Order processing method and system applied to supply chain, and storage medium
By introducing order intelligent simulation units and graph neural network models into the order processing system, the problem of poor order acceptance prediction accuracy in the existing technology is solved, and the on-time delivery of large orders and the improvement of system performance is achieved.
Patent Information
- Application Number
- CN202411488162.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-24
- Publication Date
- 2025-03-21
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The lack of accurate methods in the prior art to predict whether to accept orders when orders arrive, resulting in poor prediction accuracy, especially in the case of large orders, which may not be able to deliver goods in time, affecting customer experience.
An order processing system applied to the supply chain is designed, using order intelligent simulation units and smart order simulation models, and constructed based on graph neural network to predict the delivery cycle of goods and decide whether to accept orders or negotiate.
By accurately predicting the delivery cycle, blind order acceptance is avoided, and on-time delivery of large orders is ensured, customer experience is improved, and the computing speed and simulation performance of the order processing system are improved.
Smart Images
Figure CN119671661A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of e-commerce and artificial intelligence technology, and specifically to an order processing method, system and storage medium applied to a supply chain. Background Art
[0002] The order management system in the existing technology generally tends to focus on the generation management, delivery management, and traceability processing of goods after receiving an order. These are relatively mature technologies, but there is no accurate method in the existing technology to predict whether to accept an order when it arrives, or it only uses experience or simple parameters such as inventory and generation cycle to infer whether to accept an order, resulting in poor prediction accuracy. After accepting an order, especially a large order, the goods may not be delivered in time, affecting the customer experience.
[0003] Therefore, how to accurately determine whether to accept an order is a technical challenge. Summary of the invention
[0004] The present invention proposes the following technical solutions to address one or more technical deficiencies in the above-mentioned prior art.
[0005] An order processing system applied to a supply chain, the system comprising:
[0006] A client receives an order submitted by a customer, encrypts it, generates order data, and sends the order data to an order server. The order data includes at least: the total quantity of goods, the type of goods, the quantity of each type of goods, and the required delivery period of each type of goods;
[0007] The order server has a receiving unit, a pre-processing unit, a common order processing unit, an order intelligent simulation unit and a sending unit;
[0008] The receiving unit receives the order data from the client, and decrypts the order data and sends it to the pre-processing unit;
[0009] The preprocessing unit obtains the total quantity of goods, the type of goods, the quantity of each type of goods and the required delivery period of each type of goods from the order data, and determines whether the total quantity is greater than a first threshold value, if not, the order data is sent to the general order processing unit, if yes, the order data is sent to the order intelligent simulation unit;
[0010] The general order processing unit splits the order into multiple sub-orders based on the type of goods, the quantity of each type of goods and the warehouse where the goods are stored, and sends the multiple sub-orders to the warehouse where the goods are stored. The warehouse performs shipment processing based on the received sub-orders, generates shipment information after all sub-orders are shipped, and sends the shipment information to the client through the sending unit;
[0011] An intelligent order simulation model is run on the order intelligent simulation unit. The order intelligent simulation unit inputs the type of goods in the order data and the quantity of each type of goods into the intelligent order simulation model for processing, and then outputs the predicted delivery period of each product. The predicted delivery period of each product is compared with the required delivery period of each product. If all goods can meet the required delivery period, order acceptance information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is greater than or equal to a second threshold, negotiation information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is less than the second threshold, order rejection information is generated and sent to the client through the sending unit.
[0012] Furthermore, the magnitudes of the first threshold and the second threshold are determined based on the type and production cycle of the goods.
[0013] Furthermore, the shipping information includes at least the order number, sub-order number, shipping warehouse number, product type and quantity, and shipping date.
[0014] Furthermore, the smart order simulation model is constructed based on a graph neural network, and the composition of the graph used for the graph neural network is as follows: N warehouses where the goods are stored serve as N warehouse nodes, M production workshops serve as M workshop nodes, and K raw material suppliers serve as K raw material nodes. The N warehouse nodes, M workshop nodes, and K raw material nodes constitute the nodes of the graph. If the goods generated by a production workshop are transmitted to the corresponding warehouse, there is an edge between the corresponding workshop node and the corresponding warehouse node, otherwise, there is no edge between the two; if the raw materials of a raw material supplier are supplied to the corresponding generation workshop, there is an edge between the corresponding raw material node and the corresponding workshop node, otherwise, there is no edge between the two, wherein N≥2, M≥2, and K≥2.
