A supply chain asset financial risk monitoring method, system, device and medium

Abstract

This application relates to a method, system, device, and medium for monitoring financial risks of supply chain assets. The method includes: acquiring multi-source heterogeneous data of the target supply chain; fusing entity identifiers and relationships within the multi-source heterogeneous data to construct a supply chain knowledge graph; performing network topology feature analysis on key assets in the supply chain knowledge graph to obtain dynamic asset profile data; inputting the dynamic asset profile data into a hierarchical fusion assessment model for risk quantification to obtain preliminary risk score data; based on the preliminary risk score data and the supply chain knowledge graph, using the preliminary risk score data as the initial risk source, performing risk transmission simulation and iterative calculation on the supply chain knowledge graph to obtain supply chain transmission risk index data; and performing visual early warning and source tracing analysis on the supply chain transmission risk index data to obtain risk monitoring results. This method enables dynamic source tracing and intelligent risk monitoring of financial risks of supply chain assets.

Description

Technical Field

[0001] This invention belongs to the field of data processing technology, and in particular relates to a method, system, equipment and medium for monitoring financial risks of supply chain assets. Background Technology

[0002] With the development of supply chain finance technology, supply chain asset financial risk monitoring technology has emerged. This technology assesses the credit status and transaction behavior of each participant in the supply chain, providing financial institutions with risk warnings and decision support, thus leading to the current supply chain asset financial risk assessment methods.

[0003] Traditional technologies typically employ static credit scoring methods based on a single data source. These methods collect information such as a company's financial statements and transaction records, and then use statistical models or rule engines to conduct independent risk assessments of individual companies. The data from each stage is isolated from each other, and the assessment results only reflect the risk status at a specific point in time.

[0004] However, current methods for assessing financial risks in supply chain assets have the following problems: First, multi-source data is scattered across different business systems and supply chain links, lacking effective correlation between data, making it difficult to form a holistic understanding of the overall risks in the supply chain; second, the assessment method is mainly static, failing to capture the dynamic trends of risk changes in the supply chain network; and third, it only focuses on the independent risks of individual nodes, without considering the transmission effect of risks in the supply chain network. When a risk event occurs at a certain node, it is impossible to quantify its chain reaction on upstream and downstream related nodes, resulting in insufficient comprehensiveness and timeliness of risk monitoring. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, system, equipment, and medium for monitoring the financial risks of supply chain assets in response to the above-mentioned technical problems.

[0006] Firstly, this application provides a method for monitoring financial risks of supply chain assets, including:

[0007] S1. Acquire multi-source heterogeneous data of the target supply chain, integrate the entity identifiers and relationships in the multi-source heterogeneous data, and construct a supply chain knowledge graph; the multi-source heterogeneous data includes financial data, operational data, transaction chain data, and external environment data;

[0008] S2. Perform network topology feature analysis on key assets in the supply chain knowledge graph to obtain dynamic asset profile data;

[0009] S3. Input the dynamic asset profile data into the hierarchical fusion assessment model to quantify the risk and obtain preliminary risk score data;

[0010] S4. Based on the preliminary risk scoring data and the supply chain knowledge graph, using the preliminary risk scoring data as the initial risk source, perform risk transmission simulation and iterative calculation on the supply chain knowledge graph to obtain the supply chain transmission risk index data.

[0011] S5. Visualize and conduct source analysis on the supply chain transmission risk index data to obtain risk monitoring results; among which, risk monitoring results include risk warning information, risk root cause identification, and response suggestions.

[0012] In one embodiment, S2 includes:

[0013] S21. Based on the supply chain knowledge graph, calculate the network topology indicators for the entity nodes corresponding to key assets to obtain network correlation data; among which, the network correlation data includes degree centrality, proximity centrality and betweenness centrality.

[0014] S22. Based on financial and operational data from multi-source heterogeneous data, conduct a health assessment of key assets to obtain health index data.

[0015] S23. Extract the static attributes of key assets from multi-source heterogeneous data to obtain static attribute data;

[0016] S24. Integrate the network correlation data, health index data, and static attribute data to obtain dynamic asset profile data.

[0017] In one embodiment, the hierarchical fusion evaluation model includes a rule engine layer, a gradient boosting tree model layer, a weighted fusion layer, and a normalization layer, where S3 includes:

[0018] S31. Input the dynamic asset profile data into the rule engine layer, and perform risk filtering and marking on the assets through preset hard rule thresholds to obtain rule judgment result data;

[0019] S32. Input the dynamic asset profile data into the gradient boosting tree model layer to predict the default probability and obtain the model prediction score data.

[0020] S33. Input the rule determination result data and the model prediction score data into the weighted fusion layer for weighted fusion calculation to obtain preliminary comprehensive risk score data;

[0021] S34. Input the preliminary comprehensive risk score data into the normalization layer for normalization to obtain the preliminary risk score data.

[0022] In one embodiment, S4 includes:

[0023] S41. Based on the preliminary risk score data, map the preliminary risk score of each asset to the initial risk value of the corresponding node in the supply chain knowledge graph to obtain the initial risk data of the node.

[0024] S42. Based on the topology of the supply chain knowledge graph and the initial risk data of each node, using the initial risk data of each node as the propagation source, perform multi-round risk transmission iterative simulation calculations to obtain the iterative risk value data of each node; among which, the first... Nodes in round iteration The expression for updating the risk value is:

[0025]

[0026] In the formula, Represents a node In the Risk value after round of iterations Represents a node In the Risk value after round of iterations Indicates the iteration round number. Represents a node The set of neighboring nodes, Represents a node The neighbor node index, Represents a node To the node edge weights, Representing neighboring nodes In the Risk value after round of iterations This represents the preset risk propagation coefficient;

[0027] S43. Perform convergence judgment on the iterative risk value data of each node. When the change in risk value between two adjacent iterations is less than the preset convergence threshold, calculate the difference between the final iterative risk value and the initial risk value of each node to obtain the transmission risk added value data.

[0028] S44. Sum the initial risk data and the transmission risk added value data of the nodes to obtain the supply chain transmission risk index data.

[0029] In one embodiment, S5 includes:

[0030] S51. Compare and judge the supply chain transmission risk index data with the preset risk threshold to obtain risk warning information, and generate a risk propagation heat map based on the risk warning information;

[0031] S52. Based on the risk propagation heat map, reverse tracing is performed along the risk propagation path in the supply chain knowledge graph to locate the root cause node of the risk and obtain the risk root cause location.

[0032] S53. Based on the node type and risk attributes of the risk root cause nodes, generate response suggestions from the preset strategy knowledge base, integrate the risk warning information, risk root cause location and response suggestions to obtain risk monitoring results.

[0033] In one embodiment, S21 includes:

[0034] S211. Based on the supply chain knowledge graph, perform neighbor node statistics on the entity nodes corresponding to key assets to obtain vertex degree centrality; where the expression for vertex degree centrality is:

[0035]

[0036] In the formula, Represents a node degree centrality, Representing the first in the supply chain knowledge graph 1 node Represents a node The set of neighboring nodes, Represents a node The number of neighboring nodes, This represents the total number of nodes in the supply chain knowledge graph;

[0037] S212. Based on the supply chain knowledge graph, calculate the shortest path distance for the entity nodes corresponding to key assets to obtain proximity centrality; where the expression for proximity centrality is:

[0038]

[0039] In the formula, Represents a node Proximity centrality Representing the first in the supply chain knowledge graph 1 node This represents the total number of nodes in the supply chain knowledge graph. Representing the first in the supply chain knowledge graph 1 node Represents a node To the node The shortest path distance;

[0040] S213. Based on the supply chain knowledge graph, perform intermediate path statistics on the entity nodes corresponding to key assets to obtain the intermediary centrality; whereby the expression for intermediary centrality is:

[0041]

[0042] In the formula, Represents a node The centrality of the middle, Representing the first in the supply chain knowledge graph 1 node This represents the source node index in the supply chain knowledge graph. This represents the index of the target node in the supply chain knowledge graph. Represents a node To the node The total number of shortest paths, Represents a node To the node The shortest path passes through the nodes Quantity;

[0043] S214. Perform feature concatenation on degree centrality, proximity centrality, and betweenness centrality to obtain network correlation data.

