Supply chain collaborative deduction method based on multi-dimensional population portrait evolution and regional sales fluctuation

Abstract

The present application relates to the field of industrial internet and big data prediction technology, and discloses a supply chain collaborative deduction method based on multi-dimensional group portrait evolution and regional sales fluctuation. It aims to solve the problem that the existing prediction method does not consider the spatial transmission effect of consumption fluctuation and is separated from physical constraints such as warehouse capacity, which easily leads to delayed inventory or end warehouse explosion. The present application extracts the leading portrait evolution to construct the sales fluctuation propagation path, deduces the expected fluctuation arrival time and demand increment; combined with the upper limit of warehouse physical capacity to determine the advance inventory gap; cooperatively triggers the horizontal transfer and vertical replenishment of adjustable inventory, and based on the actual order accounting results, the unexecuted tasks are dynamically intercepted and corrected. This scheme combines dynamic portrait and spatial network, improves the accuracy of demand prediction, and effectively reduces the risk of invalid logistics loss and inventory accumulation by introducing physical constraints and closed-loop error correction mechanism.

Description

Technical Field

[0001] This invention relates to the field of industrial internet and big data prediction technology, and in particular to a supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations. Background Technology

[0002] In current supply chain management and warehousing logistics scheduling systems, the system typically relies on historical sales data to predict future market demand. Existing conventional technologies mostly employ time series analysis or basic machine learning models to independently estimate sales for a single region or warehouse, and automatically trigger replenishment or production instructions upstream when the system inventory falls below a preset safety threshold, thereby maintaining the basic operation of the supply chain.

[0003] However, with the increasing complexity of the market environment, existing technologies have gradually revealed significant limitations. On the one hand, traditional independent forecasting methods ignore the spatial transmission effect of consumption trends across different geographical regions and cannot dynamically capture the rapidly changing migration of group preferences, resulting in a severe lag in response to sudden surges in demand. On the other hand, existing supply chain extrapolation logic mostly remains at the level of pure data mathematical calculations, failing to effectively integrate real-world constraints such as the physical capacity limitations of terminal warehousing, dynamic logistics allocation costs, and interference from fictitious market transactions. This disconnect between pure data forecasting and the physical execution environment often leads to serious consequences when generated scheduling instructions are actually implemented, such as end-of-line warehouse overload and damage, deep accumulation of redundant inventory, and ineffective logistics consumption, making it difficult to achieve efficient and secure coordination of the entire supply chain. Summary of the Invention

[0004] In view of the problems existing in the prior art, the present invention is proposed.

[0005] Therefore, in order to address the problem that existing forecasting methods do not consider the spatial transmission effect of consumption fluctuations and are detached from physical constraints such as storage capacity, which can easily lead to delayed inventory preparation or end-of-line warehouse overflows, this invention provides a supply chain collaborative extrapolation method based on the evolution of multi-dimensional group profiles and regional sales fluctuations.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations, including the following steps: S1. Obtain customer behavior data, regional sales data, and supply chain status data, form a related data sequence according to product, sales region, and time, and determine the dominant profile evolution of the corresponding product in the corresponding sales region based on the changes in customer group status within adjacent statistical periods. S2. Based on the order of occurrence of the dominant profile evolution of the same product in different sales regions, and the time correspondence between the dominant profile evolution and regional sales fluctuations, determine the starting region of the fluctuation, construct the sales fluctuation propagation path, and determine the expected arrival time of the fluctuation and the expected demand increment of the target sales region in the sales fluctuation propagation path. S3. Based on the expected arrival time of the fluctuation, the expected increase in demand, and the status of supply chain nodes, determine the advance stocking gap for the target sales region; S4. According to the sales fluctuation propagation path, the adjustable inventory in the sales region that has passed the peak of the sales fluctuation is transferred in advance to the target sales region with the pre-stocking gap, and a supplementary procurement task is generated based on the remaining gap after the transfer to form a supply chain collaboration solution. S5. Based on the actual profile evolution, actual regional sales fluctuations, and the execution results of the supply chain collaboration plan, revise the expected fluctuation arrival time, expected demand increment, and unexecuted tasks corresponding to sales regions that have not yet experienced sales fluctuations.

[0007] As a preferred embodiment of the supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations described in this invention, the following steps are taken: Customer behavior data, regional sales data, and supply chain status data are acquired; a related data sequence is formed according to product, sales region, and time; and the dominant profile evolution of the corresponding product in the corresponding sales region is determined based on changes in customer group status within adjacent statistical periods. For a predefined set of sales regions, user interaction data, order transaction data, and warehousing and logistics data within those regions are extracted. These data are then aligned and linked according to product identification number, region identification code, and a unified time step to obtain an initial data sequence. For data loss due to communication interruptions, the mean data from adjacent time points is used to fill in the gaps. For data surges caused by promotions, the data at the surge points is smoothed using the historical growth multiple for the same period, and the linked data sequence is output. The smoothed customer behavior data is converted into feature values, and spatial clustering is used within continuous time windows to cluster customers with similar features, constructing a set of group clusters. Customer status features of the group clusters within two adjacent time windows are extracted and compared, and the growth rate of each group purchasing the corresponding product is calculated. The label change process of the group with the highest growth rate over time is determined as the dominant profile evolution of the corresponding product in the corresponding sales region.

