A supply chain intelligent risk management method based on DLT and LLM

Abstract

This application discloses a supply chain intelligent risk management method based on DLT and LLM, including the following steps: A. Continuous data perception and dynamic maintenance of DCRG; B. Risk identification, assessment, and verifiable interpretation generation; C. Smart contract early warning and adaptive strategy generation; D. Decision making and authorization confirmation; E. Automated execution and recording of response strategies. This application tightly integrates the deep insights of DCRG, the intelligent response of LIA-SCR, and the credible interpretation of VEP through DLT as a trusted infrastructure and smart contracts as an automated link, forming a complete closed loop from data perception, intelligent analysis, dynamic decision-making, automated execution to credible verification. DLT is not only a trusted carrier of data, but also an undeniable recording platform for collaborative behaviors such as strategy approval and result confirmation, effectively solving the problems of information silos and low collaborative efficiency in existing technologies.

Description

Technical Field

[0001] This application relates to the field of supply chain management and risk control technology, and in particular, to a supply chain intelligent risk management method based on DLT and LLM. Background Technology

[0002] Currently, supply chain risk management primarily relies on traditional Enterprise Resource Planning (ERP) and Supply Chain Management (SCM) systems, combined with human experience, regular reports, and statistical models based on historical data for risk assessment and decision-making. Some scenarios are beginning to utilize independent blockchain technology to improve data transparency, or employ basic Artificial Intelligence (AI) / Machine Learning (ML) technologies for prediction in specific stages; however, the following shortcomings still exist: ① Lag and data silos: It relies heavily on historical data and internal information, lacks real-time information, and the information of various participants is not shared, making it difficult to obtain a global and dynamic risk view; ② Superficial risk analysis: It is difficult to effectively integrate and process massive, multi-source heterogeneous (especially unstructured) data, has weak predictive ability for complex and cascading risk events, and ignores deep causal relationships; ③ Rigid and inefficient response: Risk response measures are mostly based on preset rules or manual intervention, resulting in slow response and a lack of adaptive adjustment capabilities and automated execution efficiency for specific situations; ④ Lack of transparency and trust: The decision-making process of AI models is opaque ("black box"), and cross-enterprise collaboration lacks a reliable data foundation and process auditing capabilities. Summary of the Invention

[0003] This application provides a supply chain intelligent risk management method based on DLT and LLM, which solves the technical problems of existing supply chain risk management, such as lag and data silos, superficial risk analysis, rigid and inefficient response, and lack of transparency and trust.

[0004] This application is achieved through the following solution: A supply chain intelligent risk management method based on DLT and LLM includes the following steps: A. Continuously acquire confirmed transaction data from the distributed ledger network in real time, and dynamically update the dynamic causal risk graph DCRG anchored by DLT based on the acquired transaction data. The dynamic causal risk graph is constructed and maintained in real time using LLM. B. Risk Identification, Assessment, and Verifiable Explanation Generation: After identifying potential risk events using the LLM analysis of the dynamic causal risk graph, the scope and severity of the risk events are assessed, and a verifiable explanation record for this risk assessment is generated. This hash is then anchored to the distributed ledger network via DLT transactions. ； C. Risk Warning Smart Contract Early Warning and Adaptive Strategy Generation: The risk warning smart contract generates early warning and adaptive strategies based on risk events. After obtaining the current risk context by querying the key information of the adaptive smart contract response strategy driven by LLM, LLM generates and returns the optimal recommended strategy through simulation evaluation. D. Decision Making and Authority Confirmation: The risk warning smart contract receives the optimal recommended strategy. And based on the optimal recommendation strategy Whether the preset automatic execution conditions are met or whether manual approval is required before proceeding to the corresponding execution stage; E. Automated execution and recording of response strategies: The risk warning smart contract automatically adopts or manually approves strategies based on the final determination. It executes the corresponding on-chain or off-chain actions and records the execution results back to the DLT.

[0005] Furthermore, step A specifically includes the following steps: A1. Data Injection: The data sensing and integration layer continuously obtains confirmed transaction data from the distributed ledger network in real time, including order status updates and logistics node information, and collects information from off-chain data sources through oracles. The off-chain data sources include IoT sensors, news APIs, market data, and weather services. A2. Data Fusion and Preprocessing: The data preprocessing and fusion unit cleans, formats, links entities, and aligns the collected multi-source heterogeneous data. A3, DCRG Update: The preprocessed data is sent to the intelligent analysis and decision-making layer, where the DCRG construction and inference module uses LLM to analyze new data and update the dynamic causal risk map in real time. This reflects the latest changes in the status of the supply chain and potential risks.

[0006] Furthermore, step B specifically includes the following steps: B1. Risk Event Identification: LLM analyzes the dynamic causal risk map of DCRG. Or, when directly analyzing the data stream, identify the risk probability of specific risk nodes. Or aggregate risk scores Exceeding the preset warning threshold This allows for the identification of potential risk events. ,in, Represents the set of supply chain entities and risk factor nodes. Represents the passage of time The set of potential causal relationships or dependencies of change It is the edge weight function, which quantifies the probability of risk transmission or the intensity of its impact; B2. In-depth assessment and interpretation: For identified risk events... LLM performs in-depth assessments to predict the potential scope and severity of its impact and generates human-readable natural language explanations. B3. VEP Generation and Anchoring: Simultaneously, the VEP generation module generates verifiable explanation records for this risk assessment according to the set verifiable explanation record structure. VER It includes the assessment results, explanatory text, text hash, and a detailed list of evidence sources, and calculates... VER Overall hash And anchor this hash to the distributed ledger network via DLT transactions, in conjunction with risk events. Related.

[0007] Furthermore, step C specifically includes the following steps: C1. Risk Signal Transmission: The intelligent analysis and decision-making layer will identify risk events. Key information (such as risk type, level, scope of impact summary) and The information is transmitted via an oracle to a risk warning smart contract deployed on DLT. C2. Preliminary Contract Assessment: The risk warning smart contract makes assessments based on the received risk information and built-in basic rules; C3, LIA-SCR Query: For risk events requiring intelligent decision-making, the risk warning smart contract initiates a query to the LLM strategy service via an oracle, attaching the current risk context. ; C4, LLM Strategy Recommendation: The LLM strategy service selects recommended strategies based on the risk context. The optimization objective is to reduce the amount of expected risks. Conduct simulation evaluations to generate the optimal recommendation strategy. This includes its expected utility assessment and reasons for recommendation.

