Supply chain management method, system and device based on knowledge graph and storage medium

Abstract

The invention discloses a supply chain management method, system and device based on a knowledge graph and a storage medium, and relates to the technical field of supply chains. The supply chain management method based on the knowledge graph comprises the following steps: constructing a supply chain knowledge graph based on supply chain influence information; determining a supply chain heterogeneous graph according to the supply chain knowledge graph, wherein the supply chain heterogeneous graph comprises a supplier path for selection; and determining target supplier information according to the supplier path for selection. According to the supply chain management method based on the knowledge graph, factors influencing supplier selection can be intelligently and comprehensively considered, and the supply chain knowledge graph and the supply chain heterogeneous graph are constructed according to various factors influencing supplier selection, so that high-quality suppliers can be intelligently selected through the supply chain heterogeneous graph; therefore, the intelligence and reliability of supplier selection are improved.

Description

Technical Field

[0001] The present application relates to the field of supply chain technology, and specifically to a supply chain management method, system, device and storage medium based on a knowledge graph. Background Art

[0002] The existing process of selecting suppliers for vehicle parts relies too much on manual experience, and the factors considered in the supplier selection process are too single. It fails to comprehensively consider various influencing factors and the relationship between them. Both supplier selection and supply chain optimization face great challenges. Summary of the invention

[0003] Purpose of the invention: The embodiments of the present application provide a supply chain management method, system, device and storage medium based on a knowledge graph to improve the intelligence of vehicle supplier selection.

[0004] Technical solution: A supply chain management method based on knowledge graph described in an embodiment of the present application includes:

[0005] Build a supply chain knowledge graph based on supply chain impact information;

[0006] Determine a supply chain heterogeneous graph according to the supply chain knowledge graph, wherein the supply chain heterogeneous graph includes supplier paths for selection;

[0007] The target supplier information is determined according to the supplier selection path.

[0008] In some embodiments, the method for determining the supply chain heterogeneous graph includes:

[0009] Determining node tensor information and relationship tensor information in the supply chain knowledge graph;

[0010] The supply chain knowledge graph is converted into the supply chain heterogeneous graph according to the node tensor information and the relationship tensor information.

[0011] In some embodiments, a method for determining the node tensor information and the relationship tensor information in the supply chain knowledge graph includes:

[0012] Determine node attribute information and relationship attribute information in the supply chain knowledge graph;

[0013] Determine the node tensor information according to the attribute score and importance ratio of the node attribute information;

[0014] The relationship tensor information is determined according to the attribute score and importance ratio of the relationship attribute information.

[0015] In some embodiments, the method of converting the supply chain knowledge graph into the supply chain heterogeneous graph according to the node tensor information and the relationship tensor information includes:

[0016] Replacing the nodes in the supply chain knowledge graph with the corresponding node tensor information of the supply chain heterogeneous graph;

[0017] The relationships in the supply chain knowledge graph are replaced with the corresponding relationship tensor information of the supply chain heterogeneous graph to obtain the supply chain heterogeneous graph.

[0018] In some embodiments, determining target supplier information according to the supplier selection path includes:

[0019] Dividing the supplier selection path into triples according to source node-edge-target node to obtain each triple of the supplier selection path;

[0020] Determine the path feature tensor of each triplet in the supplier selection path;

[0021] The path feature tensors of the triples reaching the same target node are concatenated and stored in the corresponding target node feature tensor, so as to obtain the concatenated tensors of each target node in the supplier selection path;

[0022] The target supplier information is determined according to the concatenated tensors of the target nodes in the supplier selection path.

[0023] In some embodiments, determining the target supplier information according to the concatenated tensors of each target node in the supplier selection path includes:

[0024] Determine the sum of the total path feature tensors of the supplier selection path according to the concatenated tensors of each target node in the supplier selection path;

[0025] The target supplier information is determined based on the total path feature tensor of each supplier selection path.

[0026] In some embodiments, determining a path feature tensor for each triple includes:

[0027] Determine a feature tensor of a source node, a feature tensor of an edge, and a feature tensor of a target node in the triple;

[0028] Determine the product of the feature tensor of the source node and the feature tensor of the edge in the triple;

[0029] The product of the feature tensor of the source node in the triple and the feature tensor of the edge is added to the feature tensor of the target node in the triple to obtain the path feature tensor of the triple.

[0030] In some embodiments, after determining the target supplier information, the method further includes:

[0031] Based on the target supplier information, construct a supplier and parts heterogeneous graph;

[0032] And determine whether to optimize suppliers based on the heterogeneous graph of suppliers and parts.

[0033] In some embodiments, judging whether to optimize a supplier based on the supplier and component heterogeneous graph includes:

[0034] Determine the dependency of the supplier and the supplier node in the component heterogeneous graph;

[0035] If the dependency of the supplier node is within the preset dependency range, there is no need to optimize the supplier; if the dependency of the supplier node is not within the preset dependency range, it is necessary to optimize the supplier.

[0036] In some embodiments, determining the dependency of the supplier with the supplier node in the component heterogeneous graph includes:

[0037] Determine the weights of all edges of supplier nodes in the heterogeneous graph of suppliers and parts;

[0038] The dependency of the supplier node is determined according to the weights of all edges of the supplier node.

[0039] In some embodiments, the supply chain impact information includes vehicles, parts, products corresponding to the parts, suppliers, assembly information, product configuration information, and supply information;

[0040] The nodes of the supply chain knowledge graph include at least vehicles, parts, products corresponding to parts and suppliers; the relationships of the supply chain knowledge graph include at least assembly information, product configuration information and supply information.

[0041] Accordingly, the embodiment of the present application also provides a supply chain management system based on a knowledge graph, including:

[0042] A building module for building a supply chain knowledge graph based on supply chain impact information;

[0043] A first determination module is used to determine a supply chain heterogeneous graph according to the supply chain knowledge graph; wherein the supply chain heterogeneous graph includes a supplier selection path;

[0044] The second determination module is used to determine the target supplier information according to the supplier selection path.

[0045] Correspondingly, an embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the knowledge graph-based supply chain management method as described above is implemented.

[0046] Correspondingly, an embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements the supply chain management method based on the knowledge graph as described above.