[0008] As a preferred embodiment of the supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations described in this invention, the method involves: determining the initial region of fluctuations based on the order of occurrence of the dominant profile evolution for the same product in different sales regions, and the temporal correspondence between the dominant profile evolution and regional sales fluctuations; constructing a sales fluctuation propagation path; and determining the expected arrival time and expected demand increment of the target sales region in the sales fluctuation propagation path. The specific steps are as follows: Extract the start time of the dominant profile evolution for the same product within the preset sales region set, and determine the sales region where the dominant profile evolution occurred earliest in the time series as the fluctuation initiation region; exclude the fluctuation initiation region from the sales region set to determine the remaining region set; starting from the fluctuation initiation region, extract the traffic connectivity index and population mobility index between the sales regions included in the remaining region set and the fluctuation initiation region; introduce the impact factors of extreme weather and geographical remoteness, deduct the corresponding traffic connectivity index and population mobility index for regions affected by the impact factors, calculate the correlation strength based on the deducted indexes, and connect the corresponding sales regions in descending order of correlation strength to form the sales fluctuation propagation path. Based on the number of node levels between the target sales region and the region where the fluctuation originates in the sales fluctuation propagation path, and combined with the average interval between sales outbreaks in adjacent regions, the expected arrival time of the fluctuation is calculated. The regional difference ratio between the target sales region and the region where the fluctuation originates in terms of socioeconomic indicators is calculated, and the actual sales growth of the region where the fluctuation originates is multiplied by the regional difference ratio to obtain the basic demand increment. If a social hotspot event related to the corresponding product is identified, a preset hotspot amplification constant is added to the basic demand increment to obtain the final expected demand increment.

[0009] As a preferred embodiment of the supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations described in this invention, the method involves determining the advance stocking gap for the target sales region based on the expected arrival time of the fluctuation, the expected increase in demand, and the status of supply chain nodes. The specific steps are as follows: The total available inventory is obtained by adding the current inventory quantity of goods in stock and the quantity of goods in transit in the target sales region; the total demand is obtained by adding the expected increase in demand and the expected consumption quantity under stable conditions; the initial stocking gap is obtained by calculating the difference between the total demand and the total available inventory; the influence factor of the maximum capacity limit of the warehouse is introduced, and if the sum of the initial stocking gap and the total available inventory is greater than the maximum capacity limit of the corresponding warehouse, the portion of the value exceeding the maximum capacity limit is removed, and the remaining gap value is determined as the advance stocking gap.

[0010] As a preferred embodiment of the supply chain collaborative deduction method based on multi-dimensional group profile evolution and regional sales fluctuations described in this invention, the method involves: according to the sales fluctuation propagation path, adjusting inventory in sales regions that have passed the peak of sales fluctuations is transferred in advance to target sales regions with pre-stocking gaps, and supplementary procurement tasks are generated based on the remaining gaps after the transfer, thus forming a supply chain collaborative solution. The specific steps are as follows: If the number of daily orders in a sales region at the forefront of the sales fluctuation propagation path is monitored to decrease for a preset number of consecutive cycles and is lower than a preset percentage of the historical peak, then the corresponding sales region is determined to have passed the peak of the sales fluctuation. The remaining inventory of the sales region that has passed the peak of the sales fluctuation is calculated, and the adjustable inventory is obtained after deducting the safety reserve quantity. The influencing factors of the comparison between transfer costs and warehousing costs are introduced. If it is calculated that the warehousing maintenance cost saved by transferring the adjustable inventory is greater than the logistics transportation cost generated by cross-regional transfer, then a transfer instruction to send the adjustable inventory to the target sales region is issued based on the principle of proximity by physical distance. The remaining gap is obtained by subtracting the inventory received through the transfer instruction plan from the pre-stocking gap; the production and processing cycle of the upstream factory and the logistics and transportation cycle to the target sales region are extracted; the production and processing cycle and the logistics and transportation cycle are subtracted sequentially from the expected fluctuation arrival time to reverse-determine the deadline for issuing procurement tasks to the upstream production factory; before the procurement deadline is reached, a supplementary procurement task order is generated based on the remaining gap; the transfer instruction and the supplementary procurement task order are integrated into the supply chain collaboration solution according to the time nodes.

[0011] As a preferred embodiment of the supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations described in this invention, the method involves: correcting the expected arrival time of sales fluctuations, expected demand increments, and unexecuted tasks for sales regions that have not yet experienced sales fluctuations, based on the actual profile evolution, actual regional sales fluctuations, and the execution results of the supply chain collaborative scheme. The specific steps are as follows: The numerical difference between the projected increase in demand and the actual number of orders is calculated periodically. A false transaction interference factor is introduced; if the numerical difference exceeds a preset tolerance threshold, it is determined that the previously collected data contains this false transaction interference factor. The historical correlation data is replaced with the actual order quantity, and the projected increase in demand and the projected arrival time of the fluctuation in the sales region following the sales fluctuation propagation path are recalculated and updated. Based on the updated projected increase in demand and the projected arrival time of the fluctuation, transfer instructions that are in transit and have not yet been handed over, and supplementary purchase orders that have not yet reached the purchase deadline and have not been issued, are retrieved from the computer system. Based on the updated values, a transportation redirection instruction to change the destination warehouse is issued to the corresponding logistics node, and a production modification instruction to adjust the production quantity is issued to the corresponding upstream factory to correct the unexecuted tasks.

[0012] The beneficial effects of this invention are: This invention constructs the propagation path of sales fluctuations by extracting the evolution of the dominant profile, and extrapolates the expected arrival time of fluctuations and the increase in demand; it determines the advance stocking gap by combining the upper limit of warehouse physical capacity; it coordinates and triggers the horizontal transfer and vertical replenishment procurement tasks of adjustable inventory, and dynamically intercepts and corrects unexecuted tasks based on the actual order accounting results. This solution combines dynamic profiles with spatial networks, improves the accuracy of demand forecasting, and effectively reduces the risk of ineffective logistics losses and inventory backlog by introducing physical constraints and closed-loop error correction mechanisms. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 Here is the overall flowchart of the method; Figure 2 The logic diagram for calculating advance stocking gaps based on physical constraints; Figure 3 A schematic diagram illustrating the principles of spatiotemporal grid mapping and group profiling evolution; Figure 4 A schematic diagram illustrating the multi-stage process of blocking correction, physical truncation, and coordinated scheduling; Figure 5 A schematic diagram illustrating the evolution of user profiles and the propagation path of sales fluctuations across multiple regions. Detailed Implementation

[0015] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0016] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0017] Secondly, the term "one embodiment" or "example" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the invention. The appearance of an embodiment in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that mutually excludes other embodiments.

