Supply chain configuration method, equipment and product

Abstract

The invention provides a supply chain configuration method, equipment and a product. The method comprises the steps that a configuration request is received, and the configuration request is sent by any one of a regional store, a supplier or a warehouse; determining a target constraint rule matched with the configuration request from a plurality of constraint rules; wherein the constraint rule comprises at least one of a service rule, a resource rule and an execution rule; inputting a configuration parameter obtained by analyzing the configuration request into a target constraint rule, and calculating to obtain a dynamic weight value of the supplier; according to the configuration parameters and historical data of the suppliers, predicting productivity gaps of the suppliers; and inputting the dynamic weight value, the productivity gap and historical data of the suppliers into a scheduling decision tree to generate a configuration result of the suppliers, the configuration result comprising a productivity distribution proportion of each supplier and a logistics routing suggestion.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and more particularly to supply chain configuration methods, equipment, and products. Background Technology

[0002] With the rapid development of e-commerce and the home decoration industry, supply chain configuration has become a key link in meeting diverse needs and coping with complex market environments.

[0003] In existing technologies, supply chain configuration methods typically rely on static rules and manual scheduling, handling order demands through pre-defined supplier rating and inventory allocation rules. However, these methods often suffer from slow response times and insufficient flexibility when facing dynamic changes (such as promotional activities and regional policy directives). For example, traditional systems struggle to adapt to supplier capacity fluctuations or sudden policy restrictions in real time, leading to fulfillment delays or cost overruns. Furthermore, existing technologies often employ a single rule system, lacking a layered and decoupled architecture, and cannot effectively balance the need for long-term stability with the requirements of real-time dynamic adjustments. Therefore, there is an urgent need for a method that can respond to configuration requests in real time and schedule efficiently to improve the flexibility, stability, and responsiveness of supply chain configuration. Summary of the Invention

[0004] This disclosure provides supply chain configuration methods, equipment, and products.

[0005] According to a first aspect of this disclosure, a supply chain configuration method is provided. The method includes: receiving a configuration request, wherein the configuration request is issued by any one of a regional store, supplier, or warehouse; determining a target constraint rule matching the configuration request from a plurality of constraint rules; wherein the constraint rule includes at least one of business rules, resource rules, and execution rules; inputting configuration parameters parsed from the configuration request into the target constraint rule to calculate a dynamic weight value for a supplier; and predicting the supplier's capacity gap based on the configuration parameters and the supplier's historical data; inputting the dynamic weight value, the capacity gap, and the supplier's historical data into a scheduling decision tree to generate a supplier configuration result, the configuration result including a capacity allocation ratio for each supplier and logistics routing suggestions.

[0006] Based on the above, upon receiving a configuration request in real time, the system matches the required rules, ensuring accurate selection of applicable rules and generating precise dynamic weight values ​​to optimize supplier allocation. Layered rules and similarity matching flexibly adapt to diverse scenarios such as promotions and production restrictions, significantly improving the accuracy of configuration results. Furthermore, the predictive model combines configuration parameters with historical supplier data to accurately predict capacity gaps. Further optimization of allocation ratios and logistics routes through a scheduling decision tree, utilizing dynamic weights and gap data, reduces fulfillment costs and delays.

[0007] According to at least one embodiment of this disclosure, the dynamic weight value of a supplier is calculated by inputting configuration parameters parsed from a configuration request into target constraint rules. This includes: inputting the configuration parameters parsed from the configuration request, including static supplier data, real-time supplier data, and regional policy parameters, into business rules to generate a supplier admission status and a static benchmark weight. The supplier admission status is used to screen qualified suppliers, and the static benchmark weight serves as the basis for weight calculation. The configuration parameters and historical default data are input into resource rules to generate a real-time adjustment factor, which reflects the supplier's real-time capabilities and performance reliability. The supplier admission status, static benchmark weight, real-time adjustment factor, and regional policy parameters are input into execution rules to generate a dynamic weight value for the supplier, which is used to evaluate supplier priority.

[0008] Based on the aforementioned publicly available solutions, it is evident that the dynamic weight calculation process significantly improves supply chain configuration performance through its ability to process real-time data. Real-time supplier data and regional policy parameters are updated in seconds via Redis's publish / subscribe mechanism, and combined with event streams to capture configuration changes, this ensures the rules engine can respond promptly to dynamic changes.

[0009] According to at least one embodiment of this disclosure, the supplier's historical data includes: capacity fluctuation coefficient, quality inspection pass rate, and logistics timeliness standard deviation; based on configuration parameters and the supplier's historical data, the supplier's capacity gap is predicted, including: converting real-time supplier data and regional policy parameters in the configuration parameters into a spatiotemporal feature matrix; and converting the capacity fluctuation coefficient, quality inspection pass rate, and logistics timeliness standard deviation into a supplier feature matrix; and inputting the spatiotemporal feature matrix and the supplier feature matrix into a long short-term memory model to predict the supplier's capacity gap.

[0010] Based on the publicly available solutions described above, the Long Short-Term Memory (LSTM) model significantly enhances the predictive power of the forecasting model by capturing the complex temporal dependencies in the spatiotemporal feature matrix and the capability characteristics in the supplier feature matrix. Through the accurate predictions and real-time data support of the LSTM model, high precision, dynamic adaptation, and efficient computation are achieved, providing powerful decision support for supply chain configuration.

[0011] According to at least one embodiment of this disclosure, the historical data of the supplier includes: supplier response speed and cost constraints; inputting dynamic weight values, capacity gaps and historical data of the supplier into a scheduling decision tree to generate the supplier configuration result includes: inputting dynamic weight values, capacity gaps, supplier response speeds and cost constraints into a scheduling decision tree to generate the supplier configuration result.

[0012] Based on the publicly available solutions, it is evident that dynamic weighting-driven scheduling decisions ensure that high-priority suppliers receive more orders, resulting in a more stable supply chain configuration and stronger resilience. Furthermore, by combining cost constraints with optimized allocation ratios and logistics routing, fulfillment costs are reduced. Through the core role of dynamic weighting in the scheduling decision tree, the efficiency, cost-effectiveness, and flexibility of supply chain configuration are significantly improved, providing reliable support for supplier allocation and logistics optimization in dynamic market environments.

[0013] According to at least one embodiment of this disclosure, the construction of the spatiotemporal feature matrix includes: generating a regional dimension based on real-time supplier data and regional policy parameters in the configuration parameters, the regional dimension including sub-regions divided according to geographical, economic or consumption characteristics, determined based on order data from the fulfillment database and service scope data from the supplier interface; generating a time dimension based on historical order data and real-time event streams, the time dimension including a prediction time window with hourly granularity to capture periodic demand and the impact of sudden events; and constructing the spatiotemporal feature matrix based on the regional dimension and the time dimension.

[0014] Based on the publicly available solutions described above, the spatiotemporal feature matrix can reflect the supply chain status in real time, ensuring the timeliness of the prediction model input. Comprehensive modeling of the complex dynamics of the supply chain using both regional and temporal dimensions helps improve the accuracy of LSTM predictions.