[0047] Beneficial effects: Compared with the prior art, the supply chain management method, system, device and storage medium based on knowledge graph in the embodiments of the present application include: constructing a supply chain knowledge graph based on supply chain impact information; determining a supply chain heterogeneous graph based on the supply chain knowledge graph, the supply chain heterogeneous graph including supplier selection paths; determining target supplier information based on the supplier selection paths. The supply chain management method based on knowledge graph provided by the present application can intelligently and comprehensively consider the factors affecting supplier selection, and construct a supply chain knowledge graph and a supply chain heterogeneous graph based on various factors affecting supplier selection, thereby intelligently selecting high-quality suppliers through the supply chain heterogeneous graph, thereby improving the intelligence and reliability of supplier selection. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0049] Figure 1 It is a flow chart of a supply chain management method based on knowledge graph provided in an embodiment of the present application;

[0050] Figure 2 It is a schematic diagram of the structure of a supply chain knowledge graph provided in an embodiment of the present application;

[0051] Figure 3 is a schematic diagram of the structure of a supply chain heterogeneous graph provided in an embodiment of the present application;

[0052] Figure 4 It is a schematic diagram of the process of converting the supply chain knowledge graph provided in the embodiment of the present application into a supply chain heterogeneous graph;

[0053] Figure 5 is a schematic diagram of the target supplier selection process provided in an embodiment of the present application;

[0054] Figure 6 is a flowchart of another supply chain management method based on knowledge graph provided in an embodiment of the present application;

[0055] Figure 7 It is a schematic diagram of an optimization process of a supplier and component heterogeneous graph provided in an embodiment of the present application;

[0056] Figure 8 It is a schematic diagram of the overall process of the supply chain management method based on the knowledge graph provided in the embodiment of the present application;

[0057] Fig. 9 It is a principle structure diagram of a supply chain management system based on knowledge graph provided in an embodiment of the present application;

[0058] Fig.10 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0059] Reference numerals:

[0060] 10 - construction module; 20 - first determination module; 30 - second determination module; 100 - vehicle supply chain management system. DETAILED DESCRIPTION

[0061] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0062] It should be understood that although the terms first, second, etc. may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another component. Therefore, the first component discussed below can be referred to as the second component without departing from the teachings of the concepts of the present application. As used herein, the term "and / or" includes any one of the associated listed items and all combinations of one or more.

[0063] Those skilled in the art will appreciate that the drawings are only schematic diagrams of example embodiments and may not be to scale. The modules or processes in the drawings are not necessarily required to implement the present application and therefore cannot be used to limit the scope of protection of the present application.

[0064] In related technologies, the main implementation process of the supplier recommendation system includes:

[0065] (1) According to the material category information, a first supplier matching the material category information is determined in a supplier relationship graph, wherein the supplier relationship graph includes a plurality of nodes and a plurality of edges, where an edge is used to connect two nodes.

[0066] (2) Nodes are used to represent suppliers or the values ​​of supplier attributes, and edges are used to represent supplier attributes and / or the relationships between nodes.

[0067] (3) Using graph algorithms, in the supplier relationship graph, associate a second supplier that matches the material category information through an association path.

[0068] (4) The first supplier and the second supplier are called target suppliers.

[0069] (5) Filter target suppliers and display the filtered target suppliers.

[0070] (6) The supplier recommendation method expands the target suppliers through graph algorithm to recommend high-quality suppliers.

[0071] The above process of selecting suppliers for vehicle parts relies too much on manual selection, and it is not possible to intelligently select the most suitable supplier based on factors such as the assembled vehicle, parts requirements, and product parameters; after selecting the supplier, it is also impossible to intelligently analyze the supply dependency of the product, which brings great challenges to supplier selection and supply chain optimization. For the existing supplier recommendation system, the factors considered in its supplier selection process are too single, and it is not possible to comprehensively consider various influencing factors and the relationship between various influencing factors to select the appropriate supplier, nor can it adjust the supplier dependency based on the supplier's supply information.

[0072] In view of this, the embodiments of the present application provide a supply chain management method, system, device and storage medium based on a knowledge graph. When constructing a supply chain knowledge graph, the present application generates nodes for influencing factors such as vehicles, parts, and products, and constructs the adaptation relationship between these influencing factors. When selecting suppliers, not only suppliers are selected by calculating the similarity between the required materials and the supplier attributes, but also the advantages and disadvantages of each influencing factor in the supply chain and the adaptation relationship between each factor are comprehensively considered, so that the supplier selection process considers all factors to find the most suitable supplier. At the same time, the present application performs supply dependency analysis and supplier dependency optimization after selecting a supplier. As a result, the present application can realize the intelligent and comprehensive selection of suitable suppliers, and improve the reliability and intelligence of vehicle supply chain management.

[0073] Figure 1 This is a flow chart of a supply chain management method based on knowledge graph provided in an embodiment of the present application. The method can be applied to a vehicle management platform to implement a comprehensive and intelligent selection process for vehicle suppliers. The method can be executed by a supply chain management system based on knowledge graph, which can be implemented by software and / or hardware, and can be configured in a processor or controller of a vehicle management platform. Figure 1, the method comprises the following steps:

[0074] Step 110: Build a supply chain knowledge graph based on supply chain impact information.

[0075] Among them, supply chain impact information refers to all factors that affect the selection of vehicle suppliers, such as parts demand, supplier qualifications, product parameters and other factors.

[0076] Among them, supply chain impact information can be obtained from a Web page (Internet page). Specifically, before building the supply chain knowledge graph, users add, supplement, and modify information that affects supplier selection through the Web page, such as assembly model, parts requirements, product parameters, supplier qualifications, etc., so that the supplier selection can be updated in time according to the current situation, thereby ensuring that the system can flexibly select a more suitable supplier according to the actual situation.

[0077] In some embodiments, supply chain influencing information includes vehicles, parts, products corresponding to parts, suppliers, assembly information, product configuration information and supply information; the nodes of the supply chain knowledge graph include vehicles, parts, products corresponding to parts and suppliers; the relationships of the supply chain knowledge graph include assembly information, product configuration information and supply information.

[0078] Among them, the means of transportation include cars, bicycles, ships, trains, etc. Exemplarily, in the technical solution of the embodiment of the present application, the means of transportation is taken as an example for explanation (the same below, no further description is given). Among them, the parts are the parts corresponding to the means of transportation. Among them, the assembly information includes information such as the assembly requirements of the means of transportation, the assembly dimensions of the means of transportation, the assembly requirements of the parts, etc. Among them, the product configuration information is information such as the product selection corresponding to the parts. Among them, the supply information is information such as the supply source of the parts.

[0079] Specifically, after obtaining the supply chain impact information, a supply chain knowledge graph is constructed based on the supply chain impact information. Since the supply chain impact information is all factors that affect the selection of vehicle suppliers, the supply chain knowledge graph constructed based on the supply chain impact information can fully reflect the various factors that affect the selection of suppliers, thereby facilitating the subsequent comprehensive and reliable selection of the best supplier.

[0080] Step 120: Determine a supply chain heterogeneous graph based on the supply chain knowledge graph; wherein the supply chain heterogeneous graph includes supplier selection paths.