[0018] Example 1 Reference Figures 1-5 This is the first embodiment of the present invention, which provides a supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations, including the following steps: S1. Obtain customer behavior data, regional sales data, and supply chain status data; form a related data sequence according to product, sales region, and time; and determine the evolution of the dominant profile of the corresponding product in the corresponding sales region based on changes in customer group status within adjacent statistical periods, including the following steps: For a pre-defined sales region set, user interaction data, order transaction data, and warehousing and logistics data within the sales regions are extracted. These data are then aligned and linked according to product identification numbers, regional identification codes, and a unified time step to obtain an initial data sequence. For data loss due to communication interruptions, the mean data from adjacent time points is used to fill in the gaps. For data surges caused by promotions, the data at the surge points is smoothed using the historical growth multiple for the same period, outputting a linked data sequence. The smoothed customer behavior data is converted into feature values. Within continuous time windows, spatial clustering is used to cluster customers with similar features, constructing group clusters. Customer status features of the groups within each group cluster set are extracted and compared within two adjacent time windows, calculating the growth rate of each group cluster's purchase of corresponding products. The label change process of the group with the highest growth rate over time (e.g., ...) is then analyzed. Figure 5 The evolution from low-frequency attention to hollow dots to high-frequency transactions to solid dots (as shown in the diagram) is used to determine the dominant profile evolution of the corresponding product in the corresponding sales region.

[0019] It should be noted that in the initial data processing and analysis phase of supply chain simulation, the system needs to eliminate spatiotemporal asynchrony and abnormal interference from heterogeneous data, and perform standardized alignment of multi-source data and extraction and mining of dynamic group features. The specific operation steps are as follows: The collection of multi-source heterogeneous data and the standardization and alignment of spatiotemporal coordinates: In a commercial operating environment, interactive data reflecting user preferences, transaction data reflecting terminal consumption capacity, and warehousing data reflecting back-end support capabilities have different sources and statistical methods. In order to establish a unified analytical foundation, the main control computer extracts the above-mentioned heterogeneous data and configures them with unified product identification codes, geographical area boundaries, and time span labels, thereby combining the discrete raw information into an initial data sequence aligned in time and space dimensions.

[0020] A smoothing mechanism to address abnormal interference: During operation, business systems may encounter partial data upload interruptions due to communication base station failures, or sudden surges in sales caused by artificial shopping festivals. Directly using raw data with gaps or artificially inflated factors would severely distort subsequent trend predictions. The system extracts the mean of regular data before and after the communication interruption to bridge these data gaps. In this process, for data surges caused by promotions, the system aims to eliminate abnormal pulses and restore the normal baseline. The mathematical formula for data smoothing is as follows: : indicates the first The smoothed real correlation data values ​​within each time step, with a value range of [value range missing]. ≥0.

[0021] : indicates the first The raw burst data values ​​collected by the system within a time step, with a value range of [value range missing]. ≥0.

[0022] : indicates the determination of the first A Boolean indicator function to determine whether a time step is a historical promotional node; when the node is a promotional activity period. =1, otherwise =0.

[0023] : Represents the growth multiple coefficient for the same period in history, and >1.

[0024] This operation ensures that when in a non-promotional period, the denominator is 1 and the original data remains unchanged; when in a promotional period, the denominator is automatically converted into a growth multiple, thereby reducing the dimensionality of abnormally high pulse data by dividing it according to the historical expansion ratio, restoring the normal consumption baseline, and ensuring that the output related data sequence can objectively reflect the stable business evolution pattern.

[0025] Customer group cluster construction and profile evolution anchoring based on spatial clustering: The system converts cleaned user behavior data into multi-dimensional feature values. Within a single time slice, the system calculates the numerical distance between customer features, aggregating similar customers with close proximity into independent clusters. By comparing the consumption trajectories of each cluster in adjacent periods, the system extracts the target customer group with the highest growth rate in the quantity of specific products purchased. The tag migration trajectory of this target customer group from low-frequency attention to high-frequency transactions is identified by the system as the current dominant profile evolution in that region, serving as a driving indicator for subsequent inferences.

[0026] S2. Based on the order of the evolution of the dominant profile for the same product in different sales regions, and the time correspondence between the evolution of the dominant profile and regional sales fluctuations, determine the region where the fluctuation originates, construct the sales fluctuation propagation path, and determine the expected arrival time and expected demand increment of the target sales region in the sales fluctuation propagation path, including the following steps: Extract the starting time of the dominant profile evolution for the same product within a pre-defined set of sales regions, and determine the sales region where the dominant profile evolution occurred earliest in the time series as the fluctuation originating region. Exclude the fluctuation originating region from the set of sales regions to determine the remaining region set. Starting from the fluctuation originating region, extract the traffic connectivity and population mobility indicators between the sales regions included in the remaining region set and the fluctuation originating region. Introduce the impact factors of extreme weather and geographical remoteness, and deduct the corresponding traffic connectivity and population mobility indicators for regions affected by these factors. Calculate the correlation strength based on the deducted indicators, and connect the corresponding sales regions in descending order of correlation strength to form the sales fluctuation propagation path. Based on the number of node levels between the target sales region and the region where the fluctuation originates in the sales fluctuation propagation path, and combined with the average interval between sales outbreaks in adjacent regions, the expected arrival time of the fluctuation is calculated. The regional difference ratio in socioeconomic indicators between the target sales region and the region where the fluctuation originates is calculated, and the actual sales growth in the region where the fluctuation originates is multiplied by the regional difference ratio to obtain the basic demand increment. If a social hotspot event related to the corresponding product is identified, a preset hotspot amplification constant is added to the basic demand increment to obtain the final expected demand increment.