[0015] According to at least one embodiment of this disclosure, determining a target constraint rule matching a configuration request from multiple constraint rules includes: parsing the configuration request through a real-time event stream to extract the request type and configuration parameters, wherein the request type includes a promotional activity request or a regional policy instruction, and the configuration parameters include a regional identifier, category information, and event type; calculating the similarity with business rules, resource rules, and execution rules based on the request type and configuration parameters, wherein the business rules are based on static supplier data and long-term regional policies, the resource rules are based on real-time supplier data and short-term regional policies, and the execution rules are based on the output of integrating the aforementioned rules; and determining at least one of the matching business rules, resource rules, or execution rules as the target constraint rule when the similarity exceeds a predetermined threshold.

[0016] As can be seen from the publicly available solutions described above, the decoupling nature of the rules results in each rule operating independently, handling different tasks and data types. Therefore, it is necessary to accurately match the target rule corresponding to the configuration request through similarity calculation and threshold judgment; otherwise, incorrect rules may be selected, leading to deviations in dynamic weight values ​​and affecting the accuracy of the configuration results. By using similarity calculation and threshold judgment, we ensure that the target rule is highly relevant to the configuration request, generating accurate dynamic weight values ​​and optimizing the configuration results.

[0017] After generating the supplier configuration result according to at least one embodiment of this disclosure, the method further includes: monitoring the propagation impact of configuration changes triggered by configuration requests through real-time event streaming, whereby configuration changes include updates to supplier dynamic weight values ​​or configuration results; automatically rolling back to the most recent stable configuration version when a cascading error is detected, whereby cascading errors include supplier allocation failures or fulfillment costs exceeding a predetermined threshold, and the stable configuration version is obtained from the fulfillment database; and using blockchain technology to record configuration change logs, whereby the logs include records of changes to configuration parameters, dynamic weight values, and configuration results to achieve audit traceability.

[0018] Based on the publicly available solutions, the circuit breaker mechanism significantly enhances the stability and traceability of supply chain configuration through real-time monitoring, automatic rollback, and blockchain recording. Because supply chain configuration involves multi-source data (static data, real-time data) and complex scenarios (such as promotions and production restrictions), configuration changes may trigger cascading errors (such as allocation failures and cost overruns). The real-time monitoring mechanism constantly monitors the propagation impact of configuration changes, detecting abnormal states within seconds. The automatic rollback mechanism quickly restores the system to a reliable state by retrieving a stable configuration version from TiDB, preventing error propagation from affecting supply chain continuity. Blockchain recording ensures that all changes are traceable and tamper-proof, facilitating error analysis and compliance auditing. This significantly improves the system's robustness and management efficiency in complex and dynamic environments.

[0019] According to at least one embodiment of this disclosure, static supplier data and historical default data are stored in a distributed database based on a sharding storage mechanism; real-time supplier data and regional policy parameters are cached in a distributed memory cache system and updated in real time through a publish / subscribe mechanism; and the coordinated update of the distributed memory cache system and the distributed database is triggered by a real-time event stream to achieve second-level data consistency of configuration parameters.

[0020] Based on the publicly available solutions mentioned above, by triggering the coordinated updates of the distributed memory caching system (Redis) and the distributed database (TiDB) through real-time event streams, efficient data access for business rules, resource rules, and execution rules is supported, improving the response speed and system performance of configuration parameter parsing and dynamic weight calculation.

[0021] After generating the supplier configuration results according to at least one embodiment of this disclosure, the method further includes: obtaining the supplier configuration results through a real-time event stream, the configuration results including the capacity allocation ratio of each supplier and logistics routing suggestions; extracting the configuration results from a distributed caching system and transmitting them to the front-end display interface through a real-time push mechanism; and displaying the configuration results in the form of a chart or map on the front-end display interface, the chart including the distribution of supplier capacity allocation ratios and the map including logistics routing path information.

[0022] Based on the publicly available solutions, the charts and maps clearly present the supplier allocation ratios and logistics routes, making it easy for managers to quickly understand the complex configuration results.

[0023] According to a second aspect of this disclosure, an electronic device is provided, comprising: a memory storing execution instructions; and a processor executing the execution instructions stored in the memory, such that the processor performs the method described in the first aspect of any embodiment of this disclosure.

[0024] According to a third aspect of this disclosure, a readable storage medium is provided, wherein executable instructions are stored therein, which, when executed by a processor, are used to implement the method described in the first aspect of any embodiment of this disclosure.

[0025] According to a fourth aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the method described in the first aspect of any embodiment of this disclosure. Attached Figure Description

[0026] The accompanying drawings illustrate exemplary embodiments of the present disclosure and, together with the description thereof, serve to explain the principles of the present disclosure. These drawings are included to provide a further understanding of the present disclosure and are incorporated in and constitute a part of this specification.

[0027] Figure 1 This is a flowchart illustrating a supply chain configuration method provided in an embodiment of the present disclosure.

[0028] Figure 2 This is a schematic diagram of the structure of the supply chain configuration system provided in an embodiment of the present disclosure.

[0029] Figure 3 This is a schematic diagram of the dynamic weight calculation process provided in an embodiment of this disclosure.

[0030] Figure 4 This is a flowchart illustrating the capacity gap prediction method provided in an embodiment of the present disclosure.

[0031] Figure 5 This is a flowchart illustrating the spatiotemporal feature matrix construction method provided in this embodiment of the disclosure.

[0032] Figure 6 This is a flowchart illustrating the method for determining target constraint rules provided in an embodiment of this disclosure.

[0033] Figure 7 This is a flowchart illustrating the abnormal circuit breaking method provided in an embodiment of this disclosure.

[0034] Figure 8This is a schematic block diagram of a supply chain configuration device according to one embodiment of the present disclosure.

[0035] Figure 9 This is a schematic block diagram of an electronic device according to one embodiment of the present disclosure. Detailed Implementation

[0036] The present disclosure will now be described in further detail with reference to the accompanying drawings and examples. It should be understood that the specific examples described herein are for illustrative purposes only and are not intended to limit the scope of the disclosure. Furthermore, it should be noted that, for ease of description, only the parts relevant to the present disclosure are shown in the accompanying drawings.

[0037] It should be noted that, where there is no conflict, the embodiments and features described in this disclosure can be combined with each other. The technical solutions of this disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0038] Figure 1 This is a flowchart illustrating a supply chain configuration method provided in an embodiment of this disclosure. Figure 1 The method shown includes steps 101 to 105. This method can be executed by an electronic device such as a server (e.g., a local server or a cloud server).

[0039] Specifically, Figure 1 The method shown includes: Step 101: Received configuration request, wherein the configuration request is issued by any one of the regional stores, suppliers or warehouses.

[0040] First, it should be noted that this supply chain configuration method is based on a decoupled supply chain configuration system. This decoupled supply chain configuration system includes a business layer, a resource layer, and an execution layer. For example... Figure 2 This is a schematic diagram of the supply chain configuration system provided in this embodiment. The three layers are decoupled from each other, meaning that changing parameters in any one layer will not affect the other two layers. Specifically, the decoupling dimension of the business layer includes branch / store strategy decoupling, that is, this layer independently configures the promotional strategies / regional policies of branches (such as the package combination rules for the 618 promotion). The decoupling dimension of the resource layer includes supplier capability decoupling, that is, this layer dynamically binds to the supplier capacity pool and automatically switches suppliers based on real-time capacity fluctuations. The decoupling dimension of the execution layer includes stock keeping unit (SKU) fulfillment path decoupling, independently configuring SKU inventory allocation and logistics routing rules by category.

