O2O retail intelligent product selection method

Abstract

The application discloses an O2O retail intelligent product selection method, and relates to the technical field of retail product selection, which comprises the following steps: embedding potential demand into the same inventory position of an existing product combination and a candidate product combination, taking the same inventory position as a common starting point, alternately advancing demand deduction and inventory replenishment along the warehouse-store replenishment time sequence, tracing the performance attribution of adjacent store orders along the distribution coverage overlapping boundary, and stripping the performance address increment to generate a candidate product area replacement record; according to the candidate product area replacement record, inversely pushing the net replacement amount between the final performance store and the supply warehouse and positively pushing the warehouse-store available capacity, aligning the replenishment boundary at the meeting position, and sequentially connecting the candidate product execution quantity to obtain a product selection allocation record, positioning the affected product selection content in the product selection allocation record, and implementing local update to generate an O2O retail intelligent product selection list. The application improves the accuracy of regional product selection quantity allocation and the stability of local update.

Description

Technical Field

[0001] This invention relates to the field of retail product selection technology, and in particular to an O2O retail intelligent product selection method. Background Technology

[0002] With the development of instant retail, in-store and home delivery integration, and warehouse-store integrated fulfillment, O2O retail product selection is gradually shifting from category configuration relying on human experience to big data processing based on orders, searches, browsing, inventory, replenishment, store coverage, and delivery fulfillment records. Existing methods typically collect online demand and offline sales data according to product identification, store identification, and business time, and then use sales forecasting, turnover analysis, and inventory constraints to evaluate the value of introducing, retaining, and eliminating products, thereby forming a candidate product list and execution quantity for regional stores.

[0003] In scenarios involving stockouts, delayed replenishment, discontinuous warehouse and store inventory, and overlapping delivery ranges of adjacent stores, existing methods typically use completed orders, actual sales volume, and final fulfillment store records directly as the basis for demand intensity. However, purchasing intentions after supply disruptions may not materialize into orders due to lack of stock, and orders originally belonging to one store may be transferred to adjacent stores. This results in observed demand being simultaneously affected by the lack of unfulfilled demand and the incremental increase in fulfillment due to relocation. The main reason is that existing processing methods establish statistical correspondences based on existing transactions and inventory snapshots, without tracing back unfulfilled demand along the supply disruption location, and without simultaneously extrapolating demand deduction, inventory replenishment, and cross-store fulfillment attribution from the same inventory starting point. Therefore, it is difficult to form a basis for candidate product replacement that reflects the region's true demand-bearing capacity. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an O2O retail intelligent product selection method that solves the problem of inaccurate cross-domain potential demand restoration caused by supply disruptions and fulfillment relocation.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides an intelligent product selection method for O2O retail, comprising: associating O2O retail product selection data with identities and aligning them with time series to generate an O2O product selection time series benchmark chain; matching online demand changes with offline fulfillment changes according to the O2O product selection time series benchmark chain to obtain demand fulfillment matching records; performing time-delay coupling variable point tracking of supply gaps and fulfillment stagnation from the demand fulfillment matching records and locating the supply blockage position; repatriating unmet demand after the supply blockage to before the supply blockage position to generate cross-domain potential demand acceptance records; and embedding potential demand into existing product combinations and candidate product combinations based on the cross-domain potential demand acceptance records. Starting from the same inventory location, demand deduction and inventory replenishment are alternately advanced along the order of warehouse and store replenishment time. The fulfillment attribution of orders from adjacent stores is traced along the overlapping boundary of delivery coverage, and the fulfillment transfer increment is stripped to generate candidate product area replacement records. Based on the candidate product area replacement records, the net replacement volume is reversed between the final fulfillment store and the supply warehouse, and the available capacity of the warehouse and store is forwarded. The replenishment boundary is aligned at the meeting point, and the candidate product execution quantity is matched one by one to obtain the product selection allocation record. The affected product selection content is located in the product selection allocation record and partial updates are implemented to generate an O2O retail intelligent product selection list.

[0007] As a preferred embodiment of the O2O retail intelligent product selection method of the present invention, the specific steps for matching changes in online demand with changes in offline fulfillment are as follows: Based on the O2O product selection timeline benchmark chain, online demand content and offline fulfillment content are arranged according to product identification and store fulfillment scope. The quantity changes of adjacent time positions are compared, and the positions with the same direction of change and continuous connection are connected as online demand change segments and offline fulfillment change segments. Using the end position of each online demand change segment as the tracking starting point, find offline fulfillment change segments with consistent product identification and overlapping store fulfillment ranges within the fulfillment response scope, retain the continuity relationship with consistent change direction, and form demand fulfillment candidate relationships; Based on the candidate relationship of demand fulfillment, the changes in online demand and the changes in offline fulfillment that have not yet been occupied are successively accepted. When a change in online demand corresponds to multiple changes in offline fulfillment, the unpaired portion is allocated in sequence. When multiple changes in online demand correspond to a change in offline fulfillment, the remaining fulfillment quantity is occupied in sequence. The matching position and the unaccepted demand are registered, and the demand fulfillment matching record is obtained.

[0008] As a preferred embodiment of the O2O retail intelligent product selection method of the present invention, the specific steps for tracking and locating the supply blockage location by performing time-delay coupling variable point tracking of supply gaps and fulfillment stagnation from the demand fulfillment matching records are as follows: Based on the demand fulfillment matching records, starting from the matching start position of unmet demand, the existing store inventory, in-transit replenishment, unmet demand, and fulfillment volume are connected in chronological order along the fulfillment response range. At each time position, the supply gap where the net available quantity fails to cover the unmet demand and the fulfillment stagnation state where the fulfillment volume has not increased are recorded, generating a supply and demand response comparison sequence. Based on the supply and demand response comparison sequence, the continuous changes of supply gap and performance stagnation before and after adjacent time positions are compared. The first position where the supply gap changes from non-existent to continuous occurrence is marked as the supply gap change point, and the first position where the performance volume changes from continuous increase to continuous non-increase is marked as the performance stagnation change point. The continuous state interval before and after each change point is retained to generate a supply and demand dual change point record. Based on the supply and demand dual change point records, each supply gap change point is used as the coupling starting point. The earliest performance stagnation change point within the performance response range and located after the supply gap change point is selected. The continuous state interval before and after the performance stagnation change point is shifted along the time axis towards the supply gap change point until the two change points are at the same time position. The response time delay corresponding to the shift distance is recorded, and the combination of change points where the supply gap interval and the performance stagnation interval continuously overlap after the shift is retained to generate a time delay coupled change point sequence. Based on the time-delay coupled variable point sequence, the response time delay is extended to the supply recovery position and the fulfillment recovery position. It is verified whether the fulfillment volume recovers and increases according to the same response time delay after the supply gap disappears. Variable point combinations with matching recovery relationships and variable point combinations where the supply gap exists continuously and the fulfillment volume stagnates continuously before the fulfillment response ends are retained. Variable point combinations where the supply gap appears briefly and there is no corresponding fulfillment stagnation are eliminated. The earliest supply gap variable point is registered as the supply interruption position.

[0009] As a preferred embodiment of the O2O retail intelligent product selection method of the present invention, the step of generating a cross-domain potential demand acceptance record before repatriating unmet demand after supply disruption to the supply disruption location includes the following specific steps: Tracing back from the point of supply disruption, the most recent time point in which net availability can cover demand and demand fulfillment matching is closed is recorded as the demand return range, and the continuous interval between the most recent time point and the time point before the supply disruption point is registered. Using the product identifier and store fulfillment range corresponding to the demand return scope as filtering conditions, the unaccepted demand after the supply blockage location is extracted, and duplicate demand is removed according to the demand event identifier. The remaining unaccepted demand is then associated with the demand return scope to generate a record of demand to be returned. Based on the pending demand records, within the demand return range, target time positions are selected where both the online demand change and the remaining capacity can accommodate the unaccepted demand. The unaccepted demand is then allocated in combination with the time interval from the supply blockage position to generate a demand return allocation sequence. Based on the demand return allocation sequence, the demand return amount at each target time position is adjusted, and the remaining demand generated by the adjustment is sequentially added to the target time position with the larger capacity. Then, the demand return amount is connected with the corresponding online demand change segment and supply blockage position and merged with the actual performance and acceptance relationship to generate a cross-domain potential demand acceptance record.

[0010] As a preferred embodiment of the O2O retail intelligent product selection method of the present invention, the specific steps of embedding potential demand into the corresponding inventory positions of existing product combinations and candidate product combinations are as follows: Extract potential demand feedback information from the cross-domain potential demand acceptance records, associate the potential demand feedback information with products in the existing product portfolio and candidate product portfolio that have the same product category and whose delivery coverage includes the accepting store, convert the product quantity according to the packaging specifications and minimum sales unit, and generate the warehouse store replenishment chain. Arrange the inventory change positions of existing product combinations and candidate product combinations along the warehouse replenishment chain, and form the cumulative net available quantity and remaining capacity before each inventory position. Record the replenishment arrival time and delivery coverage boundary, and connect the existing inventory positions and candidate inventory positions that intersect within the delivery coverage of the fulfillment response range to generate inventory position candidate pairing records. Based on the candidate pairing records of inventory locations, each pair of locations is arranged from smallest to largest according to the cumulative net availability difference. For pair of locations with the same difference, they are arranged in order of decreasing delivery coverage overlap, increasing replenishment arrival time offset, and decreasing sum of remaining capacity on both sides. The position pair at the top of each potential demand inventory anchor point is retained to generate corresponding inventory locations. Based on the same inventory position, the common minimum sales equivalent corresponding to the same potential demand event is embedded into the inventory event sequence of the existing product portfolio and the candidate product portfolio respectively. A demand deduction position with the same sort number is set before the two same inventory positions. The time position of the original inventory event remains unchanged and is associated with the cumulative net available quantity before embedding and the embedded quantity to generate potential demand same embedding record.

