Cross-border e-commerce intelligent multi-warehouse optimal order distribution and delivery system and method

Abstract

The invention discloses a cross-border e-commerce intelligent multi-warehouse optimal order distribution delivery system and method. The system comprises a warehouse inventory management module, a sales order management module, an order distribution logic management module, an inventory occupation level scoring module, an inventory occupation simulation module, a gross interest rate simulation calculation module and an optimal order distribution selection module. The method comprises the steps of obtaining and preprocessing order data; loading an order distribution rule; performing multi-dimensional scoring and sorting on the orders; simulating the occupied inventory according to the priority and calculating the gross profit rate of each order distribution scheme; the scheme meeting the rule is actually subjected to order distribution; and providing a manual intervention interface for the orders which do not meet the conditions to specify a warehouse, and triggering replenishment or allocation. According to the invention, through dynamic priority scoring and a simulation preemption mechanism, automation and optimization of order allocation decision are realized, order processing efficiency and inventory turnover rate are effectively improved, operation cost and gross interest rate are controlled, and performance capability of abnormal orders is enhanced.

Description

Technical Field

[0001] This invention relates to the field of sales order and inventory data processing technology for cross-border e-commerce platforms, specifically to an optimal multi-warehouse order splitting and delivery system and method for cross-border e-commerce. Background Technology

[0002] With the deep integration of global economic integration and internet technology, cross-border e-commerce has become an important part of international trade. To improve logistics efficiency, optimize customer experience, and reduce operating costs, large cross-border e-commerce companies generally establish distributed warehouses (such as warehouses in China and overseas) in multiple countries or regions, forming a wide-coverage multi-warehouse delivery network. In this model, an order may be shipped independently by a single warehouse, or it may be split into multiple sub-orders based on factors such as inventory distribution, logistics costs, and timeliness requirements, with different warehouses coordinating to complete the shipment. This process is called "order splitting."

[0003] However, order allocation decisions in a multi-warehouse environment are a highly complex systemic project. Traditional inventory management and order distribution methods are no longer adequate to meet the dynamic and intelligent operational needs of modern cross-border e-commerce, and mainly face the following technical challenges:

[0004] (1) Order allocation decisions rely on human experience, which is inefficient and difficult to scale: Existing systems mostly allocate orders based on simple rules (such as shipping from the nearest location) or rely on manual review, lacking a comprehensive evaluation of the multi-dimensional attributes of orders. With the surge in order volume, the slow processing speed and poor consistency of manual processing can easily lead to delivery delays or incorrect order allocation, affecting customer satisfaction and platform reputation.

[0005] (2) Inefficient allocation of inventory resources and high warehousing costs: Traditional methods often lack dynamic coordination and simulated pre-occupancy mechanisms for multiple warehouses. This can easily lead to an imbalance where popular warehouses quickly run out of inventory and are forced to replenish frequently, while other warehouses have idle inventory and low turnover rates, resulting in low overall warehousing space utilization and high leasing and holding costs.

[0006] (3) Lack of overall cost-benefit trade-off and uncontrollable gross profit margin: Order splitting decisions directly affect logistics costs, tariffs, and procurement costs. Existing solutions often only consider a single factor (such as the lowest freight cost) and lack simulation calculations and threshold control of the "overall gross profit margin of the order after splitting". This may result in a situation where a certain order splitting solution saves freight costs, but the use of high-cost warehouses or the increase in procurement costs leads to the order profit being lower than acceptable.

[0007] (4) Rigid order priority processing mechanism and insufficient response to urgent orders: Cross-border e-commerce orders are complex and diverse, including expedited orders, holiday promotion orders, influencer collaboration orders, and overdue orders. Traditional systems usually adopt a simple first-come, first-served or fixed-level system, which cannot dynamically calculate and execute priorities based on multiple dimensions of information such as payment method, special customer requirements (such as additional shipping fees), and order status. This results in high-value or urgent orders not being processed in a timely manner, increasing operational risks. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention provides an optimal multi-warehouse order splitting and delivery system and method for cross-border e-commerce. It overcomes the deficiencies of existing technologies by using dynamic priority scoring and simulated pre-occupancy mechanisms to automate and optimize order splitting decisions, effectively improving order processing efficiency and inventory turnover, controlling operating costs and gross profit margins, and enhancing the fulfillment capability of abnormal orders.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A cross-border e-commerce intelligent multi-warehouse optimal order splitting and shipping system and method, including:

[0011] Each warehouse inventory management module is used to query and record the real-time inventory quantity and shelf location information of multiple warehouses, and record the inventory occupancy status, occupancy order number and occupancy time;

[0012] The sales order management module is used to obtain order information for all unshipped sales orders and associate records with the inventory warehouse, shelf space, quantity occupied, and order status of the order.

[0013] The order allocation logic management module is used to configure the availability status, priority allocation strategy, order allocation ratio, order gross profit margin threshold of each warehouse, as well as set priority order allocation conditions and reverse order blocking conditions.

[0014] The inventory occupancy level scoring module is used to score orders based on at least one of the following: transportation method, payment type, order tag attributes, delivery time, holiday orders, and influencer orders, and to calculate the total priority score for each order.

[0015] The simulated inventory occupancy module is used to simulate the occupancy of inventory in the corresponding warehouse according to the warehouse occupancy rules configured by the order allocation logic management module, based on the total priority score from high to low. When preset conditions are met, the actual inventory is occupied and a delivery note is generated.

[0016] The gross profit margin simulation module is used to simulate and calculate the gross profit margin of multi-warehouse split order delivery schemes, and determine whether to execute split order delivery based on the set gross profit margin threshold.

[0017] The optimal order allocation module provides decision support for orders that do not meet the conditions for automatic order allocation, such as manually assigning warehouses, triggering procurement replenishment, or initiating inventory transfers.

[0018] Preferably, in the order allocation logic management module, the priority order allocation conditions include at least one of the following: order timeliness conditions, payment method conditions, order tag type conditions, order product tag type conditions, and shipping method conditions;

[0019] The reverse interception order conditions include at least one of the following: exceeding the maximum number of orders, not meeting the minimum order amount in the warehouse, or the remaining un-ordered amount being lower than a set threshold.