[0041] In practical applications, the configuration request can come from any of the following parties: regional store activity requests (such as the package combination rules for the 618 promotion or the store anniversary celebration activity rules), suppliers (such as production line shutdowns caused by sudden power outages), or warehouse SKUs.

[0042] Step 102: Determine the target constraint rule that matches the configuration request from multiple constraint rules; wherein the constraint rule includes at least one of business rules, resource rules and execution rules.

[0043] In practical applications, because the business layer, resource layer, and execution layer of this supply chain configuration system are decoupled, meaning they do not interfere with each other much, each layer generally only needs to focus on its own parameters and execute the corresponding rules. Therefore, when a configuration request is received, in order for each layer to better execute the corresponding rules, it is necessary to understand which layer's rules or which layers' rules are suitable for the configuration request to be assigned to for calculation.

[0044] By using similarity calculations and threshold judgments, we ensure that the target rules are highly relevant to the configuration request, generate accurate dynamic weight values, and optimize the configuration results.

[0045] Step 103: Input the configuration parameters parsed from the configuration request into the target constraint rules to calculate the supplier's dynamic weight value.

[0046] After parsing the configuration parameters, the matching target constraint rules can be used to execute computation tasks. It should be noted that a configuration request can match one target constraint rule or multiple target constraint rules simultaneously.

[0047] It should be noted that the dynamic weight values ​​mentioned here are updated in real time according to changes in configuration requests. This ensures that the supply chain adjusts its configuration results promptly, guaranteeing stable supply chain operation.

[0048] Step 104: Based on the configuration parameters and the supplier's historical data, predict the supplier's capacity gap.

[0049] When performing forecasting tasks, a Long Short-Term Memory (LSTM) model can be used as an alternative. Of course, other models can be selected as needed in practical applications to complete the capacity gap forecasting task. This is merely an example and does not constitute a limitation on the scheme disclosed in this application. It is assumed that by using an LSTM forecasting model, the complex time dependencies in the spatiotemporal feature matrix and the capability characteristics in the supplier feature matrix are captured, thus significantly improving the forecasting capability of the model. Through the accurate forecasting of the LSTM model and real-time data support, high accuracy, dynamic adaptation, and efficient computation are achieved, providing powerful decision support for supply chain configuration.

[0050] Step 105: Input the dynamic weight value, capacity gap and supplier historical data into the scheduling decision tree to generate supplier configuration results, which include the capacity allocation ratio and logistics routing suggestions for each supplier.

[0051] The desired configuration result can be obtained by using a scheduling decision tree. This configuration result includes the selected suppliers based on the current supply chain situation and the capacity allocation ratio of each supplier.

[0052] In addition, logistics route suggestions can be obtained. Because in practical applications, different suppliers are located in different regions, and the time and cost of transporting goods to the warehouse also vary. Therefore, to provide users with more comprehensive configuration results, multiple optional logistics routes can be offered.

[0053] In one or more embodiments of this disclosure, such as Figure 3 This is a schematic diagram of the dynamic weight calculation process provided in this embodiment of the disclosure. As described in step 103, the configuration parameters parsed from the configuration request are input into the target constraint rules to calculate the supplier's dynamic weight value. Specifically, this includes the following steps: Step 1031: The configuration parameters parsed from the configuration request, including static supplier data, real-time supplier data, and regional policy parameters, are input into the business rules to generate supplier admission status and static benchmark weight. The supplier admission status is used to screen qualified suppliers, and the static benchmark weight serves as the basis for weight calculation. Step 1032: The configuration parameters and historical default data are input into the resource rules to generate a real-time adjustment factor. The real-time adjustment factor reflects the supplier's real-time capabilities and performance reliability. Step 1033: The supplier admission status, static benchmark weight, real-time adjustment factor, and regional policy parameters are input into the execution rules to generate a supplier dynamic weight value. The dynamic weight value is used to evaluate supplier priority.

[0054] In practical applications, upon receiving a configuration request, the configuration parameters parsed from the request, including static supplier data, real-time supplier data, and regional policy parameters, are input into business rules for processing. Static supplier data is obtained from a fulfillment database (such as TiDB) and includes historical supplier performance data (such as order completion rate and service quality rating) and long-term regional policy parameters (such as environmental compliance requirements, which are updated regularly). Business rules, based on the Drools rule engine, analyze this static data to generate supplier eligibility status and static baseline weights. Supplier eligibility status is used to screen qualified suppliers, ensuring that only suppliers meeting long-term policies (such as environmental certification) participate in subsequent calculations; static baseline weights reflect the supplier's basic capabilities and serve as the initial value for dynamic weight calculations. For example, for a promotional activity request, business rules might screen reliable suppliers based on historical order completion rates and generate initial weights.

[0055] Furthermore, configuration parameters, combined with historical default data, are input into resource rules to generate real-time adjustment factors. Real-time supplier data is obtained through supplier interfaces (such as APIs) and cached in a distributed memory caching system (such as Redis), including real-time capacity ratios (such as the ratio of current capacity to total capacity) and logistical status (such as delivery timeliness). Historical default data is extracted from the fulfillment database, containing the number of supplier defaults and latency rates. Regional policy parameters are updated through real-time event streams (such as Apache Flink) to reflect short-term policy changes (such as promotional discounts and environmental production restrictions). Resource rules analyze real-time supplier data, historical default data, and short-term regional policies using the Drools engine to generate real-time adjustment factors that reflect the supplier's real-time capabilities and fulfillment reliability. For example, if a supplier's capacity decreases due to a production restriction order, the adjustment factor will reduce its weight contribution.

[0056] The supplier admission status and static baseline weights generated by business rules, the real-time adjustment factors generated by resource rules, and regional policy parameters are input into the execution rules to generate dynamic supplier weight values. The execution rules integrate multi-source data through the Drools engine, comprehensively considering admission status (screening results), static baseline weights (basic capabilities), real-time adjustment factors (dynamic capabilities), and regional policy parameters (such as the impact of promotions or production restrictions) to calculate the final dynamic weight value. The dynamic weight value is used to assess supplier priority, providing a basis for subsequent capacity forecasting and scheduling decisions. For example, in a promotional scenario, high-weight suppliers may receive more capacity allocation.

[0057] Based on the aforementioned publicly available solutions, the dynamic weight calculation process significantly improves supply chain configuration performance through its ability to process real-time data. Real-time supplier data and regional policy parameters are updated in seconds via Redis's publish / subscribe mechanism, combined with Flink event streams to capture configuration changes (such as promotional activities and production restrictions), ensuring the rule engine can respond promptly to dynamic changes. Compared to traditional timed refresh mechanisms (such as updates every 15 minutes), this publicly available solution's second-level data synchronization significantly reduces data lag and achieves rapid response. Furthermore, hierarchical rule processing (business rules, resource rules, and execution rules) balances long-term stability (static data) and short-term flexibility (real-time data) through separation of responsibilities, supporting efficient weight calculation. The parallel processing capabilities of the Drools engine further enhance computational throughput, making it suitable for high-frequency configuration request scenarios. In addition, this publicly available solution, through real-time data processing and a hierarchical rule mechanism, achieves the beneficial effects of flexible adaptation to multiple scenario requirements (such as promotions and policy changes), efficient calculation, and reliable weight evaluation, providing higher response speed and decision-making accuracy for supply chain configuration.

