A risk identification method for automobile supply chain based on knowledge graph

Abstract

The present invention relates to the field of data processing technology, and the present invention relates to a method for identifying automobile supply chain risks based on knowledge graphs, including: obtaining entity types of automobile supply chains, dividing automobile supply chain hierarchies based on entity types; each entity type contains multiple entities, and each two entities contain one or more relationships, and the entities are used as nodes and the relationships between entities are used as edges to construct a knowledge graph of the automobile supply chain; for node risk changes, the impact on the overall risk of the supply chain is calculated, and the node risk is quantified through the edge weight, initial risk change and hierarchy of the node, and the risk level is divided; finally, the initial risk change and risk level are input into a machine model for training to predict the risk level of the node to be detected in the supply chain. The present invention solves the problem of inaccurate identification of automobile supply chain risks.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology. More specifically, the present invention relates to a method for identifying risks in an automobile supply chain based on a knowledge graph. Background Art

[0002] In the modern automotive industry, the complexity of the supply chain is increasing, involving a large number of suppliers, manufacturers, distributors, logistics service providers, retailers and other participants. Each participant plays an important role in the supply chain and is interconnected. The stability of their operations directly affects the security and efficiency of the entire supply chain. However, with the acceleration of globalization, the vulnerability of the supply chain has also increased. For example, if there is a problem with a supplier in a certain link, it may lead to the interruption of the entire supply chain, thereby affecting the production and sales of the entire automotive industry. At present, supply chain risk management is one of the key areas for enterprises to maintain their competitiveness.

[0003] Traditional supply chain risk identification methods usually rely on static data analysis or expert experience to make judgments. Although these methods can provide certain guidance, they are often unable to achieve real-time and dynamic risk identification and early warning when facing complex global supply chains. For example, when a risk occurs at a certain supply chain node, how to evaluate the transmission effect of the risk on the entire supply chain and its potential impact on other nodes is always a challenge.

[0004] The patent application document with application publication number CN116777205A discloses a supply chain risk prediction method, electronic device and storage medium. The patent application document includes: obtaining material supply history data of multiple suppliers; analyzing the association relationship of each supplier based on the material supply history data, and constructing an association relationship knowledge map; performing risk prediction operations on one or more suppliers in response to risk prediction instructions, and obtaining one or more risk prediction results in risk supply chain, risk range, problem supplier and risk community. However, this method only relies on a single risk node to evaluate supply chain risks, while ignoring the impact of risk changes between nodes on different levels in the supply chain, resulting in inaccurate supply chain risk identification. Summary of the invention

[0005] In order to solve the problem of inaccurate identification of automobile supply chain risks raised in the above background technology, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for identifying automobile supply chain risks based on a knowledge graph, comprising:

[0007] Obtain the entity types of the automobile supply chain and divide each entity type into a level of the automobile supply chain; each entity type contains multiple entities, and each two entities contain one or more relationships. Entities are used as nodes and relationships between entities are used as edges to construct a knowledge graph of the automobile supply chain; calculate nodes Risks to the automotive supply chain :

[0008] ;

[0009] In the formula, Representation Node To Node The edge rights, Representation Node In time The initial risk change at Representation and Node A collection of connected nodes, Representation Node The level where Representation Node In time The initial risk change at the time; the edge weight and node and nodes The entropy weight of the relationship between them is positively correlated with the initial indicator value; nodes with risks less than a set threshold are considered low risk, and nodes with risks greater than or equal to the set threshold are considered high risk; the initial risk change of each node and its corresponding risk level are input into the machine model for training to obtain a trained machine model; the initial risk change of the node to be detected is input into the trained machine model, and the risk level of the node to be detected is output.

[0010] The above technical solution constructs a knowledge graph of the automotive supply chain, structurally represents the entities and their relationships in the supply chain, and introduces a risk assessment algorithm based on the relationship between nodes, thereby dynamically evaluating the risk level of each node in the supply chain. It can more accurately represent the complex relationships between entities in the supply chain and improve the comprehensiveness and accuracy of automotive supply chain risk identification.

[0011] Furthermore, the node To Node Edge weight for:

[0012] ;

[0013] In the formula, Representation Node With Node No. The entropy weight of a relationship, Representation Node With Node No. The initial index value of the relationship, Representation Node With Node The total number of relationships between .

[0014] The above technical solution can more accurately measure the strength of the relationship between nodes by introducing entropy weights and initial index values, so that the edge weights can reflect the influence of multiple relationship dimensions, thereby improving the accuracy and flexibility of risk calculation. This multi-dimensional weighting method can capture the complex relationship structure in the supply chain and further improve the accuracy of supply chain risk identification and early warning.