[0081] The heterogeneous supply chain graph is a graph with different types of nodes and edges. The heterogeneous supply chain graph includes multiple supplier selection paths, which can fully present all factors affecting supplier selection and provide multiple optional paths for supplier selection. The number of supplier selection paths included in the heterogeneous supply chain graph is related to the specific supply chain impact information, etc., and is not specifically limited here.

[0082] Since the supply chain knowledge graph is constructed based on supply chain impact information, the supply chain impact information can comprehensively reflect all factors that affect supplier selection. Therefore, the supply chain heterogeneous graph obtained based on the supply chain knowledge graph can comprehensively reflect all factors that affect supplier selection and provide all available supplier paths, which is conducive to the subsequent comprehensive and intelligent selection of suitable suppliers or high-quality suppliers.

[0083] Step 130: Determine target supplier information based on the supplier selection path.

[0084] The target supplier information includes the target supplier path and the target supplier. The multiple supplier paths provided by the heterogeneous supply chain graph can fully present all factors that affect supplier selection and provide multiple paths for supplier selection. Therefore, suitable suppliers or high-quality suppliers can be comprehensively and intelligently selected based on the multiple supplier paths.

[0085] In the technical solution of the embodiment of the present application, the working principle of the vehicle supply chain management method is: Figure 1 First, a supply chain knowledge graph is constructed based on the supply chain impact information. Then, a supply chain heterogeneous graph is determined based on the supply chain knowledge graph, and the supply chain heterogeneous graph includes a path for supplier selection. Finally, the target supplier information is determined based on the path for supplier selection. Therefore, this method can intelligently and comprehensively consider the factors affecting supplier selection, and construct a supply chain knowledge graph and a supply chain heterogeneous graph based on various factors affecting supplier selection. Therefore, high-quality suppliers can be intelligently selected through the supply chain heterogeneous graph, thereby improving the intelligence and reliability of supplier selection.

[0086] For example, in the technical solution of the embodiment of the present application, the supply chain management of vehicles is used as an example for explanation (hereinafter the same, no further description is given). For example, the factors affecting the selection of vehicle suppliers include: vehicles, parts, products, suppliers, assembly requirements, product selection, supply sources, inventory, warehousing, and outbound delivery.

[0087] Exemplarily, in the technical solution of the embodiment of the present application, the nodes of the supply chain knowledge graph include vehicles, parts, products (products corresponding to the parts, hereinafter referred to as products) and suppliers, and the relationships of the supply chain knowledge graph include assembly requirements, product selection and supply sources (the same below, no further details will be given).

[0088] Among them, the triple structure of the supply chain knowledge graph is "node-relationship-node", for example, "vehicle-assembly requirements-parts", "parts-product selection-product", "product-supply source-supplier" and other triples.

[0089] The supply chain influencing information may also include other factors or parameters, such as inventory, incoming warehouse, outgoing warehouse, etc. The corresponding triples, for example, constitute the triples of "parts-incoming warehouse-inventory" and "parts-outgoing warehouse-inventory".

[0090] Exemplarily, in the embodiments of the present application, supply chain influencing information includes: vehicles, parts, products, suppliers, assembly requirements, product selection and supply sources (the same below, no further details are given).

[0091] Figure 2 is a schematic diagram of the structure of a supply chain knowledge graph provided in an embodiment of the present application. Figure 2 , the nodes of the supply chain knowledge graph include: vehicles, parts (such as part a and part b), products (such as product A and product B) and suppliers (such as supplier I and supplier II), which are used to represent the characteristics of the nodes. The relationships of the supply chain knowledge graph include: assembly requirements, product selection and supply sources, which are used to represent the associated characteristics of the nodes.

[0092] Among them, the attributes of vehicles include vehicle models, configurations, batches, required parts, etc. The attributes of parts include product requirements, price standards, demand, etc. The attributes of products include product parameters, product prices, supply, origin, batches, etc. The attributes of suppliers include government rewards and punishments, business information, service quality, company public opinion, etc. The attributes of assembly requirements include assembly adaptability, importance, urgency, etc. The attributes of product selection adaptation include product adaptability, quality / effect, etc. The attributes of supply adaptation include logistics prices, consumption time, weather information, etc.

[0093] Therefore, in the process of building the supply chain knowledge graph, the factors affecting the supplier selection process are fully covered by building nodes, such as vehicles, parts, products, suppliers, assembly models, supplier qualifications and other related information, and the fitness information of these nodes is associated through relationships, such as the assembly fitness of models and parts, the product fitness of the product selection process, and the supply logistics fitness of the supplier's supply products, to ensure that the factors affecting supplier selection are fully covered. In this way, not only the pros and cons of each factor are taken into account, but also the fitness between each factor can be taken into account, making the supplier selection process more comprehensive.

[0094] Figure 3 It is a structural schematic diagram of a supply chain heterogeneous graph provided in an embodiment of the present application. In some embodiments, a method for determining a supply chain heterogeneous graph includes: determining node tensor information and relationship tensor information in a supply chain knowledge graph; and converting the supply chain knowledge graph into a supply chain heterogeneous graph according to the node tensor information and the relationship tensor information.

[0095] The node tensor information in the supply chain knowledge graph includes the feature tensor of each node in the supply chain knowledge graph, and the relationship tensor information in the supply chain knowledge graph includes the feature tensor of each relationship in the supply chain knowledge graph.

[0096] The heterogeneous supply chain graph includes multiple nodes and edges. Figure 3 , the supply chain heterogeneous graph includes points such as vehicle feature tensor, component a feature tensor, product A feature tensor, supplier I feature tensor, component b feature tensor, product B feature tensor, and supplier II feature tensor, as well as edges such as supply transportation, product selection, and assembly application. Among them, the topological structure of points and edges in the supply chain heterogeneous graph is determined by the topological structure of nodes and relationships in the supply chain knowledge graph. Since the supply chain knowledge graph is constructed based on all factors that affect supplier selection, the supply chain heterogeneous graph obtained based on the supply chain knowledge graph can comprehensively select suitable suppliers to improve the reliability and comprehensiveness of supplier selection.

[0097] In some embodiments, a method for determining node tensor information and relationship tensor information in a supply chain knowledge graph includes: determining node attribute information and relationship attribute information in the supply chain knowledge graph; determining the node tensor information based on the attribute score and importance ratio of the node attribute information; determining the relationship tensor information based on the attribute score and importance ratio of the relationship attribute information.