[0027] It should be noted that after extracting the characteristics of the underlying groups, the system needs to predict the direction and scale of the spread of consumption trends in the real physical space, and perform cross-regional propagation path construction and node-level demand quantification prediction. The specific operation steps are as follows: Anchoring the Origin Region of Sales Fluctuations and Constructing a Spatiotemporal Topological Propagation Path: A concentrated surge in sales typically exhibits spatial transmission characteristics. The system searches the global timeline to locate the first node across the entire network exhibiting the aforementioned group profile evolutionary characteristics, establishing it as the origin region of the fluctuation. The system extracts trunk logistics order volume and cross-regional travel passenger volume to measure the physical connectivity strength between the origin region and other regions nationwide. Following the path with the highest connectivity strength, the system traces level by level to find the next level of affected regional nodes, thus constructing a tree-like topological propagation path for sales fluctuations.

[0028] Dynamic correction of propagation indicators incorporating physical barriers: Theoretical connectivity data can be limited by sudden environmental factors. When the meteorological interface reports that a core trunk line encounters extreme snow and ice weather, or the target area is limited by weak transportation facilities in remote mountainous areas, the system internally uses a dynamic assessment of the actual physical connectivity between regions as the criterion. The formula for calculating the correlation tightness after incorporating physical barriers such as extreme weather and geographical remoteness is as follows: : Indicates the region and region The final correlation strength between them, with a value range of ≥0.

[0029] : Represents a basic indicator of normal traffic connectivity between regions, with a value range of . ≥0.

[0030] : This is a Boolean indicator variable representing the occurrence of a blocking event. It takes the value 1 when extreme weather or geographical blocking is detected, and 0 otherwise.

[0031] This indicates the amount of traffic capacity loss or reduction caused by physical disruption.

[0032] : Represents the weighting coefficient of the traffic connectivity index.

[0033] : Represents a basic indicator of normal population mobility between regions, with a value range of . ≥0.

[0034] This indicates the amount of population movement obstructed due to physical barriers.

[0035] : Represents the weighting coefficient of the population mobility index, and + =1.

[0036] The system assigns weights to two core flow indicators: traffic and population. When a disruption event occurs, the system proactively deducts the damage attenuation from the base indicators and uses a maximum value function to ensure that the deduction does not result in a negative number that loses its physical meaning. This dynamically outputs the corrected network topology connection weights. The propagation path constructed after weight reduction correction due to real-world disruption factors has the reliability to guide actual logistics scheduling.

[0037] Demand Spatiotemporal Quantitative Extrapolation Based on Node Level and Economic Indicator Differences: For any region to be predicted along the propagation path, the system extracts the topological hop count from the originating region and multiplies it by the time delay required for single-hop propagation in historical statistics to calculate the specific date the fluctuation arrives at the local area. When estimating the scale of the outbreak, the system aims to achieve a fusion prediction of macroeconomic constants and micro-level sudden variables. The formula for determining the final expected demand increment for the target sales region is as follows: : Indicates the final projected increase in demand for the target sales region, with a value range of . ≥0.

[0038] : Indicates the actual sales growth in the region where the fluctuation originated, with a value range of [value range missing]. ≥0.

[0039] : Indicates the number of permanent residents in the target sales region, and >0.

[0040] : Indicates the number of permanent residents in the region where the fluctuation originates, and >0.

[0041] : Represents the mapping weight of the population size factor.

[0042] : Represents the average wage income level in the target sales region, and >0.

[0043] : This represents the average wage income level in the region where the fluctuation originated, and >0.

[0044] : Represents the mapping weight of the income level factor, and + =1.

[0045] : Indicates whether there is a corresponding social hot topic related to the product. It is 1 if it exists, and 0 otherwise.

[0046] : Represents the preset hotspot amplification constant, which represents the fixed incremental conversion brought by hotspot traffic, and >0.

[0047] This calculation operation constructs a regional economic difference mapping matrix using the ratio of population to income, proportionally converts the outbreak scale of the starting region to the target region, and independently superimposes pulsed hotspot increments at the end of the formula, ultimately outputting the projected demand increment that closely matches the actual consumption potential of the region.

[0048] S3. Based on the expected arrival time of fluctuations, the expected increase in demand, and the status of supply chain nodes, determine the advance inventory gap for the target sales region, including the following steps: The total available inventory is obtained by adding the current inventory quantity of goods in stock and the quantity of goods in transit in the target sales region; the total demand is obtained by adding the expected increase in demand and the expected consumption quantity under stable conditions; the difference between the total demand and the total available inventory is calculated to obtain the initial stocking gap; the influence of the maximum capacity limit of the warehouse is introduced. If the sum of the initial stocking gap and the total available inventory is greater than the maximum capacity limit of the corresponding warehouse, the portion of the value exceeding the maximum capacity limit is removed, and the remaining gap value is determined as the advance stocking gap.

[0049] It should be noted that after obtaining the incremental demand forecast, the system needs to assess the actual carrying capacity of the supply chain infrastructure and perform physical inventory checks and stocking gap calculations. The specific steps are as follows: The system performs a comprehensive inventory check of physical inventory across the entire supply chain and establishes the boundary of total demand: It extracts existing inventory levels from physical warehouses in the target region and combines this with the volume of goods en route on trunk lines; these two quantities are combined to form the total available inventory to cope with anticipated fluctuations. Simultaneously, the system overlays the predicted surge in demand with the daily consumption base under stable conditions to generate total demand. By calculating the difference between total demand and total available inventory, the system obtains the initial stocking gap value.

