Supply chain risk quantitative evaluation method and system based on dynamic affair graph

Abstract

The invention relates to the technical field of risk analysis, in particular to a supply chain risk quantitative evaluation method and system based on a dynamic affair atlas, and the method comprises the steps: collecting multi-source heterogeneous data, constructing a four-dimensional space-time model comprising a time dimension, a geographic space dimension, a supply chain network space dimension and a risk influence space dimension, representing the supply chain event as four-dimensional spatio-temporal data; a supply chain entity is identified from the four-dimensional spatio-temporal data, risk events are extracted, a affair graph is constructed, and the affair graph takes the risk events as nodes and the evolution relation between the events as edges to calculate the relation weight between the events; calculating a probability quantized value of the risk conduction path based on the affair map, and obtaining a comprehensive risk score of the target entity; generating a risk mitigation strategy based on the comprehensive risk score; and monitoring the deviation between the actual risk occurrence condition and the prediction result, and updating the affair map and the risk mitigation strategy through adaptive parameter optimization and an incremental learning mechanism to form a self-evolutionary risk assessment system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of risk analysis, in particular to a supply chain risk quantitative evaluation method and system based on a dynamic event logic graph. BACKGROUND

[0002] With the increasing complexity of global supply chain networks, supply chain risk assessment has become an important part of enterprise management and financial regulation. In the prior art, historical data analysis, scorecard models, and other methods are used to assess supply chain risks, or risk assessment models are constructed using enterprise financial indicators, credit records, industry data, and other information. However, the above methods have obvious limitations: first, there is a lack of analysis of the evolution of risk events, and the dynamic process of risk transmission in the supply chain cannot be obtained; second, the risk quantification process is mainly based on historical statistical data, and the ability to identify new data is insufficient; third, there is a lack of effective representation of the causal and temporal relationships between multi-dimensional risk factors; event logic graphs, as a knowledge representation method, construct directed graph structures with events as nodes and evolution relationships between events as edges, and can describe the logical relationships of events such as succession, causality, and conditions. However, existing event logic graph technology lacks event abstraction modeling methods and risk quantification evaluation mechanisms for supply chain scenarios, and cannot meet the demand for accurate assessment and early warning of supply chain risks in actual business.

[0003] Therefore, a supply chain risk quantitative evaluation method and system based on a dynamic event logic graph are proposed. SUMMARY

[0004] The purpose of the present application is to provide a supply chain risk quantitative evaluation method and system based on a dynamic event logic graph, which includes collecting multi-source heterogeneous data, constructing a four-dimensional space-time model containing time dimension, geographical space dimension, supply chain network space dimension, and risk impact space dimension, and representing supply chain events as four-dimensional space-time data; identifying supply chain entities from the four-dimensional space-time data based on an entity recognition algorithm, extracting risk events, and constructing an event logic graph, which takes risk events as nodes and evolution relationships between events as edges, and calculates the relationship weights between events; calculating the probability quantization value of the risk transmission path based on the event logic graph, and obtaining the comprehensive risk score of the target entity; generating a risk mitigation strategy based on the comprehensive risk score; monitoring the deviation between the actual risk occurrence and the prediction result, and updating the event logic graph and the risk mitigation strategy through an adaptive parameter optimization and incremental learning mechanism to form a self-evolving risk assessment system.

[0005] To achieve the above purpose, the present application provides the following technical solutions:

[0006] The supply chain risk quantitative evaluation method and system based on a dynamic event logic graph include:

[0007] Collecting multi-source heterogeneous data, performing space-time coordinate labeling on the multi-source heterogeneous data, constructing a four-dimensional space-time model including time dimension, geographical space dimension, supply chain network space dimension and risk influence space dimension, and representing a supply chain event as four-dimensional space-time data;

[0008] Identifying a supply chain entity from the four-dimensional space-time data based on an entity recognition algorithm, extracting a risk event, and constructing a fact-reason graph, the fact-reason graph taking a risk event as a node, an evolution relationship between events as an edge, and a dynamic weight updating mechanism to calculate the relationship weight between events;

[0009] Based on the fact-reason graph, calculating the probability of a risk transmission path and obtaining a comprehensive risk score of a target entity;

[0010] Based on the comprehensive risk score, generating a risk mitigation strategy, the risk mitigation strategy including a supplier replacement strategy, an inventory adjustment strategy, and a transportation route optimization strategy;

[0011] Monitoring the deviation between the actual risk occurrence and the prediction result, updating the fact-reason graph and the risk mitigation strategy through an adaptive parameter optimization and an incremental learning mechanism, and forming a self-evolving risk assessment system.