[0020] Preferably, the scoring items of the inventory occupancy level scoring module include at least one or more of the following: additional shipping costs paid by the customer, the shipping method selected by the customer, the payment method, the order marking type, whether the promised delivery time has been exceeded, whether it is a 24-hour delivery order, a holiday order, an influencer order, and whether the estimated gross profit margin exceeds the set value.

[0021] Preferably, when the inventory occupancy level scoring module recalculates the priority score for already assigned orders, it uses the following formula:

[0022] Remaining order score = Original order initial score × Discount coefficient + Bonus points for exceeding the promised waiting time;

[0023] The discount coefficient ranges from 0 to 1, and the bonus points for exceeding the promised waiting time are calculated by adding a set number of points for each day exceeding the limit.

[0024] Preferably, the modules interact with each other via at least one of the following communication methods:

[0025] Inventory query interface based on RESTful API and HTTPS protocol;

[0026] Policy configuration interface based on gRPC and Protocol Buffers;

[0027] A rule execution interface based on a message queue;

[0028] Priority queue based on Redis sorted sets;

[0029] Cost accounting interface based on Apache Thrift;

[0030] Decision interface based on WebSocket.

[0031] Preferably, it also includes an exception handling and data consistency guarantee mechanism, including at least one of the following: distributed transaction control, communication retry and dead letter queue, data version control, end-to-end monitoring and logging.

[0032] This invention also discloses an optimal multi-warehouse order splitting and shipping method for cross-border e-commerce, comprising the following steps:

[0033] S1: Obtain unshipped sales order data and perform standardized preprocessing on the order data;

[0034] S2: Load the order splitting rule configuration, which includes the availability status of each warehouse, the order splitting occupancy ratio, the order gross profit margin threshold, the priority order splitting conditions, and the reverse order blocking conditions;

[0035] S3: Based on the preset scoring rules, score each order in multiple dimensions, calculate the total score of inventory occupancy level, and prioritize the orders according to the total score;

[0036] S4: Perform simulated inventory occupancy operations on orders in order of priority, and simulate the gross profit margin for each order split when there is a possibility of multi-warehouse splitting.

[0037] S5: Execute order splitting decisions based on simulation results, actually occupy inventory for orders that meet the order splitting conditions and generate corresponding delivery notes, and update order splitting status and inventory data;

[0038] S6: For orders that do not meet the conditions for automatic order splitting, add them to the order splitting restriction list and provide a manual intervention interface to support manual specification of the shipping warehouse, triggering procurement replenishment or initiating inventory transfer.

[0039] S7: Update the status of each module in the system, synchronize inventory, order and order split data to ensure data consistency and traceability.

[0040] Preferably, in step S3, the scoring dimensions include at least one of the following:

[0041] Does the customer pay additional shipping fees?

[0042] The shipping method selected in the order;

[0043] Order payment methods;

[0044] Is the order marked as an expedited order?

[0045] Has the order exceeded the promised delivery time?

[0046] Are the orders holiday orders or influencer orders?

[0047] Is the estimated gross profit margin of the order higher than the set value?

[0048] Preferably, in step S4, when simulating inventory occupancy, the simulation is performed according to the priority order of warehouse occupancy set in the order allocation logic management. Actual inventory occupancy is only executed when the inventory is sufficient and the order allocation ratio limit is met.

[0049] Preferably, in step S5, if an order meets multiple priority order splitting conditions, the order splitting is performed according to the combination of conditions, and the system records the order splitting type and execution time used.

[0050] This invention provides an intelligent multi-warehouse optimal order allocation and delivery system and method for cross-border e-commerce. It offers the following advantages: By establishing a multi-dimensional, quantifiable inventory occupancy level scoring mechanism, it can automatically and in real-time prioritize massive orders, replacing the inefficient traditional model that relies on manual experience to judge each order individually. Combined with simulated occupancy and automatic verification by a rule engine, it can calculate compliant and economical delivery plans for orders within seconds, significantly shortening the processing cycle from payment to warehouse acceptance and improving the overall throughput of warehousing operations. Decision-making based on preset, clear business rules completely avoids the arbitrariness and inconsistency that may result from manual operation. The system ensures that orders with the same characteristics always receive the same processing priority and order allocation logic, significantly reducing the operational risks of misallocation and omission, and improving the standardization and reliability of order allocation decisions. Through the simulated occupancy mechanism and multi-warehouse collaboration strategy, the system can intelligently and evenly consume inventory in each warehouse, avoiding a situation where inventory in popular warehouses is quickly depleted while other warehouses accumulate inventory. This improves overall inventory turnover and reduces the need for emergency transfers or expedited replenishment due to localized inventory shortages, thereby lowering warehousing holding costs and ineffective logistics handling costs. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in this invention or the prior art, the accompanying drawings used in the description of the prior art will be briefly introduced below.

[0052] Figure 1 A schematic diagram illustrating the system module composition and data interaction of this invention;

[0053] Figure 2 Overall flowchart of the order splitting and delivery method of the present invention;

[0054] Figure 3 A schematic diagram of the inventory occupancy level scoring mechanism in this invention;

[0055] Figure 4 The flowchart of the simulation decision-making and anomaly handling branches in this invention. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0057] Example 1, as Figures 1 to 4 As shown, this invention provides an optimal multi-warehouse order fulfillment system for cross-border e-commerce. This system adopts a microservice architecture and event-driven design pattern, is deployed in a cloud-native environment, and possesses high availability, scalability, and loose coupling characteristics. The overall architecture consists of three layers:

[0058] Access and Interface Layer: Responsible for connecting with external systems (such as e-commerce platforms, WMS, and logistics provider APIs), and managing and routing requests uniformly through the API gateway.

[0059] Business logic layer: It consists of multiple independent microservices, each of which corresponds to a core functional module of the present invention and interacts through a lightweight communication mechanism.

[0060] Data and infrastructure layer: This includes relational databases, NoSQL databases, caching, message queues, and distributed file storage, providing data persistence and infrastructure support for upper-layer services.

[0061] The system uses containerization technology (such as Docker) for encapsulation and Kubernetes for orchestration and management, enabling elastic scaling and fault self-healing of resources.

[0062] Each core functional module includes:

[0063] Inventory management modules for each warehouse:

[0064] This module acts as a data bridge between the system and physical warehousing facilities, responsible for a panoramic, real-time mapping of the inventory status of all warehouses. It is used to query and record real-time inventory quantities and shelf location information for multiple warehouses, as well as record the inventory occupancy status, occupancy order number, and occupancy time.

