Supply chain partner risk management method based on blockchain

Abstract

The present invention relates to the field of blockchain technology, and in particular to a supply chain partner risk management method based on blockchain, comprising the following steps: collecting dynamic data from multiple sources and extracting multi-dimensional risk feature vectors to achieve a comprehensive and accurate assessment of supply chain partner risks; utilizing blockchain technology to store risk metadata on-chain, thereby enhancing the transparency and credibility of risk management; dynamically adjusting risk assessment weights through a reinforcement learning algorithm to achieve intelligent and adaptive risk assessment, which is conducive to flexible adjustment of risk management strategies and improved efficiency and accuracy of risk response; and generating a risk diffusion trajectory map with a timestamp and automatically matching risk disposal strategies to achieve dynamic and optimized risk management, which is conducive to timely control of risk diffusion, reduction of risk losses, and improvement of the overall stability and competitiveness of the supply chain.

Description

Technical Field

[0001] The present invention relates to the field of blockchain technology, and in particular to a supply chain partner risk management method based on blockchain. Background Art

[0002] The current supply chain partner risk management still faces some challenges, including the following aspects: in traditional supply chain management, the data between the participants is often opaque, resulting in information asymmetry and lack of trust. It is difficult for core enterprises, suppliers, manufacturers and other parties to fully and accurately obtain the risk data of other participants, thereby affecting the accuracy of risk assessment and the effectiveness of risk management; traditional risk warning mechanisms may lack real-time monitoring and automatic warning functions, resulting in risks not being discovered and warned in a timely manner. Even if risks are discovered, they may not be responded to in a timely manner due to untimely warning information transmission or cumbersome processing procedures, and the spread of risks cannot be effectively controlled; in the supply chain, risks may be transmitted along the transaction relationship and affect other participants. However, traditional methods are difficult to effectively simulate the risk transmission path, resulting in an inability to accurately assess the impact of risks on the entire supply chain; at the same time, traditional risk management processes may involve multiple links and participants, resulting in cumbersome and inefficient processes. Due to the lack of an effective effect evaluation mechanism, it is impossible to timely understand the actual effects of risk management measures, making it difficult to continuously optimize and improve the risk management process. To this end, the present invention proposes a supply chain partner risk management method based on blockchain. Summary of the Invention

[0003] The purpose of this invention is to solve the problems in the background technology and propose a supply chain partner risk management method based on blockchain.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] Blockchain-based supply chain partner risk management methods include:

[0006] S1. Collect dynamic data from multiple sources, including transaction performance and compliance records; extract and integrate multi-dimensional risk feature vectors including credit risk, operational risk, and legal risk through data analysis methods, generate standardized risk metadata, and store them on-chain;

[0007] S2. Deploy a risk assessment and early warning smart contract on the consortium chain, extract the multi-dimensional risk feature vectors in the risk metadata certificate, and calculate a comprehensive risk score. Dynamically adjust the weights of each dimension through a reinforcement learning algorithm to achieve risk assessment and early warning functions.

[0008] S3. Obtain risk assessment and early warning results, build a supply chain topology network based on transaction data stored on the blockchain, simulate the risk transmission process through the risk value diffusion formula, and generate a time-stamped risk diffusion trajectory map for sharing;

[0009] S4. Based on the shared information of the risk diffusion trajectory map, automatically match and execute risk disposal strategies, while continuously evaluating the risk disposal effects and making dynamic adjustments to optimize risk management.

[0010] Furthermore, the process of extracting and integrating multi-dimensional risk feature vectors including credit risk, operational risk, and legal risk through data analysis methods, generating standardized risk metadata, and storing them on-chain includes:

[0011] Construct a three-dimensional risk matrix, including credit risk, operational risk, and legal risk dimensions: Calculate the credit risk characteristic vector by using the historical payment delay rate and order fulfillment rate: The calculation formula for the historical payment delay rate is: , where is the historical payment delay rate, is the total number of orders during the statistical period, is the order index within the statistical period, For the The actual payment date of the order, For the The payment date is agreed upon in the order contract; the calculation formula for the order fulfillment rate is: , where is the order fulfillment rate, To deliver the orders on time, is the total number of orders; integrating the historical payment delay rate and order fulfillment rate to obtain the credit risk characteristic vector ;

[0012] The operational risk characteristic vector is calculated based on the standard deviation of logistics timeliness and inventory turnover deviation: The calculation formula for the standard deviation of logistics timeliness is: , where is the standard deviation of logistics timeliness, For logistics index, is the number of logistics, For the Actual logistics timeliness, is the average value of logistics efficiency; the calculation formula for inventory turnover deviation is: , where is the inventory turnover deviation, is the average inventory turnover days, is the actual inventory turnover days; integrating the logistics efficiency standard deviation and inventory turnover deviation to obtain the operational risk characteristic vector ;

[0013] The legal risk feature vector is calculated based on the environmental protection penalty record and the number of days the license expires: , where is the environmental penalty rate, is the number of environmental penalties, is the total number of operating days; the formula for calculating the percentage of license expiration days is: , where is the percentage of license expiration days, is the total number of days the license is invalid; integrating the environmental penalty rate and the proportion of license invalid days to obtain the legal risk characteristic vector ;

[0014] Using the JSON-LD format, the multi-dimensional risk feature vector is serialized into a standardized risk metadata package; the serialized JSON-LD data is hashed to generate a unique data hash value; the generated JSON-LD data package is used to call the alliance chain's evidence contract; the evidence contract will record the complete JSON-LD data package and generate a digitally signed risk data evidence certificate, which contains a timestamp and data hash value; based on the risk data evidence certificate, the data hash value is used as input and written into the alliance chain's Merkle tree; the root node of the Merkle tree will be recorded on the blockchain; based on the record of the Merkle tree, a risk metadata certificate with a timestamp is generated, which contains a data source signature and a Merkle path.