[0008] Furthermore, step D specifically includes the following steps: D1. Strategy Delivery and Presentation: The recommended strategy returned by LLM The risk warning is transmitted back to the smart contract via an oracle; D2. Automated or manual approval: D2.1 Automatic Execution: If the recommended strategy is... Once the preset automatic execution conditions are met, the risk warning smart contract can directly proceed to the next execution stage; D2.2 Manual Approval: If the conditions for automatic execution are not met or the system is configured to require manual confirmation, the risk warning smart contract will recommend a strategy. The expected utility and associated VER are pushed to users with the corresponding permissions through the application interaction layer, so that users can view detailed information on the interface and verify the credibility of the explanation through the VEP mechanism; D3. On-chain decision confirmation: If manual approval is required, authorized users interact with their DLT identity through the application interaction layer to confirm the recommendation strategy. The decision to approve or reject a transaction is confirmed by signing a DLT transaction, recorded on the ledger, and the risk warning smart contract is notified.

[0009] Furthermore, step E specifically includes the following steps: E1. Contract Execution: The risk warning smart contract executes according to the final determined strategy. Perform the corresponding on-chain or off-chain actions: E1.1 On-chain actions: such as updating the status markers of related assets or orders on DLT, recording decision results, and sending messages to other related contracts; E1.2 Off-chain action triggering: Calling the oracle to trigger off-chain operations, including sending email / SMS notifications to relevant personnel, calling the API of the enterprise ERP / SCM system, and instructing physical devices; E2. Execution Result Recording: Key execution steps and results, or their hash values, are recorded back to the DLT to ensure process traceability.

[0010] Furthermore, step E also includes the step of: Multi-party collaboration and information sharing: Multi-party collaboration and information sharing are achieved through initiating collaborative processes, information sharing and interaction, and on-chain collaborative records.

[0011] Furthermore, the aforementioned multi-party collaboration and information sharing are achieved through initiating collaborative processes, information sharing and interaction, and on-chain collaborative records. Specifically, this includes the following steps: E3. Collaborative Process Initiation: For complex risk events that require multi-party collaboration, the risk warning contract or manual decision-making triggers the creation and activation of an instance of the collaborative management contract. E4. Information Sharing and Interaction: Relevant participants can access DLT data, LLM risk reports, and risk warning smart contract status information related to the event through the collaborative workspace of the application interaction layer. E5. On-chain Collaboration Records: Participants can discuss within the collaboration space and submit confirmation information, voting results, and jointly developed detailed response plans via DLT signatures, all of which are recorded by the collaboration management contract.

[0012] Furthermore, it also includes the following steps: F. Continuous monitoring and feedback learning: By continuously monitoring the actual effects of the implemented response strategies, the risk prediction model and strategy recommendation capabilities of LLM are continuously optimized, forming a learning loop.

[0013] Furthermore, step F specifically includes the following steps: F1. Effect Tracking: The system continuously monitors the development of risk events and the actual effect of the implemented response strategies, including through subsequent transaction data on DLT, IoT sensor feedback, and external information updates; F2, DCRG, and Model Iteration: Monitoring results and policy implementation effectiveness data are fed back to the intelligent analysis and decision-making layer to further update the dynamic causal risk map of DCRG. It can also be used as new training / fine-tuning data to continuously optimize the risk prediction model and strategy recommendation capabilities of LLM, forming a learning loop.

[0014] This application also provides a supply chain intelligent risk management device based on DLT and LLM, including: The data awareness and DCRG dynamic maintenance module is used to continuously acquire confirmed transaction data from the distributed ledger network in real time, and dynamically update the dynamic causal risk graph DCRG anchored by DLT based on the acquired transaction data. The dynamic causal risk graph is constructed and maintained in real time using LLM. The identification, assessment, and interpretation generation module is used for risk identification, assessment, and verifiable interpretation generation: After identifying potential risk events using the LLM analysis dynamic causal risk map, it assesses the scope and severity of the risk events' impact and generates verifiable interpretation records for this risk assessment. These hashes are then anchored to the distributed ledger network via DLT transactions. ； The early warning and adaptive strategy generation module is used for risk early warning smart contracts to generate early warning and adaptive strategies: the risk early warning smart contract generates early warning and adaptive strategies based on risk events. After obtaining the current risk context by querying the key information of the adaptive smart contract response strategy driven by LLM, LLM generates and returns the optimal recommended strategy through simulation evaluation. The decision-making and permission confirmation module is used for decision-making and permission confirmation: the risk warning smart contract receives the optimal recommendation strategy. And based on the optimal recommendation strategy Whether the preset automatic execution conditions are met or whether manual approval is required before proceeding to the corresponding execution stage; The response strategy execution and recording module is used for the automated execution and recording of response strategies: the risk warning smart contract automatically adopts or manually approves strategies based on the final determination. It executes the corresponding on-chain or off-chain actions and records the execution results back to the DLT.

[0015] This application also provides an electronic device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the DLT and LLM-based intelligent supply chain risk management method when executing the computer program.

[0016] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned DLT and LLM-based intelligent supply chain risk management method.

[0017] This application also provides a computer program product, including a computer program or computer-executable instructions, which, when executed by a processor, implement the aforementioned supply chain intelligent risk management method based on DLT and LLM.