[0015] Furthermore, the characteristic values of the N warehouse nodes are:
[0016]
[0017] The characteristic values of the M workshop nodes are:
[0018]
[0019] The characteristic values of the K raw material nodes are:
[0020]
[0021] Among them, store i Represents the characteristic value of the i-th warehouse node, storeres i represents the inventory of the i-th warehouse node, storespeed i represents the shipping speed of the i-th warehouse node, workshop j represents the characteristic value of the jth workshop node, workshopproc j represents the number of processes that can be processed in the jth workshop node, workshopprodu j represents the production speed of the jth workshop node, matsupply p represents the eigenvalue of the pth raw material node, matsupplypreT p Represents the supply speed of the pth raw material node, N≥i≥1, M≥j≥1K≥p≥1.
[0022] Furthermore, the weight of the edge between the warehouse node and the workshop node of the graph neural network is:
[0023]
[0024] The weight of the edge between the raw material node and the workshop node of the graph neural network is:
[0025]
[0026] Among them, workshopmtcust j Indicates the raw material consumption rate corresponding to the j-th workshop node.
[0027] The present invention also proposes an order processing method based on the above order processing system applied to the supply chain, characterized in that the method comprises:
[0028] In the order sending step, the client receives the order submitted by the customer, encrypts it, generates order data, and sends the order data to the order server. The order data includes at least: the total quantity of goods, the type of goods, the quantity of each type of goods, and the required delivery period of each type of goods;
[0029] An order receiving step, in which a receiving unit receives order data from the client, and decrypts the order data and sends it to the preprocessing unit;
[0030] a preprocessing step, wherein the preprocessing unit obtains the total quantity of goods, the types of goods, the quantity of each type of goods and the required delivery period of each type of goods from the order data, and determines whether the total quantity is greater than a first threshold value; if not, the order data is sent to a common order processing unit; if yes, the order data is sent to an order intelligent simulation unit;
[0031] In a common order processing step, a common order processing unit divides the order into multiple sub-orders based on the type of goods, the quantity of each type of goods and the warehouse where the goods are stored, and sends the multiple sub-orders to the warehouse where the goods are stored. The warehouse performs shipment processing based on the received sub-orders, generates shipment information after all sub-orders are shipped, and sends the shipment information to the client through the sending unit;
[0032] In the intelligent simulation step, an intelligent order simulation model is run on the order intelligent simulation unit. The order intelligent simulation unit inputs the type of goods in the order data and the quantity of each type of goods into the intelligent order simulation model for processing, and then outputs the predicted delivery period of each product. The predicted delivery period of each product is compared with the required delivery period of each product. If all goods can meet the required delivery period, order acceptance information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is greater than or equal to a second threshold, negotiation information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is less than the second threshold, order rejection information is generated and sent to the client through the sending unit.
[0033] Furthermore, the magnitudes of the first threshold and the second threshold are determined based on the type and production cycle of the goods.
[0034] Furthermore, the shipping information includes at least the order number, sub-order number, shipping warehouse number, product type and quantity, and shipping date.
[0035] Furthermore, the smart order simulation model is constructed based on a graph neural network, and the composition of the graph used for the graph neural network is as follows: N warehouses where the goods are stored serve as N warehouse nodes, M production workshops serve as M workshop nodes, and K raw material suppliers serve as K raw material nodes. The N warehouse nodes, M workshop nodes, and K raw material nodes constitute the nodes of the graph. If the goods generated by a production workshop are transmitted to the corresponding warehouse, there is an edge between the corresponding workshop node and the corresponding warehouse node, otherwise, there is no edge between the two; if the raw materials of a raw material supplier are supplied to the corresponding generation workshop, there is an edge between the corresponding raw material node and the corresponding workshop node, otherwise, there is no edge between the two, wherein N≥2, M≥2, and K≥2.
[0036] Furthermore, the characteristic values of the N warehouse nodes are:
[0037]
[0038] The characteristic values of the M workshop nodes are:
[0039]
[0040] The characteristic values of the K raw material nodes are:
[0041]
[0042] Among them, store i Represents the characteristic value of the i-th warehouse node, storeres i represents the inventory of the i-th warehouse node, storespeed i represents the shipping speed of the i-th warehouse node, workshop j represents the characteristic value of the jth workshop node, workshopproc j represents the number of processes that can be processed in the jth workshop node, workshopprodu j represents the production speed of the jth workshop node, matsupply p represents the eigenvalue of the pth raw material node, matsupplypreT p Represents the supply speed of the pth raw material node, N≥i≥1, M≥j≥1K≥p≥1.
[0043] Furthermore, the weight of the edge between the warehouse node and the workshop node of the graph neural network is:
[0044]
[0045] The weight of the edge between the raw material node and the workshop node of the graph neural network is:
[0046]
[0047] Among them, workshopmtcust j Indicates the raw material consumption rate corresponding to the j-th workshop node.