[0044] In one embodiment, S1 includes:

[0045] S11. Perform entity recognition processing on multi-source heterogeneous data to obtain entity identification data;

[0046] S12. Perform relation extraction processing on multi-source heterogeneous data to obtain relational data;

[0047] S13. Merge and align entity identification data and relationship data to obtain entity-relation triplet data;

[0048] S14. Based on entity-relationship triple data, a supply chain knowledge graph is constructed.

[0049] Secondly, this application also provides a supply chain asset financial risk monitoring system, including:

[0050] The data acquisition and graph construction module is used to acquire multi-source heterogeneous data of the target supply chain, integrate the entity identifiers and relationships in the multi-source heterogeneous data, and construct a supply chain knowledge graph; the multi-source heterogeneous data includes financial data, operational data, transaction chain data and external environment data;

[0051] The dynamic asset profiling module is used to perform network topology feature analysis on key assets in the supply chain knowledge graph to obtain dynamic asset profiling data.

[0052] The risk quantification module is used to input dynamic asset profile data into the hierarchical fusion assessment model to quantify risk and obtain preliminary risk score data.

[0053] The risk transmission simulation module is used to perform risk transmission simulation and iterative calculation on the supply chain knowledge graph based on the preliminary risk scoring data and the supply chain knowledge graph, with the preliminary risk scoring data as the initial risk source, to obtain the supply chain transmission risk index data.

[0054] The early warning and traceability module is used to visualize and analyze the risk transmission index data of the supply chain, and obtain risk monitoring results. The risk monitoring results include risk warning information, risk root cause identification and response suggestions.

[0055] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.

[0056] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.

[0057] The aforementioned supply chain asset financial risk monitoring method, system, equipment, and medium effectively address the technical challenges of fragmented and unconnected multi-source data by constructing a supply chain knowledge graph. This enables deep integration and structured expression of multi-dimensional information, including financial data, transaction records, and public opinion. Furthermore, by extracting node centrality indicators through network topology feature analysis, it accurately identifies key nodes and weak links in the supply chain network, overcoming the limitations of traditional methods that focus solely on the risk of a single enterprise. The hierarchical fusion assessment model overcomes the bottleneck of static assessment, achieving a progressive dynamic assessment from basic indicators to comprehensive risk, improving the accuracy and timeliness of risk identification. The risk transmission iterative simulation mechanism fully considers the interconnected characteristics of the supply chain network, tracking the propagation path and cumulative effects of risks between nodes, addressing the core issue of existing technologies neglecting risk transmission effects. Finally, the visualization early warning and source tracing analysis transform abstract risk assessment results into intuitive visualizations, helping decision-makers quickly locate risk sources and formulate targeted response strategies. Overall, this achieves a comprehensive upgrade of supply chain financial risk monitoring from static isolation to dynamic correlation, and from single-point assessment to network collaboration. Attached Figure Description

[0058] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1This is a flowchart illustrating a supply chain asset financial risk monitoring method in one embodiment;

[0060] Figure 2 This is a schematic diagram of the structure of a supply chain asset financial risk monitoring system in a specific embodiment;

[0061] Figure 3 This is a flowchart illustrating a supply chain asset financial risk monitoring method in a specific embodiment.

[0062] Figure 4 This is a schematic diagram of the structure of a supply chain asset financial risk monitoring system in one embodiment. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0064] In one embodiment, reference Figure 1 The document presents a flowchart illustrating a supply chain asset financial risk monitoring method provided in this application. This embodiment uses the application of this method to a risk monitoring terminal (hereinafter referred to as the terminal) as an example. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0065] S1. Obtain multi-source heterogeneous data of the target supply chain, integrate the entity identifiers and relationships in the multi-source heterogeneous data, and construct a supply chain knowledge graph.

[0066] Optionally, multi-source heterogeneous data can include financial data, operational data, transaction chain data, and external environment data. Financial data can be used to reflect the financial status of each entity's assets, and can include asset and liability data and profit data of each entity. Operational data can be used to reflect the operational efficiency of each entity, and can include capacity utilization rate, order completion rate, etc. Transaction chain data can be used to characterize the closeness of the relationship between upstream and downstream entities, and can include transaction amount, transaction frequency, and performance status. External environment data can be used to reflect the external risk environment of the supply chain, and can include industry policies, market fluctuations, and macroeconomic indicators.

[0067] For example, the risk monitoring terminal can establish communication connections with the financial management system, operation management platform, and transaction settlement system of each participant in the target supply chain through preset data interfaces to obtain corresponding data. For external environment data, the risk monitoring terminal can establish authorized communication with the official API interfaces of industry databases, policy release platforms, and macroeconomic monitoring websites through official data source API interface calling technology to obtain corresponding external environment data, ensuring the comprehensiveness and real-time nature of data acquisition.

[0068] For example, the risk monitoring terminal can use an entity alignment algorithm to match, deduplicate, and unify different identifiers representing the same entity from different data sources. At the same time, it can use a relation extraction algorithm to extract the relationships between entities, such as cooperation and supply and demand. The merged entity identifiers and relationships are organized into a standardized format that conforms to the graph database storage specifications. Based on graph database technology, a supply chain knowledge graph is constructed. The supply chain knowledge graph uses entities as nodes and relationships as edges, which can clearly present the relationship network of various entities and assets in the supply chain.

[0069] S2. Perform network topology feature analysis on key assets in the supply chain knowledge graph to obtain dynamic asset profile data.

[0070] For example, critical assets refer to assets in the supply chain that have a decisive impact on overall operations and financial security. Critical assets may include core business assets of core enterprises, core assets of key upstream and downstream nodes, etc., and can be screened and determined by the risk monitoring terminal based on preset asset importance assessment criteria. Network topology feature analysis can analyze the location, correlation strength, and influence range of critical assets in the supply chain network by extracting the topological structure information of nodes in the knowledge graph. Network topology feature analysis can be based on graph theory, reflecting the network characteristics of critical assets by quantifying the connection relationships between nodes. The risk monitoring terminal can first screen out the entity nodes corresponding to critical assets in the supply chain knowledge graph, and then, through topology feature extraction algorithms combined with the real-time update characteristics of multi-source heterogeneous data, comprehensively analyze the network correlation status and dynamic change trends of critical assets. It integrates the network characteristics, health status, and inherent attributes of critical assets to generate dynamic asset profile data that can reflect the status of critical assets in real time.

[0071] S3. Input the dynamic asset profile data into the hierarchical fusion assessment model to quantify the risk and obtain preliminary risk score data.