[0018] Compared with existing technologies, this application, through the deep and innovative integration of distributed ledger technology (DLT), large-scale language model (LLM), and smart contracts, constructs a supply chain risk early warning and intelligent collaboration system and method, which can produce the following significant beneficial effects: (1) Significantly enhance the depth and foresight of risk insights: This application utilizes a Dynamic Causal Risk Map (DCRG) constructed using LLM, which transcends traditional risk assessments based on historical data and superficial correlations. It deeply mines the complex causal transmission mechanisms within the supply chain and dynamically evolves by incorporating real-time multi-source information. This enables the system not only to identify known risks but also to predict the chain reactions and future trends of potential risks, achieving a fundamental shift from passive and lagging risk management to proactive and forward-looking risk prevention. (2) Achieving highly intelligent and adaptive risk response: By employing an LLM-driven adaptive smart contract response (LIA-SCR) mechanism, this application overcomes the shortcomings of traditional smart contract response strategies, which are rigid and lack flexibility. The system can intelligently generate or select the optimal response strategy based on the specific context of real-time risk events, making automated responses more accurate and efficient, minimizing actual risk losses, and optimizing resource allocation. (3) Significantly enhance the transparency and credibility of the AI ​​decision-making process: The unique Verifiable Explanatory Path (VEP) mechanism provides an end-to-end, DLT-anchored, and independently verifiable explanation chain for the risk analysis and strategy recommendation process of LLM. This effectively solves the "black box" problem of AI models, enabling users and regulatory agencies to understand, trust, and audit the decision-making basis of AI, significantly improving the credibility, reliability, and compliance of the entire system.

[0019] In addition to the purposes, features, and advantages described above, this application has other purposes, features, and advantages. A further detailed description of this application will be provided below with reference to the figures. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein: Figure 1 This is a schematic diagram of the overall architecture of the DLT and LLM-based intelligent supply chain risk management system according to a preferred embodiment of this application; Figure 2 This is a flowchart illustrating a preferred embodiment of the supply chain intelligent risk management method based on DLT and LLM in this application. Figure 3 This is a schematic diagram of the Dynamic Causal Risk Map (DCRG) (DCRG Example Diagram); Figure 4 This is a flowchart of LLM-driven Adaptive Smart Contract Response (LIA-SCR); Figure 5 It is a diagram of the Verifiable Explanation Record (VER) structure (VEP Record Structure Diagram); Figure 6 This is a workflow sequence diagram of another preferred embodiment of the supply chain intelligent risk management method based on DLT and LLM in this application; Figure 7 This is a schematic block diagram of an electronic device according to a preferred embodiment of this application; Figure 8 This is a schematic diagram of the internal structure of a computer device according to a preferred embodiment of this application. Detailed Implementation

[0022] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0023] Explanation of technical terms: Distributed Ledger Technology (DLT / Blockchain): A decentralized and tamper-proof database technology that provides a trusted and transparent record-keeping foundation for supply chain data.

[0024] Large Language Models (LLMs): Artificial intelligence models with powerful natural language understanding and generation capabilities, used for in-depth analysis of multi-source data, risk prediction, and insight generation.

[0025] Risk warning smart contracts: Automated scripts deployed on DLT that automatically execute risk response and coordination processes based on preset rules and on-chain / off-chain data (which can be triggered by LLM output).

[0026] Supply chain risk management: In this application, it specifically refers to the use of DLT, LLM and risk warning smart contract technologies to achieve proactive identification, prediction, assessment, response and collaborative handling of supply chain disruption risks.

[0027] It should be noted that the executing entity in this embodiment can be a computing service system with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a DLT and LLM-based intelligent supply chain risk management system capable of achieving the above functions. The following example uses a DLT and LLM-based intelligent supply chain risk management system (see...) Figure 1 Taking the execution subject as an example, this embodiment and the following embodiments will be described.

[0028] like Figure 2 As shown, a preferred embodiment of this application provides a supply chain intelligent risk management method based on DLT and LLM, including the following steps: A. Continuously acquire confirmed transaction data from the distributed ledger network in real time, and dynamically update the dynamic causal risk graph DCRG anchored by DLT based on the acquired transaction data. The dynamic causal risk graph is constructed and maintained in real time using LLM. B. Risk Identification, Assessment, and Verifiable Explanation Generation: After identifying potential risk events using the LLM analysis of the dynamic causal risk graph, the scope and severity of the risk events are assessed, and a verifiable explanation record for this risk assessment is generated. This hash is then anchored to the distributed ledger network via DLT transactions. ； C. Risk Warning Smart Contract Early Warning and Adaptive Strategy Generation: The risk warning smart contract generates early warning and adaptive strategies based on risk events. After obtaining the current risk context by querying the key information of the adaptive smart contract response strategy driven by LLM, LLM generates and returns the optimal recommended strategy through simulation evaluation. D. Decision Making and Authority Confirmation: The risk warning smart contract receives the optimal recommended strategy. And based on the optimal recommendation strategy Whether the preset automatic execution conditions are met or whether manual approval is required before proceeding to the corresponding execution stage; E. Automated execution and recording of response strategies: The risk warning smart contract automatically adopts or manually approves strategies based on the final determination. It executes the corresponding on-chain or off-chain actions and records the execution results back to the DLT.

[0029] This application addresses the core pain points of existing supply chain risk management, including poor real-time performance, lack of information transparency, insufficient depth of risk prediction, rigid response strategies, low collaborative efficiency, and the "black box" nature of AI models. It proposes an innovative technical solution that deeply integrates distributed ledger technology (DLT), large-scale language models (LLM), and smart contracts. This solution is not a simple aggregation of technologies, but rather achieves organic integration through the following three core innovative mechanisms, thereby systematically overcoming the shortcomings of existing technologies: (1) Establish and utilize a Dynamic Causal Risk Graph (DCRG) anchored by DLT to achieve in-depth risk insights and forward-looking predictions (see...). Figure 3 ).

[0030] To address the problems of traditional risk analysis, such as reliance on lagged data, emphasis on statistical correlations rather than causality, and the inability of static models to adapt to dynamic changes, this solution utilizes LLM to construct and maintain a dynamic causal risk map in real time. .in, Represents the set of supply chain entities and risk factor nodes. Represents the passage of time The set of potential causal relationships or dependencies of change It is the edge weight function, quantifying the probability of risk transmission or the strength of its impact. LLM continuously analyzes and integrates reliable transaction data from DLT. New multi-source heterogeneous off-chain evidence obtained via oracles To achieve dynamic evolution of the graph: ① Real-time update of node risk probability: Risk nodes Risk probability Based on its parent node Status and new evidence To update, LLM is used for context-aware reasoning: ; in, For LLM inference update functions, This is the function for interpreting evidence.

