Method and system for monitoring the risk of a target enterprise based on supply chain transmission

Abstract

The present application relates to the technical field of enterprise management and risk monitoring, and specifically discloses a method and system for penetrating risk monitoring of invested enterprises based on supply chain conduction, which accurately reflects the dynamic changes of the supply relationship between enterprises by constructing a supply chain meta-network containing dynamic conduction impedance tags; calculates the propagation tendency of risks in the supply chain using a risk gravity model, dynamically generates a risk propagation chain pointing to the invested enterprise, and realizes forward-looking early warning of risks; focuses on high-risk conduction paths to improve monitoring efficiency and accuracy, and solves the problem of analysis delay caused by massive data; at the same time, the system has closed-loop adaptive learning capability, and updates the supply chain meta-network parameters in reverse according to the actual risk conduction results, continuously improving the accuracy of the risk assessment model. The present application can identify potential risks in advance, provide timely and accurate risk early warning for investment institutions, and effectively improve the initiative of risk management.

Description

Technical Field

[0001] This invention belongs to the field of enterprise management and risk monitoring technology, and relates to a method and system for penetrating risk monitoring of invested enterprises based on supply chain transmission. Background Technology

[0002] In the modern business environment, a company's operations are highly dependent on its supply chain network; fluctuations in any link can impact the entire chain. For investment institutions, the value of their portfolio companies depends not only on the company itself but also heavily on the operating conditions of its upstream and downstream suppliers, customers, and other related parties. Therefore, when monitoring the risks of portfolio companies, it is essential to consider them within the entire supply chain ecosystem. This risk analysis approach, oriented towards corporate operations management and resource synergy, is a crucial component of business management.

[0003] In existing technologies, monitoring supply chain risks typically employs a solution that constructs a supply chain topology map. This solution first builds a network diagram of supply relationships between enterprises using business registration data and publicly available information. Then, the system continuously monitors publicly available risk information for each enterprise node in the network, such as financial deterioration, legal proceedings, or negative public opinion. When the risk indicators of a certain enterprise node exceed a preset threshold, the system triggers an alarm, alerting other enterprises associated with that risky enterprise that may face risks.

[0004] However, existing technical solutions have significant technical shortcomings. First, the supply chain networks they construct are typically static, failing to reflect the dynamic changes in the strength, dependence, and substitutability of supply relationships in the real world, leading to insufficient accuracy in risk assessment. Second, the solution identifies risks at the isolated node level, lacking the ability to dynamically simulate and track how risks propagate along specific paths between different enterprises. Finally, this global node monitoring approach causes the system to process a large amount of redundant information, making it difficult to focus on truly critical risk paths with high propagation potential, often resulting in delayed early warnings or a large number of invalid alerts. Summary of the Invention

[0005] In view of this, in order to solve the problems mentioned in the background technology above, a method and system for penetrating risk monitoring of invested companies based on supply chain transmission is proposed.

[0006] The objective of this invention can be achieved through the following technical solution: The first aspect of this invention provides a method for penetrating risk monitoring of invested enterprises based on supply chain transmission, including: S1, constructing a supply chain meta-network with node-to-node relationships, calculating and assigning dynamic transmission impedance labels to the node relationships to quantify the resistance to risk propagation, and generating a supply chain meta-network with transmission impedance labels.

[0007] S2. In response to the risk event signal identified by the risk source node in the supply chain meta-network, calculate the risk gravity value from the risk source node to its associated node based on the risk gravity model used to calculate the risk propagation tendency and in combination with the conduction impedance label.

[0008] S3. Based on the risk gravity value, select a path and connect nodes starting from the risk source node to dynamically generate one or more risk propagation chains pointing to the invested company node.

[0009] S4. Based on the composition of the risk propagation chain, dynamically adjust the data collection strategy, focus real-time data monitoring resources on the nodes in the risk propagation chain, and obtain real-time status feedback data of the nodes in the risk propagation chain.

[0010] S5. Utilize real-time status feedback data to adjust the growth and composition of the risk propagation chain, and generate transmission effect feedback information based on the final transmission result of the risk propagation chain. Use the transmission effect feedback information to update the transmission impedance labels of the corresponding node relationships in the supply chain meta-network.

[0011] S6. When the risk propagation chain reaches the node of the invested company, a penetrating risk warning report is generated based on the complete structure of the risk propagation chain.

[0012] The second aspect of the present invention provides a penetration-based risk monitoring system for investee companies based on supply chain transmission, comprising: a supply chain meta-network generation module, which constructs a supply chain meta-network with node-to-node relationships, calculates and assigns dynamic transmission impedance labels to the node relationships to quantify the resistance to risk propagation, and generates a supply chain meta-network with transmission impedance labels.

[0013] The risk gravity value calculation module, in response to the risk event signal identified by the risk source node in the supply chain meta-network, calculates the risk gravity value from the risk source node to its associated node based on the risk gravity model used to calculate the risk propagation tendency and in combination with the conduction impedance label.

[0014] The risk propagation chain dynamic generation module, based on the risk gravity value, selects a path and connects nodes starting from the risk source node, dynamically generating one or more risk propagation chains pointing to the invested company node.

[0015] The monitoring resource scheduling and feedback module dynamically adjusts the data collection strategy according to the composition of the risk propagation chain, focuses real-time data monitoring resources on the nodes in the risk propagation chain, and obtains real-time status feedback data of the nodes in the risk propagation chain.

[0016] The two-way feedback and learning module uses real-time status feedback data to adjust the growth and structure of the risk propagation chain, and generates transmission effect feedback information based on the final transmission result of the risk propagation chain. It then uses the transmission effect feedback information to update the transmission impedance labels of the corresponding nodes in the supply chain meta-network.

[0017] The risk warning report generation module generates a penetrating risk warning report based on the complete structure of the risk propagation chain when the risk propagation chain reaches the node of the invested company.

