A supply chain risk identification method, system, device, and medium

Abstract

A supply chain risk identification method, system, device and medium belong to the technical field of supply chain risk management. The risk identification method comprises: acquiring characteristic information of a supply chain risk event, the characteristic information of the supply chain risk event comprising event fuzzy description and corresponding parameter information; the parameter information comprising expert evaluation data and supply chain platform operation data; inputting the event fuzzy description into a supply chain language large model for matching processing, and outputting a plurality of risk index items corresponding to the event fuzzy description, the risk index items comprising events affecting supply chain capacity or transportation; and determining the risk levels of the plurality of risk index items corresponding to the parameter information. The risk identification method can improve the supply chain risk identification capability, make it possible to comprehensively perceive multi-source heterogeneous risk factors in the whole chain range and early warn key node risks, and realize dynamic monitoring and accurate identification of complex equipment supply chain operation risks.

Description

Technical Field

[0001] This application belongs to the field of supply chain risk management technology, and in particular relates to a supply chain risk identification method, system, equipment, and medium. Background Technology

[0002] Complex equipment supply chains are characterized by long chains, numerous participants, high technological complexity, and strong dependence on the external environment. The risk factors they face encompass multiple dimensions, including supply disruptions, quality defects, logistical delays, policy changes, and market fluctuations. Existing risk management methods are mostly based on static indicator systems or single-dimensional statistical analysis, making it difficult to effectively depict the dynamic relationships between risks across the entire chain. Furthermore, they heavily rely on human experience, which can easily lead to delayed risk warnings or misjudgments.

[0003] Current research employs analytic hierarchy process (AHP) or fuzzy comprehensive evaluation to quantitatively assess supply chain risks. However, these methods fail to dynamically adjust indicator weights based on real-time event data, making them ill-suited to rapidly changing risk environments. Other studies attempt to introduce machine learning models for risk identification, but these models fall short in semantic understanding of risk events, resulting in limited effectiveness in identifying small-sample and sudden risk events.

[0004] Chinese patent application CN116070906A discloses a method for risk identification and assessment based on the supply chain of complex product suppliers. It uses the complex product structure as its core, mapping the correspondence between products and suppliers to generate a directed graph of complex product suppliers. By calculating the supplier risk at each node, the edge weights between nodes in the directed graph are used as the risk transmission probability. Based on the risk transmission principle, the risk is calculated layer by layer to obtain the final supplier risk at each node, up to the root node. This invention's calculation of supplier risk for complex products can observe the overall risk changes in the supplier supply chain and the impact of supporting products on the overall situation, focusing on monitoring core and key suppliers, and promptly identifying risk nodes and weak links, thereby achieving the purpose of early warning and risk avoidance. Therefore, this patent application can calculate the supplier risk of complex products, observe the overall risk changes in the supply chain and the impact of supporting products on the overall situation, monitor core and key risk suppliers, and promptly identify risk nodes and weak links, thereby achieving the purpose of early warning and risk avoidance. However, its solution is not for risk identification of all links in the supply chain and cannot solve the problem of systematic risk identification for the entire supply chain of complex equipment.

[0005] Therefore, there is an urgent need to build a full-chain supply chain risk identification technology that integrates expert knowledge, dynamic event-driven mechanisms, and quantitative calculation methods to enhance the resilience and risk response capabilities of complex equipment supply chains. Summary of the Invention

[0006] This application aims to at least partially address one of the technical problems in related technologies. To this end, the supply chain risk identification method provided in this application can improve the systematic and quantitative modeling level of risk identification in complex equipment supply chains, making it possible to comprehensively perceive multi-source heterogeneous risk factors throughout the entire chain and provide early warning of risks at key nodes, thereby realizing dynamic monitoring, accurate identification, and scientific management of operational risks in complex equipment supply chains.

[0007] To achieve the above objectives, in a first aspect, this application provides a supply chain risk identification method, comprising:

[0008] Acquire characteristic information of supply chain risk events, which includes a fuzzy description of the event and corresponding parameter information; the parameter information includes expert evaluation data and supply chain platform operation data.

[0009] The fuzzy description of the event is input into the supply chain language big model for matching processing, and several risk indicator items corresponding to the fuzzy description of the event are output. The risk indicator items include events that affect supply chain capacity or transportation.

[0010] Based on the parameter information, the risk level of several corresponding risk indicators is determined.