[0048] The present invention further provides a computer-readable storage medium, on which a computer program code is stored. When the computer program code is executed by a computer, any of the above methods is executed.
[0049] The technical effect of the present invention is: an order processing method, system and storage medium applied to a supply chain of the present invention, the system comprises a client and an order server, the client receives an order submitted by a customer and generates order data after encryption, and sends the order data to the order server, the order data at least comprises: the total quantity of goods, the type of goods, the quantity of each type of goods and the required delivery period of each type of goods; the order server comprises a receiving unit, a preprocessing unit, a general order processing unit, an order intelligent simulation unit and a sending unit; the receiving unit receives the order data from the client, and decrypts the order data and sends it to the preprocessing unit; the preprocessing unit obtains the total quantity of goods, the type of goods, the quantity of each type of goods and the required delivery period of each type of goods from the order data, and determines whether the total quantity is greater than a first threshold value, if not, the order data is sent to the general order processing unit, if yes, the order data is sent to the order intelligent simulation unit; the general order processing unit determines the total quantity of goods, the type of goods, the quantity of each type of goods and the required delivery period of each type of goods based on the type of goods , the quantity of each type of goods and the warehouse where the goods are stored, split the order into multiple sub-orders, and send the multiple sub-orders to the warehouse where the goods are stored, the warehouse performs shipment processing based on the received sub-orders, generates shipment information after all sub-orders are shipped, and sends the shipment information to the client through the sending unit; an intelligent order simulation model runs on the order intelligent simulation unit, the order intelligent simulation unit inputs the type of goods and the quantity of each type of goods in the order data into the intelligent order simulation model for processing, and then outputs the predicted delivery cycle of each product, compares the predicted delivery cycle of each product with the required delivery cycle of each product, if all products can meet the required delivery cycle, then generates order acceptance information and sends it to the client through the sending unit, if the number of goods that meet the required delivery cycle is greater than or equal to the second threshold, then generates negotiation information and sends it to the client through the sending unit, if the number of goods that meet the required delivery cycle is less than the second threshold, then generates order rejection information and sends it to the client through the sending unit.The present invention creatively divides orders into ordinary orders and large orders based on the number of orders in the order. Ordinary orders can be ordinary retail orders, and large orders can be orders with relatively large order quantities. Different types of orders are processed differently. Ordinary orders enter the shipping warehouse for shipping processing, while large orders use the smart order simulation model running on the order intelligent simulation unit to predict the predicted delivery cycle of each product based on the type of goods and the quantity of each type of goods in the order data, and then compare them with the required delivery cycle of each product. If all goods can meet the required delivery cycle, order acceptance information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery cycle is greater than or equal to the second threshold, negotiation information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery cycle is less than the second threshold, order rejection information is generated and sent to the client through the sending unit. The present invention also creatively adds a negotiation mechanism, that is, when a part of the ordered goods cannot meet the delivery cycle, a negotiation is conducted with the customer. Negotiate and try your best to promote the order, thereby ensuring that the delivery cycle is accurately predicted based on artificial intelligence technology when accepting large orders, thereby avoiding blind acceptance of orders and ensuring the timely delivery of large orders; in the present invention, a method of constructing the graph used by the graph neural network is creatively proposed, that is, the nodes are divided into three categories: warehouse nodes, workshop nodes, and raw material nodes. The processes that can be processed in the workshop are reflected in the characteristic values of the workshop nodes, avoiding each process as a node resulting in a large number of nodes in the constructed graph, which makes the graph neural network's calculation speed slow and affects the system performance. When constructing edges, a three-layer structure of raw material node->workshop node->warehouse node is formed. There is an edge relationship between the nodes in the adjacent two layers, and there is no edge relationship between the raw material node and the warehouse node. This construction method not only conforms to the actual application scenario, but also improves the calculation speed of the graph neural network and improves the simulation performance. The specific calculation method of the node characteristic value and edge weight proposed in the present invention has good effect after actual application, can avoid blind acceptance of orders, and improve the timeliness of order delivery. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] Other features, objects and advantages of the present application will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings.
[0051] Figure 1 4 is a structural diagram of an order processing system applied to a supply chain according to an embodiment of the present invention.
[0052] Figure 2 The present invention is a flowchart of an order processing method applied to a supply chain according to an embodiment of the present invention. DETAILED DESCRIPTION
[0053] The present application will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It should also be noted that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.