[0018] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention achieves forward-looking early warning of supply chain transmission risks by dynamically generating risk propagation chains and combining them with feedback learning mechanisms. The system no longer passively monitors isolated enterprise risk signals, but actively tracks the dynamic propagation process of risks in the supply chain network. It can predict and warn of potential paths and impacts before risk events are transmitted from the source to the invested enterprises, thus advancing the risk response window and improving the foresight of risk management.

[0019] (2) This invention improves the efficiency and accuracy of risk monitoring by intelligently focusing monitoring resources on high-risk transmission paths. Unlike traditional methods that conduct homogenized and broad-based monitoring of the entire supply chain network, this invention only collects and analyzes data on dynamically generated risk transmission chains with high intensity and frequency. It concentrates limited computing and data resources on the most critical risk links, effectively solving the problems of analysis delay and noise interference caused by massive amounts of data, and achieving efficient utilization of resources.

[0020] (3) This invention constructs a closed-loop adaptive learning system, which enables the accuracy of the risk assessment model to continuously evolve. By using the transmission result of each real risk event, whether it is successful transmission or interruption, as feedback information to update the transmission impedance label of the supply chain meta-network, the system can continuously learn and correct its understanding of the transmission law of supply chain risks from practice, thereby demonstrating stronger robustness and more accurate predictive ability when facing new or sudden risk events in the future. Attached Figure Description

[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the method steps of the present invention.

[0023] Figure 2 This is a schematic diagram of the system structure connection of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please see Figure 1 The first aspect of the present invention provides a method for penetrating risk monitoring of invested enterprises based on supply chain transmission, including: S1, constructing a supply chain meta-network with node-to-node relationships, calculating and assigning dynamic transmission impedance labels to the node relationships to quantify the resistance to risk propagation, and generating a supply chain meta-network with transmission impedance labels.

[0026] In a specific embodiment of the present invention, the specific steps for calculating and assigning dynamic conduction impedance labels to node associations to quantify the resistance to risk propagation, and generating a supply chain meta-network with conduction impedance labels, include: acquiring enterprise entity data and transaction data representing the supply relationship between enterprises, and generating an original dataset.

[0027] Based on the original dataset, enterprise entities are mapped as nodes and supply relationships are mapped as edges to construct an initial supply chain network.

[0028] Based on the assessment rules used to evaluate the difficulty of risk transmission, an initial conduction impedance value is calculated for each edge of the initial supply chain network, and the initial conduction impedance value is assigned as a conduction impedance label to the corresponding edge, forming a supply chain meta-network with conduction impedance labels.

[0029] Specifically, the engineering purpose of constructing the supply chain meta-network with dynamic conduction impedance labels in this invention is to transform dispersed, multi-source heterogeneous data into a structured, quantifiable basic network model that reflects the physical characteristics of risk transmission, thus providing a computational basis for the subsequent dynamic generation of risk propagation chains.

[0030] The first step of the project aims to aggregate all the raw data required to construct the network topology. The system uses pre-defined API interfaces to batch access business registration information databases, corporate annual report databases, bidding information platforms, and third-party supply chain data services. Simultaneously, for authorized investee companies, the system securely accesses their Enterprise Resource Planning (ERP) or Customer Relationship Management (CRM) systems via a data channel to extract supplier directories, purchase order records, customer directories, and sales order records for the past 36 months. The system's data processing includes identifying company entity names from the acquired text data and aggregating related parties from transaction records, ultimately forming a raw dataset containing unique identifiers for company entities, types of inter-company transaction relationships, and transaction amounts.

[0031] The second step of the project aims to transform the original dataset into a graphical network structure. The system executes a graph-building algorithm, mapping each unique enterprise entity identifier to a network node and the supply or customer relationships between enterprises to directed edges connecting these nodes. The direction of the edges is determined by the flow of products or services, i.e., from suppliers to buyers. This algorithm generates nodes and edges by traversing all relationships in the original dataset, and its output is an initial supply chain network containing only topological connections.

[0032] The third step aims to quantify and assign values ​​to each edge in the initial supply chain network, giving it a physical sense of risk transmission resistance, thereby generating the final supply chain meta-network with transmission impedance labels. Here, the transmission impedance label is a comprehensive quantitative indicator, defined in engineering terms as the degree of obstruction encountered when a risk event propagates from one node to the next along a specific supply relationship edge. The system calculates the initial transmission impedance value for each directed edge in the initial supply chain network based on preset initial evaluation rules. This calculation is achieved through the following formula: In this formula, Representative from enterprise nodes To enterprise nodes The conduction impedance value of the directed edge. This represents the stability factor of the supply relationship. This represents the supply substitutability factor. Represents key factors in supply. , , These are preset weighting coefficients, and their sum is 1. , , This is a normalization function used to map raw data of different dimensions to a uniform range of 0 to 1, ensuring the validity of the calculation. Key factors in the formula. The negative term reflects the objective law that the more critical the supply, the less resistance there is to risk transmission. The system will calculate... As labels, they are attached to the corresponding edges to complete the construction of the supply chain meta-network with conductive impedance labels.

[0033] It should be noted that the supply relationship stability factor Supply substitutability factors and key factors of supply The specific calculation is expressed as follows: Supply relationship stability factor Based on the dispersion analysis of historical transaction records, the inverse proportional normalization function is used to calculate the formula: ,in, Represents a node With nodes The standard deviation of the trading time interval within a preset historical period is used to quantify the fluctuation of trading frequency; The coefficient of variation (i.e., the standard deviation of transaction amount divided by the mean of transaction amount) is used to quantify the fluctuation of transaction amount. and This is a preset normalized adjustment coefficient used to balance the weighting of time fluctuations and monetary fluctuations on stability. Supply substitutability factor. Through searching industry databases and performing logarithmic mapping, the formula is: ,in: This indicates the results retrieved from the industry database, related to the node. The number of active suppliers with the same standard industry classification code and located within the effective logistics radius; It is a natural logarithm function, used to smooth the marginal effect of an increase in the number of suppliers; These are standardized coefficients used to map the calculation results to a standard evaluation interval. Key supply factors. The formula for combining the proportion of comprehensive procurement with the weight of material importance level is: ,in: Represents a node From node The purchase amount of this type of material; Represents a node The total procurement cost of this type of material; This refers to the percentage of the total purchase amount. The material is assigned a importance score. If the material is marked as "strategic material" or "single source" in the BOM, the score is set to high (e.g., 1.0); otherwise, it is set to low (e.g., 0.5). and These are the weighting coefficients, and .

