Intelligent warehouse division scheduling and dynamic route planning system for cross-border e-commerce

Abstract

The invention discloses an intelligent warehouse division scheduling and dynamic route planning system for cross-border e-commerce, particularly relates to the field of logistics scheduling, is used for solving the problems of order warehouse division and transportation path optimization in cross-border e-commerce big promotion time, and is characterized in that contract freight rate and real-time tariff are fused through a double-layer cost sensing matrix, and freight rate fluctuation is captured; self-adaptive tensor decomposition is carried out to extract features on line, and the order splitting priority and the routing are cooperatively updated; injecting burst cost by a transport capacity probe, identifying a volatile flight segment in combination with a tariff amplitude and a cargo flow gradient, and remodeling a node sequence and a time window; a tracking estimation error is judged when the budget is stable, bins and paths are solidified synchronously, and transport capacity interruption is avoided; the recalculation frequency is optimized and reduced, the performance period is shortened, the punctuality rate and the fund efficiency are improved, and high toughness and transparency of the cross-border supply chain are achieved.

Description

Technical Field

[0001] This invention relates to the field of logistics scheduling, and more specifically, to an intelligent warehouse scheduling and dynamic routing planning system for cross-border e-commerce. Background Technology

[0002] During major cross-border e-commerce promotions, the intelligent warehouse dispatching and dynamic routing planning architecture needs to simultaneously determine the destination of forward warehouses and cross-border transportation routes for massive orders within an extremely short window. The architecture maps warehouse-ship segment-delivery node as a spatiotemporal network, uses a two-layer cost perception matrix to integrate the contract freight rate surface with the real-time flexible tariff surface, and then injects inventory balance, port release takt time, and capacity availability into the incremental recalculation link through rolling forecasts to keep warehouse dispatching instructions and route topology updated in sync.

[0003] However, there is a time lag between flexible capacity pricing and rigid contract freight rates. When the market suddenly sees low-priced LCL shipments or temporary chartering, contract freight rates remain unchanged due to inertia. Warehouse allocation scheduling and dynamic routing planning rely on distorted cost benchmarks, leading to deviations in warehouse network splitting and route selection from optimal practices. During execution, this necessitates a second recalculation and capacity replacement, resulting in budget overruns, disrupted loading / unloading sequences, and delivery delays. Intelligent warehouse allocation scheduling and dynamic routing planning thus expose the core technical challenge of the mismatch between rigid freight rate models and flexible tariff curves.

[0004] To address the aforementioned problems, a technical solution is provided. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an intelligent warehouse scheduling and dynamic route planning system for cross-border e-commerce. This system captures freight rate fluctuations by fusing contract freight rates and real-time tariffs through a two-layer cost-aware matrix; it extracts features online through adaptive tensor decomposition and collaboratively updates order splitting priorities and routes; it injects sudden costs using a capacity probe, identifies vulnerable flight segments by combining tariff fluctuations and cargo flow gradients, and reshapes node sequences and time windows; it tracks valuation errors through budget stability determination, synchronously solidifies warehouse allocation and routes, and avoids capacity gaps; it optimizes and reduces recalculation frequency, shortens fulfillment cycles, improves on-time performance and capital efficiency, and achieves high resilience and transparency in the cross-border supply chain, thereby solving the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The intelligent warehouse scheduling and dynamic routing planning system for cross-border e-commerce includes: a cost matrix generation module, a tensor decomposition optimization module, a capacity probe adjustment module, and a budget stability correction module. Cost matrix generation module: Generates a two-layer cost perception matrix at the warehouse allocation decision entry point. One layer records the static surface of contract freight rates, and the other layer captures the flexible capacity tariff curve in real time. The warehouse-segment mapping is locked through hash index. Tensor decomposition optimization module: Uses adaptive tensor dimensionality reduction search to perform streaming decomposition on the two-layer cost perception matrix, extracts pulse fluctuation indexes and adjusts the order splitting priority according to the index, and outputs a draft of the warehouse allocation. Capacity probe adjustment module: Injects real-time capacity probe streams into the routing topology, superimposes sudden cost increments on each trunk edge of the draft, and if the stability metric obtained by fusing price fluctuation characteristics and cargo flow gradient characteristics is lower than the standard, the edge routing reconfiguration operator is called and the node sequence and occupancy window are updated synchronously. Budget stability correction module: The budget stability arbiter calculates the segmented integral of the difference between the updated end-to-end valuation and the original valuation. If the integral exceeds the threshold, the difference is re-injected to compensate and the final allocation and path list is solidified.

[0007] In a preferred embodiment, the cost matrix generation module includes the following: By integrating the static surface of contract freight rates and the flexible capacity and tariff surface, a two-layer cost-aware matrix with a two-layer structure is generated. The static surface of contract freight rates is a two-dimensional dataset based on long-term contract price data, and the flexible capacity and tariff surface is a dynamic two-dimensional dataset based on real-time market price data. At the same time, the mapping relationship between warehouses and flight segments is locked through a hash mapping table, with the combination of warehouses and flight segments as the key and the location index in the two-layer cost-aware matrix as the value.

[0008] In a preferred embodiment, the tensor decomposition optimization module includes the following: The two-layer cost perception matrix is ​​decomposed using an adaptive tensor dimensionality reduction flow decomposition module. This decomposes the two-layer cost perception matrix into a combination of multiple low-dimensional tensors, and extracts the pulse fluctuation index of freight rate fluctuations. Priority scores are calculated based on the pulse fluctuation index and the deviation between static and dynamic freight rates, and the order splitting priorities are rearranged from high to low according to the priority scores. An elastic threshold is set based on the median and interquartile range of the priority scores, and warehouse and segment combinations with priority scores higher than or equal to the elastic threshold are selected to generate a draft warehouse splitting plan with elastic thresholds.

