An intelligent inventory scheduling management system and method for a business supply chain
By constructing a spatiotemporal consumption resonance index and reconstructing the network topology, electing a master node, intercepting replenishment requests, and virtually mapping available commitments, the system-level stockout problem caused by the correlation of inventory consumption rates in multi-level supply chain networks was solved, achieving stable operation of the supply chain network and business continuity.
Patent Information
- Application Number
- CN202611329401.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-31
- Publication Date
- 2026-11-10
AI Technical Summary
When faced with regional demand fluctuations, multi-level commercial supply chain networks are prone to system-level stockouts due to the strong correlation between inventory consumption rates among front-end warehouse nodes at the same level. Traditional scheduling schemes based on static inventory thresholds cannot identify and intervene in the consumption resonance state between nodes, resulting in the interruption of front-end business order fulfillment.
By calculating the first derivative sequence of the outbound rate of the front warehouse node and the Pearson cross-correlation coefficient, a spatiotemporal consumption resonance index is generated. A resonance weighted matrix is constructed, the network topology is reconstructed, and a master node is elected. Replenishment requests are intercepted, and the master node is used to execute terminal business orders on behalf of the client. The available commitment quantity is virtually mapped, the resonance link is cut off, and business continuity is maintained within the latency waiting window.
It enables quantitative identification and intervention of the consumption resonance state between nodes of the same level of front-end warehouses, avoids cascading stockouts, maintains the continuity of front-end business, and improves the operational stability of the multi-level supply chain network.
Smart Images

Figure CN123070853A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of supply chain warehousing and logistics and information collaborative processing technology, specifically to an intelligent inventory scheduling management system and method for a commercial supply chain. Background Technology
[0002] Currently, multi-level commercial supply chain networks typically employ a hierarchical architecture from a central warehouse to forward warehouses for goods distribution. Inventory scheduling at each forward warehouse node relies on a set static safety stock threshold. When a node's absolute physical inventory falls below the threshold, the system triggers a replenishment request to the higher-level warehouse. When the supply chain network faces regional demand fluctuations, multiple forward warehouses at the same level, geographically close, may exhibit convergent inventory depletion rates. Traditional management systems based on static thresholds only monitor the absolute inventory of a single node, lacking the ability to identify the correlation characteristics of inventory depletion rates across multiple nodes. In this situation, adjacent forward warehouse nodes may trigger replenishment alarms and send a large number of concurrent requests to the higher-level warehouse within the same timeframe, causing regional cascading stockouts in the supply chain network and impacting overall operational stability.
[0003] When dealing with such concurrent stockouts, existing systems typically employ scheduling strategies that restrict replenishment quotas for lower-level nodes or block order-taking permissions for nodes. This scheduling approach directly transmits the inventory shortage status of the underlying physical warehouse to the front-end business system, causing order fulfillment interruptions within the service area of the restricted nodes. It fails to resolve concurrent stockout issues in the physical network while maintaining the continuity of terminal business operations. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides an intelligent inventory scheduling management system and method for commercial supply chains. It solves the problem that in existing multi-level supply chain networks, when facing regional demand fluctuations, the strong correlation between the inventory consumption rates of front-end warehouse nodes at the same level can easily lead to system-level stockouts. Traditional scheduling schemes based on static inventory thresholds cannot identify and intervene in the consumption resonance state between nodes, and the execution of flow restriction or blocking scheduling can lead to the interruption of front-end business order fulfillment.