[0031] ② Dynamic adjustment of the strength of causal relationship: (Edge) weight According to the new on-chain transaction model and external information Adjustments are made to reflect the dynamic changes in causal relationships: ; in, This is the LLM weight adjustment function.

[0032] ③ Quantitative aggregation risk assessment: For specific business units Overall risk (such as orders, product lines) By aggregating its associated risk nodes Probability and Influence Weight The calculation shows that: ; in, It can be a weighted average or other function.

[0033] ④ Risk chain reaction prediction: based on graphs LLM can simulate initial risk. The propagation path is predicted, and its impact on downstream nodes is assessed. probability of influence .

[0034] ⑤ DLT anchoring of critical states: important updates to the graph, such as critical risk probabilities Exceeding the threshold Its status summary With timestamp Recorded in the DLT, the evolution process is made transparent and auditable. This DCRG mechanism directly solves the problems of superficial risk prediction, lack of real-time updates, and difficulty in traceability in existing technologies through deep causal mining, dynamic real-time updates, and DLT anchoring.

[0035] (2) Implement an LLM-Driven Adaptive SmartContract Response (LIA-SCR) strategy to achieve intelligent, flexible, and optimized risk response (see...). Figure 4).

[0036] To address the issue of traditional smart contracts having rigid response logic and being unable to adapt to specific scenarios, this solution designs the LIA-SCR mechanism, endowing smart contracts with dynamic decision-making capabilities: When risk events When a smart contract is triggered, it no longer simply executes fixed rules, but instead sends a detailed context containing the current risk to the LLM policy service via an oracle. (Query requests derived from DCRG analysis results, etc.)

[0037] ①After receiving a request, the LLM, based on the context, selects from the optional policy set. Choose the one that maximizes the expected utility strategy : ; The utility function takes into account the expected reduction in risk. (Estimated by LLM) and execution costs , These are the weighting coefficients.

[0038] ② LLM's estimation of risk reduction Based on the strategy Simulation of the effect on DCRG. Let If the expected graph state after applying the strategy is: ; in, Key risk indicators, such as the probability of risk at a specific node or an aggregated risk score, It encapsulates the simulation of policy effects using LLM.

[0039] ③ The smart contract receives the recommendation strategy Then, according to preset rules (e.g., utility) And cost Decide whether to execute automatically or... The strategy and its justification (cited via VEP) are submitted to the authorizing body for on-chain signature approval. The final adopted strategy... The basis for their decisions is recorded in the DLT.

[0040] The LIA-SCR mechanism directly addresses the problems of rigid, untargeted, and inefficient risk response strategies in existing technologies by introducing the real-time intelligent analysis and optimization capabilities of LLM.

[0041] (3) Establish a DLT-based Verifiable Explanation Path for LLM Risk Analysis (VEP) to overcome the "black box" problem of AI and enhance the transparency and credibility of the system (see Figure 5 ).

[0042] To address the issues of opacity, lack of trust, and difficulty in auditing the LLM decision-making process, this solution designs a VEP mechanism to provide a verifiable traceability path for LLM risk assessment: ① During the analysis process, the internal mechanism of LLM is designed to track and identify the key sources of evidence upon which the reasoning is based; ② The system generates a structured, verifiable explanation record (VER) for each risk assessment: ; in, Each piece of evidence was clearly recorded. Source type (e.g., DLT transactions, IoT data anchoring, oracle reports), unique identifiers (such as transaction hash, data URI) and proof required for verification (such as MerkleProof, oracle signatures); ③The overall hash of VER It is recorded in the DLT network and associated with the corresponding risk assessment event; ④ When verification is required, the user or auditor shall perform the following: The process includes: A. Verify the hash of the off-chain VER Is it related to on-chain records? Consistent; B. Verify the interpretation text hash Does it match the record in VER? C. Every piece of evidence , call The function performs independent verification, for example, for DLT transaction evidence: , including calculating the Merkle root and the Merkle root of the corresponding block header. The logic for comparison is as follows. Only when all verification steps pass can the risk assessment and its interpretation by the LLM be considered credible and verifiable.

[0043] The VEP mechanism leverages the immutability of DLT to solidify the interpretation path and evidence index, directly addressing the trust deficiency and auditing difficulties caused by the "black box" nature of AI models in existing technologies.

[0044] (4) Integration and synergy of the above mechanisms: The core of this application lies in tightly integrating the deep insights of DCRG, the intelligent response of LIA-SCR, and the trusted interpretation of VEP through DLT—a trusted infrastructure—and smart contracts—an automated link—forming a complete closed loop from data perception, intelligent analysis, dynamic decision-making, automated execution to trusted verification. DLT is not only a trusted carrier of data but also a non-repudiable recording platform for collaborative behaviors (such as policy approval and result confirmation), effectively solving the problems of information silos and low collaborative efficiency in existing technologies.

[0045] Preferably, step A specifically includes the following steps: A1. Data Injection: The data sensing and integration layer continuously obtains confirmed transaction data from the distributed ledger network in real time, including order status updates and logistics node information, and collects information from off-chain data sources through oracles. The off-chain data sources include IoT sensors, news APIs, market data, weather services, etc. A2. Data Fusion and Preprocessing: The data preprocessing and fusion unit cleans, formats, links entities, and aligns the collected multi-source heterogeneous data. A3, DCRG Update: The preprocessed data is sent to the intelligent analysis and decision-making layer, where the DCRG construction and inference module uses LLM to analyze the new data, based on the aforementioned formula (e.g. Updates (Adjustments) Real-time updates to the dynamic causal risk map This reflects the latest changes in the status of the supply chain and potential risks.