[0065] Each warehouse inventory management module maintains data synchronization with local warehouse management systems (WMS) through two main methods: 1) Proactive polling: By calling the RESTful API provided by the WMS, real-time inventory snapshots at the SKU level for each warehouse are retrieved in batches at configurable time intervals (e.g., every 30 seconds), including total quantity, available quantity, locked quantity, and specific shelf location codes. 2) Event listening: Subscribing to the WMS's inventory change message queue (such as a Kafka topic) to receive events such as inbound, outbound, and inventory adjustments in real time, achieving sub-second data updates.

[0066] Internally, a state model is maintained for each inventory unit, including states such as "Available," "Pre-occupied" (simulated occupancy), "Actual Occupancy," and "Frozen." Any simulated or actual occupancy operation is recorded as a "transaction," including transaction ID, associated order number, SKU, warehouse, quantity changed, operation type (simulated / actual), timestamp, and operation context, ensuring full traceability and rollback capability of inventory changes.

[0067] It provides a unified API for inventory query and operation. For example, the POST / api / inventory / simulate / hold interface is used to handle simulated holding requests. This operation only marks the holding request in memory or a temporary database and does not actually affect the WMS inventory. The POST / api / inventory / actual / hold interface is used to execute the actual holding after the decision is confirmed and to synchronously notify the corresponding warehouse management systems (WMS) in various regions.

[0068] Sales Order Management Module:

[0069] This module serves as the central hub for order data aggregation, standardization, and management. It retrieves order information for all unshipped sales orders and associates this information with the corresponding inventory warehouse, shelf space, quantity, and order status. Specific implementation includes:

[0070] 1) Order Acquisition: Deeply integrated with the e-commerce platform's Order Management System (OMS), it uses message middleware (such as RabbitMQ) to monitor events such as "order paid" and "order modified" in real time, or periodically pulls a list of orders to be processed. The raw order data collected covers commercial and technical attributes, such as order ID, user ID, shipping address, product details, sales price, paid shipping fees, payment gateway, coupon information, customer remarks, and internal operational tags.

[0071] 2) Data enrichment and standardization: Cleaning and enhancing raw data. For example, parsing the destination country / region code from the delivery address; mapping the corresponding procurement cost (which may be multi-level cost) to the SKU; converting sales amounts in different currencies to the base currency (such as RMB) using the real-time exchange rate; and identifying and structuring order tags (such as "urgent", "pre-sale", "influencer_order").

[0072] 3) Order Lifecycle Tracking: Create and maintain an order allocation context for each order, continuously updating its status, such as "Pending Allocation," "Simulated Occupancy," "Allocated (Partial / Full)," "Awaiting Manual Intervention," and "Closed." This context serves as the primary basis for all subsequent order allocation decisions.

[0073] Order splitting logic management module:

[0074] This module is a dynamically configurable rules engine that transforms complex business strategies into executable computer logic. It is used to configure the availability status of each warehouse, priority allocation strategies, order allocation ratios, order gross profit margin thresholds, and to set priority order allocation conditions and reverse order blocking conditions; specific implementation includes:

[0075] 1) Visual Rule Configuration: Provides an administrator backend, allowing non-technical personnel to define and combine rules through a graphical interface. Rules typically include: conditions (IF part, such as "Order Destination Country='US' AND Payment Method='PayPal'"), actions (THEN part, such as "Priority Score Increased by 50", "Attempt to ship from the 'US_Warehouse_NJ' warehouse first"), execution order, and scope of effect.

[0076] 2) Strategy Hierarchy and Priority: Rules can be organized hierarchically, for example: global rules > warehouse group rules > specific SKU rules. The rule engine (such as Drools or a self-developed engine) evaluates orders according to priority at runtime, executes all matching rules, and produces a set of decision factors (such as the final warehouse priority list and order splitting permission identifier).

[0077] 3) Reverse Interception Rules: These rules specifically address "when automatic order splitting should not occur." For example, a rule could be defined as: "If the remaining amount after order splitting is less than the product procurement cost × 0.5, then automatic order splitting should be prevented and the order marked as 'uneconomical splitting'." Such rules act as a safety valve to prevent the creation of unreasonably small sub-orders.

[0078] Inventory occupancy level scoring module:

[0079] This module quantifies abstract order attributes into comparable priority values. It scores orders based on at least one of the following: shipping method, payment type, order tag attributes, delivery time, holiday orders, and influencer orders, and calculates the total priority score for each order. Specific implementation includes:

[0080] 1) Dynamic Scoring Model: Scoring rules are dynamically loaded from the order allocation logic management module. Each scoring item is defined as a function. The system iterates through all order attributes and calls the matching function to accumulate the scores. A key implementation example is as follows: For the "order reminder" tag, not only is "whether it exists" considered, but different weights (e.g., 70, 65, 500 points) are assigned based on the source of the reminder (customer, customer service, operations) to reflect differences in internal operational priorities.

[0081] 2) Priority Recalculation Mechanism: For pending orders that have been partially split, the system will periodically (e.g., hourly) recalculate their priority. This is implemented by calling a dedicated recalculation function.

[0082] New total score = original initial total score × attenuation coefficient (0.7) + MAX(0, current time - first order time - promised time limit) × timeout penalty factor (5).

[0083] This ensures that problematic orders can be prioritized dynamically, preventing them from being "forgotten" by the system.

[0084] Simulated inventory occupancy module:

[0085] This module is the core of the system's decision execution. It simulates the allocation of inventory in the corresponding warehouses according to the warehouse occupancy rules configured in the order allocation logic management module, based on the total priority score from highest to lowest. When preset conditions are met, it actually occupies the inventory and generates a delivery note. Specifically, the simulation engine workflow includes:

[0086] a) Input: Receive a rated order and its corresponding "suggested warehouse call sequence" generated by the rules engine.

[0087] b) Sandbox environment: Creates a temporary, isolated inventory view (Snapshot) for the current decision-making cycle, initialized based on the latest real-time data provided by each warehouse inventory management module.

[0088] c) Trial holding: Following the warehouse sequence, attempt to deduct the required quantity of the order from the sandbox inventory in turn. This process requires recursively handling complex combinations of multiple SKUs and multiple warehouses.

[0089] d) Constraint checks: After each trial, immediately check a series of business constraints, such as: whether the warehouse is enabled, whether the number of sub-orders exceeds the limit after splitting orders, and whether the warehouse inventory occupancy ratio exceeds the preset threshold.