[0015] Furthermore, the process of extracting the multi-dimensional risk feature vector from the risk metadata certificate and calculating the comprehensive risk score includes:

[0016] The risk metadata certificate generated by the on-chain verification engine is called to extract the embedded JSON-LD format multi-dimensional risk feature vector and evidence credential information from the certificate. The Merkle path in the certificate is simultaneously parsed to extract the root node hash value and use it as the reference anchor point.

[0017] The risk assessment and warning smart contract reads the stored multi-dimensional risk feature vector from the temporary storage area and parses it out. It uses the multi-dimensional risk feature vector of the current cycle and the historical risk event library stored on the chain as input states.

[0018] The risk feature vector of each dimension is normalized and the dot product is calculated with the initial weight to generate a comprehensive risk score.

[0019] Furthermore, the process of dynamically adjusting the weights of each dimension through reinforcement learning algorithms to achieve risk assessment and early warning functions includes:

[0020] Introducing the supply chain risk reinforcement learning RL agent model, using the Q-learning algorithm to dynamically adjust the weight of each dimension: define the risk dimension set as , the initial weight of each dimension is ,satisfy ;

[0021] Define the state space as , represents the matching degree between the current multi-dimensional risk feature vector and the historical event database, that is, ,in is the time step, which represents the number of iterations of risk assessment and early warning smart contract learning or the discrete time point of risk events; is the L2 norm;

[0022] Define the action space as , used to adjust the weight direction, that is, to increase or decrease the dimension weight: ,in ;

[0023] The risk assessment and warning smart contract dynamically updates weights using the following Q-learning rules:

[0024] ,

[0025] Where, is the updated weight; It is an immediate reward, which is obtained based on the risk score change and the disposal cost; is the learning rate; is the discount risk factor, For the next state Candidate weight values ​​of ;

[0026] The weight adjustment results are written into the blockchain through the risk assessment and early warning smart contract, forming a risk event-weight adjustment learning mechanism: the risk assessment and early warning smart contract calls the historical transaction behavior fingerprints stored on the chain to recalculate the credit risk characteristics, operational risk characteristics, and legal risk characteristic vectors: , , ;

[0027] Compare the recalculated multidimensional risk feature vector with the original multidimensional risk feature vector generated by the three-dimensional risk matrix dimension by dimension:

[0028] ,

[0029] Where, is the risk deviation rate, is the original multidimensional risk feature vector, is the recalculated multidimensional risk feature vector, is the risk feature vector category;

[0030] Preset risk deviation thresholds β1 and β2; when the calculated risk deviation rate exceeds the risk deviation threshold, a graded warning is automatically triggered, i.e. high risk ≥ β1, medium risk [β2, β1), low risk < β2: when it is identified as a high-risk level, i.e. the risk deviation rate ≥ β1, a structured warning event package containing the risk level and risk feature category is generated and broadcast to all related nodes through the P2P network of the alliance chain, including core enterprise nodes, supplier nodes and manufacturer nodes, and the warning timestamp, triggering node ID and event hash value are recorded on the chain; when it is identified as a medium-risk level, i.e. the risk deviation rate belongs to [β2, β1), a warning notification is pushed to the risk control department of the core enterprise in the supply chain, and the node is marked as an observation object on the chain to limit its priority cooperation rights; when it is identified as a low-risk level, i.e. the risk deviation rate < β2, no warning is triggered;

[0031] Integrate risk deviation rate and warning event information, and generate new on-chain blocks through risk assessment and warning smart contracts.

[0032] Furthermore, the process of obtaining risk assessment and early warning results, building a supply chain topology network based on transaction data stored on the blockchain, simulating the risk transmission process through the risk value diffusion formula, and generating a risk diffusion trajectory map with a timestamp for sharing includes:

[0033] Each associated node in the supply chain receives the broadcasted structured warning event package through the P2P network of the alliance chain. The associated nodes include core enterprise nodes, supplier nodes, and manufacturer nodes. Each associated node parses the received warning event package, extracts key information, and stores it in the local database.