[0054] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0055] Figure 1 An order processing system applied to a supply chain of the present invention is shown, the system comprising:
[0056] A client receives an order submitted by a customer, encrypts it, generates order data, and sends the order data to an order server. The order data includes at least: the total quantity of goods, the type of goods, the quantity of each type of goods, and the required delivery period of each type of goods;
[0057] The order server has a receiving unit, a pre-processing unit, a common order processing unit, an order intelligent simulation unit and a sending unit;
[0058] The receiving unit receives the order data from the client, and decrypts the order data and sends it to the pre-processing unit;
[0059] The preprocessing unit obtains the total quantity of goods, the type of goods, the quantity of each type of goods and the required delivery cycle of each type of goods from the order data, and determines whether the total quantity is greater than a first threshold value. If not, the order data is sent to the general order processing unit; if so, the order data is sent to the order intelligent simulation unit; that is, when the total order quantity is less than or equal to the first threshold value, the order is considered to be an ordinary order, also known as a retail order, and this type of order is directly sent to the general order processing unit for processing; if the total order quantity exceeds the first threshold value, the order is a large order, and the order intelligent simulation unit needs to be used to accurately predict the delivery cycle to determine whether the order can be accepted.
[0060] The general order processing unit splits the order into multiple sub-orders based on the type of goods, the quantity of each type of goods and the warehouse where the goods are stored, and sends the multiple sub-orders to the warehouse where the goods are stored. The warehouse performs shipment processing based on the received sub-orders, generates shipment information after all sub-orders are shipped, and sends the shipment information to the client through the sending unit;
[0061] An intelligent order simulation model is run on the order intelligent simulation unit. The order intelligent simulation unit inputs the type of goods in the order data and the quantity of each type of goods into the intelligent order simulation model for processing, and then outputs the predicted delivery period of each product. The predicted delivery period of each product is compared with the required delivery period of each product. If all goods can meet the required delivery period, order acceptance information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is greater than or equal to a second threshold, negotiation information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is less than the second threshold, order rejection information is generated and sent to the client through the sending unit.
[0062] The present invention creatively divides orders into ordinary orders and large orders based on the quantity of orders in the orders. Ordinary orders can be ordinary retail orders, and large orders can be orders with relatively large order quantities. Different types of orders are processed differently. Ordinary orders enter the shipping warehouse for shipping processing, while large orders use the smart order simulation model running on the order intelligent simulation unit to predict the predicted delivery cycle of each product based on the type of goods and the quantity of each type of goods in the order data, and then compare them with the required delivery cycle of each product. If all goods can meet the required delivery cycle, order information is generated and sent to the sending unit through the sending unit. The client, if the number of goods that meet the required delivery cycle is greater than or equal to the second threshold, then a negotiation message is generated and sent to the client through a sending unit; if the number of goods that meet the required delivery cycle is less than the second threshold, then a rejection order message is generated and sent to the client through a sending unit. The present invention also creatively adds a negotiation mechanism, that is, when a part of the ordered goods cannot meet the delivery cycle, negotiate with the customer and try to facilitate the order, thereby ensuring that the delivery cycle is accurately predicted based on artificial intelligence technology when accepting large orders, thereby avoiding blind order acceptance and ensuring the timely delivery of the large order. This is an important inventive concept of the present invention.
[0063] In one embodiment, the first threshold and the second threshold are determined based on the type and production cycle of the goods. For example, for luggage and bags, due to their popularity, the inventory of the goods in the warehouse cannot be too much, so the first threshold can be set to 100-500, and the second threshold can be set to 5-10. Of course, for other types of goods, those skilled in the art can set them according to historical generation and sales data, or estimate the first and second thresholds based on LSTM.
[0064] In one embodiment, the shipping information includes at least the order number, sub-order number, shipping warehouse number, product type and quantity, and shipping date. The shipping information can be sent in the form of a list, or in other data structure forms.
[0065] In one embodiment, the smart order simulation model is built based on a graph neural network. The graph neural network (GNN) captures the complex relationships in the graph data through the interaction between nodes, and is widely used in node classification, link prediction and graph representation learning. GNN is derived from CNN, which overcomes the difficulty of its application in non-Euclidean space, and solves the problems of computational complexity and generalization ability of graph embedding technology. GNN is invariant, can process non-sequential data, and has reasoning ability. However, the current GNN faces the challenges of shallow structure, dynamic graph processing and unstructured data. The difficulty in building a graph neural network lies in determining the eigenvalues of each node in the graph and the weight values of the edges between the nodes. Only when the eigenvalues of the nodes and the weight values of the edges establish a corresponding relationship with the technical problems to be solved, the prediction results of the graph neural network will be accurate.
[0066] The graph used for the graph neural network in the present invention is composed as follows: N warehouses where the goods are stored serve as N warehouse nodes, M production workshops serve as M workshop nodes, and K raw material suppliers serve as K raw material nodes. The N warehouse nodes, M workshop nodes, and K raw material nodes constitute the nodes of the graph. If the goods generated by a production workshop are transmitted to the corresponding warehouse, there is an edge between the corresponding workshop node and the corresponding warehouse node, otherwise, there is no edge between the two; if the raw materials of a raw material supplier are supplied to the corresponding generation workshop, there is an edge between the corresponding raw material node and the corresponding workshop node, otherwise, there is no edge between the two, wherein N≥2, M≥2, and K≥2.