[0090] e) Output: Output the first complete "warehouse-SKU-quantity" allocation scheme that passes all constraint checks. If no feasible scheme is found after traversing all possible sequences, output "Simulation failed" and the reason for the failure.

[0091] Module for simulating gross profit margin calculation:

[0092] This module provides economic validation for order splitting decisions, ensuring commercial viability. It simulates and calculates gross profit margins for multi-warehouse order splitting and determines whether to execute order splitting based on a set gross profit margin threshold. Specific cost modeling and calculations include:

[0093] a) Data source: Integrate the procurement system to obtain commodity costs, integrate the logistics rate table (or API) to obtain estimated freight costs, and integrate the tariff calculation engine.

[0094] b) Real-time calculation: When the module receives the order allocation plan provided by the simulation occupancy module, it starts the calculation. The calculation formula is: Total cost = Σ(sub-order product cost) + Σ(sub-order logistics cost) + Σ(sub-order estimated tariffs and other expenses); Total revenue = (total order sales amount + customer-paid shipping fee) × real-time exchange rate; Gross profit margin = (total revenue - total cost) / total revenue × 100%.

[0095] c) Threshold Decision: Compare the calculated gross profit margin with the "acceptable range of order gross profit margin" set in the order allocation logic management module. If it is within the range, approve the plan; otherwise, reject it and return the reasons.

[0096] Select the optimal order splitting module:

[0097] This module, serving as a supplement and upgrade to the automated process, provides decision support for orders that do not meet the conditions for automatic order splitting, including manually specifying warehouses, triggering replenishment purchases, or initiating inventory transfers. Specific implementation includes:

[0098] (1) Abnormal Order Dashboard: Provides a real-time updated web interface that centrally displays all orders that have entered the "Order Segmentation Limit List". The list supports filtering and sorting by multiple dimensions such as failure reason (insufficient inventory, low gross profit margin, rule conflict, etc.), order value, and timeout time.

[0099] (2) Intelligent decision support:

[0100] a. Manual designation: Allows operators to manually override system recommendations and directly designate the shipping warehouse.

[0101] b. Replenishment suggestions: Based on the order SKU and demand time, automatically calculate and display the latest order time, estimated delivery time and associated costs for purchasing from suppliers.

[0102] c. Transfer Simulation: Simulate the transfer of inventory from other warehouses to the out-of-stock warehouse within the system, and calculate the transfer time, logistics costs, and impact on the health of the outgoing warehouse's inventory to assist human decision-making.

[0103] d. Decision Recording and Learning: All decisions, reasons, and results of human interventions are fully recorded. This data can be used for subsequent analysis and can be selectively fed back to the rule engine to optimize automated rules (for example, if a certain human decision-making pattern is found to occur frequently, it can be considered to be solidified into a new automated rule).

[0104] The modules mentioned above do not operate in isolation. For example, when the "Simulated Inventory Management Module" needs to make a decision, it requests the rule output applicable to the order from the "Order Allocation Logic Management Module," obtains a real-time inventory snapshot from the "Warehouse Inventory Management Module," and calls the "Simulated Gross Profit Margin Calculation Module" when necessary. The entire collaborative process is implemented through a combination of event bus and API calls, ensuring the system's elasticity and scalability.

[0105] In addition, lightweight JSON over HTTP / RESTful APIs are used for synchronous calls between modules (such as inventory queries). For time-consuming operations (such as gross profit margin calculations), asynchronous message queues based on RabbitMQ or Kafka are used to achieve decoupling and peak shaving. On the critical path of "simulated occupancy - actual occupancy", optimistic locking or distributed transactions (such as Seata) are used to ensure strong consistency between inventory deduction and order status updates. All operations are logged in detail, supporting rollback and audit traceability.

[0106] This invention automatically quantifies order urgency through an inventory occupancy level scoring module. It also uses a simulated inventory occupancy module and a simulated gross profit margin calculation module to evaluate the inventory feasibility and economics of various order allocation schemes in parallel within seconds. Based on simulation data, the decision-making time for processing thousands of orders can be reduced from hours to minutes, decreasing the need for manual intervention by more than 70%. Furthermore, the system automatically sorts orders according to their total priority score, ensuring that high-value, high-urgency orders are always at the front of the queue and given priority in inventory resource allocation. This dynamic priority mechanism based on multi-dimensional rules (such as payment method, timeliness, and customer tags) transforms the order processing flow from a mechanical assembly line into an intelligent "high-speed emergency lane," significantly improving the fulfillment speed of core orders and customer satisfaction compared to traditional "first-come, first-served" or simple fixed priority models.

[0107] The order allocation logic management module allows for flexible configuration of the priority and allocation ratio of each warehouse, guiding intelligent order flow distribution. Combined with a simulated occupancy mechanism, the system tends to balance the consumption of inventory across warehouses while meeting timeliness and cost requirements, avoiding uneven distribution. This effectively revitalizes the assets of peripheral warehouses or warehouses with high inventory levels, reduces temporary external warehousing costs caused by a single warehouse becoming overwhelmed, and is expected to increase overall inventory turnover by 15%-30%.

[0108] The inventory occupancy level scoring module assigns high scores to risk items such as "delivery time exceeded," driving the system to proactively prioritize such orders. The pre-positioning mechanism of the simulated inventory occupancy module ensures the feasibility of decisions and prevents the embarrassment of "no inventory after allocation." For orders that cannot be processed automatically, the anomaly dashboard and intervention tools provided by the optimal order splitting module ensure that no order is "lost," fundamentally controlling overdue and missed shipment rates. By reverse-intercepting order splitting conditions (such as minimum order amount limits), the system can automatically intercept "fragmented orders" that may generate extremely high after-sales risks due to splitting. For orders with high payment risks (such as specific credit card payments), special processing logic can be assigned through rules. These mechanisms seamlessly embed business risk control strategies into the order splitting execution process, preventing problems before they occur.