[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent inventory scheduling and management system and method for a commercial supply chain, comprising: The business data stream sampling module is used to periodically acquire the physical outbound logs of the same-level front warehouse nodes in the supply chain network at a preset discrete time step, and calculate the first derivative sequence of the outbound rate of each front warehouse node within the sliding time window. The resonance matrix construction module is used to calculate the Pearson cross-correlation coefficient of the first derivative sequence of the outbound rate of any two peer front warehouse nodes, and combine the Pearson cross-correlation coefficient with the physical distance of logistics transportation between the two nodes to generate the spatiotemporal consumption resonance index, and then construct a spatiotemporal consumption resonance weighted matrix containing the correlation of all peer front warehouse nodes. The network topology reconstruction module is used to modify the transfer edge impedance between the two nodes to infinity in the adjacency matrix of the graph theory pathfinding algorithm when the spatiotemporal consumption resonance index between the two nodes is greater than the resonance judgment threshold, so as to cut off the lateral transfer execution link, and aggregate multiple front warehouse nodes that are interconnected and whose spatiotemporal consumption resonance index is greater than the resonance judgment threshold into a resonance cluster set. The master control election and replenishment interception module is used to elect a unique node as the master control node warehouse in the resonance cluster set according to a preset weight formula, and update the remaining nodes to a controlled suspended state; when a node in the resonance cluster set triggers a stockout alarm, it intercepts the independent replenishment request of the node in the controlled suspended state, and controls the upper-level central warehouse to generate a single aggregate order to send physical goods to the master control node warehouse. The fulfillment status decoupling module is used to virtually map the corresponding share of the aggregated order to a controlled suspended status node in the form of available commitment amount in the order management system while the physical goods are in transit; and after the master control node warehouse receives the physical goods into the warehouse, it starts a delay waiting window, and within the delay waiting window, the master control node warehouse executes the terminal business orders within the jurisdiction of the controlled suspended status node on behalf of the master control node through cross-node cross-database delivery rules. The steady-state feedback recovery module is used to terminate the cross-node cross-database distribution rule and clear the virtual mapping of available commitments when the delay waiting window expires, restore the transfer edge impedance between the master control node warehouse and the controlled suspended state node, and control the master control node warehouse to transfer the remaining physical inventory to the controlled suspended state node.
[0006] Preferably, the physical outbound log obtained in the business data stream sampling module is limited to the actual physical picking and outbound movement data recorded by the warehouse execution system; the calculation logic of the first derivative of the outbound rate is as follows: extract the difference between the cumulative physical outbound quantity at the current time point and the cumulative physical outbound quantity at the previous discrete time step, and divide the difference by the discrete time step.
[0007] Preferably, the logic for generating the spatiotemporal consumption resonance index in the resonance matrix construction module is as follows: extracting nodes. With nodes Within the sliding time window, for discrete difference time series, calculate the Pearson cross-correlation coefficients of the two sets of series; obtain the nodes. With nodes The spatiotemporal consumption resonance index is obtained by multiplying the Pearson cross-correlation coefficient by the reciprocal of the physical distance of logistics transportation between them.
[0008] Preferably, the network topology reconstruction module calls the graph connectivity algorithm to traverse the spatiotemporal consumption resonance weighted matrix, identifies closed loop subgraphs in the network topology where the spatiotemporal consumption resonance index between each pair is greater than the resonance judgment threshold, and defines the closed loop subgraphs as the resonance cluster set at the control logic layer.
[0009] Preferably, the preset weight formula in the master control election and replenishment interception module is as follows: obtain the historical maximum daily throughput limit of each node in the resonance cluster set and the standard vertical trunk transportation cost from the upper-level central warehouse to the corresponding node; divide the historical maximum daily throughput limit by the standard vertical trunk transportation cost to obtain the anchor point weight value of the corresponding node, and select the node with the largest anchor point weight value as the master control node warehouse.
[0010] Preferably, when the fulfillment status decoupling module executes the cross-node cross-database delivery rule, it intercepts terminal business orders belonging to the service area of the controlled suspended node in the order management system, modifies the data structure of the terminal business order, and redirects its underlying physical fulfillment execution node identifier to the identifier of the master control node warehouse. The warehouse execution system of the master control node warehouse then performs physical picking and outbound operations based on the redirected order details.
[0011] Preferably, during the delay waiting window, due to the execution of the cross-node cross-warehouse delivery rule, the local warehousing execution system of the controlled suspended node is in a state of no physical operation, and its cumulative physical outbound volume remains constant, so that the first derivative value of the outbound rate of the controlled suspended node is forcibly maintained at zero. The cross-correlation of the first derivative sequence of the outbound rate between the controlled suspended node and the master control node warehouse is cut off by the physical operation stagnation.
[0012] Preferably, the system further includes a constrained pathfinding module; for independent front-end warehouse nodes that do not belong to any resonance cluster set and trigger a stockout alarm, the constrained pathfinding module calls a graph theory pathfinding algorithm to search for lateral allocation supply nodes. The objective function of the graph theory pathfinding algorithm is calculated as follows: the baseline allocation cost between the potential supply node and the stockout node, multiplied by a positive power of the product of the natural constant as the base, the exponent of which is the penalty coefficient, and the spatiotemporal consumption resonance exponent between the two nodes; the system selects the potential supply node with the smallest output value of the objective function as the allocation execution node.