[0067] In the present invention, a method for constructing a graph used by a graph neural network is creatively proposed, that is, nodes are divided into three categories: warehouse nodes, workshop nodes, and raw material nodes. The processes that can be processed in the workshop are reflected in the characteristic values of the workshop nodes, avoiding the situation where each process is used as a node, resulting in a large number of nodes in the constructed graph, thereby making the operation speed of the graph neural network slower and affecting the system performance. When constructing edges, a three-layer structure of raw material node->workshop node->warehouse node is formed. There is an edge relationship between the nodes in two adjacent layers, and there is no edge relationship between the raw material node and the warehouse node. This construction method not only conforms to the actual application scenario, but also improves the operation speed of the graph neural network and the simulation performance. This is another important invention point of the present invention.
[0068] In one embodiment, the characteristic values of the N warehouse nodes are:
[0069]
[0070] The characteristic values of the M workshop nodes are:
[0071]
[0072] The characteristic values of the K raw material nodes are:
[0073]
[0074] Among them, store i Represents the characteristic value of the i-th warehouse node, storeres i represents the inventory of the i-th warehouse node, storespeed i represents the shipping speed of the i-th warehouse node, workshop j represents the characteristic value of the jth workshop node, workshopproc j represents the number of processes that can be processed in the jth workshop node, workshopprodu j represents the production speed of the jth workshop node, matsupply p represents the eigenvalue of the pth raw material node, matsupplypreT p Represents the supply speed of the pth raw material node, N≥i≥1, M≥j≥1K≥p≥1.
[0075] In the present invention, the characteristic value of each node is designed based on the inventory and delivery speed of the warehouse node, the number of processable processes and production speed of the workshop node, and the supply speed of the raw material node. In particular, the number of processable processes of the workshop node is set in the characteristic value of the workshop node for consideration, avoiding the problem that each process is used as a node, resulting in a large number of nodes in the constructed graph and a slow running speed of the graph neural network. The specific calculation method of these characteristic values and the specific calculation method of the edge weights described below have been actually applied in the supply chain of a luggage company, and the effect is good, which can avoid blind order acceptance and improve the timeliness of order delivery. This is an important inventive point of the present invention.
[0076] In one embodiment, the weight of the edge between the warehouse node and the workshop node of the graph neural network is:
[0077]
[0078] The weight of the edge between the raw material node and the workshop node of the graph neural network is:
[0079]
[0080] Among them, workshopmtcust j Indicates the raw material consumption rate corresponding to the j-th workshop node.
[0081] In the present invention, a three-layer structure of raw material node -> workshop node -> warehouse node is constructed. There is an edge relationship between the nodes of two adjacent layers, but there is no edge relationship between the raw material node and the warehouse node. Based on the shipment speed of goods, the production speed of goods, and the supply speed of raw materials, a specific calculation method for the weight of each edge is proposed. Together with the specific calculation method of the above-mentioned node characteristic value, it is actually applied in the supply chain of a luggage company with good results, which can avoid blind order acceptance and improve the timeliness of order delivery. This is an important invention point of the present invention.
[0082] Figure 2 The present invention shows an order processing method applied to a supply chain of the present invention, which is based on Figure 1 System implementation, including:
[0083] In the order sending step S201, the client receives the order submitted by the customer, encrypts it, generates order data, and sends the order data to the order server. The order data includes at least: the total quantity of goods, the type of goods, the quantity of each type of goods, and the required delivery period of each type of goods;
[0084] Order receiving step S202, the receiving unit receives the order data from the client, and decrypts the order data and sends it to the pre-processing unit;
[0085] Preprocessing step S203, the preprocessing unit obtains the total quantity of goods, the type of goods, the quantity of each type of goods and the required delivery cycle of each type of goods from the order data, and determines whether the total quantity is greater than a first threshold value. If not, the order data is sent to the general order processing unit; if yes, the order data is sent to the order intelligent simulation unit; that is, when the total order quantity is less than or equal to the first threshold value, the order is considered to be an ordinary order, also known as a retail order, and this type of order is directly sent to the general order processing unit for processing; if the total order quantity exceeds the first threshold value, the order is a large order, and the order intelligent simulation unit needs to be used to accurately predict the delivery cycle to determine whether the order can be accepted.