[0011] As a preferred embodiment of the O2O retail intelligent product selection method described in this invention, the alternating process of demand deduction and inventory replenishment includes the following specific steps: Based on the potential demand embedded records, using the same inventory location as a common starting point, extract the demand deduction location and replenishment arrival location from the common starting point to the end of the fulfillment response time, merge the replenishment arrival time and remove duplicate times, divide the adjacent replenishment arrival time into deduction intervals, register the actual replenishment quantity of existing product combinations and candidate product combinations at each replenishment boundary, and generate a deduction and replenishment advancement boundary sequence. Based on the deduction and replenishment advancement boundary sequence, the unaccepted demand left over from the previous deduction interval is connected at the beginning of each deduction interval of the existing product combination and the candidate product combination. The demand deduction amount is recorded as the smaller value between the current net available quantity and the demand quantity. The undeducted part is carried into the next deduction interval, and the confirmed actual replenishment quantity is added at the end of the interval. When the demand deduction position and the replenishment boundary are at the same time, the demand formed before the replenishment confirmation is deducted first and then the actual replenishment quantity is added to generate a round-by-round deduction and replenishment record. Based on the round-by-round deduction and replenishment records, the remaining inventory and unaccepted demand of the existing product combination and the candidate product combination at the same replenishment boundary are compared round by round. The deduction and replenishment offset status is marked in combination with the demand fulfillment time limit status, and the deduction and replenishment offset status is connected to generate a deduction and replenishment offset sequence. Based on the offset sequence, the sub-round offset records are connected along the offset advancement boundary sequence. When the unaccepted demand of the existing product combination is zeroed first, a zero value is reserved for the existing product combination. When the unaccepted demand of the candidate product combination is zeroed first, a zero value is reserved for the candidate product combination. The sub-round offset records that have not yet been zeroed out continue to be connected until all unaccepted demand is zeroed out. When the fulfillment response end time is earlier than the demand zeroing position, the connection is stopped at the fulfillment response end position, and an inventory alternation advancement record is generated.

[0012] As a preferred embodiment of the O2O retail intelligent product selection method of the present invention, the specific steps of tracing the fulfillment attribution of orders from adjacent stores along the overlapping boundaries of delivery coverage and stripping the fulfillment transfer increment to generate candidate product area replacement records are as follows: Extract the earliest order with the offset status from the inventory alternation progress record, map the corresponding order's delivery address grid to the delivery coverage overlap boundary, use the obtained boundary mapping position as the tracing starting point, extract adjacent store orders segment by segment along the two extension directions of the delivery coverage overlap boundary, and associate the fulfillment time with the replenishment boundary to generate overlap boundary order records; Starting from the root order identifier in the overlapping boundary order record, trace back from the final fulfillment store to the original receiving store along the order transfer order. Connect the root order identifier, the delivery address grid and the order release position and store access position that are consistent with the product category. Compress the intermediate transfer positions that the same order passes through. Register the original receiving store, the final fulfillment store, the transfer direction and the fulfillment quantity to generate an order fulfillment migration record. Based on the order fulfillment migration records, within each overlapping boundary segment, the fulfillment volume of the candidate product combination and the existing product combination for the same root order is compared. The increased fulfillment volume of the candidate product combination is balanced with the release volume of the corresponding original receiving store and the final fulfillment store access volume. The balanced part is registered as the fulfillment transfer increment. The fulfillment transfer increment is separated from the increased fulfillment volume, and the remaining net replacement volume is associated with the candidate product and the corresponding overlapping boundary segment to generate the regional net replacement sequence. The duplicate registrations of the same order in multiple overlapping boundary segments in the regional net replacement sequence are removed. The first overlapping boundary segment that the order passes through when leaving the original receiving store's delivery coverage area and the final fulfillment store are retained. Continuous overlapping boundary segments with the same candidate products and the same transfer direction are merged and the net replacement volume and store fulfillment attribution are aggregated to generate candidate product regional replacement records.

[0013] As a preferred embodiment of the O2O retail intelligent product selection method of the present invention, the overlapping boundary of delivery coverage is obtained by mapping the delivery coverage of the receiving store and the adjacent stores to the same geographical coordinate space, and extracting the continuous edge positions of the overlapping areas and connecting them sequentially.

[0014] As a preferred embodiment of the O2O retail intelligent product selection method of the present invention, the steps of reverse-engineering the net replacement volume between the final fulfillment store and the supply warehouse and forward-engineering the available capacity of the warehouse store, aligning the filling boundary at the meeting position, matching each position to the execution quantity of candidate products, and obtaining the product selection allocation record are as follows: Based on the candidate product area replacement record, the net replacement volume is split according to the candidate product identifier and the final fulfillment store. The supply warehouse and replenishment path that can be reached before the fulfillment response ends are connected along the warehouse-store replenishment relationship. The net replacement volume is expanded into the common minimum sales equivalent position with sequential number. Starting from the final fulfillment store receiving boundary, the common minimum sales equivalent position is transmitted through the replenishment path arrival boundary, the replenishment path sending boundary and the supply warehouse sending boundary in sequence to generate the net replacement volume reverse sequence. Based on the reverse sequence of net replacement volume, the remaining inventory of the supply warehouse, the remaining inventory issued by the warehouse, the remaining inventory of the replenishment path, and the remaining inventory received by the final fulfillment store are each expanded into sequentially numbered boundary capacity positions. Starting from the boundary of the supply warehouse, the boundary capacity positions are passed sequentially to the boundary of the replenishment path, the boundary of the replenishment path, and the boundary of the final fulfillment store. Boundary capacity positions with the same number and not occupied are retained in adjacent replenishment boundaries, and the passing of boundary capacity positions that lack subsequent corresponding positions is stopped, thus generating a forward sequence of warehouse and store available capacity. Based on the reverse sequence of net replacement volume and the forward sequence of available warehouse capacity, the corresponding content is connected according to the candidate product identifier, replenishment path and arrival time. The replenishment boundary that the common minimum sales equivalent position and the boundary capacity position first pass through is registered as the meeting position. The numbering order on both sides of the meeting position is aligned and matched one by one. After the net replacement volume position is matched, the transmission of the same numbered net replacement volume position in the remaining replenishment path is stopped. The unmatched net replacement volume position continues to be transmitted to the replenishment boundary on the supply warehouse side, and the unoccupied boundary capacity position continues to be transmitted to the replenishment boundary on the final fulfillment store side, generating the replenishment boundary matching sequence. Based on the replenishment boundary matching sequence, consecutive matching positions of the same candidate product are connected in the order of replenishment boundary time. It is checked whether each matching position has a net replacement quantity reverse push mark, a warehouse / store available capacity forward push mark and the same replenishment boundary mark. Matching positions with unclosed marks are eliminated and the corresponding boundary capacity positions are released. Matching positions with closed marks are aggregated to form the candidate product execution quantity. The remaining inventory of the supply warehouse, the fulfillment remaining quantity of the replenishment path and the receiving remaining quantity of the final fulfillment store are occupied simultaneously. The supply warehouse, the final fulfillment store and the replenishment path are registered to generate the product selection allocation record.

[0015] As a preferred embodiment of the O2O retail intelligent product selection method of the present invention, the specific steps for locating the affected product selection content and implementing partial updates to generate an O2O retail intelligent product selection list are as follows: Based on the product selection allocation record, the execution quantity, supply warehouse, final fulfillment store and replenishment path of the candidate product are connected according to the candidate product identifier, and the execution quantity of the candidate product is registered as the change difference, generating a product selection change fingerprint record; Based on the fingerprint record of product selection changes, the system matches candidate products with the product selection content corresponding to the final fulfillment store in the existing O2O retail product selection content, positions the newly added candidate products in the same product category, tracks the related content affected by the change difference, and generates a record of the scope of product selection impact. Based on the product selection impact range record, form partial change content according to the product selection change fingerprint and write it to a temporary copy. Check the partial update closure status and re-incorporate the unclosed content into the product selection impact range record until the number of changes and related content are all closed, and generate a partial update closure record. Based on the partial update closure record, the version identifiers of the partial changes and the existing O2O retail product selection content are compared. Product selection content with the same version is replaced and conflicting content is rechecked. Unaffected product selection content is retained. The updated content is arranged according to the final fulfillment store, product category and effective period to generate an O2O retail smart product selection list.

[0016] The beneficial effects of this invention are as follows: by embedding cross-domain potential demand into the corresponding inventory positions of existing product combinations and candidate product combinations, demand deduction and inventory replenishment are alternately promoted along the order of warehouse and store replenishment time, and the fulfillment attribution of adjacent store orders is traced along the overlapping boundaries of delivery coverage, and the incremental fulfillment transfer is stripped, so as to realize the corresponding big data processing of potential demand, inventory supply and fulfillment scope, which is used to form candidate product regional replacement records, reduce the interference of stockouts and cross-store acceptance on the judgment of product value, and improve the accuracy of regional product selection quantity allocation and the stability of local updates. Attached Figure Description

[0017] 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.

[0018] Figure 1 A flowchart for intelligent product selection methods in O2O retail.

[0019] Figure 2 A flowchart for matching changes in online demand with changes in offline fulfillment.

[0020] Figure 3 A flowchart for tracing the location of supply disruptions.

[0021] Figure 4 A flowchart for alternating demand deduction and inventory replenishment. Detailed Implementation

[0022] 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.

[0023] 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.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides an O2O retail intelligent product selection method, including the following steps: S1. Extract product selection association data from O2O retail product selection data. This O2O retail product selection data includes online demand data, offline fulfillment data, store inventory data, supply replenishment data, and order fulfillment data. Online demand data includes product identifier, quantity demanded, time of demand, and location of demand. Offline fulfillment data includes product identifier, quantity fulfilled, fulfilling store, time of fulfillment, and location of fulfillment. Store inventory data includes inventory quantity, inventory change time, and inventory location. Supply replenishment data includes supply warehouse, replenishment quantity, replenishment time, and replenishment path. Order fulfillment data includes order identifier, product identifier, order location, and fulfillment status. Filter data within the same business cycle and with complete product identifier, business time, and quantity fields according to the product selection analysis time range to form product selection association data. Unify the identity representation of product selection association data according to the product selection business affiliation relationship, merging identity content pointing to the same product selection business object into a unique identity position, and associating the unique identity position with the corresponding product. For business content connection, missing identity content is supplemented by searching for the identity position with the shortest time interval and sufficient business quantity along the already connected product selection business relationships. For conflicting identity content, the continuity of product selection business relationships is compared and the unique connection with the highest continuity is retained, forming an O2O retail product selection data identity association record. Based on the O2O retail product selection data identity association record, the occurrence time of each product selection business content is converted to a unified time base, and the product selection business content is arranged in chronological order. For product selection business content at the same time position, the order of demand first, fulfillment second, and deduction before supplementation is determined, forming an O2O retail product selection data time sequence arrangement record. Based on the O2O retail product selection data time sequence arrangement record, the continuous change positions of the same product selection business object are linked along the unique identity position. Product selection business content that is not directly connected is connected to the existing change position with the shortest time interval and overlapping fulfillment scope, and product selection status data is registered at each change position, generating an O2O product selection time sequence base chain.