[0011] Preferably, the process further includes, before obtaining the characteristic information of supply chain risk events:

[0012] A supply chain risk indicator system is constructed to classify supply chain risks step by step according to a preset hierarchical relationship, and to refine all risk factors down to a single node enterprise of a specific product model. The final level of the supply chain risk indicator system includes several risk indicator items to assess the impact of supply chain risk events on the supply chain.

[0013] Preferably, the risk indicator includes several corresponding quantitative parameters for assessing the risk level of the risk indicator; wherein the parameter information includes data corresponding to the quantitative parameters.

[0014] Preferably, the step of determining the risk level of the corresponding risk indicator items based on the parameter information includes:

[0015] Extract several quantitative parameters corresponding to each risk indicator item from the parameter information to generate a corresponding set of quantitative parameters;

[0016] The risk indicator item and its corresponding set of quantitative parameters are input into the risk quantification rating model for processing, and the risk level of the current risk indicator item is output. The risk quantification rating model includes all risk indicator items and their corresponding sets of all quantitative parameters, and is used to classify all risk indicator items into levels.

[0017] Preferably, the process of constructing the risk quantification rating model includes:

[0018] Obtain several risk indicators corresponding to supply chain risk events;

[0019] Different quantitative formula calculation functions are constructed based on the impact type of risk indicator items to calculate the impact of risk events on supply chain capacity or transportation. The quantitative formula calculation function is a continuously differentiable function and uses data from the parameter information for calculation.

[0020] Within the domain of the specified constraints, determine the boundary conditions for the quantization formula calculation function to obtain the range of the objective function;

[0021] Obtain the parameter information of the risk event, and calculate the impact parameter value of the current risk indicator item based on the quantitative formula calculation function;

[0022] The risk level of the risk indicator item is generated based on the position of the influencing parameter value within the range of the objective function.

[0023] Preferably, the process of constructing the risk quantification rating model further includes:

[0024] Obtain the specific values ​​of several quantitative parameters corresponding to each of the several risk indicators;

[0025] Establish a correspondence between risk indicators and their corresponding quantitative parameters, as well as the risk level of the risk indicators.

[0026] Preferably, the impact types of the risk indicators include impact on supply chain capacity, transportation, and delivery time.

[0027] Preferably, the step of determining the boundary conditions of the quantization formula calculation function within the domain of the specified constraints includes: determining the boundary conditions of the quantization formula calculation function by gradient descent or function iteration.

[0028] Preferably, the step of outputting several risk indicator items corresponding to the fuzzy description of the event includes:

[0029] The event fuzzy description is encoded, distilled, and decoded through the supply chain language big model to obtain the indicator score vector.

[0030] Calculate the probability distribution of each risk indicator item based on the indicator scoring vector;

[0031] The risk indicator with the highest probability is selected as the final identification result.

[0032] Preferably, the step of calculating the probability distribution of each risk indicator item based on the indicator score vector includes: using the Softmax function to convert the indicator score vector into a probability distribution.

[0033] Preferably, the supply chain language model is trained and optimized using data from a constructed supply chain risk corpus; the supply chain risk corpus includes all risk indicators at the lowest level of the supply chain risk indicator system.

[0034] Secondly, this application provides a supply chain risk identification system, including:

[0035] The information acquisition module is used to acquire characteristic information of supply chain risk events, including a fuzzy description of the event and corresponding parameter information; the parameter information includes expert evaluation data and supply chain platform operation data.

[0036] The data analysis module is used to input the fuzzy description of the event into the supply chain language big model for matching processing, and output several risk indicator items corresponding to the fuzzy description of the event. The risk indicator items include events that affect supply chain capacity or transportation.

[0037] The calculation module is used to determine the risk level of several corresponding risk indicator items based on the parameter information.

[0038] Secondly, this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the supply chain risk identification method described in any of the preceding claims.

[0039] Thirdly, this application provides a computer-readable storage medium including a computer program that, when run on an electronic device, causes the electronic device to perform the steps of the supply chain risk identification method described in any of the preceding claims.

