Supply chain risk monitoring method, system, storage medium and computer device

Abstract

The application discloses a supply chain risk monitoring method and system, a storage medium and a computer device, and belongs to the supply chain field.The method comprises the following steps: constructing a standardized data set, setting a real-time response method, establishing a dynamic early warning rule, generating a supply-demand adjustment strategy, simulating the influence of the strategy on production, updating a supply-demand strategy plan library, and providing risk simulation and plan support for supply-demand decision-making; constructing a multi-dimensional risk assessment index, setting a risk response method, calculating a risk score, and configuring a differentiated response strategy to realize quantitative evaluation and dynamic response of supply chain risks; real-time decomposition of personalized demand into material demand, setting a driving matching method, constructing a three-dimensional priority evaluation model, dynamically adjusting material pre-occupation strategies and subsidy mechanisms, synchronously formulating phased inventory preparation and VMI inventory coordination strategies, and improving the conversion efficiency of personalized demand; obtaining supply-demand execution effects and customer satisfaction, setting an optimization method, and real-time updating of a risk quantification model and demand analysis rules.

Description

Technical Field

[0001] This invention relates to supply chain risk monitoring methods, systems, storage media, and computer equipment, and belongs to the field of supply chain technology. Background Technology

[0002] In an industrial environment deeply integrated with globalization and digitalization, supply chain procurement faces core pain points such as low collaboration efficiency, delayed supply and demand data transmission, and single-dimensional cost control due to the collaboration of multiple entities including suppliers, manufacturers, warehouses, and customers. Traditional procurement models rely on manual experience to formulate plans, resulting in fragmented processes and significant information silos, which can easily lead to problems such as inventory turnover imbalance, order response delays, and lack of dynamic supplier management, making it difficult to cope with the flexible market demand for multiple varieties and small batches under the consumption upgrade.

[0003] A Chinese patent application with publication number CN116187933A discloses a material procurement planning and supply chain management system, including a market analysis module, a customer module, a warehouse module, a factory module, and a supplier module. This system tracks production progress at key production nodes via a secure network, issuing red alerts for production nearing completion and those nearing delays. This ensures timely integration of new orders, maintains a continuous order flow, keeps the factory running, avoids order delays, and builds a trustworthy corporate reputation. Furthermore, by querying purchase orders within the procurement planning unit through the secure network, purchase orders are prioritized to prevent the signing of new urgent orders, which could lead to frequent changes in the production plan and result in orders not being completed on time, damaging the reputation. This enhances order security and allows for a comprehensive consideration of capacity and production schedule matching when signing contracts with customers, achieving stable and orderly contracted production and avoiding resource waste.

[0004] While existing technologies have enabled production progress tracking, order alerts, and procurement order sorting through commercial encryption networks, thus improving supply chain stability and order security to some extent, they have not addressed the lack of a data-driven automated decision-making loop across the entire supply chain, multi-dimensional dynamic risk assessment, and a deep collaboration mechanism for customer needs. In particular, production progress alerts, such as order timeouts and inventory criticalities, rely solely on manual intervention through alert prompts, without being integrated into an intelligent decision-making system to automatically trigger adjustments to procurement strategies. Furthermore, there is a lack of quantitative assessment models and tiered response mechanisms for risks such as supplier disruptions, transportation delays, and raw material shortages. Additionally, the personalized needs of customer modules are not mapped in real-time to bills of materials, inventory status, and production capacity data. Moreover, differentiated procurement strategies are not developed based on the demand fluctuations of specific and general customers, taking into account attributes such as material turnover rate and shelf life, such as prioritizing material reserves for specific customers and phased stocking of seasonal materials. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide a supply chain risk monitoring method, system, storage medium, and computer equipment. By constructing a data-driven automated decision-making closed loop across the entire chain, it enables automatic triggering of production progress warnings and inventory preparation strategies, establishes a multi-dimensional risk quantification assessment system, configures tiered response strategies, maps personalized customer needs with materials and inventory in real time, and formulates differentiated inventory preparation strategies based on customer and material attributes.

[0006] To achieve the above objectives, the present invention provides the following technical solution: Supply chain risk monitoring methods include: Construct a standardized dataset, set up a real-time response method, establish dynamic early warning rules, generate supply and demand adjustment strategies, simulate the impact of the supply and demand adjustment strategies on production, and dynamically update the supply and demand strategy contingency plan library. Construct multi-dimensional risk assessment indicators, set risk response methods, calculate risk scores, and configure differentiated response strategies; Personalized needs are broken down into material requirements in real time, driving matching methods are set, a three-dimensional priority evaluation model is constructed, material pre-positioning strategies and subsidy mechanisms are dynamically adjusted, and phased inventory preparation and VMI inventory coordination strategies are formulated simultaneously. Obtain the performance of supply and demand and customer satisfaction, set optimization methods, and update the risk quantification model and demand analysis rules in real time.

[0007] Specifically, the real-time response method includes: Calculate the urgency of material consumption and the risk of supplier delivery, and establish multi-level early warning rules to generate material early warning work orders; Construct a demand forecasting model, train the demand forecasting model based on historical data, and output the predicted material demand. Construct a comprehensive supplier rating system, calculate the comprehensive score for each supplier, and select the top-ranked suppliers based on their comprehensive scores. One supplier was selected as a candidate supplier; Calculate delivery risks and inventory costs, determine the objective function, and set constraints to generate a multi-objective optimization model; Using the order allocation ratio as the decision variable, the CPLEX solver is called to solve the problem. Based on the predicted material demand and relevant information of candidate suppliers, a supply and demand adjustment strategy is generated.

[0008] Specifically, the real-time response method further includes: The supply and demand adjustment strategy is sent through the interface and a smart contract is triggered to generate a supplier order confirmation receipt. Obtain the supplier's actual delivery cycle, calculate the new scheduling time, and when the new scheduling time exceeds the original planned time, scan the idle production lines, split the tasks to the idle production lines according to the order priority, and generate a production schedule. Obtain material requirements for high-priority orders, calculate the pre-allocation quantity of each material, generate an inventory pre-allocation ledger, and update the available inventory in real time after pre-allocation. Construct a three-level simulation model, including the workshop level, inventory level, and supplier level, to simulate the effects of supply and demand adjustment strategies and output a simulation effect report; Set a reward function, use the Q-learning algorithm, and update the parameters of the multi-objective optimization model based on the simulation effect report and actual execution data.