[0009] In a preferred embodiment, the tensor decomposition optimization module further includes the following: The priority score calculation process takes into account both the pulse fluctuation index and the deviation between static and dynamic fares. Specifically, the ratio of static to dynamic fares is first calculated, and then this ratio is input into an exponential function to generate an amplification factor. Subsequently, the pulse fluctuation index is multiplied by this amplification factor to obtain the priority score.

[0010] In a preferred embodiment, the capacity probe adjustment module includes the following: The latest market data is collected in real time through external interfaces to form a real-time capacity probe stream. Sudden cost increments are superimposed on each trunk line in the draft of the cargo allocation. Based on the historical freight rate sequence in the real-time capacity probe stream, the probability distribution of the freight rate change rate at adjacent times is calculated and combined with the time decay factor to generate the segment expansion and contraction entropy.

[0011] In a preferred embodiment, the capacity probe adjustment module further includes the following: The rate of change kurtosis is calculated based on the freight flow sequence of real-time freight flow data, and the freight flow emergence degree is generated by combining the maximum absolute value.

[0012] In a preferred embodiment, the capacity probe adjustment module further includes the following: The cost stabilization coefficient is generated by normalizing the segment expansion / contraction entropy and cargo flow emergence, taking the complement, and weighting the average. Based on the cost stabilization coefficient, low-stabilization segments are identified and the edge route reconfiguration operator is triggered to update the node sequence and the occupancy window.

[0013] In a preferred embodiment, the capacity probe adjustment module further includes the following: If the cost stabilization coefficient of a certain trunk line is lower than the safety threshold, the corresponding trunk line is judged as a low-stabilization flight segment.

[0014] In a preferred embodiment, the budget stabilization correction module includes the following: The impact of route reconfiguration on costs is assessed by calculating the segmented integral of the difference between the full-link valuation after reconfiguration and the original valuation. Specifically, the full link is divided into multiple key segments. For each key segment, the difference between the valuation after reconfiguration and the original valuation is calculated within a specific time window and integrated to obtain the valuation difference integral for the corresponding key segment. Then, the valuation difference integrals of all key segments are summed to obtain the total integral for the entire link.

[0015] In a preferred embodiment, the budget stabilization correction module further includes the following: When the total score exceeds the preset safety limit, the compensation difference is calculated and evenly distributed to each trunk edge in the reconfigured routing topology. The compensation difference is divided by the total number of trunk edges to obtain the adjustment amount for each trunk edge. This adjustment amount is then added to the corresponding trunk edge cost in the flexible capacity tariff surface layer of the two-layer cost perception matrix to complete the cost injection adjustment. Finally, the reconfigured routing topology and the adjusted two-layer cost perception matrix are locked as the final warehouse and route list.

[0016] The technical effects and advantages of the intelligent warehouse scheduling and dynamic routing planning system for cross-border e-commerce proposed in this invention are as follows: This invention deeply integrates the static contract freight rate with the real-time flexible tariff through a dual-layer cost perception matrix, capturing freight rate fluctuations from warehouse to flight segment in real time. An adaptive tensor decomposition mechanism extracts fluctuation features online, enabling coordinated updates of order splitting priorities and routing topology. After real-time injection of sudden cost increments into the capacity probe stream, volatile flight segments are quickly identified by integrating stability measures of tariff slope amplitude and cargo flow gradient. The edge routing reconfiguration operator instantly reshapes the node sequence and occupancy window. The budget stability determination closed-loop continuously tracks the full-link valuation and injects cost compensation the instant the error exceeds the limit, which is solidified synchronously with the warehouse allocation instructions to avoid loading and unloading peaks and capacity gaps. Thus, in scenarios with drastic fluctuations in flexible freight rates, the invention maintains dual convergence of warehousing and distribution network costs and timeliness, significantly reduces the frequency of repeated recalculation triggers, shortens the fulfillment cycle, suppresses warehouse delays and rework, improves the on-time rate of last-mile delivery and the platform's capital turnover efficiency, and achieves high resilience and high transparency throughout the cross-border supply chain. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the intelligent warehouse scheduling and dynamic routing planning system for cross-border e-commerce according to the present invention. Figure 2 This is a flowchart illustrating the budget stability correction module of the intelligent warehouse scheduling and dynamic routing planning system for cross-border e-commerce of this invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1: Figure 1 This invention presents an intelligent warehouse scheduling and dynamic routing planning system for cross-border e-commerce, comprising: a cost matrix generation module, a tensor decomposition optimization module, a capacity probe adjustment module, and a budget stability correction module.

[0020] Cost matrix generation module: Generates a two-layer cost perception matrix at the warehouse allocation decision entry point. One layer records the static surface of contract freight rates, and the other layer captures the flexible capacity tariff curve in real time. The warehouse-segment mapping is locked through hash index.

[0021] Tensor decomposition optimization module: Adaptive tensor dimensionality reduction search is used to perform streaming decomposition on the two-layer cost perception matrix, extract pulse fluctuation indexes and adjust the order splitting priority according to the index, and output a draft of the warehouse allocation with elastic threshold.