[0098] Among them, the methods for determining the node attribute information and relationship attribute information in the supply chain knowledge graph include: using a preprocessing language model to determine all attributes of each node and all attributes of each relationship in the supply chain knowledge graph. Among them, the node tensor information is determined according to the attribute score and importance ratio of the node attribute information, and the relationship tensor information is determined according to the attribute score and importance ratio of the relationship attribute information, specifically including: performing attribute scoring and assigning importance ratios to all attributes of each node and all attributes of each relationship; determining the feature tensors of each node according to the attribute scores of all attributes of each node and the importance ratios of all attributes of each node; determining the feature tensors of each relationship according to the attribute scores of all attributes of each relationship and the importance ratios of all attributes of each relationship.

[0099] Exemplarily, the specific implementation method for using a preprocessing language model to determine all attributes of each node and all attributes of each relationship in the supply chain knowledge graph is: using the preprocessing language model knowledge extraction method to extract the keywords of each node attribute and the keywords of each relationship attribute, and using the keywords to represent the attribute characteristics. For example, a relay with a rated voltage of xV (such as a voltage value of x volts) is extracted as voltage xV.

[0100] Among them, the preprocessing language models include the Bert-BiLSTM-CRF model, the Bert model, the CasRel model, the TPLinker model, etc. Exemplarily, the Bert-BiLSTM-CRF model is used for extracting each node attribute. The CasRel model, the TPLinker model, etc. are used for extracting each relationship attribute, and specific settings can be made according to the actual situation, and no specific limitations are made here.

[0101] Among them, performing attribute scoring on all attributes of each node and all attributes of each relationship specifically includes: for the attributes in the nodes and relationships that can directly judge the performance advantages and disadvantages, using a preprocessing language model to classify and score the attributes. Among them, the scoring rule is: the higher the performance requirement or the more excellent the performance, etc., the higher the corresponding score. For example, the Bert model is used to classify the attributes. For example, when inputting "the consistency difference of the battery < Y", the output is a score of "90"; when inputting "Y < the consistency difference of the battery < Z", the score is "80".

[0102] For the attributes in nodes and relationships that need to be compared to judge the quality, the preprocessing language model is used to compare the attributes and give a score. Among them, the scoring rule is: the higher the attribute matching degree, the higher the score (among which, the more matching the product, the higher the matching score). For example, the Bert model is used to compare attributes and give a score. For example, if the input "the energy attribute of the battery component node is mAh (i.e. m ampere-hours). The energy attribute of the product node is n Ah" (i.e. n ampere-hours), the output is a score of "0" (because the product parameters cannot match the component requirements, the score is 0).

[0103] Therefore, the preprocessing language model extracts keywords and scores the attributes of each node and each relationship in the supply chain knowledge graph to ensure a more intelligent division of the pros and cons of nodes and relationships (for example, the higher the node attribute score, the better the node attribute or the higher the adaptability between nodes), so as to intelligently evaluate the pros and cons of each influencing factor and the adaptability between each influencing factor, and build a heterogeneous supply chain graph in combination with the topological structure of the supply chain knowledge graph, making the supplier selection process more intelligent. In addition, it can also reduce manual labor and further ensure that the supplier selection is more intelligent.

[0104] The specific implementation method of assigning importance ratios to all attributes of each node and all attributes of each relationship is as follows: for all attributes of each node and each relationship, assign importance ratios to their attributes according to their importance. The more important the attribute, the higher the corresponding importance ratio. The sum of the importance ratios of all attributes of each node is 1, and the sum of the importance ratios of all attributes of each relationship is 1. For example, for a certain component, considering the importance of the attributes, the importance ratios of its product requirements, price standards, and demand are assigned as 0.5, 0.3, and 0.2, respectively, and 0.5+0.3+0.2=1. That is, product requirements are the most important attributes, followed by price standards, and the importance of demand attributes is the lowest. The calculation formula for determining the feature tensor of each node based on the attribute scores of all attributes of each node and the importance ratios of all attributes of each node is:

[0105] Node pro =[p1×α1,p2×α2,…p i ×α i …p n ×α n ]

[0106] α1+α2,+α i +…+α n =1

[0107] Among them, p1…p i …p n represents the attribute score (here, the node attribute score), α1…α i …α nIndicates the importance ratio of the node attribute. pro Represents the feature tensor of the node; where i is 1, 2, 3…n, and n is a positive integer.

[0108] The calculation formula for determining the feature tensor of each relationship (or edge) according to the attribute scores of all attributes of each relationship and the importance ratio of all attributes of each relationship is as follows:

[0109] Edge pro =[p1×β1,p2×β2,…,p i ×β i …p n ×β n ]

[0110] β1+β2+β i +…+β n =1

[0111] Among them, p1…p i …p n represents the attribute score (here, the relationship attribute score), β1…β i …β n Indicates the importance ratio of the relationship attribute. pro The feature tensor representing the relationship; where i is 1, 2, 3…n, and n is a positive integer.

[0112] Table 1. Feature tensor of battery BMS

[0113]

[0114] Exemplarily, taking the product node - battery BMS as an example, the node features are total voltage V, total current i, price P, supply quantity N, origin Z and batch M. The feature tensor of the battery BMS is calculated as shown in Table 1.

[0115] In some embodiments, a method for converting a supply chain knowledge graph into a supply chain heterogeneous graph based on node tensor information and relationship tensor information includes: replacing the nodes in the supply chain knowledge graph with the node tensor information of the corresponding supply chain heterogeneous graph; replacing the relationships in the supply chain knowledge graph with the relationship tensor information of the corresponding supply chain heterogeneous graph to obtain a supply chain heterogeneous graph.

[0116] Among them, the nodes in the supply chain knowledge graph are replaced with the node tensor information of the corresponding supply chain heterogeneous graph, and the relationships in the supply chain knowledge graph are replaced with the relationship tensor information of the corresponding supply chain heterogeneous graph to obtain the supply chain heterogeneous graph, which specifically includes: replacing each node in the supply chain knowledge graph with the feature tensor of each node of the corresponding supply chain heterogeneous graph; replacing each relationship in the supply chain knowledge graph with the feature tensor of each relationship (or edge) of the corresponding supply chain heterogeneous graph, and flipping the direction of each relationship (or edge) to obtain the supply chain heterogeneous graph.

[0117] Therefore, by constructing a supply chain knowledge graph, the factors affecting supplier selection are comprehensively covered, and a vehicle supplier heterogeneous graph is constructed based on the topological structure of the supply chain knowledge graph and the quality of nodes and relationships, so that high-quality suppliers can be selected based on the supply chain heterogeneous graph and the dependency on suppliers can be optimized.