[0012] Preferably, constructing the four-dimensional space-time model includes: converting multi-source heterogeneous data into a unified RDF triple format using standardization; the multi-source heterogeneous data includes business registration data, judicial litigation data, tax invoice data, and logistics transportation data; extracting timestamp information for each type of data and standardizing it into time coordinates, converting address information into latitude and longitude coordinates through a geographic coding service, mapping network topology coordinates according to the upstream and downstream positions of entities in the supply chain, and determining the risk influence radius based on the risk event type and the influence range; establishing a four-dimensional R-tree index structure to support space-time range queries and obtain four-dimensional space-time data.

[0013] Preferably, the fact-reason graph construction includes:

[0014] Entity recognition: performing named entity recognition on the text content in the four-dimensional space-time data to identify supplier, manufacturer, product, raw material, transportation route, and warehouse supply chain entity types;

[0015] Risk event extraction: based on dependency syntax analysis and semantic role labeling technology, extracting risk events such as supply interruption, quality problem, price fluctuation, and logistics delay from the four-dimensional space-time data, and representing each risk event as a four-tuple structure including risk source entity, risk type, impact object, and severity;

[0016] Relationship extraction: identifying the following relationships between risk events through text matching and rule-based reasoning: sequential relationship, causal relationship, and conditional relationship, and constructing directed edges to connect related event nodes;

[0017] Dynamic weight calculation: Calculate time decay weight combined with the time interval between events, calculate spatial influence weight according to four-dimensional spatial distance, calculate structure weight based on entity network centrality, and obtain the dynamic weight value of the edge between events.

[0018] Preferably, the comprehensive risk score acquisition step comprises:

[0019] Search all transmission paths from risk source events to target entities in the event graph, record the intermediate nodes and edge weights of each path; Calculate the transmission probability of a single path based on the dynamic weight of all edges on the path; Calculate the joint influence probability of multiple paths on the target entity, and process the overlapping influence between paths based on conditional independence assumption; Combine time urgency, spatial coverage, network criticality, influence severity and propagation speed five-dimensional indicators, and calculate the comprehensive risk score of the target entity by weighted sum of transmission probability and joint influence probability, and map it to low, medium, high and extremely high four risk levels.

[0020] Preferably, the generation of risk mitigation strategy comprises: modeling the suppliers, manufacturers and distributors in the supply chain as independent agents with state space, action space and reward function; Each agent determines its state risk level based on the current comprehensive risk score, and selects the optimal action from the pre-defined action set;

[0021] Supplier replacement strategy generation step: According to the current supplier risk level, set the screening condition, evaluate the multi-objective optimization of cost, quality and reliability of the candidate supplier, output the optimal replacement scheme and switching schedule;

[0022] Inventory adjustment strategy generation step: Calculate the inventory safety factor based on the risk level, combine the historical demand data and the supply lead time, and use the economic order quantity model to calculate the optimal inventory level;

[0023] Transportation route optimization strategy generation step: Analyze the risk nodes of the current transportation path, and use the shortest path algorithm to replan the alternative transportation route to avoid high-risk areas.

[0024] Preferably, the self-evolving risk assessment system update comprises:

[0025] Set the sliding time window, collect the actual risk event data in real time, calculate the absolute deviation of the prediction probability and the actual occurrence probability, the deviation of the prediction time and the actual time, the deviation of the prediction severity and the actual severity, when the comprehensive deviation exceeds the first threshold, adjust the time decay coefficient, the space influence coefficient and the network weight coefficient, when the comprehensive deviation exceeds the second threshold, update the event graph structure, add new event nodes and relationship edges, and recalculate the probability of risk transmission path, after each model update, perform performance verification, if the verification score is lower than the performance threshold, roll back to the previous version.

[0026] The supply chain risk quantitative evaluation system based on the dynamic event graph map comprises:

[0027] The space-time data acquisition unit is used for collecting multi-source heterogeneous data, and performing space-time coordinate labeling, constructing a four-dimensional space-time model, and representing supply chain events as four-dimensional space-time data. The event graph construction unit identifies supply chain entities from the four-dimensional space-time data, extracts risk events, and constructs an event graph. The risk assessment unit calculates the probability of risk transmission path based on the event graph, and obtains the comprehensive risk score of the target entity. The risk strategy generation unit generates a risk mitigation strategy based on the comprehensive risk score. The risk system updating unit monitors the deviation between the actual risk occurrence and the prediction result, updates the event graph and the risk mitigation strategy, and forms a self-evolving risk evaluation system.

[0028] Compared with the prior art, the beneficial effects of the present application are:

[0029] 1、The present application introduces a four-dimensional space-time model containing time dimension, geographical space dimension, supply chain network space dimension and risk influence space dimension. Through standardized processing of multi-source heterogeneous data, data from multiple fields such as industry and commerce, justice, taxation and logistics can be fused and modeled under the same space-time semantic framework to form structured four-dimensional space-time data. This model can dynamically and real-timely represent the space-time distribution and network location relationship of risk events, providing high-precision and queryable data basis for subsequent risk propagation analysis and countermeasure formulation, and improving the modeling efficiency and risk data processing capability in large-scale data environment.