[0040] Based on the above technical solution, the supply chain risk identification method of this application has at least one of the following beneficial effects compared with the prior art:

[0041] 1. This application proposes a supply chain risk identification method that combines an event-driven big language model to achieve dynamic mapping between events and risk indicators. This enables comprehensive perception of multi-source heterogeneous risk factors in complex equipment supply chains and automatic identification and risk level determination of potential risk events at key nodes. It improves the systematic, quantitative, and intelligent level of risk identification and ultimately achieves dynamic monitoring, accurate identification, and scientific management of operational risks in complex equipment supply chains, providing effective support for ensuring the supply of highly reliable equipment.

[0042] 2. This application discloses a supply chain risk identification method. By constructing a risk indicator system covering the entire chain and integrating expert knowledge and platform operation data for quantitative modeling and analysis, it improves the systematicness and quantitative modeling level of risk identification in complex equipment supply chains. This makes it possible to comprehensively perceive multi-source heterogeneous risk factors across the entire chain and provide early warning of risks at key nodes. By constructing a risk quantification rating model to achieve rapid mapping of risk levels, and using quantitative formulas to calculate the risk level classification threshold system, it achieves risk level determination. Ultimately, it realizes dynamic monitoring, accurate identification, and scientific management of operational risks in complex equipment supply chains, providing theoretical support and methodological system for high-reliability supply assurance for major equipment projects. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart of a supply chain risk identification method provided in this application;

[0045] Figure 2 This application provides a supply chain risk indicator system in a supply chain risk identification method.

[0046] Figure 3 This is a flowchart for calculating the risk level of the risk indicator items provided in this application;

[0047] Figure 4 This is a flowchart showing the mapping between real-time supply chain events and risk indicators provided in this application. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0049] The terms “first,” “second,” “third,” “fourth,” “fifth,” “sixth,” “seventh,” and “eighth,” etc. (if present), in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein.

[0050] Furthermore, the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, such that a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or apparatus.

[0051] Example 1

[0052] like Figure 1 As shown in the embodiments of this application, a supply chain risk identification method is provided, including:

[0053] S1. Obtain characteristic information of supply chain risk events, including fuzzy descriptions of the events and corresponding parameter information; the parameter information includes expert evaluation data and supply chain platform operation data.

[0054] S2. Input the fuzzy description of the event into the supply chain language big model for matching processing, and output several risk indicator items corresponding to the fuzzy description of the event. The risk indicator items include events that affect supply chain capacity or transportation.

[0055] S3. Determine the risk level of several corresponding risk indicators based on the parameter information.

[0056] Preferably, the process further includes, before obtaining the characteristic information of supply chain risk events:

[0057] A supply chain risk indicator system is constructed to classify supply chain risks level by level according to a preset hierarchical relationship, refining all risk factors down to the individual node enterprise of a specific product model. The final level of the supply chain risk indicator system includes several risk indicator items used to assess the impact of supply chain risk events on the supply chain. In specific implementation, the hierarchical relationship can be set according to actual circumstances.

[0058] One possible implementation scheme for a hierarchical relationship, and one possible implementation scheme for a supply chain risk indicator system, is as follows: Figure 2As shown, when classifying the risk indicator system, a three-tiered risk analysis framework of "model-unit-group" can be considered. Model-level risks focus on the unique risks of specific products or projects in production, R&D, and delivery, such as technological iterations, raw material supply, and production cycle fluctuations. Unit-level risks focus on the challenges faced by individual enterprises or business units in operations such as inventory management, logistics costs, and supplier cooperation. Group-level risks analyze supply chain structural risks, policy and legal risks, market supply and demand fluctuations, and global supply chain disruptions from a macro perspective, directly affecting the group's planning and sustainable development capabilities. The three-tiered system analyzes the transmission path of risks in the equipment supply chain, clarifying risk attribution and hierarchical structure. Based on the tiered risk analysis framework, risk factors are refined to the node enterprises of the model-level supply chain, constructing a four-tiered indicator classification system around five dimensions: "production-logistics-performance (compliance risks, procurement, contract risks)-attributes (the node enterprise's own cybersecurity, information security, personnel transfers)-external factors." Among them, the internal risks of node enterprises include dimensions such as "financial", "production and procurement" and "customers and delivery", while external risks cover elements such as "natural environment" and "policy changes", ultimately forming a complete and reasonably layered risk indicator system.