[0046] Preferably, step B specifically includes the following steps: B1. Risk Event Identification: LLM analyzes the dynamic causal risk map of DCRG. Or, when directly analyzing the data stream, identify the risk probability of specific risk nodes. Or aggregate risk scores Exceeding the preset warning threshold This allows for the identification of potential risk events. ,in, Represents the set of supply chain entities and risk factor nodes. Represents the passage of time The set of potential causal relationships or dependencies of change It is the edge weight function, which quantifies the probability of risk transmission or the intensity of its impact; B2. In-depth assessment and interpretation: For identified risk events... LLM performs in-depth assessments to predict the potential scope and severity of its impact and generates human-readable natural language explanations. B3. VEP Generation and Anchoring: Simultaneously, the VEP generation module follows the defined verifiable and interpretable record structure. Generate verifiable explanation records for this risk assessment. VER It includes the assessment results, explanatory text, text hash, and a detailed list of evidence sources, and calculates... VER Overall hash And anchor this hash to the distributed ledger network via DLT transactions, in conjunction with risk events. Related.

[0047] Preferably, step C specifically includes the following steps: C1. Risk Signal Transmission: The intelligent analysis and decision-making layer will identify risk events. Key information (such as risk type, level, scope of impact summary) and The information is transmitted via an oracle to a risk warning smart contract deployed on DLT. C2. Preliminary Contract Assessment: The risk warning smart contract makes assessments based on the received risk information and built-in basic rules; C3, LIA-SCR Query: For risk events requiring intelligent decision-making, the risk warning smart contract initiates a query to the LLM strategy service via an oracle, attaching the current risk context. ; C4, LLM Strategy Recommendation: The LLM strategy service selects recommended strategies based on the risk context. The optimization objective is to reduce the amount of expected risks. Conduct simulation evaluations to generate the optimal recommendation strategy. This includes its expected utility assessment and reasons for recommendation (which can be explained using the new VER).

[0048] Preferably, step D specifically includes the following steps: D1. Strategy Delivery and Presentation: The recommended strategy returned by LLM The risk warning is transmitted back to the smart contract via an oracle; D2. Automated or manual approval: D2.1 Automatic Execution: If the recommended strategy is... If the preset automatic execution conditions are met (such as risk level, cost, and utility being within the safety threshold), the risk warning smart contract can directly proceed to the next execution stage; D2.2 Manual Approval: If the conditions for automatic execution are not met or the system is configured to require manual confirmation, the risk warning smart contract will recommend a strategy. The expected utility and associated risk explanations and reasons for recommendation are pushed to users with appropriate permissions (such as risk managers and purchasing managers) through the application interaction layer, so that users can view detailed information on the interface and verify the credibility of the explanation through the VEP mechanism. D3. On-chain decision confirmation: If manual approval is required, authorized users interact with their DLT identity (wallet) through the application interaction layer to confirm the recommendation strategy. The decision to approve or reject a transaction is confirmed by signing a DLT transaction, recorded on the ledger, and the risk warning smart contract is notified.

[0049] Preferably, step E specifically includes the following steps: E1. Contract Execution: The risk warning smart contract executes according to the finalized strategy (automatic adoption or manual approval). Perform the corresponding on-chain or off-chain actions: E1.1 On-chain actions: such as updating the status markers of related assets or orders on DLT (e.g., "Risk warning in progress", "Payment suspended"), recording decision results, and sending messages to other related contracts (e.g., activating the collaborative management contract). E1.2 Off-chain action triggering: Calling the oracle to trigger off-chain operations, including sending email / SMS notifications to relevant personnel, calling the API of the enterprise ERP / SCM system (such as creating alternative supplier orders, adjusting inventory parameters); instructing physical devices (such as adjusting production line parameters, which requires careful permission design). E2. Execution Result Recording: Key execution steps and results, or their hash values, are recorded back to the DLT to ensure process traceability.

[0050] Preferably, step E further includes the step: Multi-party collaboration and information sharing: Multi-party collaboration and information sharing are achieved through initiating collaborative processes, information sharing and interaction, and on-chain collaborative records.

[0051] Specifically, the multi-party collaboration and information sharing are achieved through initiating a collaboration process, information sharing and interaction, and on-chain collaboration records, including the following steps: E3. Collaborative Process Initiation: For complex risk events that require multi-party collaboration, the risk warning contract or manual decision-making triggers the creation and activation of an instance of the collaborative management contract. E4. Information Sharing and Interaction: Relevant participants can access DLT data, LLM risk reports (including VER verification links), and risk warning smart contract status information related to the event through the collaborative workspace of the application interaction layer. E5. On-chain Collaboration Records: Participants can conduct discussions within the collaboration space (key discussion summaries can be hashed and uploaded to the chain), and submit confirmation information, voting results, and jointly developed detailed response plans via DLT signatures (plan text hashed and uploaded to the chain), which are recorded by the collaboration management contract.

[0052] Preferably, the supply chain intelligent risk management method based on DLT and LLM further includes the following steps: F. Continuous monitoring and feedback learning: By continuously monitoring the actual effects of the implemented response strategies, the risk prediction model and strategy recommendation capabilities of LLM are continuously optimized, forming a learning loop.

[0053] Specifically, step F includes the following steps: F1. Effect Tracking: The system continuously monitors the development of risk events and the actual effect of the implemented response strategies, including through subsequent transaction data on DLT, IoT sensor feedback, and external information updates; F2, DCRG, and Model Iteration: Monitoring results and policy implementation effectiveness data are fed back to the intelligent analysis and decision-making layer to further update the dynamic causal risk map of DCRG. It can also be used as new training / fine-tuning data to continuously optimize the risk prediction model and strategy recommendation capabilities of LLM, forming a learning loop.

[0054] Figure 6 This is a workflow sequence diagram of a DLT and LLM-based intelligent risk management method for supply chain, which is another preferred embodiment of this application. It belongs to the typical risk handling workflow sequence diagram of this application.

[0055] The key innovations of the above embodiments of this application include: This application addresses the shortcomings of existing technologies, and its core innovations are mainly reflected in the following aspects, which together constitute the uniqueness and advancement of this application: (a) Deep causal and dynamic evolutionary risk modeling (based on DCRG): Innovation: This paper proposes and implements a Dynamic Causal Risk Map (DCRG) based on LLM construction and real-time maintenance for the first time. This map not only represents the relationships between supply chain entities, but more importantly, it uses LLM to deeply mine and quantify the causal transmission relationships between nodes and their dynamic weights, surpassing the traditional shallow risk assessment based on statistical correlation.