[0086] In the common order processing step S204, the common order processing unit splits the order into multiple sub-orders based on the type of goods, the quantity of each type of goods and the warehouse where the goods are stored, and sends the multiple sub-orders to the warehouse where the goods are stored. The warehouse performs shipment processing based on the received sub-orders, generates shipment information after all sub-orders are shipped, and sends the shipment information to the client through the sending unit;
[0087] Intelligent simulation step S205, an intelligent order simulation model is run on the order intelligent simulation unit, the order intelligent simulation unit inputs the type of goods in the order data and the quantity of each type of goods into the intelligent order simulation model for processing, and then outputs the predicted delivery period of each product, and compares the predicted delivery period of each product with the required delivery period of each product. If all goods can meet the required delivery period, then order acceptance information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is greater than or equal to a second threshold, then negotiation information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is less than the second threshold, then order rejection information is generated and sent to the client through the sending unit.
[0088] The present invention creatively divides orders into ordinary orders and large orders based on the quantity of orders in the orders. Ordinary orders can be ordinary retail orders, and large orders can be orders with relatively large order quantities. Different types of orders are processed differently. Ordinary orders enter the shipping warehouse for shipping processing, while large orders use the smart order simulation model running on the order intelligent simulation unit to predict the predicted delivery cycle of each product based on the type of goods and the quantity of each type of goods in the order data, and then compare them with the required delivery cycle of each product. If all goods can meet the required delivery cycle, order information is generated and sent to the sending unit through the sending unit. The client, if the number of goods that meet the required delivery cycle is greater than or equal to the second threshold, then a negotiation message is generated and sent to the client through a sending unit; if the number of goods that meet the required delivery cycle is less than the second threshold, then a rejection order message is generated and sent to the client through a sending unit. The present invention also creatively adds a negotiation mechanism, that is, when a part of the ordered goods cannot meet the delivery cycle, negotiate with the customer and try to facilitate the order, thereby ensuring that the delivery cycle is accurately predicted based on artificial intelligence technology when accepting large orders, thereby avoiding blind order acceptance and ensuring the timely delivery of the large order. This is an important inventive concept of the present invention.
[0089] In one embodiment, the first threshold and the second threshold are determined based on the type and production cycle of the goods. For example, for luggage and bags, due to their popularity, the inventory of the goods in the warehouse cannot be too much, so the first threshold can be set to 100-500, and the second threshold can be set to 5-10. Of course, for other types of goods, those skilled in the art can set them according to historical generation and sales data, or estimate the first and second thresholds based on LSTM.
[0090] In one embodiment, the shipping information includes at least the order number, sub-order number, shipping warehouse number, product type and quantity, and shipping date. The shipping information can be sent in the form of a list, or in other data structure forms.
[0091] In one embodiment, the smart order simulation model is built based on a graph neural network. The graph neural network (GNN) captures the complex relationships in the graph data through the interaction between nodes, and is widely used in node classification, link prediction and graph representation learning. GNN is derived from CNN, which overcomes the difficulty of its application in non-Euclidean space, and solves the problems of computational complexity and generalization ability of graph embedding technology. GNN is invariant, can process non-sequential data, and has reasoning ability. However, the current GNN faces the challenges of shallow structure, dynamic graph processing and unstructured data. The difficulty in building a graph neural network lies in determining the eigenvalues of each node in the graph and the weight values of the edges between the nodes. Only when the eigenvalues of the nodes and the weight values of the edges establish a corresponding relationship with the technical problems to be solved, the prediction results of the graph neural network will be accurate.
[0092] The graph used for the graph neural network in the present invention is composed as follows: N warehouses where the goods are stored serve as N warehouse nodes, M production workshops serve as M workshop nodes, and K raw material suppliers serve as K raw material nodes. The N warehouse nodes, M workshop nodes, and K raw material nodes constitute the nodes of the graph. If the goods generated by a production workshop are transmitted to the corresponding warehouse, there is an edge between the corresponding workshop node and the corresponding warehouse node, otherwise, there is no edge between the two; if the raw materials of a raw material supplier are supplied to the corresponding generation workshop, there is an edge between the corresponding raw material node and the corresponding workshop node, otherwise, there is no edge between the two, wherein N≥2, M≥2, and K≥2.
[0093] In the present invention, a method for constructing a graph used by a graph neural network is creatively proposed, that is, nodes are divided into three categories: warehouse nodes, workshop nodes, and raw material nodes. The processes that can be processed in the workshop are reflected in the characteristic values of the workshop nodes, avoiding the situation where each process is used as a node, resulting in a large number of nodes in the constructed graph, thereby making the operation speed of the graph neural network slower and affecting the system performance. When constructing edges, a three-layer structure of raw material node->workshop node->warehouse node is formed. There is an edge relationship between the nodes in two adjacent layers, and there is no edge relationship between the raw material node and the warehouse node. This construction method not only conforms to the actual application scenario, but also improves the operation speed of the graph neural network and the simulation performance. This is another important invention point of the present invention.