[0059] Alternatively, a "node-edge" risk analysis framework can be adopted to systematically analyze the risk transmission path. Here, "node" represents a node enterprise in the model-level supply chain, and "edge" represents the business relationships (logistics, orders, etc.) between nodes. Based on this, all risk factors are refined to individual node enterprises, and a hierarchical four-level indicator classification system is constructed around the two main dimensions of "external supply chain risk" and "internal supply chain risk." The secondary classifications under external risk include "natural and environmental risks," "political and policy risks," "economic and financial risks," and "social and market risks." The secondary classifications under internal risk include "financial risks," "production and procurement risks," "customer and delivery risks," "corporate governance risks," and "R&D and innovation risks." Further refinement of various risk factors on this basis forms a complete and clearly hierarchical complex equipment supply chain risk indicator system.

[0060] Preferably, the risk indicator includes several corresponding quantitative parameters for assessing the risk level of the risk indicator; wherein, the parameter information includes data corresponding to the quantitative parameters, and the parameter information includes expert evaluation data and supply chain platform operation data, that is, all the quantitative parameters corresponding to the risk indicator come from the parameter information (expert evaluation data and supply chain platform operation data).

[0061] Preferably, the step of determining the risk level of the corresponding risk indicator items based on the parameter information includes:

[0062] Extract several quantitative parameters corresponding to each risk indicator item from the parameter information to generate a corresponding set of quantitative parameters;

[0063] The risk indicator item and its corresponding set of quantitative parameters are input into the risk quantification rating model for processing, and the risk level of the current risk indicator item is output. The risk quantification rating model includes all risk indicator items and their corresponding sets of all quantitative parameters, and is used to classify all risk indicator items into levels.

[0064] Preferably, the process of constructing the risk quantification rating model includes:

[0065] Obtain several risk indicators corresponding to supply chain risk events;

[0066] Different quantitative formula calculation functions are constructed based on the impact type of risk indicator items to calculate the impact of risk events on supply chain capacity or transportation. The quantitative formula calculation function is a continuously differentiable function and uses data from the parameter information for calculation.

[0067] Within the domain of the specified constraints, determine the boundary conditions for the quantization formula calculation function to obtain the range of the objective function;

[0068] Obtain the parameter information of the risk event, and calculate the impact parameter value of the current risk indicator item based on the quantitative formula calculation function;

[0069] The risk level of the risk indicator item is generated based on the position of the influencing parameter value within the range of the objective function.

[0070] Obtain the specific values ​​of several quantitative parameters corresponding to each of the several risk indicators;

[0071] Establish a correspondence between risk indicators and their corresponding quantitative parameters, as well as the risk level of the risk indicators.

[0072] like Figure 3 The flowchart shown illustrates the process of calculating the risk level of a risk indicator. On one hand, experts define the allowable range of parameter values ​​based on the actual situation; on the other hand, they define a quantization function based on the actual situation. The derivative of the quantization function is solved using analytical methods or numerical difference methods. The function variables are initialized within the expert-preset allowable parameter range to obtain the initial domain. The gradient ascent method is used to find the function's maximum value (upper bound U), and the gradient descent method is used to find the function's minimum value (lower bound L). Each iteration performs adaptive learning rate adjustment and projection constraints to ensure that the variables do not exceed the bounds. After convergence, the quantization function's value range [L, U] is obtained. Finally, the real-time event parameters are substituted into the quantization function to obtain the current influencing parameter value, and the risk level is mapped based on the position of this value within the dynamic value range.

[0073] Furthermore, the process of constructing the corresponding risk quantification rating model for risk indicators based on expert scoring also includes:

[0074] Obtain the risk indicators corresponding to supply chain risk events;

[0075] The risk level of the risk indicator item is obtained by scoring and rating its impact by experts.

[0076] Obtain the specific values ​​of several quantitative parameters corresponding to the risk indicator items;

[0077] Establish a correspondence between risk indicators and their corresponding quantitative parameters, as well as the risk level of the risk indicators.

[0078] Furthermore, the aforementioned steps of scoring and rating the impact through experts include: experts scoring the impact based on several quantitative parameters corresponding to the risk indicators, specifically using a 5-level scoring standard, where 1 point represents a very poor score and 5 points represent an excellent score, using a comprehensive scoring formula. The quantitative parameters of each expert scoring system were normalized and weighted, among which, This represents the overall risk score. Indicates the first The scores for each expert scoring indicator, Indicates the first The weights of each quantization parameter.

[0079] Preferably, the impact types of the risk indicators include impact on supply chain capacity, transportation, and delivery time.

