Supplier price management system and method based on big data analysis

Abstract

The present invention belongs to the technical field of big data analysis. The present invention discloses a supplier price management system and method based on big data analysis, including: collecting core data and auxiliary data to construct a multi-dimensional data cube; dynamically calculating the grid availability index to form a dynamic spatiotemporal grid benchmark library, and synchronously generating a supply chain topology map; receiving and quantifying business target instructions, and selecting paths through a rule engine; generating procurement strategy recommendations with confidence ratings and a dynamic priority supplier list; dynamically adjusting permissions through an ABAC model, and triggering a smart contract engine to generate orders or bargaining invitations; parallel approval through a workflow engine to generate a procurement order set and a permission change log; receiving real-time data streams and a historical case library of the Internet of Things to trigger disposal logic; matching similar historical cases to push historical optimal strategies, and synchronously updating weight rules, data cube structures, and supplier credit score thresholds to form a full-link closed-loop iteration.

Description

Technical Field

[0001] The present invention relates to the technical field of big data analysis, and more specifically, to a supplier price management system and method based on big data analysis. Background Art

[0002] With the rapid development of big data analytics and supply chain management technologies, the demand for intelligent supplier price management systems is becoming increasingly urgent. Companies urgently need to optimize procurement decisions by integrating multi-source data (such as real-time supplier production capacity, regional policies, and logistics dynamics) while also addressing the risks and uncertainties of a dynamic market environment. However, existing systems often rely on traditional data management methods and lack the ability to deeply analyze complex supply chain scenarios, making it difficult to achieve global, adaptive procurement strategy optimization.

[0003] Existing supplier price management systems suffer from a series of interrelated technical flaws, resulting in insufficient reliability and efficiency in dynamic supply chain environments. First, the system's data utilization is limited to isolated procurement records, failing to effectively integrate real-time supplier data (such as inventory and production capacity), external auxiliary data (such as regional policies and competitor prices), and historical case libraries, resulting in a lack of a holistic perspective in decision-making. Second, key processes (such as route selection and supplier screening) rely on manual price comparisons and empirical judgment, which is inefficient and prone to bias. Risk response (such as delivery delays and quality defects) lacks automated processing logic and relies on post-event manual intervention. Furthermore, routing strategy updates lag, failing to respond in real time to supply chain fluctuations (such as regional power restrictions and logistics disruptions). Supplier credit assessment and authority management lack dynamic mechanisms, making them unable to adapt to the need for authority adjustments in emergency scenarios. Furthermore, the existing system lacks closed-loop optimization capabilities, lacks knowledge accumulation after event handling, and historical cases are not integrated with AI models. This results in a reliance on manual experience for policy optimization and an inability to deliver optimal solutions through semantic matching. These issues collectively limit the system's decision-making resilience and intelligence, making it difficult to meet the complex demands of modern supply chains. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: a supplier price management system based on big data analysis, comprising:

[0005] Multidimensional data cube construction unit: This unit collects real-time core supplier data through standardized interfaces and obtains auxiliary data through third-party APIs to construct a multidimensional data cube. It also divides suppliers into clusters based on geographic grids, dynamically calculates grid availability indexes, forms a dynamic spatiotemporal grid benchmark library, and simultaneously generates a supply chain topology map.

[0006] Multi-path decision-making collaboration unit: Based on a data cube and dynamic spatiotemporal grid benchmark library, it receives and quantifies business objective instructions and selects paths through a rule engine. It uses a dynamic weighted voting model to combine path selection results to output procurement strategy recommendations with confidence ratings and a dynamic priority supplier list.

[0007] Smart Contract Execution and Permission Network Unit: Based on a dynamic priority supplier list and supply chain topology map, combined with ERP contract status and supplier credit scores, permissions are dynamically adjusted through the ABAC model, triggering the smart contract engine to generate orders or bargaining invitations. Through parallel approval through the workflow engine, a purchase order set and permission change log are generated.

[0008] Closed-loop autonomous optimization and resilience enhancement unit: Receives real-time IoT data streams and historical case libraries, triggers handling logic through a multimodal exception response engine; matches similar historical cases based on the BERT model to push historical optimal strategies, and simultaneously updates weight rules, data cube structures, and supplier credit score thresholds to form a full-link closed-loop iteration.

[0009] Furthermore, the multidimensional data cube is constructed in the following manner:

[0010] Collect real-time core data from suppliers through standardized interfaces, including quotes, inventory levels, actual production capacity, and contract status;

[0011] Obtain auxiliary data about the region through third-party APIs, including regional power restrictions, competitive product prices, designed production capacity, and logistics disruptions.

[0012] Clean and unify the collected core data and auxiliary data;

[0013] Define the dimensions and metrics of the data cube. Dimensions include time, supplier, administrative region, product, event, and contract. Metrics include average quote price, inventory availability, capacity utilization, regional risk factor, and contract default risk index.

[0014] Calculate metrics for different administrative regions based on each administrative region dimension, including:

[0015] Extract the unit price, discount rules, and currency unit of the quotation data from the core data, and calculate the average quotation price;

[0016] Extract the available inventory and the maximum inventory threshold of the inventory level in the core data, and calculate the ratio to obtain the inventory availability rate;

[0017] Take the expiration date and default clause of the contract status and calculate the contract default risk index;

[0018] Based on the processed auxiliary data, a time window is set. Taking the current moment as the benchmark, the ratio of the difference between the maximum and minimum prices of the competitor product within the past time window to the average price of the competitor product is calculated as the price fluctuation range of the competitor product.

[0019] Extract the number, duration, and impact range of regional power-rationing policies within a past time window from the auxiliary data, perform normalization on them, and then perform weighted summation to obtain the power-rationing risk value.

[0020] The price fluctuation range of competing products, the power restriction risk value, and the number of logistics interruption events in the past time window are used as coefficient indicators to calculate the administrative region risk coefficient. Each coefficient indicator is normalized and weighted using the entropy weight method. The regional risk coefficient is then calculated through weighted summation.

[0021] According to the defined dimensions and measures, dimension tables and fact tables are constructed through the star model, and foreign keys are integrated to form a cube structure as a data cube.

[0022] Furthermore, the construction method of the dynamic spatiotemporal grid benchmark library includes:

[0023] Based on the dimensions of the data cube, the supplier dimension, product dimension, and event dimension are business-associated by product type and event type, and different dynamic areas are divided according to the business associations as different supplier clusters;

[0024] Extract all data covered by the dynamic area from the data cube, and recalculate the regional risk coefficient corresponding to the dynamic area as the dynamic regional risk coefficient;

[0025] The ratio of the total actual output value of all enterprises in the dynamic area to the total designed capacity is taken as the capacity utilization rate of the dynamic area;

[0026] Extract the contract performance status of all contracts in the dynamic area and calculate the contract performance rate in the dynamic area;

[0027] Based on the capacity utilization rate, inventory availability rate and contract fulfillment rate in the dynamic area, weights are assigned to the capacity utilization rate, inventory availability rate and contract fulfillment rate respectively, and then combined with the dynamic area risk coefficient to calculate the grid availability index;

[0028] The grid availability index and the dynamic regional risk coefficient of the dynamic region are integrated as the dynamic space-time grid benchmark of the dynamic region; the dynamic space-time grid benchmarks of all dynamic regions are integrated to generate a dynamic space-time grid benchmark library.

[0029] Furthermore, the supply chain topology map is constructed in the following ways:

[0030] Based on the dimensions of the data cube, all suppliers, products, dynamic regions, and events are treated as nodes, and corresponding attributes are added to each node. The supplier node is bound to the dynamic region node. If the dynamic region risk coefficient changes, the supplier's grid availability index is synchronously triggered to update.

[0031] Extract all supply relationships, logistics paths, contract associations, and event impacts in the data cube as edges, and assign edge weights to each type of edge;

[0032] The relationship between suppliers and products is regarded as the supply relationship. The supplier's capacity utilization rate is calculated as the supplier's capacity utilization rate, and the product of the supplier's capacity utilization rate and the corresponding dynamic regional risk coefficient is taken as the basic weight of the corresponding edge.

[0033] The transportation route between two dynamic areas is used as the logistics path, the corresponding transportation cost is obtained, and the product of the dynamic area risk coefficient and the transportation cost is taken as the basic weight of the corresponding edge;

[0034] The relationship between the supplier and the contract is regarded as the contract association, and the contract default risk index is taken as the basic weight of the corresponding edge;

[0035] The relationship between the event and the affected area is taken as the event impact. The industrial chain dependence of the dynamic area is calculated using the TFR formula. The dynamic area risk coefficient and the industrial chain dependence are weighted and summed according to a preset ratio to obtain the dynamic area vulnerability. The product of the impact range of the event and the dynamic area vulnerability is taken as the basic weight of the corresponding edge.

