used to cluster these 8 total purchase days to obtain 3 clusters, and there is no restriction on the implementation process of this hierarchical clustering method. The total purchase days of cluster 1 can be distributed as 7, 9, 10, and 11, the total purchase days of cluster 2 can be distributed as 14, 15, and 16, and the total purchase days of cluster 3 can be distributed as 19.

[0069] Step 2023: Select a candidate cluster from all clusters, and the total number of purchasing days in the candidate cluster is greater than a threshold (which can be configured based on experience, such as 2, 3, 4, etc., and there is no restriction on this).

[0070] For example, assuming that the threshold is 2, since the number of total purchase days in cluster 1 is 4, which is greater than the threshold 2, cluster 1 is selected as a candidate cluster. Since the number of total purchase days in cluster 2 is 3, which is greater than the threshold 2, cluster 2 is selected as a candidate cluster. Since the number of total purchase days in cluster 3 is 1, which is not greater than the threshold 2, cluster 3 is not selected as a candidate cluster.

[0071] Step 2024: Calculate the average of all total purchase days in each candidate cluster.

[0072] For example, cluster 1 and cluster 2 are candidate clusters. The average value of all total purchase days in cluster 1 (7, 9, 10, 11) is calculated, that is, (7+9+10+11) / 4=9.25, and the average value of all total purchase days in cluster 2 (14, 15, 16) is calculated, that is, (14+15+16) / 3=15.

[0073] Step 2025: Select the minimum average value from the average values ​​corresponding to all candidate clusters, and determine the short-term time interval based on the minimum average value, such as the rounded-down value or rounded-up value of the minimum average value as the short-term time interval. For example, the short-term time interval is obtained by rounding down the minimum average value 9.25.

[0074] Step 2026: Record the correspondence between the sample matching pairs and the short-term time intervals in the second mapping table.

[0075] For example, for each sample matching pair, after obtaining the short-term time interval corresponding to the sample matching pair, the corresponding relationship between the sample matching pair and the short-term time interval (short-term LT) can be recorded in the second mapping table. At this point, the second mapping table can be maintained. In step 202, the second mapping table can be queried through the target matching pair to obtain the short-term time interval corresponding to the target matching pair.

[0076] Step 203: Determine an initial available date based on the short-term time interval and the current date. For example, the sum of the short-term time interval and the current date can be used as the initial available date of the target matching pair, and the initial available date can represent the delivery date expected by the demander, that is, the demander's expectation for the delivery date.

[0077] Step 204: Acquire a second reference feature based on the material to be purchased and the target supplier, where the second reference feature includes a first characterization vector corresponding to the material to be purchased and a second characterization vector corresponding to the target supplier.

[0078] Exemplarily, the process of acquiring the second reference feature may include the following steps:

[0079] Step 2041: construct a bipartite graph of materials and suppliers. The bipartite graph may include all material nodes and all supplier nodes. The edge weight of the bipartite graph is the number of times the supplier has delivered the material in the historical time. For example, the edge weight of the bipartite graph is the number of times the supplier has delivered the material in the past 6 months.

[0080] Step 2042: Query the material node corresponding to the material to be purchased and the supplier node corresponding to the target supplier from the bipartite graph. For example, the bipartite graph includes material nodes corresponding to each material, and the material node corresponding to the material to be purchased can be queried from these material nodes. The bipartite graph includes supplier nodes corresponding to each supplier, and the supplier node corresponding to the target supplier can be queried from these supplier nodes.

[0081] Step 2043: Based on the bipartite graph, a first characterization vector of a material node corresponding to the material to be purchased and a second characterization vector of a supplier node corresponding to the target supplier are determined using graph characterization. The closer the connection between the material to be purchased and the target supplier, the more similar the first characterization vector and the second characterization vector are.

[0082] For example, graph representation learning methods such as Node2Vec can be used to determine the first representation vector of the material node corresponding to the material to be purchased. The first representation vector is a representation of the material to be purchased, which can be a 32-dimensional representation vector or a representation vector of other dimensions. Graph representation learning methods such as Node2Vec can be used to determine the second representation vector of the supplier node corresponding to the target supplier. The second representation vector is a representation of the target supplier, which can be a 32-dimensional representation vector or a representation vector of other dimensions.

[0083] Exemplarily, the second reference feature may include only the first characterization vector and the second characterization vector. Alternatively, in addition to the first characterization vector and the second characterization vector, the second reference feature may also include but is not limited to at least one of the following features: purchase order basic features, supplier inventory features, supplier supply capability features, supplier short-term capability features, and material supply and demand features. Alternatively, the second reference feature may include but is not limited to at least one of the following features: the first characterization vector, the second characterization vector, purchase order basic features, supplier inventory features, supplier supply capability features, supplier short-term capability features, and material supply and demand features.

[0084] Regarding the basic characteristics of the purchase order, the basic characteristics of the purchase order may include but are not limited to at least one of the following: a short-term time interval (i.e., short-term LT) corresponding to the target purchase order, the short-term time interval is the difference between the expected delivery date and the purchase application creation date, see step 202, the short-term time interval has been obtained.

[0085] The configured standard time interval (ie, standard LT) corresponding to the target purchase order is shown in Table 1. The purchase application form may include the standard LT, and therefore, the standard time interval may be known.

[0086] The ratio of the short-term time interval to the standard time interval is short-term LT / standard LT.

[0087] The delivery mode (i.e., purchase type) of the materials to be purchased is shown in Table 2. The purchase order table may include the delivery mode, so the delivery mode may be known, such as the wait-for-notification mode or the due date mode.

[0088] Purchase quantity: The purchase quantity may indicate that the demander needs to purchase the quantity of the material to be purchased. As shown in Table 1, the purchase application form may include the purchase quantity, and therefore, the purchase quantity may be known.

[0089] Regarding supplier inventory characteristics, the supplier inventory characteristics may include but are not limited to at least one of the following: the target supplier's latest feedback inventory. For example, the target supplier's latest feedback is date A, and the inventory of the material to be purchased reported on date A is 1000, then the target supplier's latest feedback inventory is 1000.

[0090] The inventory after supply and demand offset after the target supplier's latest feedback. For example, assuming the current date is date B, during the period from date A to date B, if the supply and demand offset for the material to be purchased is 200, then the inventory after supply and demand offset after the target supplier's latest feedback is 800 (1000-200).

[0091] The expected delivery quantity of the target supplier in the next period of time, for example, the expected delivery quantity of the target supplier in the next 1 week, the expected delivery quantity of the target supplier in the next 2 weeks, the expected delivery quantity of the target supplier in the next 3 weeks, and the expected delivery quantity of the target supplier in the next 4 weeks.

[0092] The number of days between the target supplier's latest feedback date and the purchase application date, that is, the number of days between the target supplier's latest feedback and the purchase application creation date. For example, if the target supplier's most recent feedback is date A and the current date is date B, then the above interval can be the number of days between date B and date A.

[0093] Regarding the supplier supply capacity characteristics, the supplier supply capacity characteristics may be determined based on all historical purchase orders (such as all historical purchase orders in the past period of time (such as 6 months)). For example, all historical purchase orders corresponding to the target supplier are obtained, and the supplier supply capacity characteristics corresponding to the target supplier are determined based on these historical purchase orders. The supplier supply capacity characteristics may include but are not limited to at least one of the following: the average daily delivery quantity of the target supplier in the past period of time, such as the average daily delivery quantity of the target supplier in the past month (determined based on the historical purchase orders in the past month), the average daily delivery quantity of the target supplier in the past 2 months, the average daily delivery quantity of the target supplier in the past 3 months, the average daily delivery quantity of the target supplier in the past 4 months, and the average daily delivery quantity of the target supplier in the past 6 months.

[0094] The average delivery interval of the target supplier, that is, the historical average delivery interval of the target supplier. For example, in the past 6 months, the interval between the second delivery and the first delivery was 3 days, the interval between the third delivery and the second delivery was 6 days, and so on. The average of these delivery intervals can be calculated.

[0095] The minimum total purchasing days of the target supplier. For example, all total purchasing days of the target supplier in the past 6 months can be counted, and the minimum total purchasing days among all the total purchasing days can be determined.

[0096] The maximum total purchasing days of the target supplier. For example, all total purchasing days of the target supplier in the past 6 months can be counted, and the maximum total purchasing days among all the total purchasing days can be determined.

[0097] The average total purchasing days of the target supplier. For example, you can count all the total purchasing days of the target supplier in the past 6 months and calculate the average of all the total purchasing days.

[0098] The minimum delivery quantity of the target supplier can count the delivery quantities of all purchase orders of the target supplier in the past 6 months and determine the minimum delivery quantity among all delivery quantities.

[0099] The maximum delivery quantity of the target supplier can count the delivery quantities of all purchase orders of the target supplier in the past 6 months and determine the maximum delivery quantity among all delivery quantities.

[0100] The average delivery quantity of the target supplier can be calculated by counting the delivery quantities of all purchase orders of the target supplier in the past 6 months and calculating the average of all delivery quantities as the average delivery quantity.