[0026] It should also be noted that the O2O product selection timeline benchmark chain is a continuous chain of association formed by linking product selection business content according to a unified business identity and time sequence after identifying and aligning O2O retail product selection data. It is used to unify the corresponding positions of online demand changes and offline fulfillment changes, providing a timeline basis for demand fulfillment matching, supply disruption tracking, and potential demand re-entry, reducing demand duplication, fulfillment mismatch, and product selection deviation caused by inconsistent identities and time misalignments.

[0027] S2. Based on the O2O product selection time-series benchmark chain, match online demand changes with offline fulfillment changes, obtain demand fulfillment matching records, perform time-delay coupling variable point tracking of supply gaps and fulfillment stagnation from the demand fulfillment matching records, locate the supply blockage position, and return the unaccepted demand after the supply blockage to before the supply blockage position to generate cross-domain potential demand acceptance records.

[0028] S2.1. Based on the O2O product selection timeline benchmark chain, using product identifiers as the primary arrangement position and store fulfillment ranges as the secondary arrangement position, online demand content and offline fulfillment content with the same product identifier and consistent store fulfillment range are collected separately. Within each secondary arrangement position, adjacent time positions are connected in chronological order. For online demand content and offline fulfillment content that appear repeatedly at the same time position, their corresponding quantities are sequentially added according to their positional order in the O2O product selection timeline benchmark chain, forming an online demand content arrangement sequence and an offline fulfillment content arrangement sequence. Based on the online demand content arrangement sequence and the offline fulfillment content arrangement sequence, the difference between the quantity at the later time position and the quantity at the previous time position is calculated, and the difference is used to determine the order of occurrence. Values ​​greater than zero, differences less than zero, and differences equal to zero are recorded as increases, decreases, and stable states, respectively, forming online demand quantity change records and offline fulfillment quantity change records. Based on the online demand quantity change records and offline fulfillment quantity change records, starting from the first non-stable time position, the records are traced chronologically. Quantity change positions with the same product identification, consistent store fulfillment scope, consistent change direction, and consecutive time positions are connected sequentially. The connection is terminated when the product identification changes, the store fulfillment scope changes, the change direction changes, or the time position is broken. The connection is then restarted from the first non-stable time position after the termination position. Change segment information is recorded according to the scope of each connection, forming online demand change segments and offline fulfillment change segments.

[0029] The expression for calculating the difference between the number of positions at a later time step and the number of positions at a previous time step is: ; in, For the first The difference in the number of content categories corresponding to each time location; For the first The number of content categories corresponding to each time location after stacking; For the first The number of times the same content category is superimposed; This refers to the smallest sales unit for the corresponding product. This refers to the time position sequence number; = Content for online demand; = This refers to offline contract fulfillment content.

[0030] S2.2. Using the end position of each online demand change segment as the tracking starting point, find offline fulfillment change segments with consistent product identification and overlapping store fulfillment ranges within the fulfillment response scope, retain the continuity relationship with consistent change direction, and form demand fulfillment candidate relationships.

[0031] It should be noted that, based on the online demand tracking anchor point, the promised fulfillment duration in the corresponding order fulfillment configuration, the maximum response duration in the store fulfillment configuration, and the fulfillment end time in the actual fulfillment record are obtained. The end time of the online demand change segment is taken as the start time of the fulfillment response range, and the long time position corresponding to the promised fulfillment duration and the maximum response duration in the store fulfillment configuration is taken as the end time of the fulfillment response range. The continuous time interval between the start time and the end time is registered as the fulfillment response range.

[0032] Based on the fulfillment response range, offline fulfillment change segments within the fulfillment response range and overlapping with the online demand tracking anchor point are retrieved sequentially. The product identifiers of the online demand change segments and offline fulfillment change segments are checked segment by segment, retaining those with consistent product identifiers to form a record for the same product fulfillment segment. Based on this record, the store fulfillment ranges of the online and offline fulfillment change segments are mapped to the same delivery area, connecting store fulfillment ranges with shared coverage locations. Offline fulfillment change segments without shared coverage locations are removed, forming a record for overlapping fulfillment segments. Based on this record, the change directions of the online and offline fulfillment change segments are compared one by one, retaining connections with consistent change directions and disconnecting those with inconsistent directions. Multiple offline fulfillment change segments connected to the same online demand change segment are arranged in ascending order of time interval from the tracking start point. If the time intervals are the same, they are arranged in descending order of overlap in store fulfillment ranges. The arranged online and offline fulfillment change segments are then connected segment by segment to form a demand fulfillment candidate relationship.

[0033] S2.3. Based on the candidate relationship of demand fulfillment, arrange the acceptance relationship according to the time sequence of the end position of the online demand change segment. When the same online demand change segment connects multiple offline fulfillment changes, arrange them according to the candidate order to form a candidate acceptance order. Based on the candidate acceptance order, register the absolute amount of online demand change as the unpaired part, and register the offline fulfillment change after deducting the existing pairing amount as the unused offline fulfillment change amount. Take the smaller amount between the unpaired part and the unused offline fulfillment change amount as the pairing amount, connect the corresponding change segments and deduct the pairing amount simultaneously to form a segment-by-segment acceptance record. According to the segment-by-segment acceptance record, when there is still an unpaired part in an online demand change segment, it is sequentially allocated to the subsequent offline fulfillment change segment. When there is still a remaining fulfillment amount in an offline fulfillment change segment, the subsequent online demand change amount is sequentially accepted. Register the acceptance connection position as the pairing position, and register the remaining unpaired part after the candidate relationship processing is completed as the unaccepted demand. Combine the pairing positions and the unaccepted demand to obtain the demand fulfillment pairing record.

[0034] S2.4 Extract the starting position of the matching for unfulfilled demands from the demand fulfillment matching record. Register the position where the first unfulfilled demand appears as a common starting point and associate it with the product identifier, store fulfillment scope, and time position corresponding to the common starting point. Based on the common starting point, arrange consecutive time positions in chronological order along the fulfillment response scope. Extract the store's current inventory, in-transit replenishment, unfulfilled demand, and fulfillment volume corresponding to each time position. Connect the store's current inventory, in-transit replenishment, unfulfilled demand, and fulfillment volume at the same time position to form a supply and demand time association record. The supply and demand time correlation record summarizes the store's current inventory and replenishment in transit at each time point and compares them with the corresponding unmet demand. When the summarized available quantity is less than the unmet demand, the uncovered quantity is recorded as a supply gap. When the fulfillment quantity does not increase relative to the previous time point, the corresponding position is recorded as a fulfillment standstill. Based on the supply gap and fulfillment standstill status recorded at each time point, the supply gap position and the fulfillment standstill position are connected in chronological order, and associated with the corresponding store inventory, replenishment in transit, unmet demand, and fulfillment quantity to generate a supply and demand response comparison sequence.

[0035] S2.5. Based on the supply and demand response comparison sequence, arrange the supply gap status and fulfillment stagnation status corresponding to each time position in chronological order, and extract the state change relationship between adjacent time positions; based on the supply gap state change relationship, compare the existence of supply gaps at each time position. When there is a supply gap at the current time position and no supply gap at the previous time position, register the current time position as the initial position of the supply gap, and continue to check whether the supply gap continues to exist at subsequent time positions in chronological order, retaining the time interval where the supply gap exists continuously, and marking the first occurrence position as the supply gap change point; based on the fulfillment stagnation status change relationship, compare the adjacent time positions in chronological order. The change in fulfillment volume at a given time position is recorded as follows: when the fulfillment volume stops increasing at the current time position while the previous time position was still in a state of increasing fulfillment volume, the current time position is registered as the initial position of fulfillment stagnation. The system continues to check whether subsequent time positions maintain a state of fulfillment stagnation, retaining the continuous time intervals without increases, and marking the first occurrence of such an interval as the fulfillment stagnation change point. Based on the supply gap change point and the fulfillment stagnation change point, the system connects the continuous normal state intervals before the change point appears forward and the continuous abnormal state intervals after the change point appears backward. The system associates the supply gap state interval corresponding to the supply gap change point with the fulfillment stagnation state interval corresponding to the fulfillment stagnation change point according to time position, forming a dual supply and demand change point record.

[0036] S2.6. Based on the supply and demand dual-point records, arrange the supply gap points in chronological order and extract the corresponding set of performance stagnation points for each supply gap point. Using each supply gap point as the coupling starting point, search along the performance response range for the earliest performance stagnation point located after the supply gap point, and register the corresponding performance stagnation point as a performance stagnation point to be coupled. Based on the performance stagnation points to be coupled, extract the continuous state interval before and after the occurrence of the performance stagnation point, and move the performance stagnation point and continuous state interval forward position by position according to the interval between the time position and the supply gap point. During this process, the time position corresponding to the performance stagnation variable point is updated synchronously for each time position moved, until the performance stagnation variable point and the supply gap variable point are at the same time position. The number of time positions moved is recorded as the response time delay. Based on the response time delay, the time position overlap of the performance stagnation interval and the supply gap interval after the move is checked. The variable point combination that the supply gap state continues to exist and the performance stagnation state corresponds to the continuous state is retained. The variable point combination that does not have a continuous overlapping relationship is removed. The retained variable point combination, response time delay and corresponding state interval are connected in the order of the supply gap variable point time to generate a time delay coupled variable point sequence.

[0037] S2.7 Extract the supply gap change point, fulfillment stagnation change point, and response time delay corresponding to each set of time-delay coupled change point sequences. Continue to track the supply recovery position along the time position after the supply gap change point, and continue to track the fulfillment recovery position along the time position after the fulfillment stagnation change point. Register the position where the supply gap first disappears as the supply recovery position, and register the position where the fulfillment quantity starts to increase continuously again as the fulfillment recovery position. Based on the supply recovery position and the fulfillment recovery position, extend the time interval corresponding to the supply recovery position according to the response time delay in the time-delay coupled change point sequence. Compare the time positions of the extended supply recovery position and the fulfillment recovery position. When the fulfillment quantity starts to increase continuously again... When the supply recovery position and the extended supply recovery position are at the same time, the fulfillment recovery relationship after the supply gap recedes is recorded, and the response point combination is retained. Based on the retained recovery relationship, the state changes before the end time of the fulfillment response are checked. When the supply gap persists and the fulfillment volume does not increase, the response point combination is retained as a continuous blocking change point combination. Based on the change point combination with the matching recovery relationship and the continuous blocking change point combination, the change point combination with a short supply gap occurrence time and no corresponding fulfillment stagnation state is eliminated. The retained change point combinations are arranged in the order of the supply gap change points, and the earliest supply gap change point is registered as the supply blocking position.