[0036] Normalize the basic weights of all edges so that the sum of all basic weights is 1;

[0037] Multiply the normalized basic weight by the global adjustment coefficient of the dynamic regional risk coefficient to obtain the edge weight;

[0038] Based on the nodes and edges, a supply chain topology map is constructed.

[0039] Furthermore, the definition method of the rule engine selecting the path includes:

[0040] Decompose business goals into cost path indicators, market path indicators and risk path indicators through KPI decomposition based on the SMART principle;

[0041] The cost path indicator is calculated by calculating the weighted sum of transportation cost and inventory cost to obtain the total cost of each path, which is used as the cost path quantification;

[0042] Query all paths from the dynamic area where the starting point is located to the dynamic area where the end point is located through the supply chain topology map, and take the sum of the logistics path edge weights of the path as the transportation cost of each path;

[0043] The total product of the available inventory and the unit holding cost of each inventory point along the path is obtained through the data cube as the inventory cost of the path;

[0044] The market path indicator is the weighted sum of the on-time delivery rate and the market demand match, which is used to quantify the market path.

[0045] Based on historical order data, calculate the proportion of orders delivered on time along the route as the on-time delivery rate;

[0046] Calculate the ratio of the market demand in the dynamic area to the total actual production capacity in the dynamic area to obtain the market demand matching degree;

[0047] The risk path indicator is to calculate the path risk value of each path as the risk path quantification;

[0048] The reliability index of each supplier node or dynamic area node that the path passes through is calculated by the fault tree analysis method as the path dependence of the path to the corresponding node;

[0049] The dynamic regional risk coefficient of each dynamic area passed by the path is obtained through the dynamic space-time grid benchmark library;

[0050] Calculate the sum of the product of the dynamic regional risk coefficient and the path dependence to obtain the basic risk value of the path, and multiply the basic risk value by the preset risk weight to obtain the path risk value;

[0051] Based on the quantified cost path indicators, market path indicators, and risk path indicators, the path selection rules of the rule engine are defined as follows:

[0052] The cost path is defined as choosing the path with the lowest total cost;

[0053] The market path is defined as choosing the path with the highest market value;

[0054] The risk path is defined as the path with the lowest path risk value.

[0055] Furthermore, the procurement strategy recommendations with confidence ratings and the dynamic priority supplier list are obtained in the following manner:

[0056] Based on the selection results of the cost path, market path, and risk path, the total cost, market value, and path risk value of the path are normalized respectively. The cost score, market score, and risk score are used as inputs to the dynamic weighted voting model, which outputs procurement strategy recommendations with confidence ratings and a dynamic priority supplier list.

[0057] The dynamic weight allocation rule is defined as dynamically adjusting the path weight by the edge weight adjustment coefficient, and the total path weight is 1;

[0058] The dynamic adjustment mechanism is defined as follows: if the dynamic area risk coefficient increases and exceeds the preset dynamic area risk coefficient threshold, the risk weight will be increased by a preset ratio. If an emergency order instruction is received, the market weight will be increased by a preset ratio and the cost weight will be reduced accordingly.

[0059] Combine the cost score, market score, and risk score with the corresponding weights and calculate the weighted sum to get the comprehensive score of the path;

[0060] The confidence rating generation logic is defined as: the confidence rating is divided into high confidence, medium confidence and low confidence;

[0061] Set scoring interval thresholds, as well as cost interval thresholds and risk interval thresholds;

[0062] Based on the comprehensive score, cost score and risk score, the confidence level of the path is determined by the score interval threshold, cost interval threshold and risk interval threshold;

[0063] For each supplier node in the path, extract the performance status of all contracts corresponding to the supplier node, and take the ratio of the number of completed contracts to the total number of failed contracts as the supplier contract performance rate;

[0064] Based on the edge weights of the supply relationship in the supply chain topology graph, the path dependency of the path on the supplier node is extracted, and the weighted sum is performed with the available inventory of the supplier node and the supplier contract fulfillment rate to obtain the supplier priority score of each supplier;

[0065] Generate a dynamic priority supplier list by sorting each supplier in descending order of supplier priority score;

[0066] Set a risk area threshold. If the dynamic area risk coefficient of the dynamic area is less than the risk area threshold, it is determined to be a low-risk area. If the dynamic area risk coefficient is greater than or equal to the risk area threshold, it is determined to be a high-risk area.

[0067] The dynamic adjustment logic of the priority sorting rules that define the dynamic priority supplier list is as follows: in the urgent order scenario, the path supplier with the highest market value is given priority. If the path passes through a high-risk area, the supplier in a low-risk area is given priority.

[0068] Furthermore, the smart contract implementation method for generating an order or a bargaining invitation includes:

[0069] Obtain the supplier's ERP contract status, credit score, and rating through the supply chain topology map and ERP system interface;

[0070] Obtain the grid availability index of the dynamic area where the supplier is located from the dynamic spatiotemporal grid benchmark library;

[0071] The definition of ABAC model includes attributes and policy rules. The defined attributes include user attributes, environment attributes and operation attributes.

[0072] User attributes include supplier level and credit score; environmental attributes include current status and grid availability index; operation attributes include request type; current status includes normal status and emergency status;

[0073] Define policy rules based on attributes, including defining access rights for different levels of suppliers based on their current state;

[0074] Based on the defined ABAC model, the ABAC module of the smart contract engine generates the supplier's access control list and records the permission change log;

[0075] Define the triggering rules of the smart contract engine, including the regular procurement process and the reverse link contract;

[0076] Define the triggering condition for the regular procurement process as when the ERP contract status is valid and the supplier is in the dynamic priority supplier list;

[0077] The execution steps are defined as selecting the highest-priority supplier in the dynamic priority supplier list, calling the ERP interface of the smart contract engine, and automatically generating an order based on the preset historical order template;

[0078] The triggering condition of the reverse link contract is defined as if the grid availability index of the dynamic area where the supplier is located is lower than a preset grid availability index threshold;

[0079] The execution steps are defined as sending a negotiation invitation to the second-tier supplier based on the preset historical quotation template, including parameters and constraints;

[0080] The parameters are demand volume, dynamic regional risk factor, and order urgency, and the constraint is that the bargaining price must be lower than the preset percentage of the first-tier supplier's quotation.

[0081] Furthermore, the generation method of the purchase order set and the permission change log includes:

[0082] Based on the price negotiation invitation sent to the second-tier supplier, when the second-tier supplier accepts the price negotiation invitation, or when the order in the regular status requires manual confirmation, the parallel approval workflow engine is started;

[0083] Set approval nodes, including procurement, finance, and legal departments, and use the workflow engine to execute parallel approval operations based on the approval nodes;

[0084] Define parallel approval rules so that normal orders only require approval from the purchasing department, while emergency orders or orders requiring price negotiation require parallel approval from the purchasing, finance, and legal departments.

[0085] When all approvals are passed, the ERP contract status is updated through the smart contract engine, the historical order template is called, and the final purchase order set is generated based on the approval results;

[0086] When the emergency state is lifted, the temporary permissions of the secondary supplier are revoked through the ABAC module of the smart contract engine, and the supplier's access control list is updated, the normal status permission rules are restored, the permission change time, approval process and order execution results are recorded synchronously, and a permission change log is generated.

[0087] Furthermore, the method of synchronously updating the weight rules and the data cube structure to form a full-link closed-loop iteration includes:

[0088] Receive real-time IoT data streams, including logistics trajectories and quality inspection reports; and read the processing records and results of past events in the historical case library;

[0089] The handling logic is defined as delivery delay, quality defect, and risk event. For delivery delay, if the delay time exceeds the preset delay time threshold, liquidated damages will be calculated and the supplier's credit score will be deducted. For quality defect, if the product defect rate exceeds the preset defect rate threshold, the supplier switching mechanism will be triggered. Risk event is to identify unplanned risks through AI anomaly detection technology.

[0090] Based on the handling logic triggered by the multimodal exception response engine, the BERT model is used to analyze the historical case library, match similar historical cases with the current event handling logic, and then push the historically optimal strategy;

[0091] Based on the impact of current events, the risk weights, cost weights, and risk weights within the dynamic weight voting model are updated using the resilience enhancement principle, and the path selection rules are recalculated;

[0092] According to the type of the current event, if there is no corresponding type in the data cube dimension, the data cube dimension will be expanded; and the data cube field will be updated; the supplier credit score and corresponding threshold parameters will be updated synchronously; the current event and disposal results will be stored in the historical case library to form a closed-loop iteration.