[0034] Utilize the historical transaction records stored on the blockchain, i.e. transaction amount and transaction frequency; construct a supply chain topology network based on transaction dependencies: use identity DID identifiers as nodes of the supply chain topology network, i.e. core enterprise nodes, supplier nodes, and manufacturer nodes; use transaction amount and transaction frequency weighted values ​​as edges of the supply chain topology network; locate the warning node triggered by the warning event, i.e. the risk source node, in the constructed supply chain topology network; set the initial risk value of the risk source node to ,The weight set of edges in the supply chain topology network is ,in Representation node With node The transaction amount and transaction frequency weighted value between nodes With node The weight of the edge between them; along the edges in the supply chain topology network, the risk value is diffused according to the edge weight attenuation, where the risk value diffusion formula is:

[0035] ,

[0036] Where, For in time Time Node The risk value, For nodes The set of neighbor nodes of For nodes All neighbor nodes and nodes The sum of the weights of the edges between represents the exponential function with base e, is the risk attenuation coefficient, For nodes With node The shortest path length between For in time Time Node The risk value, The current time point; during the diffusion process, the risk status of the affected nodes is marked in real time, including risk level, multi-dimensional risk characteristics and risk value;

[0037] Based on the information recorded during the risk value diffusion process, a risk diffusion trajectory map with a timestamp is generated; the generated risk diffusion trajectory map is stored in IPFS and a corresponding CID is generated; the CID is written to the blockchain transaction log; at the same time, the trajectory map information is shared with each related node in the supply chain through the P2P network of the alliance chain.

[0038] Furthermore, based on the shared information of the risk diffusion trajectory map, the risk treatment strategy is automatically matched and executed, while the risk treatment effect is continuously evaluated and dynamically adjusted to optimize the risk management process. The process includes:

[0039] Obtain risk levels and risk feature types, and match predefined risk management rules from the on-chain policy library. Automatically generate a structured instruction package containing risk management measures, execution timelines, and responsible parties through the pre-set risk management strategy smart contract, and encrypt it with a digital signature. This instruction package will be broadcast to all relevant nodes, including core enterprise nodes, supplier nodes, and manufacturer nodes, through the consortium chain's P2P network.

[0040] After receiving the instruction package, each associated node will immediately execute the corresponding risk disposal measures and record the execution status through its own logs. The risk disposal strategy smart contract continuously monitors the logs of each associated node to verify whether the risk disposal measures are executed according to the instructions and collects the execution feedback of the risk disposal measures in real time. At the same time, it continuously captures the dynamic shared information in the risk diffusion trajectory diagram, calculates the multi-dimensional risk feature vector and compares it with the baseline before risk disposal, and quantitatively evaluates the risk reduction. The risk disposal efficiency dynamic evaluation formula is used to quantitatively evaluate the risk reduction, thereby judging the effectiveness of the risk disposal strategy:

[0041] ,

[0042] Where, The extent of risk reduction; is the integral operation of the time variable t; is the integration interval, representing the scope or time window of risk treatment; is the risk attenuation kernel function, which is used to quantify the weighted impact of risk level RH and risk feature type k on the risk treatment effect; is the strategy execution effectiveness factor; It is the risk transmission damping effect;

[0043] is the indicator function, when When , the indicator function value is 1, and the formula calculates the actual risk reduction , triggering the risk disposal mechanism; when When , the value of the exponential function is 0, the overall result of the formula is 0, and the risk treatment measure is suspension or downgrade; is the preset dynamic risk threshold.

[0044] Compared with the existing technology, the beneficial effects of the present invention are: by collecting multi-source dynamic data, integrating multi-dimensional feature vectors such as credit risk, operational risk, and legal risk, and storing them on the chain, the comprehensiveness and accuracy of risk data are ensured, a solid foundation is provided for subsequent risk assessment, and the accuracy of risk identification is improved; by deploying risk assessment and early warning smart contracts on the alliance chain, and using reinforcement learning algorithms to dynamically adjust the weights of each dimension, the intelligence and real-time nature of risk assessment are realized, the efficiency of risk assessment is improved, and the dynamic adaptability of the assessment results is enhanced; based on blockchain data, a supply chain topology network is constructed, the risk transmission process is simulated, and a risk diffusion trajectory map with a timestamp is generated, which facilitates the understanding of the risk diffusion path and timely measures to control risks, while realizing transparent sharing of risk information; based on the risk diffusion trajectory map, risk disposal strategies are automatically matched and executed, and the disposal effects are continuously evaluated, thereby realizing automation and closed-loop optimization of risk management, improving the efficiency and effectiveness of risk response, and reducing risk losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a flowchart of the blockchain-based supply chain partner risk management method proposed in this invention. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the implementation regulations described are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0047] Reference Figure 1 , a blockchain-based supply chain partner risk management method, including:

[0048] S1. Collect dynamic data from multiple sources, including transaction performance and compliance records; extract and integrate multi-dimensional risk feature vectors including credit risk, operational risk, and legal risk through data analysis methods, generate standardized risk metadata, and store them on-chain;

[0049] S2. Deploy a risk assessment and early warning smart contract on the consortium chain, extract the multi-dimensional risk feature vectors in the risk metadata certificate, and calculate a comprehensive risk score. Dynamically adjust the weights of each dimension through a reinforcement learning algorithm to achieve risk assessment and early warning functions.

[0050] S3. Obtain risk assessment and early warning results, build a supply chain topology network based on transaction data stored on the blockchain, simulate the risk transmission process through the risk value diffusion formula, and generate a time-stamped risk diffusion trajectory map for sharing;

[0051] S4. Based on the shared information of the risk diffusion trajectory map, automatically match and execute risk disposal strategies, while continuously evaluating the risk disposal effects and making dynamic adjustments to optimize risk management.