[0030] 2, The application proposes a technical route for constructing a dynamic event logic graph based on risk events, and establishes a causal evolution network between risk events through entity recognition, risk event extraction, relationship recognition and dynamic edge weight updating. The dynamic quantification of the transmission and influence between events makes the event logic graph have real explainability and reasoning feasibility in representing the risk propagation path. Combined with the five indexes of time urgency, space coverage, network criticality, propagation speed and influence severity, the risk exposure degree of the target entity is comprehensively scored and graded, which adapts to the trend of dynamic evolution of risk with time and space, and has a significant improvement effect in risk quantization accuracy and propagation mechanism restoration.

[0031] 3, The application introduces multi-agent modeling, and regards the suppliers, manufacturers and distributors in the supply chain as independent agents with state space, action space and reward function, respectively, selects the optimal risk response behavior according to the risk state, and generates individualized and executable mitigation strategies. The strategies such as supplier replacement, inventory adjustment and transportation path optimization form a dynamic response scheme with strong operability through the linkage analysis of the current risk level, historical data and business objectives, and use multi-objective optimization and shortest path algorithms to enhance the elastic recovery ability of the supply chain when facing risks. The application has the ability of sustainable learning, self-correction and real-time iteration of strategies, which improves the intelligent level of risk control and the ability to adapt to complex environmental changes. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 The supply chain risk quantitative evaluation method based on the dynamic event logic graph provided by the application is shown in the flowchart;

[0033] Figure 2 The supply chain risk quantitative evaluation system structure diagram based on the dynamic event logic graph provided by the application is shown in the flowchart;

[0034] Figure 3 The comprehensive risk score acquisition step diagram provided by the application is shown in the flowchart. DETAILED DESCRIPTION

[0035] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0036] The supply chain risk quantitative evaluation method based on the dynamic event logic graph provided by the application is applied to the supply chain risk quantitative evaluation system based on the dynamic event logic graph, and the method flowchart is referred to Figure 1The system structure diagram includes a space-time data acquisition unit, a matter logic graph construction unit, a risk assessment unit, a risk strategy generation unit, and a risk system updating unit, and specific reference is made to Figure 2 The specific technical solutions are as follows:

[0037] Multi-source heterogeneous data is collected, the multi-source heterogeneous data is marked with space-time coordinates, a four-dimensional space-time model containing time dimension, geographical space dimension, supply chain network space dimension and risk influence space dimension is constructed, and a supply chain event is represented as four-dimensional space-time data;

[0038] The four-dimensional space-time model is constructed by converting multi-source heterogeneous data into a unified RDF triple format through standardization; the multi-source heterogeneous data includes business registration data, judicial litigation data, tax invoice data and logistics transportation data; timestamp information is extracted for each type of data and standardized as time coordinates T, address information is converted into longitude and latitude coordinates (X, Y) through a geographic coding service, network topology coordinates N are mapped according to the upstream and downstream positions of entities in the supply chain, and risk influence radius R is determined based on the type and influence range of the risk event; the above four dimensions are combined to form (T, (X, Y), N, R) coordinates, a four-dimensional R-tree index structure is established to support space-time range queries, and four-dimensional space-time data is obtained.

[0039] In this embodiment, by constructing a four-dimensional space-time model containing time, geographical space, supply chain network and risk influence, effective integration and analysis of heterogeneous data such as business registration, judicial litigation, tax invoice and logistics transportation are realized. The RDF triple standardized data is combined with the four-dimensional R-tree index structure to support efficient space-time range queries, significantly improving the accuracy and timeliness of supply chain risk assessment and providing strong support for enterprise decision-making.

[0040] The four-dimensional R-tree index structure is optimized by using a distributed edge computing architecture, including: designing a geographic location-based sharding strategy, dividing global supply chain data into multiple geographic partitions according to longitude and latitude, and deploying edge computing nodes in each partition; constructing a hierarchical index structure, including a global space-time index layer, a regional aggregation index layer and a local real-time index layer, supporting second-level risk event retrieval; implementing a dynamic load balancing mechanism to adaptively adjust the allocation of computing resources according to the risk event density and query frequency of each region; ensuring data consistency between distributed nodes, using the PBFT algorithm to handle node failures and network partition problems; designing a streaming data processing engine to support incremental index updating and rapid propagation of real-time risk event streams.