[0101] For the supplier's short-term capability characteristics, the supplier's short-term capability characteristics can be determined based on candidate historical purchase orders among all historical purchase orders, and the short-term time interval corresponding to the candidate historical purchase orders is less than a threshold. For example, all historical purchase orders corresponding to the target supplier are obtained (such as all historical purchase orders in the past period of time (such as 6 months)). For each historical purchase order, if the short-term time interval (i.e., short-term LT) corresponding to the historical purchase order is less than a threshold (which can be configured based on experience), the historical purchase order is used as a candidate historical purchase order. If the short-term time interval corresponding to the historical purchase order is not less than the threshold, the historical purchase order is not used as a candidate historical purchase order. On this basis, multiple candidate historical purchase orders can be selected from all historical purchase orders. Then, the supplier's short-term capability characteristics corresponding to the target supplier can be determined based on these candidate historical purchase orders.

[0102] For example, the supplier's short-term capability characteristics may include but are not limited to at least one of the following:

[0103] The target supplier's average daily delivery volume over the past period of time, such as the target supplier's average daily delivery volume over the past month (determined based on all candidate historical purchase orders over the past month), the target supplier's average daily delivery volume over the past 2 months, the target supplier's average daily delivery volume over the past 3 months, the target supplier's average daily delivery volume over the past 4 months, and the target supplier's average daily delivery volume over the past 6 months.

[0104] The average delivery days interval of the target supplier. For example, based on all candidate historical purchase orders in the past 6 months, the historical average delivery days interval of the target supplier can be determined.

[0105] The minimum total procurement days of the target supplier, the maximum total procurement days of the target supplier, and the average total procurement days of the target supplier. For example, the total procurement days corresponding to all candidate historical procurement orders of the target supplier in the past 6 months can be counted, and then the minimum total procurement days among all total procurement days, the maximum total procurement days among all total procurement days, and the average of all total procurement days can be determined.

[0106] The minimum delivery quantity of the target supplier, the maximum delivery quantity of the target supplier, and the average delivery quantity of the target supplier. For example, the delivery quantities corresponding to all candidate historical purchase orders of the target supplier in the past 6 months can be counted, and then the minimum delivery quantity among all delivery quantities, the maximum delivery quantity among all delivery quantities, and the average of all delivery quantities can be calculated as the average delivery quantity.

[0107] Regarding the material supply and demand characteristics, the material supply and demand characteristics include at least one of the following: the demand quantity of the materials to be purchased in the future period (also known as demand distribution, such as the demand quantity of the materials to be purchased in the next week, the demand quantity of the materials to be purchased in the next 2 weeks, the demand quantity of the materials to be purchased in the next 3 weeks, and the demand quantity of the materials to be purchased in the next 4 weeks), the supply quantity of the materials to be purchased in the future period (also known as supply distribution, such as the supply quantity of the materials to be purchased in the next week, the supply quantity of the materials to be purchased in the next 2 weeks, the supply quantity of the materials to be purchased in the next 3 weeks, and the supply quantity of the materials to be purchased in the next 4 weeks).

[0108] For example, the demand quantity indicates the quantity of materials to be purchased that the demander needs, and the supply quantity indicates the quantity of materials to be purchased that the demander supplies itself, that is, the quantity not purchased from the target supplier. Assuming the demand quantity is 100 and the supply quantity is 30, then the demander needs to obtain 100 materials to be purchased, the demander supplies 30 materials to be purchased, and the remaining 70 materials to be purchased need to be purchased from the target supplier.

[0109] Step 205: input the second reference feature into the trained second prediction model so that the second prediction model outputs an initial available quantity based on the second reference feature. The initial available quantity indicates that the demand side needs to purchase the quantity of the materials to be purchased, that is, the demand side's expected value for the purchase quantity of the materials to be purchased.

[0110] For example, the second prediction model can be pre-trained, and the second prediction model is used to predict the available quantity. The second prediction model can be a deep learning model or a neural network model. There is no restriction on the network structure of the second prediction model, as long as it can predict the available quantity. For example, the second prediction model can be a Lightgbm (Light GradientBoosting Machine) model.

[0111] Exemplarily, a training data set may be constructed based on historical purchase orders over a period of time (such as 6 months) in the past. The training data set may include a large amount of sample data. For each sample data, the sample data may include a second reference feature corresponding to the sample matching pair and a true label corresponding to the sample matching pair.

[0112] For the sample matching pair (sample material and sample supplier), step 204 can be used to obtain the second reference feature corresponding to the sample matching pair, which will not be repeated here. In addition, the second reference feature corresponding to the sample matching pair can be consistent with the second reference feature corresponding to the target matching pair. For example, in step 204, the second reference feature corresponding to the target matching pair includes the first characterization vector, the second characterization vector, the purchase order basic feature and the supplier inventory feature, then the second reference feature corresponding to the sample matching pair includes the first characterization vector, the second characterization vector, the purchase order basic feature and the supplier inventory feature.

[0113] For the true label corresponding to the sample matching pair, the true label is the short-term inventory quantity (that is, the actual inventory quantity corresponding to the short-term LT). For example, for this sample matching pair, the actual inventory quantity within the short-term LT (that is, the actual delivery quantity of the supplier) is obtained, and this actual inventory quantity is used as the true label.

[0114] Exemplarily, the Lightgbm model can be trained based on the training data set to obtain a trained Lightgbm model, which is called the second prediction model, and there is no restriction on the model training process.

[0115] Exemplarily, based on the trained second prediction model, in step 205, the second reference feature corresponding to the target matching pair can be input to the second prediction model, and the second prediction model outputs the initial available quantity based on the second reference feature, that is, the initial available quantity in the short term LT. At this point, the initial available quantity corresponding to the target matching pair can be predicted by a supervised learning method (i.e., the Lightgbm model).

[0116] Step 206: Generate a target purchase application form (also called a virtual purchase application form) for the materials to be purchased. The target purchase application form may include, but is not limited to, the materials to be purchased, the target supplier, the initial availability date, and the initial availability quantity. See Table 3 for an example of a target purchase application form.

[0117] Table 3

[0118]

[0119] Compared with the purchase application form in Table 1, the short-term LT, initial available date and initial available quantity in Table 3 are all predicted, while the standard LT, expected delivery date and purchase quantity in Table 1 are all pre-configured by the demander. Therefore, the target purchase application form in Table 3 can be called a virtual purchase application form.

[0120] In Table 3, Order1 etc. may represent the serial number of the purchase application, Matnr1 etc. may represent the unique identifier of the material to be purchased, and Vender1 etc. may represent the unique identifier of the target supplier. In Table 3, the supplier inventory quantity may or may not exist.

[0121] The short-term LT (short-term time interval) may indicate the time required from the start of an order to its early completion, that is, the short-term LT is less than the standard LT, and the target supplier is required to complete the order in advance. In this embodiment, the short-term LT may indicate the time interval between the initial availability date and the current date, and therefore, the sum of the short-term time interval and the current date may be used as the initial availability date. The determination method of the short-term LT may refer to step 202.

[0122] The initial availability date indicates the expected delivery date of the demander. Compared with the expected delivery date in Table 1, the initial availability date is an expected earlier delivery date, that is, the material planner hopes that the purchased materials can be delivered within the short-term interval LT, rather than within the standard interval LT.

[0123] The initial available quantity indicates the quantity of materials to be purchased that the demander needs to purchase, that is, the demander's expectation of the materials to be purchased. The initial available quantity indicates the quantity that can be delivered within the short time interval LT.

[0124] In summary, we can obtain the short-term supply capacity of materials, predict the short-term LT and short-term warehousing quantity of materials, obtain a virtual purchase requisition (target purchase requisition form), and display the short-term LT, initial available date (algorithm-predicted available date) and initial available quantity (algorithm-predicted available quantity) through the target purchase requisition form.

[0125] In one possible implementation, for the procurement process, there is a difference between the supply and demand table data pulled daily by the APS system and the actual inventory data. The data difference is largely caused by the actual operation process. Therefore, the uncertainty in the business can be estimated and the gap between business documents and real scenarios can be reduced.

[0126] For example, taking an unconfirmed purchase order as an example, the uncertainty of the supply and demand table data is explained. After a purchase order is generated, since the purchase contract has just been signed and the specific delivery date has not been determined with the supplier, the purchase order is called an unconfirmed purchase order. The expected availability date of the purchase order in the supply and demand table is the expected delivery date signed on the purchase contract. When there is a gap in a certain material, the material planner formally sends a pick-up application to the supplier. The buyer will communicate with the supplier to negotiate and determine the expected delivery date (hereinafter referred to as the negotiated availability date). The supplier will then contact the logistics company for delivery. At this time, the purchase order is called a confirmed purchase order. The expected availability date (negotiated availability date) of the purchase order in the supply and demand table is the expected delivery date negotiated by both parties. Finally, after the material is delivered to the company's warehouse, the actual delivery date (i.e., the actual warehousing date) and the actual delivery quantity (i.e., the actual warehousing quantity) are obtained.

[0127] In summary, when outstanding orders are in an unconfirmed state, there may be a large gap between the estimated availability date in the supply and demand table and the final actual availability date. Therefore, the gap between the two dates needs to be reduced.

[0128] Based on this, in this embodiment, it is necessary to obtain a predicted availability date, and the predicted availability date is used instead of the negotiated availability date. The predicted availability date can reduce the gap with the actual warehousing date.