[0093] The supplier price management method based on big data analysis is implemented based on the supplier price management system based on big data analysis, and includes:

[0094] S1: Collect real-time core supplier data through standardized interfaces and obtain auxiliary data through third-party APIs to build a multi-dimensional data cube. Divide supplier clusters by geographic grid, dynamically calculate grid availability indexes, form a dynamic spatiotemporal grid benchmark library, and simultaneously generate a supply chain topology map.

[0095] S2: Based on a data cube and dynamic spatiotemporal grid benchmark library, it receives and quantifies business objective instructions and selects paths through a rule engine. It uses a dynamic weighted voting model to combine path selection results and outputs procurement strategy recommendations with confidence ratings and a dynamic prioritized supplier list.

[0096] S3: Based on a dynamic priority supplier list and supply chain topology map, combined with ERP contract status and supplier credit scores, permissions are dynamically adjusted through the ABAC model, triggering the smart contract engine to generate orders or bargaining invitations. Through parallel approval through the workflow engine, a purchase order set and permission change log are generated.

[0097] S4: Receives real-time IoT data streams and historical case libraries, triggers handling logic through a multimodal exception response engine; pushes historical optimal strategies based on matching similar historical cases with the BERT model, and simultaneously updates weight rules, data cube structures, and supplier credit score thresholds to form a full-link closed-loop iteration.

[0098] The technical effects and advantages of the supplier price management system and method based on big data analysis of the present invention are as follows:

[0099] The present invention integrates supplier core data (such as quotations, inventory), external data (such as regional policies, competitor prices) and historical case libraries to construct a multi-dimensional cube and dynamic spatiotemporal grid benchmark library, evaluate regional risks and path dependencies in real time, and solve the problem of lagging static analysis in traditional systems; then, through fault tree analysis, path risks are quantified, weights are dynamically adjusted (such as increasing market weights in emergency situations), and procurement strategies with confidence ratings are generated, giving priority to matching low-risk suppliers. The decision-making dimensions are more comprehensive and better than traditional manual price comparisons; then, based on the ABAC model and smart contracts, order generation, secondary supplier negotiation and dynamic adjustment of permissions (such as parallel approval of emergency orders by three departments) are automatically triggered, reducing manual intervention; finally, real-time responses to logistics and quality inspection anomalies (such as automatic credit deductions and supplier switching) are implemented, the optimal strategy is pushed through historical case matching using the BERT model, and data dimensions are dynamically expanded to form a "data-analysis-decision-optimization" closed loop, continuously improving the system's adaptability; the present invention realizes the automation, intelligence and self-evolution of supplier price management, improves procurement efficiency, cost control and risk resistance, and ensures supply chain stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0100] Figure 1 Schematic diagram of the supplier price management system based on big data analysis of the present invention;

[0101] Figure 2 This is a flow chart of a dynamic weighted voting model for a supplier price management system based on big data analysis according to the present invention;

[0102] Figure 3 Schematic diagram of the supplier price management method based on big data analysis of the present invention. DETAILED DESCRIPTION

[0103] 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 described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0104] Example 1

[0105] See also Figure 1 and Figure 2 As shown, the supplier price management system based on big data analysis in this embodiment includes:

[0106] Multidimensional data cube construction unit: This unit collects real-time core supplier data through standardized interfaces and obtains auxiliary data through third-party APIs to construct a multidimensional data cube. It also divides suppliers into clusters based on geographic grids, dynamically calculates grid availability indexes, forms a dynamic spatiotemporal grid benchmark library, and simultaneously generates a supply chain topology map.

[0107] Multi-path decision-making collaboration unit: Based on a data cube and dynamic spatiotemporal grid benchmark library, it receives and quantifies business objective instructions and selects paths through a rule engine. It uses a dynamic weighted voting model to combine path selection results to output procurement strategy recommendations with confidence ratings and a dynamic priority supplier list.

[0108] Smart Contract Execution and Permission Network Unit: Based on a dynamic priority supplier list and supply chain topology map, combined with ERP contract status and supplier credit scores, permissions are dynamically adjusted through the ABAC model, triggering the smart contract engine to generate orders or bargaining invitations. Through parallel approval through the workflow engine, a purchase order set and permission change log are generated.

[0109] Closed-loop autonomous optimization and resilience enhancement unit: Receives real-time IoT data streams and historical case libraries, triggers handling logic through a multimodal exception response engine; matches similar historical cases based on the BERT model to push historical optimal strategies, and simultaneously updates weight rules, data cube structures, and supplier credit score thresholds to form a full-link closed-loop iteration.

[0110] Collect real-time core supplier data through standardized interfaces (such as ERP systems, IoT sensors, and supplier contract systems), including quotes (including unit price, quantity, currency unit, discount rules, and effective date. For example, Supplier A might quote a certain electronic component at "$2.50 USD per thousand units." Some quotes also include volume discounts or special offers based on long-term contracts), inventory levels (including real-time inventory quantity, available inventory, in-transit inventory, and minimum / maximum inventory thresholds, indicating the actual inventory level of a product in the warehouse at a specific point in time, usually expressed in quantity (such as number of pieces or tons)), actual production capacity, and contract status (including contract number, contracting party, effective date, expiration date, payment terms, performance status, and breach of contract clauses, covering the validity, term, and rights and obligations of both parties).

[0111] Obtain auxiliary data on the region through third-party APIs (such as government policy databases, competitor price monitoring APIs, and logistics event monitoring systems), including regional power restriction policies (including policy name, effective date, duration, impact range (i.e., percentage of affected areas, such as provincial vs. county levels), competitor prices (including competitor product models, prices, promotions, and historical price trends, which refers to the price levels set by competitors for similar products or services in the market. These data can help companies understand market pricing trends and formulate their own pricing strategies. In addition to direct price information, some data also include promotions and discount information), design production capacity, and logistics disruption events (including event type, affected area, duration, and recovery status). It records various abnormal situations that occur during the logistics process, such as road closures caused by natural disasters, port strikes, and transportation delays caused by traffic accidents. This type of data is crucial for assessing supply chain risks and can help companies prepare countermeasures in advance to reduce losses. It also adopts a layered caching mechanism to update core data in real time and regularly update auxiliary data (for example, it is obtained once every 7 days and cached for 7 days after each acquisition). It should be noted that auxiliary data is low-frequency data and relies on third-party APIs (such as government announcements and logistics platforms). There is a certain probability of delays or incomplete data, so it cannot be updated in real time. It needs to be distinguished from the core data that is updated in real time. Through regular caching and updating, the real-time nature of the data can be guaranteed.

[0112] For collected core and auxiliary data, data cleaning and unification are performed, including cleaning outliers (such as sudden price changes), filling missing values ​​(such as using industry averages through the mean / difference method), deduplication, aligning timestamps, and deploying lightweight adapter clusters (such as Kafka-based message queues for real-time processing of core data and Redis for caching auxiliary data). Data formats are unified (for example, core data is accessed through a real-time streaming pipeline (Kafka), with fields including supplier ID, quote, inventory value, and fulfillment rate; auxiliary data is accessed through a batch loading channel (FTP / SFTP), with fields including region code, policy level, and average price of competing products). This includes converting different currencies (such as USD and CNY) into a unified unit (such as USD) using a real-time exchange rate API and converting date fields (effective date, expiration date, and event occurrence time) into a unified format (such as YYYY-MM-DD).

[0113] Define the dimensions and metrics of the data cube. The dimensions include the time dimension (the time attribute of the data, used for trend analysis, year (2025) → quarter (Q2) → month (April) → day (16th), the supplier dimension (supplier attributes, used to evaluate supplier performance, supplier ID → supplier name → industry → contract status (effective / expired / defaulted)), the administrative region dimension (geographic location, used to analyze regional risks and the impact of logistics disruptions, country (China) → province (Sichuan Province) → county (Chengdu City) → policy impact range (provincial / county level)), and the product dimension (product attributes, used to analyze quotations). Inventory relevance (product model → product category (e.g., chemical raw materials, electronic components) → product specifications (e.g., PVC-A1)), event dimension (external event attributes, used to calculate regional risk, event type (power outage, logistics disruption) → event severity (high / medium / low) → recovery status (not recovered / recovered)), and contract dimension (key contract attributes, used to assess performance risk, contract number → contracting party → payment terms (30% advance payment) → breach of contract clause (5% / day overdue penalty)). Metrics include average quote price, inventory availability, capacity utilization, regional risk factor, and contract breach risk index.