[0041] Based on an entity recognition algorithm, supply chain entities are identified from the four-dimensional space-time data, risk events are extracted, and a matter logic graph is constructed, the matter logic graph taking risk events as nodes and evolution relationships between events as edges, and using a dynamic weight updating mechanism to calculate the relationship weight between events;

[0042] The matter graph construction includes:

[0043] Entity recognition: perform named entity recognition on the text content in the four-dimensional spatio-temporal data to identify supplier, manufacturer, product, raw material, transportation route, and warehouse supply chain entity types;

[0044] Risk event extraction: based on dependency syntax analysis and semantic role labeling technology, extract risk events of supply interruption, quality problem, price fluctuation, and logistics delay from the four-dimensional spatio-temporal data, each risk event is represented as a four-tuple structure containing risk source entity, risk type, impact object, and severity;

[0045] Relationship extraction: identify the following relationships between risk events through text matching and rule reasoning: sequential relationship, causal relationship, and conditional relationship, and construct directed edges connecting related event nodes;

[0046] Dynamic weight calculation: combine the time interval between events to calculate the time decay weight, calculate the spatial influence weight according to the four-dimensional spatial distance, and calculate the structure weight based on the entity network centrality, and comprehensively obtain the dynamic weight value of the edge between events.

[0047] In this embodiment, by accurately identifying supply chain entities and extracting risk events from four-dimensional spatio-temporal data, a matter graph is constructed with risk events as nodes and evolution relationships between events as edges, realizing the structured representation and dynamic analysis of supply chain risks. Adopting a dynamic weight updating mechanism, considering time, space and structure factors, accurately calculating the relationship weight between events, improving the timeliness and accuracy of risk assessment. Provide efficient risk management support for enterprises, help to discover and respond to potential risks in time, and ensure the stable operation of the supply chain.

[0048] Based on the matter graph, the probability of the risk transmission path is calculated, and the comprehensive risk score of the target entity is obtained;

[0049] The comprehensive risk score acquisition step refers to Figure 3 , including:

[0050] Search all transmission paths from the risk source event to the target entity in the event graph, record the intermediate nodes and edge weights passed by each path; obtain the transmission probability of a single path based on the conditional probability formula of the dynamic weight of all edges on the path; calculate the joint influence probability of multiple paths on the target entity, and process the overlapping influence between paths based on the conditional independence assumption; combine the time urgency, spatial coverage, network criticality, influence severity and propagation speed five-dimensional indicators, and use weighted summation to calculate the comprehensive risk score of the target entity based on the transmission probability and the joint influence probability, and map it to four risk levels of low, medium, high and extremely high; in this embodiment, the comprehensive risk of the target entity is evaluated by calculating the risk transmission path probability based on the event graph and combining multi-dimensional indicators. By searching all paths, calculating single path probability, and processing the joint influence of multiple paths, the comprehensiveness and accuracy are ensured. The comprehensive risk score is mapped to four risk levels, which improves the efficiency of risk management and the stability of the supply chain.

[0051] Based on the comprehensive risk score, a risk mitigation strategy is generated, which includes a supplier replacement strategy, an inventory adjustment strategy, and a transportation route optimization strategy;

[0052] The generation of the risk mitigation strategy includes: modeling the suppliers, manufacturers and distributors in the supply chain as independent agents with state space, action space and reward function; each agent determines its state risk level based on the current comprehensive risk score and selects the optimal action from the predefined action set;

[0053] The supplier replacement strategy generation step: according to the current supplier risk level, set the screening conditions, evaluate the multi-objective optimization of cost, quality and reliability of the candidate suppliers, and output the optimal replacement scheme and switching schedule;

[0054] The inventory adjustment strategy generation step: calculate the inventory safety factor based on the risk level, combine the historical demand data and the supply lead time, and use the economic order quantity model to calculate the optimal inventory level;

[0055] The transportation route optimization strategy generation step: analyze the risk nodes of the current transportation path, and use the shortest path algorithm to replan an alternative transportation route that avoids high-risk areas.

[0056] In this embodiment, by constructing an agent model, a risk mitigation strategy is generated based on the comprehensive risk score to improve the efficiency of supply chain management. The supplier replacement strategy selects candidate suppliers based on the risk level, performs multi-objective optimization of cost, quality and reliability, quickly outputs the optimal replacement scheme, and reduces the risk of interruption; the inventory adjustment strategy uses risk level and historical data to calculate safety factor and optimize inventory level, balancing sudden demand and cost control; the transportation route optimization strategy analyzes the risk nodes and uses the shortest path algorithm to plan an alternative route, enhancing transportation stability.

[0057] The intelligent agent is enhanced modeling by digital twin technology, including: creating a digital twin model for the supply chain, synchronizing the state, behavior and environmental parameters of the physical entity in real time; establishing a real-time data acquisition system based on an Internet of Things sensor network, including RFID tags, GPS locators, temperature and humidity sensors, vibration sensors, forming a two-way mapping of physical-digital space; designing a predictive digital twin algorithm, combining physical modeling, data-driven modeling and hybrid modeling to predict the future state evolution trajectory of the entity; constructing a virtual interaction protocol between twins, simulating the complex interaction between supply chain entities, supporting risk propagation simulation in multiple scenarios; realizing an autonomous decision-making mechanism based on digital twins, and the intelligent agent can pre-empt the execution effect of the risk mitigation strategy in the virtual environment, and optimize the decision parameters before applying them to the supply chain.