[0072] For example, the hierarchical fusion assessment model can be a multi-dimensional risk assessment model, which may include a rule engine layer, a gradient boosting tree model layer, a weighted fusion layer, and a normalization layer. The risk monitoring terminal can first standardize the dynamic asset profile data to eliminate the differences in the dimensions of different feature dimensions, and then input it into the rule engine layer to perform preliminary risk filtering and labeling of assets through preset hard rule thresholds; then, the data is input into the gradient boosting tree model layer for default probability prediction to obtain model prediction score data; finally, through the processing of the weighted fusion layer and the normalization layer, standardized preliminary risk score data is output; wherein, the preliminary risk score data can be used to reflect the individual risk level of each key asset.

[0073] S4. Based on the preliminary risk scoring data and the supply chain knowledge graph, using the preliminary risk scoring data as the initial risk source, perform risk transmission simulation and iterative calculation on the supply chain knowledge graph to obtain the supply chain transmission risk index data.

[0074] For example, preliminary risk score data can serve as the initial risk source, corresponding to the initial risk level of each key asset node in the supply chain knowledge graph. The risk transmission simulation iteration calculation can be based on complex network propagation theory, simulating the transmission process of risk between nodes in the supply chain, considering the impact of the strength of the association between nodes on risk transmission. The risk monitoring terminal can first map the preliminary risk score of each asset to the initial risk value of the corresponding node in the supply chain knowledge graph, determining the risk transmission path and transmission weight between nodes. The edge weights can be determined by the degree of association between nodes in the supply chain knowledge graph. Subsequently, multiple rounds of risk transmission simulation calculations are performed according to a preset iterative formula. Through multiple rounds of iterative calculations, a complete simulation of risk transmission in the supply chain network can be achieved, thereby obtaining the supply chain transmission risk index data for each node.

[0075] S5. Visualize and conduct source analysis on the supply chain transmission risk index data to obtain risk monitoring results.

[0076] Optionally, risk monitoring results may include risk warning information, risk source identification, and response recommendations.

[0077] For example, visual early warning can be achieved by using data visualization technology to display supply chain transmission risk index data in the form of heat maps, line graphs, etc. The risk monitoring terminal can preset risk warning thresholds of different levels, compare the transmission risk index of each node with the warning threshold, and generate corresponding risk warning information. The risk warning information can be used to clearly indicate high-risk, medium-risk, and low-risk nodes. Source tracing analysis can be performed by the risk monitoring terminal based on the relationship of the supply chain knowledge graph, starting from a high-risk node and tracing back along the risk transmission path to locate the root cause node of the risk, clarifying the cause of the risk. The cause of the risk may include abnormal financial conditions of the node itself, poor operation and management, or external environmental influences. The risk monitoring terminal has a built-in preset strategy knowledge base that can store response plans corresponding to different types of risk root causes. Based on the risk root cause location results, the risk monitoring terminal can match corresponding response suggestions from the strategy knowledge base. By integrating risk warning information, risk root cause location, and response suggestions, a complete risk monitoring result can be obtained, providing accurate risk decision support for relevant parties.

[0078] The aforementioned supply chain asset financial risk monitoring method effectively addresses the problems of isolated data, static assessment, and inability to quantify dynamic transmission in traditional risk monitoring by integrating multi-source heterogeneous data and constructing a supply chain knowledge graph. It comprehensively integrates financial, operational, transactional, and external environmental information to generate a global risk view. Employing network topology features and dynamic asset profiling elevates risk assessment from independent analysis of a single enterprise to network-based analysis of the supply chain, enhancing the comprehensiveness and accuracy of risk identification. Risk quantification through a hybrid model combining rule engines and layered machine learning improves assessment reliability. Dynamic simulation of risk transmission based on knowledge graphs and network propagation models quantifies the scope and intensity of risk diffusion, enabling proactive early warning. By relying on visualization and path tracing to automatically locate the root causes of risks and generate response suggestions, it transforms passive early warning into proactive handling, enhancing the systematic, dynamic, and intelligent nature of supply chain asset financial risk monitoring and improving risk management efficiency and decision support capabilities.

[0079] To further illustrate the solutions of this application, a specific embodiment is described below in conjunction with the above-mentioned supply chain asset financial risk monitoring method. Figure 2 The diagram shown is a structural schematic of the supply chain asset financial risk monitoring system 10, which includes:

[0080] The data acquisition and graph construction module 11 is used to collect multi-dimensional data and construct a supply chain knowledge graph.

[0081] The asset profiling module 12 is used to calculate and generate dynamic asset profiles containing specific network topology indicators based on the supply chain knowledge graph.

[0082] The hybrid risk quantification module 13 adopts a layered architecture, which integrates a rule engine component, a machine learning model component, and a network propagation calculation unit in sequence to calculate various risk indicators.

[0083] The early warning and report generation module 14 is used to realize risk visualization, early warning push and intelligent report generation.

[0084] To further illustrate the solutions of this application, a specific embodiment applied to the aforementioned supply chain asset financial risk monitoring system 10 will be described below, such as... Figure 3 The diagram shows a flowchart of a supply chain asset financial risk monitoring method, which includes the following steps:

[0085] S101: Collect and integrate multi-dimensional data of core enterprises and their related upstream and downstream enterprises in the target supply chain network; the multi-dimensional data includes at least financial data, operational data, transaction chain data and external environment data; based on the entity identification and relationship in the multi-dimensional data, construct a dynamically updated supply chain knowledge graph;

[0086] Specifically, the system collects data through APIs, direct database connections, and other methods, and constructs a supply chain knowledge graph. The graph uses enterprises as nodes and relationships such as transactions and equity as edges.

[0087] S102: Based on the supply chain knowledge graph, construct dynamic asset profiles for key assets in the supply chain network; the dynamic asset profiles include the asset's static attributes, health indicators calculated based on real-time data, and specific network connectivity indicators determined based on its network topology position in the supply chain knowledge graph; the specific network connectivity indicators include at least one or more of vertex degree centrality, proximity centrality, and betweenness centrality.

[0088] Specifically, the system constructs dynamic profiles for assets (such as accounts receivable) in the knowledge graph. A key improvement lies in the concretization of network connectivity metrics into computable topological indicators:

[0089] (1) Degree centrality: Calculate the number of trading partners directly connected to the enterprise to which the asset belongs, reflecting its direct influence.

[0090] (2) Proximity centrality: Calculate the reciprocal of the average shortest path distance from the firm to all other firms in the graph, reflecting the efficiency of its information or risk propagation.

[0091] (3) Intermediation centrality: Calculate the number of times the firm is located on the shortest path between other firms in the graph, reflecting its importance as a "hub" or "bottleneck".

[0092] For example, a debtor with high "intermediation centrality" of an account receivable means that it occupies a key bridging position in the supply chain network, and its default may trigger a wider chain reaction.

[0093] S103: Based on dynamic asset profiling, a hybrid assessment model combining a rule engine and a machine learning model is used to quantify risk, resulting in an asset liquidity risk score, a related entity credit risk score, and a supply chain transmission risk index. The hybrid assessment model employs a hierarchical decision-making logic: first, the rule engine filters and labels risks based on preset hard rules; second, the machine learning model makes probability predictions on the filtered assets; and finally, the rule outputs and model predictions are weighted and fused to form a preliminary comprehensive risk score. The supply chain transmission risk index is obtained by using a supply chain knowledge graph as a risk propagation network and the preliminary comprehensive risk score as the initial risk source, through iterative simulation calculations using a network propagation model.