[0009] Specifically, the risk response method includes: Construct a three-tiered risk factor database covering suppliers, logistics and transportation, and materials and inventory; Collect historical risk event data and label the corresponding risk levels. At the same time, extract risk factor data before the event occurs to construct a training dataset. Using the XGBoost algorithm, with standardized indicators from the three-level risk factor library as input, a risk quantification model is established, and the probability of each risk event is output. Based on real-time data, the risk quantification model is invoked to generate a real-time risk scoring table; Construct a mapping table between risk levels and response strategies, set specific countermeasures for different risk types and levels, and generate risk response instructions.

[0010] Specifically, the risk response method further includes: The supply and demand adjustment strategy is optimized according to the recommended strategy in the risk response instruction, and the inventory pre-occupancy ledger is updated synchronously to generate a material priority adjustment notification. Relevant information for each risk response operation is automatically recorded on the blockchain to generate a risk event log; Based on the pre-set periodic summary of risk event logs, the risk quantification model is retrained and the risk factor weights are adjusted. Based on digital twins, extreme scenarios are simulated to simulate the risk transmission effect and generate risk contingency plans.

[0011] Specifically, the driver matching method includes: Get personalized requirements and generate unique numbers, then clean them and convert them into work orders; NLP is used to identify key product parameters and convert them into standard material codes. Combined with a material database, a customized bill of materials is generated, including standard materials, special materials and alternative materials. Obtain standard material inventory and historical stock preparation and delivery cycle of special materials to make a feasibility assessment. When there is no inventory and the stock preparation and delivery cycle is overdue, trigger the alternative solution recommendation engine and generate a demand feasibility assessment report. Extract historical transaction data to calculate value scores and generate a value stratification table; Calculate the urgency coefficient and capacity feasibility score, and generate an order priority queue.

[0012] Specifically, the driver matching method further includes: Identify specific customer orders, extract the required quantity and obtain the pre-allocation quantity for the order, set safety stock, and trigger the reserve call mechanism when the pre-allocation quantity exceeds the limit; Read the candidate suppliers, select the specific supplier, and generate a specific customer order processing form; Identify core customer orders, calculate inventory levels using the EOQ model, set adjustment ranges based on historical order fluctuations, select and dynamically adjust primary and alternative suppliers according to on-time delivery rates, and generate core customer inventory execution orders. Identify regular customer orders, summarize material orders weekly, prepare goods in a unified manner after the minimum order quantity is reached, set payment terms, and generate a consolidated preparation order for regular customers. Predict peak demand for seasonally scarce materials and develop phased inventory strategies, generate seasonal material inventory plans, monitor material shelf life and trigger processing procedures. For high-frequency, general-purpose materials, VMI mode is used to set replenishment thresholds and generate replenishment instructions. During the supplier certification period, acquire suppliers with similar technologies, include them in the supplier pool after certification, and update the pool regularly. The system acquires information on demand changes, retrieves data, and uses digital twin technology to simulate and output material shortages, inventory turnover rate changes, and peak production line load rates, thereby generating the optimal adjustment plan.

[0013] Specifically, the optimization method includes: Define core performance indicators for material efficiency, risk control, and customer value; establish a correlation matrix between decision-making strategies and performance indicators; and identify outliers. Adjust the parameters and weight configurations of the multi-objective optimization model and risk quantification model, conduct retrospective analysis and update inventory preparation, risk response, and customer-tiered inventory preparation strategies, encrypt and store key decision data, and construct a knowledge graph.

[0014] The supply chain risk monitoring system includes: a response module, a risk assessment module, a matching module, and a feedback module; The response module is used to collect data from smart workshops, warehouses and suppliers in real time, clean and integrate the data to generate a standardized dataset, establish dynamic early warning rules, generate supply and demand adjustment strategies through multi-objective optimization and trigger their execution, simulate supply chain scenarios based on historical data, optimize and dynamically update the supply and demand strategy contingency plan library through reinforcement learning, and output decision parameters. The risk assessment module is used to construct multi-dimensional risk assessment indicators, calculate risk scores using machine learning and real-time data, and configure differentiated response strategies. The matching module is used to break down customers’ personalized needs into material requirements, connect to the material database to match inventory and preparation delivery cycles, build a three-dimensional priority evaluation model, dynamically adjust the special customer expedited order strategy, formulate a phased preparation and VMI inventory coordination strategy, and realize hierarchical management and control of customers, demand, and supply and demand. The feedback module is used to obtain the performance of supply and demand and customer satisfaction, update the risk quantification model and demand analysis rules, optimize algorithm parameters through transfer learning, and select best practice cases.

[0015] A medium storing a computer program that, when executed by a processor, implements the aforementioned supply chain risk monitoring method.

[0016] A computer device, comprising: Memory, used to store instructions; A processor is configured to execute the instructions, causing the device to perform operations that implement the supply chain risk monitoring method described above.

[0017] The beneficial effects of this invention are: By implementing a data-driven, automated decision-making loop across the entire supply chain, the system addresses the issue of traditional systems relying solely on manual intervention for production progress warnings. It achieves a closed-loop linkage from data collection and dynamic early warning to the automatic generation and execution of inventory preparation strategies, eliminating delays caused by manual intervention and improving decision-making efficiency and scientific rigor. Through a multi-dimensional risk assessment system covering suppliers, logistics, and materials, machine learning quantifies risk levels and configures differentiated response strategies, compensating for the lack of systematic assessment and response to risks such as supplier disruptions and transportation delays. For customer needs, intelligent analysis, hierarchical control, and flexible inventory preparation strategies map personalized requirements to bill of materials, inventory, and capacity data in real time. Differentiated strategies are developed based on customer value and material attributes, resolving the issues of crude hierarchical demand control and insufficient coordination. This enables the supply chain to transform from early warning observation to proactive intervention, and from production-driven sales to demand-driven procurement, comprehensively enhancing the supply chain's agility, resilience, and customer demand responsiveness. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of supply chain risk monitoring methods. Figure 2 This is a flowchart of the real-time response method of the present invention; Figure 3 This is a flowchart of the driving matching method of the present invention; Figure 4 This is a structural diagram of a supply chain risk monitoring system. Detailed Implementation