[0013] Preferably, the order management system and the warehouse execution system of the same-level front warehouse node are deployed separately at the data layer. The available commitment quantity only modifies the order acceptance available quantity field in the order management system and does not generate physical inventory deduction flow in the warehouse execution system.
[0014] A method for intelligent inventory scheduling and management in a commercial supply chain includes the following steps: S1: Periodically acquire the physical outbound logs of the same-level front warehouse nodes in the supply chain network at a preset discrete time step, and calculate the first derivative sequence of the outbound rate within the sliding time window. S2: Calculate the Pearson cross-correlation coefficient of the first derivative sequence of the outbound rate of any two front warehouse nodes of the same level and combine it with the reciprocal of the physical distance of logistics transportation to generate the spatiotemporal consumption resonance index and construct the spatiotemporal consumption resonance weighted matrix. S3: When the spatiotemporal consumption resonance index between two nodes is greater than the resonance judgment threshold, the transfer edge impedance between the two nodes is modified to infinity in the adjacency matrix of the graph theory pathfinding algorithm, and multiple front warehouse nodes that are interconnected and whose spatiotemporal consumption resonance index is greater than the resonance judgment threshold are aggregated into a resonance cluster set. S4: Elect the only node in the resonance cluster set as the master node based on the preset weight formula, and update the remaining nodes to the controlled suspended state; S5: Intercept replenishment requests from nodes in a controlled suspended state, and control the upper-level central warehouse to generate aggregated orders and send physical goods to the master control node warehouse; S6: During the period when the physical goods are in transit, the corresponding share of the aggregated order is virtually mapped to the controlled suspended state node in the form of available commitment in the order management system; S7: After the main control node warehouse receives the physical goods into the warehouse, a delay waiting window is started. Within the delay waiting window, the main control node warehouse executes the terminal business orders within the jurisdiction of the controlled suspended node by means of the cross-node cross-database delivery rule. S8: When the delay waiting window timer expires, terminate the cross-node cross-database distribution rule and clear the virtual mapping, restore the allocation edge impedance, and control the master control node warehouse to allocate the remaining physical inventory to the controlled suspended state node.
[0015] This invention provides an intelligent inventory scheduling and management system and method for a commercial supply chain. It has the following beneficial effects: 1. This invention calculates the first derivative sequence of the outbound rate by extracting physical outbound logs, and generates a spatiotemporal consumption resonance index by combining cross-correlation coefficients and logistics transportation distance, thereby constructing a resonance weighted matrix. This mechanism transforms the inventory monitoring benchmark from a static stock index to a dynamic consumption rate correlation, enabling the system to quantitatively identify the coupling state of demand fluctuations between peer-level front-end warehouse nodes, providing an objective data basis for subsequent blocking of resonance links and reconstruction of network topology.
[0016] 2. During the replenishment interception and targeted delivery of physical goods to the resonance cluster set, this invention virtually maps the available commitment quantity in the order management system and, within the delay waiting window, the master control node warehouse performs cross-database delivery of terminal business orders. This mechanism utilizes the data separation architecture between the information system and physical warehousing to maintain the availability of the front-end business interface during the phase when the underlying physical operations of the controlled suspended node are stagnant, thus solving the problem that traditional flow-limiting scheduling intervention directly leads to order fulfillment interruption.