[0058] In one or more embodiments of this disclosure, such as Figure 4 This is a flowchart illustrating the capacity gap prediction method provided in this embodiment. The historical data of the suppliers includes: capacity fluctuation coefficient, quality inspection pass rate, and logistics timeliness standard deviation. As described in step 104, based on the configuration parameters and the historical data of the suppliers, the capacity gap of the suppliers is predicted, specifically including the following steps: Step 1041: Converting real-time supplier data and regional policy parameters from the configuration parameters into a spatiotemporal feature matrix. And, Step 1042: Converting the capacity fluctuation coefficient, quality inspection pass rate, and logistics timeliness standard deviation into a supplier feature matrix. Step 1043: Inputting the spatiotemporal feature matrix and the supplier feature matrix into a long short-term memory model to predict the capacity gap of the suppliers.

[0059] The historical supplier data mentioned here includes capacity fluctuation coefficient, quality inspection pass rate, and logistics timeliness standard deviation, all extracted from fulfillment databases (such as TiDB). The capacity fluctuation coefficient reflects the stability of the supplier's capacity changes over time in historical order data, calculated, for example, by the standard deviation of historical order volume and actual supply. The quality inspection pass rate indicates the stability of the supplier's product quality, derived from historical quality inspection records. The logistics timeliness standard deviation measures the volatility of the supplier's delivery time, calculated based on historical logistics records. These data provide a long-term, stable characteristic basis for assessing supplier capabilities.

[0060] When constructing the spatiotemporal feature matrix, the configuration parameters parsed from the configuration request, including real-time supplier data and regional policy parameters, are converted into the spatiotemporal feature matrix. Real-time supplier data is obtained through supplier interfaces (such as APIs) and cached in a distributed memory caching system (such as Redis), containing real-time capacity ratios (such as the ratio of current capacity to total capacity) and logistics status (such as delivery timeliness). Regional policy parameters are obtained from the regional policy configuration table through real-time event streams (such as Apache Flink), containing promotional activity factors (such as discount levels) or regional policy directives (such as environmental protection production restriction orders).

[0061] The spatiotemporal feature matrix includes the following dimensions: Regional dimension: Sub-regions are defined based on geographical, economic, or consumption characteristics (e.g., Shanghai, East China), determined by order data from the fulfillment database and the service scope of suppliers. Time dimension: Measured at the hourly level (e.g., a 24-hour forecast window), capturing cyclical demand patterns and the impact of unforeseen events (e.g., a surge in promotional orders). Furthermore, it may also include a feature dimension: Combining supplier dynamic weight values, order volume, promotional sensitivity, and other features obtained earlier, reflecting changes in demand and policy.

[0062] Historical supplier data, including capacity fluctuation coefficients, quality inspection pass rates, and logistics timeliness standard deviations, is transformed into a supplier feature vector. This matrix is ​​extracted from the fulfillment database and generated through statistical analysis. For example, the capacity fluctuation coefficient is calculated based on historical order and supply fluctuations, the quality inspection pass rate is based on quality inspection records, and the logistics timeliness standard deviation is based on delivery time statistics. These features reflect the supplier's long-term capability characteristics and are used as input to a Long Short-Term Memory (LSTM) model.

[0063] After obtaining the spatiotemporal feature matrix and supplier feature vector through the steps described above, these are input into a Long Short-Term Memory (LSTM) model for prediction. The LSTM model captures the time dependence of regional demand and the dynamic changes in supplier capacity through its gating mechanisms (such as input gate, forget gate, and output gate), predicting the supplier capacity gap in the future (e.g., 24 hours, 48 ​​hours, or 72 hours). The capacity gap can be understood as the difference between the predicted demand and the supplier's actual capacity. For example, in a promotional campaign scenario, LSTM predicts the gap amount for a specific supplier based on order surges in the spatiotemporal feature matrix and capacity fluctuations in the supplier feature matrix.

[0064] It should be noted that the LSTM model is trained using supervised learning methods, with training samples sourced from the fulfillment database and real-time event streams. Specific training samples include: Historical order data: extracted from TiDB, containing regional dimensions (geographical and economic features), time dimensions (hourly order volume), and feature dimensions (historical demand, promotional factors). Supplier historical data: extracted from TiDB, including capacity fluctuation coefficients, quality inspection pass rates, and logistics timeliness standard deviations, reflecting supplier capabilities. Regional policy records: obtained from the regional policy configuration table (Redis), containing historical promotional activities and policy directives (such as production restriction orders), supplemented by real-time changes via Flink event streams. The training process optimizes LSTM parameters through backpropagation to minimize the error between the predicted gap and the actual gap. The model is updated weekly based on the latest data to ensure predictive capabilities adapt to market changes. Training data covers multiple scenarios (such as promotions and policy changes) to improve the model's generalization ability.

[0065] Based on the aforementioned publicly available solutions, the LSTM prediction model significantly enhances its predictive capabilities by capturing the complex time dependencies in the spatiotemporal feature matrix and the capability characteristics in the supplier feature matrix. Compared to traditional statistical models (such as linear regression), LSTM effectively models the periodic fluctuations in regional demand (such as daily peaks) and sudden events (such as surges in promotional orders), as well as the dynamic changes in supplier capabilities (such as capacity reductions), significantly improving prediction accuracy. Real-time supplier data and regional policy parameters are updated second-by-second via Redis pub / sub mechanisms and Flink event streams, ensuring that the model input reflects the latest market conditions, thereby achieving dynamic adaptation to multiple scenario demands (such as promotions and production restrictions). High-precision capacity gap prediction provides a reliable basis for scheduling decisions, optimizing supplier allocation ratios and logistics routes, and reducing fulfillment delays and cost overruns. Furthermore, hierarchical data processing (historical data stored in TiDB and real-time data cached in Redis) combined with the parallel computing capabilities of LSTM improves prediction efficiency and is suitable for high-frequency configuration scenarios. This publicly available solution achieves high accuracy, dynamic adaptation, and efficient computation through the accurate prediction of the LSTM model and real-time data support, providing powerful decision support for supply chain configuration.

[0066] In one or more embodiments of this disclosure, the supplier's historical data includes: supplier response speed and cost constraints. As described in step 105, inputting dynamic weight values, capacity gaps, and the supplier's historical data into the scheduling decision tree to generate the supplier's configuration result includes: inputting dynamic weight values, capacity gaps, supplier response speeds, and cost constraints into the scheduling decision tree to generate the supplier's configuration result.

[0067] In practical applications, supplier historical data, including response speed and cost constraints, is extracted from fulfillment databases (such as TiDB). Response speed reflects the average time from order receipt to fulfillment, such as delivery time calculated based on historical logistics records. Cost constraints include supplier quotations, logistics costs, and historical default costs, reflecting economic constraints. This data is updated regularly from the fulfillment database (e.g., daily or weekly), providing a stable basis for supplier capability assessment in scheduling decisions.