[0050] A gap-cutting mechanism based on the physical space limits of warehousing: Conventional replenishment logic can easily lead to goods arriving at the warehouse exceeding its capacity and resulting in open-air storage losses. To mitigate this physical execution risk, the system internally prioritizes absolutely avoiding the risk of physical warehouse overload. Its pre-stocking gap-cutting calculation formula, based on the maximum warehouse capacity limit, is as follows: : Represents the final determined and output value of the advance stocking gap, with a value range of 0 ≤ ≤ .

[0051] This represents the total demand, calculated by adding the projected increase in demand to the amount of stable consumption. Its value ranges from [value missing]. ≥0.

[0052] This represents the total available inventory, consisting of the combined quantities of goods currently in stock and in transit. The value range is [value range missing]. ≥0.

[0053] : Indicates the maximum capacity limit set by the building structure of the corresponding target warehouse, and ≥ .

[0054] The calculation operation first calculates the theoretical initial inventory gap to prevent negative demand when there is excess inventory. Then, it performs a physical cutoff judgment, comparing the theoretical gap with the remaining available physical space in the warehouse. Once the theoretical gap exceeds the physical limit, the system automatically discards the excess part and outputs the gap value that strictly meets the physical constraints, thereby blocking the risk of backlog and warehouse overload at the end of the supply chain.

[0055] S4. Following the sales fluctuation propagation path, adjustable inventory in sales regions that have passed the peak of sales fluctuations is transferred in advance to target sales regions with pre-stocking gaps. Based on the remaining gaps after the transfer, supplementary procurement tasks are generated to form a supply chain collaboration solution, including the following steps: If the number of daily orders in a sales region at the forefront of the sales fluctuation propagation path decreases for a consecutive preset number of cycles and falls below a preset percentage of the historical peak, the corresponding sales region is determined to have passed the peak of the sales fluctuation. The remaining inventory in the sales region that has passed the peak is calculated, and the adjustable inventory is obtained after deducting the safety reserve quantity. Introducing the impact of the comparison between transfer costs and warehousing costs, if the calculated warehousing maintenance cost savings from transferring adjustable inventory are greater than the logistics transportation costs incurred by cross-regional transfers, a transfer instruction is issued to send the adjustable inventory to the target sales region based on the principle of proximity by physical distance. The remaining gap is obtained by subtracting the inventory quantity planned to be received through the transfer instruction from the pre-stocked inventory gap. The production and processing cycle of the upstream factory and the logistics transportation cycle to the target sales region are extracted. The production and processing cycle and the logistics transportation cycle are subtracted sequentially from the expected arrival time of the fluctuation to reverse-determine the deadline for issuing procurement tasks to the upstream production factory. Before the procurement deadline arrives, a supplementary procurement task order is generated based on the remaining gap. The transfer instruction and the supplementary procurement task order are integrated into a supply chain collaboration solution based on the time nodes.

[0056] It should be noted that after identifying the safety stock gaps at each node, the system prioritizes activating static assets within the supply chain network and coordinating external production capacity, executing horizontal cross-regional inventory transfers and vertical procurement plan generation. The specific operational steps are as follows: Determining Sales Regions During Sales Decline and Releasing Adjustable Inventory: Regions at the forefront of sales fluctuation propagation typically enter the sales decline phase earlier. The system monitors the transaction curves of these regions. When daily order throughput declines for several consecutive periods, and the current value falls below the set lower limit of the historical peak, the system determines that the region faces the risk of unsold inventory. The system then clears the inventory in this region, reserving a safety margin to support stable sales, and converts the remaining goods into adjustable inventory that can be responded to global scheduling.

[0057] Introducing Horizontal Inventory Smoothing and Transfer Scheduling Based on Economic Benefit Accounting: For adjustable inventory, the system is driven by maximizing the overall economic benefits of allocation. Before triggering a horizontal inventory transfer instruction, the allocation triggering constraint inequality for financial accounting is as follows: : Represents the total quantity of adjustable inventory to be transferred, and It is a positive integer.

[0058] This represents the daily storage and maintenance cost per unit of goods in the original warehouse, with a value range of [value range missing]. >0.

[0059] This indicates the number of days the surplus goods are expected to remain unsold in their original location if no transfer occurs, with a value ranging from [value missing]. >0.

[0060] This indicates the total fixed toll fee calculated based on the trunk line mileage for this inter-regional logistics dispatch, with a value range of [value missing]. >0.

[0061] This indicates the total dynamic fuel cost calculated based on the trunk line mileage for this inter-regional logistics dispatch, with a value range of [value missing]. >0.

[0062] This judgment constraint operation quantifies the total amount of accumulated warehousing waste generated by statically storing goods in a declining region, and compares it with the dynamic logistics fulfillment cost required to perform a cross-regional physical transfer. The system state machine is activated only when the warehousing savings on the left are significantly greater than the transportation costs on the right, and then a physical transfer instruction is issued based on the optimization criterion of the shortest physical distance.

[0063] Vertical Supplement Procurement and Solution Collaboration for Remaining Gaps Based on Time-Axis Reverse Calculation: After the internal network completes the horizontal allocation of existing inventory, the system initiates an external procurement process for any remaining unmet needs. The system extracts the actual production cycle time (production processing cycle) of upstream factories and the trunk line delay (logistics transportation cycle) for delivery to the target region. Using the predicted arrival time of fluctuations as a baseline, it performs reverse calculations on the timeline to deduce the latest order placement node (procurement deadline) to ensure no stockouts occur. Before the deadline, the system automatically generates supplementary procurement task orders matching the remaining gap size and integrates internal transfer instructions with external procurement orders into a seamless supply chain collaboration solution.