[0056] Value: It enables a shift in the analysis of supply chain risks from static, isolated analysis to dynamic, systematic insights, allowing for more accurate prediction of the cascading effects and potential impacts of risks.

[0057] (b) Adaptive decision-making capability of smart contracts (based on LIA-SCR): Innovation: An LLM-driven Adaptive Smart Contract Response (LIA-SCR) mechanism was designed. By allowing smart contracts to query the LLM to obtain optimal policy recommendations based on real-time context when risks are triggered, the limitation of traditional smart contracts that can only execute fixed rules is broken.

[0058] Value and role: It endows automated risk response processes with unprecedented intelligence, flexibility and optimization capabilities, making response measures more precise and effective.

[0059] (c) Credibility and transparency of AI risk analysis (based on VEP): Innovation: A Verifiable Explanatory Path (VEP) mechanism based on DLT was created. This mechanism provides an end-to-end, independently verifiable chain of evidence for LLM risk assessment results, and records the key anchor points of the explanation path on an immutable DLT.

[0060] Value and impact: It effectively solves the "black box" problem of LLM and significantly improves the transparency, auditability and user trust of AI-driven risk management systems.

[0061] (d) Deep native integration of DLT and LLM: Innovation: This application does not simply connect DLT and LLM as two independent systems, but achieves deep native integration of the two at the data level (DLT provides credible facts, LLM performs in-depth interpretation), the logic level (LLM empowers smart contract decision-making), and the trust level (DLT anchors LLM processes and results) through mechanisms such as DCRG, LIA-SCR, and VEP.

[0062] Value and role: It fully leverages the complementary advantages of the two technologies, resulting in a synergistic effect of 1+1>2, and constructs a new paradigm of risk management that is significantly superior in function and performance to a single technology or a simple combination.

[0063] (e) A closed loop of end-to-end automation and trusted collaboration: Innovation: Integrating data perception, intelligent analysis, dynamic decision-making, automated execution, result recording, trusted verification, and multi-party collaboration into a closed-loop system based on DLT trusted infrastructure.

[0064] Value: While ensuring full transparency, traceability, and immutability, it maximizes the automation level and collaborative efficiency of risk management.

[0065] The core of this invention lies in using the deep integration of DLT, LLM, and smart contracts to solve supply chain risk management problems; however, the specific implementation method is not unique. Without departing from the basic principles and scope of this invention, various alternative technical implementations or variations are possible, for example: (a) Alternatives to the choice of distributed ledger platforms: While the foregoing embodiments may be geared towards enterprise-grade consortium blockchains (such as Hyperledger Fabric), the present invention is equally applicable to other types of DLT platforms, such as: 1. Other consortium blockchain technologies: such as R3 Corda, whose unique state objects and peer-to-peer communication model may be suitable for certain financial or transaction-intensive supply chain scenarios.

[0066] 2. High-performance public blockchains or their Layer 2 networks: For scenarios requiring greater transparency or broader participation, consider using Layer 2 solutions such as Ethereum (e.g., Polygon, Arbitrum, Optimism) or high-performance public blockchains with enterprise-grade features (e.g., Solana, Avalanche), but these should be combined with stronger privacy protection mechanisms (e.g., zero-knowledge proofs ZKP).

[0067] 3. Directed Acyclic Graph (DAG) technology: For specific processes that require extremely high concurrent transaction processing capabilities (such as anchoring large amounts of IoT data), DLT based on DAG (such as Hedera Hashgraph, IOTA) can be considered.

[0068] (b) Alternatives to Large Language Model (LLM) implementations: 1. Pure API Call Mode: The system can rely entirely on external commercial LLM APIs (such as OpenAI GPT series, Anthropic Claude, Google Gemini / PaLM) for risk analysis and strategy recommendation. In this mode, the innovation focus of this invention will be more on: secure and reliable interaction protocols with DLT and oracles, effective anonymization and privacy protection technologies for input data, complex prompt engineering design for supply chain risks, and verification and fault tolerance mechanisms for API return results; 2. Hybrid LLM Architecture: A hybrid model can be adopted, for example, using a locally deployed, fine-tuned smaller model to handle internal sensitive data and perform core DCRG updates, while calling powerful external APIs to process public information (such as news analysis, general knowledge Q&A) or perform complex policy generation, to balance cost, performance and privacy; 3. Application of Federated Learning: For scenarios involving multiple parties and highly sensitive data, federated learning can be used to perform distributed training or fine-tuning of LLM. Each party updates the model parameters locally using its own data, exchanging only encrypted or aggregated model update information, thereby jointly building a more powerful risk model without sharing the original data. 4. Model Ensemble: Multiple different LLMs (local or API) can be used simultaneously to analyze the same risk event, and the results can be integrated through voting, weighted averaging or other ensemble methods to improve the robustness and accuracy of the analysis.

[0069] Another preferred embodiment of this application also provides a supply chain intelligent risk management device based on DLT and LLM, including: The data awareness and DCRG dynamic maintenance module is used to continuously acquire confirmed transaction data from the distributed ledger network in real time, and dynamically update the dynamic causal risk graph DCRG anchored by DLT based on the acquired transaction data. The dynamic causal risk graph is constructed and maintained in real time using LLM. The identification, assessment, and interpretation generation module is used for risk identification, assessment, and verifiable interpretation generation: After identifying potential risk events using the LLM analysis dynamic causal risk map, it assesses the scope and severity of the risk events' impact and generates verifiable interpretation records for this risk assessment. These hashes are then anchored to the distributed ledger network via DLT transactions. ； The early warning and adaptive strategy generation module is used for risk early warning smart contracts to generate early warning and adaptive strategies: the risk early warning smart contract generates early warning and adaptive strategies based on risk events. After obtaining the current risk context by querying the key information of the adaptive smart contract response strategy driven by LLM, LLM generates and returns the optimal recommended strategy through simulation evaluation. The decision-making and permission confirmation module is used for decision-making and permission confirmation: the risk warning smart contract receives the optimal recommendation strategy. And based on the optimal recommendation strategy Whether the preset automatic execution conditions are met or whether manual approval is required before proceeding to the corresponding execution stage; The response strategy execution and recording module is used for the automated execution and recording of response strategies: the risk warning smart contract automatically adopts or manually approves strategies based on the final determination. It executes the corresponding on-chain or off-chain actions and records the execution results back to the DLT.