[0080] Specifically, the steps for determining the boundary conditions of the quantization formula calculation function within the domain of the specified constraints include: determining the boundary conditions of the quantization formula calculation function using the gradient descent method or the function iteration method.

[0081] The calculation process for the risk level when the impact type of the risk indicator item is on the supply chain's capacity or transportation (such as an earthquake) is as follows:

[0082] Constructing the quantization formula R i (x), where x represents the variable affecting production capacity or transportation, R i (x) is a continuously differentiable function, expressed as: , where R i 产能 R represents the calculated impact of production capacity. i 运输 This indicates the calculated value of the transportation impact. post_capacity This indicates the production capacity of the enterprise at the post-event stage. pre_capacity This indicates the production capacity of the enterprise at the pre-conference stage. post_time Indicates the time required for subsequent material transportation. pre_time Indicates the time required for prior material transportation;

[0083] Define the bounded domain (i.e., the domain of definition) of the variables for calculating the impact of production capacity and transportation.

[0084] Within the domain of the specified constraints, the boundary conditions of the function calculated by the quantization formula are determined by the gradient descent method, and the range of the objective function is obtained.

[0085] Obtain the parameter information of the risk event, and calculate the impact parameter value of the current risk indicator item based on the quantitative formula calculation function;

[0086] The risk level of the risk indicator item is generated based on the position of the influencing parameter value within the range of the objective function.

[0087] The goal of the gradient descent algorithm is to determine the boundary conditions of the risk quantification function within a specified constrained bounded domain, thereby defining the effective discrimination interval for risk levels: lower boundary. upper boundary The lower bound is approximated by gradient descent, and the upper bound is achieved by negating the function and applying gradient descent (equivalent to gradient ascent). Each iteration is performed according to... x (t+1) =proj D ( x (t) -η ),in For learning rate, Let be the gradient vector, t+1 and t represent the current time and the previous time, respectively, and D represent the domain of the independent variable. This involves projecting the variable vector back to the domain to maintain constraint feasibility; after obtaining the boundary values, risk mapping is performed based on the actual risk data. R real Represents the actual risk value, the mapped value This is the risk level output by the current event quantification formula. If the current event is not a risk event, then The current event is a risk event and its risk level is [risk level]. .

[0088] When the impact of a risk indicator falls under the category of delivery time impact (such as delayed completion), the quantitative formula is calculated as: Risk Sensitivity Coefficient * exp((Actual Project Completion Time - Planned Completion Time) / (Expected Lateness Days + Standard Deviation of Lateness Days)). This is because a longer lag time generally indicates higher risk; a 1-day lag might be low risk, but a 2-day lag is high risk. In this case, gradient descent can be used to quickly obtain the upper and lower boundary values ​​by solving for the derivative.

[0089] Specifically, this step involves quantitatively modeling and analyzing a risk indicator system by integrating expert knowledge with the characteristics of complex equipment supply chain operational events. Based on this constructed risk indicator system, the impact of risk indicators on the supply chain is quantified from two core dimensions: "capacity" and "transportation." It should be noted that for quantitative parameters that cannot be specifically calculated or assessed, such as whether a node enterprise experiences delivery interruptions or whether a supplier is unable to provide raw materials or products, experts conduct assessments to determine when a risk event occurs, thus establishing a certain level of risk.

[0090] Preferably, the step of outputting several risk indicator items corresponding to the fuzzy description of the event includes:

[0091] The event fuzzy description is encoded, distilled, and decoded through the supply chain language big model to obtain the indicator score vector.

[0092] Calculate the probability distribution of each risk indicator item based on the indicator scoring vector;

[0093] The risk indicator with the highest probability is selected as the final identification result.

[0094] Preferably, the step of calculating the probability distribution of each risk indicator item based on the indicator score vector includes: using the Softmax function to convert the indicator score vector into a probability distribution.

[0095] Specific implementation methods include obtaining a length of [length missing] after operations such as encoding, distillation, and decoding of the fuzzy event description through a large language model. Indicator scoring vector ,in and The distribution represents the weight matrix and bias terms. This represents the high-level semantics described by the input event model; by ,in, It is a risk event The score, It is a risk indicator item The predicted probability. The risk identification algorithm calculates the probability distribution of each risk indicator and selects the one with the highest probability as the final identification result, thereby achieving a precise mapping from fuzzy input of risk node events to specific risk indicator items.