[0019] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0020] Example 1 refer to Figures 1 to 3 As shown in the figure, this embodiment introduces a supply chain risk monitoring method, including the following steps: Real-time capture of production progress data from smart workshops, material consumption data from warehouses, and delivery data from suppliers; data cleaning and integration to generate standardized datasets; real-time response methods; dynamic early warning rules; automatic generation of supply and demand adjustment strategies through multi-objective optimization; synchronous triggering of execution; achieving automated real-time response from production data to procurement decisions; eliminating delays caused by manual intervention; improving decision-making efficiency; based on historical production and procurement data, simulating a three-tier supply chain scenario involving workshops, inventory, and suppliers; pre-simulating the impact of various supply and demand adjustment strategies on production; through reinforcement learning optimization; dynamically updating the supply and demand strategy contingency plan library based on simulation results and actual execution effects; outputting optimized decision parameters; providing risk simulation and contingency plan support for procurement decisions; enhancing the scientific and forward-looking nature of decision-making. Construct multi-dimensional risk assessment indicators covering suppliers, logistics, and materials; set risk response methods; use machine learning and real-time data to calculate risk scores to quantify risk levels; and configure differentiated response strategies to achieve quantitative assessment and dynamic response to supply chain risks, thereby enhancing risk resilience. We break down customers' personalized needs into material requirements in real time, set up driving matching methods, connect to the global material database to intelligently match inventory and preparation delivery cycles, and build a three-dimensional priority evaluation model by combining customer value, order urgency, and capacity load. We dynamically adjust the material pre-positioning strategy and supplier subsidy mechanism for special customers' urgent orders, and formulate phased preparation and VMI inventory coordination strategies for seasonal or general materials. This enables refined management of customer segmentation, demand segmentation, and procurement segmentation, improves the on-time delivery rate of special customer orders, and increases the efficiency of personalized demand conversion. To obtain the execution results of supply and demand and customer satisfaction, optimization methods are set, and risk quantification models and demand parsing rules are updated in real time. Algorithm parameters are dynamically optimized through transfer learning, and best practice cases are selected to enable the system to continuously evolve with the accumulation of business data. Among them, the demand parsing rules rely on the HanLP natural language processing tool to complete the semantic parsing of personalized customer needs. Combined with the parameter mapping relationship of the Neo4j material knowledge graph, the natural language needs are transformed into standardized material requirement parameters. At the same time, through the TensorFlow transfer learning module, the semantic matching weight of the demand parsing rules is dynamically optimized based on new procurement scenario data to continuously improve the accuracy of demand parsing.

[0021] Specifically, real-time response methods include: Extracting near Daily material consumption data is used to calculate the average daily consumption rate. Safety stock and average daily consumption are obtained from standardized datasets. The safety stock turnover rate is calculated by the ratio between safety stock and average daily consumption. The material consumption urgency index (CSI) is calculated by the ratio between the average daily consumption rate and the safety stock turnover rate to measure the urgency of inventory consumption. By obtaining historical delayed delivery rates and current supplier capacity utilization rates, the supplier delivery risk index (DRI) is calculated, transforming complex business risks into quantifiable metrics; the expression is shown below:

[0022] In the formula, This represents the current capacity utilization rate, which is the percentage of total capacity currently allocated to suppliers, and is reported by suppliers in real time. Historical delayed delivery rate, which is the percentage of delayed orders from the corresponding supplier in the past 6 months out of the total number of orders; Risk classification management is achieved through tiered early warning systems, ensuring that resources are prioritized for high-urgency scenarios. Multi-level early warning rules are established based on CSI and DRI, and corresponding material early warning work orders are generated according to the warning level, indicating material information, warning level, and suggested processing time. Among these multi-level early warning rules, urgency thresholds are set (including...). , ), risk thresholds (including , ),and , Extract the CSI values ​​for all stockout events 7 days prior to their occurrence, calculate the probability distribution, and take the 75th quantile as the probability distribution. 95th percentile as Extract the DRI values ​​for the 3 days preceding all delivery delay events, calculate the probability distribution, and take the 70th quantile as the probability distribution. 90th percentile as ,when and At this point, the warning level is low-risk, requiring no manual intervention. The system is automatically monitored, generating daily inventory and delivery status reports, and executing procurement plans according to the standard stock preparation and delivery cycle. or At this time, the warning level is a medium-risk warning. and At this time, the warning level is high-risk. A demand forecasting model is built using Long Short-Term Memory (LSTM) networks. Based on historical material consumption data, the model is trained to predict future demand. Daily material requirements; A comprehensive supplier evaluation system is constructed, taking into account the on-time delivery rate, price competitiveness, and quality pass rate of different suppliers. A weighted summation method is used to calculate the comprehensive score of each supplier, and the top-ranked suppliers are selected. The system selects suppliers as candidates, providing reliable demand forecasts and a range of high-quality suppliers for procurement decisions. The weights for on-time delivery rate, price competitiveness, and quality pass rate are determined based on the analytic hierarchy process (AHP). The value range is 3-5, with a maximum of 5 for special materials and a minimum of 3 for general materials, to ensure diversity and competitiveness in supplier selection; Delivery risk is obtained by summing the products of each supplier's delivery cycle and order ratio, and inventory preparation cost is obtained by summing the products of expedited purchase premium and order quantity. Taking into account both delivery risk and inventory preparation cost, an objective function is determined and constraints are set, such as the total order quantity being the predicted demand multiplied by a safety factor, where the safety factor is determined according to the warning level. For example, the safety factor is 1.1 for low-risk warnings, 1.3 for medium-risk warnings, and 1.5 for high-risk warnings. The safety factor under different warning levels is determined by historical stockout rate statistics to ensure that demand fluctuations and delivery delays are covered at a 95% confidence level, and the order ratio of a single supplier does not exceed 70%. A balance is sought between cost and delivery risk to ensure the rationality of procurement decisions, thereby generating parameters for a multi-objective optimization model. The mixed integer programming algorithm is used to solve the multi-objective optimization model. The order allocation ratio is used as the decision variable and the CPLEX solver is called to solve the model. Based on the predicted material demand and relevant information of candidate suppliers, the order allocation ratio and transportation method of each supplier are determined, and the supply and demand adjustment strategy is generated, which includes supplier allocation ratio, order quantity, delivery cycle, transportation method and expedited premium amount. The supply and demand adjustment strategy is sent to suppliers via the application programming interface (API). A regular order is sent to the main supplier with a 3-day grace period. If no confirmation is received from the main supplier within 72 hours, an expedited order is automatically sent to the alternative supplier. At the same time, the unconfirmed order from the main supplier is cancelled to avoid duplicate purchases. The smart contract is triggered, which requires confirmation of production scheduling within 72 hours. Otherwise, a 1% penalty is automatically deducted. This enables the rapid and accurate issuance of procurement instructions. Finally, a supplier order confirmation receipt is generated, which includes the acceptance or rejection status, for use in production scheduling adjustments. The actual delivery cycle of suppliers is obtained from the supply and demand adjustment strategy, and the new scheduling time is calculated. When the new scheduling time exceeds the original planned time, idle production lines are scanned, and tasks are split into idle production lines according to order priority to ensure the continuity and efficiency of production. This ensures that the production plan and the procurement progress are closely coordinated and a production schedule is generated, which includes the time nodes of each process and the production line allocation results. Among them, the customer priority order is special customer orders, ordinary orders, and trial orders. Based on the supply and demand adjustment strategy, obtain the material requirements of high-priority orders, such as the material requirements of special customer orders, calculate the pre-allocation quantity of each material, generate an inventory pre-allocation ledger, and record the pre-allocation time, order number, material ID, and pre-allocation quantity to ensure the material supply for high-priority orders. The available inventory is updated in real time after pre-allocation to avoid duplicate allocation; the expression is as follows:

[0023] In the formula, This refers to the pre-allocation of materials. For material requirements, This refers to the pre-occupancy ratio; A three-level simulation model is constructed, including the workshop level, inventory level, and supplier level. The workshop level simulates the delayed impact of equipment downtime on material demand. The inventory level dynamically calculates the inventory turnover rate under different procurement strategies. The supplier level simulates the delivery delay caused by the risk of geopolitical fluctuations. The supply and demand adjustment strategy is input into the three-level simulation model, and several Monte Carlo simulations are run. The simulation effect report is output, including the probability of supply disruption risk, cost increase, and production line waiting time. The implementation effect of the procurement strategy is evaluated in advance, and potential problems are identified. Taking into account the probability of supply disruption risk, cost increase, and on-time delivery rate, a reward function is set up. Negative numbers represent penalties, and positive numbers represent rewards. The weights reflect the company's preference for risk and efficiency. After every 100 decisions, a Q-learning algorithm is used. The simulation results report and actual execution data are input into the algorithm to update the parameters of the multi-objective optimization model for use in the next decision. This adjusts the weights of supplier allocation ratios and transportation methods, continuously optimizing the procurement strategy to adapt to the ever-changing supply chain environment. The expression of the reward function is shown below:

[0024] In the formula, The total reward value, , , These are weighting coefficients, ranked according to the priority of supply disruption risk > cost control > delivery performance. For example, they are set to 0.4, 0.3, and 0.3 respectively. To determine the probability of supply disruption, As the percentage increase in costs, , This represents the actual total cost of this procurement. This represents the average cost of materials to prepare over the past three months. On-time delivery rate is the ratio of the number of orders delivered within the contractually agreed delivery period to the total number of orders in this procurement.

[0025] Specifically, risk response methods include: To address the issue of traditional risk assessment having only one dimension, a hierarchical indicator system is used to cover potential risks across the entire supply chain, enabling systematic risk identification. Starting from three levels—suppliers, logistics and transportation, and materials and inventory—a series of key risk factors are identified, and these indicators are mapped to a standardized dataset with unified units to generate a three-level risk factor library. Collect historical risk event data and label the corresponding risk levels. At the same time, extract risk factor data from the 30 days before the event to construct a training dataset, where positive samples represent risk events and negative samples represent risk-free events. Using the XGBoost algorithm, a risk quantification model is established with standardized indicators from a three-level risk factor library as input. The model outputs the probability of each risk event and thus determines the risk level. The optimal parameters are determined through grid search to reduce the risk of overfitting and improve the model accuracy. In order to avoid poor generalization ability in new scenarios, 5-fold cross-validation is used to evaluate the model performance. When the difference between the accuracy of the training set and the accuracy of the test set exceeds 10%, an L2 regularization term is added. Based on real-time data, the risk quantification model is invoked every hour to calculate and output the current risk level of each supplier, material, and logistics route, forming a real-time risk scoring table to achieve dynamic quantitative assessment of risk. A mapping table between risk levels and response strategies is constructed, specific countermeasures are set for different risk types and levels, and risk response instructions containing recommended strategies are automatically sent to the procurement decision engine based on the risk level in the real-time risk scoring table, so as to ensure that a quick and accurate response can be made when risks occur. The supply and demand adjustment strategy is optimized based on the recommended strategy in the risk response instruction, and the inventory pre-occupancy ledger is updated simultaneously, the inventory allocation is adjusted, the material priority adjustment notification is generated, and the production schedule is adjusted according to the new material priority to prioritize the production material requisition in high-risk scenarios, ensuring that risk control measures are closely coordinated with procurement, production, and inventory management. By leveraging blockchain technology, relevant information for each risk response operation is automatically recorded on the chain, such as activating alternative suppliers, releasing safety stock, generating risk event logs, and recording in detail the operation time, the strategies adopted, and the impact, such as reducing the amount of losses due to supply disruptions. Monthly risk event logs are compiled, feature data from high-risk scenarios are extracted, the risk quantification model is retrained, and the weights of risk factors are adjusted, such as increasing the weights of certain key risk factors. The risk quantification model is continuously optimized to improve the accuracy and relevance of risk assessment and achieve continuous improvement in risk assessment and response. Based on digital twins, various extreme scenarios are preset, such as supplier factory fires or port blockades lasting 30 days. The current supply and demand adjustment strategies and real-time risk scoring table data are input to simulate the transmission effect of risks in the supply chain under extreme scenarios. Based on the simulation results, the corresponding risk response strategies are optimized, such as increasing the minimum allocation ratio of alternative suppliers, and extreme scenario risk contingency plans are generated and added to the risk response strategy library to enhance the supply chain's ability to cope with extreme risks and improve its resilience and risk resistance.