[0058] Monitoring the deviation between actual risk occurrence and prediction results, updating the affair-logic graph and risk mitigation strategy through adaptive parameter optimization and incremental learning mechanism, forming a self-evolving risk assessment system.

[0059] The self-evolving risk assessment system update includes:

[0060] Setting a sliding time window, collecting real-time risk event data, calculating the absolute deviation of predicted probability and actual occurrence probability, the deviation of predicted time and actual time, and the deviation of predicted severity and actual severity; wherein the absolute deviation of predicted probability and actual occurrence probability, the deviation of predicted time and actual time, and the deviation of predicted severity and actual severity are weighted and summed to obtain a comprehensive deviation;

[0061] When the comprehensive deviation exceeds a first threshold, adjust the time decay coefficient, the spatial influence coefficient and the network weight coefficient; for the comprehensive deviation exceeding a second threshold, update the affair-logic graph structure, add new event nodes and relationship edges, and recalculate the probability of the risk transmission path; after each model update, perform performance verification, and if the verification score is lower than the performance threshold, roll back to the previous version.

[0062] In this embodiment, through the self-evolving risk assessment system, dynamic monitoring and prediction of supply chain risks are realized. The sliding time window is used to collect data in real time, calculate the prediction and actual deviation, automatically adjust the model parameters and update the affair-logic graph, ensuring the timeliness and accuracy of risk assessment. The incremental learning mechanism reduces the computational overhead, and the performance verification and rollback mechanism ensures system stability. The system can actively identify emerging risks, optimize risk mitigation strategies, and improve supply chain resilience and decision-making efficiency.

[0063] An intelligent decision mechanism based on a graph neural network includes: constructing a graph attention network-based matter graph embedding model to automatically learn the latent feature representation of event nodes and relationship edges; designing a multi-agent deep reinforcement learning framework, modeling each node of the supply chain as a collaborative agent, and optimizing the global risk control strategy through a shared reward mechanism; establishing an adversarial generative network-based risk scenario generator to simulate the supply chain response capability under extreme risk conditions and generate risk propagation paths for strategy robustness verification; and realizing cross-domain risk knowledge transfer to enable the model to quickly adapt to emerging risk types and industry-specificities.

[0064] The application integrates multi-source heterogeneous data by constructing a four-dimensional space-time model, combines matter graphs with graph neural networks to learn risk features, and improves the accuracy and timeliness of supply chain risk assessment. Multi-agent deep reinforcement learning optimizes the global risk control strategy, adversarial generative networks and digital twin technology enhance the robustness and predictive ability of the strategy, and meta-learning enables cross-domain knowledge transfer. Distributed four-dimensional R-tree indexing and blockchain consensus mechanisms support millisecond-level queries and data consistency, adaptive parameter optimization and incremental learning form a self-evolution system, and quickly adapt to emerging risks.

[0065] Embodiment two:

[0066] Multi-source heterogeneous data including business registration data, judicial litigation data, tax invoice data, and logistics transportation data are collected. Based on the data standardization engine technology of Data Fusion Intelligence, the above heterogeneous data is converted into a unified RDF triple format.

[0067] In the specific implementation process, the business registration data includes fields such as enterprise basic information, equity structure, and business scope; the judicial litigation data covers case types, party information, and judgment results; the tax invoice data records transaction amounts, commodity categories, and transaction times; and the logistics transportation data includes transportation routes, cargo types, and transportation time efficiency. Through OCR recognition technology and data cleaning conversion algorithms, data quality and format uniformity are ensured.

[0068] Timestamp information is extracted for each type of data and standardized to time coordinates T, with an hour-level time accuracy that accurately reflects the timing characteristics of risk events. Through geocoding services, enterprise registration addresses, court jurisdiction addresses, and logistics node addresses are converted into latitude and longitude coordinates (X, Y) to cover geographic location information within the target area.

[0069] According to the upstream and downstream position of enterprises in the industrial chain, based on the operation data index system precipitated by Shumeng Zhilian, the enterprise is mapped to network topology coordinates N. Network topology coordinates reflect the hierarchical position and importance of enterprises in the entire industrial network. Based on the type of risk events and historical impact range data, calculate the risk impact radius R, form a complete coordinate (T, (X, Y), N, R).

[0070] A four-dimensional R-tree index structure is established to support spatio-temporal range queries within the target area. The index structure is layered according to administrative divisions, with corresponding index nodes established at the provincial, municipal, and county levels. The query response time is controlled within seconds, meeting the business needs of real-time risk monitoring.

[0071] Based on multi-dimensional evaluation technology, the text content in four-dimensional spatio-temporal data is named entity recognition. In the industrial intelligence space environment, the system can identify the following types of supply chain entities: upstream raw material suppliers, midstream manufacturing enterprises, downstream sales distributors, key product categories, core raw material categories, main transportation routes, and important warehouse facilities.