[0068] In this scheduling decision tree, dynamic weight values, capacity gaps, supplier response speeds, and cost constraints are used as inputs. For example... Figure 3 As illustrated in the example, the dynamic weight value integrates static supplier data (historical performance), real-time supplier data (capacity ratio, logistics status), historical default data, and regional policy parameters (such as promotion factors and production restriction orders) to reflect the supplier's overall priority. As mentioned earlier, the capacity gap here is predicted using a Long Short-Term Memory (LSTM) model, representing the difference between demand and the supplier's actual capacity in the next 24, 48, or 72 hours. Response speed and cost constraints are obtained from the fulfillment database, providing timeliness and economic constraints respectively. Real-time supplier data and dynamic weight values ​​are updated in real-time through a distributed memory caching system (such as Redis) using a publish / subscribe mechanism, while regional policy parameters are synchronized through a real-time event stream (such as Apache Flink) to ensure the real-time nature of the input data.

[0069] In one alternative approach, the scheduling decision tree can employ the C4.5 algorithm. By analyzing historical events (such as order fulfillment records and abnormal allocation scenarios), it constructs decision paths and generates the optimal supplier configuration results. Dynamic weight values ​​represent the current supplier priority, helping to quickly find better suppliers and ensuring that high-weight suppliers (such as those with strong overall capabilities and excellent real-time status) receive priority access to more capacity allocation.

[0070] Capacity gaps are used to constrain total allocation, ensuring that forecasted demand is met. Response speed and cost constraints further optimize allocation ratios and logistics routing, such as selecting suppliers and delivery routes with fast response times and low costs. The decision tree, through multi-branch analysis, comprehensively considers weights, gaps, timeliness, and cost to generate configuration results, including the capacity allocation ratios of each supplier (e.g., supplier A 50%, supplier B 30%, supplier C 20%) and logistics routing suggestions (e.g., prioritizing local delivery or cross-regional allocation).

[0071] It should be noted that this publicly disclosed solution integrates suppliers' static capabilities (historical performance), real-time status (capacity, logistics), and policy impacts (promotional weighting, production restriction weighting) to provide a comprehensive prioritization basis for the decision tree. Compared to traditional fixed-priority scheduling methods, dynamic weights can adapt to market changes in real time (such as a surge in promotional orders or the impact of production restriction orders), ensuring that high-priority suppliers are allocated reasonably, thereby optimizing fulfillment efficiency. Furthermore, dynamic weights, combined with capacity gaps, response speed, and cost constraints, enable the decision tree to balance timeliness and economy, generating the lowest-cost and most efficient configuration result. For example, in promotional scenarios, dynamic weights prioritize suppliers with sufficient real-time capacity, reducing fulfillment delays; in production restriction scenarios, weight adjustments avoid allocating to restricted suppliers, reducing default risk.

[0072] This publicly disclosed solution employs dynamic weight-driven scheduling decisions. Dynamic weighting ensures that high-priority suppliers receive more orders, resulting in a more stable supply chain configuration and stronger resilience. Furthermore, it optimizes allocation ratios and logistics routes based on cost constraints, reducing fulfillment costs. Through the core role of dynamic weighting in the scheduling decision tree, it significantly improves the efficiency, cost-effectiveness, and flexibility of supply chain configuration, providing reliable support for supplier allocation and logistics optimization in dynamic market environments.

[0073] In one or more embodiments of this disclosure, such as Figure 5 This is a flowchart illustrating the spatiotemporal feature matrix construction method provided in this embodiment of the disclosure. Figure 5 As can be seen, the method specifically includes the following steps: Step 501: Based on real-time supplier data and regional policy parameters in the configuration parameters, generate a regional dimension. The regional dimension includes sub-regions divided according to geographical, economic, or consumption characteristics, determined based on order data from the fulfillment database and service scope data from the supplier interface. Step 502: Based on historical order data and real-time event streams, generate a time dimension. The time dimension includes prediction time windows with hourly granularity to capture periodic demand and the impact of sudden events. Step 503: Construct the spatiotemporal feature matrix using the regional dimension and the time dimension.

[0074] Based on real-time supplier data and regional policy parameters parsed from the configuration request, the regional dimension of the spatiotemporal feature matrix is ​​generated. Real-time supplier data is obtained through supplier interfaces (such as APIs) and cached in a distributed memory caching system (such as Redis). This data includes real-time capacity ratios (such as the ratio of current capacity to total capacity) and logistics status (such as delivery timeliness), and is updated hourly via a publish / subscribe mechanism. Regional policy parameters are obtained from the regional policy configuration table (Redis) and synchronized through a real-time event stream (such as Apache Flink). These parameters include promotional activity factors (such as discount levels) or regional policy directives (such as environmental protection production restrictions). The regional dimension is divided into sub-regions based on geographical features (such as provinces / cities), economic features (such as developed / underdeveloped regions), or consumption features (such as high-frequency / low-frequency purchase areas). This is determined based on historical order data from a fulfillment database (such as TiDB) and service scope data from the supplier interface. For example, order data for Shanghai may show high consumption characteristics, and the supplier's service scope covers East China, generating corresponding regional dimension features.

[0075] The time dimension in spatiotemporal features can be derived from historical order data and real-time event streams. Historical order data is extracted from the fulfillment database and includes order volume by region and time granularity (e.g., hourly), reflecting cyclical demand patterns (e.g., daily peaks). The real-time event stream is processed by Flink to capture sudden events (e.g., surges in promotional orders, production restriction orders), updating the time dimension data in real time. The time dimension can be granular at the hourly level, constructing prediction time windows (e.g., the next 24 hours, 48 ​​hours, or 72 hours), capable of capturing the impact of cyclical demand (e.g., weekend order peaks) and sudden events (e.g., sudden surges in demand due to promotions).

[0076] By leveraging regional and temporal dimensions, and combining them with other features (such as dynamic weights, order volume, and promotional sensitivity), a spatiotemporal feature matrix is ​​constructed. The regional dimension provides the spatial distribution characteristics of the supply chain, while the temporal dimension provides the dynamic temporal characteristics of demand. The combination of these two dimensions forms a multidimensional feature matrix that comprehensively describes the real-time status and demand patterns of the supply chain. The spatiotemporal feature matrix is ​​updated in real-time through a distributed caching system and event streams, ensuring the timeliness of the input data and supporting the accurate predictions of the LSTM model.

[0077] Based on the publicly available solutions described above, the spatiotemporal feature matrix can reflect the supply chain status in real time, ensuring the timeliness of the prediction model input. Comprehensive modeling of the complex dynamics of the supply chain using both regional and temporal dimensions helps improve the accuracy of LSTM predictions.

[0078] In one or more embodiments of this disclosure, such as Figure 6This is a flowchart illustrating the method for determining target constraint rules provided in this embodiment of the disclosure. As described in step 102, determining the target constraint rule matching the configuration request from multiple constraint rules specifically includes the following steps: Step 1021: Parsing the configuration request through a real-time event stream to extract the request type and configuration parameters. The request type includes a promotional activity request or a regional policy instruction, and the configuration parameters include a regional identifier, category information, and event type. Step 1022: Calculating the similarity with business rules, resource rules, and execution rules based on the request type and configuration parameters. Business rules are based on static supplier data and long-term regional policies, resource rules are based on real-time supplier data and short-term regional policies, and execution rules are based on the output integrating the aforementioned rules. Step 1023: When the similarity exceeds a predetermined threshold, determining at least one of the matching business rules, resource rules, or execution rules as the target constraint rule.