[0114] Calculate metrics for different administrative regions based on each administrative region dimension, including:

[0115] Extract the unit price, discount rules, and currency unit of the quotation data from the core data, and calculate the average quotation price;

[0116] The average price is calculated by extracting the discount rate from the discount rules, multiplying the unit price by (1-discount rate), and then comparing the resulting value with the currency conversion factor to obtain the quoted price. The currency conversion factor is the exchange rate ratio between different currencies (for example, if the unified currency format is USD and the real-time conversion ratio between CNY and USD is 7.308, then 1 CNY divided by 7.308 equals 0.1368 USD).

[0117] Extract the available inventory and the maximum inventory threshold of the inventory level in the core data, and calculate the ratio to obtain the inventory availability rate;

[0118] Take the expiration date and default clause of the contract status and calculate the contract default risk index;

[0119] The calculation method is to subtract the current date from the due date to get the remaining days. The remaining days are then multiplied by the penalty rate in the breach clause to get the contract default risk index.

[0120] Based on the processed auxiliary data, set a time window (e.g., 10 days, 30 days), and using the current moment as the benchmark, calculate the ratio of the difference between the maximum and minimum prices of the competitor product within the past time window to the average price of the competitor product as the price fluctuation range of the competitor product;

[0121] Extract the number, duration, and impact range of regional power curtailment policies within a past time window from auxiliary data, perform normalization processing (normalize to the range of 0-1 to eliminate dimensional differences), and then perform weighted summation to obtain the power curtailment risk value;

[0122] The price fluctuation range of competing products, the power restriction risk value, and the number of logistics interruption events in the past time window are used as coefficient indicators to calculate the administrative region risk coefficient. Each coefficient indicator is normalized and weighted using the entropy weight method. The regional risk coefficient is then calculated through weighted summation.

[0123] Among them, the method of assigning weights to each coefficient indicator through the entropy weight method is:

[0124] Collect historical data for each coefficient indicator (such as regional risk event data for the past three years) and calculate the information entropy value of each indicator using the information entropy formula of the entropy weight method. The lower the entropy value, the greater the difference in the indicator and the higher the weight. The weight calculation formula of the entropy weight method is used to calculate the weight of each coefficient indicator based on the entropy value;

[0125] Based on the defined dimensions and metrics, a star schema is used to construct a dimension table for each dimension (for example, the dimension table for the time dimension is Year (2025) → Quarter (Q2) → Month (April) → Day (16th)). Event types (such as power restrictions and logistics disruptions) are added to each dimension table. Event dimensions are then associated with different analysis scenarios (such as power restrictions and logistics disruptions) by event type. A primary key (such as TimeKey) and descriptive fields (such as country and supplier name) are added to each dimension table.

[0126] Assign a unique ID to each dimension, which serves as the foreign key for each dimension table. All foreign keys are integrated to build a fact table, which is then linked to the dimension table through the foreign keys. All metrics are then aggregated by time and region (i.e., the hierarchy is calculated by multiplying time by region, e.g., the quarterly average quote price in East China is the average of all quote prices in East China over a quarter (three months)).

[0127] Based on the dimension table and fact table, combine the dimensions and measures into a cube structure as a data cube. For example, Cube[Time=2021-04][Region=Sichuan][Supplier=Company A][Product=PVC-A1][Event=Power Restriction][Contract=CT2025001] contains measure values ​​(such as average bid price and risk factor).

[0128] Based on the dimensions of the data cube, the supplier dimension, product dimension, and event dimension are business-associated by product type and event type. Different dynamic regions are divided according to business associations (for example, using ArcGIS tools, the total area is divided into different spatial units based on different business-associated regions, with the GIS grid as the smallest unit), forming different supplier clusters (for example, East China - electronic components cluster, South China - logistics hub area);

[0129] It should be noted that the division of dynamic regions is updated in real time and can be aggregated into large regions (such as the East China Industrial Cluster) or differentiated into small grids (such as the logistics hub of a city's high-tech zone) as needed;

[0130] Extract all data covered by the dynamic area from the data cube, and recalculate the regional risk coefficient corresponding to the dynamic area as the dynamic regional risk coefficient;

[0131] The ratio of the total actual output value of all enterprises in the dynamic area to the total designed capacity is taken as the capacity utilization rate of the dynamic area;

[0132] Extract the contract performance status of all contracts in the dynamic area and calculate the contract performance rate in the dynamic area;

[0133] One method for calculating the contract fulfillment rate is: by analyzing the fulfillment status of all contracts in the dynamic area, the ratio of fulfillment completion to fulfillment failure is taken as the contract fulfillment rate;

[0134] Based on the capacity utilization rate, inventory availability rate, and contract fulfillment rate within the dynamic region, weights are assigned to the capacity utilization rate, inventory availability rate, and contract fulfillment rate (e.g., weights are set based on industry standards or expert experience). The capacity utilization rate, inventory availability rate, and contract fulfillment rate are multiplied by the corresponding weights, and the sum is divided by the sum of the weights to obtain a weighted average. The weighted average is then multiplied by the complementary value of the dynamic region risk coefficient to obtain a grid availability index. This grid availability index can reflect the resource status within the dynamic region.

[0135] The relationship is:

[0136] ;

[0137] in, represents the capacity utilization rate, represents the capacity utilization weight, represents the inventory availability rate, Inventory availability weight, represents the contract fulfillment rate, represents the weight of the contract fulfillment rate, represents the dynamic regional risk coefficient;

[0138] It should be noted that the integration of capacity utilization, inventory availability, and contract fulfillment rate comprehensively reflects the dynamic availability of the supply chain, avoiding the one-sidedness of a single indicator. The indicator priorities can be adjusted for different business scenarios through weighting (e.g., the logistics hub area focuses on inventory weighting, and the production area focuses on capacity weighting).

[0139] Since the higher the risk factor of a dynamic region (e.g., power outages or logistics disruptions), the lower the availability of the supply chain, the availability index of high-risk regions will be significantly reduced by multiplying the compensation value of the dynamic region risk factor. This directly reflects the inhibitory effect of risk on the supply chain (e.g., availability in power-limited regions is directly discounted). In addition, the dynamic region risk factor can be updated in real time based on dynamic region updates (e.g., sudden logistics disruptions). The formula result automatically reflects risk changes, supporting rapid decision-making.

[0140] This calculation method achieves a dynamic, comprehensive and explainable evaluation of supply chain availability through weighted comprehensive evaluation + risk suppression while ensuring calculation efficiency. It is particularly suitable for supply chain optimization and risk warning in grid management scenarios.

[0141] Integrate the grid availability index and dynamic regional risk coefficient of the dynamic region as the dynamic space-time grid benchmark of the dynamic region; integrate the dynamic space-time grid benchmarks of all dynamic regions to generate a dynamic space-time grid benchmark library;

[0142] According to the dimensions of the data cube, all suppliers, products, dynamic regions, and events are treated as nodes, and corresponding attributes are added to each node. The supplier node is bound to the dynamic region node. If the dynamic region risk coefficient changes, the supplier's grid availability index is triggered to update synchronously.

[0143] Supplier attributes include ID, name, dynamic region, contract status, and capacity utilization. Product attributes include model, supplier, inventory status, and region. Dynamic region attributes include geographic location, dynamic region risk factor, logistics efficiency, and capacity load. Event attributes include type (such as power rationing), impact range (dynamic region), and timestamp.

[0144] Extract all supply relationships, logistics paths, contract associations, and event impacts in the data cube as edges, and assign edge weights to each type of edge;

[0145] The relationship between suppliers and products is regarded as the supply relationship. The capacity utilization rate of each supplier is calculated as the supplier capacity utilization rate. The product of the supplier capacity utilization rate and the dynamic regional risk coefficient is then taken as the basic weight of the corresponding edge.

[0146] The transportation route between two dynamic areas is used as the logistics path, and the corresponding transportation cost is obtained. The product of the dynamic area risk coefficient and the transportation cost is taken as the basic weight of the corresponding edge;

[0147] The relationship between the supplier and the contract is regarded as the contractual association, and the contract default risk index between the supplier and the corresponding contract is taken as the basic weight of the corresponding edge;

[0148] The relationship between the event and the affected area is taken as the event impact. Based on the data cube, the total external dependence is calculated using the TFR (total external dependence) formula to obtain the total external dependence as the industrial chain dependence. The dynamic regional risk coefficient and the industrial chain dependence are weighted and summed to obtain the dynamic regional vulnerability. The product of the impact range of the event and the dynamic regional vulnerability is taken as the basic weight of the corresponding edge.