[0129] See also Figure 3 FIG. 1 is a flowchart of a method for generating a purchase order, and the process may include:

[0130] Step 301: Obtain the material to be purchased and the target supplier, and the target supplier provides the material to be purchased to the demander. For the convenience of description, the combination of the material to be purchased and the target supplier can be called a target matching pair, that is, the target matching pair can include the material to be purchased and the target supplier.

[0131] Step 302: Determine whether the material to be purchased is a non-infinite supply material.

[0132] If not, that is, the material to be purchased is an unlimited supply material, step 303 can be executed.

[0133] If yes, that is, the material to be purchased is a non-infinite supply material, step 304 may be executed.

[0134] Exemplarily, the process of determining an infinite supply material and a non-infinite supply material may include the following steps:

[0135] Step 3021: For each material, statistics are collected on the warehousing data of the material in a historical period (such as the past month). The warehousing data may include but is not limited to the first ratio value and / or the second ratio value.

[0136] Exemplarily, the first ratio value may be the ratio of the number of days that materials are put into storage to the historical time. Assuming the historical time is 1 month, such as 30 days, the number of days that materials are put into storage within 30 days may be counted. For example, if the number of days that materials are put into storage is 10 days, it means that there are material put into storage operations on 10 days out of 30 days, and the first ratio value may be 10 / 30. For another example, statistics may also be performed based on working days. Assuming the historical time is 1 month, and there are 21 working days, the number of days that materials are put into storage within 21 working days may be counted. For example, if the number of days that materials are put into storage is 10 days, it means that there are material put into storage operations on 10 days out of 21 days, and the first ratio value may be 10 / 21.

[0137] Exemplarily, the second ratio value may be a ratio value of the candidate quantity to the historical time. For each day in the historical time, if the actual material storage quantity of the K days after the day is greater than the expected material storage quantity, the candidate quantity may be increased by 1, otherwise, the candidate quantity remains unchanged, and K may be a positive integer.

[0138] For example, assuming that the historical time is 1 month, such as 30 days, and K is 3, then for the first day of the month, the actual material inventory quantity from the 2nd day to the 4th day (i.e. [T+1, T+3]) can be counted (indicating that the number of materials actually entered in these 3 days), and the expected material inventory quantity from the 2nd day to the 4th day can be counted (indicating that the number of materials expected to be entered in these 3 days). If the actual material inventory quantity is greater than the expected material inventory quantity, the candidate quantity can be increased by 1, and the initial value of the candidate quantity can be 0. If the actual material inventory quantity is not greater than the expected material inventory quantity, the candidate quantity can be kept unchanged.

[0139] For the second day of a month, the actual material incoming quantity and the expected material incoming quantity for the third to fifth days can be counted. If the actual material incoming quantity is greater than the expected material incoming quantity, the candidate quantity is increased by 1. If the actual material incoming quantity is not greater than the expected material incoming quantity, the candidate quantity remains unchanged.

[0140] By analogy, after performing the above operation for 30 days in a month, the final number of candidates can be obtained. If the number of candidates is 10, the second ratio value can be 10 / 30.

[0141] For another example, the second ratio value may be calculated based on working days. The implementation method is similar, except that the actual material storage quantity and the estimated material storage quantity are both for working days, which will not be described in detail here.

[0142] The actual quantity of materials entering the warehouse can be obtained based on the warehouse entry document of the warehouse entry process.

[0143] For the estimated quantity of materials entering the warehouse, an APS supply and demand table can be provided in advance. The APS supply and demand table includes cross-sectional data of daily fixed-point storage, such as the latest inventory, outstanding demand, outstanding supply and other information at the time of daily fixed-point storage. On this basis, the APS supply and demand table can include the estimated quantity of materials entering the warehouse every day (that is, the number of materials expected to be entered into the warehouse). Therefore, the estimated quantity of materials entering the warehouse can be obtained based on the APS supply and demand table.

[0144] Step 3022: Based on the storage data of each material, an abnormal data analysis method is used to analyze all materials to obtain an abnormal data candidate set, and the abnormal data candidate set includes multiple abnormal materials.

[0145] Exemplarily, the abnormal data analysis method is used to analyze abnormal materials from all materials. For example, the first proportion values ​​of most materials are within a certain interval, and only a few materials have first proportion values ​​that are not within the interval, and these few materials are abnormal materials. There is no limitation on how to analyze abnormal materials from all materials in this embodiment, as long as abnormal materials can be analyzed from all materials.

[0146] For example, the abnormal data analysis method can be the Isolation Forest algorithm. The incoming data of each material can be input into the Isolation Forest algorithm. The Isolation Forest algorithm is used to analyze all materials to obtain an abnormal data candidate set, which includes multiple abnormal materials. The Isolation Forest algorithm is an efficient anomaly detection algorithm used to identify outliers in data. The Isolation Forest algorithm identifies anomalies by isolating outliers in the data. The core idea of ​​the algorithm is that abnormal data is usually different from most data and can be easily isolated. The Isolation Forest algorithm randomly selects a feature and randomly selects a value range of the feature to split the data, and repeats this process until all data is isolated.

[0147] In summary, the incoming data of each material can be input into the isolation forest algorithm. The isolation forest algorithm constructs a binary tree by randomly selecting features and partitioning points, and divides the material into multiple subsets until each subset contains only one material or reaches the maximum depth of the tree. Abnormal materials are usually divided into leaf nodes quickly in the tree, and their path length is short. Obviously, through the isolation forest algorithm, all materials can be analyzed to obtain abnormal data candidate sets, which include multiple abnormal materials.

[0148] Step 3023: for each abnormal material in the abnormal data candidate set, if the storage data of the abnormal material meets the preset conditions, the abnormal material is determined as an unlimited supply material; if the storage data of the abnormal material does not meet the preset conditions, the abnormal material is determined as a non-infinite supply material.

[0149] For example, if the warehousing data includes a first ratio value and a second ratio value, then when the first ratio value of the abnormal material is greater than a first threshold value (which can be configured based on experience), and the second ratio value of the abnormal material is greater than a second threshold value (which can be configured based on experience), the warehousing data of the abnormal material meets the preset condition. Otherwise, the warehousing data of the abnormal material does not meet the preset condition. If the warehousing data includes a first ratio value, then when the first ratio value of the abnormal material is greater than the first threshold value, the warehousing data of the abnormal material meets the preset condition. Otherwise, the warehousing data of the abnormal material does not meet the preset condition. If the warehousing data includes a second ratio value, then when the second ratio value of the abnormal material is greater than the second threshold value, the warehousing data of the abnormal material meets the preset condition. Otherwise, the warehousing data of the abnormal material does not meet the preset condition.

[0150] Exemplarily, only step 3022 may be performed without performing step 3023. Based on this, after obtaining the abnormal data candidate set, each abnormal material in the abnormal data candidate set is used as an infinite supply material.

[0151] Exemplarily, only step 3023 may be executed without executing step 3022. Based on this, for each material, if the warehousing data of the material meets the preset conditions, the material is determined as an unlimited supply material; if the warehousing data of the material does not meet the preset conditions, the material is determined as a non-infinite supply material.

[0152] Step 3024: The remaining materials except the unlimited supply materials among all the materials are determined as non-infinite supply materials. For example, unlimited supply materials refer to the target supplier having a large amount of materials, and the demander does not need to wait, and the target supplier can provide materials at any time. Non-infinite supply materials refer to the target supplier not having a large amount of materials, and the demander needs to wait, and the target supplier cannot provide materials at any time.

[0153] In summary, for each material, it can be known whether the material is an unlimited supply material or a non-infinite supply material. Based on this, for the materials to be purchased, it can be known whether the materials to be purchased are unlimited supply materials or non-infinite supply materials. For unlimited supply materials (i.e. unlimited advance materials), the anomaly detection algorithm is used to identify materials that are frequently over-delivered recently. After manual confirmation or without manual confirmation, they are defined as unlimited advance materials, and it is believed that the supply capacity of such materials has no upper limit.

[0154] Step 303: If the material to be purchased is an unlimited supply material, the predicted availability date is determined based on the current date. For example, the date after the current date plus 1 is used as the predicted availability date, and the predicted availability date is T+1.

[0155] For example, if the material to be purchased is an unlimited supply material, the delivery mode of the material to be purchased is an unlimited material mode, that is, the target supplier has a large number of materials to be purchased, and the demander does not need to wait, and the target supplier can provide the materials to be purchased at any time. Based on this, the predicted availability date is T+1 day, and T represents the current date.

[0156] Step 304: If the material to be purchased is a non-infinite supply material, the delivery mode of the material to be purchased is obtained, and the delivery mode can be a delivery date mode or a waiting notification mode. If the delivery mode is a delivery date mode, step 305 can be executed; if the delivery mode is a waiting notification mode, step 306 can be executed.

[0157] Step 305: obtain a first reference feature based on the material to be purchased and the target supplier, and determine the forecast lead time based on the first reference feature; determine the forecast available date based on the forecast lead time and the negotiated available date; wherein the forecast available date may be the forecasted delivery date.

[0158] Exemplarily, the process of obtaining the predicted availability date may include the following steps:

[0159] Step 3051: construct a bipartite graph of materials and suppliers. The bipartite graph may include all material nodes and all supplier nodes. The edge weight of the bipartite graph is the number of times the supplier has delivered the material in the historical time.