[0034] It should also be noted that the normalization adjustment coefficient The value is usually taken as 0.1. The value is 1.0, based on the standard deviation of the transaction time interval ( ) is usually measured in "days", and its numerical magnitude is significantly larger than the coefficient of variation in monetary terms, which is between 0 and 1. ), by setting This can eliminate the bias of dimensional differences in stability calculations, ensuring a balance between the contributions of temporal regularity and monetary stability to the results; standardization coefficient The value is usually taken as 0.23, based on the logarithmic function. It exhibits diminishing marginal utility; when the number of effective alternative suppliers reaches approximately 75 (in a perfectly competitive market), Multiplying by 0.23 approaches the normalization upper limit of 1.0, which aligns with the economic principle that "after the number of suppliers reaches a certain scale, the marginal contribution of new additions to risk reduction rapidly diminishes"; weighting coefficient The value is 0.6. The value is 0.4. Based on the "Pareto Principle" and the principle of risk exposure, the proportion of procurement amount (objective hard data) directly determines the scale of economic loss when the risk occurs and should have a dominant weight. The strategic level (subjective qualitative label) serves as a key correction item. Together, they constitute a composite evaluation system that "focuses on financial exposure while taking into account material attributes".

[0035] In a specific embodiment of the present invention, the weighting coefficient is typically taken as a value of , , The evaluation criteria follow the risk assessment logic of "behavior-driven, business-based, and environment-dependent": the supply relationship stability factor is based on real historical performance data between nodes and is the most direct empirical indicator for predicting future reliability, reflecting the inherent resilience of the supply chain, and is therefore given the highest weight of 0.5; the supply criticality factor is directly related to the economic lifeline and production continuity of the enterprise, determining the level of damage caused by the breakage of the supply connection, and belongs to the core business dimension, so it is given the second highest weight of 0.3; while the supply substitutability factor reflects the fault tolerance buffer capacity of the external market. Although it can dilute risks, it is a subordinate environmental variable compared to direct transaction behavior and business impact, so it is given the lowest weight of 0.2. The three factors together constitute a comprehensive evaluation system that emphasizes both historical performance and business consequences and market elasticity.

[0036] S2. In response to the risk event signal identified by the risk source node in the supply chain meta-network, calculate the risk gravity value from the risk source node to its associated node based on the risk gravity model used to calculate the risk propagation tendency and in combination with the conduction impedance label.

[0037] In a specific embodiment of the present invention, the specific steps for calculating the risk gravity value from the risk source node to its associated node include: obtaining the business association strength data between the risk source node and each of its associated nodes.

[0038] Obtain the risk buffer capacity parameter for each associated node, which characterizes its ability to withstand risks.

[0039] The business association strength data, risk buffer capacity parameters, and conduction impedance labels of the edges connecting the risk source node and the associated node are input into the risk gravity model for calculation, and the risk gravity value is output.

[0040] Specifically, the step of calculating the risk gravity value based on a preset risk gravity model and conduction impedance label in this invention aims to transform an abstract risk event into a quantifiable, directional propagation tendency indicator, providing a decisive input for the subsequent dynamic generation of the risk propagation chain path selection. Here, the risk gravity value, in engineering terms, is the magnitude of the instantaneous "traction" of a risk event spreading from the risk source node to its directly related downstream nodes.

[0041] The first step of the project aims to quantify the strength of economic ties between risk source nodes and related nodes. The system first retrieves all directly connected downstream nodes within the supply chain meta-network tagged with conducted impedance, based on the unique identifier of the risk source node. For each downstream related node, the system queries its internal database to extract pre-calculated and stored business association strength data. This data is typically a dimensionless value obtained after normalization, based on the proportion of the risk source node's total transaction amount with that related node to the related node's total procurement amount over the past 24 months.

[0042] The second step of the project aims to quantify the ability of each related node to withstand external risk shocks. For each related node, the system extracts a pre-defined risk buffer capability parameter from its node attribute data. This risk buffer capability parameter is a composite indicator, calculated using a pre-defined weighted average model based on multiple sub-indicators, including the current ratio and quick ratio from the node's latest financial statements, as well as safety stock levels and the number of backup suppliers from its inventory management system. The model is also normalized to eliminate the influence of different dimensions.

[0043] The third step aims to comprehensively consider the above factors and calculate the final risk gravity value. The system invokes a preset risk gravity model to perform calculations for each potential transmission path from the risk source node to its associated nodes. This model is implemented using the following formula: In this formula, For the final calculation of the risk source node To its associated nodes The risk gravity value. Data representing the strength of business associations is obtained in the first step. Represents associated nodes The risk buffer capacity parameter is obtained in the second step. Represents the connection node With nodes The conduction impedance label value of the edge is read directly from the supply chain meta-network with conduction impedance labels. , , These are the preset calibration coefficients for the model, which are dimensionless weight values, typically set between 0.1 and 1.0. They are optimized and calibrated by the system through backtesting of historical risk events, and are used to adjust the contribution of each input parameter to the final result.