[0149] Among them, the calculation formula of TFR is TFR equal to the average of import dependence (FIR) and export dependence (FMR);

[0150] Normalize the basic weights of all edges so that the sum of all basic weights is 1;

[0151] Multiply the normalized basic weight by the global adjustment coefficient of the dynamic regional risk coefficient to obtain the edge weight;

[0152] The global adjustment coefficient of the dynamic regional risk coefficient is calculated as follows: the weighted sum of the dynamic regional risk coefficient and the contract default risk index is added to 1 to obtain the global adjustment coefficient;

[0153] Based on nodes and edges, a supply chain topology map is constructed. Visualization tools (such as Tableau) are used to display node relationships and risk transmission paths (e.g., "power restrictions in Sichuan → reduced production capacity of supplier A → shortage in Guangdong"). The risk transmission path is used to identify high-risk nodes and their impact range. High-risk nodes can be defined as nodes with a corresponding dynamic regional risk coefficient higher than a preset threshold. When the risk in a certain area is too high, a cascade supplier is recommended (e.g., activating backup supplier B).

[0154] Using the SMART-based KPI hierarchical decomposition method, business objectives are broken down into different KPI indicators, including cost path indicators, market path indicators, and risk path indicators.

[0155] It should be noted that the business goal instructions received must follow the SMART principle to ensure that the goals are achievable. The SMART principle includes specific, measurable, achievable, relevant and time-bound;

[0156] Specific: Clarify the target content. For example, "reduce total supply chain costs" should be refined into "reduce logistics costs in East China by 10%."

[0157] Measurable: Quantifiable goals, such as "Increase on-time delivery from 85% to 95%."

[0158] Achievable: Combining resources and capabilities, such as "reducing costs by optimizing logistics routes rather than building new warehouses";

[0159] Relevant: Aligned with corporate strategy. For example, minimizing risk must be linked to improving supply chain resilience.

[0160] Time-bound: Set a timeframe, such as “Complete cost optimization within Q3”;

[0161] The exemplary decomposition process of the KPI hierarchical decomposition method for decomposing business objectives is as follows:

[0162] Business goal (strategic level): Reduce total supply chain costs by 10% (annual goal);

[0163] Specific goals (tactical level): Cost path: Optimize logistics routes to reduce transportation costs by 5%, and reduce inventory holding costs by 3% through inventory sharing; Market path: Improve on-time delivery rate to 95% and increase demand matching by 10%; Risk path: Reduce the probability of regional cascading failure to below 0.1;

[0164] KPI indicators (execution level): Cost path: transportation cost / inventory cost (specific values); Market path: on-time delivery rate (percentage), demand matching (index); Risk path: dynamic regional risk coefficient, path dependence;

[0165] Quantify different KPI indicators, including:

[0166] The cost path indicator is quantified as follows: the core indicators of the cost path indicator are defined as transportation cost and inventory cost;

[0167] Using the data cube and supply chain topology map, query all paths from the dynamic area where the starting point is located to the dynamic area where the end point is located (for example, from dynamic area 1 to dynamic area 2). For each path, obtain the logistics path edge weight of each segment on the path as the unit transportation cost of the segment (for example, if path 1 passes through dynamic areas 1-2-3, the logistics path edge weight of 1-2 is used as the unit cost of segment 1-2, and the logistics path edge weight of 2-3 is used as the unit cost of segment 2-3). Take the sum of the unit transportation costs of all segments on the path as the transportation cost of the path;

[0168] Use the data cube to obtain the available inventory at each inventory point within the dynamic region that the path passes through, and simultaneously obtain the unit holding cost of the corresponding inventory point (for example, the storage fee is ¥5 / item / day). For each inventory point, the product of the available inventory and the unit holding cost is taken as the single inventory cost of the inventory point. The sum of the single inventory costs of all inventory points along all paths is taken as the inventory cost of the path.

[0169] The transportation cost and inventory cost of each path are weighted and summed to obtain the total cost of each path, which is used as the quantified cost path indicator. (The weights in the weighted summation process can be dynamically adjusted. For example, if cost is a higher priority, the transportation or inventory weight can be increased (such as a transportation weight of 0.7 and an inventory weight of 0.3). If inventory risk is greater, the inventory weight can be increased to prioritize inventory cost control. The weights can be dynamically adjusted based on real-time data (for example, if the dynamic regional risk factor increases, the transportation weight can be reduced to avoid the risk area).

[0170] The market path indicator is quantified as follows: The core indicators of the market path indicator are defined as on-time delivery rate and market demand matching;

[0171] Extract historical order data from the data cube and calculate the ratio of on-time delivery orders to the total number of orders for a single route as the on-time delivery rate. Obtain market demand for the dynamic region (this is forecast data, which can be obtained through market data forecasts for the region or directly obtained through third-party APIs). Combined with the actual total production capacity within the dynamic region, calculate the ratio of the actual total production capacity to market demand to obtain the market demand matching degree.

[0172] The weighted sum of the on-time delivery rate and market demand matching is used to obtain the market value of each path, which is used as the quantified market path indicator.

[0173] The risk path indicators are quantified as follows: the core indicators of the risk path indicators are defined as the dynamic regional risk coefficient and path dependence;

[0174] By combining the fault tree analysis (FTA) with the supply chain topology map, the reliability index of each supplier node or dynamic area node that the path passes through is calculated as the path dependence of the path to the corresponding node;

[0175] The exemplary steps of the fault tree analysis method are:

[0176] The failure of supplier nodes and dynamic region nodes (such as supplier supply interruption and supplier late delivery) is taken as the top event and decomposed into underlying faults that may lead to failure (such as equipment aging and human error). Logic gates (such as AND gates and OR gates) are constructed to connect the failure events. The probability of each fault is quantified, and the total probability of node failure is calculated through probabilistic calculation as a reliability indicator.

[0177] The dynamic regional risk coefficient of each dynamic area passed by each path is obtained through the dynamic space-time grid benchmark library;

[0178] For each dynamic region node and supplier node that the path passes through, calculate the product of the dynamic region risk coefficient and the path dependency of the path on the node. Add up the product results of all dynamic regions that the path passes through to obtain the basic risk value of the path. Multiply the basic risk value by a preset risk weight (such as 0.3 or 0.5, which is dynamically adjusted based on business objectives (e.g., a higher weight in a risk-priority scenario)) to obtain the path risk value, which serves as a quantified risk path indicator.

[0179] It should be noted that the preset risk weights reflect the priority of business objectives. The path risk value is the same as the path risk at the end of the month. The lower the path risk, the lower the path risk is. To avoid the risk of cascading failure, the path with the lowest risk value is selected.

[0180] Construct a multi-objective optimization model to perform multi-objective optimization on the quantified cost path indicators, market path indicators, and risk path indicators to balance the conflicts among cost, market, and risk;

[0181] The objective function of the multi-objective optimization model is:

[0182] ;

[0183] The constraints are defined as logistics path capacity limitations and supplier availability;

[0184] Generate Pareto optimal solution sets through genetic algorithms, balance conflicts among cost, market, and risk, and adjust corresponding weights in real time according to business objectives;

[0185] When the regional risk coefficient changes, the path risk value is recalculated and optimization is triggered; when a logistics interruption occurs, the path weight is dynamically adjusted;

[0186] For cost path selection, the rule is defined as selecting the path with the lowest total cost;

[0187] For market path selection, the rule is defined as choosing the path with the highest market value;

[0188] For risk path selection, the rule is defined as selecting the path with the lowest path risk value;

[0189] Taking the selection results of cost path, market path and risk path as input, a dynamic weighted voting model is used to output procurement strategy recommendations with confidence ratings and a dynamic priority supplier list;

[0190] The dynamic weight allocation rule is defined as dynamically adjusting the path weight by the edge weight adjustment coefficient, and the total path weight is 1;

[0191] The dynamic adjustment mechanism is defined as follows: if the dynamic area risk coefficient increases and exceeds the preset dynamic area risk coefficient threshold (such as 0.2), the risk weight will be increased by a preset ratio. If an emergency order instruction is received, the market weight will be increased by a preset ratio (such as 0.3) and the cost weight will be reduced accordingly.