[0070] The DLT and LLM-based intelligent supply chain risk management device provided in this embodiment adopts the DLT and LLM-based intelligent supply chain risk management method in the above embodiments, solving the technical problems of existing supply chain risk management such as lag and data silos, superficial risk analysis, rigid and inefficient response, and lack of transparency and trust. Compared with the prior art, the beneficial effects of the DLT and LLM-based intelligent supply chain risk management device provided in this embodiment are the same as those of the DLT and LLM-based intelligent supply chain risk management method provided in the above embodiments. Moreover, other technical features in the DLT and LLM-based intelligent supply chain risk management device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0071] like Figure 7 As shown, a preferred embodiment of this example also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the supply chain intelligent risk management method based on DLT and LLM in the above embodiment.

[0072] This embodiment provides an electronic device that employs the DLT and LLM-based intelligent supply chain risk management method described in the above embodiments. This addresses the technical problems of existing supply chain risk management, such as lag, data silos, superficial risk analysis, rigid and inefficient response, and lack of transparency and trust. Compared with the prior art, the beneficial effects of the electronic device provided in this embodiment are the same as those of the DLT and LLM-based intelligent supply chain risk management method provided in the above embodiments. Furthermore, other technical features of the electronic device are the same as those disclosed in the methods of the above embodiments, and will not be elaborated upon here.

[0073] like Figure 8 As shown in the preferred embodiment, this embodiment also provides a computer device, which may be a terminal or a liveness detection server, and its internal structure diagram may be as follows. Figure 8 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used to communicate with other external computer devices via a network connection. When the computer program is executed by the processor, it implements the steps of the aforementioned DLT and LLM-based intelligent supply chain risk management method.

[0074] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the solution of this embodiment, and does not constitute a limitation on the computer device to which the solution of this embodiment is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0075] The computer equipment provided in this application adopts the DLT and LLM-based intelligent supply chain risk management method in the above embodiments, which solves the technical problems of existing supply chain risk management, such as lag and data silos, superficial risk analysis, rigid and inefficient response, and lack of transparency and trust. Compared with the prior art, the beneficial effects of the computer equipment provided in this embodiment are the same as those of the DLT and LLM-based intelligent supply chain risk management method provided in the above embodiments, and other technical features in the electronic equipment are the same as those disclosed in the above embodiments, which will not be repeated here.

[0076] A preferred embodiment of this example also provides a storage medium, which includes a stored program that, when the program is executed, controls the device where the storage medium is located to perform the steps of the supply chain intelligent risk management method based on DLT and LLM in the above embodiment.

[0077] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0078] If the functions described in this embodiment are implemented as software functional units and sold or used as independent products, they can be stored in one or more computing device-readable storage media. Based on this understanding, the parts of this embodiment that contribute to the prior art or the technical solution can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a computing device (which may be a personal computer, server, mobile computing device, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this embodiment. The aforementioned storage media include: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0079] Those skilled in the art will understand that embodiments of this example can be provided as methods, systems, or computer program products. Therefore, this example can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this example can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in this example can be implemented using various computer languages, such as the object-oriented programming language C++ and the embedded programming language C.

[0080] This embodiment is described with reference to flowchart illustrations and / or block diagrams of the method, apparatus (system), and computer program product according to this embodiment. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0081] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0082] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0083] This embodiment also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the supply chain intelligent risk management method based on DLT and LLM as described above.

[0084] The computer program product provided in this embodiment addresses the technical problems of existing supply chain risk management, such as lag, data silos, superficial risk analysis, rigid and inefficient responses, and a lack of transparency and trust. Compared with the prior art, the beneficial effects of the computer program product provided in this embodiment are the same as those of the DLT and LLM-based intelligent supply chain risk management methods provided in the above embodiments, and will not be elaborated upon here.

[0085] Obviously, those skilled in the art can make various modifications and variations to this embodiment without departing from the spirit and scope of this embodiment. Therefore, if these modifications and variations of this embodiment fall within the scope of the claims of this embodiment and their equivalents, this embodiment is also intended to include these modifications and variations.

Claims

1. A supply chain intelligent risk management method based on DLT and LLM, characterized in that, Including the following steps: A. Continuously acquire confirmed transaction data from the distributed ledger network in real time, and dynamically update the dynamic causal risk graph DCRG anchored by DLT based on the acquired transaction data. The dynamic causal risk graph is constructed and maintained in real time using LLM. B. Risk Identification, Assessment, and Verifiable Explanation Generation: After identifying potential risk events using the LLM analysis of the dynamic causal risk graph, the scope and severity of the risk events are assessed, and a verifiable explanation record for this risk assessment is generated. This hash is then anchored to the distributed ledger network via DLT transactions. ； C. Risk Warning Smart Contract Early Warning and Adaptive Strategy Generation: The risk warning smart contract generates early warning and adaptive strategies based on risk events. After obtaining the current risk context by querying the key information of the adaptive smart contract response strategy driven by LLM, LLM generates and returns the optimal recommended strategy through simulation evaluation. D. Decision Making and Authority Confirmation: The risk warning smart contract receives the optimal recommended strategy. And based on the optimal recommendation strategy Whether the preset automatic execution conditions are met or whether manual approval is required before proceeding to the corresponding execution stage; E. Automated execution and recording of response strategies: The risk warning smart contract automatically adopts or manually approves strategies based on the final determination. It executes the corresponding on-chain or off-chain actions and records the execution results back to the DLT.