[0072] The entity recognition algorithm combines the application scenario index evaluation model precipitated by Shumeng Zhilian, which can accurately distinguish the functional positioning and importance of different types of enterprises in the industrial chain. Through association analysis with business registration information, the accuracy of entity recognition is further verified.

[0073] Based on dependency syntax analysis and semantic role labeling technology, risk events related to the industrial chain are extracted from four-dimensional spatio-temporal data. Risk event types include supply disruption, quality problem, price fluctuation, logistics delay, financial chain rupture, and policy change.

[0074] Each risk event is represented as a four-tuple structure containing risk source entities, risk types, impact objects, and severity levels. For example, the event of a raw material supplier stopping production due to environmental problems is represented as a four-tuple structure: (Supplier A, Supply Disruption, Downstream Manufacturing Enterprises B, C, D, High Risk Level). The severity level is quantitatively evaluated based on historical data and expert decision systems.

[0075] Through text matching and rule-based reasoning technology, the continuity relationship, causal relationship, and conditional relationship between risk events are identified. In the industrial intelligence space of Shumeng Zhilian, the relationship extraction algorithm can identify complex associations across regions and industries, and construct directed edges connecting related event nodes.

[0076] The dynamic weight calculation comprehensively considers three dimensions: the time decay weight is calculated according to the time interval between events, the shorter the time interval, the higher the weight; the spatial influence weight is calculated based on four-dimensional spatial distance, the closer the geographical location and the closer the industry chain level, the higher the weight of the event; the structure weight is calculated based on the centrality of entities in the network, and the events related to core enterprises have higher propagation weight.

[0077] In the constructed rationale map, the system uses a depth-first search algorithm to find all possible transmission paths from the risk source event to the target entity. During the search process, the intermediate nodes and corresponding edge weight values of each path are recorded to ensure the completeness and accuracy of the path search.

[0078] According to the regional characteristics of the number of fusion intelligence services, the system can identify cross-provincial and cross-city risk transmission paths to provide decision support for regional industrial coordinated development. The search depth is set to six degrees of separation, which ensures the comprehensiveness of the search and controls the computational complexity within an acceptable range.

[0079] Multiply the dynamic weight values of all edges in each risk transmission path found to get the transmission probability of the path. Consider the decay effect of path length on transmission probability during the calculation process, and the longer the path, the lower the transmission probability. Through probability normalization processing, ensure that all path probability values are between 0 and 1.

[0080] When there are multiple transmission paths from the same risk source to the same target entity, use the conditional independence assumption to handle the overlapping influence between paths. Calculate the joint influence probability of multiple paths through the union formula in probability theory to avoid the problem of high probability values caused by repeated calculations.

[0081] Combine the five-dimensional indicators of time urgency, spatial coverage, network criticality, impact severity, and propagation speed to calculate the comprehensive risk score of the target entity using weighted summation. The weight coefficients are adjusted and optimized according to the practical experience of the number of fusion intelligence in different regions and the expert decision system.

[0082] Map the comprehensive risk score to four risk levels: low, medium, high, and extremely high, which correspond to different risk response strategies and resource allocation schemes. Low-risk level uses routine monitoring, medium-risk level increases monitoring frequency, high-risk level starts the early warning mechanism, and extremely high-risk level immediately executes emergency response.

[0083] Model the suppliers, manufacturers, and distributors in the industry intelligence space as independent agents with state space, action space, and reward function. The state space includes key indicators such as the financial situation, production capacity, inventory level, and order situation of the enterprise. The action space covers executable operations such as adjusting production plans, changing procurement strategies, optimizing inventory allocation, and adjusting sales strategies.

[0084] The reward function design comprehensively considers multiple objectives such as cost control, service quality, risk reduction, and maximum revenue, and quantitatively calculates them through the multi-dimensional evaluation technology of Shumujilian. Based on the current comprehensive risk score, each agent determines its state risk level and selects the optimal action combination from the predefined action set.

[0085] According to the risk level of the target supplier, the system automatically sets the screening conditions for the candidate suppliers. The screening conditions include geographical location restrictions, capacity requirements, quality standards, price ranges, and multiple dimensions. For the candidate suppliers that pass the preliminary screening, a multi-objective optimization evaluation of cost, quality, and reliability is conducted; the cost evaluation includes procurement cost, transportation cost, switching cost, and other factors; the quality evaluation is based on historical supply records, certification qualifications, customer evaluations, and other indicators; the reliability evaluation considers factors such as financial stability, production stability, and timely supply. Through a multi-objective optimization algorithm, the optimal replacement plan and detailed switching schedule are output to ensure supply chain continuity.