[0052] It should be further explained that, during the specific implementation process, the process of collecting dynamic data from multiple sources (including transaction performance and compliance records), extracting and integrating multi-dimensional risk feature vectors including credit risk, operational risk, and legal risk through data analysis methods, generating standardized risk metadata, and storing them on the chain is as follows:

[0053] Construct a three-dimensional risk matrix, including credit risk, operational risk, and legal risk dimensions: Calculate the credit risk characteristic vector by using the historical payment delay rate and order fulfillment rate: The calculation formula for the historical payment delay rate is: , where is the historical payment delay rate, is the total number of orders during the statistical period, is the order index within the statistical period, For the The actual payment date of the order, For the The payment date is agreed upon in the order contract; the calculation formula for the order fulfillment rate is: , where is the order fulfillment rate, To deliver the orders on time, is the total number of orders; integrating the historical payment delay rate and order fulfillment rate to obtain the credit risk characteristic vector ;

[0054] The operational risk characteristic vector is calculated based on the standard deviation of logistics timeliness and inventory turnover deviation: The calculation formula for the standard deviation of logistics timeliness is: , where is the standard deviation of logistics timeliness, For logistics index, is the number of logistics, For the Actual logistics timeliness, is the average value of logistics efficiency; the calculation formula for inventory turnover deviation is: , where is the inventory turnover deviation, is the average inventory turnover days, is the actual inventory turnover days; integrating the logistics efficiency standard deviation and inventory turnover deviation to obtain the operational risk characteristic vector ;

[0055] The legal risk feature vector is calculated based on the environmental protection penalty record and the number of days the license expires: , where is the environmental penalty rate, is the number of environmental penalties, is the total number of operating days; the formula for calculating the percentage of license expiration days is: , where is the percentage of license expiration days, is the total number of days the license is invalid; integrating the environmental penalty rate and the proportion of license invalid days to obtain the legal risk characteristic vector ;

[0056] The actual payment date, contractually agreed payment date, number of orders delivered on time, and total number of orders are obtained by connecting to the supplier's ERP system through an API gateway and extracting transaction data and related historical records. Actual logistics timeliness, average logistics timeliness, average inventory turnover days, and actual inventory turnover days are obtained through statistical analysis of logistics fulfillment data captured in real time by IoT devices (such as RFID and GPS) and inventory data extracted from the supplier's ERP or SCM system. Environmental protection penalty records, total operating days, and license expiration days are obtained by crawling and analyzing legal compliance data through access to a government regulatory platform.

[0057] Using the JSON-LD format, the multi-dimensional risk feature vector is serialized into a standardized risk metadata package; the serialized JSON-LD data is hashed to generate a unique data hash value; the generated JSON-LD data package is used to call the alliance chain's evidence contract; the evidence contract will record the complete JSON-LD data package and generate a digitally signed risk data evidence certificate, which contains a timestamp and data hash value to ensure the integrity and verifiability of the data; based on the risk data evidence certificate, the data hash value is used as input and written into the Merkle tree of the alliance chain; the root node of the Merkle tree will be recorded on the blockchain, providing a lightweight verification method for the data; based on the record of the Merkle tree, a risk metadata certificate with a timestamp is generated, which contains a data source signature and a Merkle path (Merkle root hash value, node hash value) to prove the integrity and authenticity of the data.

[0058] It should be further explained that in the specific implementation process, a risk assessment and early warning smart contract is deployed on the consortium chain, the multi-dimensional risk feature vector in the risk metadata certificate is extracted, and a comprehensive risk score is calculated; the weight of each dimension is dynamically adjusted through the reinforcement learning algorithm to realize the risk assessment and early warning function. The process is as follows:

[0059] The risk metadata certificate generated by the on-chain verification engine is called to extract the embedded JSON-LD format multi-dimensional risk feature vector (credit, operational, and legal risk features) and evidence credential information (data hash value, digital signature, timestamp) from the certificate. The Merkle path in the certificate is simultaneously parsed to extract the root node hash value (this root node hash value is stored on the chain), providing a benchmark anchor point for subsequent data authenticity verification.

[0060] The risk assessment and warning smart contract reads the stored multi-dimensional risk feature vector from the temporary storage area and parses it out. It uses the multi-dimensional risk feature vector of the current cycle (credit, operational, and legal risk features) and the historical risk event library stored on the chain as input states.

[0061] Normalize the risk feature vector for each dimension (e.g., map the historical payment delay rate from [0, 100%] to [0, 1]), perform a dot product calculation with the initial weights, and generate a comprehensive risk score (formula: comprehensive risk score = initial credit risk weight × credit risk feature value + initial operational risk weight × operational risk feature value + initial legal risk weight × legal risk feature value).