[0118] Figure 4 Schematic diagram of the process of converting the supply chain knowledge graph provided in the embodiment of the present application into a supply chain heterogeneous graph. Figure 4 , the whole conversion process is as follows: first, based on the supply chain knowledge graph, attribute keyword extraction is performed, and the keywords of node and relationship attributes are extracted by knowledge extraction to represent their attribute characteristics. Then, attribute keyword classification and scoring are performed for the attributes in nodes and relationships that can directly judge the performance characteristics. The preprocessing language model is used to classify and score the attributes of nodes and relationships. The better the performance of the parts or product attributes, the higher the score. For the attributes in nodes and relationships that need to be compared to judge the quality, attribute keyword comparison and analysis are performed. For example, the preprocessing language model is used to compare the corresponding attributes of products and parts and give scores. The higher the matching degree between product attributes and parts required attributes, the higher the score. Then, importance is assigned. The importance ratio is assigned to the attributes of nodes and relationships according to the importance. The more important the attribute, the higher the importance ratio. The sum of the importance ratios of all attributes in each node or relationship is 1. Secondly, the node feature tensor and the relationship feature tensor are calculated. Finally, the supply chain knowledge graph is converted into a supply chain heterogeneous graph according to the node feature tensor and the relationship feature tensor.

[0119] In some embodiments, target supplier information is determined based on a supplier selection path, including: dividing the supplier selection path into triples according to source node-edge-target node to obtain each triple of the supplier selection path; determining the path feature tensor of each triple in the supplier selection path; performing a splicing operation on the path feature tensors of the triples reaching the same target node and storing them in the corresponding target node feature tensor to obtain the splicing tensor of each target node in the supplier selection path; and determining the target supplier information based on the splicing tensor of each target node in the supplier selection path.

[0120] Among them, the heterogeneous graph of the supply chain includes multiple supplier selection paths, and the target supplier information includes the target supplier path and the target supplier. Exemplarily, the specific implementation method of determining the target supplier information includes: dividing each supplier selection path into triples according to the source node-edge-target node to obtain each triple of each supplier selection path; calculating the path feature tensor of each triple in each supplier selection path according to the path propagation direction; according to the path propagation direction, sequentially performing splicing operations on the path feature tensors of the triples reaching the same target node and storing them in the corresponding target node feature tensor to obtain the splicing tensor of each target node in each supplier selection path; determining the target supplier path and the target supplier according to the splicing tensor of each target node in each supplier selection path.

[0121] Among them, the heterogeneous supply chain graph can provide multiple supplier paths for selection. Figure 3 There are multiple paths of "supplier->product->parts->vehicle" as shown. Each supplier selection path is divided into "source node-edge-target node" triples, that is, divided into three triplets of "supplier->supply transportation->product", "product->product selection->parts", and "parts->assembly application->vehicle".

[0122] After dividing each supplier selection path into triples, the path features are focused through message propagation, and the path feature tensors of each triple in each supplier selection path are calculated in turn according to the path propagation direction. And according to the path propagation direction, the path feature tensors of the triples reaching the same target node are spliced ​​in turn and stored in the corresponding target node feature tensor to obtain the splicing tensor of each target node in each supplier selection path.

[0123] Among them, the process of determining the path feature tensor of each triple includes: determining the feature tensor of the source node, the feature tensor of the edge, and the feature tensor of the target node in the triple; determining the product of the feature tensor of the source node and the feature tensor of the edge in the triple; adding the product of the feature tensor of the source node and the feature tensor of the edge in the triple to the feature tensor of the target node of the triple to obtain the path feature tensor of the triple.

[0124] Among them, the calculation formula of the path feature tensor of each triple is:

[0125] Dest pathpro =Sour pathpro ×Path weight +Dest pro

[0126] Among them, Dest pathproThe path feature tensor representing the target node of the triple is the path feature tensor of the entire triple; Sour pathpro Represents the source node path feature tensor of the triple (i.e., the concatenation of the path feature tensors of all triplets with the triple source node as the target node), Path weight Represents the path weight feature tensor, that is, the feature tensor of the path; Dest pro A feature tensor representing the triplet target node.

[0127] For example, suppose you need to calculate the path feature tensor of the triple "product-product selection-component", and its previous path is the triple "supplier-supplier transportation-product". From this, we can see that the overlapping node of the two triples is "product", so the path feature tensor of the previous path is gathered at the "product" node and serves as the source node path feature of the subsequent path. The feature tensor of product selection is the path weight feature tensor of this path, and the feature tensor of parts is the feature tensor of the target node of this path. Therefore, the path feature tensor of the "product-product selection-component" triple can be expressed as:

[0128] Component path feature tensor = product path feature tensor × product selection weight + component feature tensor

[0129] Exemplary, combined Figure 3 , according to the path propagation direction, the path feature tensors of the triples reaching the same target node are concatenated in turn and stored in the corresponding target node feature tensor, and the concatenated tensors of each target node in each supplier selection path are obtained as follows:

[0130] Dest pathpro_che =[Dest pathpro_che_1 ,…Dest pathpro_che_i …Dest pathpro_che_n ]

[0131] Dest pathpro_che_i =[Dest pathpro_ling_1 ,…Dest pathpro_ling_j …Dest pathpro_ling_n ]

[0132] Dest pathpro_long_j =[Dest pathpro_chan_1 ,…Dest pathpro_chan_k …Dest pathpro_chan_n ]

[0133] Dest pathpro_chan_k =[Dest pathpro_gong_1 ,…Dest pathpro_gong_o …Dest pathpro_gong_n ]

[0134] Among them, Dest pathpro Represents the path feature tensor of the target node. The subscripts "che", "ling", "chan", and "gong" represent vehicle nodes, component nodes, product nodes, and supplier nodes, respectively. Among them, "i", "j", "k", and "o" represent serial numbers, for example, Dest path_che_i A path feature tensor representing the i-th supplier path to the vehicle node.

[0135] The splicing operation is performed because there are multiple paths to each target node, and the path feature tensors of all paths need to be spliced ​​to represent the path features of the node. For example, because a vehicle requires multiple parts, there are multiple paths to the vehicle. The path feature tensors of all paths to the vehicle are spliced ​​and merged into Dest pathpro_che The same goes for the rest, so I won’t go into details.

[0136] In some embodiments, target supplier information is determined based on the concatenated tensors of each target node in the supplier selection path, including: determining the total path feature tensor sum of the supplier selection path based on the concatenated tensors of each target node in the supplier selection path; and determining the target supplier information based on the total path feature tensor sum of each supplier selection path.

[0137] Exemplarily, a specific implementation method for determining the target supplier information includes: calculating the total path feature tensor sum of each supplier selection path based on the concatenated tensor of each target node in each supplier selection path; selecting the path with the largest sum value from the total path feature tensor sum of each supplier selection path as the target supplier path, and selecting the supplier corresponding to the target supplier path as the target supplier.