[0192] Based on the selection results of the cost path, market path and risk path, the total cost, market value and path risk value of the path are normalized respectively, and the normalized results are used as the cost score (range 0-1, the lower the cost, the higher the score), the market score (range 0-1, the higher the market value, the higher the score) and the risk score (range 0-1, the lower the risk value, the higher the score);

[0193] The cost score, market score, and risk score are combined with the weights of the corresponding path indicators, and the comprehensive score of the path is obtained through weighted summation;

[0194] The confidence rating generation logic is defined as follows: the confidence rating classification criteria are defined as high confidence, medium confidence, and low confidence;

[0195] Set the score interval threshold (e.g., [0.8, 0.5]), the cost interval threshold (e.g., 0.7), and the risk interval threshold (e.g., 0.7);

[0196] If the comprehensive score of the path is greater than or equal to the maximum value of the score interval threshold, and the cost score is greater than or equal to the maximum value of the cost interval threshold and the risk score is greater than or equal to the maximum value of the risk interval threshold, then the path is judged to be high confidence;

[0197] If the comprehensive score of the path is greater than or equal to the minimum value of the score interval threshold and less than the maximum value of the score interval threshold, and at the same time satisfies the conditions that the cost score is greater than or equal to the minimum value of the cost interval threshold and the risk score is greater than or equal to the minimum value of the risk interval threshold, then the path is determined to have medium confidence;

[0198] If the comprehensive score of the path is less than the minimum value of the score interval threshold, or the cost score is less than the minimum value of the cost interval threshold, or the risk score is less than the minimum value of the risk interval threshold, the path is judged to be of low confidence;

[0199] For each supplier node that the path passes through, extract the contract performance of all suppliers corresponding to the supplier node, and take the ratio of contract fulfillment to contract failure as the supplier contract fulfillment rate;

[0200] The path dependency of the path to the supplier node, the available inventory of the supplier node, and the supplier contract fulfillment rate are weighted and summed to obtain the supplier priority score of each supplier;

[0201] Arrange each supplier in descending order based on their supplier priority score into a dynamic priority supplier list;

[0202] Set a risk area threshold. If the dynamic area risk coefficient of a dynamic area is less than the risk area threshold, it is determined to be a low-risk area. If the dynamic area risk coefficient of a dynamic area is greater than or equal to the risk area threshold, it is determined to be a high-risk area.

[0203] The dynamic adjustment logic of the priority sorting rules for the dynamic priority supplier list is defined as follows: for urgent orders, suppliers with high market value paths are prioritized (for example, supplier B has a high market match), and for high-risk areas, suppliers in low-risk areas are prioritized (for example, supplier A is located in a low-risk area);

[0204] For example, the output dynamic priority supplier list is: priority 1 is supplier 1 (path 1, dynamic regional risk factor 0.2, sufficient inventory); priority 2 is supplier 3 (path 2, low cost but dynamic regional risk factor 0.4); priority 3 is supplier 5 (path 3, high-risk area, backup supplier);

[0205] The output format of the defined procurement strategy recommendations is:

[0206] High-confidence strategy: recommended path (Path 1 (lowest cost and controllable risk)), confidence rating (high confidence (confidence 0.9)), implementation recommendation (prefer supplier 1, for example, cost savings of 15% and risk reduction of 20%));

[0207] Alternative strategy: Path 2 (medium confidence, high market value but high cost);

[0208] A visual decision-making aid multi-path comparison chart can be constructed:

[0209] Use Tableau or GIS tools to display the total cost, market value, and risk value comparison of the path;

[0210] High-risk paths are marked in red, and high-market-value paths are marked in green;

[0211] Through Pareto frontier analysis, the optimal set of cost-risk solutions is displayed to assist in trade-off decision-making (e.g., "Path 1 is the Pareto optimal solution with the lowest cost and lowest risk");

[0212] Based on the dynamic priority supplier list and supply chain topology map, obtain each supplier's ERP contract status (including the current contract validity period, used credit limit, and payment terms), as well as the supplier's credit score (based on the supplier's historical performance (such as on-time rate, quality compliance rate)) and grade (such as level 1, level 2);

[0213] Define ABAC model rules including attributes and policy rules. Define attributes including user attributes, environment attributes and operation attributes.

[0214] User attributes include supplier level and credit score (e.g., ≥80 for high credit); environmental attributes include current status and grid availability index (e.g., <0.3 for low availability); operational attributes include request type (e.g., order generation, price negotiation, data access); current status includes normal status and emergency status;

[0215] Define policy rules based on attributes, including defining access rights for different levels of suppliers based on their current state;

[0216] An example policy rule defines that, under normal circumstances, first-tier suppliers can access ERP contracts and initiate orders, while second-tier suppliers can only view public quotation templates and cannot initiate orders. In emergency situations, second-tier suppliers can temporarily grant contract permissions (for example, valid for 24 hours), but hide high-risk data (such as inventory levels and dynamic regional risk factors).

[0217] Based on the defined ABAC model, the smart contract engine's ABAC module generates the supplier's access control list (ACL, which contains the supplier's access rights (such as order generation, bargaining invitation) and effectiveness conditions) and records the permission change log;

[0218] Define the triggering rules of the smart contract engine, including the regular procurement process and the reverse link contract;

[0219] Define the regular procurement process to automatically generate an order when the ERP contract status is valid and the supplier is in the dynamic priority supplier list;

[0220] The execution steps are defined as selecting the highest priority supplier based on the dynamic priority supplier list, calling the ERP interface through the smart contract engine, and automatically generating an order based on the historical order template with thresholds (such as product quantity, price, and delivery time);

[0221] The order information is synchronized to the ERP system, and the contract status (such as used credit + order amount) is updated. The order hash value is recorded on the blockchain to ensure that it cannot be tampered with.

[0222] The reverse link contract is defined as initiating price negotiation with the secondary supplier when the grid availability index of the dynamic area where the supplier is located is lower than the preset grid availability index threshold;

[0223] The execution steps are defined as monitoring the grid availability index. If it is lower than the preset grid availability index threshold, the reverse link contract is triggered and a bargaining invitation is sent to the secondary supplier based on the preset historical quotation template, including parameters and constraints;

[0224] The parameters are demand volume, dynamic regional risk factor, and order urgency, and the constraint is that the price must be lower than the first-tier supplier's quote by a preset percentage (e.g., 120% to avoid arbitrage).

[0225] Based on the price negotiation invitation sent to the second-tier supplier, when the price negotiation invitation is accepted by the second-tier supplier or the order in the regular status requires manual confirmation, the workflow engine (such as the Power Automate process) is started;

[0226] Set approval nodes, including procurement, finance, and legal departments, and use the workflow engine to execute parallel approval operations based on the approval nodes;

[0227] Procurement Department: Review order prices and supplier priorities; Finance Department: Review budget compliance and payment terms; Legal Department: Review contract terms and risk compliance;

[0228] The parallel approval rules are defined as follows: regular orders only require approval from the purchasing department, while emergency orders or orders requiring price negotiation require parallel approval from the purchasing, finance, and legal departments.

[0229] The approval logic is: under normal circumstances, the application will take effect after approval by the purchasing department at a single node;

[0230] Emergency or bargaining orders: require unanimous approval from all three departments;

[0231] If any department's approval time expires (e.g. no response within 48 hours), the automatic urging mechanism will be triggered;

[0232] When all approvals are passed, the ERP contract status is updated (updated to being executed) through the smart contract engine, the historical order template is called, and the final purchase order set is generated based on the approval results and recorded in the log;

[0233] Real-time monitoring of events that lead to emergency situations (e.g., dynamic regional risk coefficient exceeding a threshold). Once the emergency situation is resolved (e.g., the dynamic regional risk coefficient returns to normal), the ABAC module of the smart contract engine revokes the temporary permissions of the secondary supplier, updates the supplier's access control list, and simultaneously records permission changes, approval processes, and order execution results, generating a permission change log.

[0234] It should be noted that the deployment environment of the smart contract engine can be based on a hop-count network (such as StarChain) or Ethereum, supporting parallel execution and fast confirmation. Regular orders in the trigger conditions can call the smart contract interface through the ERP system, and the reverse link can be triggered by IoT sensors or network data (such as breaking the availability index threshold).

[0235] Attributes of dynamically adjusted ABAC permissions, such as supplier credit scores and network availability indexes, are stored in blockchains or trusted databases. Policies can calculate permission decisions in real time based on the ABAC module.

[0236] These can be integrated into a workflow engine integration, using Power Automate or a custom workflow engine to support multi-departmental parallel branching and shorten approval time;

[0237] Gas limits can also be set through Gas to avoid interruptions in contract execution. If the approval process times out or there is a conflict of authority, the contract will automatically roll back and notify the administrator. High-risk data (such as inventory levels) is only visible to first-tier suppliers.

[0238] Receive real-time IoT data streams, including logistics trajectories (e.g., transportation time and location) and quality inspection reports (e.g., defect rates and batch data); and read the processing records and results of past events (e.g., delivery delays and quality defects) in the historical case library;

[0239] The handling logic is defined as delivery delay, quality defect, and risk event. For delivery delay, if the delay time exceeds the preset delay time threshold (such as 3 days), liquidated damages will be calculated and the supplier's credit score will be deducted (such as 5 points). For quality defect, if the product defect rate exceeds the preset defect rate threshold, a supplier switching mechanism (such as activating an alternative first-tier supplier or a second-tier supplier) will be triggered. Risk event is to identify unplanned risks (such as equipment failure) through AI anomaly detection technology (such as the LogBERT model).