[0094] Specifically, the layered and integrated architecture used in the risk quantification process includes:

[0095] 1. Rule Engine Layer: Executes hard rule judgments. For example, if the rule "If the counterparty is on the national list of dishonest persons subject to enforcement, it is marked as high risk" is triggered, the asset will receive an extremely high base risk score or be directly blocked.

[0096] 2. Machine Learning Model Layer: For assets that are not directly identified as extremely high-risk at the rule layer, their dynamic profiles (including the aforementioned network indicators) are input into trained models such as Gradient Boosting Tree (GBDT) to predict their risk scores, such as default probability.

[0097] 3. Fusion Layer: The Boolean labels or ratings output by the rule layer and the probability ratings output by the model layer are weighted and calculated using preset weights to form a preliminary comprehensive risk score for the asset.

[0098] 4. Network Propagation Simulation Layer: The preliminary comprehensive risk scores of each asset / enterprise obtained in the previous step are used as the initial "infection intensity" and loaded onto the corresponding nodes in the knowledge graph. Multiple iterative simulations are conducted using propagation algorithms such as the Susceptibility-Infection (SI) model: In each simulation, high-risk nodes will "infect" their neighboring nodes with a certain probability, increasing the risk values ​​of the neighboring nodes. After the simulation converges, each node receives a transmission risk bonus. Ultimately, the supply chain transmission risk index can be derived by combining the initial score of the target asset node with its transmission bonus.

[0099] S104: Based on the risk quantification results, perform visual early warning, and when the risk level exceeds the threshold, perform source analysis based on the risk propagation path simulated by the supply chain knowledge graph and network propagation model, and automatically generate a risk report containing the root cause of the risk and response suggestions.

[0100] Specifically, the system provides visualized early warnings and intelligent reporting. When the supply chain transmission risk index of an asset exceeds the standard, the system automatically traces back along the "infection path" simulated in S103, locating the initial and most impactful risk source node in the knowledge graph. The reporting engine combines the type of the source node (e.g., "second-tier raw material supplier") and the cause of the risk (e.g., "operational anomaly") to generate specific suggestions from the strategy library, such as "It is recommended to check the list of alternative suppliers and increase the diversity of procurement channels."

[0101] The aforementioned supply chain asset financial risk monitoring method is applied to the supply chain asset financial risk monitoring system. By integrating multi-dimensional data to construct a knowledge graph, it introduces quantifiable network topology indicators (such as betweenness centrality) into asset profiling for the first time, elevating risk assessment from isolated attribute analysis to network structure analysis, providing crucial structural input for understanding risk transmission. A hierarchical hybrid assessment architecture of "rule hard filtering - model soft prediction - result weighted fusion" is adopted. This architecture retains the clarity and strong interpretability of business rules while leveraging the ability of machine learning models to handle complex nonlinear relationships, making the risk quantification process more robust and credible. The initial risk score is used as the "source of infection," and dynamic propagation simulation (such as using the SI model) is performed on the real topology of the knowledge graph to quantify the potential scope and intensity of risk diffusion along the supply chain, achieving advanced simulation and early warning of systemic risks. A visual interface combined with a transmission heatmap intuitively displays the source and diffusion path of risks. The system can automatically trace back to the root risk point based on the simulated path and generate targeted reports, elevating traditional risk alarms to solutions with diagnostic and decision support value.

[0102] In an optional embodiment, S2 includes:

[0103] S21. Based on the supply chain knowledge graph, calculate the network topology indicators for the entity nodes corresponding to key assets to obtain network correlation data.

[0104] Optionally, network connectivity data includes degree centrality, proximity centrality, and betweenness centrality.

[0105] Optionally, network topology metrics can be calculated based on graph theory principles, reflecting the network connectivity of key assets by quantifying the positional characteristics of nodes in the supply chain knowledge graph. Degree centrality measures the number of direct connections between a node and other nodes; proximity centrality measures the ease with which a node can reach all other nodes; and betweenness centrality measures the mediating role of a node in the shortest paths between other nodes. The risk monitoring terminal can calculate these three metrics using corresponding algorithms, and then perform feature concatenation to obtain network connectivity data.

[0106] S22. Based on financial and operational data from multi-source heterogeneous data, conduct a health assessment of key assets to obtain health index data.

[0107] Optionally, financial data can be derived from financial statements, cash flow, and debt information of various participants in the supply chain. Operational data can be derived from production plan execution, order fulfillment rates, and inventory turnover data of each participant. The risk monitoring terminal can acquire these two types of data in real time through a data interface and preprocess them to eliminate data noise and outliers. Health assessment can be achieved by analyzing asset profitability and solvency reflected in financial data, combined with asset operational efficiency reflected in operational data, to construct a multi-dimensional health assessment system and comprehensively evaluate the current health status of key assets. The risk monitoring terminal can use a preset assessment algorithm to extract and quantify features from financial and operational data, converting various data into unified-dimensional assessment indicators. These indicators are then integrated through weighted summation and other methods to obtain health indicator data that reflects the current health status of key assets. This health indicator data can dynamically reflect the operational and financial health level of key assets.

[0108] S23. Extract the static attributes of key assets from multi-source heterogeneous data to obtain static attribute data.

[0109] Optionally, static attribute data refers to the inherent attribute information of key assets that does not change significantly over time or through transactions. Distinguished from dynamically changing network correlation data and health indicator data, static attribute data can reflect the inherent characteristics of key assets. The risk monitoring terminal can use a preset attribute extraction algorithm to filter relevant data conforming to the definition of static attributes from multi-source heterogeneous data. Specifically, this can include the entity type corresponding to the key asset, asset ownership information, asset type, core qualification certifications, and registration information. Entity types can include core enterprises, suppliers, and distributors, and asset types can include fixed assets and current assets. The risk monitoring terminal can standardize the extracted static attribute data, unifying data formats and expression standards, eliminating differences in the expression of static attributes from different data sources, and ensuring the consistency and accuracy of the static attribute data.

[0110] S24. Integrate the network correlation data, health index data, and static attribute data to obtain dynamic asset profile data.

[0111] Optionally, the risk monitoring terminal can first standardize the network correlation data, health index data, and static attribute data to eliminate the differences in the dimensions of different data and avoid the excessive influence of a certain type of data on the overall characteristics. Then, a feature splicing algorithm is used to orderly integrate the degree centrality, proximity centrality, and betweenness centrality in the network correlation data, various health assessment indicators in the health index data, and various inherent attributes in the static attribute data to generate a feature vector in a unified format. At the same time, combined with the real-time update characteristics of multi-source heterogeneous data, the integrated feature data is dynamically calibrated to ensure that the feature data can reflect the status changes of key assets in real time, thereby obtaining dynamic asset profile data.

[0112] In an optional embodiment, the hierarchical fusion evaluation model includes a rule engine layer, a gradient boosting tree model layer, a weighted fusion layer, and a normalization layer, where S3 includes:

[0113] S31. Input the dynamic asset profile data into the rule engine layer, and perform risk filtering and marking on the assets through preset hard rule thresholds to obtain the rule judgment result data.

[0114] Optionally, the rule engine layer is the foundational layer of the layered fusion assessment model. Based on preset risk rules, the rule engine layer performs preliminary screening and risk labeling of dynamic asset profile data. The preset hard rule thresholds are formulated based on industry risk management standards, supply chain operation norms, and historical risk cases. These hard rule thresholds can cover risk assessment standards for key assets across core dimensions such as finance and operations. They can be used to quickly identify clearly high-risk and risk-free assets. The risk monitoring terminal can compare each characteristic indicator in the dynamic asset profile data with the preset hard rule thresholds one by one. Assets that meet the high-risk rule are marked as high-risk, those that meet the risk-free rule are marked as risk-free, and the remaining assets are marked as pending assessment. The rule judgment basis for each asset is recorded, thus obtaining the rule judgment result data.