[0064] S5. Based on the actual evolution of the sales profile, actual regional sales fluctuations, and the execution results of the supply chain collaboration plan, revise the expected arrival time of fluctuations, expected demand increments, and unexecuted tasks for sales regions that have not yet experienced sales fluctuations, including the following steps: The system calculates the numerical difference between the projected increase in demand and the actual number of orders generated periodically. It introduces the possibility of fictitious transactions as a source of interference; if the difference exceeds a preset tolerance threshold, it determines that fictitious transactions exist in the previously collected data. It replaces historical correlation data with actual order quantities, recalculating and updating the projected increase in demand and the expected arrival time of fluctuations for sales regions following the sales fluctuation propagation path. Based on the updated projected increase in demand and the expected arrival time of fluctuations, it retrieves transfer instructions that are en route and have not yet been handed over, as well as supplementary purchase orders that have not yet reached their purchase deadline and have not been issued, from the computer system. Based on the updated values, it issues transportation redirection instructions to the corresponding logistics nodes to change the destination warehouse and production modification instructions to the corresponding upstream factories to adjust production quantities, thus correcting any unexecuted tasks.

[0065] It should be noted that during the implementation of the supply chain collaboration solution, the system must possess closed-loop control capabilities to handle abnormal interference and prediction distortion, dynamically recalculate global parameters, and intercept and modify physical execution tasks. The specific operation steps are as follows: Dynamic monitoring of factors interfering with fraudulent transactions and reconstruction of the underlying prediction baseline: In the business environment, malicious order-boosting and other illegal traffic injection behaviors can lead to significantly overestimations in the initial prediction. During the implementation of the solution, the system performs high-frequency monitoring, verifying and comparing the estimated order increase generated by the simulation with the actual order volume successfully settled by the business system. When the difference between the estimated high sales volume and the actual low sales volume exceeds the preset tolerance threshold, the system triggers an abnormal intervention mechanism, determining that the underlying data has been interfered with by fraudulent transactions. The system truncates the original simulation logic, calls up real low-end order data to overwrite the historical sequence, restarts the calculation process, and significantly lowers the estimated demand increase in downstream regions along the propagation path and corrects the expected arrival time.

[0066] For redirection of in-transit logistics and order intervention at the production front end: After parameter reconstruction, the system's error correction mechanism is directly sent to the physical execution end. The system searches the supply chain control network: For transportation tasks that have been dispatched but have not yet completed physical handover in the trunk network, the system issues transportation redirection instructions to the vehicle terminal, requiring the vehicle to change its destination and be diverted to other safe throughput nodes to prevent deep stagnation and inventory backlog; at the same time, the system intercepts purchase orders that have not yet reached their deadlines and sends production modification instructions to upstream factories to reduce or cancel production capacity, thereby building a physical fault-tolerant defense line for supply chain management in the face of false fluctuations.

[0067] Example 2 To further clarify the technical solution of the present invention, the following is a second embodiment of the present invention, using a specific business scenario of a large consumer electronics company conducting full-network supply chain collaborative scheduling for a new intelligent cooling device as an example, to provide a more detailed description of the technical solution of the present invention.

[0068] In this embodiment, with the changing seasons, the company's newly released smart cooling equipment experienced an initial surge in demand in some southern regions. To avoid regional inventory depletion or physical backlog caused by excessive allocation due to the sudden increase in demand, the company deployed the supply chain collaborative simulation system of this invention.

[0069] Step S1: Correlation of multi-source data and anchoring of dominant profile After the target product was launched, the system began extracting user interaction data, order transaction data, and warehousing and logistics data from various regions, and then aligned these data according to calendar days and city identification codes. During the data smoothing phase, for data gaps in City A caused by a short-term server failure, the system automatically extracted the average data from two adjacent days to fill in the gaps; for data surges in City B due to regional promotional activities, the system used the typical growth multiple of similar historical promotions to perform dimensionality reduction on the surge data. The system then transformed the cleaned data into feature values ​​and performed spatial clustering. The system comparison revealed that in City S, a group with basic portable fan purchase records showed a significantly higher growth rate in searching for and purchasing this high-priced smart cooling device within a specific time window, far exceeding other groups. The system then identified this upward migration trajectory of customer consumption characteristics as the dominant profile evolution of the target product in City S.

[0070] Step S2: Construction of propagation path and quantitative deduction of demand The system retrieved the global timeline and confirmed that City S was the earliest region in the entire network to experience this dominant profile evolution, thus establishing it as the starting point of the fluctuation. Starting with City S, the system extracted the frequency of trunk freight departures and passenger flow at transportation hubs between it and other regions nationwide. When introducing environmental disruption factors, the meteorological interface reported that the trunk highway leading to the neighboring City C was hit by torrential rain and freezing disasters. The system immediately and significantly reduced the traffic connectivity index to City C in its calculations. After weight reduction correction, the system calculated that City H, which was unaffected by the disaster, had the strongest correlation with City S, thus establishing City H as the next-level core node in the sales fluctuation propagation path. Based on the topological hierarchy, the system calculated that the expected arrival time of the fluctuation from City S to City H was 5 days. When quantifying the scale of the outbreak, the system extracted economic indicators such as the resident population and average wage income of City H, and calculated the basic demand increment as 8,500 units according to regional differences. At this time, the system detected a high-profile social hot topic related to the product on social media platforms in City H. The system automatically added a preset hot topic amplification constant of 2,000 units, and finally determined that the expected increase in demand in City H was 10,500 units.