[0015] Furthermore, it also includes using entropy weight method to calculate the node With Node No. The entropy weight of a relationship , specifically:

[0016] ;

[0017] In the formula, Representation Node With Node Between The information entropy of a relationship, Representation Node With Node The total number of relationships between .

[0018] The above technical solution introduces the entropy weight method to calculate the entropy weight of the relationship between nodes, and uses information entropy to measure uncertainty to dynamically adjust the contribution of each relationship to the edge weight. By adjusting the relationship weight of information entropy, the role of key and stable relationships in risk identification is enhanced, thereby improving the rationality and accuracy of risk assessment, making risk identification more in line with the actual situation of the supply chain.

[0019] Furthermore, it also includes preprocessing the initial indicator value by using Z-score standardization.

[0020] The above technical solution, by using Z-score standardization preprocessing on the initial indicator values, makes the initial indicator values ​​of different relationships comparable, eliminates the influence of different dimensions on the calculation, and effectively improves the stability of data processing and the accuracy of calculation results. By converting each indicator value into a standard normal distribution, the solution further enhances the accuracy of supply chain risk assessment.

[0021] Furthermore, the entity types include: suppliers, manufacturers, distributors, retailers, logistics service providers and consumers.

[0022] Furthermore, the supplier's entities include: the supplier's financial status, main business and parts quality; the manufacturer's entities include: the manufacturer's production capacity and product quality;

[0023] The entities of distributors include: the distributor's distribution network, market coverage and logistics capabilities; the entities of retailers include: the retailer's sales and supply chain; the entities of logistics service providers include: the logistics service provider's transportation capacity, warehousing capacity and transportation costs; the entities of consumers include: the consumer's purchasing power, consumption habits and purchasing behavior.

[0024] Furthermore, the relationships include: supply relationship, manufacturing relationship, distribution relationship, retail relationship, logistics relationship and purchasing relationship.

[0025] Furthermore, the machine model is a random forest model.

[0026] The above technical solution uses the random forest model, which can improve the accuracy and robustness of supply chain risk prediction by utilizing its efficient processing ability for complex data and strong generalization ability. By integrating the results of multiple decision trees, the random forest model can reduce the risk of overfitting, while improving the model's adaptability to different types of data and its ability to accurately judge potential risks.

[0027] Furthermore, the initial risk change is the node difference between the current time point and the previous time point, and the initial index value is the correlation between the nodes.

[0028] The beneficial effects of the present invention are:

[0029] The present invention constructs an automobile supply chain model based on knowledge graph, uses the relationship between nodes and edges, entropy weights, and initial risk changes to accurately assess the risk impact of each node on the supply chain, and improves the stability of data processing through Z-score standardization. Combined with the random forest model for risk classification, the automation and efficiency of supply chain risk identification are achieved, and comprehensive analysis of entity relationships at different levels can be performed, significantly improving the accuracy and reliability of supply chain risk warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0031] Figure 1 is a flowchart schematically illustrating a method for identifying automobile supply chain risks based on a knowledge graph according to an embodiment of the present invention;

[0032] Figure 2 It is a schematic diagram schematically illustrating an automobile supply chain knowledge graph according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0034] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0035] An embodiment of a method for identifying risks in an automobile supply chain based on a knowledge graph.

[0036] like Figure 1 As shown, a flow chart of a method for identifying automobile supply chain risks based on a knowledge graph in an embodiment of the present invention includes the following steps:

[0037] S1: Obtain the entity types of the automobile supply chain, divide the levels according to the entity types, and build the automobile supply chain knowledge graph.

[0038] Among them, the types of automobile supply chain entities include suppliers, manufacturers, distributors, retailers, logistics service providers and consumers. Manufacturers can be automobile OEMs, parts manufacturers, body manufacturers, etc. Suppliers can be first-tier parts suppliers, second-tier parts suppliers and third-tier parts suppliers, etc. Distributors can be regional distributors, franchise distributors and independent distributors, retailers can be used car dealers, parts retailers, etc. Logistics service providers can be railway freight companies, airlines, shipping companies, etc. Consumers can be individual consumers, corporate customers and public institutions, etc. Suppliers' entities include: suppliers' financial status, main business, production cycle, settlement methods and parts quality, etc. Manufacturers' entities include: manufacturers' production capacity, product quality, R&D capabilities and partnerships, etc. Distributors' entities include: distributors' distribution networks, market coverage, inventory management capabilities and logistics capabilities, etc. Retailers' entities include: retailers' sales, customer satisfaction and supply chain, logistics service providers' entities include: logistics service providers' transportation capacity, warehousing capacity, regional weather and transportation costs, and consumers' entities include: consumers' purchasing power, consumption habits and purchasing behavior, etc. Of course, the entity types and entities in the automobile supply chain can also be set according to actual conditions.