[0062] Introducing the supply chain risk reinforcement learning RL agent model, using the Q-learning algorithm to dynamically adjust the weight of each dimension: define the risk dimension set as , the initial weight of each dimension is ( is the risk dimension index), satisfying ;

[0063] Define the state space as , represents the matching degree between the current multi-dimensional risk feature vector and the historical event database, that is,

[0064] ,

[0065] in is the time step, which represents the number of iterations of risk assessment and early warning smart contract learning or the discrete time point of risk events; is the L2 norm;

[0066] Define the action space as , used to adjust the weight direction, that is, to increase or decrease the dimension weight:

[0067] ,

[0068] in, is a specific action in the action space, each action Corresponding to a specific weight adjustment direction, is the set of action spaces, ;

[0069] The risk assessment and warning smart contract dynamically updates weights using the following Q-learning rules:

[0070] ,

[0071] Where, is the updated weight; For instant rewards, based on the risk score change and the disposal cost: , is the weight factor of the comprehensive risk score change, is the weight factor of disposal cost, The change in the overall risk score ( ), is the weight-adjusted compliance cost ( , is the preset cost coefficient);

[0072] is the learning rate, which is used to control the weight distribution between new experience (immediate reward) and old knowledge (historical Q value). In supply chain risk management, It can be adjusted dynamically according to the volatility of the risk environment. For example, if the stability of the supply chain decreases (risk events increase), the To adapt to new risk patterns more quickly; if the environment is stable, it can be reduced Adjust with smooth weights;

[0073] is the discount risk factor, and , which is used to reflect the discount effect of long-term risk-return, and its value is positively correlated with the stability of the supply chain: , is the baseline discount rate, which is used to reflect the default long-term return decay rate; is the stability adjustment coefficient, which is used to control the amplifying effect of consecutive risk-free periods on the discount rate; is the time period threshold, which represents the unit of measurement for the consecutive risk-free period (e.g., month); for example, if the supply chain is risk-free for 12 consecutive months (number of consecutive risk-free periods = 12), then , indicating that future benefits are fully valued; if risks frequently occur in the supply chain in the near future (e.g., the number of consecutive risk-free cycles = 0), then , focusing more on immediate risk mitigation;

[0074] For the next state The candidate weight value is the new weight that the risk assessment and early warning smart contract may adopt after adjusting the action space A (increasing or decreasing the weight) after the state transition (such as risk event triggering or no event occurring);

[0075] The weight adjustment results are written into the blockchain through the risk assessment and early warning smart contract, forming a risk event-weight adjustment learning mechanism: the risk assessment and early warning smart contract calls the historical transaction behavior fingerprint stored on the chain (the historical transaction behavior fingerprint is generated by extracting the root node hash value corresponding to the Merkle path in the risk metadata certificate. These hash values ​​specifically include order receipt hash, payment timestamp hash, logistics track hash, etc., which serve as unique identifiers of key transaction behaviors and together constitute the core elements of the historical transaction behavior fingerprint) to recalculate the credit risk feature, operational risk feature, and legal risk feature vectors: , , ;

[0076] Compare the recalculated multidimensional risk feature vector with the original multidimensional risk feature vector generated by the three-dimensional risk matrix dimension by dimension:

[0077] ,

[0078] Where, is the risk deviation rate, is the original multidimensional risk feature vector, is the recalculated multidimensional risk feature vector, is the risk feature vector category;

[0079] Preset risk deviation thresholds β1 and β2; when the calculated risk deviation rate exceeds the risk deviation threshold, a graded warning is automatically triggered, i.e., high risk ≥ β1, medium risk [β2, β1), and low risk < β2: When identified as a high-risk level, i.e., a risk deviation rate ≥ β1, a structured warning event package containing the risk level and risk feature category is generated and broadcast to all related nodes through the P2P network of the alliance chain, including core enterprise nodes, supplier nodes, and manufacturer nodes. At the same time, the warning timestamp, triggering node ID, and event hash value are recorded on the chain; when identified as a medium-risk level, i.e., the risk deviation rate belongs to [β2, β1), an early warning notification (including risk level, specific risk feature category, and recommended measures, such as a recommendation to increase the sampling ratio by 10%) is pushed to the risk control department of the core enterprise in the supply chain, and the node is marked as an observation object on the chain, restricting its priority cooperation rights (such as suspending the allocation of new orders); when identified as a low-risk level, i.e., a risk deviation rate < β2, no warning is triggered;

[0080] The risk deviation rate and warning event information are integrated and packaged into a new on-chain block through the risk assessment and warning smart contract. The block contains the hash value of the previous block, the version number of the RL agent's weight adjustment strategy, the list of receiving node IDs of the warning event, and the delivery timestamp, ensuring the traceability and immutability of the scoring process.

[0081] It should be further explained that, in the specific implementation process, the process of obtaining risk assessment and early warning results, building a supply chain topology network based on transaction data stored on the blockchain, simulating the risk transmission process through the risk value diffusion formula, and generating a risk diffusion trajectory map with a timestamp for sharing is as follows:

[0082] Each associated node in the supply chain receives the broadcasted structured warning event package through the P2P network of the alliance chain. The associated nodes include core enterprise nodes, supplier nodes, and manufacturer nodes. Each associated node parses the received warning event package, extracts key information, and stores it in the local database for subsequent risk transmission path tracking.