[0115] S32. Input the dynamic asset profile data into the gradient boosting tree model layer to predict the probability of default and obtain the model prediction score data.

[0116] Optionally, the gradient boosting tree model layer is the core prediction layer of the hierarchical fusion evaluation model. The gradient boosting tree model layer is constructed using the gradient boosting decision tree (GBDT) algorithm. The risk monitoring terminal can first divide the dynamic asset profile data into feature input vectors and input them into the trained gradient boosting tree model. The model predicts the default probability of financial risks of key assets by weighting and combining each feature. Then, the default probability is converted into standardized model prediction score data. The model prediction score data can reflect the individual risk level of key assets and make up for the deficiency of the rule engine layer, which can only make qualitative judgments.

[0117] S33. Input the rule determination result data and model prediction score data into the weighted fusion layer for weighted fusion calculation to obtain preliminary comprehensive risk score data.

[0118] Optionally, the risk monitoring terminal can use the Analytic Hierarchy Process (AHP) to perform a weighted summation of the scores corresponding to the rule judgment results and the model prediction scores according to the preset weight allocation standards. The weight coefficients of the rule judgment results and the model prediction scores are preset according to the focus of risk assessment to ensure that the fused results can reflect both the risk characteristics under the hard rules and the accuracy of the model predictions, and finally obtain preliminary comprehensive risk score data.

[0119] S34. Input the preliminary comprehensive risk score data into the normalization layer for normalization to obtain the preliminary risk score data.

[0120] Optionally, the risk monitoring terminal can use a linear normalization algorithm to take the maximum and minimum values ​​in the preliminary comprehensive risk score data as reference benchmarks, perform linear transformation on all score data, eliminate the influence of different feature dimensions on the dimensionality, ensure that the preliminary risk scores of each key asset are comparable, and avoid the interference of extreme values ​​on subsequent risk transmission calculations, and finally output standardized preliminary risk score data.

[0121] In an optional embodiment, S4 includes:

[0122] S41. Based on the preliminary risk score data, map the preliminary risk score of each asset to the initial risk value of the corresponding node in the supply chain knowledge graph to obtain the initial risk data of the node.

[0123] Optionally, preliminary risk scoring data can be used to reflect the initial risk level of each key asset. Node initial risk data can be a concrete mapping of the preliminary risk scoring data in a supply chain knowledge graph. This mapping principle can be based on the one-to-one correspondence between entity nodes and key assets in the supply chain knowledge graph, directly associating the preliminary risk score of each key asset with its corresponding entity node in the knowledge graph, serving as the initial risk benchmark for that node. The risk monitoring terminal can establish the association between preliminary risk scoring data and entity nodes in the supply chain knowledge graph through a preset mapping algorithm, ensuring that each entity node has a corresponding initial risk value. For nodes not included in the preliminary risk scoring scope, a preset initial risk value is used as the base value, thereby generating complete node initial risk data.

[0124] S42. Based on the topology of the supply chain knowledge graph and the initial risk data of the nodes, and using the initial risk data of the nodes as the propagation source, perform multi-round risk transmission iterative simulation calculations to obtain the iterative risk value data of each node.

[0125] For example, the risk monitoring terminal can perform multi-round iterative simulation calculations of risk transmission based on the topology of the supply chain knowledge graph and the initial risk data of the nodes, using the initial risk data of the nodes as the propagation source, to obtain the iterative risk value data of each node. The topology of the supply chain knowledge graph can be used to characterize the path and intensity of risk transmission, and the edge weights between nodes can be used to reflect the efficiency of risk transmission between related nodes. The edge weights can be determined by factors such as the closeness of transactions between nodes, the length of cooperation, and the degree of dependence. The iterative simulation calculation of risk transmission simulates the transmission process of risk in the supply chain network. In each iteration, the risk value of each node is updated by being affected by the risk values ​​of its neighboring nodes, and the definitions of each parameter are consistent with those described above.

[0126] Optionally, the risk monitoring terminal can calculate the risk value of each node in each iteration according to the expression for updating the risk value, record the iteration result of each round, until the convergence condition is met, thereby obtaining the iterative risk value data of each node. Wherein, the... Nodes in round iteration The expression for updating the risk value can be:

[0127]

[0128] In the above expression, Represents a node In the Risk value after round of iterations Represents a node In the Risk value after round of iterations Indicates the iteration round number. Represents a node The set of neighboring nodes, Represents a node The neighbor node index, Represents a node To the node edge weights, Representing neighboring nodes In the Risk value after round of iterations This represents the preset risk propagation coefficient.

[0129] S43. Perform convergence judgment on the iterative risk value data of each node. When the change in risk value between two adjacent iterations is less than the preset convergence threshold, calculate the difference between the final iterative risk value and the initial risk value of each node to obtain the transmission risk added value data.

[0130] Optionally, convergence judgment can be used to ensure that the risk transmission simulation reaches a stable state and avoid meaningless iterative calculations. The preset convergence threshold can be formulated based on the accuracy requirements of risk monitoring and can be used to determine whether the iterative process has reached stability. The risk monitoring terminal can calculate the absolute difference of the risk values ​​of each node in two adjacent iterations and compare this difference with the preset convergence threshold. If the change in the risk values ​​of all nodes is less than the preset convergence threshold, the iteration is determined to be converged and the calculation is stopped. The terminal also calculates the difference between the final iterative risk value and the initial risk value of each node. This difference is the transmission risk added value data, which can be used to reflect the additional risk level generated by the transmission of supply chain risks at each node.

[0131] S44. Sum the initial risk data and the transmission risk added value data of the nodes to obtain the supply chain transmission risk index data.

[0132] Optionally, the supply chain transmission risk index data is a comprehensive reflection of the initial risk of each node and the additional risk brought about by transmission, which can comprehensively reflect the overall risk level of each node after the transmission of supply chain risks. The risk monitoring terminal can sum the initial risk data and the transmission risk added value data of each node one by one through a preset summation algorithm to ensure that the risk index of each node can accurately reflect the combined impact of its own risk and transmission risk. At the same time, the summation results are standardized to unify the data format, and finally, complete supply chain transmission risk index data can be generated.

[0133] In an optional embodiment, S5 includes:

[0134] S51. Compare and judge the supply chain transmission risk index data with the preset risk threshold to obtain risk warning information, and generate a risk propagation heat map based on the risk warning information.

[0135] Optionally, preset risk thresholds can be based on industry risk management standards and supply chain operation objectives, and are divided into different risk levels to distinguish high, medium, and low-risk nodes. The risk monitoring terminal can compare the supply chain transmission risk index of each node with the preset risk thresholds one by one. Based on the comparison results, it determines the risk level of each node, generates corresponding risk warning information, and clearly marks the risk level, risk index, and warning priority of each node. The risk propagation heatmap can be generated through data visualization technology, based on a supply chain knowledge graph, and uses different color gradients to intuitively present the risk level of each node. The darker the color, the higher the risk level. It also marks the risk transmission path between nodes, facilitating a clear understanding of the overall risk distribution and propagation in the supply chain, and providing relevant personnel with intuitive risk references.

[0136] S52. Based on the risk propagation heat map, reverse tracing is performed along the risk propagation path in the supply chain knowledge graph to locate the root cause node of the risk and obtain the risk root cause location.