[0039] In this embodiment, the automobile supply chain is assumed to be a pyramid structure. Each entity type is a level of the automobile supply chain, and there are one or more relationships between every two entities. The relationships include: supply relationship, manufacturing relationship, distribution relationship, retail relationship, logistics relationship and purchasing relationship. For example, the financial status of the supplier has a supply relationship with the production capacity of the manufacturer, and the logistics capacity of the distributor has a retail relationship and a logistics relationship with the supply chain of the retailer. The knowledge graph of the automobile supply chain is constructed by taking entities as nodes and the relationships between entities as edges.

[0040] S2: The risks that changes in computing node risks may have on the entire automotive supply chain.

[0041] In one embodiment, the computing node Risks to the automotive supply chain :

[0042] ;

[0043] In the formula, Representation Node To Node The edge rights, Representation Node In time The initial risk change at Representation and Node A collection of connected nodes, Representation Node The level where Representation Node In time The initial risk change at time ; the edge weight and node and nodes The entropy weight of the relationship and the initial index value are positively correlated;

[0044] node To Node Edge weight for:

[0045] ;

[0046] In the formula, Representation Node With Node No. The entropy weight of a relationship, Representation Node With Node No. The initial index value of the relationship, Representation Node With Node The total number of relationships between .

[0047] It also includes the entropy weight method to calculate the node With Node No. The entropy weight of a relationship , specifically:

[0048] ;

[0049] In the formula, Representation Node With Node Between The information entropy of a relationship.

[0050] like Figure 2 As shown, a schematic diagram of the automobile supply chain knowledge graph of an embodiment of the present invention.

[0051] For example: Figure 2 As shown, assume that the automobile supply chain contains three nodes: node A, node B and node C, and node A is at the first level, node B is at the second level, and node C is at the third level. Get the node A at time The initial risk change at time is 0.05, and node B is at time The initial risk change at time is 0.02, and node C is The initial risk change at 0.03, if the α edge between node A and node B has a manufacturing relationship and a distribution relationship, and the β edge between node A and node C has a manufacturing relationship and a distribution relationship, substituting into the above entropy weight calculation formula, the entropy weight of the manufacturing relationship between node A and node B is 0.4, the entropy weight of the manufacturing relationship between node A and node C is 0.3, the entropy weight of the distribution relationship between node A and node B is 0.6, and the entropy weight of the distribution relationship between node A and node C is 0.7;

[0052] The initial index value of the manufacturing relationship between node A and node B is 0.5, the initial index value of the distribution relationship between node A and node B is 0.7, the initial index value of the manufacturing relationship between node A and node C is 0.6, and the initial index value of the distribution relationship between node A and node C is 0.8. Substituting them into the above edge weight calculation formula, the edge weight for node A and node B is 0.62, and the edge weight between node B and node C is 0.74. Substituting them into the above automobile supply chain risk formula, the overall risk of the supply chain caused by node A is 0.0285.

[0053] In one embodiment, the initial indicator value is the correlation between nodes, and the correlation is obtained by quantifying the relationship between nodes through a quantitative evaluation method or a risk factor analysis method. For example, the initial indicator value of the manufacturing relationship between node A and node B after quantification through a quantitative evaluation method is 0.5, and the initial indicator value of the distribution relationship between node A and node B after quantification through a quantitative evaluation method is 0.7.

[0054] And a node contains multiple risk factors, and the sum of the risk level changes of the risk factors is the initial risk change of the node. There are three ways to obtain the initial risk change. The first is to artificially increase the initial risk change; the second is to obtain the initial risk change based on the node difference between the current time point and the previous time point. For example, the sales of a supplier at the previous time point is 0, and the current time point is 0.25. The sales state changes from 0 to 0.25 from the previous time point to the current time point, and the initial risk change of the node is 0.25; the third is to build a neural network prediction model, divide the weight according to the influence of each risk factor on the node, predict the probability of risk factor changes, multiply the probability of each risk factor by the weight, and add them to get the total risk level. The second method is used in this embodiment.

[0055] S3: Divide the risk levels based on the risks, input the initial risk change and its corresponding risk level into the machine model for training, and obtain a trained machine model.

[0056] Specifically, traverse each node in the supply chain at time The risk generated to the automotive supply chain at that time is low risk if the risk is less than the set threshold, and high risk if the risk is greater than or equal to the set threshold.

[0057] In this embodiment, the threshold is set to 0.6, which can of course be set according to actual conditions. Finally, the initial risk change of each node and its corresponding risk level are input into the machine model for training to obtain a trained machine model; the initial risk change of the node to be detected is input into the trained machine model to output the risk level of the node to be detected. Among them, the machine model is a random forest model, and the specific training process is:

[0058] A training set is obtained; the training set includes the initial risk change of each node and its corresponding risk level; and the labels are manually calibrated.