[0026] Specifically, the driving matching methods include: Acquire customers' personalized needs, including product parameters, order urgency, and expected delivery time, and automatically generate unique numbers for each need for subsequent tracking and management. Clean up personalized needs, filter out vague or incomplete information, and transform personalized needs into structured customer requirement work orders. This provides a clear and accurate data foundation for subsequent processes, improving processing efficiency and accuracy. By using NLP to identify key product parameters in customer work orders and converting these parameters into standard material codes, combined with an existing material database, the required materials are determined based on the product structure. Standard materials, special materials, and alternative materials are distinguished, thereby generating a customized bill of materials that includes the type, code, required quantity, suggested supplier, and historical stock preparation and delivery cycle for each material. This achieves accurate conversion from customer needs to material requirements, providing a key basis for procurement decisions. Obtain the inventory quantity of standard materials and the historical stock preparation and delivery cycle of special materials from supplier delivery data to make a feasibility assessment of the materials; If a special material is out of stock and the lead time for preparation and delivery is longer than the customer's required delivery date, the alternative recommendation engine is automatically triggered. This engine sorts and filters alternatives based on both cost and delivery time, and then outputs the recommended solutions. The system provides alternative solutions and generates a demand feasibility assessment report, which includes material shortages, costs, delivery times and performance of the alternative solutions, and delivery risk levels, providing multiple options for procurement decisions and ensuring timely delivery. Extract recent historical transaction data from customers, taking into account the most recent purchase time, purchase frequency, and spending amount. Calculate customer value scores using the RFM model, and rank customers according to their value scores, selecting the top performers. The clients are specialized clients and intermediaries. Customers are categorized as core customers, and the remaining customers are considered ordinary customers, generating a customer value stratification table; among them... , ; An urgency coefficient is calculated by comparing the standard delivery cycle with the customer-required delivery cycle to measure the urgency of the order. The overall equipment efficiency and the current actual production line load rate are also obtained. The production line load rate score is calculated by subtracting the current actual production line load rate from 1. A capacity feasibility score is then calculated by weighting the overall equipment efficiency and the production line load rate score. The weighting is based on a production logic that prioritizes equipment operational stability over production line idle time. In this embodiment, the weight of overall equipment efficiency is 0.6, and the weight of the production line load rate score is 0.4. Taking into account customer value score, urgency coefficient, and capacity feasibility score, an order priority score is calculated using a weighted formula to generate an order priority queue. This ensures that urgent orders from high-value customers are processed first, optimizing resource allocation. The weights used to calculate the order priority score are retrieved from the priority weight configuration table based on customer level, which is obtained from the customer value stratification table. The priority weight configuration table stores the priority weights corresponding to each customer level. Multiple weight combinations of customer value score, urgency coefficient, and capacity feasibility score are set for each customer level. Customer satisfaction and capacity utilization under different combinations are statistically analyzed. The weight combination with the highest sum of customer satisfaction and capacity utilization is selected as the configuration weight for the corresponding customer level, forming the priority weight configuration table. Customer transaction data from the past 12 months is obtained, and the three indicators of recent purchase time, purchase frequency, and consumption amount are calculated. Using the RFM model, customer value scores are obtained. Customers are sorted from high to low value scores, and the top 20% are classified as special customers, the middle 30% as core customers, and the remaining 50% as ordinary customers, generating a customer value stratification table. To address the issue of slow order response from high-value customers, on-time delivery rates for specific customer orders are improved by pre-positioning inventory and ensuring delivery through supplier privileges. Orders identified as belonging to specific customers are processed by extracting the required quantity from a customized bill of materials and obtaining the corresponding pre-positioned quantity from the inventory pre-positioning ledger. Based on historical demand fluctuations and supplier delivery cycles, safety stock of corresponding materials is set to cover the regular order fluctuations of all customers. If the pre-allocated amount does not exceed the safety stock, the safety stock will be used to meet the current orders of specialized customers. Specialized customer orders can only pre-allocate 70% of the safety stock, with the remaining 30% used to cover the urgent needs of general customers, preventing supply disruptions for general customers due to the safety stock being fully pre-allocated. If the pre-allocated amount exceeds the safety stock, the reserve allocation mechanism will be automatically triggered, and the reserve allocation amount will be calculated. Furthermore, each call should not exceed 15% of the total reserve to avoid excessive consumption of reserve resources; the expression is as follows:

[0027] In the formula, As a pre-reserved quantity, For safety stock, This is used to filter and output the maximum value from the input parameters to determine whether the material pre-occupancy exceeds the safety stock and to calculate the basic reserve to be called up. The rules for setting the total amount of special reserves include 30 days of average consumption for special materials, 15 days of average consumption for core materials, and no special reserves for ordinary materials. The total reserve amount called up in a single instance shall not exceed 15%. This limit is determined based on the possibility of multiple emergency calls to ensure that there are still reserves available in the event of consecutive risk events. When the cumulative call up exceeds 50% of the total reserve amount, emergency procurement to replenish the reserve shall be initiated immediately. Read the candidate suppliers, prioritize the supplier ranked first as the special supplier, and generate a special customer order processing form, which includes inventory pre-positioning records and supplier allocation results; Orders identified as belonging to core customers are used to calculate the required inventory levels using the Economic Order Quantity (EOQ) model, and an adjustment range is set based on the standard deviation of the customer's historical order fluctuations. Obtain the historical on-time delivery rate for each supplier, select suppliers with a historical on-time delivery rate greater than the delivery threshold as primary suppliers, and the remaining suppliers as backup suppliers, and allocate resources to the primary suppliers. Order volume, allocated to alternative suppliers Order volume is calculated, and the primary and alternative suppliers are dynamically adjusted monthly based on supplier delivery performance to balance inventory costs and supply risks. This generates a core customer inventory execution order, which includes order allocation ratios, delivery cycles, and cost budgets. The delivery threshold is set by those skilled in the art based on average industry standards, such as 95%. To determine the allocation ratio, since the main supplier offers lower prices but carries concentrated risk, and alternative suppliers serve as risk buffers, this embodiment adopts a [missing information - likely a specific supplier's pricing strategy]. ; Orders from regular customers are identified and summarized weekly for similar materials. If the total quantity of materials summarized in the current week reaches the minimum order quantity, a unified purchase is made; otherwise, it is postponed to the following week. Simultaneously, the purchase frequency is reduced, and payment terms are set, such as a 60-day payment period, with a 2% prepayment discount offered to incentivize customers to pay early, thus reducing inventory costs. This generates a consolidated inventory list for regular customers, including consolidated order details and payment terms. The minimum order quantity is calculated based on the Economic Order Quantity (EOQ) model. The peak demand for seasonally scarce materials is predicted over a future period using a cyclical forecasting model. Based on the forecast results, a phased inventory preparation strategy is formulated, comprising three phases: Phase 1 (the three months leading up to the peak season) and Phase 2 (the inventory preparation amount being 1 / 3 of the predicted peak value). Phase 2, which is one month before the peak season, involves stocking up at a level that is projected to be the peak. Phase 3, during the peak season, involves stocking up at a rate that is twice the predicted peak. times, , , It can be dynamically adjusted by those skilled in the art based on real-time sales figures, and Historical inventory preparation data for the same season over the past three years was obtained. With the dual optimization objectives of maximizing inventory turnover and minimizing stockout rate, a grid-based optimization fitting method was used to obtain the baseline allocation coefficients for three-stage inventory preparation. Using a natural week as the rolling update cycle, actual terminal sales were collected for the current period. The difference between real-time sales and predicted demand was used as the numerator, and the predicted demand as the denominator to calculate the real-time sales deviation rate. Simultaneously, the inventory preparation response coefficients for each stage were obtained, characterizing the sensitivity of the inventory preparation ratio at each stage to sales deviation. Since stage 1 has a longer inventory preparation cycle, it has the greatest adjustment flexibility and is most sensitive to sales... The deviation response is the most sensitive, so the inventory response coefficient is set to 1.2. Phase 2 has a moderate inventory cycle and medium response sensitivity, so the inventory response coefficient is set to 1. Phase 3 is the peak season, with tight supplier capacity and limited adjustment space, resulting in the lowest response sensitivity, so the inventory response coefficient is set to 0.6. The temporary allocation coefficient for each phase is calculated by multiplying the inventory response coefficient by the real-time sales deviation rate. Adding 1 to the temporary allocation coefficient and then multiplying by the baseline allocation coefficient yields the theoretical allocation coefficient for the corresponding phase. To ensure that the total inventory level across the three phases matches the peak demand, even... The three theoretical allocation coefficients obtained from independent calculations are normalized to obtain the final phased inventory preparation coefficient. , , For example, take , , This generates a seasonal material stocking plan; the periodic prediction model is obtained by training and fitting the Prophet time series prediction model with a large amount of historical data. The system synchronously monitors the shelf life of materials. When the remaining shelf life is less than the shelf life threshold, it automatically marks the material and triggers a processing flow to reduce stagnant inventory and generate a stagnant material processing order. If the material is marketable, a promotional plan is automatically generated, such as a 30% discount clearance sale. If the material is not marketable, a donation or scrapping application is triggered. The shelf life threshold is set by those skilled in the art based on industry-standard practices. For example, the shelf life threshold for electronic components is 3 months. The shelf life threshold can be adjusted for different material types: 1 month for food materials, 6 months for mechanical parts, and 2 months for chemical materials. For high-frequency, general-purpose materials, a Vendor Managed Inventory (VMI) model is adopted, with a replenishment threshold set. When the inventory level falls below the replenishment threshold, an order is automatically placed, targeting the top-ranked supplier among the candidate suppliers, and the replenishment order quantity is calculated. This generates a replenishment order; the expression is shown below:

[0028] In the formula, This represents the average daily consumption. The replenishment ratio is set to 1.5 by default, meaning the replenishment quantity is 1.5 times the average daily consumption. When the inventory level is below 50% of the safety stock, the replenishment ratio increases to 2.0. When the inventory level is above 150% of the safety stock, the replenishment ratio decreases to 1.0. High-frequency general-purpose materials are those whose average daily consumption exceeds a preset consumption threshold. The consumption threshold is set by those skilled in the art based on the balance point between inventory turnover cost and preparation cost, that is, the critical value of average daily consumption that minimizes total inventory cost is calculated through the Economic Order Quantity (EOQ) model. Within a pre-defined supplier certification period, several suppliers with similar technologies are identified and their parameters are certified for consistency. Suppliers with parameter consistency exceeding a pre-defined threshold are added to the supplier pool. The certification status is updated quarterly, and suppliers failing certification are removed from the pool. The expression for parameter consistency is as follows:

[0029] In the formula, For parameter consistency, For the number of key parameters, For the supplier's sample Parameter values, for The standard parameter values ​​are defined, and the parameter consistency threshold is 95%, meaning that only suppliers with a parameter deviation rate of less than 5% can be included in the candidate pool. The system acquires information on changes in input requirements, such as increased order volume from specific customers. It then retrieves inventory, capacity, and supplier data from a standardized dataset and uses digital twin technology to simulate and output material shortages, changes in inventory turnover, and peak production line load rates. Finally, it generates optimal adjustment plans, such as activating special reserves, placing additional orders with alternative suppliers, or outsourcing some processes.

[0030] Specifically, the optimization methods include: Define core performance indicators, including material efficiency, risk control, and customer value, and clarify the corresponding calculation formulas, such as decision response timeliness, on-time delivery rate of emergency orders, and major risk loss rate; A correlation matrix between decision-making strategies and performance indicators is established to compare the actual implementation effects of the strategies with the target values, evaluate the effectiveness of each strategy, and use techniques such as the Isolation Forest algorithm to identify outliers in the data, generate a list of abnormal events, locate problem areas in the system operation, and provide direction for subsequent optimization. The parameters of the Isolation Forest algorithm include 100 trees, a subsample size of 256, a maximum tree depth of 8, and an anomaly score threshold of 0.6. Anomalies with an anomaly score greater than 0.6 are identified as outliers. Based on the execution log of the supply and demand adjustment strategy, the multi-objective optimization model in the procurement decision-making closed loop is adjusted. When the actual delivery cycle of a supplier is consistently higher than the model prediction, the safety factor of the corresponding supplier's delivery cycle is increased. By using reinforcement learning algorithms, with decision response timeliness and cost deviation rate as reward functions, the weighting of supplier allocation ratios is adjusted to improve the accuracy and efficiency of procurement decisions. Add full data on newly occurring supply chain risk events, including event type, trigger time, and impact scope, to the risk quantification model. Retrain the model parameters. If a risk indicator in the risk quantification model fails to identify the risk in an actual supply chain risk event (i.e., the risk quantification model does not trigger the corresponding warning), increase the weight of the corresponding indicator. This enhances the risk quantification model's ability to identify and warn of various risks. Each time a risk is not successfully identified, the weight of the corresponding indicator increases by 0.1, with an upper limit of 0.5 and a lower limit of 0.05, to avoid model imbalance caused by excessively high or low weights for a single indicator. After three consecutive successful risk identifications, the weights are restored to their initial values. The risk quantification model's risk identification process includes: building a risk indicator library based on the supply chain scenario, collecting business data for the corresponding indicator dimensions, substituting the business data into the risk quantification model to match the feature vectors of historical risk events, calculating the actual matching degree, and triggering a risk warning if the actual matching degree is greater than the risk identification threshold. It also involves extracting feature matching degree samples from the 30 days prior to the occurrence of all risk events, calculating their probability distribution, and using the 90th quantile as the risk identification threshold to ensure a recall rate of no less than 90% for historical risk events. Based on customer demand response reports, adjust the weighting of priority assessments in flexible procurement; when specific customers are more concerned about on-time delivery, increase the weighting of customer value scores and decrease the weighting of capacity scores, so that procurement strategies are more in line with customer needs. The system performs retrospective analysis on existing procurement, risk response, and customer-tiered procurement strategies. Based on actual performance, it adjusts trigger thresholds or adds constraints. A rule engine detects conflicts between different strategies in the strategy library and generates conflict resolution solutions, such as prioritizing risk strategies over inventory strategies and updating all strategies synchronously. Trigger threshold adjustments include: statistically analyzing the accuracy and false alarm rates of individual strategies; if the accuracy is below 80% for three consecutive periods, the corresponding trigger threshold is lowered by 5%; if the false alarm rate is above 30% for three consecutive periods, the corresponding trigger threshold is raised by 5%; otherwise, the trigger threshold is not adjusted. Key decision-making data, such as early warning trigger values, final procurement strategies, and actual performance indicators, are encrypted and stored on the blockchain to generate an immutable decision-making blockchain log, enabling full-chain traceability of the decision-making process. Natural language processing (NLP) technology is used to extract high-frequency and effective strategies from the decision-making blockchain log, build a supply chain best practice case library, and store practical experience in a structured way for easy querying and reuse. Construct a knowledge graph that includes decision-making strategies, performance indicators, and influencing factors to show the relationships between these factors and support intelligent question-and-answer functionality. For example, when a user asks how to improve the satisfaction of specific clients while reducing costs, the knowledge graph can automatically recommend a balancing strategy to promote knowledge sharing and experience transfer, thereby improving the overall decision-making level.