[0022] Capacity probe adjustment module: Injects real-time capacity probe streams into the routing topology, superimposes sudden cost increments on each trunk edge of the draft, and if the stability metric obtained by fusing price fluctuation characteristics and freight flow gradient characteristics is lower than the standard, it calls the edge routing reconfiguration operator and synchronously updates the node sequence and occupancy window.

[0023] Budget stability correction module: The budget stability arbiter calculates the segmented integral of the difference between the updated end-to-end valuation and the original valuation. If the integral exceeds the threshold, the difference is re-injected to compensate and the final allocation and path list is solidified.

[0024] With the rapid development of cross-border e-commerce in the context of globalization, especially during peak sales periods, order volumes experience explosive growth, placing extremely high demands on the responsiveness and processing capacity of logistics systems. Traditional logistics scheduling methods rely on static planning and fixed contract freight rates, making it difficult to cope with sudden order peaks and real-time fluctuations in market freight rates. This limitation often leads to a surge in warehousing pressure, frequent transportation delays, and even cost overruns and decreased fulfillment efficiency. To address this pain point, intelligent warehouse scheduling and dynamic routing planning systems for cross-border e-commerce have emerged, aiming to achieve efficient processing of massive orders through technological innovation and ensure the stability and timeliness of the logistics chain amidst market fluctuations.

[0025] In cross-border e-commerce promotional scenarios, intelligent warehouse scheduling and dynamic route planning need to determine suitable forward warehouse locations and cross-border transportation routes for massive orders within an extremely short time window. However, flexible freight rates in the market (such as temporary LCL or charter prices) change rapidly, while traditional contract freight rates have a certain lag. The time difference between the two can cause scheduling decisions to deviate from the optimal state. To address this, the cost matrix generation module constructs a two-layer cost perception matrix to integrate static contract freight rates with dynamic flexible freight rates, providing a data foundation for subsequent dynamic adjustments. This process is not only the technical starting point of the entire invention but also a key step in solving the mismatch between rigid freight rate models and real-time rates.

[0026] The cost matrix generation module includes the following: The cost matrix generation module aims to construct a two-layered cost-awareness matrix by integrating long-term contract price data and real-time market price data: a static contract freight rate surface and a flexible capacity and tariff surface. This provides real-time and accurate cost basis for warehouse allocation decisions. Simultaneously, by constructing a hash-based mapping table, it binds warehouse-segment combinations to their corresponding cost data locations, improving data access efficiency. Achieving this technical goal ensures that the system can balance comprehensive cost data and real-time access speed when processing large-scale orders.

[0027] S1.1, Data Collection and Classification: In the data collection and classification phase, the first step is to collect price data from long-term contracts signed with logistics providers to form a static table of contract freight rates.

[0028] The static contract freight rate surface is a two-dimensional dataset, specifically represented as a matrix of rows and columns. Each row represents a warehouse, and each column represents a flight segment. Each element in the matrix represents the fixed freight rate from the corresponding warehouse to the corresponding flight segment. This dataset originates from long-term cooperation agreements, ensuring stability and reflecting a fixed cost benchmark. Secondly, through a real-time data interface, dynamically changing capacity and price information in the market is continuously acquired, forming a flexible capacity and price surface. This flexible capacity and price surface is also a two-dimensional dataset with the same structure as the static contract freight rate surface, but its values ​​are updated in real-time according to changes in market supply and demand, reflecting the volatility of immediate freight rates.

[0029] It acquires both static and dynamic cost data, providing a raw data source for constructing a comprehensive cost basis; it can simultaneously capture the stability of long-term agreements and the flexibility of market prices.

[0030] S1.2, Construction of the two-layer cost-awareness matrix: In the construction phase of the two-layer cost perception matrix, the static surface of contract freight rates and the surface of flexible capacity rates are integrated into a cost perception matrix with a two-layer structure.

[0031] The two-layer cost perception matrix is ​​a three-dimensional dataset composed of two superimposed two-dimensional matrices: the first layer is the static surface of contract freight rates, and the second layer is the surface of flexible capacity rates.

[0032] Each layer is a cost matrix consisting of warehouses and flight segments. The values ​​in the first layer are derived from long-term contract price data and are fixed; the values ​​in the second layer are derived from real-time market price data and are dynamic. Each element in the two-layer cost-awareness matrix represents the freight rate in the first and second layers, respectively, for a specific combination of warehouses and flight segments. Through this two-layer structure, the system can simultaneously retain information on both static and dynamic costs.

[0033] By overlaying two layers of data structure, it fully integrates long-term and real-time cost perspectives; it provides a comprehensive cost data framework, enabling warehouse allocation decisions to select appropriate cost bases based on different scenarios.

[0034] S1.3, Hash key locks warehouse-segment mapping: In the warehouse-segment mapping stage using hash keys, a hash-based mapping table is constructed to optimize data access efficiency. The keys of the hash mapping table consist of combinations of warehouses and segments, while the values ​​are the position index of that combination in the two-layer cost-aware matrix. Specifically, for each warehouse-segment combination, the hash mapping table records its corresponding row and column numbers, allowing the system to directly locate the corresponding element in the two-layer cost-aware matrix without traversing the entire matrix. The construction process of the hash mapping table includes: First, traverse the two-layer cost-aware matrix to extract all combinations of warehouses and flight segments; then, generate a unique key for each combination and associate it with its row and column positions in the matrix; finally, store the correspondence between these key values ​​and positions.