[0109] By simulating the "simulated pre-occupancy-verification-actual submission" transactional process used in the inventory occupancy module, and combining it with distributed transactions or optimistic locking mechanisms, fatal errors such as "overselling inventory" or "duplicate order occupancy" that may occur in high-concurrency scenarios are fundamentally avoided, ensuring the accuracy of the most critical inventory data in e-commerce operations. The introduction of asynchronous message queues decouples modules, preventing a brief failure of a single module from causing the entire order allocation process to collapse. Exception handling and data consistency guarantee mechanisms (such as retries, dead-letter queues, and degradation strategies) ensure that the system can self-repair or gracefully degrade in the event of local network fluctuations or service anomalies, maintaining the availability of core services.

[0110] Example 2: This invention also discloses an optimal multi-warehouse order splitting and shipping method for cross-border e-commerce, characterized by the following steps:

[0111] S1: Order data acquisition and preprocessing; specifically including:

[0112] S11: Triggering and Input: The process is triggered by the scheduler or event listener. The sales order management module then pulls all sales order data with a status of "pending processing" or "pending order allocation" from the connected e-commerce platform's order management system (OMS).

[0113] S12: Data Extraction: Obtain the core fields of the order, including but not limited to: unique order number, shipping address (country, postal code), product details (SKU, quantity), sales amount (original currency), payment method, customer-selected logistics service level, order creation time, the platform's promised latest delivery time, and all manually or system-added tags.

[0114] S13: Data Standardization: Transform heterogeneous raw data into a unified format within the system. For example, map country names to standard country codes, convert various currency amounts to a base accounting currency (such as RMB) using real-time exchange rates, and map vague logistics service descriptions to "transportation mode" enumeration values ​​defined within the system.

[0115] S14: Rich Data: Link the product master data to obtain the procurement cost of each SKU (which may use a weighted average cost depending on the procurement batch). Then, load the processed order data into the "Pending Order Pool" for this batch processing. Each order object in the pool contains all the structured information needed for subsequent decision-making.

[0116] S2: Order splitting rule configuration and loading; specifically including:

[0117] S21: Rule Loading: After system initialization or rule changes, the unit logic management module loads all currently effective configuration rules into the rule engine in memory. These rules exist in interpretable data structures (such as decision trees and rule sets) to ensure high-speed execution.

[0118] S22: Strategy Ready: This step enables the entire process to run based on the latest business strategy, which includes: the start / stop status and default priority of each warehouse, the order splitting ratio and amount threshold, the gross profit margin requirements for global and specific order types, and the detailed rule expressions that constitute the conditions for priority order splitting and reverse interception.

[0119] S3: Inventory occupancy priority score; specifically including:

[0120] S31: Scoring Calculation: The inventory occupancy level scoring module iterates through each order in the "Pending Order Pool". For each order, it submits its attributes (such as shipping method = "express", and tags containing "holiday promotion") as facts to the rules engine. The engine matches the scores according to the scoring rules loaded in step S2, performs the corresponding scoring operation, and finally returns the "total inventory occupancy priority score" for that order.

[0121] S32: Queue Sorting: After all orders have been scored, the system sorts them in descending order based on their total scores, generating a "priority processing queue" for this process. The order with the highest total score is placed at the front of the queue and will receive priority for inventory allocation attempts. This queue is dynamic; if an order is re-scored in subsequent steps, its position may change.

[0122] S4: Simulated inventory holding and gross profit margin calculation (core decision-making loop);

[0123] The system processes each order sequentially, starting from the head of the "priority processing queue." For the current order:

[0124] 1) Obtain candidate warehouse sequence: Consult the order allocation logic management module to obtain a recommended warehouse sequence list based on the order destination country, product SKU, warehouse status and priority strategy, for example: [Warehouse A (overseas warehouse), Warehouse B (domestic central warehouse)].

[0125] 2) Simulation and verification; specifically including:

[0126] a. Single-warehouse fulfillment attempt: The system first attempts to fully fulfill the order using the first warehouse in the list (e.g., warehouse A). In a "simulated environment" in memory or database, the available inventory of that warehouse is checked and pre-deducted.

[0127] b. Business rule verification: Check whether this operation violates any configuration rules, such as whether it causes the warehouse to exceed the upper limit for the proportion of split orders for a single order.

[0128] c. Gross Profit Margin Simulation Calculation: Call the gross profit margin simulation calculation module to calculate the estimated order gross profit margin based on this shipping plan (all orders are shipped from warehouse A).

[0129] d. Decision point: If inventory is sufficient, rule validation is passed, and gross profit margin is ≥ set threshold, then the solution is marked as "feasible solution," the current order's trial loop is terminated, and step S5 is initiated. Otherwise, the simulated inventory is released, and the next warehouse in the list is tried.

[0130] 3) Multi-warehouse order splitting attempt: If a single warehouse cannot fulfill the order (insufficient inventory), the system will initiate "order splitting logic": attempting to use a combination of the top N warehouses in the list to jointly fulfill the order (e.g., warehouse A sends part of the order, and warehouse B sends the remaining part). Similarly, simulated inventory usage, rule validation, and gross profit margin calculation are performed for each splitting combination.

[0131] 4) Looping and Termination: Loop through the warehouse list and possible splitting combinations until a "feasible solution" is found. If no feasible solution is found after traversing all possible combinations, mark the order as "no feasible solution found" in this round of processing.

[0132] S5: Order allocation decision and actual inventory usage;

[0133] For orders where a "feasible solution" is successfully found in step S4, the system initiates a database transaction. The actual operation includes:

[0134] 1) Convert the inventory pre-occupied in the simulated environment into formal inventory lock requests to each warehouse inventory management module, actually deduct the available inventory quantity, and record the occupation details.

[0135] 2) Generate one or more sub-shipment orders based on the plan, with each shipment order associated with a specific warehouse and product details.

[0136] 3) Update the status of the sales order to "splitting" and record the details of the splitting scheme used (e.g., splitting type, warehouse used, quantity shipped from each warehouse, and calculated gross profit margin).

[0137] After completing all the above database operations, commit the transaction to ensure data consistency. The shipping order information is automatically sent to the corresponding warehouse's WMS via the interface, triggering subsequent picking, packing, and outbound processes.

[0138] S6: Exceptional Order Processing and Manual Intervention; specifically includes:

[0139] S61: List Aggregation: Move all orders marked as "No feasible solution found" in step S4 into a dedicated "Order Splitting Limitation List". The system will record a clear failure reason code for each order (e.g., "Insufficient inventory in all candidate warehouses", "Gross profit margin of any order splitting solution is below the threshold").