[0031] Example 2 Please see Figure 4 Another embodiment of the present invention provides a supply chain risk monitoring system, comprising: a response module, a risk assessment module, a matching module, and a feedback module; The response module is used to collect production progress data from the smart workshop, material consumption data from the warehouse, and delivery data from suppliers in real time. It cleans and integrates the data to generate standardized datasets, establishes dynamic early warning rules, automatically generates supply and demand adjustment strategies through multi-objective optimization, and triggers execution synchronously. This achieves automated real-time response from production data to procurement decisions, eliminates delays caused by manual intervention, and improves decision-making efficiency. Based on historical production and procurement data, it simulates a three-tier supply chain scenario involving the workshop, inventory, and suppliers, and pre-analyzes the impact of various supply and demand adjustment strategies on production. Through reinforcement learning optimization, it dynamically updates the supply and demand strategy contingency plan library based on simulation results and actual execution effects, and outputs optimized decision parameters. The risk assessment module is used to build multi-dimensional risk assessment indicators covering suppliers, logistics, and materials. It uses machine learning and real-time data to calculate risk scores to quantify risk levels and configure differentiated response strategies. The matching module is used to break down customers' personalized needs into material requirements in real time, connect to the global material database to intelligently match inventory and preparation delivery cycles, and build a three-dimensional priority evaluation model by combining customer value, order urgency, and capacity load. It dynamically adjusts the material pre-positioning strategy and supplier subsidy mechanism for special customers' urgent orders, and formulates phased preparation and VMI inventory coordination strategies for seasonal or general materials. It achieves refined management of customer segmentation, demand segmentation, and procurement segmentation, improves the on-time delivery rate of special customer orders, and improves the conversion efficiency of personalized needs. The feedback module is used to obtain the performance of supply and demand and customer satisfaction, update the risk quantification model and demand analysis rules in real time, dynamically optimize algorithm parameters through transfer learning, select best practice cases, and enable the system to continuously evolve as business data accumulates.

[0032] Example 3 An embodiment of the present invention provides a storage medium storing a computer program, which, when executed by a processor, implements the aforementioned supply chain risk monitoring method.

[0033] Example 4 A computer device according to an embodiment of the present invention includes a processor and the aforementioned storage medium. The processor executes a computer program in the storage medium to implement the aforementioned supply chain risk monitoring method.

[0034] In summary, this invention captures data from smart workshops, warehouses, and suppliers in real time, cleans and integrates this data to generate a standardized dataset, calculates CSI and DRI using real-time response methods, establishes multi-level early warning rules, generates material early warning work orders, uses LSTM to predict material demand, constructs a comprehensive supplier rating system, determines multi-objective optimization model parameters, solves and generates supply and demand adjustment strategies, sends these strategies to suppliers to adjust production schedules and pre-position safety stock, and optimizes procurement strategies through digital twin simulation and Q-learning algorithms. Secondly, it constructs a three-level risk factor library, uses the XGBoost algorithm to establish a risk quantification model, generates a real-time risk scoring table, configures response strategies, and optimizes supply and demand adjustment strategies, continuously improving risk assessment and response through blockchain and digital twin technologies. Next, it acquires personalized customer needs, generates customer demand work orders, converts them into customized bills of materials, calculates customer value scores, urgency coefficients, and capacity feasibility scores, generates an order priority queue, formulates procurement strategies for different customer levels, manages seasonal and high-frequency general-purpose materials, certifies alternative suppliers, and uses digital twin technology to handle demand changes. Finally, it defines core performance indicators, evaluates the effectiveness of strategies, adjusts model parameters and strategies, and achieves continuous system evolution through blockchain and knowledge graphs.

[0035] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A supply chain risk monitoring method, characterized in that, include: Construct a standardized dataset, set up a real-time response method, establish dynamic early warning rules, generate supply and demand adjustment strategies, simulate the impact of the supply and demand adjustment strategies on production, and dynamically update the supply and demand strategy contingency plan library. Construct multi-dimensional risk assessment indicators, set risk response methods, calculate risk scores, and configure differentiated response strategies; Personalized needs are broken down into material requirements in real time, driving matching methods are set, a three-dimensional priority evaluation model is constructed, material pre-positioning strategies and subsidy mechanisms are dynamically adjusted, and phased inventory preparation and VMI inventory coordination strategies are formulated simultaneously. Obtain the performance of supply and demand and customer satisfaction, set optimization methods, and update the risk quantification model and demand analysis rules in real time.

2. The supply chain risk monitoring method according to claim 1, characterized in that: The real-time response method includes: Calculate the urgency of material consumption and the risk of supplier delivery, and establish multi-level early warning rules to generate material early warning work orders; Construct a demand forecasting model, train the demand forecasting model based on historical data, and output the predicted material demand. Construct a comprehensive supplier rating system, calculate the comprehensive score for each supplier, and select the top-ranked suppliers based on their comprehensive scores. One supplier was selected as a candidate supplier; Calculate delivery risks and inventory costs, determine the objective function, and set constraints to generate a multi-objective optimization model; Using the order allocation ratio as the decision variable, the CPLEX solver is called to solve the problem. Based on the predicted material demand and relevant information of candidate suppliers, a supply and demand adjustment strategy is generated.