[0079] As mentioned earlier, configuration requests are initiated by stores, suppliers, or warehouse SKUs. These requests are parsed using real-time event streaming to extract the request type and configuration parameters. Request types include promotional activity requests (such as promotional orders initiated by stores) and regional policy directives (such as production restriction orders initiated by suppliers or warehouse SKUs). Configuration parameters include regional identifiers (such as Shanghai or East China), category information (such as bathroom fixtures or furniture), and event types (such as promotions or production restrictions). The parsed parameters are stored in a regional policy configuration table (a distributed in-memory cache system, such as Redis), and are updated sub-seconds via Flink event streaming to ensure the latest market demands and policy changes are captured.

[0080] Based on the parsed request type and configuration parameters, the Drools rule engine calculates the similarity with business rules, resource rules, and execution rules. Business rules, based on static supplier data (such as historical performance and order completion rates, stored in the TiDB fulfillment database) and long-term regional policies (such as environmental certification requirements), are suitable for filtering supplier access status and generating static baseline weights. Resource rules, based on real-time supplier data (such as real-time capacity ratios and logistics status, cached in Redis) and short-term regional policies (such as promotional factors and production restrictions, updated via Flink), generate real-time adjustment factors. Execution rules integrate the outputs of business rules and resource rules to calculate dynamic weight values. Similarity calculation is implemented through Drools' DSL conditional logic, comparing the degree of match between request parameters (such as region, category, and event type) and rule conditions (such as policy constraints and supplier status).

[0081] When the similarity exceeds a predetermined threshold, at least one match is determined as the target constraint rule. The threshold is defined by the rule triggering conditions of the Drools engine, ensuring that the matched rule is highly relevant to the request. The matching result may be a single rule (such as a promotional request triggering a business rule) or multiple rules (such as a production restriction order triggering resource rules and execution rules). The target constraint rule is used for subsequent dynamic weight value calculation to ensure that the weight reflects the specific needs of the request.

[0082] As can be seen from the publicly available solutions described above, the decoupling nature of the rules results in each rule operating independently, handling different tasks and data types. Therefore, it is necessary to accurately match the target rule corresponding to the configuration request through similarity calculation and threshold judgment; otherwise, incorrect rules may be selected, leading to deviations in dynamic weight values ​​and affecting the accuracy of the configuration results. By using similarity calculation and threshold judgment, we ensure that the target rule is highly relevant to the configuration request, generating accurate dynamic weight values ​​and optimizing the configuration results.

[0083] In one or more embodiments of this disclosure, such as Figure 7 This is a flowchart illustrating the abnormal circuit breaker method provided in this embodiment. After generating the supplier's configuration result, the method specifically includes: Step 701: Monitoring the propagation impact of configuration changes triggered by configuration requests through real-time event streaming. Configuration changes include updates to supplier dynamic weight values ​​or configuration results. Step 702: When a cascading error is detected, automatically rolling back to the most recent stable configuration version. Cascading errors include supplier allocation failures or fulfillment costs exceeding a predetermined threshold. The stable configuration version is obtained from the fulfillment database. Step 703: Recording configuration change logs using blockchain technology. The logs include records of changes to configuration parameters, dynamic weight values, and configuration results to enable audit traceability.

[0084] After generating vendor configuration results, the propagation impact of configuration changes triggered by configuration requests is monitored via real-time event streaming (such as Apache Flink). Configuration changes include updates to vendor dynamic weight values ​​or configuration results (such as allocation ratio adjustments). Propagation impacts involve order fulfillment status (such as latency rates), cost changes (such as logistics costs), or allocation effectiveness (such as vendor capacity shortages). These changes are captured in real-time, with real-time weight values ​​and configuration status obtained from a distributed in-memory caching system (such as Redis), and data synchronization is achieved within seconds via a publish / subscribe mechanism.

[0085] When a cascading error is detected, the system automatically rolls back to the most recent stable configuration version. Cascading errors include supplier allocation failures (such as allocation to a supplier with no capacity) or fulfillment costs exceeding predetermined thresholds (such as logistics costs exceeding the budget limit). Error detection is based on real-time fulfillment data (such as order delay rates and cost overruns) analyzed by Flink event stream analytics and predefined thresholds (such as cost thresholds based on historical average costs). The stable configuration version is retrieved from a fulfillment database (such as TiDB) and includes previous dynamic weight values, allocation ratios, and logistics routes to ensure supply chain continuity and stability. For example, if a promotional activity results in allocation to a low-capacity supplier, triggering an allocation failure, the system rolls back to the stable allocation ratio of the previous day and resumes normal fulfillment.

[0086] All configuration changes, including configuration parameters (regional identifiers, category information), dynamic weight values, and configuration results, are recorded in a distributed ledger using blockchain technology. The change log employs an immutable blockchain structure, containing timestamps, change types (such as weight adjustments and allocation updates), and change data (such as adjusted proportions). Blockchain records are synchronized through distributed nodes, ensuring data security and traceability, and supporting auditing and tracing. For example, managers can use blockchain logs to trace weight changes triggered by promotional activities, analyze the causes of cascading errors, or verify compliance.

[0087] Based on the publicly available solutions, the circuit breaker mechanism significantly enhances the stability and traceability of supply chain configuration through real-time monitoring, automatic rollback, and blockchain recording. Because supply chain configuration involves multi-source data (static data, real-time data) and complex scenarios (such as promotions and production restrictions), configuration changes may trigger cascading errors (such as allocation failures and cost overruns). The real-time monitoring mechanism constantly monitors the propagation impact of configuration changes, detecting abnormal states within seconds. The automatic rollback mechanism quickly restores the system to a reliable state by retrieving a stable configuration version from TiDB, preventing error propagation from affecting supply chain continuity. Blockchain recording ensures that all changes are traceable and tamper-proof, facilitating error analysis and compliance auditing. This significantly improves the system's robustness and management efficiency in complex and dynamic environments.

[0088] In one or more embodiments of this disclosure, the method further includes storing static supplier data and historical default data in a distributed database based on a sharding storage mechanism; caching real-time supplier data and regional policy parameters in a distributed memory caching system and updating them in real time through a publish / subscribe mechanism; and triggering the coordinated update of the distributed memory caching system and the distributed database through a real-time event stream to achieve second-level data consistency of configuration parameters.

[0089] Static supplier data and historical default data are stored in a distributed database (such as TiDB) using a sharding mechanism. Static supplier data includes the supplier's historical performance (such as order completion rate and service quality score) and long-term regional policies (such as environmental certification requirements), and is updated regularly (daily or monthly) from the fulfillment database. Historical default data includes the number of defaults and latency rates, reflecting the supplier's fulfillment reliability. TiDB optimizes query performance and data distribution by partitioning data by region, category, or supplier ID through sharding.

[0090] Real-time supplier data and regional policy parameters are cached in a distributed in-memory caching system (such as Redis) and updated in real time via a publish / subscribe mechanism. Real-time supplier data is obtained through supplier interfaces (such as APIs) and includes real-time capacity ratios (such as the ratio of current capacity to total capacity) and logistics status (such as delivery timeliness), updated hourly via the publish / sub mechanism. Regional policy parameters are obtained from the regional policy configuration table and include promotional activity factors (such as discount levels) or regional policy directives (such as environmental protection production restrictions), updated via real-time event streams (such as Apache Flink).