[0160] Step 3052: Query the material node corresponding to the material to be purchased and the supplier node corresponding to the target supplier from the bipartite graph. For example, the bipartite graph includes material nodes corresponding to each material, and the material node corresponding to the material to be purchased can be queried from these material nodes. The bipartite graph includes supplier nodes corresponding to each supplier, and the supplier node corresponding to the target supplier can be queried from these supplier nodes.

[0161] Step 3053: Based on the bipartite graph, a first characterization vector of a material node corresponding to the material to be purchased and a second characterization vector of a supplier node corresponding to the target supplier are determined by using graph characterization. The closer the connection between the material to be purchased and the target supplier, the more similar the first characterization vector and the second characterization vector are.

[0162] Exemplarily, steps 3051 to 3053 may refer to step 204 and will not be described in detail here.

[0163] Step 3054, obtain the first reference feature. For example, the first reference feature may only include the first characterization vector and the second characterization vector. Alternatively, in addition to the first characterization vector and the second characterization vector, the first reference feature may also include but is not limited to at least one of the following features: purchase order basic features, historical delivery features, supplier historical delivery features, and material historical delivery features. Alternatively, the first reference feature may include but is not limited to at least one of the following features: the first characterization vector, the second characterization vector, purchase order basic features, historical delivery features, supplier historical delivery features, and material historical delivery features.

[0164] With respect to the purchase order basic features, the purchase order basic features may include but are not limited to at least one of the following: the number of days between the generation date of the target purchase order and the current date (i.e., the number of days between the purchase order creation date and the current date). When generating the first reference feature of the target matching pair, the generation date of the target purchase order may be the current date, and therefore, the number of days between the generation date and the current date may be 0. However, in the training process of the first prediction model (see the subsequent process for the training process), the target purchase order is a historical purchase order, and the generation date of the historical purchase order is earlier than the current date, and therefore, the number of days between the generation date and the current date is not 0.

[0165] The number of days between the demander's expected delivery date and the current date (i.e. the number of days between the expected delivery date and the current date), and the number of days between the supplier's expected delivery date and the current date (i.e. the number of days between the supplier's feedback date and the current date). For example, during the procurement process, the demander and the target supplier need to negotiate the delivery date, and the negotiated delivery date is the negotiated availability date.

[0166] In the process of negotiating the delivery date, the delivery date expected by the demander is called the demander's expected delivery date. Based on whether the supplier agrees to the delivery date, it can be known that the demander's expected delivery date may or may not be the final negotiated available date.

[0167] In the process of negotiating the delivery date, the delivery date expected by the target supplier is called the supplier's expected delivery date. Based on whether the demander agrees to the delivery date, it can be known that this supplier's expected delivery date may or may not be the final negotiated available date.

[0168] The purchase quantity of the materials to be purchased can be obtained by referring to Table 1 or Table 2. The purchase quantity indicates that the demander needs to purchase the materials to be purchased in this quantity, that is, the demander's expected value.

[0169] Standard LT, as shown in Table 1, can be known as the standard LT (ie, standard time interval).

[0170] For the historical delivery feature, the historical delivery feature can also be called the historical lead time sequence statistical feature. The historical delivery feature includes the first statistical feature of the target matching pair. The target matching pair includes the material to be purchased and the target supplier. The first statistical feature is obtained by counting the lead time and / or the total days of procurement time of the historical purchase orders of the target matching pair. For example, all historical purchase orders of the target matching pair in the past period of time (such as 6 months) can be obtained. For each historical purchase order, the lead time corresponding to the historical purchase order can be determined. Assuming that the negotiated available date in the historical purchase order is A, and the actual warehousing date corresponding to the historical purchase order is B, the lead time corresponding to the historical purchase order is AB. Multiple historical purchase orders will correspond to multiple lead times, and these lead times are combined into a lead time sequence. On this basis, the first statistical feature can be obtained by counting the lead time sequence. For example, the first statistical feature can be obtained by extracting features from the lead time sequence through a feature extractor, or by extracting features from the lead time sequence through a feature engineering method, and there is no restriction on this.

[0171] For example, all historical purchase orders of the target match in the past period of time can be obtained. For each historical purchase order, the total purchase time corresponding to the historical purchase order can be determined (i.e., the difference between the actual warehousing date and the purchase order creation date). Multiple historical purchase orders will correspond to multiple total purchase time, and these total purchase time are combined into a time sequence. On this basis, the time sequence can be statistically analyzed to obtain the first statistical feature, and there is no restriction on this statistical method.

[0172] With respect to the supplier's historical delivery characteristics, the supplier's historical delivery characteristics include a second statistical characteristic of the target supplier, and the second statistical characteristic is obtained by counting the lead time days and / or the total days of procurement time of the target supplier's historical purchase orders. For example, all historical purchase orders of the target supplier in the past period of time can be obtained, and for each historical purchase order, the lead time days corresponding to the historical purchase order can be determined, and the multiple lead time days corresponding to multiple historical purchase orders can be combined into an lead time sequence, and the second statistical characteristic can be obtained by counting the lead time sequence. The total procurement time corresponding to the historical purchase order can be determined, and the multiple total procurement time corresponding to multiple historical purchase orders can be combined into a time sequence, and the second statistical characteristic can be obtained by counting the time sequence.

[0173] With respect to the historical delivery characteristics of materials, the historical delivery characteristics of materials may include a third statistical characteristic of the materials to be purchased, and the third statistical characteristic is obtained by counting the number of days in advance and / or the total number of days it takes to purchase the historical purchase orders of the materials to be purchased. For example, all historical purchase orders of the materials to be purchased in the past period of time may be obtained, and for each historical purchase order, the number of days in advance corresponding to the historical purchase order may be determined, and the multiple days in advance corresponding to the multiple historical purchase orders may be combined into a sequence of days in advance, and the third statistical characteristic may be obtained by counting the sequence of days in advance. The total number of days it takes to purchase the historical purchase order may be determined, and the multiple days in total number of days it takes to purchase the multiple historical purchase orders may be combined into a sequence of days in time, and the third statistical characteristic may be obtained by counting the sequence of days in time.

[0174] Step 3055: input the first reference feature to the trained first prediction model so that the first prediction model outputs the predicted lead time days based on the first reference feature. The predicted lead time days indicates how many days in advance the delivery can be completed based on the negotiated available date. This advance date is the predicted lead time days.

[0175] For example, a first prediction model can be pre-trained, and the first prediction model is used to predict the number of days in advance. The first prediction model can be a deep learning model or a neural network model, and there is no restriction on the network structure, as long as it can predict the number of days in advance. For example, the first prediction model can be a Lightgbm model.

[0176] Exemplarily, a training data set may be constructed based on historical purchase orders over a period of time (such as 6 months) in the past. The training data set may include a large amount of sample data. For each sample data, the sample data may include a first reference feature corresponding to the sample matching pair and a true label corresponding to the sample matching pair.

[0177] For the sample matching pair (sample material and sample supplier), steps 3051-3054 can be used to obtain the first reference feature corresponding to the sample matching pair. The sample matching pair is a matching pair whose actual number of deliveries is not less than a given threshold, that is, the sample matching pair is a valid sample. The first reference feature corresponding to the sample matching pair can be consistent with the first reference feature corresponding to the target matching pair. For example, when the first reference feature corresponding to the target matching pair includes the first characterization vector, the second characterization vector, and the purchase order basic feature, the first reference feature corresponding to the sample matching pair includes the first characterization vector, the second characterization vector, and the purchase order basic feature.

[0178] For the true label corresponding to the sample matching pair, the true label is the number of days in advance. For example, for the sample matching pair, the actual warehousing date and the negotiated availability date corresponding to the purchase order of the sample matching pair are obtained, and the difference between the negotiated availability date and the actual warehousing date is used as the true label.

[0179] Exemplarily, the Lightgbm model can be trained based on the training data set to obtain a trained Lightgbm model, which is called the first prediction model, and there is no restriction on the model training process.

[0180] Exemplarily, based on the trained first prediction model, in step 3055, the first reference feature corresponding to the target matching pair can be input to the first prediction model, and the first prediction model determines the prediction lead time based on the first reference feature and outputs the prediction lead time. At this point, the prediction lead time corresponding to the target matching pair can be predicted by the supervised learning method (i.e., the Lightgbm model).

[0181] Step 3056: Determine the predicted available date based on the predicted days in advance and the negotiated available date.

[0182] Exemplarily, as shown in Table 1, the purchase application form may include an expected delivery date, which may be configured by the material planner. On this basis, the expected delivery date may be used as the initial availability date, which may represent the delivery date expected by the demander. However, the target supplier may agree to the initial availability date or may reject the initial availability date. Therefore, it is necessary to negotiate the delivery date between the demander and the target supplier, and the negotiated delivery date is the negotiated availability date, that is, the negotiated availability date is the delivery date negotiated between the demander and the target supplier. In summary, if the target supplier agrees to the initial availability date, the initial availability date may be determined as the negotiated availability date, or if the target supplier rejects the initial availability date, the initial availability date may be adjusted to obtain the negotiated availability date.