[0094] In one embodiment, the characteristic values of the N warehouse nodes are:
[0095]
[0096] The characteristic values of the M workshop nodes are:
[0097]
[0098] The characteristic values of the K raw material nodes are:
[0099]
[0100] Among them, store i Represents the characteristic value of the i-th warehouse node, storeres i represents the inventory of the i-th warehouse node, storespeed i represents the shipping speed of the i-th warehouse node, workshop j represents the characteristic value of the jth workshop node, workshopproc j represents the number of processes that can be processed in the jth workshop node, workshopprodu j represents the production speed of the jth workshop node, matsupply p represents the eigenvalue of the pth raw material node, matsupplypreT p Represents the supply speed of the pth raw material node, N≥i≥1, M≥j≥1K≥p≥1.
[0101] In the present invention, the characteristic value of each node is designed based on the inventory and delivery speed of the warehouse node, the number of processable processes and production speed of the workshop node, and the supply speed of the raw material node. In particular, the number of processable processes of the workshop node is set in the characteristic value of the workshop node for consideration, avoiding the problem that each process is used as a node, resulting in a large number of nodes in the constructed graph and a slow running speed of the graph neural network. The specific calculation method of these characteristic values and the specific calculation method of the edge weights described below have been actually applied in the supply chain of a luggage company, and the effect is good, which can avoid blind order acceptance and improve the timeliness of order delivery. This is an important inventive point of the present invention.
[0102] In one embodiment, the weight of the edge between the warehouse node and the workshop node of the graph neural network is:
[0103]
[0104] The weight of the edge between the raw material node and the workshop node of the graph neural network is:
[0105]
[0106] Among them, workshopmtcust jIndicates the raw material consumption rate corresponding to the j-th workshop node.
[0107] In the present invention, a three-layer structure of raw material node -> workshop node -> warehouse node is constructed. There is an edge relationship between the nodes of two adjacent layers, but there is no edge relationship between the raw material node and the warehouse node. Based on the shipment speed of goods, the production speed of goods, and the supply speed of raw materials, a specific calculation method for the weight of each edge is proposed. Together with the specific calculation method of the above-mentioned node characteristic value, it is actually applied in the supply chain of a luggage company with good results, which can avoid blind order acceptance and improve the timeliness of order delivery. This is an important invention point of the present invention.
[0108] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.
[0109] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the devices described in the various embodiments of the present application or certain parts of the embodiments.
[0110] Finally, it should be noted that the above embodiments are only intended to illustrate rather than limit the technical solutions of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the present invention can still be modified or replaced by equivalents. Any modification or partial replacement that does not depart from the spirit and scope of the present invention should be included in the scope of the claims of the present invention.
Claims
1. An order processing system applied to a supply chain, characterized in that: The system includes: A client receives an order submitted by a customer, encrypts it, generates order data, and sends the order data to an order server. The order data includes at least: the total quantity of goods, the type of goods, the quantity of each type of goods, and the required delivery period of each type of goods; The order server has a receiving unit, a pre-processing unit, a common order processing unit, an order intelligent simulation unit and a sending unit; The receiving unit receives the order data from the client, and decrypts the order data and sends it to the pre-processing unit; The preprocessing unit obtains the total quantity of goods, the type of goods, the quantity of each type of goods and the required delivery period of each type of goods from the order data, and determines whether the total quantity is greater than a first threshold value, if not, the order data is sent to the general order processing unit, if yes, the order data is sent to the order intelligent simulation unit; The general order processing unit splits the order into multiple sub-orders based on the type of goods, the quantity of each type of goods and the warehouse where the goods are stored, and sends the multiple sub-orders to the warehouse where the goods are stored. The warehouse performs shipment processing based on the received sub-orders, generates shipment information after all sub-orders are shipped, and sends the shipment information to the client through the sending unit; An intelligent order simulation model is run on the order intelligent simulation unit. The order intelligent simulation unit inputs the type of goods in the order data and the quantity of each type of goods into the intelligent order simulation model for processing, and then outputs the predicted delivery period of each product. The predicted delivery period of each product is compared with the required delivery period of each product. If all goods can meet the required delivery period, order acceptance information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is greater than or equal to a second threshold, negotiation information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is less than the second threshold, order rejection information is generated and sent to the client through the sending unit.
2. The system according to claim 1, characterized in that The first threshold and the second threshold are determined based on the type and production cycle of the product.
3. The system according to claim 2, characterized in that The shipping information includes at least the order number, sub-order number, shipping warehouse number, product type and quantity, and shipping date.