[0059] The training set is input into the pre-built random forest model for training. During the training process, the loss between the output prediction value and the label is calculated; the model parameters are adjusted using the gradient descent method to minimize the prediction error; the parameters of the random forest model are iteratively adjusted until the loss is less than a certain value or the set number of training times is reached, and finally a trained random forest model is obtained.

[0060] S4: Input the risk change of the node to be detected into the trained machine model, and output the risk level of the node to be detected.

[0061] In one embodiment, it is assumed that the knowledge graph of the automotive supply chain has been constructed, including entities such as suppliers, manufacturers, distributors, retailers, logistics service providers, and consumers and their relationships. Each entity type is divided into different levels of the supply chain. For example, manufacturers are the upstream level of suppliers, while retailers are the downstream level of suppliers. Supplier entities include: supplier's financial status, main business, and parts quality; manufacturer entities include: manufacturer's production capacity and product quality, and the supplier's financial status has a supply relationship with the manufacturer's production capacity. All these entities and the relationships between them are represented as nodes and edges, forming a detailed automotive supply chain knowledge graph.

[0062] Then, the initial risk changes of all nodes in the knowledge graph of the automobile supply chain are collected. The collected initial risk change data of all nodes in the automobile supply chain are input into the trained random forest model to obtain the risk level of the node to be detected in the automobile supply chain.

[0063] In this embodiment, after determining the risk level of the node to be detected, a reminder may be given, such as a reminder to the relevant management personnel in the form of an alarm or a text message, and adaptive adjustments may be made to each node in a timely manner.

[0064] The solution of the present invention can effectively capture the complex entity relationships and hierarchical structures in the supply chain and achieve accurate risk assessment and classification by constructing an automobile supply chain risk identification method based on knowledge graphs, using technologies such as entropy weight method, Z-score standardized preprocessing and random forest model. In particular, the ability to identify potential risks is enhanced by calculating and standardizing the entropy weight of multi-dimensional relationships, and the random forest model further improves the prediction accuracy and generalization ability of the model, thereby providing a more reliable risk warning mechanism for supply chain management.

[0065] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0066] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for identifying automobile supply chain risks based on knowledge graph, characterized in that: include: Obtain the entity types of the automobile supply chain and divide each entity type into a level of the automobile supply chain; each entity type contains multiple entities, and each two entities contain one or more relationships. Entities are used as nodes and relationships between entities as edges to construct a knowledge graph of the automobile supply chain; calculate nodes Risks to the automotive supply chain : ； In the formula, Representation Node To Node The edge rights, Representation Node In time The initial risk change at Representation and Node A collection of connected nodes, Representation Node The level where Representation Node In time The initial risk change at time Said , where Representation Node With Node No. The entropy weight of a relationship, Representation Node With Node No. The initial index value of the relationship, Representation Node With Node The total number of relationships between The initial risk change is the difference between the nodes at the current time point and the previous time point, and the initial indicator value is the correlation between the nodes; Said , where Representation Node With Node Between The information entropy of a relationship, Representation Node With Node The total number of relationships between Nodes with risks less than a set threshold are responded to as low risk, and nodes with risks greater than or equal to the set threshold are responded to as high risk; the initial risk change of each node and its corresponding risk level are input into the machine model for training to obtain a trained machine model; the initial risk change of the node to be detected is input into the trained machine model, and the risk level of the node to be detected is output.

2. The method for identifying automobile supply chain risks based on knowledge graph according to claim 1 is characterized in that: It also includes preprocessing the initial indicator values ​​using Z-score standardization.

3. The method for identifying automobile supply chain risks based on knowledge graph according to claim 1 is characterized in that: The entity types include: suppliers, manufacturers, distributors, retailers, logistics service providers and consumers.

4. The method for identifying automobile supply chain risks based on knowledge graph according to claim 3 is characterized in that: The supplier's entity includes: the supplier's financial status, main business and parts quality; the manufacturer's entity includes: the manufacturer's production capacity and product quality; The entities of distributors include: the distributor's distribution network, market coverage and logistics capabilities; the entities of retailers include: the retailer's sales and supply chain; the entities of logistics service providers include: the logistics service provider's transportation capacity, warehousing capacity and transportation costs; the entities of consumers include: the consumer's purchasing power, consumption habits and purchasing behavior.

5. The method for identifying automobile supply chain risks based on knowledge graph according to claim 1 is characterized in that: The relationships include: supply relationship, manufacturing relationship, distribution relationship, retail relationship, logistics relationship and purchasing relationship.

6. The method for identifying automobile supply chain risks based on knowledge graph according to claim 1 is characterized in that: The machine model is a random forest model.

Images