[0140] S62: Manual Intervention Platform: This list is displayed to operations personnel in real time through a graphical interface. Interface provided:

[0141] 1) Diagnostic View: Clearly displays order details, detailed reasons for failed attempts, and key results from simulation calculations.

[0142] 2) Intervention tools: Provide a series of manual decision-making buttons, such as: "Force delivery from warehouse XX" (covering system logic), "Create emergency procurement request with one click", and "Initiate inter-warehouse transfer suggestion" (the system recommends transfer routes based on inventory distribution and cost calculation).

[0143] S63: Closed-loop feedback: After the operations staff makes an intervention decision, the decision is received by the system as a new, high-priority temporary rule. The system can automatically or manually trigger the reprocessing of the specific order starting from step S3 or S4, thus forming a closed loop of "automatic failure → manual diagnosis → decision intervention → system re-execution".

[0144] S7: System status update and data synchronization; specifically including:

[0145] S71: Status Summary: After this round of batch processing is completed, the system automatically updates the monitoring dashboard and summarizes key indicators, such as: total number of orders processed this time, number of orders successfully automatically split, number of orders transferred to the exception list, and inventory consumption overview of each warehouse.

[0146] S72: Data Synchronization: Through message middleware (such as Kafka) or API callbacks, important status change events (such as "orders have been split" and "inventory has been locked") are notified in real time to relevant upstream and downstream systems, including OMS, financial settlement system, customer service system, etc., to ensure the final consistency of data status throughout the entire business chain.

[0147] S73: Logs and Traceability: Fully record the start time, end time, summary logs for each step, and all exception information of this workflow instance for performance analysis, audit traceability, and system optimization.

[0148] This embodiment establishes a multi-dimensional, quantifiable inventory occupancy level scoring mechanism, enabling automatic and real-time prioritization of massive orders, replacing the inefficient traditional model that relies on manual experience for order-by-order judgment. Combined with simulated occupancy and automatic verification by a rule engine, it can calculate compliant and economical delivery plans for orders within seconds, significantly shortening the order processing cycle from payment to warehouse receipt and improving overall warehousing throughput. Decisions are made based on preset, clear business rules (such as priority conditions and interception conditions), completely avoiding the arbitrariness and inconsistencies that can result from manual operation. The system ensures that orders with the same characteristics always receive the same processing priority and order allocation logic, significantly reducing the operational risks of misallocation and omission, and improving the standardization and reliability of order allocation decisions.

[0149] By simulating inventory occupancy and employing a multi-warehouse collaboration strategy, the system can intelligently and evenly deplete inventory across all warehouses, preventing situations where popular warehouses are quickly exhausted while other warehouses experience inventory buildup. This improves overall inventory turnover and reduces the need for emergency transfers or expedited replenishment due to localized inventory shortages, thereby lowering warehousing holding costs and ineffective logistics handling costs.

[0150] By constructing a closed-loop anomaly handling and manual intervention mechanism, order fulfillment resilience is greatly enhanced, and operational risks are reduced. The system does not simply discard orders that cannot be automatically processed; instead, it clearly categorizes them into an "order allocation limit list" and explicitly marks the reasons for failure (such as insufficient inventory or low gross profit margin). This provides operations personnel with precise early warning and intervention entry points, transforming reactive firefighting into proactive pre-emptive or in-process management, effectively avoiding the serious risks of orders becoming silent due to system inability to process them, ultimately leading to overdue or missed shipments. The provided manual intervention interfaces (such as forcibly specifying warehouses and triggering replenishment) do not bypass the system but are fed back to the system as higher-priority inputs, driving it to re-execute automated processes. This forms a highly efficient closed loop of "automatic processing as the primary method and manual judgment as a supplementary method," leveraging the computational efficiency of machines while retaining the flexibility of humans in handling complex anomalies, significantly improving the final order fulfillment rate and customer satisfaction.

[0151] Example 3: To make the technical solution, workflow, and beneficial effects of the present invention clearer and more specific, the implementation process of the system and method of the present invention will be described in detail below with reference to a typical application scenario.

[0152] 1. Scene setting

[0153] Platform and Warehouses: A cross-border e-commerce platform has shipping warehouses in Jiaxing, China (CNJX), the East Coast of the United States (USEC), and the United Kingdom (UK).

[0154] Products and Inventory:

[0155] SKU: New Autumn / Winter Women's Coat, Style Number "COAT-2023".

[0156] Attributes: This SKU is available in different sizes (S, M, L) and colors (black - BLK, khaki - KHK). The inventory system manages inventory by "SKU-size-color".

[0157] Real-time inventory snapshot:

[0158] CNJX warehouse: BLK-S(20), BLK-M(15), BLK-L(10); KHK-S(25), KHK-M(20), KHK-L(15).

[0159] USEC warehouse: BLK-S(5), BLK-M(8), BLK-L(4); KHK-S(2), KHK-M(6), KHK-L(3).

[0160] UK warehouse: BLK-S(8), BLK-M(10), BLK-L(5); KHK-S(12), KHK-M(15), KHK-L(8).

[0161] System configuration (preset via the order splitting logic management module):

[0162] Priority allocation strategy: North American orders will be prioritized for USEC warehouses, European orders will be prioritized for UK warehouses, and other regions and backup orders will use CNJX warehouses.

[0163] Order splitting ratio and rules: A single order can be split into a maximum of 2 sub-orders (to consider customer experience and avoid excessive packages). Special rule: Items of the same style and color but different sizes within the same order should be shipped together as much as possible to avoid confusion for customers.

[0164] Gross profit margin threshold: 25%.

[0165] Example of scoring rules:

[0166] Orders containing "pre-sale items": +50 points (stock needs to be locked first).

[0167] Orders marked as "Social Media Influencer Promotion": +100 points (guaranteed promotion timeliness).

[0168] Customer is a "Black Card Member": +60 points.

[0169] The order was placed during the Black Friday sale: +40 points.

[0170] The product is a "seasonal featured item": +30 points.

[0171] Reverse interception rules: Sub-orders with incomplete sizes (such as only sending one size) require manual review.

[0172] Order information: The system has received an order (Order-F123) from Paris, France.

[0173] Products: 1 piece of COAT-2023 (BLK, size S); 1 piece of COAT-2023 (BLK, size M).

[0174] Amount: Total order amount €280 (equivalent to ¥2184 at an exchange rate of 7.8).