[0035] Hash technology can reduce the time complexity of data lookup from linear to near constant level; it significantly improves the system's response speed when processing large-scale orders, and is especially suitable for scenarios with high real-time requirements.

[0036] S1.4, Data Association and Validation: In the data association and verification phase, a hash mapping table is used to bind the freight rate data in the static contract freight rate surface and the flexible capacity and tariff surface with specific warehouse and flight segment combinations. The specific operation is as follows: Based on the key values ​​in the hash map table, the position of each warehouse and flight segment combination in the two-layer cost-awareness matrix is ​​found, and the fare values ​​from the first and second layers are filled into the corresponding positions. Subsequently, a consistency check is performed on the data in the two-layer cost-awareness matrix to ensure that the fare data from the first and second layers are comparable in terms of units and completeness. For example, it checks whether the currency units of all fare data are consistent, whether there are missing or outliers, and corrects or marks any inconsistent data.

[0037] By linking and verifying data, the accuracy and availability of data in the two-layer cost perception matrix are ensured, thus improving the reliability of cost data and avoiding decision-making biases caused by data errors.

[0038] The cost matrix generation module outputs a two-layer cost-aware matrix and a hash map. The two-layer cost-aware matrix contains two layers of freight rate information: the first layer is a static contract freight rate surface, and the second layer is a flexible capacity tariff surface, providing static and dynamic cost perspectives respectively. The hash map records the position index of each warehouse and flight segment combination within the two-layer cost-aware matrix, enabling rapid data access later. This provides a structured and efficient organization of cost data, offering a comprehensive and quickly searchable basis for subsequent warehouse allocation decisions.

[0039] The cost matrix generation module constructs a two-layer cost perception matrix that includes a static surface of contract freight rates and a flexible capacity and tariff surface. This enables the integration of long-term contract prices with real-time market prices, meets the needs of warehouse allocation decisions for multi-dimensional cost information, and improves the comprehensiveness and adaptability of decision-making.

[0040] The cost matrix generation module has successfully constructed a two-layer cost-awareness matrix, integrating static contract freight rates and dynamic flexible capacity tariffs, and locking the warehouse-segment mapping with hash keys. However, real-time fluctuations in market freight rates, especially the time lag between fluctuations in flexible capacity tariffs and contract freight rates, can lead to distorted cost benchmarks in cargo allocation decisions, resulting in cost overruns and performance delays. Therefore, the tensor decomposition optimization module needs to analyze the freight rate fluctuation characteristics in the two-layer cost-awareness matrix to optimize order splitting priorities and cargo allocation drafts, thereby improving the flexibility and adaptability of decision-making.

[0041] The tensor decomposition optimization module includes the following: The goal of the Tensor Decomposition Optimization (TDO) module is to optimize order allocation priorities and generate preliminary warehouse allocation plans (i.e., draft allocation proposals) by analyzing the freight rate fluctuation characteristics in the two-layer cost perception matrix. The two-layer cost perception matrix contains information on both static and dynamic costs, reflecting freight rate changes in warehouse and flight segment combinations under different market conditions. The TDO module identifies the intensity and frequency of freight rate fluctuations and uses these characteristics to adjust the selection order of warehouse and flight segment combinations, thereby improving the adaptability and flexibility of the decision-making process to market changes and ultimately providing an optimized basis for order allocation.

[0042] S2.1, Adaptive Tensor Dimensionality Reduction Flow Decomposition: In the adaptive tensor dimensionality reduction flow decomposition stage, the two-layer cost-aware matrix is ​​first treated as a dataset containing three dimensions: warehouse, flight segment, and cost layer. The cost layer includes static and dynamic costs. For this three-dimensional dataset, dimensionality reduction decomposition is performed, breaking it down into a combination of multiple low-dimensional datasets to extract the main patterns of fare fluctuations. The decomposition process employs an adaptive strategy, dynamically adjusting the complexity of the decomposition based on the characteristics of fare fluctuations in the input data. This ensures accurate capture of the main features of fare fluctuations while avoiding overfitting due to excessive decomposition complexity. From the multiple low-dimensional datasets obtained through decomposition, key indicators of fare fluctuations are extracted, termed the impulse fluctuation index. The calculation process for the impulse fluctuation index is as follows: For each warehouse and route combination, the temporal variation of the flexible capacity tariff surface is analyzed. Specifically, this is achieved by calculating the maximum rate of change of flexible capacity tariffs over time and combining this with the fluctuation frequency of the combination for a comprehensive evaluation. The fluctuation frequency is determined by statistically analyzing the number of fluctuations in the time series and using a logarithmic function to amplify the impact of high-frequency fluctuations. Finally, the value of the pulse fluctuation index is derived to reflect the intensity and frequency of freight rate fluctuations.

[0043] By using dimensionality reduction decomposition and key indicator extraction, the main characteristics of freight rate fluctuations can be accurately identified.

[0044] S2.2, Reorder order splitting priority: During the order splitting priority reordering phase, a priority score is calculated for each warehouse and segment combination based on the impulse fluctuation index. The priority score calculation process comprehensively considers both the impulse fluctuation index and the deviation between static and dynamic fares. Specifically, first, the ratio of static to dynamic fares is calculated. This ratio is then input into an exponential function to generate a magnification factor. This factor is designed to significantly increase its value when the deviation between static and dynamic fares is large. Subsequently, the impulse fluctuation index is multiplied by this magnification factor to obtain the priority score. This ensures that warehouse and segment combinations with larger or higher fare fluctuations receive higher priority scores. Next, all warehouse and segment combinations are sorted in descending order of priority score, generating a reordered order splitting priority sequence. During order allocation, warehouse and segment combinations with the highest priority scores are prioritized to fully utilize low-price capacity opportunities in the market or mitigate potential cost risks from fare fluctuations.