[0071] Step S3: Calculation of inventory gap based on physical constraints For City H, the system inventory shows 1,000 units currently in stock and 500 units in transit, resulting in a total available inventory of 1,500 units. Combining this with the expected consumption of 1,000 units under stable conditions, the system calculates the total demand at the anticipated peak demand point to be 11,500 units, theoretically resulting in an initial inventory gap of 10,000 units. Upon entering the physical constraint verification phase, the system reads that the maximum building capacity of the corresponding warehouse in City H is 8,000 units. If shipments are made directly based on the initial gap, 3,500 units will exceed the actual warehouse capacity, leading to physical inconvenience and losses. After triggering the cutoff mechanism, the system removes the portion exceeding the limit, forcibly correcting and locking the final advance inventory gap at 6,500 units.

[0072] Step S4: Horizontal Inventory Transfer and Vertical Replenishment Procurement Coordination After identifying the shortage in City H, the system inspected the front-end nodes of the transmission path. The system detected that City G, at the forefront of the path, had seen a continuous decline in daily order volume for four consecutive periods, falling below 40% of its historical peak, indicating that City G had passed its peak sales fluctuation. After deducting the safety reserve quantity for City G, the system released 3,000 units of adjustable inventory. The financial accounting process then proceeded: if this inventory were left in City G, the estimated static warehousing maintenance cost would be 8,000 yuan; while the total dynamic logistics transportation cost for cross-regional transfer to City H would be 4,500 yuan. Since the warehousing savings outweighed the transfer transportation costs, the system issued a transfer instruction to send these 3,000 units to City H based on the principle of proximity. At this point, the remaining shortage was 3,500 units. The system extracted the upstream OEM's production and processing cycle of 3 days and the logistics transportation cycle to City H of 2 days. Using the estimated arrival time of the fluctuation (5 days) to work backwards along the timeline, the current date was determined as the deadline for issuing the purchase order to the OEM. The system then generated a supplementary purchase order for 3,500 units, which, together with the aforementioned transfer instructions, constituted a complete supply chain collaboration solution.

[0073] Step S5: Dynamic error correction and physical interception to deal with fraudulent transactions During the implementation period, the system conducted high-frequency checks and discovered that the actual number of completed payment orders in City H was only 50 units, a significant discrepancy exceeding the system's tolerance threshold compared to the previously projected increase in demand. Based on this, the system determined that malicious order manipulation and other fraudulent transactions had interfered with the initial baseline data. The system immediately replaced the historical data sequence with the actual order quantity, triggering a recalculation and drastically reducing the projected increase in demand for City H to 800 units. To prevent deep inventory buildup, the system urgently searched the entire network execution status: it located 3,000 units transferred from City G that were in transit and had not yet been handed over, while 3,500 supplementary purchase orders sent to the OEM factory had not yet entered the actual production scheduling stage. The system immediately issued a transportation redirection instruction to the corresponding logistics nodes, diverting the goods to nearby secondary nodes with actual throughput capacity; simultaneously, it issued a production modification instruction to the OEM factory, directly adjusting the production quantity of the 3,500 units to 0, thus successfully correcting and rectifying the unexecuted tasks at the physical execution level.

[0074] In summary, this invention constructs the propagation path of sales fluctuations by extracting the evolution of the dominant profile, and extrapolates the expected arrival time of fluctuations and the increase in demand; it determines the advance stocking gap by combining the upper limit of warehouse physical capacity; it coordinates the horizontal transfer and vertical replenishment procurement tasks of adjustable inventory, and dynamically intercepts and corrects unexecuted tasks based on the actual order accounting results. This solution combines dynamic profiling with spatial networks, improving the accuracy of demand forecasting, and effectively reduces the risk of ineffective logistics losses and inventory backlog by introducing physical constraints and a closed-loop error correction mechanism.

[0075] It should be noted that 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations, characterized in that, Includes the following steps: S1. Obtain customer behavior data, regional sales data, and supply chain status data, form a related data sequence according to product, sales region, and time, and determine the dominant profile evolution of the corresponding product in the corresponding sales region based on the changes in customer group status within adjacent statistical periods. S2. Based on the order of occurrence of the dominant profile evolution of the same product in different sales regions, and the time correspondence between the dominant profile evolution and regional sales fluctuations, determine the starting region of the fluctuation, construct the sales fluctuation propagation path, and determine the expected arrival time of the fluctuation and the expected demand increment of the target sales region in the sales fluctuation propagation path. S3. Based on the expected arrival time of the fluctuation, the expected increase in demand, and the status of supply chain nodes, determine the advance stocking gap for the target sales region; S4. According to the sales fluctuation propagation path, the adjustable inventory in the sales region that has passed the peak of the sales fluctuation is transferred in advance to the target sales region with the pre-stocking gap, and a supplementary procurement task is generated based on the remaining gap after the transfer to form a supply chain collaboration solution. S5. Based on the actual profile evolution, actual regional sales fluctuations, and the execution results of the supply chain collaboration plan, revise the expected fluctuation arrival time, expected demand increment, and unexecuted tasks corresponding to sales regions that have not yet experienced sales fluctuations.

2. The supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations as described in claim 1, characterized in that, Step S1 involves acquiring customer behavior data, regional sales data, and supply chain status data, forming a related data sequence based on product, sales region, and time, including: For a preset set of sales regions, user interaction data, order transaction data, and warehousing and logistics data within the sales regions are extracted. These data are then correlated and aligned according to product identification number, region identification code, and a unified time step to obtain an initial data sequence. For data loss due to communication interruptions, the average data from adjacent time points is used to fill in the gaps. For data surges caused by promotions, the data at the surge points is smoothed using the historical growth multiple for the same period, and the correlated data sequence is output.