[0137] Optionally, risk root cause localization can be based on the relationships within a supply chain knowledge graph. Starting from high-risk nodes, it traces the source of risk transmission backward to identify the initial node and core cause of the risk in the entire supply chain. The risk monitoring terminal can use high-risk nodes in the risk propagation heatmap as a starting point and, through a pre-set tracing algorithm, progressively query backward along the risk transmission path to track the risk source of each high-risk node, investigate the risk transmission impact of its upstream related nodes, until it finds the root cause node that has not been affected by the transmission from other nodes and has generated its own initial risk. At the same time, it records the transmission process of risk from the root cause node to each high-risk node, clarifying the type and specific cause of the risk root cause, thereby obtaining a complete risk root cause localization result.

[0138] S53. Based on the node type and risk attributes of the risk root cause nodes, generate response suggestions from the preset strategy knowledge base, integrate the risk warning information, risk root cause location and response suggestions to obtain risk monitoring results.

[0139] Optionally, the preset strategy knowledge base can be pre-stored in the risk monitoring terminal. This knowledge base can cover different types of risk root causes and corresponding response plans for different node types. It can include risk prevention and control measures, emergency response procedures, and optimization suggestions. The data in the strategy knowledge base can come from industry best practices, risk management cases, and expert experience summaries. By extracting the node type and risk attributes of the risk root cause nodes, the risk monitoring terminal can use keyword matching algorithms to match corresponding response suggestions in the preset strategy knowledge base, ensuring that the suggestions are targeted and actionable. Subsequently, the risk warning information, risk root cause location results, and response suggestions are integrated to generate standardized risk monitoring results, providing precise decision support for supply chain risk management.

[0140] In an optional embodiment, S21 includes:

[0141] S211. Based on the supply chain knowledge graph, perform neighbor node statistics on the entity nodes corresponding to key assets to obtain the degree centrality of the nodes.

[0142] Alternatively, the expression for vertex degree centrality can be:

[0143]

[0144] For example, in the above expression for vertex degree centrality, Represents a node degree centrality, Representing the first in the supply chain knowledge graph 1 node Represents a node The set of neighboring nodes, that is, the set of nodes with which the node is located. All nodes that have a direct relationship Represents a node The number of neighboring nodes, This represents the total number of nodes in the supply chain knowledge graph.

[0145] For example, the risk monitoring terminal can traverse the supply chain knowledge graph, count the number of neighboring nodes of each key asset's corresponding entity node, and then calculate the degree centrality of each node according to the above expression for degree centrality. The higher the degree centrality value, the closer the direct connection between the node and other nodes, and the stronger its direct influence in the supply chain network.

[0146] S212. Based on the supply chain knowledge graph, calculate the shortest path distance for the entity nodes corresponding to key assets to obtain proximity centrality.

[0147] Alternatively, the expression for proximity centrality can be:

[0148]

[0149] In the above expressions for near-centrality, Represents a node Proximity centrality Representing the first in the supply chain knowledge graph 1 node This represents the total number of nodes in the supply chain knowledge graph. Representing the first in the supply chain knowledge graph 1 node Represents a node To the node The shortest path distance is the path length with the fewest edges among the associated paths between two nodes. Proximity centrality can be used to measure how easily a node can reach all other nodes in a supply chain knowledge graph; proximity centrality reflects the reachability of a node in the network.

[0150] For example, the risk monitoring terminal can use the shortest path algorithm to calculate the shortest path distance from the node corresponding to the key asset to all other nodes, and then substitute it into the above expression for proximity centrality to calculate the proximity centrality. The higher the proximity centrality value, the stronger the reachability of the node in the supply chain network, and the faster it can receive and transmit information and risks.

[0151] S213. Based on the supply chain knowledge graph, perform intermediary path statistics on the entity nodes corresponding to key assets to obtain intermediary centrality.

[0152] Alternatively, the expression for betweenness centrality can be:

[0153]

[0154] In the above expression of betweenness centrality, Represents a node The centrality of the middle, Representing the first in the supply chain knowledge graph 1 node This represents the source node index in the supply chain knowledge graph. This represents the index of the target node in the supply chain knowledge graph. Represents a node To the node The total number of shortest paths, Represents a node To the node The shortest path passes through the nodes The number of nodes. Betweenness centrality can be used to measure the mediating role of a node in the shortest path between other nodes. Betweenness centrality can reflect a node's ability to control the transmission of information and risk in the supply chain network.

[0155] For example, the risk monitoring terminal, based on the supply chain knowledge graph, performs mediation path statistics on the entity nodes corresponding to key assets to obtain mediation centrality. The risk monitoring terminal can calculate mediation centrality by statistically analyzing the shortest paths between all source nodes and target nodes, as well as the number of paths passing through the node corresponding to the key asset, and substituting these into the aforementioned mediation centrality expression. A higher mediation centrality value indicates a stronger mediation role for the node in the supply chain network and a stronger ability to control the transmission of risks and information.

[0156] S214. Perform feature concatenation on degree centrality, proximity centrality, and betweenness centrality to obtain network correlation data.

[0157] Optionally, the risk monitoring terminal can first standardize the three indicators—degree centrality, proximity centrality, and betweenness centrality—to eliminate differences in dimensions and prevent any single indicator from interfering with the overall data due to variations in numerical range. Then, a feature concatenation algorithm can be used to combine the three indicators in a predetermined order, generating network correlation data containing features across these three dimensions. This network correlation data can comprehensively reflect the network position, correlation strength, and mediating role of key asset nodes within the supply chain knowledge graph.

[0158] In an optional embodiment, S1 includes:

[0159] S11. Perform entity recognition processing on multi-source heterogeneous data to obtain entity identification data.

[0160] Optionally, entity recognition refers to identifying entities with independent meaning from multi-source heterogeneous data. Entities may include core enterprises, upstream and downstream enterprises, key assets, and trading entities in the supply chain. The risk monitoring terminal can identify various entities from financial data, operational data, transaction chain data, and external environment data and assign them unique entity identifiers based on named entity recognition algorithms in natural language processing, through a preset entity type dictionary and machine learning model. This ensures that the identifier of the same entity is consistent across different data sources, thereby obtaining entity identifier data.

[0161] S12. Perform relation extraction processing on multi-source heterogeneous data to obtain relational data.

[0162] Optionally, relationship extraction refers to extracting the relationships between different entities from multi-source heterogeneous data. These relationships can include cooperative relationships, supply and demand relationships, transaction relationships, and subordinate relationships. For different types of multi-source heterogeneous data, the risk monitoring terminal can adopt corresponding relationship extraction strategies. For example, it can extract transaction relationships between enterprises from transaction chain data, ownership relationships between enterprises and assets from financial data, and cooperative relationships between enterprises and suppliers from operational data. The extracted relationships can be standardized and labeled to clarify the relationship type and strength, thereby generating structured relationship data.

[0163] S13. Merge and align entity identification data and relationship data to obtain entity-relationship triplet data.

[0164] Optionally, entity identifier fusion and alignment can be used to address inconsistencies in the identifiers of the same entity across different data sources. For example, a company might use its full name in financial data but its abbreviation in transaction chain data. The risk monitoring terminal can use an entity alignment algorithm to match and unify entity identifiers from different data sources, ensuring that the same entity corresponds to a unique entity identifier. For instance, the risk monitoring terminal can use preset fusion rules to clean, match, and integrate entity identifier data and relationship data, combining entities, relationships, and relationship attributes to form entity-relationship triples (i.e., entity 1 - relationship - entity 2), ultimately resulting in standardized entity-relationship triple data.