[0183] Alternatively, as shown in Table 3, the target purchase application form may include an initial availability date, which is predicted by an algorithm, that is, the delivery date expected by the demander is predicted, rather than the material planner configuration. On this basis, the delivery date needs to be negotiated between the demander and the target supplier, and the negotiated delivery date is the negotiated availability date. In summary, if the target supplier agrees to the initial availability date, the initial availability date can be determined as the negotiated availability date, or if the target supplier rejects the initial availability date, the initial availability date can be adjusted to obtain the negotiated availability date.

[0184] For example, if the demander's expected delivery date is greater than or equal to the sum of the standard LT and the current date, it means that the current scenario is a normal scenario, and the initial available date configured by the material planner can be used to determine the negotiated available date, as shown in Table 1. Alternatively, if the demander's expected delivery date is less than the sum of the standard LT and the current date, it means that the current scenario is a short-term delivery scenario, and the initial available date predicted by the algorithm can be used (see Figure 3 The forecast process shown in the figure) determines the negotiation available date, see Table 3.

[0185] For example, after obtaining the negotiated availability date and the predicted lead time, the difference between the negotiated availability date and the predicted lead time can be used as the predicted availability date, which can be the predicted delivery date, that is, the delivery date predicted by the algorithm, rather than the configured delivery date.

[0186] Exemplarily, the predicted available date can also be determined in the following manner: the difference between the negotiated available date and the predicted number of days in advance is taken as date 1, the sum of the current date and a fixed time (such as 4 or 5 days, etc.) is taken as date 2 (i.e., T+4), and the larger value of date 1 and date 2 is taken as the predicted available date.

[0187] From the above, we can see that for the delivery date mode, some suppliers can deliver goods in advance, considering that they have a good relationship with the demand side, or the supplier hopes to get a good supplier rating. To this end, the supervised learning method is used to estimate the next advance delivery days for each matching pair (material, supplier), so as to determine the predicted available date based on the negotiated available date and the predicted advance days, that is, the advanced delivery date.

[0188] Step 306: Obtain logistics days based on the materials to be purchased and the target supplier, and determine a predicted availability date based on the delivery notice date and the logistics days. The predicted availability date may be the predicted delivery date.

[0189] For example, the first mapping table can be queried through the target matching pair to obtain the logistics days corresponding to the target matching pair, which can include the material to be purchased and the target supplier. The logistics days can be the number of days required for the material to be delivered from the target supplier to the demander. For example, the period from when the target supplier receives the purchase order to when the demander finishes warehousing the material is the logistics days.

[0190] For example, a first mapping table may be maintained in advance, and the first mapping table may include a correspondence between matching pairs (i.e., combinations of materials and suppliers) and logistics days. On this basis, the first mapping relationship may be queried through a target matching pair to obtain the logistics days corresponding to the target matching pair.

[0191] Exemplarily, the maintenance process for the first mapping table may include the following steps:

[0192] Step 3061: Based on historical purchase orders, obtain sample matching pairs and historical logistics time series corresponding to the sample matching pairs. The sample matching pairs may include sample materials and sample suppliers (i.e., the combination of sample materials and sample suppliers is called a sample matching pair). The historical logistics time series may include multiple historical logistics days, and each historical logistics day represents the number of days required for the sample material to be delivered from the sample supplier to the demander.

[0193] For example, for each matching pair (material, supplier), the total number of historical purchase orders of the matching pair in the past period of time (such as 6 months) can be counted. If the total number of historical purchase orders is less than a preset threshold, the historical purchase orders of the matching pair are filtered. If the total number of historical purchase orders is greater than or equal to the preset threshold, the matching pair is used as a sample matching pair (i.e., a valid matching pair).

[0194] Each historical purchase order of the sample matching pair corresponds to a historical logistics day. For example, from the time when the sample supplier obtains the historical purchase order (that is, the creation date of the historical purchase order is the start time) to the end of the demander's material entry into the warehouse for the historical purchase order (that is, the actual entry date is the end time), this period of time is the historical logistics day. The historical logistics day can include the communication time of the two parties and the transportation time of the logistics company. Based on all the historical purchase orders of the sample matching pair, the historical logistics time series corresponding to the sample matching pair is obtained, and the historical logistics time series includes multiple historical logistics days.

[0195] Step 3062: For each sample matching pair, the next logistics day of the sample matching pair can be predicted based on the multiple historical logistics days corresponding to the sample matching pair. For example, the next logistics day can be obtained by analyzing multiple historical logistics days, and there is no restriction on this process.

[0196] For example, abnormal data (i.e., abnormal historical logistics days) can be removed from the historical logistics time series. This process is optional, and subsequent analysis can also be performed based on all historical logistics days. For example, the 3Sigma criterion and / or the box plot method can be used to remove abnormal data from the historical logistics time series. There is no restriction on this process, as long as the abnormal historical logistics days in the historical logistics time series can be found and removed, the processed historical logistics time series can be obtained.

[0197] Exemplarily, based on the historical logistics time series without abnormal data or the historical logistics time series with abnormal data removed, multiple historical logistics days in the historical logistics time series can be obtained, and the next logistics days can be predicted based on the multiple historical logistics days. For example, a non-parametric statistical method is used to predict the next logistics days. The non-parametric statistical method can be the SAA (Sample Average Approximation) method or other methods, and there is no restriction on this. The SAA method estimates the true distribution by sampling from the historical sample distribution, and the sampling probability of recent samples is assumed to be high.

[0198] Step 3063: Record the correspondence between the sample matching pairs and the next logistics days in the first mapping table.

[0199] For example, for each sample matching pair, after obtaining the next logistics days corresponding to the sample matching pair, the corresponding relationship between the sample matching pair and the next logistics days can be recorded in the first mapping table. At this point, the first mapping table can be maintained. In step 306, the first mapping table can be queried through the target matching pair to obtain the logistics days corresponding to the target matching pair.

[0200] After obtaining the logistics days corresponding to the target matching pair, the predicted availability date can be determined based on the delivery notification date and the logistics days. For example, the sum of the delivery notification date and the logistics days is used as the predicted availability date of the target matching pair, and the predicted availability date can be the predicted delivery date.

[0201] After the demander negotiates the delivery date with the target supplier, the demander usually reserves a period of time for the target supplier to produce the materials to be purchased during this period. After a period of time, the demander sends a delivery notice to the target supplier. The date of sending the delivery notice is called the delivery notice date. In order to deliver the materials to be purchased to the demander before the negotiated availability date, the delivery notice date is earlier than the negotiated availability date.

[0202] Exemplarily, the predicted available date can also be determined in the following manner: the sum of the delivery notice date and the logistics days is taken as date 1, and the negotiated available date between the demander and the target supplier is taken as date 2. Based on this, the smaller value of date 1 and date 2 can be used as the predicted available date.

[0203] In summary, it can be seen that for the waiting notification mode, considering that the communication time and logistics time required for materials are less than 7 days, the next logistics days for each matching pair (material, supplier) are estimated through non-parametric statistical methods, so as to determine the predicted available date based on the next logistics days and the delivery notification date.

[0204] Step 307: Generate a target purchase order (also called a virtual purchase order) for the materials to be purchased. The target purchase order may include, but is not limited to, the materials to be purchased, the target supplier, the predicted availability date, and the negotiated availability date. See Table 4 for an example of a target purchase order.

[0205] Table 4

[0206]

[0207] Compared with the purchase order table in Table 2, the predicted available date in Table 4 is predicted, while the negotiated available date in Table 2 is pre-configured by the demander. Therefore, the target purchase order (also called the target purchase table) in Table 4 can be called a virtual purchase order (also called a virtual purchase table).

[0208] In Table 4, Order1 etc. may represent the serial number of the purchase application, Matnr1 etc. may represent the unique identifier of the material to be purchased, and Vender1 etc. may represent the unique identifier of the target supplier.

[0209] The delivery mode of the material to be purchased (material delivery mode) can be the waiting notification mode (abbreviated as waiting notification), the delivery date mode (abbreviated as delivery date) and the unlimited material mode (abbreviated as unlimited material). For example, if the material to be purchased is an unlimited supply material, the delivery mode of the material to be purchased is the unlimited material mode. If the material to be purchased is a non-infinite supply material, the delivery mode of the material to be purchased is the waiting notification mode or the delivery date mode, that is, the delivery mode of the material to be purchased can be pre-configured as the waiting notification mode or the delivery date mode.

[0210] For the negotiated available quantity, the negotiated available quantity is the delivery quantity negotiated between the demander and the supplier. As shown in Table 1, the purchase application form may include the purchase quantity, and the purchase quantity is used as the initial available quantity, indicating that the demander needs to purchase the quantity of the materials to be purchased. However, the target supplier may agree to the initial available quantity or may reject the initial available quantity. Therefore, the material quantity needs to be negotiated between the demander and the target supplier, and the negotiated material quantity is the negotiated available quantity, that is, the negotiated available quantity can be the material quantity negotiated between the demander and the target supplier. Alternatively, as shown in Table 3, the target purchase application form may include the initial available quantity, and the initial available quantity can be predicted by an algorithm, that is, the material quantity expected by the demander is predicted, rather than the material planner configuration. On this basis, the material quantity needs to be negotiated between the demander and the target supplier, and the negotiated material quantity is the negotiated available quantity. In summary, if the target supplier agrees to the initial available quantity, the initial available quantity can be determined as the negotiated available quantity, or if the target supplier rejects the initial available quantity, the initial available quantity can be adjusted to obtain the negotiated available quantity.