[0240] Based on the handling logic triggered by the multimodal exception response engine, the BERT model is used to analyze the historical case library, matching similar historical cases with the current event handling logic (such as delivery delays and quality defects in a certain region), and then delivering the historically optimal strategy (for example, switching to supplier 1 results in a 100% increase in costs but a 20% decrease in risk).

[0241] Based on the impact of current events, the risk weights, cost weights, and risk weights within the dynamic weight voting model are updated using the resilience enhancement principle, and the path selection rules are recalculated (e.g., the lowest risk value is prioritized).

[0242] Based on the type of the current event, if there is no corresponding type in the data cube dimension, the data cube dimension will be expanded (for example, a cross-border policy risk dimension will be added based on the policy backup mechanism); and the data cube field will be updated (for example, a new supplier credit score change record will be added); and the supplier credit score and corresponding threshold parameters will be updated simultaneously;

[0243] Store current events and handling results in the historical case library to form a closed-loop iteration.

[0244] Example 2

[0245] See also Figure 3 As shown, for the parts not described in detail in this embodiment, please refer to the description of Example 1. A supplier price management method based on big data analysis is provided, including:

[0246] S1: Collect real-time core supplier data through standardized interfaces and obtain auxiliary data through third-party APIs to build a multi-dimensional data cube. Divide supplier clusters by geographic grid, dynamically calculate grid availability indexes, form a dynamic spatiotemporal grid benchmark library, and simultaneously generate a supply chain topology map.

[0247] S2: Based on a data cube and dynamic spatiotemporal grid benchmark library, it receives and quantifies business objective instructions and selects paths through a rule engine. It uses a dynamic weighted voting model to combine path selection results and outputs procurement strategy recommendations with confidence ratings and a dynamic prioritized supplier list.

[0248] S3: Based on a dynamic priority supplier list and supply chain topology map, combined with ERP contract status and supplier credit scores, permissions are dynamically adjusted through the ABAC model, triggering the smart contract engine to generate orders or bargaining invitations. Through parallel approval through the workflow engine, a purchase order set and permission change log are generated.

[0249] S4: Receives real-time IoT data streams and historical case libraries, triggers handling logic through a multimodal exception response engine; pushes historical optimal strategies based on matching similar historical cases with the BERT model, and simultaneously updates weight rules, data cube structures, and supplier credit score thresholds to form a full-link closed-loop iteration.

[0250] Example 3

[0251] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the operation mode of the supplier price management method based on big data analysis provided above is implemented.

[0252] Since the electronic device introduced in this embodiment is the electronic device used to implement the supplier price management method based on big data analysis in the embodiment of this application, based on the supplier price management method based on big data analysis introduced in the embodiment of this application, those skilled in the art can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application will not be described in detail here. As long as those skilled in the art implement the electronic device used by the supplier price management method based on big data analysis in the embodiment of this application, it falls within the scope of protection to be provided by this application.

[0253] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0254] 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 embodiment. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for users of ordinary skill in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. The supplier price management system based on big data analysis is characterized by: include: Multi-dimensional data cube construction unit: collects real-time core data from suppliers through standardized interfaces and obtains auxiliary data through third-party APIs to build multi-dimensional data cubes; Divide supplier clusters by geographic grid, dynamically calculate grid availability index, form a dynamic spatiotemporal grid benchmark library, and simultaneously generate a supply chain topology map; Multi-path decision-making coordination unit: Based on the data cube and dynamic spatiotemporal grid benchmark library, it receives and quantifies business target instructions and selects paths through the rule engine; Adopting a dynamic weighted voting model, combined with path selection results, it outputs procurement strategy recommendations with confidence ratings and a dynamic priority supplier list; The procurement strategy recommendations with confidence ratings and the dynamic priority supplier list are obtained in the following ways: Based on the selection results of the cost path, market path, and risk path, the total cost, market value, and path risk value of the path are normalized respectively. The cost score, market score, and risk score are used as inputs to the dynamic weighted voting model, which outputs procurement strategy recommendations with confidence ratings and a dynamic priority supplier list. The dynamic weight allocation rule is defined as dynamically adjusting the path weight by the edge weight adjustment coefficient, and the total path weight is 1; The dynamic adjustment mechanism is defined as follows: if the dynamic area risk coefficient increases and exceeds the preset dynamic area risk coefficient threshold, the risk weight will be increased by a preset ratio. If an emergency order instruction is received, the market weight will be increased by a preset ratio and the cost weight will be reduced accordingly. Combine the cost score, market score, and risk score with the corresponding weights and calculate the weighted sum to get the comprehensive score of the path; The confidence rating generation logic is defined as: the confidence rating is divided into high confidence, medium confidence and low confidence; Set scoring interval thresholds, as well as cost interval thresholds and risk interval thresholds; Based on the comprehensive score, cost score and risk score, the confidence level of the path is determined by the score interval threshold, cost interval threshold and risk interval threshold; For each supplier node in the path, extract the performance status of all contracts corresponding to the supplier node, and take the ratio of the number of completed contracts to the total number of failed contracts as the supplier contract performance rate; Based on the edge weights of the supply relationship in the supply chain topology graph, the path dependency of the path on the supplier node is extracted, and the weighted sum is performed with the available inventory of the supplier node and the supplier contract fulfillment rate to obtain the supplier priority score of each supplier; Generate a dynamic priority supplier list by sorting each supplier in descending order of supplier priority score; Set a risk area threshold. If the dynamic area risk coefficient of the dynamic area is less than the risk area threshold, it is determined to be a low-risk area. If the dynamic area risk coefficient is greater than or equal to the risk area threshold, it is determined to be a high-risk area. The dynamic adjustment logic for the priority sorting rules of the dynamic priority supplier list is as follows: in urgent order scenarios, suppliers with the highest market value are prioritized; if the path passes through a high-risk area, suppliers in a low-risk area are prioritized; Smart Contract Execution and Permission Network Unit: Based on a dynamic priority supplier list and supply chain topology map, combined with ERP contract status and supplier credit scores, permissions are dynamically adjusted through the ABAC model, triggering the smart contract engine to generate orders or bargaining invitations. Through parallel approval through the workflow engine, a purchase order set and permission change log are generated. Closed-loop autonomous optimization and resilience enhancement unit: Receives real-time IoT data streams and historical case libraries, triggers handling logic through a multimodal exception response engine; matches similar historical cases based on the BERT model to push historical optimal strategies, and simultaneously updates weight rules, data cube structures, and supplier credit score thresholds to form a full-link closed-loop iteration.

2. The supplier price management system based on big data analysis according to claim 1 is characterized in that: The multidimensional data cube is constructed in the following manner: Collect real-time core data from suppliers through standardized interfaces, including quotes, inventory levels, actual production capacity, and contract status; Obtain auxiliary data about the region through third-party APIs, including regional power restrictions, competitive product prices, designed production capacity, and logistics disruptions. Clean and unify the collected core data and auxiliary data; Define the dimensions and metrics of the data cube. Dimensions include time, supplier, administrative region, product, event, and contract. Metrics include average quote price, inventory availability, capacity utilization, regional risk factor, and contract default risk index. Based on the processed auxiliary data, a time window is set. Taking the current moment as the benchmark, the price fluctuation range of the competing product is calculated as the ratio of the difference between the maximum and minimum prices of the competing product in the past time window to the average price of the competing product. Extract the number, duration, and impact range of regional power restriction policies within a past time window from the auxiliary data, perform normalization processing on them, and obtain the power restriction risk value; The price fluctuation range of competing products, the power restriction risk value, and the number of logistics interruption events in the past time window are used as coefficient indicators to calculate the administrative region risk coefficient. Each coefficient indicator is normalized to obtain the regional risk coefficient. According to the defined dimensions and measures, dimension tables and fact tables are constructed through the star model, and foreign keys are integrated to form a cube structure as a data cube.

3. The supplier price management system based on big data analysis according to claim 2 is characterized in that: The construction method of the dynamic spatiotemporal grid benchmark library includes: Based on the dimensions of the data cube, the supplier dimension, product dimension, and event dimension are business-associated by product type and event type, and different dynamic areas are divided according to the business associations as different supplier clusters; Extract all data covered by the dynamic area from the data cube, and re-obtain the regional risk coefficient corresponding to the dynamic area as the dynamic regional risk coefficient; The ratio of the total actual output value of all enterprises in the dynamic area to the total designed capacity is taken as the capacity utilization rate of the dynamic area; Extract the contract performance status of all contracts in the dynamic area and obtain the contract performance rate in the dynamic area; Based on the capacity utilization rate, inventory availability rate and contract fulfillment rate in the dynamic area, weights are assigned to the capacity utilization rate, inventory availability rate and contract fulfillment rate respectively, and then combined with the dynamic area risk coefficient to obtain the grid availability index; The grid availability index and the dynamic regional risk coefficient of the dynamic region are integrated as the dynamic space-time grid benchmark of the dynamic region; the dynamic space-time grid benchmarks of all dynamic regions are integrated to generate a dynamic space-time grid benchmark library.