[0138] Specifically, in each path of "supplier->product->parts->vehicle", all elements in the total path feature tensor are added and summed to obtain the total path feature tensor sum of the path. Thus, the total path feature tensor sum of each supplier selection path is calculated. The total path feature tensor sum of each supplier selection path is compared, and the path with the largest sum value is selected as the target supplier path, and the supplier corresponding to the path with the largest sum value is selected as the target supplier.

[0139] The calculation formula for the sum of the total path feature tensors of each supplier selection path is:

[0140] Dest pathpro_che_isum =Dest pathpro_ling_1 +…+Dest pathpro_ling_j +……+Dest pathpro_ling_n

[0141] Among them, Dest pathpro_che_isum is the sum of the total path feature tensors of one of the supplier paths for selection, that is, the sum of all elements of the path feature tensor of the target node vehicle, Dest pathpro_ling_n It is the path feature tensor of the nth path to the component node for selecting a supplier.

[0142] Therefore, the present application transmits the scoring information of the nodes and relationships of the path (i.e., the pros and cons of the nodes and relationships of the path) to the path end node - the vehicle in sequence through message propagation and message integration, ensuring that the path information is fully summarized at the path end node, thereby ensuring that the supplier selection process considers more comprehensive information, and further ensuring that the optimal path and the optimal supplier are selected.

[0143] It should be noted that in the technical solution of the above embodiment of the present application, the path information is calculated and the optimal path is selected by message propagation. In addition, the optimal path can also be selected by other graph algorithms, such as Graph Neural Network (GNN).

[0144] Figure 5 is a schematic diagram of the target supplier selection process provided in the embodiment of the present application. For example, please refer to Figure 5 ,Based on the heterogeneous graph of the supply chain, first, the path is divided. In each path of "supplier-product-parts-vehicle", the "source node-edge-target node" triples are divided. Then, the message propagation, the source node feature tensor of each triple is cross-multiplied with the weight feature tensor of the edge in the direction of path propagation, and added to the node feature tensor of the target node to obtain the path feature tensor. Secondly, the message integration, the path feature tensor is stored in the target node feature tensor in the direction of path propagation, and the path feature tensor reaching the same target node is spliced. Then, the path feature tensor is summed. In each path of "supplier->product->parts->vehicle", all spliced ​​tensors reaching the vehicle node are summed to obtain the total path feature tensor sum of the corresponding path. Finally, the path and supplier selection, compare and select the path with the highest total path feature tensor sum, and the supplier on this path is used as the target supplier of the part of this model.

[0145] Figure 6 is a flow chart of another supply chain management method based on knowledge graph provided in the embodiment of the present application. Figure 6 , the method comprises the following steps:

[0146] Step 210: Build a supply chain knowledge graph based on supply chain impact information.

[0147] Step 220: Determine a supply chain heterogeneous graph based on the supply chain knowledge graph; wherein the supply chain heterogeneous graph includes supplier paths for selection.

[0148] Step 230: Determine target supplier information based on the supplier selection path.

[0149] Step 240: Based on the target supplier information, a supplier and component heterogeneous graph is constructed.

[0150] Specifically, the target supplier can be determined based on the target supplier information. After the target supplier is determined, a supplier and component heterogeneous graph is constructed for the selected supplier based on the supply information (such as product supply quantity and price). The supplier selection can be adjusted based on product price, transportation cost, supply quantity and other information to optimize the dependence of components on products and suppliers.

[0151] Among them, the specific construction method of the supplier and parts heterogeneous graph can refer to the construction method of the supply chain heterogeneous graph, and the details will not be repeated here.

[0152] When constructing a heterogeneous graph of suppliers and parts, a "supplier-part" edge is established, where the edge points from the supplier to the part, and the weight of the edge is the cost of purchasing the part from the supplier, that is:

[0153] Weight = quantity of supplied products * (product price + logistics price)

[0154] Step 250: Determine whether to optimize the supplier based on the supplier and component heterogeneous graph.

[0155] In some embodiments, determining whether to optimize a supplier is based on a heterogeneous graph of suppliers and parts, including: determining the dependency of supplier nodes in the heterogeneous graph of suppliers and parts; if the dependency of the supplier node is within a preset dependency range, there is no need to optimize the supplier; if the dependency of the supplier node is not within the preset dependency range, the supplier needs to be optimized.

[0156] Specifically, the dependency of the supplier node is calculated, and the dependency of all supplier nodes is compared to determine whether it is within the preset dependency range. If it is within the preset dependency range, there is no need to adjust the node supplier. If it is lower than or exceeds the preset dependency range, it is necessary to adjust the node supplier, such as adjusting the node supplier's supply products and supply volume, to reduce the risk of dependence on suppliers.

[0157] In some embodiments, determining the dependency of a supplier node in a heterogeneous graph of suppliers and parts includes: determining the weights of all edges of the supplier node in the heterogeneous graph of suppliers and parts; and determining the dependency of the supplier node based on the weights of all outgoing edges of the supplier node.

[0158] Calculating the dependency (or dependency centrality) of the supplier node according to the weights of all edges of the supplier node includes: calculating the sum of the weights of all outgoing edges of the supplier node, and taking the sum of the weights of all outgoing edges of the supplier node as the dependency of the supplier node. In addition, other calculation methods can also be used to calculate the dependency, such as eigenvector centrality.

[0159] For example, a range of reliance on a single supplier (i.e., a preset reliance range) is set to tilt the reliance of the product on suppliers toward high-quality suppliers, or reduce excessive reliance on a single supplier. The specific range value of the preset reliance range can be set according to actual conditions and is not specifically limited here.

[0160] Figure 7 is a schematic diagram of an optimization process of a supplier and component heterogeneous graph provided in an embodiment of the present application. Figure 7 First, construct a product dependency graph, that is, construct a heterogeneous graph of suppliers and parts. Then, establish a supplier-parts edge, where the edge points from the supplier to the part, and the weight of the edge is the supplier's charge = the number of products supplied * (product price + logistics price). Secondly, calculate the supplier node dependency, that is, calculate the sum of the weights of all outgoing edges. Secondly, set the preset dependency range, such as setting the minimum dependency x1 and the maximum dependency x2. Finally, optimize the dependency relationship and compare the dependency of each supplier node with the size of the preset dependency range [x1-x2]. If the supplier in this range meets the standard, no adjustment is required. If the supplier is lower than or exceeds this range, the supply quantity needs to be adjusted. Therefore, after the supplier has been selected, the heterogeneous graph of suppliers and parts is constructed according to the supply information (such as product supply quantity and price structure), the preset dependency range of the supplier is set and the supplier dependency is calculated, and the supply quantity of the supplier that is not within the preset dependency range is adjusted based on this, the risk of dependence on the supplier can be effectively reduced.