[0091] Based on the publicly available solutions mentioned above, by triggering the coordinated updates of the distributed memory caching system (Redis) and the distributed database (TiDB) through real-time event streams, efficient data access for business rules, resource rules, and execution rules is supported, improving the response speed and system performance of configuration parameter parsing and dynamic weight calculation.

[0092] In one or more embodiments of this disclosure, after generating the supplier configuration results, the method further includes: obtaining the supplier configuration results through a real-time event stream, the configuration results including the capacity allocation ratio of each supplier and logistics routing suggestions; extracting the configuration results from a distributed caching system and transmitting them to the front-end display interface through a real-time push mechanism; and displaying the configuration results in the form of a chart or map on the front-end display interface, the chart including the distribution of supplier capacity allocation ratios and the map including logistics routing path information.

[0093] Supplier configuration results are obtained through real-time event streaming. These results include the capacity allocation ratios for each supplier (e.g., supplier A 50%, supplier B 30%) and logistics routing suggestions (e.g., local delivery, cross-regional transfer). The real-time event stream captures configuration changes (e.g., ratio adjustments, routing optimizations) in real time, ensuring the results reflect the latest supply chain status. The extracted results are transmitted to the front-end display interface via a real-time push mechanism, ensuring managers receive the latest configurations immediately. On the front-end display interface, the configuration results are presented in chart or map format. Charts (e.g., bar charts, pie charts) show the distribution of supplier capacity allocation ratios, intuitively reflecting the allocation weight of each supplier; maps display logistics routing path information (e.g., delivery routes, cross-regional transfer routes), clearly presenting the logistics plan. The interface supports dynamic updates to reflect real-time configuration changes.

[0094] Based on the publicly available solutions, the charts and maps clearly present the supplier allocation ratios and logistics routes, making it easy for managers to quickly understand the complex configuration results.

[0095] Based on any of the above embodiments, this disclosure also provides a supply chain configuration apparatus. This apparatus can be applied to a server. Figure 8 This is a schematic block diagram of a supply chain configuration apparatus according to one embodiment of this disclosure. Figure 8 As shown, the supply chain configuration device includes: a receiving module 81 for receiving configuration requests, wherein the configuration request is issued by any one of the regional stores, suppliers or warehouses.

[0096] The determination module 82 is used to determine the target constraint rule that matches the configuration request from multiple constraint rules; wherein the constraint rule includes at least one of business rules, resource rules and execution rules.

[0097] The weight calculation module 83 is used to input the configuration parameters parsed from the configuration request into the target constraint rules to calculate the dynamic weight value of the supplier.

[0098] The prediction module 84 is used to predict the supplier's capacity gap based on configuration parameters and the supplier's historical data.

[0099] The generation module 85 is used to input dynamic weight values, capacity gaps and historical data of suppliers into the scheduling decision tree to generate the configuration results of suppliers. The configuration results include the capacity allocation ratio of each supplier and logistics routing suggestions.

[0100] The weight calculation module 83 is used to parse configuration parameters obtained from the configuration request, including static supplier data, real-time supplier data, and regional policy parameters, input them into business rules, and generate supplier access status and static benchmark weights. The supplier access status is used to screen qualified suppliers, and the static benchmark weights serve as the basis for weight calculation. The configuration parameters and historical default data are input into resource rules to generate real-time adjustment factors, which reflect the supplier's real-time capabilities and performance reliability. The supplier access status, static benchmark weights, real-time adjustment factors, and regional policy parameters are input into execution rules to generate supplier dynamic weight values, which are used to evaluate supplier priority.

[0101] Optionally, the supplier's historical data includes: capacity fluctuation coefficient, quality inspection pass rate, and logistics timeliness standard deviation. Prediction module 84 is used to convert real-time supplier data and regional policy parameters in the configuration parameters into a spatiotemporal feature matrix; and to convert the capacity fluctuation coefficient, quality inspection pass rate, and logistics timeliness standard deviation into a supplier feature matrix; using the spatiotemporal feature matrix and the supplier feature matrix as inputs into a long short-term memory model, the supplier's capacity gap is predicted.

[0102] Optionally, the supplier's historical data includes supplier response speed and cost constraints. The generation module 85 is used to input dynamic weight values, capacity gaps, supplier response speeds, and cost constraints into the scheduling decision tree to generate the supplier configuration results.

[0103] The prediction module 84 is used to generate a regional dimension based on real-time supplier data and regional policy parameters in the configuration parameters. The regional dimension includes sub-regions divided by geographical, economic or consumption characteristics and is determined based on order data from the fulfillment database and service scope data from the supplier interface. Based on historical order data and real-time event streams, a time dimension is generated. The time dimension includes a prediction time window with hourly granularity to capture the impact of periodic demand and sudden events.

[0104] The determination module 82 is used to parse configuration requests through real-time event streams, extract request types and configuration parameters. Request types include promotional activity requests or regional policy instructions, and configuration parameters include regional identifiers, category information, and event types. Based on the request type and configuration parameters, it calculates the similarity with business rules, resource rules, and execution rules. Business rules are based on static supplier data and long-term regional policies, resource rules are based on real-time supplier data and short-term regional policies, and execution rules are based on the output of integrating the aforementioned rules. When the similarity exceeds a predetermined threshold, at least one of the matching business rules, resource rules, or execution rules is determined as the target constraint rule.

[0105] The circuit breaker module 86 is used to monitor the propagation impact of configuration changes triggered by configuration requests through real-time event streaming. Configuration changes include updates to supplier dynamic weight values ​​or configuration results. When a cascading error is detected, it automatically rolls back to the most recent stable configuration version. Cascading errors include supplier allocation failures or fulfillment costs exceeding a predetermined threshold. The stable configuration version is obtained from the fulfillment database. Blockchain technology is used to record configuration change logs, which include records of changes to configuration parameters, dynamic weight values, and configuration results to achieve audit traceability. (The circuit breaker mechanism enhances the stability and traceability of supply chain configuration through real-time monitoring, automatic rollback, and blockchain recording.)

[0106] Storage module 87 is used to store static supplier data and historical default data in a distributed database based on a sharding storage mechanism; cache real-time supplier data and regional policy parameters in a distributed memory cache system and update them in real time through a publish / subscribe mechanism; and trigger the linkage update between the distributed memory cache system and the distributed database through a real-time event stream to achieve second-level data consistency of configuration parameters.

[0107] Display module 88 is used to obtain supplier configuration results through real-time event stream. The configuration results include the capacity allocation ratio of each supplier and logistics routing suggestions. The configuration results are extracted from the distributed cache system and transmitted to the front-end display interface through a real-time push mechanism. The configuration results are displayed on the front-end display interface in the form of charts or maps. The charts include the distribution of supplier capacity allocation ratios, and the maps include the path information of logistics routes.

[0108] The specific implementation process of the functions and roles of each module in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0109] The entity executing the supply chain configuration method in the specific embodiments of this disclosure can be an electronic device such as a server (including a local server or a cloud server).

[0110] Therefore, based on any of the above embodiments, this disclosure also provides an electronic device that can execute the supply chain configuration method of any of the embodiments described above.

[0111] Figure 9 This is a schematic block diagram of an electronic device according to one embodiment of the present disclosure.