[0211] For example, if the demander's expected delivery date is greater than or equal to the sum of the standard LT and the current date, it means that the current scenario is a normal scenario, and the initial available quantity configured by the material planner can be used to determine the negotiated available quantity, as shown in Table 2. Alternatively, if the demander's expected delivery date is less than the sum of the standard LT and the current date, it means that the current scenario is a short-term delivery scenario, and the initial available quantity predicted by the algorithm can be used to determine the negotiated available quantity (see Figure 3 The prediction process is shown in Table 4.

[0212] Predicted availability date: The predicted availability date is obtained by algorithm prediction. If the material to be purchased is an unlimited supply material, the predicted availability date is determined in step 303. If the material to be purchased is a non-unlimited supply material and the delivery mode is the delivery date mode, the predicted availability date is determined in step 305. If the material to be purchased is a non-unlimited supply material and the delivery mode is the wait notification mode, the predicted availability date is determined in step 306.

[0213] In summary, the material supply capacity can be predicted, which includes predicting the available date, obtaining a virtual purchase order (target purchase order), and displaying the predicted available date through the target purchase order.

[0214] In one possible implementation, combining Figure 2 Target Purchase Requisition Form and Figure 3 The target purchase order can output the material supply capacity forecast report, which consists of two sub-reports. 1 is the forecast report of the available date of outstanding purchase orders, that is, the target purchase order in Table 4. The target purchase order is used to correct the estimated available date of unconfirmed purchase documents in the system, that is, to correct it by predicting the available date. 2 is the material short-term supply capacity forecast report, that is, the target purchase application form in Table 3. The target purchase application form is based on the forecast of the supplier's supply capacity and creates a non-existent virtual purchase application document.

[0215] In the present application, a method for predicting material supply capacity is proposed. Figure 4 FIG. 1 is a flow chart of a material supply capacity prediction method, which may include the following steps:

[0216] Step 401: Obtain the whole process data of material procurement. The whole process data may include supply and demand table, purchase application table, purchase order table, inbound delivery form, material movement voucher table, material master data table, etc.

[0217] For example, the supply and demand table can include the cross-sectional data of the daily fixed-point storage of the APS system, including the latest inventory, outstanding demand, outstanding supply and other information at the time of daily fixed-point storage. The purchase application form can include purchase application related information, such as materials, purchase quantity, expected delivery date, etc. The purchase order table can include purchase order related information, such as materials, suppliers, purchase quantity, expected delivery date, etc. The inbound delivery order table can include supplier delivery record information, such as materials, suppliers, expected delivery date, etc. The material movement voucher table can include supplier delivery warehousing information, such as materials, suppliers, actual warehousing date, etc. The material master data table can include material procurement LT, material delivery mode, etc.

[0218] Step 402: predict the available time of unconfirmed purchase orders of different delivery modes (i.e., predict the available time) to obtain virtual purchase orders. Figure 3 As shown, no further description is given here.

[0219] Step 403: Forecast the short-term LT, initial availability date and initial availability quantity of different materials to obtain a virtual purchase requisition. Figure 2 As shown, no further description is given here.

[0220] Step 404: Based on the virtual purchase order and the virtual purchase application, a material supply capacity forecast report is output. The material supply capacity forecast report consists of two sub-reports: 1 is an outstanding purchase order available date forecast report, i.e., a target purchase order; 2 is a material short-term supply capacity forecast report, i.e., a target purchase application form.

[0221] In this embodiment, the material supply capacity prediction process involves virtual pull and virtual replenishment. Virtual pull is used to predict the available time of unconfirmed purchase orders of different delivery modes. Figure 3 As shown, virtual replenishment orders are used to predict the short-term LT and short-term warehousing quantity of different materials, see Figure 2 shown.

[0222] In this embodiment, starting from the business process of raw material procurement, material supply capacity prediction is performed for each business link (purchase application-purchase pickup-logistics distribution). Different prediction technologies are used in different business links, such as: supervised learning, clustering algorithm, anomaly detection, etc., and finally a material supply capacity prediction report is output.

[0223] As can be seen from the above technical solutions, in the embodiment of the present application, the demander can respond to the user's order in time, thereby improving the order fulfillment rate. By predicting the material supply capacity (such as predicting the available date), a method for estimating the uncertainty of the raw material supply capacity is given to reduce the gap between the target purchase order and the actual scenario, that is, the predicted available date is closer to the actual delivery date, reducing the uncertainty of the material supply capacity, effectively reducing the time of the entire link of urgent orders, and improving the work efficiency of planners. It is possible to provide the material plan with a material supply capacity forecast report for a period of time in the future (the forecast report on the available date of outstanding purchase orders and the forecast report on the short-term supply capacity of materials), and provide issuance suggestions and assistance when the material plan performs actual material extraction operations. Starting from the entire business process of raw material procurement, all purchase orders are divided into three different delivery modes, and purchase orders of different delivery modes use different forecasting technologies, for example, the available time forecast of delivery documents by delivery date based on supervised learning, the available time forecast of delivery documents based on non-parametric statistics, and the available time forecast of unlimited supply documents based on anomaly detection.

[0224] Taking into account that the supplier delivery time LT of different raw materials varies greatly, and the supplier delivery time LT is the decisive factor in the final supplier delivery quantity, a two-stage algorithm of "short-term LT estimation + short-term supply forecast" is proposed. The short-term LT of each material is identified based on the clustering algorithm, and the short-term supply capacity of each material is predicted based on the supervised learning method, so as to more accurately predict the short-term supply capacity of different matching pairs.

[0225] Based on the same application concept as the above method, a purchase order generation device is proposed in the embodiment of the present application, see Figure 5 FIG. 1 is a schematic diagram of the structure of the device, and the device may include:

[0226] The acquisition module 51 is used to acquire the delivery mode of the material to be purchased if the material to be purchased is a non-infinite supply material, and the delivery mode is a delivery date mode or a wait notification mode; wherein the delivery date mode means that the material is delivered according to the negotiated availability date, and the negotiated availability date is the delivery date negotiated between the demander and the supplier; the wait notification mode means that the material is delivered according to the delivery notification date, and the delivery notification date is the date when the demander sends the delivery notification to the supplier, and the delivery notification date is earlier than the negotiated availability date;

[0227] The determination module 52 is used for obtaining a first reference feature based on the material to be purchased and the target supplier if the delivery mode is the delivery date mode, determining the forecast lead time based on the first reference feature; determining the forecast availability date based on the forecast lead time and the negotiated availability date; if the delivery mode is the wait notification mode, obtaining logistics days based on the material to be purchased and the target supplier, the logistics days being the days required for the material to be delivered from the target supplier to the demander; determining the forecast availability date based on the delivery notification date and the logistics days; wherein the forecast availability date is the forecasted delivery date;

[0228] The generating module 53 is used to generate a target purchase order for the material to be purchased, wherein the target purchase order includes the material to be purchased, the target supplier, the predicted availability date and the negotiated availability date.

[0229] Exemplarily, the determination module 52 obtains a first reference feature based on the material to be purchased and the target supplier, and determines the forecast lead time based on the first reference feature specifically for:

[0230] Construct a bipartite graph of materials and suppliers, wherein the bipartite graph includes all material nodes and all supplier nodes, and the edge weight of the bipartite graph is the number of times the supplier has delivered the material in the historical time;

[0231] Querying the material node corresponding to the material to be purchased and the supplier node corresponding to the target supplier from the bipartite graph; based on the bipartite graph, using graph representation to determine a first representation vector of the material node and a second representation vector of the supplier node; wherein the closer the connection between the material to be purchased and the target supplier, the more similar the first representation vector and the second representation vector;

[0232] Acquire a first reference feature, where the first reference feature includes the first characterization vector and the second characterization vector; input the first reference feature into a trained first prediction model, and have the first prediction model output the predicted advance days based on the first reference feature.

[0233] Exemplarily, the first reference feature also includes at least one of the following features:

[0234] The basic features of the purchase order include at least one of the following: the number of days between the generation date of the target purchase order and the current date, the number of days between the demander's expected delivery date and the current date, the number of days between the supplier's expected delivery date and the current date, and the purchase quantity of the material to be purchased;

[0235] Historical delivery characteristics, the historical delivery characteristics comprising a first statistical characteristic of a target matching pair, the target matching pair comprising the material to be purchased and the target supplier, the first statistical characteristic being obtained by counting the number of days in advance and / or the total number of days it takes to purchase a historical purchase order of the target matching pair;

[0236] The supplier's historical delivery characteristics include the second statistical characteristics of the target supplier, which is obtained by counting the number of lead times and / or the total number of days required for purchase of the target supplier's historical purchase orders; the material's historical delivery characteristics include the third statistical characteristics of the material to be purchased, which is obtained by counting the number of lead times and / or the total number of days required for purchase of the material to be purchased.