3. The supply chain risk monitoring method according to claim 2, characterized in that: The real-time response method further includes: The supply and demand adjustment strategy is sent through the interface and a smart contract is triggered to generate a supplier order confirmation receipt. Obtain the supplier's actual delivery cycle, calculate the new scheduling time, and when the new scheduling time exceeds the original planned time, scan idle production lines, split tasks to idle production lines according to order priority, and generate a production schedule. Obtain material requirements for high-priority orders, calculate the pre-allocation quantity of each material, generate an inventory pre-allocation ledger, and update the available inventory in real time after pre-allocation. Construct a three-level simulation model, including the workshop level, inventory level, and supplier level, to simulate the effects of supply and demand adjustment strategies and output a simulation effect report; Set a reward function, use the Q-learning algorithm, and update the parameters of the multi-objective optimization model based on the simulation effect report and actual execution data.

4. The supply chain risk monitoring method according to claim 3, characterized in that: The risk response methods include: Construct a three-tiered risk factor database covering suppliers, logistics and transportation, and materials and inventory; Collect historical risk event data and label the corresponding risk levels. At the same time, extract risk factor data before the event occurs to construct a training dataset. Using the XGBoost algorithm, with standardized indicators from the three-level risk factor library as input, a risk quantification model is established, and the probability of each risk event is output. Based on real-time data, the risk quantification model is invoked to generate a real-time risk scoring table; Construct a mapping table between risk levels and response strategies, set specific countermeasures for different risk types and levels, and generate risk response instructions.

5. The supply chain risk monitoring method according to claim 4, characterized in that: The risk response method also includes: The supply and demand adjustment strategy is optimized according to the recommended strategy in the risk response instruction, and the inventory pre-occupancy ledger is updated synchronously to generate a material priority adjustment notification. Relevant information for each risk response operation is automatically recorded on the blockchain to generate a risk event log; Based on the pre-set periodic summary of risk event logs, the risk quantification model is retrained and the risk factor weights are adjusted. Based on digital twins, extreme scenarios are simulated to simulate the risk transmission effect and generate risk contingency plans.

6. The supply chain risk monitoring method according to claim 5, characterized in that: The drive matching method includes: Get personalized requirements and generate unique numbers, then clean them and convert them into work orders; NLP is used to identify key product parameters and convert them into standard material codes. Combined with a material database, a customized bill of materials is generated, including standard materials, special materials and alternative materials. Obtain standard material inventory and historical stock preparation and delivery cycle of special materials to make a feasibility assessment. When there is no inventory and the stock preparation and delivery cycle is overdue, trigger the alternative solution recommendation engine and generate a demand feasibility assessment report. Extract historical transaction data to calculate value scores and generate a value stratification table; Calculate the urgency coefficient and capacity feasibility score, and generate an order priority queue.

7. The supply chain risk monitoring method according to claim 6, characterized in that: The driver matching method further includes: Identify specific customer orders, extract the required quantity and obtain the pre-allocation quantity for the order, set safety stock, and trigger the reserve call mechanism when the pre-allocation quantity exceeds the limit; Read the candidate suppliers, select the specific supplier, and generate a specific customer order processing form; Identify core customer orders, calculate inventory levels using the EOQ model, set adjustment ranges based on historical order fluctuations, select and dynamically adjust primary and alternative suppliers according to on-time delivery rates, and generate core customer inventory execution orders. Identify regular customer orders, summarize material orders weekly, prepare goods in a unified manner after the minimum order quantity is reached, set payment terms, and generate a consolidated preparation order for regular customers.

8. The supply chain risk monitoring method according to claim 7, characterized in that: The driver matching method further includes: Predict peak demand for seasonally scarce materials and develop phased inventory strategies, generate seasonal material inventory plans, monitor material shelf life and trigger processing procedures. For high-frequency, general-purpose materials, VMI mode is used to set replenishment thresholds and generate replenishment instructions. During the supplier certification period, acquire suppliers with similar technologies, include them in the supplier pool after certification, and update the pool regularly. The system acquires information on demand changes, retrieves data, and uses digital twin technology to simulate and output material shortages, inventory turnover rate changes, and peak production line load rates, thereby generating the optimal adjustment plan.

9. The supply chain risk monitoring method according to claim 8, characterized in that: The optimization method includes: Define core performance indicators for material efficiency, risk control, and customer value; establish a correlation matrix between decision-making strategies and performance indicators; and identify outliers. Adjust the parameters and weight configurations of the multi-objective optimization model and risk quantification model, conduct retrospective analysis and update inventory preparation, risk response, and customer-tiered inventory preparation strategies, encrypt and store key decision data, and construct a knowledge graph.

10. A supply chain risk monitoring system, used to implement the supply chain risk monitoring method as described in any one of claims 1-9, characterized in that, include: Response module, risk assessment module, matching module, and feedback module; The response module is used to collect data from smart workshops, warehouses and suppliers in real time, clean and integrate the data to generate a standardized dataset, establish dynamic early warning rules, generate supply and demand adjustment strategies through multi-objective optimization and trigger their execution, simulate supply chain scenarios based on historical data, optimize and dynamically update the supply and demand strategy contingency plan library through reinforcement learning, and output decision parameters. The risk assessment module is used to construct multi-dimensional risk assessment indicators, calculate risk scores using machine learning and real-time data, and configure differentiated response strategies. The matching module is used to break down customers’ personalized needs into material requirements, connect to the material database to match inventory and preparation delivery cycles, build a three-dimensional priority evaluation model, dynamically adjust the special customer expedited order strategy, formulate a phased preparation and VMI inventory coordination strategy, and realize hierarchical management and control of customers, demand, and supply and demand. The feedback module is used to obtain the performance of supply and demand and customer satisfaction, update the risk quantification model and demand analysis rules, optimize algorithm parameters through transfer learning, and select best practice cases.

11. A medium, characterized in that, The medium stores a computer program, which, when executed by a processor, implements the supply chain risk monitoring method as described in any one of claims 1-9.

12. A computer device, characterized in that, include: Memory, used to store instructions; A processor for executing the instructions, causing the device to perform operations that implement the supply chain risk monitoring method as described in any one of claims 1 to 9.

Images