[0096] Preferably, the supply chain language model is trained and optimized using data from a constructed supply chain risk corpus; the supply chain risk corpus includes all risk indicators at the lowest level of the supply chain risk indicator system.

[0097] like Figure 4The diagram illustrates the mapping process between real-time events and risk indicators in the supply chain. When the supply chain system captures an unstructured event description (i.e., a vague description of a risk event, such as "Node A's factory partially collapsed due to a strong earthquake, causing production line shutdown"), the text is first transformed into a risk semantic vector through a domain-enhanced word embedding layer. Key terms ("strong earthquake," "factory collapse") are injected into specialized vectors in the supply chain risk knowledge base. This vector sequence is input into a multi-level Transformer encoder, which contains multiple layers (four are shown in the diagram). Each layer enhances the model's training stability through residual connections and layer normalization. During the encoding process, spatiotemporal context (such as earthquake duration and geographical coordinates) is simultaneously fused to output a deep semantic representation. Subsequently, the model enters a knowledge distillation enhancement layer: the soft labels generated by the teacher model (pre-trained industry-wide model) are compared with the predictions of the student model using KL divergence calculation to form a weighted distillation loss that constrains the model's optimization direction, ensuring the inheritance of domain knowledge. After CLS tagging and linear transformation, the global feature vector of the event is passed to the mapping layer for semantic similarity matching with the risk indicator library (e.g., calculating the cosine similarity with indicators such as "earthquake-induced factory collapse" and "equipment damage"), thus achieving the mapping of risk indicator items. Finally, backpropagation is driven by a joint loss function (e.g., cross-entropy classification loss + distillation regularization term), and the model parameters are updated by combining gradient pruning and elastic weight solidification strategies, achieving a dynamic mapping from fuzzy event descriptions to accurate risk indicators. This process, through a triple mechanism of domain pre-training, distillation enhancement, and feature focusing, solves key challenges in supply chain event description such as ambiguous terminology, lack of context, and sudden pattern recognition, forming a continuously self-optimizing risk cognition engine.

[0098] Example 2

[0099] This application provides a supply chain risk identification system, including:

[0100] The information acquisition module is used to acquire characteristic information of supply chain risk events, including a fuzzy description of the event and corresponding parameter information; the parameter information includes expert evaluation data and supply chain platform operation data.

[0101] The data analysis module is used to input the fuzzy description of the event into the supply chain language big model for matching processing, and output several risk indicator items corresponding to the fuzzy description of the event. The risk indicator items include events that affect supply chain capacity or transportation.

[0102] The calculation module is used to determine the risk level of several corresponding risk indicator items based on the parameter information.

[0103] Furthermore, the supply chain risk identification system also constructs a complex equipment supply chain risk indicator system covering all links in the supply chain. The system identifies and measures multi-dimensional risk factors that the entire chain may face. The supply chain risk indicator system classifies supply chain risks level by level according to a preset hierarchical relationship, refining all risk factors down to the single-node enterprise of a specific product model. The final level of the supply chain risk indicator system includes several risk indicator items used to assess the impact of supply chain risk events on the supply chain. Each risk indicator item includes several corresponding quantitative parameters used to assess the risk level of the risk indicator item; wherein, the parameter information includes data corresponding to the quantitative parameters.

[0104] Furthermore, the step of determining the risk level of the corresponding risk indicator items based on the parameter information includes:

[0105] Extract several quantitative parameters corresponding to each risk indicator item from the parameter information to generate a corresponding set of quantitative parameters;

[0106] The risk indicator item and its corresponding set of quantitative parameters are input into the risk quantification rating model for processing, and the risk level of the current risk indicator item is output. The risk quantification rating model includes all risk indicator items and their corresponding sets of all quantitative parameters, and is used to classify all risk indicator items into levels.

[0107] Furthermore, based on an event-driven large language model and combined with the established risk level classification threshold system, a mapping relationship between risk events and risk indicator items is established. The threshold space corresponding to the risk indicator items is dynamically parsed to achieve automatic identification and level determination of risk events.

[0108] Regarding the construction of the risk level classification threshold system, based on the quantitative modeling of risk indicators, an optimization method based on gradient descent algorithm is adopted to construct a classification threshold system for various risk indicators. The event quantification function is set as a continuously differentiable function, with variables constrained within a bounded domain. Gradient descent and gradient ascent are used to approximate its lower and upper boundaries, forming a risk level discrimination interval. Ultimately, this achieves accurate mapping and classification of event risk values, providing a quantitative basis for subsequent risk identification and response.