[0165] S14. Based on entity-relationship triple data, a supply chain knowledge graph is constructed.

[0166] Optionally, a supply chain knowledge graph can be a graph-based knowledge representation, using entities as nodes and relationships as edges to clearly present the network of connections between entities and assets in the supply chain. The risk monitoring terminal can use graph database technology to store and organize entity-relationship triplet data according to a graph structure, generating a visualized network topology. For example, the risk monitoring terminal can use standardized entity-relationship triplet data input, through a graph generation algorithm, to map entities to nodes in the graph and relationships to edges between nodes. Simultaneously, relevant attribute information from multi-source heterogeneous data can be associated with corresponding nodes and edges, enriching the content of the knowledge graph. Ultimately, a supply chain knowledge graph that comprehensively reflects the entities, assets, and relationships within the supply chain can be constructed.

[0167] The aforementioned supply chain asset financial risk monitoring method effectively addresses the problems of isolated data, static assessment, and inability to quantify transmission in traditional risk monitoring by integrating multi-source data and dynamically constructing a supply chain knowledge graph. It comprehensively integrates financial, operational, transactional, and external environmental information to generate a global risk view. Employing network topology features and dynamic asset profiling elevates risk assessment from independent analysis of a single enterprise to network-based analysis of the supply chain, enhancing the comprehensiveness and accuracy of risk identification. A hybrid model integrating rule engines and machine learning is used for risk quantification, balancing rule rigor with model prediction accuracy, resulting in more stable and reliable assessment results. Risk transmission simulation using a knowledge graph network quantifies the scope and intensity of risk diffusion, enabling proactive early warning. Visualization and path tracing automatically locate the root causes of risks and generate response suggestions, transforming passive early warning into proactive handling. This enhances the systematic, dynamic, and intelligent level of supply chain asset financial risk monitoring, providing reliable support for risk management.

[0168] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0169] Based on the same inventive concept, this application also provides a supply chain asset financial risk monitoring system for implementing the aforementioned supply chain asset financial risk monitoring method. The solution provided by this system is similar to the implementation scheme described in the above method; therefore, the specific limitations of one or more embodiments of the supply chain asset financial risk monitoring system provided below can be found in the limitations of the supply chain asset financial risk monitoring method described above, and will not be repeated here.

[0170] In one exemplary embodiment, such as Figure 4 As shown, a schematic diagram of the structure of a supply chain asset financial risk monitoring system 20 is provided, including:

[0171] The data acquisition and graph construction module 21 is used to acquire multi-source heterogeneous data of the target supply chain, integrate the entity identifiers and relationships in the multi-source heterogeneous data, and construct a supply chain knowledge graph; wherein, the multi-source heterogeneous data includes financial data, operational data, transaction chain data and external environment data;

[0172] The dynamic asset profile construction module 22 is used to perform network topology feature analysis on key assets in the supply chain knowledge graph to obtain dynamic asset profile data.

[0173] Risk quantification module 23 is used to input dynamic asset profile data into the hierarchical fusion assessment model for risk quantification and obtain preliminary risk score data;

[0174] The risk transmission simulation module 24 is used to perform risk transmission simulation and iterative calculation on the supply chain knowledge graph based on the preliminary risk scoring data and the supply chain knowledge graph, with the preliminary risk scoring data as the initial risk source, to obtain the supply chain transmission risk index data.

[0175] The early warning and traceability module 25 is used to perform visual early warning and traceability analysis on the supply chain transmission risk index data to obtain risk monitoring results; among which, the risk monitoring results include risk early warning information, risk root cause identification and response suggestions.

[0176] Furthermore, the dynamic asset profile building module 22 can also be used for:

[0177] S21. Based on the supply chain knowledge graph, calculate the network topology indicators for the entity nodes corresponding to key assets to obtain network correlation data; among which, the network correlation data includes degree centrality, proximity centrality and betweenness centrality.

[0178] S22. Based on financial and operational data from multi-source heterogeneous data, conduct a health assessment of key assets to obtain health index data.

[0179] S23. Extract the static attributes of key assets from multi-source heterogeneous data to obtain static attribute data;

[0180] S24. Integrate the network correlation data, health index data, and static attribute data to obtain dynamic asset profile data.

[0181] Furthermore, the hierarchical fusion evaluation model includes a rule engine layer, a gradient boosting tree model layer, a weighted fusion layer, and a normalization layer. The risk quantification module 23 can also be used for:

[0182] S31. Input the dynamic asset profile data into the rule engine layer, and perform risk filtering and marking on the assets through preset hard rule thresholds to obtain rule judgment result data;

[0183] S32. Input the dynamic asset profile data into the gradient boosting tree model layer to predict the default probability and obtain the model prediction score data.

[0184] S33. Input the rule determination result data and the model prediction score data into the weighted fusion layer for weighted fusion calculation to obtain preliminary comprehensive risk score data;

[0185] S34. Input the preliminary comprehensive risk score data into the normalization layer for normalization to obtain the preliminary risk score data.

[0186] Furthermore, the risk transmission simulation module 24 can also be used for:

[0187] S41. Based on the preliminary risk score data, map the preliminary risk score of each asset to the initial risk value of the corresponding node in the supply chain knowledge graph to obtain the initial risk data of the node.

[0188] S42. Based on the topology of the supply chain knowledge graph and the initial risk data of the nodes, and using the initial risk data of the nodes as the propagation source, perform multiple rounds of risk transmission iterative simulation calculations to obtain the iterative risk value data of each node.

[0189] S43. Perform convergence judgment on the iterative risk value data of each node. When the change in risk value between two adjacent iterations is less than the preset convergence threshold, calculate the difference between the final iterative risk value and the initial risk value of each node to obtain the transmission risk added value data.

[0190] S44. Sum the initial risk data and the transmission risk added value data of the nodes to obtain the supply chain transmission risk index data.

[0191] Furthermore, the early warning and source tracing module 25 can also be used for:

[0192] S51. Compare and judge the supply chain transmission risk index data with the preset risk threshold to obtain risk warning information, and generate a risk propagation heat map based on the risk warning information;

[0193] S52. Based on the risk propagation heat map, reverse tracing is performed along the risk propagation path in the supply chain knowledge graph to locate the root cause node of the risk and obtain the risk root cause location.

[0194] S53. Based on the node type and risk attributes of the risk root cause nodes, generate response suggestions from the preset strategy knowledge base, integrate the risk warning information, risk root cause location and response suggestions to obtain risk monitoring results.

[0195] Furthermore, the dynamic asset profile building module 22 can also be used for:

[0196] S211. Based on the supply chain knowledge graph, perform neighbor node statistics on the entity nodes corresponding to key assets to obtain the degree centrality of nodes.

[0197] S212. Based on the supply chain knowledge graph, calculate the shortest path distance for the entity nodes corresponding to key assets to obtain proximity centrality.

[0198] S213. Based on the supply chain knowledge graph, perform intermediary path statistics on the entity nodes corresponding to key assets to obtain intermediary centrality.

[0199] S214. Perform feature concatenation on degree centrality, proximity centrality, and betweenness centrality to obtain network correlation data.