[0045] By quantifying the characteristics of freight rate fluctuations and static and dynamic freight rate deviations, the priority of order allocation was optimized, thereby improving the system's sensitivity to changes in market freight rates and its response efficiency.

[0046] S2.3, Generate a draft of the sub-warehouse with elastic thresholds: In the stage of generating the draft sub-classification with a flexible threshold, a flexible threshold is first set based on the priority scores. The process of setting the flexible threshold is as follows: calculate the median value of all priority scores, that is, take the middle value after sorting all scores by size, as the benchmark; then, calculate the interquartile range of the priority scores, that is, the difference between the 75th percentile value and the 25th percentile value in the score sequence, reflecting the degree of dispersion of the scores; next, multiply the interquartile range by an adjustment coefficient (usually taken as an empirical value of 1.5) to obtain the adjustment amount, and add this adjustment amount to the median value to generate the final flexible threshold.

[0047] After setting a flexibility threshold, warehouse and segment combinations with priority scores higher than or equal to this threshold are included in the allocation draft, and orders are assigned to these combinations according to the rearranged order splitting priority sequence. The allocation draft includes a list of recommended warehouse and segment combinations and corresponding order allocation suggestions.

[0048] By using flexible thresholds to select the optimal combination of warehouses and flight segments, the rationality of the draft is ensured; this approach can both capture cost optimization opportunities in the market and effectively control the risks caused by freight rate fluctuations, providing high-quality input for subsequent processing.

[0049] The key technical feature of the tensor decomposition optimization module lies in its adaptive tensor dimensionality reduction flow decomposition of the two-layer cost perception matrix. This extracts the key indicator of freight rate fluctuations—the impulse fluctuation index—and reorders order allocation priorities based on this index, ultimately generating a draft allocation plan with elastic thresholds. This effectively captures the real-time volatility characteristics of market freight rates and optimizes order allocation priorities through quantification and sorting; it enhances the adaptability of allocation decisions to market changes and strengthens the targeting of cost optimization.

[0050] The tensor decomposition optimization module extracts freight rate fluctuation characteristics and generates a draft allocation plan with elastic thresholds through adaptive tensor dimensionality reduction flow decomposition, laying the initial topology for route planning. However, during cross-border e-commerce promotions, real-time fluctuations in market freight rates and unforeseen events (such as low-price consolidation or temporary cargo bookings) may cause the trunk-side costs in the draft allocation plan to deviate from reality, affecting the stability of route planning. If not adjusted in time, low-stability segments will lead to budget overruns and delivery delays. Therefore, the capacity probe adjustment module needs to incorporate real-time capacity data, overlay sudden cost increases, and identify unstable segments and optimize routes by quantifying segment cost fluctuations and cargo flow concentration characteristics to ensure the adaptability and cost control of the warehousing and distribution network.

[0051] The capacity probe adjustment module includes the following: The goal of the capacity probe adjustment module is to introduce real-time capacity data to overlay sudden cost increments onto each trunk line in the cargo allocation draft, and to identify unstable segments and optimize route planning by quantifying segment cost fluctuations and cargo flow concentration characteristics, thereby improving the system's stability and adaptability. Specifically, it introduces real-time capacity probe streams to obtain the latest freight rate information and calculates sudden cost increments; then, it calculates segment expansion / contraction entropy and cargo flow emergence to assess segment volatility and cargo flow suddenness; finally, it combines the two to calculate a cost stabilization coefficient, identifies low-stability segments, and triggers route reconfiguration.

[0052] S3.1 introduces a real-time capacity probe stream and overlays it with sudden cost increments: In the stage of introducing a real-time capacity probe stream and overlaying sudden cost increments, the latest market capacity prices, availability, and information on sudden events are first collected in real time through an external interface to form a real-time capacity probe stream. For each trunk line in the draft distribution plan, the latest freight rate at the current moment is extracted from the real-time capacity probe stream and compared with the original cost of that trunk line in the two-layer cost perception matrix. If the latest freight rate is higher than the original cost, the difference between the latest freight rate and the original cost is calculated, and this difference is used as the sudden cost increment; if the latest freight rate is lower than or equal to the original cost, the sudden cost increment is zero. Then, the sudden cost increment is added to the original cost to obtain the updated trunk line cost.

[0053] Real-time capacity probe streams can reflect immediate changes in market freight rates and ensure that trunk-side costs are consistent with actual market conditions by overlaying sudden cost increases; this improves the accuracy of cost calculations and avoids routing planning errors caused by market fluctuations not being reflected in a timely manner.

[0054] S3.2, Price Volatility Characteristics: Price volatility characteristics include segment expansion and contraction entropy. In the stage of calculating segment expansion and contraction entropy, segment expansion and contraction entropy is used to quantify the complexity and uncertainty of cost fluctuations on the trunk line within a specific time window.

[0055] First, the historical fare sequence within the time window is used from the real-time capacity probe stream to calculate the rate of change in fares between each adjacent moment. Next, these rates of change are grouped and their frequencies are statistically analyzed to form a probability distribution. Then, based on the concept of information entropy, the product of the logarithm of the frequency of each rate of change and the frequency itself is calculated, and all products are summed to obtain the local entropy index. To highlight the importance of recent fluctuations, a time decay factor is introduced, assigning different weights to the local entropy value at each moment according to its temporal proximity, and then performing a weighted summation to obtain the final segment expansion / contraction entropy.