[0109] The Supply Chain Language Model (Risk Identification Mapping) leverages the capabilities of a large language model, combined with an established risk indicator system and risk level classification system, to construct a mapping relationship between real-time supply chain events and risk indicator items. Input events include parameter information and fuzzy event descriptions. These fuzzy descriptions undergo an encoding-distillation-decoding process, outputting corresponding indicator score vectors, which are then used to calculate the scores and predicted probabilities of risk indicator items. The maximum probability criterion is used to identify risk indicator items, and the established quantitative model and risk level discrimination algorithm are invoked to achieve accurate identification and response from fuzzy events to specific risk levels.

[0110] Example 3

[0111] This application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the program to implement the steps of the supply chain risk identification method described above, and achieves the following functions: The supply chain risk identification method, combined with an event-driven large language model, realizes dynamic mapping between events and risk indicators, thereby achieving comprehensive perception of multi-source heterogeneous risk factors in the complex equipment supply chain and automatic identification and risk level determination of potential risk events at key nodes, improving the systematic, quantitative, and intelligent level of risk identification, and ultimately realizing dynamic monitoring, accurate identification, and scientific management of operational risks in the complex equipment supply chain, providing effective support for the supply guarantee of high-reliability equipment. By constructing a risk indicator system covering the entire supply chain and integrating expert knowledge with platform operation data for quantitative modeling and analysis, the systemic and quantitative modeling level of risk identification in complex equipment supply chains is improved, making it possible to comprehensively perceive multi-source heterogeneous risk factors across the entire supply chain and provide early warning of risks at key nodes. By constructing a risk quantification rating model to achieve rapid mapping of risk levels and using quantitative formulas to calculate risk level classification thresholds, risk level determination is achieved. Ultimately, dynamic monitoring, accurate identification, and scientific management of operational risks in complex equipment supply chains are realized, providing theoretical support and methodological system for high-reliability supply assurance of major equipment projects.

[0112] Example 4

[0113] This application provides a computer-readable storage medium including a computer program. When the computer program is run on an electronic device, the electronic device executes the steps of the supply chain risk identification method described above and achieves the following functions: combining an event-driven large language model to realize dynamic mapping between events and risk indicators, thereby achieving comprehensive perception of multi-source heterogeneous risk factors in complex equipment supply chains and automatic identification and risk level determination of potential risk events at key nodes, improving the systematic, quantitative and intelligent level of risk identification, and ultimately realizing dynamic monitoring, accurate identification and scientific management of operational risks in complex equipment supply chains, providing effective support for the supply guarantee of highly reliable equipment. By constructing a risk indicator system covering the entire supply chain and integrating expert knowledge with platform operation data for quantitative modeling and analysis, the systemic and quantitative modeling level of risk identification in complex equipment supply chains is improved, making it possible to comprehensively perceive multi-source heterogeneous risk factors across the entire supply chain and provide early warning of risks at key nodes. By constructing a risk quantification rating model to achieve rapid mapping of risk levels and using quantitative formulas to calculate risk level classification thresholds, risk level determination is achieved. Ultimately, dynamic monitoring, accurate identification, and scientific management of operational risks in complex equipment supply chains are realized, providing theoretical support and methodological system for high-reliability supply assurance of major equipment projects.

[0114] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0115] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0116] The foregoing has described specific embodiments of the present invention. In some cases, the described actions or steps may be performed in a different order than those shown in the embodiments and the desired results may still be achieved. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0117] In the description of the embodiments of the present invention, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In the embodiments of the present invention, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in the embodiments of the present invention, as well as the features of the different embodiments or examples.

[0118] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of the invention pertain.