[0086] Based on the risk level of the target entity, the corresponding inventory safety factor is calculated. Low-risk level uses standard safety factor, medium-risk level increases safety factor, high-risk and extremely high-risk level significantly increases safety factor. Combined with historical demand data and supply lead time, the economic order quantity model is used to calculate the optimal inventory level.

[0087] The inventory adjustment strategy considers the balance between storage cost, shortage cost, and ordering cost, and determines the optimal order quantity and order point through mathematical optimization methods. The system dynamically adjusts the inventory strategy parameters based on real-time monitoring of risk changes, achieving intelligent and adaptive inventory management.

[0088] Analyze the risk nodes in the current transportation path, including traffic congestion-prone areas, natural disaster-prone areas, and strictly regulated areas. Use the shortest path algorithm to replan alternative transportation routes that avoid high-risk areas, while considering transportation cost, transportation time, transportation reliability, and other constraints.

[0089] The transportation route optimization strategy provides multiple alternative schemes, including the shortest time route, the lowest cost route, the highest reliability route, etc., for decision-makers to choose according to actual conditions. The system monitors the risk status changes of each transportation route in real time and adjusts the route planning and vehicle scheduling arrangements in a timely manner.

[0090] A thirty-day sliding time window is set to collect real-time risk event data within the service area of Shumujilian. Through data sharing mechanisms established with government departments, industry associations, and key enterprises, first-hand risk event information is obtained.

[0091] The system calculates the absolute deviation of the predicted probability and the actual occurrence probability, the deviation of the predicted time and the actual time, and the deviation of the predicted severity and the actual severity. The deviation calculation adopts a weighted average method, giving higher weight to recent events to ensure the timeliness of the evaluation results.

[0092] When the comprehensive deviation exceeds a preset first threshold, the system automatically starts a parameter optimization program. The optimization content includes adjusting the time decay coefficient, the spatial influence coefficient, and the network weight coefficient. Parameter adjustment adopts optimization algorithms such as gradient descent to determine the optimal parameter value by minimizing the prediction deviation.

[0093] During the parameter optimization process, the system retains multiple historical versions of parameter configurations, forming a parameter evolution trajectory. By comparing and analyzing the prediction effects under different parameter configurations, the optimal parameter combination scheme is identified to improve the accuracy of risk assessment.

[0094] When the comprehensive deviation exceeds a second threshold, the system performs a structure update operation of the affair graph. The update content includes adding newly discovered event nodes, establishing newly identified relationship edges, and adjusting the weight values of existing edges. The structure update adopts an incremental learning method to avoid the computational overhead caused by rebuilding the entire graph.

[0095] New event nodes are automatically generated through entity recognition and event extraction algorithms, and new relationship edges are automatically established through relationship extraction algorithms. The system simultaneously re-trains the probability prediction model, uses the updated affair graph data to train new model parameters, and improves the model's ability to identify emerging risk patterns.

[0096] After each model update, the system automatically performs performance verification testing. The verification test uses an independent test data set to evaluate the updated model's performance in accuracy, recall rate, precision, and other indicators. If the verification score is lower than the preset performance threshold, the system automatically rolls back to the previous stable version. In this scheme, all thresholds are obtained through historical data and expert experience.

[0097] The version management mechanism records the detailed information of each update, including the update time, update content, performance indicators, etc. Through version comparison and analysis, the evolution rules and trends of the model are identified to provide reference for subsequent optimization and improvement.

[0098] Although embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A supply chain risk quantification assessment method based on dynamic event graphs, characterized in that, include: Collect multi-source heterogeneous data, label the multi-source heterogeneous data with spatiotemporal coordinates, and construct a four-dimensional spatiotemporal model that includes time dimension, geographic space dimension, supply chain network space dimension and risk impact space dimension, and represent supply chain events as four-dimensional spatiotemporal data; Based on the entity recognition algorithm, supply chain entities are identified from the four-dimensional spatiotemporal data, risk events are extracted and a process graph is constructed. The process graph uses risk events as nodes and the evolutionary relationship between events as edges, and a dynamic weight update mechanism is used to calculate the relationship weight between events. Based on the aforementioned event graph, the probability of risk transmission paths is calculated to obtain the comprehensive risk score of the target entity; Based on the comprehensive risk score, risk mitigation strategies are generated, including supplier replacement strategies, inventory adjustment strategies, and transportation route optimization strategies. By monitoring the deviation between the actual occurrence of risks and the predicted results, and updating the underlying logic graph and risk mitigation strategies through adaptive parameter optimization and incremental learning mechanisms, a self-evolving risk assessment system is formed.