[0038] S2.8. Tracing back from the supply interruption point, the most recent time position where net available supply can cover demand and demand fulfillment matching is closed, and registering the continuous interval between the most recent time position and the time position before the supply interruption point as the demand return range.

[0039] It should be noted that, starting from the time position before the supply interruption point, the supply and demand tracking sequence is traced one by one from the back to the front along the time position. The net supply and demand are compared at each time position. When the net supply is greater than or equal to the demand, the demand coverage position is marked.

[0040] According to the demand coverage location, verify the demand fulfillment matching record. When all the corresponding online demand changes have entered the matching location and there is no unaccepted demand, mark the demand fulfillment matching as closed and connect the demand coverage location with the demand fulfillment matching closure to form a return start verification record. According to the return start verification record, search from back to front along the time position for the first position that simultaneously satisfies the requirement that the net available quantity can cover the demand and the demand fulfillment matching is closed, and register the first position as the most recent time position. Starting from the most recent time position, connect adjacent time positions one by one in chronological order, stopping when connecting to the time position before the supply blockage position, and register the continuous interval formed by the connection as the demand return range.

[0041] S2.9. Using the product identifier and store fulfillment range corresponding to the demand return range as filtering conditions, trace backward from the supply blockage position to the fulfillment response end position in the demand fulfillment matching record, extract the demand quantity that has not entered the matching position, and aggregate them according to the demand event identifier to form a record of unaccepted demand after the blockage. Based on the record of unaccepted demand after the blockage, locate the supply recovery position in the supply and demand tracking sequence, verify the demand event identifier and the fulfilled quantity that have entered the matching position after the supply recovery position, deduct the re-accepted demand quantity from the unaccepted demand according to the same demand event identifier, and stop when the deduction reaches zero to form a record of remaining demand; based on the record of remaining demand, merge the multiple registration contents of the same demand event identifier, retain the earliest demand position and the last updated remaining unaccepted demand quantity, and remove the intermediate duplicate demand to form a deduplicated demand record; based on the deduplicated demand record, connect the remaining unaccepted demand to the corresponding demand return range according to the relationship of consistent product identifier and intersection of store fulfillment range to generate a record of demand to be returned.

[0042] S2.10. Based on the pending demand record, enter the corresponding demand return range item by item according to the demand event identifier, extract the online demand change and remaining capacity at each time position in chronological order, deduct the occupied amount that has been matched and returned from the online demand change, and deduct the locked return amount from the remaining capacity to form a location capacity reserve record.

[0043] Based on the location-based capacity record, when both the processed online demand change and the remaining capacity are greater than zero, the smaller of the online demand change and the remaining capacity is taken as the allocable upper limit, and the corresponding time location is registered as the target time location, forming a target time location record. Based on the target time location record, the time interval between each target time location and the supply interruption location is recorded, arranged in ascending order of time interval, and if the time intervals are the same, arranged in descending order of allocable upper limit. Following the arrangement order, the smaller of the unallocated unaccepted demand and the allocable upper limit is taken as the demand return amount, and the unaccepted demand and the allocable upper limit are deducted simultaneously. The target time location, demand return amount, and unallocated unaccepted demand are connected in sequence to generate a demand return allocation sequence.

[0044] S2.11. Based on the demand return allocation sequence, check the demand return amount at each target time position one by one. Use the smaller of the corresponding online demand change and the remaining capacity as the allowable return amount. When the demand return amount exceeds the allowable return amount, reduce the demand return amount to the allowable return amount and register the reduced part as the remaining demand to form a return amount adjustment record.

[0045] Based on the return volume adjustment record, the allowable return volume is subtracted from the adjusted demand return volume to obtain the capacity reserve. Target time positions are arranged from largest to smallest capacity reserve. If the capacity reserves are the same, they are arranged from smallest to largest time interval from the supply interruption position. The smaller amount of remaining demand and capacity reserve is then added to the target time position, and the demand return volume and capacity reserve are updated synchronously to form a remaining demand addition record. Based on the remaining demand addition record, the demand return volume is connected to the corresponding online demand change segment according to the demand event identifier, and then the online demand change segment is connected to the supply interruption position to form a potential demand connection record. The actual performance acceptance relationship is extracted from the demand fulfillment matching record. The potential demand connection record and the actual performance acceptance relationship are merged according to the demand event identifier and time position, and duplicate actual performance acceptance relationships are eliminated to generate a cross-domain potential demand acceptance record.

[0046] It should also be noted that the cross-domain potential demand acceptance record is a demand acceptance record formed by reverting unaccepted demand after supply disruption to the time period before the disruption, and linking it with the online demand change period, the supply disruption location, and the actual fulfillment relationship. It is used to distinguish between actual fulfillment demand and demand loss caused by insufficient supply, restore the potential demand acceptance capacity of different stores within the fulfillment scope, provide a basis for subsequent same-location inventory embedding, regional replacement volume verification, and candidate product selection, and reduce the interference of supply disruption and fulfillment relocation on product selection judgment.

[0047] S3. Based on the cross-domain potential demand acceptance record, the potential demand is embedded into the same inventory position of the existing product combination and the candidate product combination. Taking the same inventory position as a common starting point, the demand deduction and inventory replenishment are alternately promoted along the order of warehouse and store replenishment time. The fulfillment ownership of adjacent store orders is traced along the overlapping boundary of delivery coverage and the fulfillment transfer increment is stripped to generate candidate product area replacement record.

[0048] S3.1 Extract potential demand return information item by item from the cross-domain potential demand acceptance records, connect the demand return volume with the accepting store according to the demand event identifier, and arrange the connected content in chronological order to form a potential demand return sequence; based on the potential demand return sequence, extract the product category, delivery coverage, and historical fulfillment relationship from the existing product portfolio, and filter candidate products according to the conditions of consistent product category, delivery coverage including accepting store, and existence of supply replenishment relationship, and gather the products that meet the filtering conditions to form a candidate product portfolio; based on the candidate product portfolio, associate the potential demand return volume in the cross-domain potential demand acceptance records, register the potential demand return volume corresponding to the product as the candidate product execution demand, and establish a correspondence with the candidate product identifier. This process involves: creating candidate product demand records; searching for products with the same category in both existing and candidate product combinations; mapping the product delivery coverage boundary to the receiving store location; retaining products whose delivery coverage includes the receiving store, thus creating product association records; converting potential demand return quantities into minimum sales unit quantities based on product packaging specifications, and then converting the minimum sales unit quantities into product packaging quantities, thus creating product quantity conversion records; and connecting supply warehouses and receiving stores along the warehouse-store replenishment relationship corresponding to the matching products, connecting the converted product quantities to the corresponding replenishment locations, and arranging the replenishment locations according to the replenishment arrival time, thus generating a warehouse-store replenishment chain.

[0049] S3.2 Extract the inventory change positions of existing product combinations and candidate product combinations along the warehouse replenishment chain, arrange them from front to back according to the time of inventory change, and arrange the demand deduction positions first and then the replenishment arrival positions within the same time period to form an inventory position arrangement record.

[0050] Based on the inventory location arrangement record, starting from the initial inventory quantity, the preceding replenishment quantity is accumulated position by position, and the preceding demand-occupied quantity is deducted. The quantity reserved before each inventory location occurs is recorded as the cumulative net availability. Then, the locked fulfillment quantity is deducted from the cumulative net availability, and negative values ​​are recorded as zero, forming the remaining capacity. The cumulative net availability, remaining capacity, replenishment arrival time, and delivery coverage boundary are connected to the corresponding inventory location to form an inventory location status record. Based on the inventory location status record, existing inventory locations and candidate inventory locations within the fulfillment response range are filtered, and the delivery coverage boundary is mapped to the same delivery area. When there are common coverage locations, existing inventory locations and candidate inventory locations are connected, and the connection relationship is arranged according to the replenishment arrival time to generate inventory location candidate pairing records.

[0051] S3.3 Extract the demand event identifier, receiving store, and inventory location association relationship corresponding to the inventory location candidate pairing record from the inventory location candidate pairing record. Bind the receiving store corresponding to the same demand event identifier to the first inventory location with a replenishment relationship, and connect the demand event identifier, receiving store, and the first inventory location to form a potential demand inventory anchor point. Based on the potential demand inventory anchor points, aggregate and connect the existing inventory locations and candidate inventory locations of the same potential demand inventory anchor point to form an inventory anchor point pairing group. For each location pair in the inventory anchor point pairing group, expand the cumulative net available quantity on both sides into continuous quantity locations according to the minimum sales unit, align them position by position starting from the first quantity location, and gather the quantity locations that exceed the common alignment range into non-overlapping quantity locations. Record the quantity of the non-overlapping quantity locations as the cumulative net available quantity difference. When the cumulative net available quantity on both sides is zero, record the cumulative net available quantity difference as zero. When only one side has a cumulative net available quantity difference, record the difference as zero. When the quantity is zero, all quantity locations corresponding to the cumulative net available quantity on the other side are registered as non-overlapping quantity locations; the delivery coverage boundaries on both sides are mapped to the same delivery area and the common coverage part is aggregated to obtain the delivery coverage overlap range; the replenishment arrival time is obtained by subtracting the earlier replenishment arrival time from the later replenishment arrival time; the remaining acceptance quantities on both sides are added to form a location pair comparison record; according to the location pair comparison record, in each inventory anchor pairing group, they are first arranged in ascending order of cumulative net available quantity difference; when the cumulative net available quantity difference is the same, they are arranged in descending order of delivery coverage overlap range, ascending order of replenishment arrival time offset, and descending order of the sum of remaining acceptance quantities on both sides; when all arrangement conditions are the same, they are arranged in the original order in the inventory location candidate pairing record; the position pair at the top of each potential demand inventory anchor is retained, and the existing inventory positions and candidate inventory positions in the retained position pairs are connected to the potential demand inventory anchor to generate corresponding inventory positions.

[0052] It should be noted that, based on the corresponding inventory positions, paired inventory positions are located in the inventory event sequences of the existing product mix and the candidate product mix, respectively. The common minimum sales equivalent corresponding to the same potential demand event is connected to the two inventory positions, and the sorting positions of the two inventory positions are aligned. The two inventory positions are registered as a common starting point. From the common starting point, the warehouse-store replenishment relationship of the existing product mix and the candidate product mix is ​​tracked in chronological order. The replenishment arrival positions within the fulfillment response range are extracted, and all replenishment arrival positions are arranged according to the replenishment arrival time. Positions with the same replenishment arrival time are connected as the same replenishment boundary. Replenishment arrival positions appearing on one side are set to zero values ​​on the other side. Then, the two inventory event sequences are connected in chronological order of each replenishment boundary. Thus, with the corresponding inventory positions as a common starting point, the process proceeds in chronological order of warehouse-store replenishment time.