[0017] 3. This invention utilizes cross-regional dispatching to restrict the actual physical outbound operations of nodes in a controlled suspended state, forcing their outbound rate's first derivative to zero to eliminate the cross-correlation of consumption frequencies between nodes. Furthermore, a positive power-law penalty term based on the resonance exponent is added to the objective function of the allocation and routing algorithm for independent nodes. This mechanism, at the underlying logic level, severs the cascading stockout associations between nodes and enables the system to automatically avoid high-resonance-risk nodes during regular supply node searches, thereby improving the operational stability of the multi-level supply chain network in the face of regional demand fluctuations. Attached Figure Description
[0018] Figure 1 This is a schematic diagram of the system architecture of the present invention; Figure 2 This is a schematic diagram of the method steps of the present invention. Detailed Implementation
[0019] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0020] Please see the appendix Figure 1 A smart inventory scheduling and management system and method for a commercial supply chain, comprising: The business data stream sampling module is used to periodically acquire the physical outbound logs of the same-level front warehouse nodes in the supply chain network at a preset discrete time step, and calculate the first derivative sequence of the outbound rate of each front warehouse node within the sliding time window. The resonance matrix construction module is used to calculate the Pearson cross-correlation coefficient of the first derivative sequence of the outbound rate of any two peer front warehouse nodes, and combine the Pearson cross-correlation coefficient with the physical distance of logistics transportation between the two nodes to generate the spatiotemporal consumption resonance index, and then construct a spatiotemporal consumption resonance weighted matrix containing the correlation of all peer front warehouse nodes. The network topology reconstruction module is used to modify the transfer edge impedance between the two nodes to infinity in the adjacency matrix of the graph theory pathfinding algorithm when the spatiotemporal consumption resonance index between the two nodes is greater than the resonance judgment threshold, so as to cut off the lateral transfer execution link, and aggregate multiple front warehouse nodes that are interconnected and whose spatiotemporal consumption resonance index is greater than the resonance judgment threshold into a resonance cluster set. The master control election and replenishment interception module is used to elect a unique node as the master control node warehouse in the resonance cluster set according to a preset weight formula, and update the remaining nodes to a controlled suspended state; when a node in the resonance cluster set triggers a stockout alarm, it intercepts the independent replenishment request of the node in the controlled suspended state, and controls the upper-level central warehouse to generate a single aggregate order to send physical goods to the master control node warehouse. The fulfillment status decoupling module is used to virtually map the corresponding share of the aggregated order to a controlled suspended status node in the form of available commitment amount in the order management system while the physical goods are in transit; and after the master control node warehouse receives the physical goods into the warehouse, it starts a delay waiting window, and within the delay waiting window, the master control node warehouse executes the terminal business orders within the jurisdiction of the controlled suspended status node on behalf of the master control node through cross-node cross-database delivery rules. The steady-state feedback recovery module is used to terminate the cross-node cross-database distribution rule and clear the virtual mapping of available commitments when the delay waiting window expires, restore the allocation edge impedance between the master control node warehouse and the controlled suspended state node, and control the master control node warehouse to allocate the remaining physical inventory to the controlled suspended state node. The constrained pathfinding module, for independent front-end warehouse nodes that do not belong to any resonance cluster set and trigger a stockout alarm, calls a graph theory pathfinding algorithm to search for lateral allocation supply nodes. The objective function of the graph theory pathfinding algorithm is calculated as follows: the baseline allocation cost between the potential supply node and the stockout node, multiplied by the natural constant as the base, the exponent of which is the penalty coefficient, and the positive power of the product of the spatiotemporal consumption resonance exponent between the two nodes; the system selects the potential supply node with the smallest output value of the objective function as the allocation execution node.
[0021] The system's underlying infrastructure relies on a hardware communication environment that includes an order management system, a warehouse execution system, and multiple supply chain network nodes. These supply chain network nodes consist of a centrally managed upper-level warehouse and multiple parallel, peer-level forward warehouses. Supported by a time-series database and a scheduling engine, the system executes various business data processing and control logic modules.
[0022] The system uses a business data stream sampling module with a fixed discrete time step. The system periodically retrieves physical outbound logs and absolute physical inventory levels for each inventory unit from the warehouse execution system at each forward warehouse node. The physical outbound logs are limited to actual physical picking and outbound movement data recorded by the warehouse execution system, excluding logical inventory deduction records from the order management system. The system also extracts the cumulative physical outbound quantity within the current time step. And calculate the front warehouse node. First derivative of the outbound rate within the current time window The calculation logic for the first derivative of the outbound rate is the sum of the cumulative physical outbound volume at the current time point and the cumulative physical outbound volume at the previous time step. The difference divided by the discrete time step The system will continue. The results of calculating the first derivative of the outbound rate within each time step are stored in a time series, forming a corresponding front-end warehouse node with a length of [missing information]. The discrete difference time series. When there is no physical shipment operation at the corresponding node, the first derivative of its outbound rate is zero.