[0175] Attributes: The customer is a Black Card member, this coat is a promotional item for a current social media influencer, and the order has been marked by the operations team as an "urgent order for the same style as the influencer".

[0176] Procurement cost: ¥400 per coat.

[0177] 2. System Intelligent Order Allocation and Execution Process

[0178] Steps S1-S3: Order Acquisition and Rating Sorting:

[0179] The sales order management module captures Order-F123 and identifies it as containing two specific attribute items: COAT-2023-BLK-S and COAT-2023-BLK-M.

[0180] The inventory occupancy level scoring module calculates the total priority score:

[0181] Base score: 100 points

[0182] Black Card Member: +60 points

[0183] Influencer Promotion: +100 points

[0184] Season's featured item: +30 points

[0185] Priority score = 100 + 60 + 100 + 30 = 290 points

[0186] The order was prioritized with a high score of 290.

[0187] Step S4: Simulate the decision-making process

[0188] 1) First attempt (ship all items from the nearest warehouse):

[0189] a. Based on the "European orders prioritize UK warehouse" strategy, simulate the use of 1 BLK-S and 1 BLK-M item in the UK warehouse.

[0190] b. Inventory check: UK warehouse has 8 BLK-S units in stock and 10 BLK-M units in stock. Simulated usage was successful.

[0191] c. Triggering consolidated shipment rule check: Two items of the same style and color are shipped from the same warehouse, which fully comply with the "consolidate shipments as much as possible" rule.

[0192] d. Simulate and calculate gross profit margin:

[0193] Logistics costs (from the UK to France, express delivery): Assuming ¥200.

[0194] Total cost = Procurement cost (2 × 400 = ¥800) + Logistics cost (¥200) = ¥1000.

[0195] Gross profit margin = (¥2184 - ¥1000) / ¥2184 ≈ 54.2%

[0196] e. Decision: Inventory, rules, and gross profit margin are all approved. This is the optimal solution.

[0197] The system will not try other order splitting schemes and will proceed directly to step S5.

[0198] Step S5: Actual Occupation and Execution:

[0199] 1) The system actually occupies 1 COAT-2023-BLK-S and BLK-M inventory in the UK warehouse.

[0200] 2) Generate a shipping order containing two items to be shipped from the UK warehouse to Paris, France.

[0201] 3) The order status is updated to "Order split / Type: Single warehouse shipment / Warehouse: UK".

[0202] 3. Scenario Variation 1: Initially, there is a partial stockout in the warehouse, triggering a game between intelligent order allocation and rules:

[0203] Assuming other conditions remain unchanged, but the BLK-M inventory in the UK warehouse is 0.

[0204] First attempt failed: The system recorded "Insufficient stock in the preferred warehouse (BLK-M missing)".

[0205] Second attempt (intelligent order splitting, while simultaneously evaluating consolidated shipping rules):

[0206] The system needs to split orders. It evaluates two order splitting options:

[0207] Option A (Nearest + Remote Warehouse): S codes are shipped from the UK warehouse, and M codes are shipped from the CNJX warehouse.

[0208] Option B (cross-overseas warehouses): S codes are sent from the UK warehouse, and M codes are sent from the USEC warehouse (inter-overseas warehouse transfer).

[0209] Simulation of Scheme A:

[0210] Inventory check: Passed.

[0211] Rule check: Violation of the rule "same style and color should be shipped together as much as possible" will generate a warning mark in the system, but it is not absolutely prohibited.

[0212] Gross profit margin calculation: Assuming UK to France costs ¥100, CNJX to France costs ¥150, and total logistics cost is ¥250. Gross profit margin = (2184 - 800 - 250) / 2184 ≈ 51.9%. Passed.

[0213] Simulation of Option B:

[0214] Inventory check: Passed.

[0215] Rule check: Also violates the consolidated shipment rule.

[0216] Gross profit margin calculation: Assuming USEC shipping to France has extremely high costs, at ¥350. Total logistics costs: ¥450.

[0217] Gross profit margin = (2184 - 800 - 450) / 2184 ≈ 42.8%. Passes but is lower than Plan A.

[0218] System decision: Both options A and B are feasible. Based on the secondary principle of "cost priority" (or comprehensive scoring), the system selects option A because it has a higher gross margin and the total cost of shipping from the Chinese warehouse is generally more controllable. However, the system will record in the order notes "Due to inventory distribution, orders of the same color were forced to be split."

[0219] 4. Scenario Variation Two: Gross Profit Margin Anomalies and Manual Intervention:

[0220] Assuming the above scenario variant A is adopted, but the platform sets an extremely high "logistics risk surcharge" for France on that day, the logistics cost from CNJX to France skyrockets to ¥400.

[0221] The total logistics cost for Option A is ¥100 (UK) + ¥400 (CNJX) = ¥500.

[0222] Gross profit margin = (2184 - 800 - 500) / 2184 ≈ 40.5%.

[0223] Gross profit margin verification: 40.5% > 25% threshold, still passes. Orders will be automatically executed according to Plan A.

[0224] In a more extreme case: if logistics costs cause the gross profit margin to fall below 25%, then Option A will be rejected.

[0225] The system will try option B (UK+USEC). If its gross profit margin is also below the threshold, all automatic order splitting options will fail.

[0226] Proceed to step S6 (human intervention):

[0227] The order has been added to the "Order Splitting Limit List" because: "The gross profit margin of all feasible order splitting schemes is lower than the threshold".

[0228] The operations staff saw this high-priority order (influencer + black card member) in the optimal order allocation module. The system provides analysis:

[0229] "The current optimal simulation scheme is: UK(S) + CNJX(M), with a gross profit margin of 22%, which is below the threshold."

[0230] "Recommendations: a) Manually authorize shipment according to this plan (given the high priority of the order); b) Immediately check if there is any BLK-M inventory available for allocation in the UK or nearby warehouses."

[0231] The operations staff selected suggestion b, found inventory in a nearby warehouse, immediately created an emergency transfer order from the nearby warehouse to the UK warehouse, and set Order-F123 so that the UK warehouse would consolidate and ship the goods after the transfer was completed. The system synchronously updated the order status to "awaiting transfer, reserved inventory".

[0232] In summary, this embodiment demonstrates how the system of the present invention effectively addresses the specific challenges of order segmentation for apparel categories:

[0233] 1) Refined inventory management: The system accurately connects to the three-level inventory of "SKU-size-color" to ensure the accuracy of order splitting operations.