[0044] In a specific embodiment of the present invention, and in a preferred embodiment, the typical value of the preset calibration coefficient is usually: , , Its value is determined in accordance with the model robustness principle of "global dominance, local correction, and noise suppression": A higher value is chosen for the global gain coefficient to strengthen the dominant role of the graph topology in risk transmission and ensure the consistency of the model's predictions at the macro level. The intermediate value of the baseline bias coefficient is used to preserve the fundamental influence of the inherent attributes of nodes and prevent the network structure from being completely smoothed out. Taking a lower value as the adjustment coefficient actually acts as a damping effect of "low-pass filtering". In historical backtesting, it can effectively filter out occasional market fluctuation noise and prevent the model from overfitting due to its sensitivity to short-term abnormal data, thereby ensuring that the risk warning output by the system has a very high confidence and stability.

[0045] In a specific embodiment of the present invention, the risk gravity model is configured such that the output risk gravity value is positively correlated with the input business association strength data, and negatively correlated with the input risk buffering capacity parameter and conduction impedance label.

[0046] Specifically, the further definition of the risk gravity model calculation formula in this invention aims to clarify the functional relationship between the input variables and output results within the model, ensuring that the model's calculation logic conforms to the objective laws of the business and physical world, thereby guaranteeing the effectiveness and reliability of the risk gravity value as a decision-making basis. This definition clarifies the direction of correlation between the risk gravity value and the three core input variables. In the aforementioned risk gravity model formula, that is... This correlation is achieved through the formula structure.

[0047] First, the numerator of the formula is ,in This represents the strength of business association data. Since this item is located in the numerator and all other variables are positive, the risk gravity value... Data on the strength of relevance to business There is a positive correlation. This aligns with objective laws, namely, the closer the economic link between the risk source and the related nodes—for example, the higher the proportion of procurement—the greater the direct impact and pull of the risk event. Secondly, the denominator of the formula is... This includes risk buffering capacity parameters. and conduction impedance label value Since both variables are in the addition terms of the denominator, increasing them will increase the overall denominator, thus affecting the final risk gravity value. It decreases. Therefore, the risk gravity value... Risk buffer capacity parameter and conduction impedance label value All three show a negative correlation. This aligns with objective laws: the more financially healthy, well-stocked, or backup suppliers a related node possesses—that is, the stronger its risk buffer—the more resilient it is to external risk shocks, thus weakening the gravitational pull of risk transmission. Similarly, the more unstable and easily replaceable the supply relationship between two nodes—that is, the higher the transmission resistance—the more difficult it is for risk to propagate along that path, and the less the gravitational pull. Through this explicit mathematical structure, the risk gravity model ensures that its calculations accurately reflect the interaction between the intrinsic driving forces and resistances of risk transmission, providing a solid logical foundation for the system to subsequently select the most probable risk propagation path.

[0048] S3. Based on the risk gravity value, select a path and connect nodes starting from the risk source node to dynamically generate one or more risk propagation chains pointing to the invested company node.

[0049] S4. Based on the composition of the risk propagation chain, dynamically adjust the data collection strategy, focus real-time data monitoring resources on the nodes in the risk propagation chain, and obtain real-time status feedback data of the nodes in the risk propagation chain.

[0050] In a specific embodiment of the present invention, the specific steps for obtaining real-time status feedback data of nodes on the risk propagation chain include: identifying all nodes in the risk propagation chain and marking the latest extended node of the risk propagation chain as the growth front node.

[0051] Based on all nodes and the growth front node, a differentiated data collection strategy containing different monitoring priorities is generated.

[0052] Based on a differentiated data acquisition strategy, data acquisition operations are performed, and the acquired raw data is processed into structured real-time status feedback data.

[0053] Specifically, the present invention focuses real-time data monitoring resources on the risk propagation chain and obtains status feedback data. Its engineering purpose is to dynamically switch the system's data collection mode from static, homogeneous full-network scanning to proactive tracking of high-risk paths with resource-intensive methods, so as to obtain the most valuable real-time intelligence for risk evolution judgment with the lowest resource consumption.

[0054] The first step of the project aims to accurately identify targets that require enhanced monitoring. Upon receiving the risk propagation chain data structure generated in the previous step, the system first executes a node parsing program. This program traverses the risk propagation chain, extracting the unique enterprise identifiers of all nodes within the chain to form a list of high-priority monitoring targets. Simultaneously, the program specifically marks the most recently added node in the chain at the current time, defining it as a growth frontier node and assigning it the highest monitoring level.

[0055] The second step of the project aims to execute specific resource reallocation actions. The system has a built-in data acquisition strategy management module that maintains the monitoring task queues and resource quotas for all network nodes. Upon receiving a list of high-priority monitoring targets, this module triggers a strategy adjustment procedure. This procedure adjusts the real-time data monitoring resources, such as the system's data acquisition API call frequency, web crawler bandwidth, and database query priority, from the default balanced allocation mode to a differentiated allocation mode. For nodes not on any risk propagation chain, their monitoring frequency is reduced to the baseline level, for example, a public information poll every 6 to 12 hours. For nodes already on the risk propagation chain but not at the forefront of growth, their monitoring frequency is increased to an enhanced level, for example, every 1 to 2 hours. For nodes marked as being at the forefront of growth, the system allocates the highest priority monitoring resources, and the API call frequency for their associated public opinion monitoring keywords can be increased to once every 5 to 10 minutes to achieve near real-time tracking.

[0056] The third step of the project aims to perform data collection and generate structured feedback data. Based on the adjusted data collection strategy, the system's data collection agent continuously captures multi-dimensional real-time data from nodes in the high-priority monitoring target list through designated interfaces. These data sources include, but are not limited to, financial news release interfaces, public legal proceedings platforms, corporate announcement updates, and specific topic streams from key social media platforms. The collected raw data stream then enters a data processing pipeline for cleaning, deduplication, entity linking, and sentiment polarity assessment, ultimately being formatted into structured data records containing node identifiers, data sources, timestamps, and summaries of core events. These processed, structured data records, directly reflecting the latest dynamics of on-chain nodes, collectively constitute the state feedback data used for decision-making in the next step.