2. The supply chain intelligent risk management method based on DLT and LLM according to claim 1, characterized in that, Step A specifically includes the following steps: A1. Data Injection: The data sensing and integration layer continuously obtains confirmed transaction data from the distributed ledger network in real time, including order status updates and logistics node information, and collects information from off-chain data sources through oracles. The off-chain data sources include IoT sensors, news APIs, market data, and weather services. A2. Data Fusion and Preprocessing: The data preprocessing and fusion unit cleans, formats, links entities, and aligns the collected multi-source heterogeneous data. A3, DCRG Update: The preprocessed data is sent to the intelligent analysis and decision-making layer, where the DCRG construction and inference module uses LLM to analyze new data and update the dynamic causal risk map in real time. This reflects the latest changes in the status of the supply chain and potential risks.

3. The intelligent supply chain risk management method based on DLT and LLM according to claim 2, characterized in that, Step B specifically includes the following steps: B1. Risk Event Identification: LLM analyzes the dynamic causal risk map of DCRG. Or, when directly analyzing the data stream, identify the risk probability of specific risk nodes. Or aggregate risk scores Exceeding the preset warning threshold This allows for the identification of potential risk events. ,in, Represents the set of supply chain entities and risk factor nodes. Represents the passage of time The set of potential causal relationships or dependencies of change It is the edge weight function, which quantifies the probability of risk transmission or the intensity of its impact; B2. In-depth assessment and interpretation: For identified risk events... LLM performs in-depth assessments to predict the potential scope and severity of its impact and generates human-readable natural language explanations. B3. VEP Generation and Anchoring: Simultaneously, the VEP generation module generates verifiable explanation records for this risk assessment according to the set verifiable explanation record structure. VER It includes the assessment results, explanatory text, text hash, and a detailed list of evidence sources, and calculates... VER Overall hash And anchor this hash to the distributed ledger network via DLT transactions, in conjunction with risk events. Related.

4. The intelligent supply chain risk management method based on DLT and LLM according to claim 3, characterized in that, Step C specifically includes the following steps: C1. Risk Signal Transmission: The intelligent analysis and decision-making layer will identify risk events. Key information and The information is transmitted via an oracle to a risk warning smart contract deployed on DLT. C2. Preliminary Contract Assessment: The risk warning smart contract makes assessments based on the received risk information and built-in basic rules; C3, LIA-SCR Query: For risk events requiring intelligent decision-making, the risk warning smart contract initiates a query to the LLM strategy service via an oracle, attaching the current risk context. ; C4, LLM Strategy Recommendation: The LLM strategy service selects recommended strategies based on the risk context. The optimization objective is to reduce the amount of expected risks. Conduct simulation evaluations to generate the optimal recommendation strategy. This includes its expected utility assessment and reasons for recommendation.

5. The supply chain intelligent risk management method based on DLT and LLM according to claim 4, characterized in that, Step D specifically includes the following steps: D1. Strategy Delivery and Presentation: The recommended strategy returned by LLM The risk warning is transmitted back to the smart contract via an oracle; D2. Automated or manual approval: D2.1 Automatic Execution: If the recommended strategy is... Once the preset automatic execution conditions are met, the risk warning smart contract can directly proceed to the next execution stage; D2.2 Manual Approval: If the conditions for automatic execution are not met or the system is configured to require manual confirmation, the risk warning smart contract will recommend a strategy. The expected utility and associated VER are pushed to users with the corresponding permissions through the application interaction layer, so that users can view detailed information on the interface and verify the credibility of the explanation through the VEP mechanism; D3. On-chain decision confirmation: If manual approval is required, authorized users interact with their DLT identity through the application interaction layer to confirm the recommendation strategy. The decision to approve or reject a transaction is confirmed by signing a DLT transaction, recorded on the ledger, and the risk warning smart contract is notified.

6. The intelligent supply chain risk management method based on DLT and LLM according to claim 5, characterized in that, Step E specifically includes the following steps: E1. Contract Execution: The risk warning smart contract executes according to the final determined strategy. Perform the corresponding on-chain or off-chain actions: E1.1 On-chain actions: such as updating the status markers of related assets or orders on DLT, recording decision results, and sending messages to other related contracts; E1.2 Off-chain action triggering: Calling the oracle to trigger off-chain operations, including sending email / SMS notifications to relevant personnel, calling the API of the enterprise ERP / SCM system, and instructing physical devices; E2. Execution Result Recording: Key execution steps and results, or their hash values, are recorded back to the DLT to ensure process traceability.

7. The supply chain intelligent risk management method based on DLT and LLM according to claim 1 or 6, characterized in that, Step E further includes the following step: Multi-party collaboration and information sharing: Multi-party collaboration and information sharing are achieved through initiating collaborative processes, information sharing and interaction, and on-chain collaborative records.

8. The intelligent supply chain risk management method based on DLT and LLM according to claim 7, characterized in that, The multi-party collaboration and information sharing are achieved through initiating collaboration processes, information sharing and interaction, and on-chain collaboration records. Specifically, the steps include: E3. Collaborative Process Initiation: For complex risk events that require multi-party collaboration, the risk warning contract or manual decision-making triggers the creation and activation of an instance of the collaborative management contract. E4. Information Sharing and Interaction: Relevant participants can access DLT data, LLM risk reports, and risk warning smart contract status information related to the event through the collaborative workspace of the application interaction layer. E5. On-chain Collaboration Records: Participants can discuss within the collaboration space and submit confirmation information, voting results, and jointly developed detailed response plans via DLT signatures, all of which are recorded by the collaboration management contract.

9. The intelligent supply chain risk management method based on DLT and LLM according to claim 1, characterized in that, It also includes the following steps: F. Continuous monitoring and feedback learning: By continuously monitoring the actual effects of the implemented response strategies, the risk prediction model and strategy recommendation capabilities of LLM are continuously optimized, forming a learning loop.

10. The intelligent supply chain risk management method based on DLT and LLM according to claim 9, characterized in that, Step F specifically includes the following steps: F1. Effect Tracking: The system continuously monitors the development of risk events and the actual effect of the implemented response strategies, including through subsequent transaction data on DLT, IoT sensor feedback, and external information updates; F2, DCRG, and Model Iteration: Monitoring results and policy implementation effectiveness data are fed back to the intelligent analysis and decision-making layer to further update the dynamic causal risk map of DCRG. It can also be used as new training / fine-tuning data to continuously optimize the risk prediction model and strategy recommendation capabilities of LLM, forming a learning loop.

Images