[0056] The higher the value of the segment expansion / contraction entropy, the more complex and uncertain the freight rate fluctuations along the trunk line. Segment expansion / contraction entropy comprehensively describes the characteristics of freight rate fluctuations through statistical and weighted methods; it provides a scientific basis for segment stability assessment and facilitates the identification of fluctuation risks that may affect route planning.

[0057] S3.3, Cargo Flow Gradient Characteristics: The freight flow gradient feature includes calculating freight flow emergentness, which measures the suddenness and concentration of freight flow along trunk lines. First, using real-time freight flow data within a specific time window, the rate of change of freight flow between each adjacent time point is calculated. Next, the average of these rates of change is calculated, and the fourth power average and squared average of the differences between the rates of change and the average are calculated separately. The ratio of the fourth power average to the squared average is then taken to obtain the kurtosis index. Finally, the logarithm of the kurtosis index is taken and multiplied by the maximum absolute value in the rate of change sequence to obtain the freight flow emergentness.

[0058] A higher cargo flow emergent value indicates a more concentrated and sudden change in cargo flow. Cargo flow emergent value, through kurtosis analysis and amplification, can highlight abnormal fluctuations in cargo flow, providing quantitative support for identifying flight segments that may experience cost overruns due to sudden increases in cargo flow.

[0059] S3.4, Calculate the cost-based voltage stabilization coefficient: In the cost stabilization coefficient calculation stage, the segment expansion / contraction entropy and cargo flow emergentness are first normalized to unify their dimensions. Next, the complements of the normalized segment expansion / contraction entropy and cargo flow emergentness are taken, i.e., one minus the normalized value, to reflect stability. Then, according to preset weighting coefficients, a weighted average is calculated on the complements of the segment expansion / contraction entropy and the complements of the cargo flow emergentness to obtain the cost stabilization coefficient.

[0060] The cost stabilization coefficient ranges from zero to one; a smaller value indicates poorer cost stability along the main route. By normalizing and weighting the average, the impact of cost fluctuations and cargo flow concentration on route stability is comprehensively considered, generating an intuitive indicator to facilitate subsequent identification of routes requiring optimization.

[0061] S3.5 identifies low-voltage stable flight segments and triggers edge route reconfiguration: During the identification of low-voltage stable flight segments and the triggering of edge route reconfiguration, a safety threshold is set based on business requirements. If the cost stability coefficient of a trunk line edge is lower than this safety threshold, the trunk line edge is identified as a low-voltage stable flight segment. For low-voltage stable flight segments, the system calls the edge route reconfiguration operator and, based on the updated trunk line edge cost and cost stability coefficient, uses a heuristic search algorithm to find a more stable alternative path. Specifically, it traverses the available paths from the origin to the destination, comprehensively considers path cost and stability, selects the optimal path, and updates the node sequence and placeholder time window in the route topology.

[0062] By identifying low-pressure flight segments and optimizing routes, the uncertainty caused by market fluctuations and sudden cargo flows can be reduced; the reliability of route planning can be enhanced, ensuring the efficient operation of the warehousing and distribution network in a dynamic environment.

[0063] The technical feature of the capacity probe adjustment module lies in its ability to introduce real-time capacity probe streams, superimpose sudden cost increments, and calculate a cost stabilization coefficient by integrating segment expansion / contraction entropy and cargo flow emergence. This identifies low-stability segments and triggers edge route reconfiguration. It can respond in real-time to changes in market freight rates and cargo flow, quantifying the stability and risk of segments; improving the adaptability and stability of route planning; and significantly reducing the risk of budget overruns and delivery delays caused by market uncertainty. This provides efficient support for warehousing and distribution network optimization in cross-border e-commerce promotional scenarios.

[0064] The capacity probe adjustment module utilizes the capacity probe stream to overlay sudden cost increases and trigger edge route reconfiguration, updating the node sequence and occupancy window. However, real-time fluctuations in market freight rates and unforeseen events may cause the reconfigured end-to-end cost to deviate from the original estimate. Without correction, this will distort the allocation of cargo space and route lists, leading to budget overruns and performance delays. Therefore, the budget stabilization correction module needs to quantify the difference between the reconfigured end-to-end estimate and the original estimate, and adjust the two-layer cost perception matrix as necessary to ensure the cost stability and execution reliability of the final solution.

[0065] The goal of the budget stabilization correction module is to assess the impact of route reconfiguration on costs by quantifying the difference between the end-to-end valuation after reconfiguration and the original valuation. When the difference exceeds the preset safety limit, the module compensates and adjusts the two-layer cost perception matrix, and finally solidifies the warehouse allocation and route list to ensure cost stability and execution reliability amidst market freight rate fluctuations.

[0066] like Figure 2 As shown, the budget stabilization correction module includes the following: S4.1, Obtain the end-to-end valuation after reconfiguration and the original valuation: In the stage of obtaining the full-link valuation after reconfiguration and the original valuation, the original valuation and the full-link valuation after reconfiguration are calculated first. The original valuation is the sum of the original costs of each trunk line in the draft allocation plan, reflecting the expected costs without considering market fluctuations. The original costs are derived from the two-layer cost perception matrix, specifically the fusion value of the static surface of the contract freight rate and the flexible capacity tariff surface for each trunk line. The full-link valuation after reconfiguration is the sum of the adjusted costs of each trunk line in the route topology updated by the capacity probe adjustment module. The adjusted costs have been superimposed with the sudden cost increment, reflecting the actual costs after market fluctuations.