[0053] S3.4 Based on the location of the corresponding inventory, the potential demand return quantity is associated with the potential demand event identifier. According to the conversion relationship between the minimum sales unit of the existing product mix and the candidate product mix, the potential demand return quantity is converted into the minimum measurement quantity that can be expressed by both sides, forming a common minimum sales equivalent record.

[0054] Based on the common minimum sales equivalent record, the inventory event sequences corresponding to the existing product mix and the candidate product mix are located respectively. The two corresponding inventory positions and the previous inventory event are set as embedding intervals to form corresponding embedding position records. According to the corresponding embedding position records, demand deduction positions are inserted in the two embedding intervals respectively. The two demand deduction positions are assigned the same sort number according to the order of potential demand events, and the subsequent sort numbers are sequentially extended to keep the time position of the original inventory events unchanged. The common minimum sales equivalent is used as the embedding quantity and connected to the two demand deduction positions respectively. At the same time, the cumulative net available quantity before embedding of the two corresponding inventory positions is recorded. The potential demand event, demand deduction position, cumulative net available quantity before embedding and embedding quantity are connected to generate potential demand corresponding embedding records.

[0055] S3.5. Based on the potential demand co-embedded records, locate the co-located inventory positions corresponding to the existing product combination and the candidate product combination, register the two co-located inventory positions as a common starting point, and align the two inventory event sequences to the common starting point.

[0056] Starting from a common starting point, the system traces chronologically to the end time of the fulfillment response, extracting the demand deduction positions and replenishment arrival positions for both existing and candidate product combinations to form a deduction / replenishment position record. Based on this record, the replenishment arrival times corresponding to the replenishment arrival positions on both sides are aggregated and arranged chronologically. Identical replenishment arrival times are merged into a single replenishment boundary, and duplicate times are removed. Simultaneously, the common starting point and the end time of the fulfillment response are registered as the first and last boundaries, forming a unified replenishment boundary record. Based on this unified replenishment boundary record, adjacent replenishment boundaries are divided into deduction intervals. Demand deduction positions are connected to their corresponding deduction intervals, and if a demand deduction position and a replenishment boundary occur at the same time, it is connected to the preceding deduction interval. The actual replenishment quantity for both existing and candidate product combinations is registered at each replenishment boundary, with zero values ​​recorded when no replenishment occurs. The deduction intervals and replenishment boundaries are connected chronologically to generate a deduction / replenishment advancement boundary sequence.

[0057] S3.6. Based on the deduction and compensation advancement boundary sequence, advance from the first deduction interval in the existing commodity portfolio and candidate commodity portfolio respectively. Connect the cumulative net available quantity before embedding the corresponding inventory position to the starting position of the first deduction interval, and connect the unaccepted demand and remaining net available quantity left over from the previous deduction interval to the starting position of subsequent deduction intervals to form the interval starting record.

[0058] Based on the interval start record, the demand quantities at each demand deduction position within the current deduction interval are collected and merged with the unaccepted demand left over from the previous deduction interval. The minimum value between the current net available quantity and the merged demand quantity is recorded as the demand deduction quantity. The demand deduction quantity is deducted from the merged demand quantity, and the undeducted portion is carried over to the next deduction interval. At the same time, the demand deduction quantity is deducted from the current net available quantity, forming an interval deduction record. Based on the interval deduction record, the confirmed actual replenishment quantity is added at the end position of the interval. The net available quantity after the addition is connected to the next deduction interval. When the demand deduction position and the replenishment boundary are at the same time, the demand formed before the replenishment confirmation is processed first, and then the actual replenishment quantity is added. The demand deduction quantity, unaccepted demand, and actual replenishment quantity of each round are connected in the time sequence of the deduction interval to generate a round-by-round deduction and replenishment record.

[0059] S3.7. Based on the round-by-round replenishment records, align the same replenishment boundaries of the existing product combination and the candidate product combination according to the time sequence of the replenishment boundaries, and extract the remaining inventory, unaccepted demand and associated root order identifiers corresponding to the same replenishment boundaries respectively.

[0060] Calculate the remaining inventory difference and unmet demand difference for each identical replenishment boundary. Subtract the remaining inventory of the existing product combination from the remaining inventory of the candidate product combination, and record the resulting quantity as the remaining inventory difference. Subtract the unmet demand of the existing product combination from the unmet demand of the candidate product combination, and record the resulting quantity as the unmet demand difference. Associate the changes in remaining inventory and unmet demand of the corresponding orders according to the root order identifier to form a replenishment quantity comparison record. Based on the replenishment quantity comparison record, compare the replenishment boundary time with the demand fulfillment deadline. Check the unmet demand status of each order according to the root order identifier. When both sides have zero unmet demand, mark the corresponding root order identifier as fulfillment closed. When at least one side has non-zero unmet demand and has not exceeded the demand fulfillment deadline, mark the corresponding root order identifier as pending within the deadline. When the demand fulfillment deadline is exceeded, the corresponding root order is marked as fulfillment timeout, forming a demand fulfillment deadline status. Based on the demand fulfillment deadline status, the deduction offset direction is determined according to the difference between the unaccepted demand and the remaining inventory corresponding to the root order. When the difference between the unaccepted demand and the remaining inventory is less than zero, the acceptance is marked to move forward; when the difference between the unaccepted demand and the remaining inventory is greater than zero, the acceptance is marked to move backward; when the difference between the unaccepted demand and the remaining inventory is equal to zero, the inventory is marked as surplus; when the difference between the unaccepted demand and the remaining inventory is equal to zero, the inventory is marked as insufficient; when both differences are zero, the deduction is marked as synchronized. The root order, the replenishment boundary position, and the corresponding deduction offset status are associated to form an order-related deduction offset record. The deduction offset statuses in the order-related deduction offset record are connected according to the time sequence of the replenishment boundary, preserving the continuous change relationship corresponding to the root order, to generate a deduction offset sequence.

[0061] The expression for calculating the remaining inventory difference and the unmet demand difference for each identical replenishment boundary is as follows: ; in, For the first The difference in the amount of compensation at the same compensation boundary; For the first The remaining inventory difference at the same replenishment boundary; For the first The difference in unaccepted demand at the same fill-in boundary; For candidate product combinations in the 1st Remaining inventory at the same replenishment boundary; For the existing product mix in the first Remaining inventory at the same replenishment boundary; For candidate product combinations in the 1st Unaccepted requests with the same fill-in boundary; For the existing product mix in the first Unaccepted requests at the same boundary.

[0062] It should also be noted that the difference in unmet demand at the same replenishment boundary is obtained by calculating the quantitative difference between the unmet demand of the candidate product combination at the same replenishment boundary and the unmet demand of the existing product combination at the same replenishment boundary, thus obtaining the change in demand met by the candidate product combination relative to the existing product combination.

[0063] S3.8. Based on the offset sequence, each offset state is connected to the same fill boundary in the offset advancement boundary sequence according to the fill boundary time. Then, the sub-round offset records corresponding to the existing product combination and the candidate product combination are connected to both sides of the fill boundary to form the same boundary advancement record.

[0064] Based on the boundary advancement record, the process proceeds round by round from the common starting point along the deduction and replenishment advancement boundary sequence. After each replenishment boundary connection is completed, the unmet demand of the existing product combination and the candidate product combination is checked. When the unmet demand of the existing product combination reaches zero first, zero values ​​are continuously set for the existing product combination at subsequent replenishment boundaries, and the deduction and replenishment records of the candidate product combination where the unmet demand has not yet reached zero continue to be connected. When the unmet demand of the candidate product combination reaches zero first, zero values ​​are continuously set for the candidate product combination in the same way, and the deduction and replenishment records of the existing product combination where the unmet demand has not yet reached zero continue to be connected. When all unmet demands reach zero, the connection stops at the corresponding replenishment boundary. If the fulfillment response end time arrives first, the remaining unmet demands are registered at the fulfillment response end position and the connection stops. All connection contents are collected to generate an inventory alternation advancement record.

[0065] S3.9. Arrange the corresponding deduction / supplementation offset states of orders in the order association deduction / supplementation progress record according to the time sequence of the supplementation boundary. Select the first deduction / supplementation offset state that is not synchronized with the deduction / supplementation, and register the corresponding order identifier as an offset order to form an offset order record. Based on the offset order record, extract the order receiving address grid corresponding to the order identifier, map the order receiving address grid and the delivery coverage overlap boundary to the same grid area, find the delivery coverage overlap boundary segment that intersects with the order receiving address grid, and if there are multiple intersecting boundary segments, arrange them in descending order of overlap length. If the overlap length is the same, arrange them in the original order and retain the boundary segment at the top of the list. If there are no intersecting boundary segments, project the center position of the order receiving address grid to the nearest delivery coverage overlap boundary segment. The first and second boundary segments are retained in the original order, and the intersection and projection positions are registered as boundary mapping positions. Using the boundary mapping positions as the starting point for tracing, open boundaries are advanced to the corresponding endpoints along two extension directions, and closed boundaries are advanced to the reconnected boundary mapping positions along clockwise and counterclockwise directions respectively. Fork boundaries are advanced according to the turning angle from small to large for branches that have not yet been traversed, and stop when reaching the endpoint and connecting to the tracked boundary segment. Adjacent store orders that fall into the overlapping area of ​​the order delivery address grid are extracted along the traversed boundary segments to form boundary order tracking records. According to the boundary order tracking records, the order fulfillment time is connected to the supplementary boundary with the closest time and no later than the order fulfillment time. If the order fulfillment time is the same as the supplementary boundary, the previous supplementary boundary is connected to generate overlapping boundary order records.

[0066] S3.10. Using the root order identifier in the overlapping boundary order record as the starting point for tracing, the fulfillment content of the same order is aggregated according to the root order identifier. The final fulfillment store corresponding to the root order identifier is located, and from the final fulfillment store, the store access location and the previous order release location are searched sequentially from the last to the first from the order transfer order until the earliest order release location is found. The store corresponding to the earliest order release location is registered as the original receiving store, forming an order transfer tracking record. Based on the order transfer tracking record, the root order identifier, delivery address grid, and product category of adjacent order release locations and store access locations are checked segment by segment. When the three items are consistent, the order is connected. A single release location and a store access location are used to form a transfer location connection record. Based on the transfer location connection record, all transfer locations of the same order are connected in the order of order transfer. The earliest order release location of the original receiving store and the last store access location of the final fulfillment store are retained. Intermediate transfer locations within the continuous connection range are deleted, and the final actual fulfillment quantity is used as the fulfillment quantity to form a migration compression record. Based on the migration compression record, the transfer direction is registered according to the extension direction from the original receiving store to the final fulfillment store according to the delivery coverage overlap boundary. The original receiving store, the final fulfillment store, and the fulfillment quantity are also registered to generate an order fulfillment migration record.