[0234] 2) Category-specific rule adaptation: By setting category-specific rules such as "shipping the same style and color together", we can balance operational efficiency and customer experience.

[0235] 3) Prioritize high-value orders: Through high-weight scoring items, ensure that scarce inventory and transportation capacity are prioritized to serve high-value, high-impact orders.

[0236] 4) Complex decision-making game: When the inventory distribution is not ideal, it can intelligently weigh and play games among multiple constraints such as "consolidated delivery rules", "logistics costs" and "gross profit margin" to select the relatively optimal solution.

[0237] 5) Provide business-oriented intervention support: When the system encounters hard conflicts (such as insufficient gross profit margin), it can provide operational staff with decision support with business insights (such as suggested allocation routes), rather than just technical error reports.

[0238] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A cross-border e-commerce intelligent multi-warehouse optimal order splitting and delivery system, characterized in that: include: Each warehouse inventory management module is used to query and record the real-time inventory quantity and shelf location information of multiple warehouses, and record the inventory occupancy status, occupancy order number and occupancy time; The sales order management module is used to obtain order information for all unshipped sales orders and associate records with the inventory warehouse, shelf space, quantity occupied, and order status of the order. The order allocation logic management module is used to configure the availability status, priority allocation strategy, order allocation ratio, order gross profit margin threshold of each warehouse, as well as set priority order allocation conditions and reverse order blocking conditions. The inventory occupancy level scoring module is used to score orders based on at least one of the following: transportation method, payment type, order tag attributes, delivery time, holiday orders, and influencer orders, and to calculate the total priority score for each order. The simulated inventory occupancy module is used to simulate the occupancy of inventory in the corresponding warehouse according to the warehouse occupancy rules configured by the order allocation logic management module, based on the total priority score from high to low. When preset conditions are met, the actual inventory is occupied and a delivery note is generated. The gross profit margin simulation module is used to simulate and calculate the gross profit margin of multi-warehouse split order delivery schemes, and determine whether to execute split order delivery based on the set gross profit margin threshold. The optimal order allocation module provides decision support for orders that do not meet the conditions for automatic order allocation, such as manually assigning warehouses, triggering procurement replenishment, or initiating inventory transfers.

2. The optimal multi-warehouse order splitting and shipping system for cross-border e-commerce according to claim 1, characterized in that: In the order allocation logic management module, the priority order allocation conditions include at least one of the following: order timeliness conditions, payment method conditions, order tag type conditions, order product tag type conditions, and shipping method conditions; The reverse interception order conditions include at least one of the following: exceeding the maximum number of orders, not meeting the minimum order amount in the warehouse, or the remaining un-ordered amount being lower than a set threshold.

3. The optimal multi-warehouse order splitting and shipping system for cross-border e-commerce according to claim 1, characterized in that: The scoring items of the inventory occupancy level scoring module include at least one or more of the following: additional shipping costs paid by the customer, the shipping method selected by the customer, the payment method, the order marking type, whether the promised delivery time has been exceeded, whether it is a 24-hour delivery order, a holiday order, an influencer order, and whether the estimated gross profit margin exceeds the set value.

4. The optimal multi-warehouse order splitting and shipping system for cross-border e-commerce according to claim 1, characterized in that: When the inventory occupancy level scoring module recalculates the priority score for already assigned orders, it uses the following formula: Remaining order score = Original order initial score × Discount coefficient + Bonus points for exceeding the promised waiting time; The discount coefficient ranges from 0 to 1, and the bonus points for exceeding the promised waiting time are calculated by adding a set number of points for each day exceeding the limit.

5. The optimal multi-warehouse order splitting and shipping system for cross-border e-commerce according to claim 1, characterized in that: The modules interact with each other through at least one of the following communication methods: Inventory query interface based on RESTful API and HTTPS protocol; Policy configuration interface based on gRPC and Protocol Buffers; A rule execution interface based on a message queue; Priority queue based on Redis sorted sets; Cost accounting interface based on Apache Thrift; Decision interface based on WebSocket.

6. The optimal multi-warehouse order splitting and shipping system for cross-border e-commerce according to claim 1, characterized in that: It also includes exception handling and data consistency guarantee mechanisms, including at least one of the following: distributed transaction control, communication retry and dead letter queue, data version control, end-to-end monitoring and logging.

7. A method for optimal order splitting and shipping in cross-border e-commerce using intelligent multi-warehouse systems, characterized by: Includes the following steps: S1: Obtain unshipped sales order data and perform standardized preprocessing on the order data; S2: Load the order splitting rule configuration, which includes the availability status of each warehouse, the order splitting occupancy ratio, the order gross profit margin threshold, the priority order splitting conditions, and the reverse order blocking conditions; S3: Based on the preset scoring rules, score each order in multiple dimensions, calculate the total score of inventory occupancy level, and prioritize the orders according to the total score; S4: Perform simulated inventory occupancy operations on orders in order of priority, and simulate the gross profit margin for each order split when there is a possibility of multi-warehouse splitting. S5: Execute order splitting decisions based on simulation results, actually occupy inventory for orders that meet the order splitting conditions and generate corresponding delivery notes, and update order splitting status and inventory data; S6: For orders that do not meet the conditions for automatic order splitting, add them to the order splitting restriction list and provide a manual intervention interface to support manual specification of the shipping warehouse, triggering procurement replenishment or initiating inventory transfer. S7: Update the status of each module in the system, synchronize inventory, order and order split data to ensure data consistency and traceability.

8. The method for splitting orders and shipping according to claim 7, characterized in that: In step S3, the scoring dimensions include at least one of the following: Does the customer pay additional shipping fees? The shipping method selected in the order; Order payment methods; Is the order marked as an expedited order? Has the order exceeded the promised delivery time? Are the orders holiday orders or influencer orders? Is the estimated gross profit margin of the order higher than the set value? 9. The method for splitting orders and shipping according to claim 7, characterized in that: In step S4, when simulating inventory occupancy, the simulation is performed according to the priority order of warehouse occupancy set in the order allocation logic management. Actual inventory occupancy is only executed when the inventory is sufficient and the order allocation ratio limit is met.

10. The method for splitting orders and shipping according to claim 7, characterized in that: In step S5, if an order meets multiple priority order splitting conditions, the order splitting is performed according to the combination of conditions, and the system records the order splitting type and execution time used.

Images