[0067] The calculation process is as follows: the original costs of all trunk lines in the draft distribution plan are added together to obtain the original valuation; the adjusted costs of all trunk lines in the routing topology output by the capacity probe adjustment module are added together to obtain the full-link valuation after reconfiguration.

[0068] By comparing the original valuation with the full-chain valuation after reallocation, the specific impact of market fluctuations on costs can be quantified; this provides an accurate data basis for cost deviation assessment and ensures the reliability of subsequent analysis.

[0069] S4.1, Calculate the piecewise integral of the valuation difference: In the segmented integration stage of calculating the valuation difference, the entire supply chain is divided into multiple key segments, such as from warehouse to port, port to delivery node, etc., to refine the assessment of cost deviations. For each key segment, the difference between the reconfigured valuation and the original valuation within a specific time window is calculated over time, and these differences are integrated to obtain the valuation difference integral for that key segment.

[0070] The specific steps are as follows: First, determine the time window, which is set according to the time range of market fluctuations; then, calculate the difference between the valuation after reallocation and the original valuation at each time point; next, sum these differences over the time window to obtain the integral value of that key segment. Subsequently, add up the integrals of the valuation differences of all key segments one by one to obtain the total integral of the entire chain, which represents the cumulative value of the overall cost deviation.

[0071] Segmented integration can accurately capture the cost change trends of different links in market fluctuations; it provides comprehensive and detailed quantitative indicators of cost deviation, which facilitates subsequent judgment and adjustment needs.

[0072] S4.2, Determine if the safety limit has been exceeded: In the stage of determining whether the safety limit has been exceeded, a safety limit is preset based on business needs, which serves as the upper limit of the allowable valuation difference.

[0073] The specific operation involves comparing the total score across the entire value chain with the safety limit. If the total score exceeds the safety limit, the cost deviation after reconfiguration is deemed to exceed an acceptable range, requiring compensation adjustments. The safety limit is set based on the business's tolerance for cost fluctuations, ensuring budget control is maintained amidst market volatility.

[0074] By setting preset safety limits, abnormal situations of cost overruns can be automatically identified, improving the automation level of budget management and avoiding financial risks caused by excessive cost deviations.

[0075] S4.3, inject compensation difference back into the two-layer cost perception matrix: During the stage of injecting compensation difference into the two-layer cost perception matrix, the compensation difference is first calculated, which is the difference between the total integral of the entire link and the safety limit, representing the amount of cost that needs to be adjusted.

[0076] The specific operation is as follows: Subtract the safety limit from the total score to obtain a positive difference as the compensation difference. Then, distribute the compensation difference evenly to each trunk edge in the reconfigured routing topology. Specifically, divide the compensation difference by the total number of trunk edges to obtain the adjustment amount for each trunk edge. Then, add this adjustment amount to the corresponding trunk edge cost in the flexible capacity tariff surface layer of the two-layer cost awareness matrix to complete the cost reinjection adjustment.

[0077] By injecting compensation for the difference, cost deviations caused by market fluctuations can be corrected; maintaining consistency between the two-layer cost perception matrix and the actual market situation provides accurate cost basis for subsequent execution.

[0078] S4.4, solidify the final warehouse allocation and route list: In the final warehouse allocation and route list solidification stage, the reconfigured routing topology and adjusted two-layer cost awareness matrix output by the capacity probe adjustment module are locked as the final warehouse allocation and route list, serving as the basis for subsequent order execution and transportation scheduling. The warehouse allocation and route planning in the reconfigured routing topology, along with the cost data in the adjusted two-layer cost awareness matrix, are integrated into a fixed scheme and stored. By solidifying the list, the stability and executability of decisions are ensured; a clear and optimized solution is provided for the actual operation of the warehousing and distribution network, avoiding repeated adjustments and execution chaos caused by market fluctuations.

[0079] The technical feature of the budget stabilization correction module lies in calculating the piecewise integral of the difference between the reconfigured end-to-end valuation and the original valuation. When the integral exceeds a safety limit, it injects compensation into the two-layer cost perception matrix, ultimately solidifying the warehouse allocation and route list. This allows for accurate assessment of the specific impact of market fluctuations on costs and corrects deviations through an automatic adjustment mechanism. It significantly improves the budget stability and fulfillment reliability of the warehousing and distribution network under scenarios of drastic fluctuations in flexible freight rates, providing efficient cost control and execution assurance for high-concurrency scenarios such as cross-border e-commerce promotions.

[0080] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0081] It should be noted that the system of the present invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting a variety of hardware environments and usage requirements.