[0119] The above embodiments are merely preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for identifying risks in a complex equipment supply chain, characterized in that, include: Construct a risk indicator system for complex equipment supply chains to classify supply chain risks level by level according to a preset hierarchical relationship, and refine all risk factors down to the single node enterprise of a specific product model. Acquire characteristic information of supply chain risk events, which includes a fuzzy description of the event and corresponding parameter information; the parameter information includes expert evaluation data and supply chain platform operation data. The fuzzy description of the event is input into the supply chain language big model for matching processing, and several risk indicator items corresponding to the fuzzy description of the event are output. The risk indicator items include events that affect supply chain capacity or transportation. The supply chain language model is trained and optimized using data from the constructed supply chain risk corpus; the supply chain risk corpus includes all risk indicators at the lowest level of the supply chain risk indicator system. Based on the parameter information, determine the risk level of several corresponding risk indicators: obtain several quantitative parameters corresponding to each risk indicator from the parameter information, and generate a corresponding set of quantitative parameters; The risk indicator item and its corresponding set of quantitative parameters are input into the risk quantification rating model for processing, and the risk level of the current risk indicator item is output. The risk quantification rating model includes all risk indicator items and their corresponding sets of all quantitative parameters, and is used to classify all risk indicator items into levels.

2. The method for identifying risks in the complex equipment supply chain according to claim 1, characterized in that, The final level of the supply chain risk indicator system includes several risk indicator items used to assess the impact of supply chain risk events on the supply chain.

3. The method for identifying risks in the complex equipment supply chain according to claim 2, characterized in that, The risk indicator includes several corresponding quantitative parameters used to assess the risk level of the risk indicator; wherein, the parameter information includes data corresponding to the quantitative parameters.

4. The method for identifying risks in the complex equipment supply chain according to claim 3, characterized in that, The construction process of the risk quantification rating model includes: obtaining several risk indicators corresponding to supply chain risk events; constructing different quantification formula calculation functions based on the impact type of the risk indicator items to calculate the impact of risk events on supply chain capacity or transportation, wherein the quantification formula calculation function is a continuously differentiable function and uses data from the parameter information for calculation; determining the boundary conditions of the quantification formula calculation function within the domain of specified constraints to obtain the target function range; obtaining the parameter information of the risk event and calculating the impact parameter value of the current risk indicator item based on the quantification formula calculation function; and generating the corresponding risk level of the risk indicator item based on the interval position of the impact parameter value within the target function range.

5. The method for identifying risks in the complex equipment supply chain according to claim 4, characterized in that, The construction process of the risk quantification rating model also includes: obtaining the specific values ​​of several quantitative parameters corresponding to several risk indicators; establishing a correspondence between the risk indicators and their corresponding quantitative parameters, as well as the risk level of the risk indicators.

6. The method for identifying risks in the complex equipment supply chain according to claim 4, characterized in that, The impact types of the risk indicators include the impact on supply chain capacity, transportation, and delivery time.

7. The method for identifying risks in the complex equipment supply chain according to claim 4, characterized in that, The steps for determining the boundary conditions of the quantization formula calculation function within the domain of a specified constraint include: determining the boundary conditions of the quantization formula calculation function using the gradient descent method or the function iteration method.

8. The method for identifying risks in a complex equipment supply chain according to claim 1, characterized in that, The steps for outputting several risk indicator items corresponding to the fuzzy description of the event include: the fuzzy description of the event is encoded, distilled, and decoded through the supply chain language big model to obtain an indicator score vector; the probability distribution of each risk indicator item is calculated based on the indicator score vector; and the risk indicator item with the highest probability is selected as the final identification result.

9. The method for identifying risks in the complex equipment supply chain according to claim 8, characterized in that, The steps for calculating the probability distribution of each risk indicator item based on the indicator score vector include: using the Softmax function to convert the indicator score vector into a probability distribution.

10. A complex equipment supply chain risk identification system, used to implement the complex equipment supply chain risk identification method according to any one of claims 1-9, characterized in that, include: The information acquisition module is used to acquire characteristic information of supply chain risk events, including fuzzy descriptions of the events and corresponding parameter information. The parameter information includes expert evaluation data and supply chain platform operation data; The data analysis module is used to input the fuzzy description of the event into the supply chain language big model for matching processing, and output several risk indicator items corresponding to the fuzzy description of the event. The risk indicator items include events that affect supply chain capacity or transportation. The calculation module is used to determine the risk level of several corresponding risk indicator items based on the parameter information.

11. A computer device comprising a memory and a processor, the memory being used to store a computer program, characterized in that, The processor is used to execute the computer program to implement the steps of the complex equipment supply chain risk identification method according to any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when run on a computer or processor, causes the computer or processor to perform the steps of the complex equipment supply chain risk identification method according to any one of claims 1 to 9.

Images