[0023] The system calculates the spatiotemporal consumption resonance index between peer-level front-end warehouse nodes using a resonance matrix construction module. The system extracts any two peer-level front-end warehouse nodes from the network. With nodes In the sliding time window For discrete difference time series within a given time period, calculate the Pearson cross-correlation coefficient between the two sets of series. The system obtains node information through a basic information database. With nodes Physical distance of logistics transportation between The system will use Pearson cross-correlation coefficients. Multiplying this by the reciprocal of the physical distance for logistics transportation generates the spatiotemporal consumption resonance index between the two nodes. Based on the calculation results of all peer-level front-end nodes in the network, the system generates and periodically updates a spatiotemporal consumption resonance weighted matrix containing the relationships between all nodes in memory. .
[0024] The system reads the spatiotemporal consumption resonance weighted matrix through the network topology reconstruction module. The system extracts the spatiotemporal consumption resonance index between any two nodes in the matrix. And compare it with the resonance judgment threshold set by the system. A comparison is performed. If the comparison result exceeds the resonance determination threshold... At that time, the system places the nodes in the adjacency matrix of the graph theory pathfinding algorithm. With nodes Interchange edge impedance The value was changed from the basic transportation cost to infinity, and the lateral allocation execution link between the two nodes was severed at the control layer of the scheduling system. The system called the graph connectivity algorithm to traverse the updated resonance weighted matrix. When multiple interconnected nodes were identified in the network topology and the spatiotemporal consumption resonance index between any two nodes was greater than the resonance judgment threshold, the algorithm was activated. When a set of nodes is generated, the system aggregates the set of nodes into a resonant cluster set at the control logic layer. .
[0025] The system uses the master control election and replenishment interception modules to target the formation of resonance cluster sets. The nodes are then reconstructed hierarchically. The system obtains the nodes within the resonance cluster set respectively. The highest historical daily throughput limit And the standard vertical trunk transportation costs from the central warehouse to the corresponding nodes. The system divides the historical maximum daily throughput by the standard transportation cost of the vertical trunk line to obtain the anchor point weight value of the corresponding node. The system selects the node with the largest anchor weight value as the master control node and sets up the resonance cluster. The status flags of the remaining front-end nodes are updated to a controlled suspended state. When the resonance cluster set... Absolute physical inventory of any front-end warehouse node Below safety stock alarm threshold At that time, the system intercepts replenishment request data packets sent from controlled suspended nodes to the central warehouse. The system aggregates the resonant cluster set. The total demand generated by all nodes within the system is The system manages a single aggregated order and controls the central warehouse to distribute all physical goods to the main control node warehouse based on the aggregated order.
[0026] The system handles business order requests during the physical replenishment transit period through a fulfillment status decoupling module. The system extracts the total quantity of aggregated orders. Based on the historical sales ratio of each node within the resonance cluster, an available commitment quantity is generated. In the order management system, the corresponding share of the available commitment quantity is virtually mapped to nodes in a controlled suspended state. The available inventory displayed in the business interface of the order management system for nodes in the controlled suspended state is updated to the available commitment quantity value. When the master control node warehouse receives a physical goods receipt instruction from the central warehouse, the system initiates a cross-node cross-database delivery control process and triggers a process lasting for a specified duration. The delay waiting window.
[0027] During this delayed waiting window, the system intercepts terminal business orders belonging to the controlled suspended node service area in the order management system. The system modifies the data structure of the terminal business orders, redirecting their underlying physical fulfillment node identifier to the master node warehouse identifier. The master node warehouse's storage execution system performs physical picking and outbound operations based on the redirected order details. During this delayed waiting window, the local storage execution system of the controlled suspended node is in a state of no physical operations, and its actual physical outbound volume... By maintaining a constant value, the first derivative of the outbound rate of this node is forced to remain at zero. This state physically eliminates the cross-correlation of consumption acceleration characteristics between the controlled suspended node and the master node's warehouse.
[0028] The system monitors the duration of the latency wait window through the steady-state feedback recovery module. When the latency wait window timer reaches... At this time, the system terminates the cross-node cross-warehouse delivery routing intervention rule in the order management system and clears the virtual mapping record of available commitments. The system restores the transfer edge impedance between the master control node warehouse and the controlled suspended state node in the graph theory pathfinding algorithm to the normal logistics cost parameters. The system issues a lateral transfer instruction to control the master control node warehouse to transfer the remaining physical inventory to the controlled suspended state node. After the warehouse execution system of the controlled suspended state node receives the physical inventory, it resumes the independent fulfillment execution process, and the system resets the resonant cluster set. The aggregation state.