[0237] Exemplarily, when the determination module 52 obtains the logistics days based on the material to be purchased and the target supplier, it is specifically used to: query the first mapping table through the target matching pair to obtain the logistics days corresponding to the target matching pair; wherein the target matching pair includes the material to be purchased and the target supplier, and the first mapping table includes the corresponding relationship between the matching pair and the logistics days;

[0238] The maintenance process of the first mapping table includes: obtaining a sample matching pair and a historical logistics time series corresponding to the sample matching pair based on a historical purchase order, wherein the sample matching pair includes a sample material and a sample supplier, and the historical logistics time series includes a plurality of historical logistics days, and each historical logistics day indicates the number of days required for the sample material to be delivered from the sample supplier to the demander;

[0239] Predicting the next logistics day of the sample matching pair based on the multiple historical logistics days;

[0240] The correspondence between the sample matching pairs and the next logistics days is recorded in the first mapping table.

[0241] Exemplarily, the determination module 52 is also used to determine the predicted available date based on the current date if the material to be purchased is an unlimited supply material; wherein, when the determination module 52 determines the unlimited supply material and the non-infinite supply material, it is specifically used to: for each material, count the storage data of the material in the historical time, and the storage data includes a first ratio value and / or a second ratio value; wherein, the first ratio value is the ratio value of the number of days the material is stored in the warehouse to the historical time, and the second ratio value is the ratio value of the candidate quantity to the historical time; for each day in the historical time, if the actual material storage quantity of the K days after the day is greater than the expected material storage quantity, then the candidate quantity is increased by 1, otherwise, the candidate quantity is kept unchanged; based on the storage data of each material, all materials are analyzed by using an abnormal data analysis method to obtain an abnormal data candidate set, and the abnormal data candidate set includes multiple abnormal materials;

[0242] For each abnormal material in the abnormal data candidate set, if the storage data of the abnormal material meets the preset condition, the abnormal material is determined as an unlimited supply material; wherein, if the storage data includes a first ratio value and a second ratio value, when the first ratio value is greater than a first threshold value, and the second ratio value is greater than a second threshold value, the storage data of the abnormal material meets the preset condition;

[0243] The remaining materials among all materials except the infinite supply materials are determined as non-infinite supply materials.

[0244] Exemplarily, the acquisition module 51 is further used to query the second mapping table through the target matching pair to obtain the short-term time interval corresponding to the target matching pair; determine the initial available date based on the short-term time interval and the current date, and determine the initial available date as the negotiated available date, or adjust the initial available date to obtain the negotiated available date; wherein the target matching pair includes the material to be purchased and the target supplier, and the second mapping table includes the correspondence between the matching pair and the short-term time interval; wherein the maintenance process of the second mapping table includes: based on historical purchase orders, obtaining a sample matching pair and a purchase day sequence corresponding to the sample matching pair, the sample matching pair includes a sample material and a sample supplier, the purchase day sequence includes a plurality of total purchase days, and the total purchase days represent the difference between the actual warehousing date of the sample material and the purchase application date of the sample material;

[0245] Clustering the plurality of total purchase days to obtain at least one cluster; selecting a candidate cluster from all the clusters, wherein the number of total purchase days in the candidate cluster is greater than a threshold;

[0246] Calculate the average value of all total purchase days in each candidate cluster, select the minimum average value from the average values ​​corresponding to all candidate clusters, and determine the short-term time interval based on the minimum average value;

[0247] The corresponding relationship between the sample matching pair and the short-term time interval is recorded in the second mapping table.

[0248] Exemplarily, the target purchase order also includes a negotiated available quantity, which is the delivery quantity negotiated between the demander and the supplier. The acquisition module 51 is also used to obtain a second reference feature based on the material to be purchased and the target supplier in the process of acquiring the negotiated available quantity, and the second reference feature includes a first characterization vector and a second characterization vector; wherein, based on the constructed bipartite graph of materials and suppliers, a graph representation is used to determine the first characterization vector of the material node corresponding to the material to be purchased, and the second characterization vector of the supplier node corresponding to the target supplier is determined; the second reference feature is input into the trained second prediction model so that the second prediction model outputs an initial available quantity based on the second reference feature; the initial available quantity is determined as the negotiated available quantity, or the initial available quantity is adjusted to obtain the negotiated available quantity.

[0249] Exemplarily, the second reference feature includes at least one of the following features:

[0250] The purchase order basic characteristics include at least one of the following: a short-term time interval corresponding to the target purchase order, a configured standard time interval corresponding to the target purchase order, a ratio of the short-term time interval to the standard time interval, and a delivery mode of the material to be purchased;

[0251] Supplier inventory characteristics, the supplier inventory characteristics including at least one of the following: the target supplier's latest feedback inventory, the target supplier's latest feedback inventory after supply and demand offset, the target supplier's expected delivery quantity in the future, and the number of days between the target supplier's latest feedback date and the purchase application date;

[0252] Supplier supply capability characteristics, the supplier supply capability characteristics are determined based on all historical purchase orders; the supplier supply capability characteristics include at least one of the following: the average daily delivery quantity of the target supplier over a period of time, the average delivery interval of the target supplier, the minimum total procurement days of the target supplier, the maximum total procurement days of the target supplier, the average total procurement days of the target supplier, the minimum delivery quantity of the target supplier, the maximum delivery quantity of the target supplier, and the average delivery quantity of the target supplier;

[0253] a supplier short-term capability characteristic, wherein the supplier short-term capability characteristic is determined based on candidate historical purchase orders among all historical purchase orders, and the short-term time interval corresponding to the candidate historical purchase orders is less than a threshold value; the supplier short-term capability characteristic comprises at least one of the following: the average daily delivery quantity of the target supplier over a period of time, the average delivery interval of the target supplier, the minimum total purchase days of the target supplier, the maximum total purchase days of the target supplier, the average total purchase days of the target supplier, the minimum delivery quantity of the target supplier, the maximum delivery quantity of the target supplier, and the average delivery quantity of the target supplier;

[0254] Material supply and demand characteristics, the material supply and demand characteristics include at least one of the following: the demand quantity of the material to be purchased in a future period of time, and the supply quantity of the material to be purchased in a future period of time.

[0255] Based on the same application concept as the above method, an electronic device is proposed in the embodiment of the present application, see Figure 6 As shown, it includes: a processor 61 and a machine-readable storage medium 62, wherein the machine-readable storage medium 62 stores machine-executable instructions that can be executed by the processor 61; the processor 61 is used to execute the machine-executable instructions to implement the purchase order generation method disclosed in the above example of this application.

[0256] Based on the same application concept as the above method, an embodiment of the present application also provides a machine-readable storage medium, on which a number of computer instructions are stored. When the computer instructions are executed by a processor, the purchase order generation method disclosed in the above example of the present application can be implemented.

[0257] The above-mentioned machine-readable storage medium may be any electronic, magnetic, optical or other physical storage device, which may contain or store information, such as executable instructions, data, etc. For example, the machine-readable storage medium may be: RAM (Radom Access Memory), volatile memory, non-volatile memory, flash memory, storage drive (such as hard disk drive), solid state drive, any type of storage disk (such as optical disk, DVD, etc.), or similar storage medium, or a combination thereof.

[0258] Based on the same application concept as the above method, an embodiment of the present application also provides a computer program product, which may include a computer program, and when the computer program is executed by a processor, it implements the purchase order generation method disclosed in the above example of the present application.

[0259] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the embodiments of the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0260] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for generating a purchase order, characterized in that: The method comprises: If the material to be purchased is a material with non-infinite supply, the delivery mode of the material to be purchased is obtained, and the delivery mode is a delivery date mode or a wait notification mode; wherein the delivery date mode means that the material is delivered according to the negotiated available date, and the negotiated available date is the delivery date negotiated between the demander and the supplier; the wait notification mode means that the material is delivered according to the delivery notification date, and the delivery notification date is the date when the demander sends the delivery notification to the supplier, and the delivery notification date is earlier than the negotiated available date; If the delivery mode is the delivery date mode, a first reference feature is obtained based on the material to be purchased and the target supplier, and the forecast lead time is determined based on the first reference feature; the forecast available date is determined based on the forecast lead time and the negotiated available date; if the delivery mode is the wait notification mode, the logistics days are obtained based on the material to be purchased and the target supplier, and the logistics days are the days required for the material to be delivered from the target supplier to the demander; the forecast available date is determined based on the delivery notification date and the logistics days; wherein the forecast available date is the predicted delivery date; Generate a target purchase order for the material to be purchased, wherein the target purchase order includes the material to be purchased, the target supplier, the predicted availability date, and the negotiated availability date; Wherein, a first reference feature is obtained based on the material to be purchased and the target supplier, and the forecast lead time days are determined based on the first reference feature, including: constructing a bipartite graph of materials and suppliers, the bipartite graph includes all material nodes and all supplier nodes, and the edge weight of the bipartite graph is the number of deliveries of the material by the supplier in the historical time; querying the material node corresponding to the material to be purchased and the supplier node corresponding to the target supplier from the bipartite graph; based on the bipartite graph, using graph representation to determine the first representation vector of the material node and the second representation vector of the supplier node; wherein, the closer the connection between the material to be purchased and the target supplier, the more similar the first representation vector and the second representation vector are; obtaining a first reference feature, the first reference feature includes the first representation vector and the second representation vector; inputting the first reference feature into a trained first prediction model, and the first prediction model outputs the forecast lead time days based on the first reference feature.