[0161] Figure 8 is a schematic diagram of the overall process of the supply chain management method based on the knowledge graph provided in the embodiment of the present application, for example, see Figure 8First, the user inputs the factors that affect the supplier selection through the web page, such as parts demand, supplier qualifications, product parameters and other information. Then, the supply chain knowledge graph is constructed according to the vehicle, the parts for assembling the vehicle, the product selection corresponding to the parts, and the supplier of the product to form an "entity-relationship-entity" triple, and its attribute features are added to the entity and relationship respectively. Among them, the nodes of the supply chain knowledge graph include "vehicle", "parts", "product", and "supplier", and the relationships of the supply chain knowledge graph include "assembly requirements", "product selection", "supply source", etc. Secondly, the nodes and relationships of the supply chain knowledge graph are converted into a supply chain heterogeneous graph. The topological relationship between the nodes and edges of the supply chain heterogeneous graph is determined by the topological structure of the nodes and relationships in the supply chain knowledge graph. Secondly, the feature tensor of the points in the supply chain heterogeneous graph is determined by the attribute score and importance of the nodes in the supply chain knowledge graph, and the feature tensor of the weight of the edge in the supply chain heterogeneous graph is determined by the attribute score and importance of the relationship in the supply chain knowledge graph. Secondly, in each path of "supplier-product-parts-vehicle", starting from the supplier as the source node, the node feature tensors and edge weight feature tensors on the path are passed to the target node "vehicle" in turn through message propagation. Then, according to the path information passed to the target node, the appropriate path and supplier (i.e., target supplier path and target supplier) are selected. After the target supplier is selected, the attributes of each node and relationship in the supply chain knowledge graph are adjusted according to the assembly effect of the parts and after-sales traceability, so as to optimize the supplier selection according to the actual use situation and make the next supplier selection more in line with the actual use requirements. For example, the attributes of the nodes and relationships of the supply chain knowledge graph can be changed according to the actual use of the parts, such as the use effect and after-sales service. (For example, the influencing factors can be filled in through the web interface, and these influencing factors can also be changed). Finally, for the selected suppliers, a heterogeneous graph of suppliers and parts is constructed, and the supplier selection is adjusted according to information such as product price, transportation cost, and supply quantity, so as to optimize the dependence of parts on products and suppliers.

[0162] It can be seen that this application uses each influencing factor of supplier selection as a node, and the adaptation relationship between each influencing factor as a relationship to construct a supply chain knowledge graph; the attributes of the nodes and relationships of the supply chain knowledge graph are evaluated and scored through a preprocessing language model to represent information such as product quality, assembly fitness, and supply fitness. A supply chain heterogeneous graph is generated based on the topological structure of the supply chain knowledge graph, the pros and cons of each influencing factor, and the adaptation relationship between the influencing factors, and message propagation and message integration are performed on the supply chain heterogeneous graph to ensure that the optimal supplier path and supplier are selected intelligently and comprehensively according to each influencing factor. In addition, a heterogeneous graph of suppliers and parts is constructed based on factors such as product price and supply volume, and the product dependency of each supplier is calculated. The supply dependency is optimized based on the supplier dependency range to reduce the risk of dependency on suppliers.

[0163] Fig. 9 This is a schematic diagram of the principle structure of a supply chain management system based on a knowledge graph provided in the embodiment of the present application. Correspondingly, the embodiment of the present application also provides a supply chain management system based on a knowledge graph, please refer to Fig. 9 The supply chain management system 100 based on the knowledge graph includes: a construction module 10, which is used to construct a supply chain knowledge graph based on supply chain impact information; a first determination module 20, which is used to determine a supply chain heterogeneous graph according to the supply chain knowledge graph; wherein the supply chain heterogeneous graph includes supplier paths for selection; a second determination module 30, which is used to determine target supplier information according to the supplier paths for selection.

[0164] The technical solution of the embodiment of the present application provides a vehicle supply chain management system, which includes a construction module for constructing a supply chain knowledge graph based on supply chain impact information; a first determination module for determining a supply chain heterogeneous graph based on the supply chain knowledge graph; wherein the supply chain heterogeneous graph includes a supplier selection path; and a second determination module for determining target supplier information based on the supplier selection path. The system can intelligently and comprehensively consider the factors that affect supplier selection, and construct a supply chain knowledge graph and a supply chain heterogeneous graph based on various factors that affect supplier selection, thereby intelligently selecting high-quality suppliers through the supply chain heterogeneous graph, thereby improving the intelligence and reliability of vehicle supplier selection.

[0165] In some embodiments, the first determination module 20 includes: a first determination unit, configured to determine node tensor information and relationship tensor information in the supply chain knowledge graph; and a conversion unit, configured to convert the supply chain knowledge graph into a supply chain heterogeneous graph according to the node tensor information and the relationship tensor information.

[0166] In some embodiments, the first determination unit is also used to determine node attribute information and relationship attribute information in the supply chain knowledge graph; determine node tensor information based on the attribute score and importance ratio of the node attribute information; and determine relationship tensor information based on the attribute score and importance ratio of the relationship attribute information.

[0167] In some embodiments, the conversion unit is also used to replace the nodes in the supply chain knowledge graph with the corresponding node tensor information of the supply chain heterogeneous graph; replace the relationships in the supply chain knowledge graph with the corresponding relationship tensor information of the supply chain heterogeneous graph to obtain the supply chain heterogeneous graph.

[0168] In some embodiments, the second determination module 30 includes: a triple division unit, which is used to divide the supplier selection path into triples according to the source node-edge-target node to obtain each triple of the supplier selection path. A second determination unit is used to determine the path feature tensor of each triple in the supplier selection path. A splicing and storage unit is used to splice the path feature tensors of the triples reaching the same target node and store them in the corresponding target node feature tensor to obtain the splicing tensor of each target node in the supplier selection path; a third determination unit is used to determine the target supplier information according to the splicing tensor of each target node in the supplier selection path.

[0169] In some embodiments, the third determination unit is also used to determine the total path feature tensor sum of the supplier selection path based on the splicing tensor of each target node in the supplier selection path; and determine the target supplier information based on the total path feature tensor sum of each supplier selection path.

[0170] In some embodiments, the second determination unit is also used to determine the feature tensor of the source node, the feature tensor of the edge, and the feature tensor of the target node in the triplet; determine the product of the feature tensor of the source node and the feature tensor of the edge in the triplet; add the product of the feature tensor of the source node and the feature tensor of the edge in the triplet to the feature tensor of the target node of the triplet to obtain the path feature tensor of the triplet.

[0171] In some embodiments, the supply chain management system based on the knowledge graph further includes: a supplier and component heterogeneous graph construction module, which is used to construct a supplier and component heterogeneous graph based on target supplier information; and an optimization judgment module, which is used to judge whether to optimize the supplier based on the supplier and component heterogeneous graph.