[0029] For independent front-end warehouse nodes that do not belong to any resonance cluster set and trigger the safety stock alarm threshold for stockouts The system generates normal-state lateral allocation instructions through a constrained pathfinding module. The system then invokes a graph theory-based pathfinding algorithm to locate potential supplier nodes. When searching for allocation sources, an objective function with a resonance penalty term is invoked. This objective function is calculated based on the supply node. With out-of-stock nodes Baseline transfer cost between Multiply by the natural constant The base is the base and the exponent is the penalty coefficient. Resonance index of spatiotemporal consumption between two nodes The objective function formula for the negative power of the product is: ; Based on the calculation results of the above objective function, the system identifies the potential supply node with the smallest output value as the final horizontal transfer node to execute subsequent transfer business processes.
[0030] Please see the appendix Figure 2A smart inventory scheduling and management method for a commercial supply chain includes the following steps: Step S1: Periodically acquire the physical outbound logs of the same-level front warehouse nodes in the supply chain network with a preset discrete time step, and calculate the first derivative sequence of the outbound rate within the sliding time window.
[0031] The system pre-sets the discrete time step (e.g., 1 hour or 1 calendar day) and the sliding time window length. The system obtains the actual physical movement data of picking and outbound completion recorded by the peer front warehouse node in the warehouse execution system through interface calls, excluding virtual data that is only created at the order level but has not been physically outbound.
[0032] In terms of computational logic, the system extracts the difference between the cumulative physical outbound volume at the current time point and the cumulative physical outbound volume at the previous discrete time step, and divides this difference by the discrete time step to obtain the dynamic outbound rate at the current moment. Further, it extracts the difference in dynamic outbound rates between adjacent time steps, divides it by the discrete time step, and obtains the first derivative of the outbound rate. The system sorts multiple first derivative values calculated consecutively within the sliding time window by time, forming a sequence of first derivatives of the outbound rate. This step transforms the static absolute inventory level indicator into a dynamic characteristic quantity representing the acceleration of inventory consumption at a given node.
[0033] Step S2: Calculate the Pearson cross-correlation coefficient of the first derivative sequence of the outbound rate of any two same-level front warehouse nodes, and combine it with the reciprocal of the physical distance of logistics transportation to generate the spatiotemporal consumption resonance index and construct the spatiotemporal consumption resonance weighted matrix.
[0034] The system traverses the peer-to-peer front-end nodes in the network, for any node With nodes The first derivative sequence of the outbound rate within the same sliding time window is extracted. A statistical algorithm is then used to calculate the Pearson cross-correlation coefficient between the two sequences, which is used to objectively quantify the trend synergy of the two nodes in demand fluctuations.
[0035] Subsequently, the system obtains the node. With nodes The physical distance for logistics transportation between nodes is calculated, and its reciprocal is multiplied by the Pearson cross-correlation coefficient to obtain the spatiotemporal consumption resonance index. The system then fills the calculated spatiotemporal consumption resonance indices of all node pairs into a two-dimensional matrix to generate a spatiotemporal consumption resonance weighted matrix.
[0036] Step S3: When the spatiotemporal consumption resonance index between two nodes is greater than the resonance judgment threshold, the transfer edge impedance between the two nodes is modified to infinity in the adjacency matrix of the graph theory pathfinding algorithm, and multiple front warehouse nodes that are interconnected and whose spatiotemporal consumption resonance index is greater than the resonance judgment threshold are aggregated into a resonance cluster set.
[0037] The system iterates through the spatiotemporal consumption resonance weighting matrix based on a set resonance determination threshold. For node pairs with an exponent greater than this threshold, the system modifies the data in the corresponding pathfinding adjacency matrix in the network topology graph of the control logic layer, forcibly setting the transfer edge impedance value between the two nodes to infinity. The physical meaning of this operation is to cut off the execution link of lateral transfer between nodes.
[0038] Subsequently, the system calls the connected graph search algorithm to identify closed node subgraphs in the adjacency matrix, and divides multiple front warehouse nodes that are connected to each other with high resonance exponents into independent resonance cluster sets, thus isolating them from scheduling interactions with other regular nodes.