[0083] Utilize the historical transaction records stored on the blockchain, i.e. transaction amount and transaction frequency; construct a supply chain topology network based on transaction dependencies: use identity DID identifiers as nodes of the supply chain topology network (i.e. core enterprise nodes, supplier nodes, manufacturer nodes); use transaction amount and transaction frequency weighted values ​​as edges of the supply chain topology network; locate the warning nodes triggered by warning events, i.e. risk source nodes (e.g. supplier nodes) in the constructed supply chain topology network; set the initial risk value of the risk source node to ,The weight set of edges in the supply chain topology network is ,in Representation node With node The transaction amount and transaction frequency weighted value between them; along the edges in the supply chain topology network, the risk value is attenuated and diffused according to the edge weight (transaction amount, transaction frequency weighted value). The risk value diffusion formula is:

[0084] ,

[0085] Where, For in time Time Node The risk value, For nodes The set of neighbor nodes of For nodes All neighbor nodes and nodes The sum of the weights of the edges between For nodes a neighbor node of For nodes With node The weight of the edge between represents the exponential function with base e, is the risk attenuation coefficient, For nodes With node The shortest path length between For in time Time Node The risk value, The current time point; the larger the edge weight, the more frequent the transactions and closer the cooperation between the two nodes, and the smaller the attenuation of the risk value during the diffusion process, that is, the greater the possibility of risk transmission; during the diffusion process, the risk status of the affected nodes (such as supplier nodes and manufacturer nodes) is marked in real time, including risk level (high risk, medium risk and low risk), multi-dimensional risk characteristics (credit risk, operational risk and legal risk) and risk value size;

[0086] Based on the information recorded during the risk value diffusion process, a timestamp risk diffusion trajectory diagram is generated; this trajectory diagram is used to show the path and process of risk transmission from the early warning node to other nodes along the supply chain topology network, as well as the risk status of each node at different time points; the generated risk diffusion trajectory diagram is stored in IPFS (InterPlanetary File System) and a corresponding CID (Content Identifier) ​​is generated; the CID is written to the blockchain transaction log to ensure the integrity and non-tamperability of the trajectory diagram; at the same time, the trajectory diagram information is shared with each related node in the supply chain through the P2P network of the alliance chain.

[0087] It should be further explained that, in the specific implementation process, based on the shared information of the risk diffusion trajectory map, the risk treatment strategy is automatically matched and executed, while the risk treatment effect is continuously evaluated and dynamically adjusted to optimize the risk management process.

[0088] Obtain risk levels (high, medium, low) and risk characteristics (credit, operational, and legal risk characteristics), and match predefined risk management rules from the on-chain policy library (e.g., high-risk suppliers must implement a combination of order suspension and full margin). The risk management strategy smart contract automatically generates a structured instruction package containing risk management measures (e.g., order volume reduction ratio, inspection frequency increase), execution time (immediately or within 72 hours), and responsible party (core enterprise or third-party quality inspection agency), and encrypts it with a digital signature. The instruction package will be broadcast to all relevant nodes, including core enterprise nodes, supplier nodes, and manufacturer nodes, via the consortium chain's P2P network.

[0089] After receiving the instruction package, each associated node will immediately execute the corresponding risk disposal measures and record the execution status through its own logs. The risk disposal strategy smart contract continuously monitors the logs of each associated node to verify whether the risk disposal measures are executed according to the instructions and collects the execution feedback of the risk disposal measures in real time. At the same time, it continuously captures the dynamic shared information in the risk diffusion trajectory diagram, calculates the multi-dimensional risk feature vector and compares it with the baseline before risk disposal, and quantitatively evaluates the risk reduction. The risk disposal efficiency dynamic evaluation formula is used to quantitatively evaluate the risk reduction, thereby judging the effectiveness of the risk disposal strategy:

[0090] ,

[0091] Where, The extent of risk reduction; is the integral operation of the time variable t; is the integration interval, representing the scope or time window of risk treatment;

[0092] is the risk attenuation kernel function, which is used to quantify the weighted impact of risk level RH and risk feature type k on risk treatment effect, where , 、 is the preset weight coefficient, is the risk level sensitivity coefficient, is the risk feature type impact function;

[0093] is the strategy execution effectiveness factor, which is used to combine the strategy matching degree sm and the execution quality fq, and punish the deviation between strategy and execution. ,in 、 is the preset strategy matching degree sm and execution quality fq weight coefficient, 、 is the nonlinear adjustment index, is the policy-execution deviation penalty coefficient;

[0094] The risk transmission damping effect is used to simulate the risk transmission inhibition in the supply chain network through the node risk change rate and damping coefficient. , is the risk transmission rate constant, is the initial risk value, is the risk damping coefficient of the supply chain topology network node, is the supply chain topology network node index, is the number of supply chain topology network nodes, is the node risk change rate;

[0095] is the indicator function, when When , the indicator function value is 1, and the formula calculates the actual risk reduction , triggering risk management mechanisms (such as upgrading the warning level, expanding the scope of strategy implementation); when When , the value of the exponential function is 0, the overall result of the formula is 0, and the risk treatment measures may be suspended or downgraded; It is a preset dynamic risk threshold; if the risk treatment measures fail to effectively reduce the risk or the risk escalates further, a higher level of warning will be automatically triggered and the risk treatment strategy will be re-matched.

[0096] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0097] It should be understood that determining B based on A does not mean determining B based solely on A. B can also be determined based on A and / or other information.