[0067] S3.11. Based on the order fulfillment migration record, collect the fulfillment content of candidate product combinations and existing product combinations according to overlapping boundary segments and root order identifiers, and associate the candidate fulfillment quantity, original receiving store, and final fulfillment store in the candidate product order fulfillment record. Expand the candidate fulfillment quantity and existing product combination fulfillment quantity of the same root order into continuous fulfillment quantity positions according to the common minimum sales equivalent, and overlap them one by one starting from the first fulfillment quantity position. Retain the fulfillment quantity positions of the candidate product combination that are not overlapped by the existing product combination to form a candidate fulfillment increment record; based on the candidate fulfillment increment record, combine the release quantity of the original receiving store and the access quantity of the final fulfillment store according to... The common minimum sales equivalent is expanded into corresponding quantity positions. The candidate fulfillment increment positions, the original receiving store release positions, and the final fulfillment store access positions are aligned one by one. The portion of the three types of positions that can be aligned together is registered as the fulfillment transfer increment and associated with the corresponding root order identifier to form a transfer balancing record. According to the transfer balancing record, the corresponding fulfillment transfer increment positions are marked in the candidate fulfillment increment positions. The unmarked fulfillment quantity positions are aggregated into the remaining net replacement quantity, and the corresponding root order identifier, original receiving store, final fulfillment store, candidate product identifier, and overlapping boundary segment information are retained. The overlapping boundary segments are connected in the extension order to generate a regional net replacement sequence.

[0068] S3.12. Collect the registration information of the same order across multiple overlapping boundary segments from the regional net replacement sequence according to the root order identifier, and arrange the corresponding overlapping boundary segments along the transfer direction to form an order boundary tracking record. Based on the order boundary tracking record, trace the delivery coverage location segment by segment along the arranged overlapping boundary segments, starting from the original receiving store. Register the overlapping boundary segment through which the order passes when it moves from the original receiving store's delivery coverage area to outside the delivery coverage area as the first overlapping boundary segment. Retain the first overlapping boundary segment and the final fulfillment store, and remove duplicate registration information of the same order in subsequent overlapping boundary segments to form... Order deduplication and replacement records: Based on the order deduplication and replacement records, overlapping boundary segments are collected according to candidate products and delivery directions. When the end point of one overlapping boundary segment connects to the start point of the next overlapping boundary segment, they are connected to form a continuous overlapping boundary segment. The connection stops when the candidate product changes, the delivery direction changes, or the boundary endpoints are broken, forming a continuous boundary replacement record. Based on the continuous boundary replacement record, continuous overlapping boundary segments with the same candidate products and the same delivery direction are merged. The corresponding net replacement volume is assigned to the merged overlapping boundary segment item by item, and the final fulfillment store is registered as the store fulfillment affiliation, generating a candidate product area replacement record.

[0069] It should also be noted that the candidate product regional replacement record is a record formed by excluding the cross-boundary duplicate registration and fulfillment relocation effects of the same order, and is based on the regional replacement relationship of candidate products, transfer direction and continuous overlapping boundaries. It is used to indicate the net replacement demand that candidate products can undertake within the fulfillment scope of different stores and the store fulfillment attribution, to provide a basis for the allocation of execution quantity and partial updates of product selection under warehouse and store inventory constraints, to improve the accuracy of regional product selection judgment, and to avoid misjudging order transfer as new demand brought about by candidate products.

[0070] S4. Based on the candidate product area replacement record, reverse the net replacement volume between the final fulfillment store and the supply warehouse and forward the available capacity of the warehouse store. Align the filling boundary at the meeting position, match the candidate product execution quantity bit by bit, obtain the product selection allocation record, locate the affected product selection content in the product selection allocation record and implement partial updates, and generate the O2O retail intelligent product selection list.

[0071] S4.1. Based on the candidate product regional replacement records, aggregate the regional net replacement volume according to the candidate product identifier and the final fulfillment store. Merge and register the net replacement volumes of the same candidate product identifier and the same final fulfillment store, and associate them with the corresponding supply warehouse, replenishment path, and overlapping boundary segment. Based on the aggregated net replacement volume, search for the replenishment path from the supply warehouse to the final fulfillment store along the warehouse-store replenishment relationship, extract the supply warehouses and replenishment paths that can be replenished before the fulfillment response end time, and remove the replenishment relationships that cannot meet the fulfillment response end time, forming a valid warehouse-store replenishment relationship record. Based on the valid warehouse-store replenishment relationship record, split the net replacement volume into consecutive numbers according to the minimum sales unit corresponding to the candidate product. The quantity locations are numbered sequentially from the beginning to form a net replacement quantity number sequence. Taking the final fulfillment store receiving boundary as the starting position of the net replacement quantity number sequence, the replenishment locations are tracked step by step along the logistics transmission direction of the replenishment path. The quantity locations are sequentially passed through the replenishment path arrival boundary, the replenishment path departure boundary, and the supply warehouse departure boundary, and the corresponding quantity location number is registered at each boundary location. According to the transmission order of each boundary location, the quantity locations corresponding to the final fulfillment store receiving boundary, the replenishment path arrival boundary, the replenishment path departure boundary, and the supply warehouse departure boundary are connected. The quantity location number, candidate product identifier, final fulfillment store, supply warehouse, and replenishment path are associated to generate a net replacement quantity reverse sequence.

[0072] S4.2. Based on the net replacement volume reverse sequence, extract the corresponding candidate product identifier, supply warehouse, replenishment path, and final fulfillment store's transfer location from the net replacement volume reverse sequence, and obtain the supply warehouse inventory balance, warehouse dispatch balance, replenishment path fulfillment balance, and final fulfillment store receiving balance respectively; based on the obtained balances, divide the supply warehouse inventory balance, warehouse dispatch balance, replenishment path fulfillment balance, and final fulfillment store receiving balance into continuous capacity positions according to the minimum sales unit corresponding to the candidate product, and number them sequentially according to the number of capacity positions to form the supply warehouse boundary capacity position, warehouse dispatch boundary capacity position, replenishment path boundary capacity position, and final fulfillment store receiving boundary capacity position; taking the supply warehouse dispatch boundary as the capacity transfer starting point, according to the replenishment path transfer direction... The system sequentially retrieves the boundary capacity locations of the warehouse, the replenishment path, the replenishment path, and the final fulfillment store. Boundary capacity locations with the same number that are not occupied by other net replacement quantities are connected, and these connected boundary capacity locations are registered as available capacity locations. Based on the transmission relationship of available capacity locations, capacity location transmission stops when a corresponding boundary capacity location with the same number does not exist for subsequent replenishment boundaries, while retaining the connection relationship of the boundary capacity locations with the already completed numbers. The retained boundary capacity locations are arranged according to the transmission order from the supply warehouse's boundary to the final fulfillment store's receiving boundary. The boundary capacity location number, candidate product identifier, supply warehouse, replenishment path, and final fulfillment store are associated to generate a forward sequence of warehouse-store available capacity.

[0073] S4.3. Based on the reverse sequence of net replacement volume and the forward sequence of available warehouse / store capacity, extract the common minimum sales equivalent (MSE) position and boundary capacity position corresponding to the same candidate product identifier, replenishment path, and arrival time. Arrange the MSE and boundary capacity positions according to the transmission direction from the supply warehouse to the final fulfillment store. Based on the arranged MSE and boundary capacity positions, compare the replenishment boundaries passed by both sides one by one. Register the position where the MSE and boundary capacity positions first meet at the same replenishment boundary as the meeting position, and record the MSE position number and boundary capacity position number corresponding to the meeting position. Based on the number records on both sides of the meeting position, align the MSE positions one by one in ascending order of number. When the common minimum sales equivalent position number matches the boundary capacity position number and the boundary capacity position is not occupied, the two positions are connected and registered as a matching position. Based on the matching position, the net replacement quantity position that has been matched is stopped from continuing to be passed along the remaining replenishment path, and the net replacement quantity position that has not been matched continues to be passed along the supply warehouse direction to the replenishment boundary on the supply warehouse side. At the same time, the unoccupied boundary capacity position continues to be passed along the final fulfillment store direction to the replenishment boundary on the final fulfillment store side. The meeting position, matching position, unmatched net replacement quantity position and unoccupied boundary capacity position are connected in the order of replenishment boundary time. The candidate product identifier, replenishment path, arrival time and position number are associated to generate the replenishment boundary matching sequence.

[0074] S4.4. Based on the fill-in boundary mating sequence, arrange the consecutive mating positions corresponding to the same candidate product in chronological order of the fill-in boundary, and extract the candidate product identifier, fill-in boundary position, common minimum sales equivalent position, and boundary capacity position corresponding to each mating position; based on the arranged mating positions, check one by one whether each mating position is simultaneously associated with the net replacement quantity inverse sequence, the warehouse / store available capacity forward sequence, and the same fill-in boundary mark in the fill-in boundary mating sequence. When a mating position has all three types of marks, register the mating position as a closed mating position; when any mark is missing, register the mating position as an unclosed mating position and remove the boundary capacity position corresponding to the unclosed mating position. The occupancy relationship is set, and the released boundary capacity position is reconnected to the warehouse-store available capacity forward sequence. Based on the closed matching position, the continuous matching positions corresponding to the same candidate product are collected according to the candidate product identifier. The common minimum sales equivalent quantity in the closed matching positions is summarized to form the candidate product execution quantity. Based on the candidate product execution quantity, the corresponding inventory occupancy relationship is checked along the replenishment path from the supply warehouse to the final fulfillment store. The candidate product execution quantity is deducted from the supply warehouse inventory balance, the replenishment path fulfillment balance, and the final fulfillment store receiving balance, respectively. The supply warehouse, the final fulfillment store, the replenishment path, and the corresponding candidate product identifier are registered simultaneously. The continuous closed matching positions, the candidate product execution quantity, and the warehouse-store replenishment relationship are connected to generate the product selection allocation record.