[0172] In some embodiments, the optimization judgment module is also used to determine the dependency between suppliers and supplier nodes in the component heterogeneous graph; if the dependency of the supplier node is within a preset dependency range, there is no need to optimize the supplier; if the dependency of the supplier node is not within the preset dependency range, the supplier needs to be optimized.

[0173] In some embodiments, the optimization judgment module is also used to determine the weights of all edges of the supplier node in the supplier and component heterogeneous graph; and determine the dependency of the supplier node based on the weights of all edges of the supplier node.

[0174] In some embodiments, supply chain impact information includes vehicles, parts, products corresponding to parts, suppliers, assembly information, product configuration information and supply information; the nodes of the supply chain knowledge graph include at least vehicles, parts, products corresponding to parts and suppliers; the relationships of the supply chain knowledge graph include at least assembly information, product configuration information and supply information.

[0175] Accordingly, the present application also provides an electronic device, see Fig.10 , Fig.10 The structure diagram of the electronic device of the embodiment of the present application is illustrated. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the supply chain management method based on the knowledge graph are implemented. Since the supply chain management method based on the knowledge graph is described in detail above, it will not be repeated here.

[0176] Accordingly, an embodiment of the present application further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned supply chain management method based on the knowledge graph are implemented. Since the above-mentioned supply chain management method based on the knowledge graph is described in detail, it will not be repeated here.

[0177] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0178] The above is a detailed introduction to the knowledge graph-based supply chain management method, system, device and storage medium provided in the embodiments of the present application, and specific examples are used to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the technical solution and core idea of ​​the present application; ordinary technicians in this field should understand that they can still modify the technical solutions recorded in the aforementioned embodiments, or replace some of the technical features therein with equivalents; and these modifications or replacements do not make the essence of the corresponding technical solution deviate from the scope of the technical solution of the embodiments of the present application.

Claims

1. A supply chain management method based on knowledge graph, characterized in that: include: Build a supply chain knowledge graph based on supply chain impact information; Determine a supply chain heterogeneous graph according to the supply chain knowledge graph, wherein the supply chain heterogeneous graph includes supplier paths for selection; The target supplier information is determined according to the supplier selection path.

2. The supply chain management method based on knowledge graph according to claim 1 is characterized in that: The method for determining the supply chain heterogeneous graph includes: Determining node tensor information and relationship tensor information in the supply chain knowledge graph; The supply chain knowledge graph is converted into the supply chain heterogeneous graph according to the node tensor information and the relationship tensor information.

3. The supply chain management method based on knowledge graph according to claim 2 is characterized in that: The method for determining the node tensor information and the relationship tensor information in the supply chain knowledge graph includes: Determine node attribute information and relationship attribute information in the supply chain knowledge graph; Determine the node tensor information according to the attribute score and importance ratio of the node attribute information; The relationship tensor information is determined according to the attribute score and importance ratio of the relationship attribute information.

4. The supply chain management method based on knowledge graph according to claim 2 is characterized in that: The method for converting the supply chain knowledge graph into the supply chain heterogeneous graph according to the node tensor information and the relationship tensor information includes: Replacing the nodes in the supply chain knowledge graph with the corresponding node tensor information of the supply chain heterogeneous graph; The relationships in the supply chain knowledge graph are replaced with the corresponding relationship tensor information of the supply chain heterogeneous graph to obtain the supply chain heterogeneous graph.

5. The supply chain management method based on knowledge graph according to claim 1 is characterized in that: Determining target supplier information according to the supplier selection path includes: Dividing the supplier selection path into triples according to source node-edge-target node to obtain each triple of the supplier selection path; Determine the path feature tensor of each triplet in the supplier selection path; The path feature tensors of the triples reaching the same target node are concatenated and stored in the corresponding target node feature tensor, so as to obtain the concatenated tensors of each target node in the supplier selection path; The target supplier information is determined according to the concatenated tensors of the target nodes in the supplier selection path.

6. The supply chain management method based on knowledge graph according to claim 5 is characterized in that: The step of determining the target supplier information according to the concatenated tensors of each target node in the supplier selection path includes: Determine the sum of the total path feature tensors of the supplier selection path according to the concatenated tensors of each target node in the supplier selection path; The target supplier information is determined based on the total path feature tensor of each supplier selection path.

7. The supply chain management method based on knowledge graph according to claim 5 is characterized in that: Determine the path feature tensor for each triple, including: Determine a feature tensor of a source node, a feature tensor of an edge, and a feature tensor of a target node in the triple; Determine the product of the feature tensor of the source node and the feature tensor of the edge in the triple; The product of the feature tensor of the source node in the triple and the feature tensor of the edge is added to the feature tensor of the target node in the triple to obtain the path feature tensor of the triple.

8. The supply chain management method based on knowledge graph according to claim 1, characterized in that: After determining the target supplier information, it also includes: Based on the target supplier information, construct a supplier and parts heterogeneous graph; And determine whether to optimize suppliers based on the heterogeneous graph of suppliers and parts.

9. The supply chain management method based on knowledge graph according to claim 8 is characterized in that: The step of judging whether to optimize a supplier based on the heterogeneous graph of suppliers and parts includes: Determine the dependency of the supplier and the supplier node in the component heterogeneous graph; If the dependency of the supplier node is within the preset dependency range, there is no need to optimize the supplier; if the dependency of the supplier node is not within the preset dependency range, it is necessary to optimize the supplier.

10. The supply chain management method based on knowledge graph according to claim 9, characterized in that: Determining the dependency of the supplier and the supplier node in the component heterogeneous graph includes: Determine the weights of all edges of supplier nodes in the heterogeneous graph of suppliers and parts; The dependency of the supplier node is determined according to the weights of all edges of the supplier node.

11. The supply chain management method based on knowledge graph according to claim 1, characterized in that: The supply chain impact information includes vehicles, parts, products corresponding to parts, suppliers, assembly information, product configuration information and supply information; The nodes of the supply chain knowledge graph include at least vehicles, parts, products corresponding to parts and suppliers; the relationships of the supply chain knowledge graph include at least assembly information, product configuration information and supply information.

12. A supply chain management system based on knowledge graph, characterized in that: include: A building module for building a supply chain knowledge graph based on supply chain impact information; A first determination module is used to determine a supply chain heterogeneous graph according to the supply chain knowledge graph; wherein the supply chain heterogeneous graph includes a supplier selection path; The second determination module is used to determine the target supplier information according to the supplier selection path.

13. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the knowledge graph-based supply chain management method as described in any one of claims 1 to 11 is implemented.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the knowledge graph-based supply chain management method as described in any one of claims 1 to 11 is implemented.

Images