[0098] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0099] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A blockchain-based supply chain partner risk management method, characterized by: S1. Collect dynamic data from multiple sources, including transaction performance and compliance records; extract and integrate multi-dimensional risk feature vectors including credit risk, operational risk, and legal risk through data analysis methods, generate standardized risk metadata, and store them on-chain; S2. Deploy a risk assessment and early warning smart contract on the consortium chain, extract the multi-dimensional risk feature vector in the risk metadata certificate, and calculate the comprehensive risk score; Dynamically adjust the weights of each dimension through reinforcement learning algorithms to achieve risk assessment and early warning functions; S3. Obtain risk assessment and early warning results, build a supply chain topology network based on transaction data stored on the blockchain, simulate the risk transmission process through the risk value diffusion formula, and generate a time-stamped risk diffusion trajectory map for sharing; The process of obtaining risk assessment and early warning results, building a supply chain topology network based on transaction data stored on the blockchain, simulating the risk transmission process through the risk value diffusion formula, and generating a risk diffusion trajectory map with a timestamp for sharing includes: Each associated node in the supply chain receives the broadcasted structured warning event package through the P2P network of the alliance chain. The associated nodes include core enterprise nodes, supplier nodes, and manufacturer nodes. Each associated node parses the received warning event package, extracts key information, and stores it in the local database. Utilize the historical transaction records stored on the blockchain, i.e. transaction amount and transaction frequency; construct a supply chain topology network based on transaction dependencies: use identity DID identifiers as nodes of the supply chain topology network, i.e. core enterprise nodes, supplier nodes, and manufacturer nodes; use transaction amount and transaction frequency weighted values ​​as edges of the supply chain topology network; locate the warning node triggered by the warning event, i.e. the risk source node, in the constructed supply chain topology network; set the initial risk value of the risk source node to ,The weight set of edges in the supply chain topology network is ,in Representation node With node The transaction amount and transaction frequency weighted value between nodes With node The weight of the edges between them; along the edges in the supply chain topology network, the risk value is diffused according to the edge weight attenuation. During the diffusion process, the risk status of the affected nodes is marked in real time, including risk level, multi-dimensional risk characteristics and risk value size; Based on the information recorded during the risk value diffusion process, a timestamped risk diffusion trajectory map is generated. The generated risk diffusion trajectory map is stored in IPFS and a corresponding CID is generated. The CID is written into the blockchain transaction log. At the same time, the trajectory map information is shared with all related nodes in the supply chain through the consortium chain's P2P network. S4. Based on the shared information of the risk diffusion trajectory map, automatically match and execute risk disposal strategies, while continuously evaluating the risk disposal effects and making dynamic adjustments to optimize risk management.

2. The blockchain-based supply chain partner risk management method according to claim 1, characterized in that: The process of extracting and integrating multi-dimensional risk feature vectors including credit risk, operational risk, and legal risk through data analysis methods, generating standardized risk metadata, and storing them on-chain includes: Construct a three-dimensional risk matrix, including credit risk, operational risk, and legal risk dimensions: Calculate the credit risk characteristic vector by using the historical payment delay rate and order fulfillment rate: The calculation formula for the historical payment delay rate is: , where is the historical payment delay rate, is the total number of orders during the statistical period, is the order index within the statistical period, For the The actual payment date of the order, For the The payment date is agreed upon in the order contract; the calculation formula for the order fulfillment rate is: , where is the order fulfillment rate, To deliver the orders on time, is the total number of orders; integrating the historical payment delay rate and order fulfillment rate to obtain the credit risk characteristic vector ; The operational risk characteristic vector is calculated based on the standard deviation of logistics timeliness and inventory turnover deviation: The calculation formula for the standard deviation of logistics timeliness is: , where is the standard deviation of logistics timeliness, For logistics index, is the number of logistics, For the Actual logistics timeliness, is the average value of logistics efficiency; the calculation formula for inventory turnover deviation is: , where is the inventory turnover deviation, is the average inventory turnover days, is the actual inventory turnover days; integrating the logistics efficiency standard deviation and inventory turnover deviation to obtain the operational risk characteristic vector ; The legal risk feature vector is calculated based on the environmental protection penalty record and the number of days the license expires: , where is the environmental penalty rate, is the number of environmental penalties, is the total number of operating days; the formula for calculating the percentage of license expiration days is: , where is the percentage of license expiration days, is the total number of days the license is invalid; integrating the environmental penalty rate and the proportion of license invalid days to obtain the legal risk characteristic vector ; Using the JSON-LD format, the multi-dimensional risk feature vector is serialized into a standardized risk metadata package; the serialized JSON-LD data is hashed to generate a unique data hash value; the generated JSON-LD data package is used to call the alliance chain's evidence contract; the evidence contract will record the complete JSON-LD data package and generate a digitally signed risk data evidence certificate, which contains a timestamp and data hash value; based on the risk data evidence certificate, the data hash value is used as input and written into the alliance chain's Merkle tree; the root node of the Merkle tree will be recorded on the blockchain; based on the record of the Merkle tree, a risk metadata certificate with a timestamp is generated, which contains a data source signature and a Merkle path.

3. The blockchain-based supply chain partner risk management method according to claim 1, characterized in that: The process of extracting the multi-dimensional risk feature vector from the risk metadata certificate and calculating the comprehensive risk score includes: The risk metadata certificate generated by the on-chain verification engine is called to extract the embedded JSON-LD format multi-dimensional risk feature vector and evidence credential information from the certificate. The Merkle path in the certificate is simultaneously parsed to extract the root node hash value and use it as the reference anchor point. The risk assessment and warning smart contract reads the stored multi-dimensional risk feature vector from the temporary storage area and parses it out. It uses the multi-dimensional risk feature vector of the current cycle and the historical risk event library stored on the chain as input states. The risk feature vector of each dimension is normalized and the dot product is calculated with the initial weight to generate a comprehensive risk score.