[0057] S5. Utilize real-time status feedback data to adjust the growth and composition of the risk propagation chain, and generate transmission effect feedback information based on the final transmission result of the risk propagation chain. Use the transmission effect feedback information to update the transmission impedance labels of the corresponding node relationships in the supply chain meta-network.

[0058] In a specific embodiment of the present invention, the specific steps for adjusting the growth and structure of the risk propagation chain using real-time status feedback data include: updating the risk buffering capacity parameters of the growth frontier nodes on the risk propagation chain based on the real-time status feedback data.

[0059] Using the updated risk buffering capacity parameters, the risk gravity value starting from the growth front node is recalculated.

[0060] Based on the recalculated risk gravity value, it is determined whether the risk propagation chain stops growing, changes its path, or continues to extend, thereby achieving dynamic adjustment of the risk propagation chain.

[0061] Specifically, the step in this invention of dynamically adjusting the risk propagation chain based on state feedback data and updating the supply chain meta-network in reverse has the engineering purpose of achieving closed-loop adaptive learning of the analysis model, that is, using the latest real-time intelligence to correct the current risk transmission prediction and solidify the experience of this event to improve the accuracy of the entire system's cognition of future events.

[0062] The first step of the project aims to update risk propagation predictions in real time using new intelligence. Upon receiving status feedback data generated in the previous step, the system immediately correlates it with the growth front nodes in the risk propagation chain. For example, if the status feedback data indicates that a growth front node has issued a clarification announcement or activated a backup supplier, the system triggers an instantaneous update procedure for the risk buffer capacity parameter, temporarily increasing the node's risk buffer capacity parameter. Subsequently, the system calls the same risk gravity model as in step S2, using the updated parameters to recalculate the risk gravity values ​​from the growth front node to all its downstream associated nodes.

[0063] The second step aims to dynamically adjust the risk propagation chain based on the updated predictions. The system compares the newly calculated risk gravity values ​​with preset risk propagation thresholds. If all newly calculated risk gravity values ​​are lower than the preset risk propagation threshold, it indicates that the risk has been effectively contained at that node. The system will then perform a chain pruning operation, stopping the continued growth of the risk propagation chain along all relevant paths and marking the chain state of that node as "blocked." If the risk gravity values ​​of some paths still exceed the preset risk propagation threshold, the chain will continue to grow along these paths, but paths that no longer meet the conditions due to the decrease in risk gravity values ​​have been excluded, thus achieving dynamic correction of the chain growth direction.

[0064] In a specific embodiment of the present invention, the preset risk propagation threshold is typically set to 0.05. The value is determined based on the principle of balancing "effectiveness boundary and computational convergence": this value corresponds to the 5% significance level commonly used in statistics, meaning that when the risk energy is attenuated through multiple levels of network topology during the transmission process, if its remaining influence is lower than this critical value, it is considered in a physical sense to have been digested by the enterprise's safety stock or capital redundancy, etc., and belongs to "background noise" rather than an effective threat. At the same time, at the algorithm level, this risk propagation threshold acts as a "pruning filter" to prevent excessive traversal, effectively blocking the infinite spread of weak risks, avoiding the exponential explosion of computational paths due to the "butterfly effect", and ensuring that the system's computing power is focused on those truly destructive backbone risk links that may trigger chain reactions, thereby achieving the optimal trade-off between the coverage of early warning and computational efficiency.

[0065] It should be noted that generating transmission effect feedback information based on the final transmission result of the risk propagation chain specifically includes the following: After the entire risk propagation chain's lifecycle ends, whether the transmission successfully reaches the invested company node or is blocked midway, the system will execute a transmission review procedure. This procedure will trace back the complete path of the chain and the final state of each node, generating transmission effect feedback information containing the initial predicted path, the actual transmission path, and information on key blocking points. The system inputs the transmission effect feedback information into a transmission impedance label update module. This update module adjusts the transmission impedance label values ​​of the corresponding edges in the supply chain meta-network according to the following rules: ,in, This is the updated conduction impedance label value. This is the value before the update. It is an adjustment amount, calculated as follows: if the feedback information of the transmission effect shows that the risk has been successfully passed through the edge. ,but A negative value reduces its impedance; conversely, if the risk is on the edge... If the blockage is blocked, then If it is a positive value, its impedance is increased. The absolute value is proportional to the certainty or impact level of the event, ensuring that significant and certain events have a higher weight for model correction.

[0066] In a specific embodiment of the present invention, the specific steps of using the transmission effect feedback information to reverse update the transmission impedance label of the corresponding node association in the supply chain meta-network include: distinguishing from the transmission effect feedback information the edges that successfully transmit risk and the edges that block the transmission of risk.

[0067] Based on the learning rate used to control the model's learning rate, the conduction impedance label value of the edge that successfully conducts the risk is lowered.

[0068] Based on the preset base learning rate, the conduction impedance label value of the edge blocking the transmission of risk is increased, and the magnitude of the increase is adjusted in combination with the event confidence factor that characterizes the degree of determinism of the blocking.

[0069] Specifically, the step of adjusting the conduction impedance label value using conduction effect feedback information in this invention aims to transform the evolution result of a single risk event into a persistent correction of the parameters of the basic network model, thereby achieving adaptive evolution and cognitive enhancement of the supply chain meta-network, making its understanding of risk transmission increasingly closer to the real world.

[0070] The first step of the project aims to enhance the understanding of validated risk transmission paths. After receiving feedback on the transmission effect, the system analyzes the path segments marked as 'successfully transmitted.' For each edge in that path segment, if it was included in the risk propagation chain during the initial prediction, the system initiates a resistance reduction procedure for that edge. The core of this procedure is to identify a known "weak path" and reduce its resistance value in the model so that similar risks can be more easily predicted in the future.