4. The system according to claim 3, characterized in that The smart order simulation model is constructed based on a graph neural network. The composition of the graph used for the graph neural network is as follows: N warehouses where the goods are stored serve as N warehouse nodes, M production workshops serve as M workshop nodes, and K raw material suppliers serve as K raw material nodes. The N warehouse nodes, M workshop nodes, and K raw material nodes constitute the nodes of the graph. If the goods generated by a production workshop are transmitted to the corresponding warehouse, there is an edge between the corresponding workshop node and the corresponding warehouse node, otherwise, there is no edge between the two; if the raw materials of a raw material supplier are supplied to the corresponding generation workshop, there is an edge between the corresponding raw material node and the corresponding workshop node, otherwise, there is no edge between the two, wherein N≥2, M≥2, and K≥2.
5. The system according to claim 4, characterized in that The characteristic values of the N warehouse nodes are: The characteristic values of the M workshop nodes are: The characteristic values of the K raw material nodes are: Among them, store i Represents the characteristic value of the i-th warehouse node, storeres i represents the inventory of the i-th warehouse node, storespeed i represents the shipping speed of the i-th warehouse node, workshop j represents the characteristic value of the jth workshop node, workshopproc j represents the number of processes that can be processed in the jth workshop node, workshopprodu j represents the production speed of the jth workshop node, matsupply p represents the eigenvalue of the pth raw material node, matsupplypreT p Represents the supply speed of the pth raw material node, N≥i≥1, M≥j≥1K≥p≥1. The weight of the edge between the warehouse node and the workshop node of the graph neural network is: The weight of the edge between the raw material node and the workshop node of the graph neural network is: Among them, workshopmtcust j Indicates the raw material consumption rate corresponding to the j-th workshop node.
6. An order processing method for an order processing system applied to a supply chain based on any one of claims 1 to 5, characterized in that: The method includes: In the order sending step, the client receives the order submitted by the customer, encrypts it, generates order data, and sends the order data to the order server. The order data includes at least: the total quantity of goods, the type of goods, the quantity of each type of goods, and the required delivery period of each type of goods; An order receiving step, in which a receiving unit receives order data from the client, and decrypts the order data and sends it to the preprocessing unit; a preprocessing step, wherein the preprocessing unit obtains the total quantity of goods, the types of goods, the quantity of each type of goods and the required delivery period of each type of goods from the order data, and determines whether the total quantity is greater than a first threshold value; if not, the order data is sent to a common order processing unit; if yes, the order data is sent to an order intelligent simulation unit; In a common order processing step, a common order processing unit divides the order into multiple sub-orders based on the type of goods, the quantity of each type of goods and the warehouse where the goods are stored, and sends the multiple sub-orders to the warehouse where the goods are stored. The warehouse performs shipment processing based on the received sub-orders, generates shipment information after all sub-orders are shipped, and sends the shipment information to the client through the sending unit; In the intelligent simulation step, an intelligent order simulation model is run on the order intelligent simulation unit. The order intelligent simulation unit inputs the type of goods in the order data and the quantity of each type of goods into the intelligent order simulation model for processing, and then outputs the predicted delivery period of each product. The predicted delivery period of each product is compared with the required delivery period of each product. If all goods can meet the required delivery period, order acceptance information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is greater than or equal to a second threshold, negotiation information is generated and sent to the client through the sending unit. If the number of goods that meet the required delivery period is less than the second threshold, order rejection information is generated and sent to the client through the sending unit.
7. The method according to claim 6, characterized in that The first threshold and the second threshold are determined based on the type and production cycle of the product.
8. The method according to claim 7, characterized in that The shipping information includes at least the order number, sub-order number, shipping warehouse number, product type and quantity, and shipping date.
9. The method according to claim 8, characterized in that The smart order simulation model is constructed based on a graph neural network. The composition of the graph used for the graph neural network is as follows: N warehouses where the goods are stored serve as N warehouse nodes, M production workshops serve as M workshop nodes, and K raw material suppliers serve as K raw material nodes. The N warehouse nodes, M workshop nodes, and K raw material nodes constitute the nodes of the graph. If the goods generated by a production workshop are transmitted to the corresponding warehouse, there is an edge between the corresponding workshop node and the corresponding warehouse node, otherwise, there is no edge between the two; if the raw materials of a raw material supplier are supplied to the corresponding generation workshop, there is an edge between the corresponding raw material node and the corresponding workshop node, otherwise, there is no edge between the two, wherein N≥2, M≥2, and K≥2.
10. A computer-readable storage medium having a computer program code stored thereon, wherein when the computer program code is executed by a computer, the method of any one of claims 1 to 5 is executed.