2. The supply chain risk quantification assessment method based on dynamic event graphs according to claim 1, characterized in that: The construction of the four-dimensional spatiotemporal model includes: standardizing multi-source heterogeneous data into a unified RDF triple format; the multi-source heterogeneous data includes business registration data, judicial litigation data, tax invoice data, and logistics and transportation data; extracting timestamp information for each type of data and standardizing it into time coordinates; converting address information into latitude and longitude coordinates through geocoding services; mapping the upstream and downstream positions of entities in the supply chain to network topology coordinates; determining the risk impact radius based on the risk event type and impact range; and establishing a four-dimensional R-tree index structure to support spatiotemporal range queries to obtain four-dimensional spatiotemporal data.

3. The supply chain risk quantification assessment method based on dynamic event graphs according to claim 1, characterized in that: The construction of the principle graph includes: Entity recognition: Named entity recognition is performed on text content in four-dimensional spatiotemporal data to identify entity types such as suppliers, manufacturers, products, raw materials, transportation routes, and warehouse supply chains; Risk event extraction: Based on dependency parsing and semantic role labeling technology, risk events such as supply disruption, quality problems, price fluctuations and logistics delays are extracted from four-dimensional spatiotemporal data. Each risk event is represented as a four-tuple structure containing risk source entity, risk type, affected object and severity. Relationship extraction: Identify sequential, causal, and conditional relationships among risk events through text matching and rule-based reasoning, and construct directed edges to connect related event entities; Dynamic weight calculation: The time decay weight is calculated by combining the time interval between events, the spatial influence weight is calculated based on the four-dimensional spatial distance, and the structural weight is calculated based on the centrality of the entity network. The dynamic weight values ​​of the edges between events are obtained by combining the results.

4. The supply chain risk quantification assessment method based on dynamic event graphs according to claim 1, characterized in that: The steps for obtaining the comprehensive risk score include: Search the event graph for all transmission paths from the risk source event to the target entity, and record the intermediate nodes and edge weights of each path. Calculate the transmission probability of a single path by applying the conditional probability formula to the dynamic weights of all edges on the path. Handle the overlapping effects between paths by assuming conditional independence, and calculate the joint impact probability of multiple paths on the target entity. Combining five dimensions of indicators—time urgency, spatial coverage, network criticality, impact severity, and propagation speed—calculate the comprehensive risk score of the target entity by weighted summation of the transmission probability and joint impact probability, and map it to four risk levels: low, medium, high, and extremely high.

5. The supply chain risk quantification assessment method based on dynamic event graph as described in claim 1, characterized in that: The risk mitigation strategy generation steps include: modeling suppliers, manufacturers, and distributors in the supply chain as independent intelligent agents with state space, action space, and reward function; each intelligent agent determines its own state risk level based on the current comprehensive risk score and selects the optimal action from a predefined set of actions; the supplier replacement strategy generation steps include: setting screening conditions based on the current supplier risk level, performing multi-objective optimization evaluation of candidate suppliers in terms of cost, quality, and reliability, and outputting the optimal replacement scheme and switching schedule; and the inventory adjustment strategy generation steps include: calculating the inventory safety factor based on the risk level, combining historical demand data and supply lead time, and using the economic order quantity model to calculate the optimal inventory level. The steps for generating a transportation route optimization strategy are as follows: Analyze the risk nodes of the current transportation route and use the shortest path algorithm to replan alternative transportation routes that avoid high-risk areas.

6. The supply chain risk quantification assessment method based on dynamic event graph as described in claim 1, characterized in that: The updated risk assessment system for self-evolution includes: A sliding time window is set to collect real-time data on actual risk events, and the absolute deviation between the predicted probability and the actual probability, the deviation between the predicted time and the actual time, and the deviation between the predicted severity and the actual severity are calculated. When the overall deviation exceeds the first threshold, the time decay coefficient, spatial influence coefficient, and network weight coefficient are adjusted. When the overall deviation exceeds the second threshold, the event graph structure is updated, new event nodes and relationship edges are added, and the probability of risk transmission paths is recalculated. Performance verification is performed after each model update, and if the verification score is lower than the performance threshold, the system is rolled back to the previous version.

7. A supply chain risk quantification assessment system based on dynamic event graphs, characterized in that, Implementing the supply chain risk quantification assessment method based on dynamic event graph as described in claim 1 includes: The system comprises the following components: a spatiotemporal data acquisition unit, which collects multi-source heterogeneous data, annotates the multi-source heterogeneous data with spatiotemporal coordinates, constructs a four-dimensional spatiotemporal model, and represents supply chain events as four-dimensional spatiotemporal data; a causal graph construction unit, which identifies supply chain entities from the four-dimensional spatiotemporal data, extracts risk events, and constructs a causal graph; a risk assessment unit, which calculates the probability of risk transmission paths based on the causal graph and obtains a comprehensive risk score for the target entity; a risk strategy generation unit, which generates risk mitigation strategies based on the comprehensive risk score; and a risk system update unit, which monitors the deviation between the actual risk occurrence and the predicted results, updates the causal graph and risk mitigation strategies, and forms a self-evolving risk assessment system.

Images