[0071] The second step aims to identify and enhance the actual barriers to risk propagation. If the propagation feedback indicates that a predicted risk propagation path is blocked at a certain edge due to feedback data from downstream nodes (e.g., the node successfully activates a backup supplier), the system initiates an impedance increase procedure for that blocked edge. This aims to solidify the "firewall" effect verified in this event into the model, increasing its resistance value and preventing the system from overestimating the risk propagation capability of that path in the future.

[0072] Both programs share the same core adjustment algorithm to ensure consistency and controllability in the learning process. Specific adjustment amounts... Calculated using the following formula: In this formula, This is the amount of adjustment that this event caused to the conduction impedance label value. It is a transmission direction factor; when the risk is successfully transmitted through this edge... The value is negative 1, used to reduce impedance; when the risk is blocked on this side, The value is positive 1, which is used to increase the impedance. It is the preset base learning rate, usually set to 0.1, used to control the speed of the entire model's learning and adaptation, and to prevent excessive oscillation of model parameters caused by a single event. This is the event confidence factor, a dimensionless value between 0 and 1, derived from the confidence assessment of status feedback data. For example, the event confidence factor corresponding to a formal legal judgment or a listed company announcement. It can be set from 0.9 to 1.0, while sentiment fluctuations originating from a single social media platform may correspond to... The value is only 0.2 to 0.4. This formula ensures that the direction of adjustment is determined by the actual transmission results, while the magnitude of adjustment is weighted by the degree of certainty of the event, thereby achieving accurate, robust, and continuous self-optimization of the supply chain meta-network.

[0073] S6. When the risk propagation chain reaches the node of the invested company, a penetrating risk warning report is generated based on the complete structure of the risk propagation chain.

[0074] In a specific embodiment of the present invention, the specific steps for generating a penetrating risk warning report based on the complete structure of the risk propagation chain include: extracting the sequence of nodes and edges from the risk source node to the invested enterprise node from the finally generated risk propagation chain to form a risk transmission path.

[0075] In the risk transmission path, identify the edges where the conduction impedance label value changes abruptly and their connection points to generate a list of key bottleneck nodes.

[0076] By combining the risk transmission path and the list of key bottleneck nodes, the potential impact of risk events on the nodes of the invested companies is quantitatively calculated.

[0077] The risk transmission path, a list of key bottleneck nodes, and the degree of potential impact are integrated to generate a penetrating risk warning report.

[0078] Specifically, the step of generating a penetrating risk warning report based on the final generated risk propagation chain in this invention aims to transform the complex analysis process and data within the system into a highly structured and interpretable intelligence product that provides direct decision support value for investment managers.

[0079] The first step of the project aims to clearly present the complete propagation path of the risk. When the system determines that the risk propagation chain has reached the invested company node, it triggers the report generation module. The report generation module first calls the path extraction algorithm. This algorithm traverses the finally confirmed risk propagation chain data structure, starting from the risk source node and extracting the unique identifiers of all traversed nodes and the identifiers of the edges connecting them, according to the node connection order. These identifier sequences, after being formatted, constitute the risk transmission path in the early warning report, intuitively answering the question, "Where did the risk come from, and who did it pass through?"

[0080] The second step of the process aims to identify and highlight vulnerable links in the risk transmission path. The report generation module then executes a bottleneck identification procedure. This procedure traverses each edge in the risk transmission path, reading its associated conduction impedance label value within the supply chain meta-network. Simultaneously, it calculates the average and standard deviation of the conduction impedance labels across the entire path. The procedure marks edges with conduction impedance labels higher than the average—for example, exceeding the average by more than one standard deviation—and their connected upstream and downstream nodes as critical bottleneck nodes. These nodes are specifically highlighted in the report because they represent key locations where risk may be amplified or abruptly change during transmission.

[0081] The third step of the project aims to quantify and assess the potential economic losses caused by risk events to the ultimate goal. The module initiates an impact assessment model. This model first extracts procurement dependence data for each node along the risk transmission path from the attribute database of the invested enterprise node; that is, the percentage of its total costs procured from each node. Then, based on the type of risk event at the risk source node, the model matches an initial impact coefficient from a pre-set risk scenario library; for example, the initial impact coefficient for a core component supplier's production halt is 0.8. This impact coefficient is attenuated and corrected by the risk buffering capacity parameters of each node along the risk transmission path. Finally, the model estimates the potential impact level using the following simplified formula: In this formula, Representing risk to the target investee company The ultimate potential impact is an estimate representing a percentage decrease in revenue or profit. Risk source node The initial influence coefficient. Representative target company For risk sources The overall procurement dependence is a value that combines direct and indirect procurement relationships. Target company Its own risk buffering capacity parameter.

[0082] The fourth step aims to integrate all the above analysis results into a complete report. The system will extract the risk transmission paths, list the marked key bottleneck nodes, and calculate the potential impact, filling in the information according to a preset report template. The report will ultimately be output in the form of a visual graph combined with a text summary. Users can clearly see the dynamic path of risk propagation and quickly obtain detailed information on key nodes and a quantified risk exposure assessment, thus completing the entire penetrating risk early warning process.

[0083] In a specific embodiment of the present invention, the method further includes: when multiple risk propagation chains exist simultaneously, calculating the total risk gravity value for each risk propagation chain by accumulating the risk gravity values ​​of each node on each risk propagation chain.

[0084] Multiple risk propagation chains are sorted according to their total risk gravity value to generate a risk propagation chain priority queue.

[0085] Based on the priority queue of the risk propagation chain, the real-time data monitoring resources allocated to each risk propagation chain are configured differently.

[0086] Specifically, the step of differentiated resource allocation in this invention when multiple risk propagation chains exist simultaneously aims to ensure that, in complex scenarios facing concurrent risk events, the system's limited monitoring resources can be dynamically and prioritized for allocation to the most threatening risk transmission path, thereby achieving continuous focus on key risks.