[0075] S4.5. Based on the product selection allocation record, connect the execution quantity, supply warehouse, final fulfillment store and replenishment path of the candidate product according to the candidate product identifier, and register the execution quantity of the candidate product as the change difference, and generate a product selection change fingerprint record.

[0076] It should be noted that, based on the product selection allocation record, the corresponding allocation content is collected according to the candidate product identifier. The execution quantity, supply warehouse, final fulfillment store and replenishment path of the candidate product are respectively connected to the candidate product identifier. When the same candidate product identifier corresponds to multiple warehouse-store replenishment relationships, an independent connection branch is set according to the allocation sequence number in the product selection allocation record to form a candidate product execution association record.

[0077] Based on the candidate product execution association record, the execution quantity of candidate products in each connection branch is directly registered as the change difference, and the change difference is connected to the corresponding supply warehouse, replenishment path, and final fulfillment store to form a product selection change content record. Based on the product selection change content record, the character format of the candidate product identifier and the warehouse-store replenishment relationship is unified. The connection content is arranged in a fixed order of candidate product identifier, supply warehouse, replenishment path, final fulfillment store, and change difference. Separator marks are set between adjacent connection content. The arranged connection content is combined into a product selection change fingerprint, and the product selection change fingerprint is associated with the corresponding product allocation position to generate a product selection change fingerprint record.

[0078] S4.6 Extract candidate product identifiers, final fulfillment stores, and change differences from the product selection change fingerprint records. Perform joint matching between candidate product identifiers and final fulfillment stores in the existing O2O retail product selection content. Register product selections where both information matches as product matching positions, forming a product matching position record. Based on the product matching position record, register candidate products that do not form a product matching position as new candidate products. Search for the same product category in the existing O2O retail product selection content corresponding to the final fulfillment store, locate the end position of the same product category along the product selection order, and record the end position. The end position is then registered as the insertion position, forming a new product selection insertion record. Based on the product selection matching position record and the new product selection insertion record, the product selection matching position and the insertion position are used as the starting point for tracking. The related content referencing the same candidate product identifier and the final fulfillment store is tracked layer by layer along the warehouse-store replenishment relationship in the product selection change fingerprint record. The related content that directly carries the change difference and depends on the change difference is connected to the tracking starting point until there are no new related connection positions, forming a product selection association tracking record. The product selection matching position, insertion position and product selection association tracking record are collected to generate a product selection impact range record.

[0079] S4.7. Based on the product selection impact range record, copy the existing O2O retail product selection content covered by the product selection impact range record to a temporary copy item by item, retaining the original arrangement position and version identifier; process the temporary copy item by item according to the product selection change fingerprint, register the execution quantity of candidate products at the product selection matching position, establish the product selection content corresponding to the newly added candidate products at the insertion position, and update the associated content connected by the product selection change fingerprint to the corresponding candidate products and warehouse store supplementation relationship, forming partial change content and writing it to the temporary copy; based on the written temporary copy, check the partial update closure status along each product selection change fingerprint, mark it as closed when the execution quantity of candidate products has entered the corresponding product selection position and all associated content is connected to the updated product selection content, mark it as unclosed when there are missing connection positions, forming a partial update verification record; based on the partial update verification record, locate the missing connection positions along the association relationship of the unclosed content, re-incorporate the missing connection positions into the product selection impact range record, and write the expanded product selection content into the temporary copy and check again until the change quantity and associated content are all closed, connect the final partial change content with the closed status, and generate a partial update closure record.

[0080] S4.8. Based on the partial update closure record, extract the product selection location and version identifier corresponding to each partial change. Locate the same product selection location in the existing O2O retail product selection content and compare the two version identifiers item by item to form a version verification record. According to the version verification record, replace the corresponding product selection content with the partial change content for product selection locations with the same version to form a version consistency update record. For product selection locations with version conflicts, use the latest version in the existing O2O retail product selection content as the verification basis, rewrite the corresponding partial change content according to the product change fingerprint, and verify the version change. After the update is closed, the version identifier is compared again. If the versions are consistent, the corresponding product selection content is replaced, forming a version conflict update record. The version consistent update record and the version conflict update record are merged, and the product selection content that is not covered by the product selection impact record is retained from the existing O2O retail product selection content, forming a product selection list update record. Based on the product selection list update record, the updated content is first arranged according to the final fulfillment store, and then the updated content of the same category is collected according to the product selection category. Within each product selection category, the updated content is arranged in chronological order of the effective time period, generating an O2O retail smart product selection list.

[0081] It should also be noted that the O2O retail intelligent product selection list is the final product selection content formed after restoring potential cross-regional demand, removing the incremental fulfillment from relocation, and completing the allocation of warehouse and store execution constraints. It records the execution arrangements and effective scope of candidate products in the corresponding fulfillment stores. It is used to transform potential demand identification, regional replacement judgment, and inventory fulfillment restrictions into directly executable product selection criteria, supporting store product selection adjustments, warehouse and store replenishment arrangements, and partial content updates. It reduces product selection deviations caused by supply disruptions and order reassignment, and improves the matching degree between product mix and actual regional demand.

[0082] In summary, this invention achieves big data processing of potential demand, inventory supply, and fulfillment scope by embedding cross-domain potential demand into the corresponding inventory positions of existing and candidate product combinations, alternately advancing demand deduction and inventory replenishment along the order of warehouse and store replenishment, and tracing the fulfillment attribution of adjacent store orders along the overlapping boundaries of delivery coverage, and stripping the incremental fulfillment transfer. This results in the formation of regional replacement records for candidate products, reducing the interference of stockouts and cross-store transfers on the judgment of product value, and improving the accuracy of regional product selection quantity allocation and the stability of local updates.

[0083] 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. An O2O retail intelligent product selection method, characterized in that, include: Identity-linked and time-series aligned O2O retail product selection data to generate an O2O product selection time-series benchmark chain; Based on the O2O product selection timeline benchmark chain, online demand changes and offline fulfillment changes are matched to obtain demand fulfillment matching records. From the demand fulfillment matching records, the time lag coupling change point tracking of supply gaps and fulfillment stagnation is performed and the supply blockage position is located. The unaccepted demand after the supply blockage is returned to the position before the supply blockage, generating cross-domain potential demand acceptance records. Based on the cross-domain potential demand acceptance record, the potential demand is embedded in the same inventory position of the existing product combination and the candidate product combination. Taking the same inventory position as a common starting point, the demand deduction and inventory replenishment are alternately promoted along the order of warehouse and store replenishment time. The fulfillment attribution of adjacent store orders is traced along the delivery coverage overlap boundary and the fulfillment transfer increment is stripped to generate candidate product area replacement record. Based on the candidate product area replacement records, the net replacement volume is calculated backward between the final fulfillment store and the supply warehouse, and the available capacity of the warehouse store is calculated forward. The boundary is aligned at the meeting point, and the candidate product execution quantity is matched bit by bit. The product selection allocation record is obtained, the affected product content is located in the product selection allocation record and a partial update is implemented to generate an O2O retail intelligent product selection list.

2. The O2O retail intelligent product selection method as described in claim 1, characterized in that, The specific steps regarding changes in online demand and offline fulfillment for matching are as follows: Based on the O2O product selection timeline benchmark chain, online demand content and offline fulfillment content are arranged according to product identification and store fulfillment scope. The quantity changes of adjacent time positions are compared, and the positions with the same direction of change and continuous connection are connected as online demand change segments and offline fulfillment change segments. Using the end position of each online demand change segment as the tracking starting point, find offline fulfillment change segments with consistent product identification and overlapping store fulfillment ranges within the fulfillment response scope, retain the continuity relationship with consistent change direction, and form demand fulfillment candidate relationships; Based on the candidate relationship of demand fulfillment, the changes in online demand and the changes in offline fulfillment that have not yet been occupied are successively accepted. When a change in online demand corresponds to multiple changes in offline fulfillment, the unpaired portion is allocated in sequence. When multiple changes in online demand correspond to a change in offline fulfillment, the remaining fulfillment quantity is occupied in sequence. The matching position and the unaccepted demand are registered, and the demand fulfillment matching record is obtained.

3. The O2O retail intelligent product selection method as described in claim 2, characterized in that, The specific steps for tracking and locating the supply blockage point by performing time-delay coupling of supply gaps and fulfillment stagnation from the demand fulfillment matching records are as follows: Based on the demand fulfillment matching records, starting from the matching start position of unmet demand, the existing store inventory, in-transit replenishment, unmet demand, and fulfillment volume are connected in chronological order along the fulfillment response range. At each time position, the supply gap where the net available quantity fails to cover the unmet demand and the fulfillment stagnation state where the fulfillment volume has not increased are recorded, generating a supply and demand response comparison sequence. Based on the supply and demand response comparison sequence, the continuous changes of supply gap and performance stagnation before and after adjacent time positions are compared. The first position where the supply gap changes from non-existent to continuous occurrence is marked as the supply gap change point, and the first position where the performance volume changes from continuous increase to continuous non-increase is marked as the performance stagnation change point. The continuous state interval before and after each change point is retained to generate a supply and demand dual change point record. Based on the supply and demand dual change point records, each supply gap change point is used as the coupling starting point. The earliest performance stagnation change point within the performance response range and located after the supply gap change point is selected. The continuous state interval before and after the performance stagnation change point is shifted along the time axis towards the supply gap change point until the two change points are at the same time position. The response time delay corresponding to the shift distance is recorded, and the combination of change points where the supply gap interval and the performance stagnation interval continuously overlap after the shift is retained to generate a time delay coupled change point sequence. Based on the time-delay coupled variable point sequence, the response time delay is extended to the supply recovery position and the fulfillment recovery position. It is verified whether the fulfillment volume recovers and increases according to the same response time delay after the supply gap disappears. Variable point combinations with matching recovery relationships and variable point combinations where the supply gap exists continuously and the fulfillment volume stagnates continuously before the fulfillment response ends are retained. Variable point combinations where the supply gap appears briefly and there is no corresponding fulfillment stagnation are eliminated. The earliest supply gap variable point is registered as the supply interruption position.