4. The blockchain-based supply chain partner risk management method according to claim 2, characterized in that: The process of dynamically adjusting the weights of each dimension through reinforcement learning algorithms to achieve risk assessment and early warning functions includes: Introducing the supply chain risk reinforcement learning RL agent model, using the Q-learning algorithm to dynamically adjust the weight of each dimension: define the risk dimension set as , the initial weight of each dimension is ,satisfy ; Define the state space as , represents the matching degree between the current multi-dimensional risk feature vector and the historical event database, that is, ,in is the time step, which represents the number of iterations of risk assessment and early warning smart contract learning or the discrete time point of risk events; is the L2 norm; Define the action space as , used to adjust the weight direction, that is, to increase or decrease the dimension weight: ,in ; The risk assessment and warning smart contract dynamically updates weights using the following Q-learning rules: ， Where, is the updated weight; It is an immediate reward, which is obtained based on the risk score change and the disposal cost; is the learning rate; is the discount risk factor, For the next state Candidate weight values ​​of ; The weight adjustment results are written into the blockchain through the risk assessment and early warning smart contract, forming a risk event-weight adjustment learning mechanism: the risk assessment and early warning smart contract calls the historical transaction behavior fingerprints stored on the chain to recalculate the credit risk characteristics, operational risk characteristics, and legal risk characteristic vectors: , , ; Compare the recalculated multidimensional risk feature vector with the original multidimensional risk feature vector generated by the three-dimensional risk matrix dimension by dimension: ， Where, is the risk deviation rate, is the original multidimensional risk feature vector, is the recalculated multidimensional risk feature vector, is the risk feature vector category; Preset risk deviation thresholds β1 and β2; when the calculated risk deviation rate exceeds the risk deviation threshold, a graded warning is automatically triggered, i.e. high risk ≥ β1, medium risk [β2, β1), low risk < β2: when it is identified as a high-risk level, i.e. the risk deviation rate ≥ β1, a structured warning event package containing the risk level and risk feature category is generated and broadcast to all related nodes through the P2P network of the alliance chain, including core enterprise nodes, supplier nodes and manufacturer nodes, and the warning timestamp, triggering node ID and event hash value are recorded on the chain; when it is identified as a medium-risk level, i.e. the risk deviation rate belongs to [β2, β1), a warning notification is pushed to the risk control department of the core enterprise in the supply chain, and the node is marked as an observation object on the chain to limit its priority cooperation rights; when it is identified as a low-risk level, i.e. the risk deviation rate < β2, no warning is triggered; Integrate risk deviation rate and warning event information, and generate new on-chain blocks through risk assessment and warning smart contracts.

5. The blockchain-based supply chain partner risk management method according to claim 1, characterized in that: The risk value diffusion formula is: ， Where, For in time Time Node The risk value, For nodes The set of neighbor nodes of For nodes All neighbor nodes and nodes The sum of the weights of the edges between represents the exponential function with base e, is the risk attenuation coefficient, For nodes With node The shortest path length between For in time Time Node The risk value, The current time point.

6. The blockchain-based supply chain partner risk management method according to claim 1, characterized in that: Based on the shared information of the risk diffusion trajectory map, the risk treatment strategy is automatically matched and executed, while the risk treatment effect is continuously evaluated and dynamically adjusted to optimize the risk management process. The process includes: Obtain risk levels and risk feature types, and match predefined risk management rules from the on-chain policy library. Automatically generate a structured instruction package containing risk management measures, execution timelines, and responsible parties through the pre-set risk management strategy smart contract, and encrypt it with a digital signature. This instruction package will be broadcast to all relevant nodes, including core enterprise nodes, supplier nodes, and manufacturer nodes, through the consortium chain's P2P network. After receiving the instruction package, each associated node will immediately execute the corresponding risk disposal measures and record the execution status through its own logs. The risk disposal strategy smart contract continuously monitors the logs of each associated node to verify whether the risk disposal measures are executed according to the instructions and collects the execution feedback of the risk disposal measures in real time. At the same time, it continuously captures the dynamic shared information in the risk diffusion trajectory diagram, calculates the multi-dimensional risk feature vector and compares it with the baseline before risk disposal, and quantitatively evaluates the risk reduction. The risk disposal efficiency dynamic evaluation formula is used to quantitatively evaluate the risk reduction, thereby judging the effectiveness of the risk disposal strategy: ， Where, The extent of risk reduction; is the integral operation of the time variable t; is the integration interval, representing the scope or time window of risk treatment; is the risk attenuation kernel function, which is used to quantify the weighted impact of risk level RH and risk feature type k on the risk treatment effect; is the strategy execution effectiveness factor; It is the risk transmission damping effect; is the indicator function, when When , the indicator function value is 1, and the formula calculates the actual risk reduction , triggering the risk disposal mechanism; when When , the value of the exponential function is 0, the overall result of the formula is 0, and the risk treatment measure is suspension or downgrade; is the preset dynamic risk threshold.

Images