[0087] The first step of the project aims to quantitatively assess the threat level of all currently active risk propagation chains. The system will periodically, for example, every 15 minutes, launch a concurrent risk assessment program. This program first retrieves all instances of risk propagation chains in the system memory that are in a "growing" state. For each risk propagation chain, the program performs a cumulative summation calculation to obtain an indicator representing its current overall threat level, namely the total risk gravity value. This total risk gravity value comprehensively reflects the cumulative propagation momentum of the risk propagation chain since its source; the higher the value, the stronger the penetration power of the risk event represented by the chain and the smoother the propagation path.

[0088] The second step of the project aims to assign a resource priority to each risk propagation chain based on the quantified threat level. After calculating the total risk gravity of all active risk propagation chains, the concurrent risk assessment program sorts these totals in descending order. Based on the sorting results, the system classifies all risk propagation chains into different levels; for example, the top 10% are classified as "high-risk," 10% to 50% as "medium-risk," and the rest as "low-risk." Each level corresponds to a preset real-time data monitoring resource priority coefficient.

[0089] The third step of the project aims to execute specific resource reallocation actions. The system transmits the priority coefficients of each risk propagation chain to the data acquisition strategy management module. Based on these coefficients, the data acquisition strategy management module dynamically adjusts the monitoring resource quotas allocated to each node on the chain. For example, for a "high-risk" risk propagation chain, the monitoring frequency of its nodes may be increased by an additional 50% on top of the enhanced or maximum frequency. For a "low-risk" chain, its monitoring frequency may be moderately reduced, but still above the baseline level. In this way, the system ensures that when multiple risks evolve simultaneously, the most valuable real-time monitoring resources, such as computing power and API calls, can always prioritize serving the risk chains with the strongest propagation momentum and the greatest potential threat, achieving intelligent resource optimization and focus in complex situations.

[0090] Reference Figure 2 The second aspect of the present invention provides a penetration-based risk monitoring system for invested enterprises based on supply chain transmission, including: a supply chain meta-network generation module, a risk gravity value calculation module, a risk propagation chain dynamic generation module, a monitoring resource scheduling feedback module, a two-way feedback and learning module, and a risk warning report generation module.

[0091] The supply chain meta-network generation module is connected to the risk gravity value calculation module, the risk gravity value calculation module is connected to the risk propagation chain dynamic generation module, the risk propagation chain dynamic generation module is connected to the monitoring resource scheduling feedback module, the supply chain meta-network generation module, the risk propagation chain dynamic generation module, and the monitoring resource scheduling feedback module are all connected to the two-way feedback and learning module, and the two-way feedback and learning module is connected to the risk early warning report generation module.

[0092] The supply chain meta-network generation module constructs a supply chain meta-network with node-to-node relationships, calculates and assigns dynamic conduction impedance labels to the node relationships to quantify the resistance to risk propagation, and generates a supply chain meta-network with conduction impedance labels.

[0093] The risk gravity value calculation module, in response to the risk event signal identified by the risk source node in the supply chain meta-network, calculates the risk gravity value from the risk source node to its associated node based on the risk gravity model used to calculate the risk propagation tendency and in combination with the conduction impedance label.

[0094] The risk propagation chain dynamic generation module, based on the risk gravity value, selects a path and connects nodes starting from the risk source node, dynamically generating one or more risk propagation chains pointing to the invested company node.

[0095] The monitoring resource scheduling and feedback module dynamically adjusts the data collection strategy according to the composition of the risk propagation chain, focuses real-time data monitoring resources on the nodes in the risk propagation chain, and obtains real-time status feedback data of the nodes in the risk propagation chain.

[0096] The two-way feedback and learning module uses real-time status feedback data to adjust the growth and structure of the risk propagation chain, and generates transmission effect feedback information based on the final transmission result of the risk propagation chain. It then uses the transmission effect feedback information to update the transmission impedance labels of the corresponding nodes in the supply chain meta-network.

[0097] The risk warning report generation module generates a penetrating risk warning report based on the complete structure of the risk propagation chain when the risk propagation chain reaches the node of the invested company.

[0098] The core of this invention lies in establishing a standardized and quantifiable supply chain network risk assessment process through a risk gravity model and dynamic transmission impedance labeling. This standardized service enables investment institutions to screen and issue penetrating risk warning reports. Essentially, it provides a universal post-investment evaluation system for innovation and entrepreneurship and standardized risk control screening services. By constructing, implementing, and providing such a penetrating risk control management mechanism with unified quantitative evaluation standards, it achieves forward-looking early warning of supply chain transmission risks, effectively solves the problems of analysis delays and noise interference caused by massive amounts of data, achieves efficient resource utilization, and demonstrates stronger robustness and more accurate predictive capabilities when facing new or sudden risk events in the future.

[0099] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for monitoring investee company risks based on supply chain transmission, characterized in that: include: S1. Construct a supply chain meta-network with node-to-node relationships, calculate and assign dynamic conduction impedance labels to the node relationships to quantify the resistance to risk propagation, and generate a supply chain meta-network with conduction impedance labels. S2. In response to the risk event signal identified by the risk source node in the supply chain meta-network, calculate the risk gravity value from the risk source node to its associated node based on the risk gravity model used to calculate the risk propagation tendency and in combination with the conduction impedance label. S3. Based on the risk gravity value, select a path and connect nodes starting from the risk source node to dynamically generate one or more risk propagation chains pointing to the invested company node. S4. Based on the composition of the risk propagation chain, dynamically adjust the data collection strategy, focus real-time data monitoring resources on the nodes in the risk propagation chain, and obtain real-time status feedback data of the nodes in the risk propagation chain. S5. Utilize real-time status feedback data to adjust the growth and composition of the risk propagation chain, and generate propagation effect feedback information based on the final propagation result of the risk propagation chain. Use the propagation effect feedback information to update the propagation impedance label of the corresponding node association relationship in the supply chain meta-network in reverse. S6. When the risk propagation chain reaches the node of the invested company, a penetrating risk warning report is generated based on the complete structure of the risk propagation chain.