2. The method according to claim 1, characterized in that: The first reference feature also includes at least one of the following features: The basic features of the purchase order include at least one of the following: the number of days between the generation date of the target purchase order and the current date, the number of days between the demander's expected delivery date and the current date, the number of days between the supplier's expected delivery date and the current date, and the purchase quantity of the material to be purchased; Historical delivery characteristics, the historical delivery characteristics comprising a first statistical characteristic of a target matching pair, the target matching pair comprising the material to be purchased and the target supplier, the first statistical characteristic being obtained by counting the number of days in advance and / or the total number of days it takes to purchase a historical purchase order of the target matching pair; The supplier's historical delivery characteristics include the second statistical characteristics of the target supplier, which is obtained by counting the number of lead times and / or the total number of days required for purchase of the target supplier's historical purchase orders; the material's historical delivery characteristics include the third statistical characteristics of the material to be purchased, which is obtained by counting the number of lead times and / or the total number of days required for purchase of the material to be purchased.

3. The method according to claim 1, characterized in that The obtaining of logistics days based on the material to be purchased and the target supplier includes: querying a first mapping table through a target matching pair to obtain logistics days corresponding to the target matching pair; wherein the target matching pair includes the material to be purchased and the target supplier, and the first mapping table includes a corresponding relationship between matching pairs and logistics days; The maintenance process of the first mapping table includes: obtaining a sample matching pair and a historical logistics time series corresponding to the sample matching pair based on a historical purchase order, wherein the sample matching pair includes a sample material and a sample supplier, and the historical logistics time series includes a plurality of historical logistics days, and each historical logistics day indicates the number of days required for the sample material to be delivered from the sample supplier to the demander; Predicting the next logistics day of the sample matching pair based on the multiple historical logistics days; The correspondence between the sample matching pairs and the next logistics days is recorded in the first mapping table.

4. The method according to claim 1, characterized in that: The method further includes: if the material to be purchased is an unlimited supply material, determining the predicted availability date based on the current date; The process of determining unlimited supply materials and non-infinite supply materials includes: For each material, statistics are collected on the incoming data of the material in the historical time, and the incoming data includes a first ratio value and / or a second ratio value; wherein the first ratio value is the ratio value of the number of days the material has been in storage to the historical time, and the second ratio value is the ratio value of the candidate quantity to the historical time; for each day in the historical time, if the actual incoming quantity of the material in the K days after the day is greater than the expected incoming quantity of the material, the candidate quantity is increased by 1, otherwise, the candidate quantity is kept unchanged; Based on the incoming data of each material, an abnormal data analysis method is used to analyze all materials to obtain an abnormal data candidate set, wherein the abnormal data candidate set includes multiple abnormal materials; For each abnormal material in the abnormal data candidate set, if the storage data of the abnormal material meets the preset condition, the abnormal material is determined as an unlimited supply material; wherein, if the storage data includes a first ratio value and a second ratio value, when the first ratio value is greater than a first threshold value, and the second ratio value is greater than a second threshold value, the storage data of the abnormal material meets the preset condition; The remaining materials among all materials except the infinite supply materials are determined as non-infinite supply materials.

5. The method according to any one of claims 1 to 4, characterized in that: The process of acquiring the negotiated available date includes: querying a second mapping table through a target matching pair to obtain a short-term time interval corresponding to the target matching pair; determining an initial available date based on the short-term time interval and a current date, and determining the initial available date as the negotiated available date, or adjusting the initial available date to obtain the negotiated available date; wherein the target matching pair includes the material to be purchased and the target supplier, and the second mapping table includes a corresponding relationship between the matching pair and the short-term time interval; The maintenance process of the second mapping table includes: obtaining a sample matching pair and a purchase day sequence corresponding to the sample matching pair based on historical purchase orders, wherein the sample matching pair includes a sample material and a sample supplier, and the purchase day sequence includes a plurality of total purchase days, and the total purchase days represent the difference between the actual storage date of the sample material and the purchase application date of the sample material; Clustering the plurality of total purchase days to obtain at least one cluster; selecting a candidate cluster from all the clusters, wherein the number of total purchase days in the candidate cluster is greater than a threshold; Calculate the average value of all total purchase days in each candidate cluster, select the minimum average value from the average values ​​corresponding to all candidate clusters, and determine the short-term time interval based on the minimum average value; The corresponding relationship between the sample matching pair and the short-term time interval is recorded in the second mapping table.

6. The method according to any one of claims 1 to 4, characterized in that: The target purchase order also includes a negotiated available quantity, which is the delivery quantity negotiated between the demander and the supplier. The process of obtaining the negotiated available quantity includes: A second reference feature is obtained based on the material to be purchased and the target supplier, the second reference feature including a first characterization vector and a second characterization vector; wherein, based on the constructed bipartite graph of materials and suppliers, a first characterization vector of a material node corresponding to the material to be purchased is determined by using graph characterization, and a second characterization vector of a supplier node corresponding to the target supplier is determined; The second reference feature is input into a trained second prediction model so that the second prediction model outputs an initial available quantity based on the second reference feature; the initial available quantity is determined as the negotiated available quantity, or the initial available quantity is adjusted to obtain the negotiated available quantity.

7. The method according to claim 6, characterized in that The second reference feature includes at least one of the following features: The purchase order basic characteristics include at least one of the following: a short-term time interval corresponding to the target purchase order, a configured standard time interval corresponding to the target purchase order, a ratio of the short-term time interval to the standard time interval, and a delivery mode of the material to be purchased; Supplier inventory characteristics, the supplier inventory characteristics including at least one of the following: the target supplier's latest feedback inventory, the target supplier's latest feedback inventory after supply and demand offset, the target supplier's expected delivery quantity in the future, and the number of days between the target supplier's latest feedback date and the purchase application date; Supplier supply capability characteristics, the supplier supply capability characteristics are determined based on all historical purchase orders; the supplier supply capability characteristics include at least one of the following: the average daily delivery quantity of the target supplier over a period of time, the average delivery interval of the target supplier, the minimum total procurement days of the target supplier, the maximum total procurement days of the target supplier, the average total procurement days of the target supplier, the minimum delivery quantity of the target supplier, the maximum delivery quantity of the target supplier, and the average delivery quantity of the target supplier; a supplier short-term capability characteristic, wherein the supplier short-term capability characteristic is determined based on candidate historical purchase orders among all historical purchase orders, and the short-term time interval corresponding to the candidate historical purchase orders is less than a threshold value; the supplier short-term capability characteristic comprises at least one of the following: the average daily delivery quantity of the target supplier over a period of time, the average delivery interval of the target supplier, the minimum total purchase days of the target supplier, the maximum total purchase days of the target supplier, the average total purchase days of the target supplier, the minimum delivery quantity of the target supplier, the maximum delivery quantity of the target supplier, and the average delivery quantity of the target supplier; Material supply and demand characteristics, the material supply and demand characteristics include at least one of the following: the demand quantity of the material to be purchased in a future period of time, and the supply quantity of the material to be purchased in a future period of time.

8. A purchase order generating device, characterized in that: The device comprises: an acquisition module, for acquiring a delivery mode of the material to be purchased if the material to be purchased is a non-infinite supply material, wherein the delivery mode is a delivery date mode or a wait notification mode; wherein the delivery date mode indicates that the material is delivered according to a negotiated availability date, which is a delivery date negotiated between the demander and the supplier; and the wait notification mode indicates that the material is delivered according to a delivery notification date, which is a date when the demander sends a delivery notification to the supplier, and the delivery notification date is earlier than the negotiated availability date; A determination module, for, if the delivery mode is the delivery date mode, obtaining a first reference feature based on the material to be purchased and the target supplier, determining the forecast lead time based on the first reference feature; determining the forecast availability date based on the forecast lead time and the negotiated availability date; if the delivery mode is the wait notification mode, obtaining logistics days based on the material to be purchased and the target supplier, the logistics days being the days required for the material to be delivered from the target supplier to the demander; determining the forecast availability date based on the delivery notification date and the logistics days; wherein the forecast availability date is the forecasted delivery date; A generating module, used for generating a target purchase order for the material to be purchased, wherein the target purchase order includes the material to be purchased, the target supplier, the predicted availability date and the negotiated availability date; Among them, the determination module obtains a first reference feature based on the material to be purchased and the target supplier, and is specifically used to determine the forecast lead time based on the first reference feature: construct a bipartite graph of materials and suppliers, the bipartite graph includes all material nodes and all supplier nodes, and the edge weight of the bipartite graph is the number of deliveries of the material by the supplier in the historical time; query the material node corresponding to the material to be purchased and the supplier node corresponding to the target supplier from the bipartite graph; based on the bipartite graph, use graph representation to determine the first representation vector of the material node and the second representation vector of the supplier node; wherein, the closer the connection between the material to be purchased and the target supplier, the more similar the first representation vector and the second representation vector; obtain a first reference feature, the first reference feature includes the first representation vector and the second representation vector; input the first reference feature into a trained first prediction model, and the first prediction model outputs the forecast lead time based on the first reference feature.

9. An electronic device, characterized in that: include: a processor and a machine-readable storage medium storing machine-executable instructions executable by the processor; The processor is used to execute machine executable instructions to implement the method described in any one of claims 1-7.

Images