4. The supplier price management system based on big data analysis according to claim 3 is characterized in that: The supply chain topology map is constructed in the following ways: Based on the dimensions of the data cube, all suppliers, products, dynamic regions, and events are treated as nodes, and corresponding attributes are added to each node. The supplier node is bound to the dynamic region node. If the dynamic region risk coefficient changes, the supplier's grid availability index is synchronously triggered to update. Extract all supply relationships, logistics paths, contract associations, and event impacts in the data cube as edges, and assign edge weights to each type of edge; The relationship between suppliers and products is regarded as the supply relationship. The supplier's capacity utilization rate is calculated as the supplier's capacity utilization rate, and the product of the supplier's capacity utilization rate and the corresponding dynamic regional risk coefficient is taken as the basic weight of the corresponding edge. The transportation route between two dynamic areas is used as the logistics path, the corresponding transportation cost is obtained, and the product of the dynamic area risk coefficient and the transportation cost is taken as the basic weight of the corresponding edge; The relationship between the supplier and the contract is regarded as the contractual association, and the contract default risk index is taken as the basic weight of the corresponding edge; The relationship between the event and the affected area is taken as the event impact. The industrial chain dependence of the dynamic area is calculated using the TFR formula. The dynamic area risk coefficient and the industrial chain dependence are weighted and summed according to a preset ratio to obtain the dynamic area vulnerability. The product of the impact range of the event and the dynamic area vulnerability is taken as the basic weight of the corresponding edge. Normalize the basic weights of all edges so that the sum of all basic weights is 1; Multiply the normalized basic weight by the global adjustment coefficient of the dynamic regional risk coefficient to obtain the edge weight; Based on the nodes and edges, a supply chain topology map is constructed.

5. The supplier price management system based on big data analysis according to claim 4 is characterized in that: The definition method of the rule engine selection path includes: Decompose business goals into cost path indicators, market path indicators and risk path indicators through KPI decomposition based on the SMART principle; The cost path indicator is calculated by calculating the weighted sum of transportation cost and inventory cost to obtain the total cost of each path, which is used as the cost path quantification; The market path indicator is the weighted sum of the on-time delivery rate and the market demand match, which is used to quantify the market path. The risk path indicator is to calculate the path risk value of each path as the risk path quantification; The reliability index of each supplier node or dynamic area node that the path passes through is calculated by the fault tree analysis method as the path dependence of the path to the corresponding node; The dynamic regional risk coefficient of each dynamic area passed by the path is obtained through the dynamic space-time grid benchmark library; Calculate the sum of the product of the dynamic regional risk coefficient and the path dependence to obtain the basic risk value of the path, and multiply the basic risk value by the preset risk weight to obtain the path risk value; Based on the quantified cost path indicators, market path indicators, and risk path indicators, the path selection rules of the rule engine are defined as follows: The cost path is defined as choosing the path with the lowest total cost; The market path is defined as choosing the path with the highest market value; The risk path is defined as the path with the lowest path risk value.

6. The supplier price management system based on big data analysis according to claim 5 is characterized in that: The smart contract implementation methods for generating orders or bargaining invitations include: Obtain the supplier's ERP contract status, credit score, and rating through the supply chain topology map and ERP system interface; Obtain the grid availability index of the dynamic area where the supplier is located from the dynamic spatiotemporal grid benchmark library; The definition of ABAC model includes attributes and policy rules. The defined attributes include user attributes, environment attributes and operation attributes. User attributes include supplier level and credit score; environmental attributes include current status and grid availability index; operation attributes include request type; current status includes normal status and emergency status; Define policy rules based on attributes, including defining access rights for different levels of suppliers based on their current state; Based on the defined ABAC model, the ABAC module of the smart contract engine generates the supplier's access control list and records the permission change log; Define the triggering rules of the smart contract engine, including the regular procurement process and the reverse link contract; Define the triggering condition for the regular procurement process as when the ERP contract status is valid and the supplier is in the dynamic priority supplier list; The execution steps are defined as selecting the highest-priority supplier in the dynamic priority supplier list, calling the ERP interface of the smart contract engine, and automatically generating an order based on the preset historical order template; The triggering condition of the reverse link contract is defined as if the grid availability index of the dynamic area where the supplier is located is lower than a preset grid availability index threshold; The execution steps are defined as sending a negotiation invitation to the second-tier supplier based on the preset historical quotation template, including parameters and constraints; The parameters are demand volume, dynamic regional risk factor, and order urgency, and the constraint is that the bargaining price must be lower than the preset percentage of the first-tier supplier's quotation.

7. The supplier price management system based on big data analysis according to claim 6 is characterized in that: The generation method of the purchase order set and the permission change log includes: Based on the price negotiation invitation sent to the second-tier supplier, when the second-tier supplier accepts the price negotiation invitation, or when the order in the regular status requires manual confirmation, the parallel approval workflow engine is started; Set approval nodes, including procurement, finance, and legal departments, and use the workflow engine to execute parallel approval operations based on the approval nodes; Define parallel approval rules so that normal orders only require approval from the purchasing department, while emergency orders or orders requiring price negotiation require parallel approval from the purchasing, finance, and legal departments. When all approvals are passed, the ERP contract status is updated through the smart contract engine, the historical order template is called, and the final purchase order set is generated based on the approval results; When the emergency state is lifted, the temporary permissions of the secondary supplier are revoked through the ABAC module of the smart contract engine, and the supplier's access control list is updated, the normal status permission rules are restored, the permission change time, approval process and order execution results are recorded synchronously, and a permission change log is generated.

8. The supplier price management system based on big data analysis according to claim 7 is characterized in that: The method of synchronously updating the weight rules and the data cube structure to form a full-link closed-loop iteration includes: Receive real-time IoT data streams, including logistics trajectories and quality inspection reports; and read the processing records and results of past events in the historical case library; The handling logic is defined as delivery delay, quality defect, and risk event. For delivery delay, if the delay time exceeds the preset delay time threshold, liquidated damages will be calculated and the supplier's credit score will be deducted. For quality defect, if the product defect rate exceeds the preset defect rate threshold, the supplier switching mechanism will be triggered. Risk event is to identify unplanned risks through AI anomaly detection technology. Based on the handling logic triggered by the multimodal exception response engine, the BERT model is used to analyze the historical case library, match similar historical cases with the current event handling logic, and then push the historically optimal strategy; Based on the impact of current events, the risk weights, cost weights, and risk weights within the dynamic weight voting model are updated using the resilience enhancement principle, and the path selection rules are recalculated; According to the type of the current event, if there is no corresponding type in the data cube dimension, the data cube dimension will be expanded; and the data cube field will be updated; the supplier credit score and corresponding threshold parameters will be updated synchronously; the current event and disposal results will be stored in the historical case library to form a closed-loop iteration.

9. A supplier price management method based on big data analysis, which is implemented based on the supplier price management system based on big data analysis according to any one of claims 1 to 8, characterized in that: include: S1: Collect real-time core data from suppliers through standardized interfaces and obtain auxiliary data through third-party APIs to build a multi-dimensional data cube; Divide supplier clusters by geographic grid, dynamically calculate grid availability index, form a dynamic spatiotemporal grid benchmark library, and simultaneously generate a supply chain topology map; S2: Based on the data cube and dynamic spatiotemporal grid benchmark library, it receives and quantifies business goal instructions and selects paths through the rule engine; Adopting a dynamic weighted voting model, combined with path selection results, it outputs procurement strategy recommendations with confidence ratings and a dynamic priority supplier list; S3: Based on a dynamic priority supplier list and supply chain topology map, combined with ERP contract status and supplier credit scores, permissions are dynamically adjusted through the ABAC model, triggering the smart contract engine to generate orders or bargaining invitations. Through parallel approval through the workflow engine, a purchase order set and permission change log are generated. S4: Receives real-time IoT data streams and historical case libraries, triggers handling logic through a multimodal exception response engine; pushes historical optimal strategies based on matching similar historical cases with the BERT model, and simultaneously updates weight rules, data cube structures, and supplier credit score thresholds to form a full-link closed-loop iteration.

Images