3. The supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations as described in claim 2, characterized in that, Step S1, which determines the evolution of the dominant profile of a corresponding product in a corresponding sales region based on changes in customer group status within adjacent statistical periods, includes: The smoothed customer behavior data is transformed into feature values. Within a continuous time window, spatial clustering is used to cluster customers with similar features, constructing a set of group clusters. The customer status features of the group clusters contained in the set are extracted and compared within two adjacent time windows, and the growth rate of the group clusters' purchase of corresponding products is calculated. The label change process of the group cluster with the highest growth rate in the time series is determined as the dominant profile evolution of the corresponding product in the corresponding sales region.

4. The supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations as described in claim 3, characterized in that, Step S2, which involves determining the originating region of the fluctuation and constructing the propagation path of the sales fluctuation, includes: Extract the start time of the dominant profile evolution for the same product within the preset sales region set, and determine the sales region where the dominant profile evolution occurred earliest in the time series as the fluctuation initiation region; exclude the fluctuation initiation region from the sales region set to determine the remaining region set; starting from the fluctuation initiation region, extract the traffic connectivity index and population mobility index between the sales regions included in the remaining region set and the fluctuation initiation region; introduce the impact factors of extreme weather and geographical remoteness, deduct the corresponding traffic connectivity index and population mobility index for regions affected by the impact factors, calculate the correlation strength based on the deducted indexes, and connect the corresponding sales regions in descending order of correlation strength to form the sales fluctuation propagation path.

5. The supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations as described in claim 4, characterized in that, Step S2, which determines the expected arrival time and expected demand increment of the target sales region in the sales fluctuation propagation path, includes: Based on the number of node levels between the target sales region and the region where the fluctuation originates in the sales fluctuation propagation path, and combined with the average interval between sales outbreaks in adjacent regions, the expected arrival time of the fluctuation is calculated. The regional difference ratio between the target sales region and the region where the fluctuation originates in terms of socioeconomic indicators is calculated, and the actual sales growth of the region where the fluctuation originates is multiplied by the regional difference ratio to obtain the basic demand increment. If a social hotspot event related to the corresponding product is identified, a preset hotspot amplification constant is added to the basic demand increment to obtain the final expected demand increment.

6. The supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations as described in claim 5, characterized in that, Determining the advance stocking gap in the target sales region in step S3 includes: The total available inventory is obtained by adding the current inventory quantity of goods in stock and the quantity of goods in transit in the target sales region; the total demand is obtained by adding the expected increase in demand and the expected consumption quantity under stable conditions; the initial stocking gap is obtained by calculating the difference between the total demand and the total available inventory; the influence factor of the maximum capacity limit of the warehouse is introduced, and if the sum of the initial stocking gap and the total available inventory is greater than the maximum capacity limit of the corresponding warehouse, the portion of the value exceeding the maximum capacity limit is removed, and the remaining gap value is determined as the advance stocking gap.

7. The supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations as described in claim 6, characterized in that, In step S4, according to the sales fluctuation propagation path, the adjustable inventory in sales regions that have passed the peak of sales fluctuation is transferred in advance to the target sales regions where there is a pre-stocking gap, including: If the number of daily orders in a sales region at the forefront of the sales fluctuation propagation path is monitored to decrease for a preset number of consecutive cycles and is lower than a preset percentage of the historical peak, then the corresponding sales region is determined to have passed the peak of the sales fluctuation. The remaining inventory of the sales region that has passed the peak of the sales fluctuation is calculated, and the adjustable inventory is obtained after deducting the safety reserve quantity. The influencing factors of the comparison between transfer costs and warehousing costs are introduced. If it is calculated that the warehousing maintenance cost saved by transferring the adjustable inventory is greater than the logistics transportation cost generated by cross-regional transfer, then a transfer instruction to send the adjustable inventory to the target sales region is issued based on the principle of proximity by physical distance.

8. The supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations as described in claim 7, characterized in that, In step S4, a supplementary procurement task is generated based on the remaining gap after the transfer, forming a supply chain collaboration solution, including: The remaining gap is obtained by subtracting the inventory received through the transfer instruction plan from the pre-stocking gap; the production and processing cycle of the upstream factory and the logistics and transportation cycle to the target sales region are extracted; the production and processing cycle and the logistics and transportation cycle are subtracted sequentially from the expected fluctuation arrival time to reverse-determine the deadline for issuing procurement tasks to the upstream production factory; before the procurement deadline is reached, a supplementary procurement task order is generated based on the remaining gap; the transfer instruction and the supplementary procurement task order are integrated into the supply chain collaboration solution according to the time nodes.

9. The supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations as described in claim 8, characterized in that, Step S5, which corrects the expected arrival time of sales fluctuations and the expected increase in demand for sales regions that have not yet experienced sales fluctuations, includes: The numerical difference between the projected increase in demand and the actual number of orders is calculated periodically. False transaction interference factors are introduced, and when the numerical difference exceeds a preset tolerance threshold, it is determined that the previously collected data contains such false transaction interference factors. The historical related data is replaced with the actual number of orders, and the projected increase in demand and the projected arrival time of the fluctuation in the sales region ranked later in the sales fluctuation propagation path are recalculated and updated.

10. The supply chain collaborative extrapolation method based on multi-dimensional group profile evolution and regional sales fluctuations as described in claim 9, characterized in that, The step S5, which involves correcting unexecuted tasks, includes: Based on the updated projected demand increment and the expected arrival time of fluctuations, the computer system is retrieved for transfer instructions that are in transit and have not yet been handed over, as well as supplementary purchase orders that have not yet been issued and have not yet reached the purchase deadline. Based on the updated values, a transportation redirection instruction to change the destination warehouse is issued to the corresponding logistics node, and a production modification instruction to adjust the production quantity is issued to the corresponding upstream factory to complete the correction of the unexecuted tasks.

Images