2. The investee company penetration risk monitoring method based on supply chain transmission as described in claim 1, characterized in that, The specific steps for calculating node relationships and assigning dynamic conduction impedance tags to quantify risk propagation resistance, thereby generating a supply chain meta-network with conduction impedance tags, include: Acquire enterprise entity data and transaction data representing supply relationships between enterprises to generate the raw dataset; Based on the original dataset, enterprise entities are mapped as nodes and supply relationships are mapped as edges to construct an initial supply chain network; Based on the assessment rules used to evaluate the difficulty of risk transmission, an initial conduction impedance value is calculated for each edge of the initial supply chain network, and the initial conduction impedance value is assigned as a conduction impedance label to the corresponding edge, forming a supply chain meta-network with conduction impedance labels.

3. The investee company penetration risk monitoring method based on supply chain transmission as described in claim 1, characterized in that, The specific steps for calculating the risk gravity value from the risk source node to its associated node include: Obtain data on the business correlation strength between the risk source node and each of its associated nodes; Obtain the risk buffer capacity parameter for each associated node, which characterizes its ability to withstand risks; The business association strength data, risk buffer capacity parameters, and conduction impedance labels of the edges connecting the risk source node and the associated node are input into the risk gravity model for calculation, and the risk gravity value is output.

4. The investee company penetration risk monitoring method based on supply chain transmission as described in claim 3, characterized in that, The risk gravity model is configured such that the output risk gravity value is positively correlated with the input business correlation strength data, and negatively correlated with the input risk buffering capacity parameter and conduction impedance label.

5. The investee company penetration risk monitoring method based on supply chain transmission according to claim 1, characterized in that, The specific steps for obtaining real-time status feedback data of nodes on the risk propagation chain include: Identify all nodes in the risk propagation chain and mark the latest extended node in the risk propagation chain as the growth front node; Based on all nodes and the growth front node, a differentiated data collection strategy containing different monitoring priorities is generated; Based on a differentiated data acquisition strategy, data acquisition operations are performed, and the acquired raw data is processed into structured real-time status feedback data.

6. The investee company penetration risk monitoring method based on supply chain transmission according to claim 1, characterized in that, The specific steps for adjusting the growth and structure of the risk propagation chain using real-time status feedback data include: Update the risk buffering capacity parameters of the growth front nodes on the risk propagation chain based on real-time status feedback data. Using the updated risk buffer capacity parameters, the risk gravity value starting from the growth front node is recalculated; Based on the recalculated risk gravity value, it is determined whether the risk propagation chain stops growing, changes its path, or continues to extend, thereby achieving dynamic adjustment of the risk propagation chain.

7. The investee company penetration risk monitoring method based on supply chain transmission as described in claim 6, characterized in that, The specific steps for using conduction effect feedback information to reverse update the conduction impedance label of the corresponding node association in the supply chain meta-network include: From the feedback information on the transmission effect, distinguish between the edges that successfully transmit risk and the edges that block the transmission of risk; Based on the learning rate used to control the model's learning rate, the conduction impedance label value of the edge that successfully conducts the risk is lowered. Based on the preset base learning rate, the conduction impedance label value of the edge blocking the transmission of risk is increased, and the magnitude of the increase is adjusted in combination with the event confidence factor that characterizes the degree of determinism of the blocking.

8. The investee company penetration risk monitoring method based on supply chain transmission according to claim 1, characterized in that, The specific steps for generating a penetrating risk warning report based on the complete structure of the risk propagation chain include: From the final generated risk propagation chain, extract the sequence of nodes and edges from the risk source node to the invested company node to form the risk transmission path; In the risk transmission path, identify the edges where the conduction impedance label value changes abruptly and their connection points to generate a list of key bottleneck nodes; By combining the risk transmission path and the list of key bottleneck nodes, the potential impact of risk events on the nodes of the invested companies is quantitatively calculated. The risk transmission path, a list of key bottleneck nodes, and the degree of potential impact are integrated to generate a penetrating risk warning report.

9. The investee company penetration risk monitoring method based on supply chain transmission according to claim 1, characterized in that, Also includes: When multiple risk propagation chains exist simultaneously, the total risk gravity value of each risk propagation chain is calculated by summing the risk gravity values ​​of each node on each risk propagation chain. Multiple risk propagation chains are sorted according to the total risk gravity value to generate a risk propagation chain priority queue; Based on the priority queue of the risk propagation chain, the real-time data monitoring resources allocated to each risk propagation chain are configured differently.

10. A portfolio company penetration-based risk monitoring system based on supply chain transmission, characterized in that: include: The supply chain meta-network generation module constructs a supply chain meta-network with node-to-node relationships, calculates and assigns dynamic conduction impedance labels to the node relationships to quantify the resistance to risk propagation, and generates a supply chain meta-network with conduction impedance labels. The risk gravity value calculation module, in response to the risk event signal identified by the risk source node in the supply chain meta-network, calculates the risk gravity value from the risk source node to its associated node based on the risk gravity model used to calculate the risk propagation tendency and in combination with the conduction impedance label. The risk propagation chain dynamic generation module, based on the risk gravity value, selects a path and connects nodes starting from the risk source node, dynamically generating one or more risk propagation chains pointing to the invested company node. The monitoring resource scheduling and feedback module dynamically adjusts the data collection strategy according to the composition of the risk propagation chain, focuses real-time data monitoring resources on the nodes in the risk propagation chain, and obtains real-time status feedback data of the nodes in the risk propagation chain. The two-way feedback and learning module uses real-time status feedback data to adjust the growth and composition of the risk propagation chain, and generates transmission effect feedback information based on the final transmission result of the risk propagation chain. It uses the transmission effect feedback information to update the transmission impedance label of the corresponding node relationship in the supply chain meta-network in reverse. The risk warning report generation module generates a penetrating risk warning report based on the complete structure of the risk propagation chain when the risk propagation chain reaches the node of the invested company.

Images