[0200] Furthermore, the data acquisition and map construction module 21 can also be used for:

[0201] S11. Perform entity recognition processing on multi-source heterogeneous data to obtain entity identification data;

[0202] S12. Perform relation extraction processing on multi-source heterogeneous data to obtain relational data;

[0203] S13. Merge and align entity identification data and relationship data to obtain entity-relation triplet data;

[0204] S14. Based on entity-relationship triple data, a supply chain knowledge graph is constructed.

[0205] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the supply chain asset financial risk monitoring method as described above.

[0206] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0207] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0208] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for monitoring financial risks of supply chain assets, characterized in that, The method includes: S1. Obtain multi-source heterogeneous data of the target supply chain, and fuse the entity identifiers and relationships in the multi-source heterogeneous data to construct a supply chain knowledge graph; wherein, the multi-source heterogeneous data includes financial data, operational data, transaction chain data and external environment data; S2. Perform network topology feature analysis on the key assets in the supply chain knowledge graph to obtain dynamic asset profile data; S3. Input the dynamic asset profile data into the hierarchical fusion assessment model to quantify the risk and obtain preliminary risk score data; S4. Based on the preliminary risk scoring data and the supply chain knowledge graph, using the preliminary risk scoring data as the initial risk source, perform risk transmission simulation and iterative calculation on the supply chain knowledge graph to obtain supply chain transmission risk index data. S5. Visualize and perform source tracing analysis on the supply chain transmission risk index data to obtain risk monitoring results; wherein, the risk monitoring results include risk warning information, risk root cause identification, and response suggestions.

2. The method according to claim 1, characterized in that, S2 includes: S21. Based on the supply chain knowledge graph, network topology indicators are calculated for the entity nodes corresponding to the key assets to obtain network correlation data; wherein, the network correlation data includes degree centrality, proximity centrality and betweenness centrality. S22. Based on the financial data and operational data in the multi-source heterogeneous data, conduct a health assessment of the key assets to obtain health index data. S23. Extract the static attributes of the key assets from the multi-source heterogeneous data to obtain static attribute data; S24. The network correlation data, the health index data, and the static attribute data are integrated to obtain the dynamic asset profile data.

3. The method according to claim 2, characterized in that, The hierarchical fusion evaluation model includes a rule engine layer, a gradient boosting tree model layer, a weighted fusion layer, and a normalization layer. S3 includes: S31. Input the dynamic asset profile data into the rule engine layer, and perform risk filtering and marking on the assets through preset hard rule thresholds to obtain rule judgment result data; S32. Input the dynamic asset profile data into the gradient boosting tree model layer to predict the default probability and obtain the model prediction score data. S33. Input the rule determination result data and the model prediction score data into the weighted fusion layer for weighted fusion calculation to obtain preliminary comprehensive risk score data; S34. Input the preliminary comprehensive risk score data into the normalization layer for normalization to obtain the preliminary risk score data.

4. The method according to claim 1, characterized in that, S4 includes: S41. Based on the preliminary risk score data, map the preliminary risk score of each asset to the initial risk value of the corresponding node in the supply chain knowledge graph to obtain the initial risk data of the node. S42. Based on the topology of the supply chain knowledge graph and the initial risk data of the nodes, using the initial risk data of the nodes as the propagation source, perform multi-round risk transmission iterative simulation calculations to obtain the iterative risk value data of each node; wherein, the first... Nodes in round iteration The expression for updating the risk value is: In the formula, Represents a node In the Risk value after round of iterations Represents a node In the Risk value after round of iterations Indicates the iteration round number. Represents a node The set of neighboring nodes, Represents a node The neighbor node index, Represents a node To the node edge weights, Representing neighboring nodes In the Risk value after round of iterations This represents the preset risk propagation coefficient; S43. Perform convergence judgment on the iterative risk value data of each node. When the change in risk value between two adjacent iterations is less than the preset convergence threshold, calculate the difference between the final iterative risk value and the initial risk value of each node to obtain the transmission risk added value data. S44. The initial risk data of the node and the transmission risk added value data are summed to obtain the supply chain transmission risk index data.

5. The method according to claim 1, characterized in that, S5 includes: S51. Compare and judge the supply chain transmission risk index data with the preset risk threshold to obtain risk warning information, and generate a risk propagation heat map based on the risk warning information; S52. Based on the risk propagation heatmap, reverse tracing is performed along the risk propagation path in the supply chain knowledge graph to locate the root cause node of the risk and obtain the risk root cause location. S53. Based on the node type and risk attribute of the risk root cause node, generate response suggestions from the preset strategy knowledge base, integrate the risk warning information, the risk root cause location and the response suggestions to obtain the risk monitoring result.

6. The method according to claim 2, characterized in that, S21 includes: S211. Based on the supply chain knowledge graph, perform neighbor node statistics on the entity nodes corresponding to the key assets to obtain the vertex degree centrality; wherein, the expression for the vertex degree centrality is: In the formula, Represents a node degree centrality, Represents the first in the supply chain knowledge graph 1 node Represents a node The set of neighboring nodes, Represents a node The number of neighboring nodes, This represents the total number of nodes in the supply chain knowledge graph. S212. Based on the supply chain knowledge graph, calculate the shortest path distance for the entity nodes corresponding to the key assets to obtain proximity centrality; wherein, the expression for proximity centrality is: In the formula, Represents a node Proximity centrality Represents the first in the supply chain knowledge graph 1 node This represents the total number of nodes in the supply chain knowledge graph. Represents the first in the supply chain knowledge graph 1 node Represents a node To the node The shortest path distance; S213. Based on the supply chain knowledge graph, perform mediation path statistics on the entity nodes corresponding to the key assets to obtain mediation centrality; wherein, the expression for mediation centrality is: In the formula, Represents a node The centrality of the middle, Represents the first in the supply chain knowledge graph 1 node This represents the source node index in the supply chain knowledge graph. This represents the index of the target node in the supply chain knowledge graph. Represents a node To the node The total number of shortest paths, Represents a node To the node The shortest path passes through the nodes Quantity; S214. Perform feature concatenation on the degree centrality, proximity centrality, and betweenness centrality to obtain the network correlation data.

7. The method according to claim 1, characterized in that, S1 includes: S11. Perform entity recognition processing on the multi-source heterogeneous data to obtain entity identification data; S12. Perform relation extraction processing on the multi-source heterogeneous data to obtain relational data; S13. Merge and align the entity identifier data and the association relationship data to obtain entity-relation triplet data; S14. Based on the entity-relationship triple data, the supply chain knowledge graph is constructed.

8. A supply chain asset financial risk monitoring system, characterized in that, The system includes: The data acquisition and graph construction module is used to acquire multi-source heterogeneous data of the target supply chain, integrate the entity identifiers and relationships in the multi-source heterogeneous data, and construct a supply chain knowledge graph; wherein, the multi-source heterogeneous data includes financial data, operational data, transaction chain data, and external environment data; The dynamic asset profile construction module is used to perform network topology feature analysis on key assets in the supply chain knowledge graph to obtain dynamic asset profile data. The risk quantification module is used to input the dynamic asset profile data into the hierarchical fusion assessment model to quantify the risk and obtain preliminary risk score data. The risk transmission simulation module is used to perform risk transmission simulation iterative calculation on the supply chain knowledge graph, based on the preliminary risk scoring data and the supply chain knowledge graph, with the preliminary risk scoring data as the initial risk source, to obtain supply chain transmission risk index data. The early warning and traceability module is used to perform visual early warning and traceability analysis on the supply chain transmission risk index data to obtain risk monitoring results; wherein, the risk monitoring results include risk early warning information, risk root cause identification and response suggestions.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images