[0082] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0083] It should be noted that, in this document, the use of relational terms such as "first" and "second" is merely to distinguish one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0084] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An intelligent warehouse distribution and dynamic routing system for cross-border e-commerce, characterized in that, Comprise: Cost matrix generation module, tensor decomposition optimization module, capacity probe adjustment module and budget stability correction module; The cost matrix generation module: in the warehouse decision entry generates a double-layer cost perception matrix, one layer records the static surface of the contract freight rate, and the other layer captures the flexible capacity cost curve in real time, and locks the warehouse-segment mapping through hash index; The tensor decomposition optimization module: using adaptive tensor dimensionality reduction search to flow decomposition of double-layer cost perception matrix, extract pulse fluctuation index and adjust the single priority according to the index high and low, output the warehouse draft; The capacity probe adjustment module: the real-time capacity probe flow is injected into the routing topology, and the sudden cost increment is superimposed on each trunk edge of the draft, if the stability measure obtained by fusing the price amplitude characteristics and the freight flow gradient characteristics is lower than the standard, the edge routing reconfiguration operator is called and the node sequence and placeholder time window are updated synchronously; The budget stability correction module: the segmented integral of the difference between the updated full-link estimate and the original estimate is calculated by the budget stability determinator, if the integral is above the threshold, the compensation difference is injected back and the final warehouse and path list are solidified. 2.The cross-border e-commerce oriented intelligent warehouse dispatching and dynamic routing planning system of claim 1, wherein, The cost matrix generation module includes the following: By integrating the static surface of the contract freight rate and the flexible capacity cost curve, a double-layer cost perception matrix containing two layers is generated, wherein the static surface of the contract freight rate is a two-dimensional data set based on long-term contract price data, and the flexible capacity cost curve is a dynamic two-dimensional data set based on real-time market price data; at the same time, the mapping relationship between the warehouse and the segment is locked through the hash mapping table, taking the combination of the warehouse and the segment as the key and the position index in the double-layer cost perception matrix as the value. 3.The cross-border e-commerce oriented intelligent warehouse dispatching and dynamic routing planning system of claim 2, characterized in that, The tensor decomposition optimization module includes the following: Through the adaptive tensor dimensionality reduction flow decomposition module, the double-layer cost perception matrix is decomposed into a combination of multiple low-dimensional tensors, and the pulse fluctuation index of the freight rate fluctuation is extracted; According to the pulse fluctuation index and the deviation between the static freight rate and the dynamic freight rate, the priority score is calculated, and the single priority is rearranged from high to low according to the priority score; according to the median value of the priority score and the interquartile range, an elastic threshold is set, and the warehouse and segment combination with a priority score higher than or equal to the elastic threshold is selected to generate a warehouse draft with an elastic threshold. 4.The cross-border e-commerce oriented intelligent warehouse dispatching and dynamic routing planning system of claim 3, characterized in that, The tensor decomposition optimization module further includes the following: The calculation process of the priority score considers the pulse fluctuation index and the deviation between the static freight rate and the dynamic freight rate; specifically, first, the ratio of the static freight rate to the dynamic freight rate is calculated, then the ratio is input into an exponential function to generate an amplification factor; subsequently, the pulse fluctuation index is multiplied by the amplification factor to obtain the priority score. 5.The cross-border e-commerce oriented intelligent warehouse dispatching and dynamic routing planning system of claim 4, characterized in that, The capacity probe adjustment module includes the following: Through the external interface, the latest market data is collected in real time to form a real-time capacity probe flow, and a sudden cost increment is added to each trunk edge in the warehouse draft; based on the historical freight rate sequence in the real-time capacity probe flow, the probability distribution of the adjacent time freight rate change rate is calculated and combined with the time decay factor to generate the segment inflation and shrink entropy. 6.The cross-border e-commerce oriented intelligent warehouse dispatching and dynamic routing planning system of claim 5, wherein, The capacity probe adjustment module further includes the following: Based on the freight flow data, the change rate kurtosis is calculated from the freight flow sequence and combined with the maximum absolute value to generate the freight flow emergence degree. 7.The cross-border e-commerce oriented intelligent warehouse dispatching and dynamic routing planning system of claim 6, wherein, The capacity probe adjustment module further includes the following: The cost stability coefficient is generated by taking the complement of the normalized expansion and contraction entropy and cargo flow emergence degree of the voyage section and weighted average, the low stability voyage section is identified according to the cost stability coefficient, and the edge routing reconfiguration operator is triggered, and the node sequence and the placeholder time window are updated. 8.The cross-border e-commerce oriented intelligent warehouse dispatching and dynamic routing planning system of claim 7, wherein, The capacity probe adjustment module further includes the following: If the cost stability coefficient of a certain line edge is lower than the safety threshold, the corresponding trunk line edge is determined as a low stability voyage section. 9.The cross-border e-commerce oriented intelligent warehouse dispatching and dynamic routing planning system of claim 8, wherein, The budget stability correction module includes the following: The influence of routing reconfiguration on cost is evaluated by calculating the segmented integral of the difference between the estimated value of the whole link after reconfiguration and the original estimated value, and the specific operation is to divide the whole link into multiple key sections, calculate the difference between the estimated value after reconfiguration and the original estimated value in a specific time window for each key section, and perform integral processing to obtain the estimated value difference integral of the corresponding key section, and then sum up the estimated value difference integrals of all key sections to obtain the total integral of the whole link.

10. The intelligent bin-oriented scheduling and dynamic routing planning system for cross-border e-commerce of claim 9, wherein, The budget stability correction module further includes the following: When the total integral is greater than the preset safety limit, the compensation difference is calculated and evenly distributed to each trunk line edge in the reconfigured routing topology, the compensation difference is divided by the total number of trunk line edges to obtain the adjustment amount of each trunk line edge, and the adjustment amount is added to the corresponding trunk line edge cost of the flexible capacity cost surface layer in the double-layer cost perception matrix to complete the cost back-annotation adjustment; finally, the reconfigured routing topology and the adjusted double-layer cost perception matrix are locked as the final warehouse and path list. ​

Images