[0112] The hardware architecture of the electronic device 1000 can be implemented using a bus architecture. The bus architecture can include any number of interconnect buses and bridges, depending on the specific application of the hardware and overall design constraints. Bus 1100 connects various circuits, including one or more processors 1200, memory 1300, and / or hardware modules. Bus 1100 can also connect various other circuits 1400, such as peripheral devices, voltage regulators, power management circuits, external antennas, etc.

[0113] Bus 1100 can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Component (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, only one connection line is used in this diagram, but this does not imply that there is only one bus or only one type of bus.

[0114] This disclosure also provides a readable storage medium storing a computer program that, when executed by a processor, is used to implement the methods described above. A "readable storage medium" can be any means capable of containing, storing, communicating, propagating, or transmitting a program for use by or in conjunction with an instruction execution system, apparatus, or device. More specific examples of a readable storage medium include: an electrical connection with one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and portable read-only memory (CDROM), etc.

[0115] This disclosure also provides a computer program product, the methods of which can be implemented wholly or partially through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented wholly or partially as a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed, all or part of the processes or functions of this disclosure are performed.

[0116] Computer programs or instructions can be stored in a readable storage medium or transferred from one readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The readable storage medium can be any available medium capable of access, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; an optical medium, such as a digital video optical disc; or a semiconductor medium, such as a solid-state drive. The computer-readable storage medium can be a volatile or non-volatile storage medium, or it can include both volatile and non-volatile types of storage media.

[0117] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0118] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0119] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0120] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0121] In the description of this specification, the references to terms such as "one embodiment / mode," "some embodiments / modes," "example," "specific example," or "some examples," etc., refer to specific features, structures, or characteristics described in connection with that embodiment / mode or example, which are included in at least one embodiment / mode or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment / mode or example. Moreover, the specific features, structures, or characteristics described may be combined in any suitable manner in one or more embodiments / modes or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments / modes or examples described in this specification, as well as the features of different embodiments / modes or examples.

[0122] Those skilled in the art should understand that the above embodiments are merely for illustrating the present disclosure and are not intended to limit the scope of the disclosure. Those skilled in the art can make other changes or modifications based on the above disclosure, and these changes or modifications still fall within the scope of the present disclosure.

Claims

1. A supply chain configuration method, characterized in that, The method includes: The received configuration request, wherein the configuration request is issued by any one of the regional stores, suppliers or warehouses; A target constraint rule matching the configuration request is determined from a plurality of constraint rules; wherein the constraint rule includes at least one of business rules, resource rules, and execution rules. The configuration parameters parsed from the configuration request are input into the target constraint rules to calculate the supplier's dynamic weight value; and, Based on the configuration parameters and the supplier's historical data, predict the supplier's capacity gap; The dynamic weight value, the capacity gap, and the supplier's historical data are input into the scheduling decision tree to generate the supplier configuration result, which includes the capacity allocation ratio and logistics routing suggestions for each supplier.

2. The supply chain configuration method according to claim 1, characterized in that, The step of inputting the configuration parameters parsed from the configuration request into the target constraint rules to calculate the supplier's dynamic weight value includes: The configuration parameters obtained from the configuration request, including static supplier data, real-time supplier data, and regional policy parameters, are input into business rules to generate supplier access status and static benchmark weight. The supplier access status is used to screen qualified suppliers, and the static benchmark weight serves as the basis for weight calculation. The configuration parameters and historical default data are input into the resource rules to generate a real-time adjustment factor, which reflects the supplier’s real-time capabilities and performance reliability. The supplier admission status and static benchmark weight, the real-time adjustment factor, and the regional policy parameters are input into the execution rules to generate a supplier dynamic weight value, wherein the dynamic weight value is used to evaluate supplier priority.

3. The supply chain configuration method according to claim 1, characterized in that, The supplier's historical data includes: capacity fluctuation coefficient, quality inspection pass rate, and logistics timeliness standard deviation; The step of predicting the supplier's capacity gap based on the configuration parameters and the supplier's historical data includes: Based on the real-time supplier data and regional policy parameters in the configuration parameters, a spatiotemporal feature matrix is ​​converted; and... Based on the capacity fluctuation coefficient, the quality inspection pass rate, and the logistics timeliness standard deviation, a supplier feature matrix is ​​converted. The spatiotemporal feature matrix and the supplier feature matrix are input into the long short-term memory model to predict the supplier's capacity gap.

4. The supply chain configuration method according to claim 1, characterized in that, The supplier's historical data includes: supplier response speed and cost constraints; The step of inputting the dynamic weight value, the capacity gap, and the supplier's historical data into the scheduling decision tree to generate the supplier configuration result includes: The dynamic weight value, the capacity gap, the supplier response speed, and the cost constraint are input into the scheduling decision tree to generate the supplier configuration result.

5. The supply chain configuration method according to claim 3, characterized in that, The construction of the spatiotemporal feature matrix includes: Based on the real-time supplier data and regional policy parameters in the configuration parameters, a regional dimension is generated. The regional dimension includes sub-regions divided according to geographical, economic or consumption characteristics, and is determined based on order data from the fulfillment database and service scope data from the supplier interface. Based on historical order data and real-time event streams, a time dimension is generated, which includes a prediction time window with hourly granularity to capture the impact of periodic demand and sudden events. The spatiotemporal feature matrix is ​​constructed based on the region dimension and the time dimension.

6. The supply chain configuration method according to claim 1, characterized in that, The step of determining the target constraint rule that matches the configuration request from multiple constraint rules includes: The configuration request is parsed by real-time event stream to extract the request type and configuration parameters. The request type includes a promotional activity request or a regional policy instruction. The configuration parameters include a regional identifier, category information, and event type. Based on the request type and configuration parameters, the similarity with business rules, resource rules and execution rules is calculated. The business rules are based on static supplier data and long-term regional policies, the resource rules are based on real-time supplier data and short-term regional policies, and the execution rules are based on the output of integrating the aforementioned rules. When the similarity exceeds a predetermined threshold, at least one of the matching business rules, resource rules, or execution rules is determined as the target constraint rule.

7. The supply chain configuration method according to claim 1, characterized in that, After generating the vendor configuration results, the following is also included: The propagation impact of configuration changes triggered by the configuration request is monitored through real-time event streaming. These configuration changes include updates to vendor dynamic weight values ​​or configuration results. When a cascading error is detected, the system automatically rolls back to the most recent stable configuration version. The cascading error includes supplier allocation failure or fulfillment costs exceeding a predetermined threshold. The stable configuration version is obtained from the fulfillment database. Blockchain technology is used to record the configuration changes. The logs include records of changes to configuration parameters, dynamic weight values, and configuration results, in order to enable audit traceability.

8. The supply chain configuration method according to claim 1, characterized in that, Also includes: Static supplier data and historical default data are stored in a distributed database based on a sharding storage mechanism; Real-time supplier data and regional policy parameters are cached in a distributed memory caching system and updated in real time through a publish / subscribe mechanism; By triggering the coordinated updates of the distributed memory caching system and the distributed database through real-time event streams, the data consistency of configuration parameters can be achieved within seconds.

9. An electronic device, characterized in that, include: The memory stores execution instructions; as well as A processor that executes execution instructions stored in the memory, causing the processor to perform the method of any one of claims 1 to 8.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 8.

Images