4. The O2O retail intelligent product selection method as described in claim 1 or 3, characterized in that, Before returning unmet demand after supply disruption to the location of supply disruption, a record of potential cross-domain demand is generated. The specific steps are as follows: Tracing back from the point of supply disruption, the most recent time point in which net availability can cover demand and demand fulfillment matching is closed is recorded as the demand return range, and the continuous interval between the most recent time point and the time point before the supply disruption point is registered. Using the product identifier and store fulfillment range corresponding to the demand return scope as filtering conditions, the unaccepted demand after the supply blockage location is extracted, and duplicate demand is removed according to the demand event identifier. The remaining unaccepted demand is then associated with the demand return scope to generate a record of demand to be returned. Based on the pending demand records, within the demand return range, target time positions are selected where both the online demand change and the remaining capacity can accommodate the unaccepted demand. The unaccepted demand is then allocated in combination with the time interval from the supply blockage position to generate a demand return allocation sequence. Based on the demand return allocation sequence, the demand return amount at each target time position is adjusted, and the remaining demand generated by the adjustment is sequentially added to the target time position with the larger capacity. Then, the demand return amount is connected with the corresponding online demand change segment and supply blockage position and merged with the actual performance and acceptance relationship to generate a cross-domain potential demand acceptance record.

5. The O2O retail intelligent product selection method as described in claim 4, characterized in that, The specific steps for embedding potential demand into the corresponding inventory positions of existing and candidate product combinations are as follows: Extract potential demand feedback information from the cross-domain potential demand acceptance records, associate the potential demand feedback information with products in the existing product portfolio and candidate product portfolio that have the same product category and whose delivery coverage includes the accepting store, convert the product quantity according to the packaging specifications and minimum sales unit, and generate the warehouse store replenishment chain. Arrange the inventory change positions of existing product combinations and candidate product combinations along the warehouse replenishment chain, and form the cumulative net available quantity and remaining capacity before each inventory position. Record the replenishment arrival time and delivery coverage boundary, and connect the existing inventory positions and candidate inventory positions that intersect within the delivery coverage of the fulfillment response range to generate inventory position candidate pairing records. Based on the candidate pairing records of inventory locations, each pair of locations is arranged from smallest to largest according to the cumulative net availability difference. For pair of locations with the same difference, they are arranged in order of decreasing delivery coverage overlap, increasing replenishment arrival time offset, and decreasing sum of remaining capacity on both sides. The position pair at the top of each potential demand inventory anchor point is retained to generate corresponding inventory locations. Based on the same inventory position, the common minimum sales equivalent corresponding to the same potential demand event is embedded into the inventory event sequence of the existing product portfolio and the candidate product portfolio respectively. A demand deduction position with the same sort number is set before the two same inventory positions. The time position of the original inventory event remains unchanged and is associated with the cumulative net available quantity before embedding and the embedded quantity to generate potential demand same embedding record.

6. The O2O retail intelligent product selection method as described in claim 1, characterized in that, The specific steps for alternating demand deduction and inventory replenishment are as follows: Based on the potential demand embedded records, using the same inventory location as a common starting point, extract the demand deduction location and replenishment arrival location from the common starting point to the end of the fulfillment response time, merge the replenishment arrival time and remove duplicate times, divide the adjacent replenishment arrival time into deduction intervals, register the actual replenishment quantity of existing product combinations and candidate product combinations at each replenishment boundary, and generate a deduction and replenishment advancement boundary sequence. Based on the deduction and replenishment advancement boundary sequence, the unaccepted demand left over from the previous deduction interval is connected at the beginning of each deduction interval of the existing product combination and the candidate product combination. The demand deduction amount is recorded as the smaller value between the current net available quantity and the demand quantity. The undeducted part is carried into the next deduction interval, and the confirmed actual replenishment quantity is added at the end of the interval. When the demand deduction position and the replenishment boundary are at the same time, the demand formed before the replenishment confirmation is deducted first and then the actual replenishment quantity is added to generate a round-by-round deduction and replenishment record. Based on the round-by-round deduction and replenishment records, the remaining inventory and unaccepted demand of the existing product combination and the candidate product combination at the same replenishment boundary are compared round by round. The deduction and replenishment offset status is marked in combination with the demand fulfillment time limit status, and the deduction and replenishment offset status is connected to generate a deduction and replenishment offset sequence. Based on the offset sequence, the sub-round offset records are connected along the offset advancement boundary sequence. When the unaccepted demand of the existing product combination is zeroed first, a zero value is reserved for the existing product combination. When the unaccepted demand of the candidate product combination is zeroed first, a zero value is reserved for the candidate product combination. The sub-round offset records that have not yet been zeroed out continue to be connected until all unaccepted demand is zeroed out. When the fulfillment response end time is earlier than the demand zeroing position, the connection is stopped at the fulfillment response end position, and an inventory alternation advancement record is generated.

7. The O2O retail intelligent product selection method as described in claim 6, characterized in that, The specific steps for tracing the fulfillment attribution of orders from adjacent stores along the overlapping boundaries of delivery coverage and stripping the fulfillment transfer increment to generate candidate product area replacement records are as follows: Extract the earliest order with the offset status from the inventory alternation progress record, map the corresponding order's delivery address grid to the delivery coverage overlap boundary, use the obtained boundary mapping position as the tracing starting point, extract adjacent store orders segment by segment along the two extension directions of the delivery coverage overlap boundary, and associate the fulfillment time with the replenishment boundary to generate overlap boundary order records; Starting from the root order identifier in the overlapping boundary order record, trace back from the final fulfillment store to the original receiving store along the order transfer order. Connect the root order identifier, the delivery address grid and the order release position and store access position that are consistent with the product category. Compress the intermediate transfer positions that the same order passes through. Register the original receiving store, the final fulfillment store, the transfer direction and the fulfillment quantity to generate an order fulfillment migration record. Based on the order fulfillment migration records, within each overlapping boundary segment, the fulfillment volume of the candidate product combination and the existing product combination for the same root order is compared. The increased fulfillment volume of the candidate product combination is balanced with the release volume of the corresponding original receiving store and the final fulfillment store access volume. The balanced part is registered as the fulfillment transfer increment. The fulfillment transfer increment is separated from the increased fulfillment volume, and the remaining net replacement volume is associated with the candidate product and the corresponding overlapping boundary segment to generate the regional net replacement sequence. The duplicate registrations of the same order in multiple overlapping boundary segments in the regional net replacement sequence are removed. The first overlapping boundary segment that the order passes through when leaving the original receiving store's delivery coverage area and the final fulfillment store are retained. Continuous overlapping boundary segments with the same candidate products and the same transfer direction are merged and the net replacement volume and store fulfillment attribution are aggregated to generate candidate product regional replacement records.

8. The O2O retail intelligent product selection method as described in claim 1, characterized in that, The overlapping boundary of the delivery coverage is obtained by mapping the delivery coverage of the receiving store and the adjacent stores to the same geographic coordinate space, and extracting the continuous edge positions of the overlapping areas and connecting them sequentially.

9. The O2O retail intelligent product selection method as described in claim 8, characterized in that, The process involves reverse-engineering the net replacement volume between the final fulfillment store and the supply warehouse, forward-engineering the available capacity of the warehouse, aligning the replenishment boundary at the meeting point, and matching each position with the candidate product execution quantity to obtain the product selection allocation record. The specific steps are as follows: Based on the candidate product area replacement record, the net replacement volume is split according to the candidate product identifier and the final fulfillment store. The supply warehouse and replenishment path that can be reached before the fulfillment response ends are connected along the warehouse-store replenishment relationship. The net replacement volume is expanded into the common minimum sales equivalent position with sequential number. Starting from the final fulfillment store receiving boundary, the common minimum sales equivalent position is transmitted through the replenishment path arrival boundary, the replenishment path sending boundary and the supply warehouse sending boundary in sequence to generate the net replacement volume reverse sequence. Based on the reverse sequence of net replacement volume, the remaining inventory of the supply warehouse, the remaining inventory issued by the warehouse, the remaining inventory of the replenishment path, and the remaining inventory received by the final fulfillment store are each expanded into sequentially numbered boundary capacity positions. Starting from the boundary of the supply warehouse, the boundary capacity positions are passed sequentially to the boundary of the replenishment path, the boundary of the replenishment path, and the boundary of the final fulfillment store. Boundary capacity positions with the same number and not occupied are retained in adjacent replenishment boundaries, and the passing of boundary capacity positions that lack subsequent corresponding positions is stopped, thus generating a forward sequence of warehouse and store available capacity. Based on the reverse sequence of net replacement volume and the forward sequence of available warehouse capacity, the corresponding content is connected according to the candidate product identifier, replenishment path and arrival time. The replenishment boundary that the common minimum sales equivalent position and the boundary capacity position first pass through is registered as the meeting position. The numbering order on both sides of the meeting position is aligned and matched one by one. After the net replacement volume position is matched, the transmission of the same numbered net replacement volume position in the remaining replenishment path is stopped. The unmatched net replacement volume position continues to be transmitted to the replenishment boundary on the supply warehouse side, and the unoccupied boundary capacity position continues to be transmitted to the replenishment boundary on the final fulfillment store side, generating the replenishment boundary matching sequence. Based on the replenishment boundary matching sequence, consecutive matching positions of the same candidate product are connected in the order of replenishment boundary time. It is checked whether each matching position has a net replacement quantity reverse push mark, a warehouse / store available capacity forward push mark and the same replenishment boundary mark. Matching positions with unclosed marks are eliminated and the corresponding boundary capacity positions are released. Matching positions with closed marks are aggregated to form the candidate product execution quantity. The remaining inventory of the supply warehouse, the fulfillment remaining quantity of the replenishment path and the receiving remaining quantity of the final fulfillment store are occupied simultaneously. The supply warehouse, the final fulfillment store and the replenishment path are registered to generate the product selection allocation record.

10. The O2O retail intelligent product selection method as described in claim 9, characterized in that, The process of locating the affected product selections and implementing partial updates to generate an O2O retail smart product selection list involves the following steps: Based on the product selection allocation record, the execution quantity, supply warehouse, final fulfillment store and replenishment path of the candidate product are connected according to the candidate product identifier, and the execution quantity of the candidate product is registered as the change difference, generating a product selection change fingerprint record; Based on the fingerprint record of product selection changes, the system matches candidate products with the product selection content corresponding to the final fulfillment store in the existing O2O retail product selection content, positions the newly added candidate products in the same product category, tracks the related content affected by the change difference, and generates a record of the scope of product selection impact. Based on the product selection impact range record, form partial change content according to the product selection change fingerprint and write it to a temporary copy. Check the partial update closure status and re-incorporate the unclosed content into the product selection impact range record until the number of changes and related content are all closed, and generate a partial update closure record. Based on the partial update closure record, the version identifiers of the partial changes and the existing O2O retail product selection content are compared. Product selection content with the same version is replaced and conflicting content is rechecked. Unaffected product selection content is retained. The updated content is arranged according to the final fulfillment store, product category and effective period to generate an O2O retail smart product selection list.

Images