[0039] Step S4: Elect the only node in the resonance cluster set as the master node warehouse according to the preset weight formula, and update the remaining nodes to the controlled suspended state.
[0040] For the identified resonant cluster set, the system obtains the equipment and cost parameters of each node in the set, including the historical maximum daily throughput limit of each node, and the standard vertical trunk transportation cost from the upper-level central warehouse to the corresponding node.
[0041] The system executes the weighting formula: dividing the historical maximum daily throughput by the standard transportation cost of the vertical trunk line, the anchor point weight value of each node is calculated. The system selects the node with the largest anchor point weight value and updates its status flag to the master control node warehouse, while updating the status flags of the remaining nodes in the set to the controlled suspended state.
[0042] Step S5: Intercept the replenishment request of the controlled suspended node, and control the upper-level central warehouse to generate an aggregated order and send the physical goods to the master control node warehouse.
[0043] When any node within the resonance cluster triggers a safety stock shortage alarm, the system intercepts independent replenishment instruction messages sent from controlled, suspended nodes to the central warehouse. The system aggregates the shortage demand from all nodes within the resonance cluster, generating a single aggregated order containing the total demand. The system redirects the destination of this aggregated order to the master control node warehouse and drives the upper-level central warehouse to perform physical picking, loading, and shipping operations only to that master control node warehouse.
[0044] Step S6: While the physical goods are in transit, the corresponding share of the aggregated order is virtually mapped to a controlled suspended state node in the form of available commitments in the order management system.
[0045] During the transit phase of aggregated physical goods transported to the master control node warehouse via trunk line, the system invokes the inventory field modification interface of the Order Management System (OMS). Based on the previously summarized demand ratios of each node, the system divides the physical quantity in the aggregated order into corresponding shares and adds them to the OMS available inventory field of the controlled suspended node in the form of Available-To-Promise (ATP). This operation only modifies the data at the order access layer and does not trigger the inbound flow of the underlying physical inventory in the Warehouse Execution System (WMS), thus maintaining the continuous order-taking capability of the terminal business channels.
[0046] Step S7: After the master control node warehouse receives the physical goods into the warehouse, a delay waiting window is started. Within the delay waiting window, the master control node warehouse executes the terminal business orders within the jurisdiction of the controlled suspended node by means of the cross-node cross-database delivery rule.
[0047] Once the WMS at the master control node warehouse confirms the physical receipt of goods, the system starts a built-in timer to begin a delay waiting window. During this window, the system activates the cross-node cross-warehouse dispatch rule: for terminal business orders belonging to the service jurisdiction of controlled suspended nodes, the system intercepts the order at the OMS layer and redirects its underlying physical fulfillment execution node to the master control node warehouse. The WMS at the master control node warehouse then executes the physical picking and outbound operations based on the redirection instructions.
[0048] During this period, the local WMS of the controlled suspended node is in a state of no physical outbound operations, and its cumulative physical outbound volume remains constant. According to the calculation logic of step S1, the first derivative value of the outbound rate of the controlled suspended node is objectively limited to zero, thereby eliminating its cross-correlation with the outbound sequences of other nodes in the resonance cluster and eliminating the consumption resonance state that would trigger cascading stockouts.
[0049] Step S8: When the delay waiting window ends, terminate the cross-node cross-database distribution rule and clear the virtual mapping, restore the allocation edge impedance, and control the master control node warehouse to allocate the remaining physical inventory to the controlled suspended node.
[0050] When the timer for the waiting window reaches zero, the system sends a command to OMS to terminate the redirection logic of the cross-node cross-database proxy rule and deduct the excess available commitment virtual mapping data in the controlled suspended state node.
[0051] Simultaneously, the system restores the transfer edge impedance between the master control node warehouse and the controlled suspended nodes from infinity to the baseline transfer cost value in the pathfinding adjacency matrix, reconnecting the lateral transfer link. Finally, based on the remaining actual shortage quantity of each suspended node, the system generates lateral transfer instructions, controlling the master control node warehouse to allocate and distribute the remaining physical inventory to the physical warehouses of each controlled suspended node, thus restoring the entire supply chain network to normal operation. Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A smart inventory scheduling management system and method for a commercial supply chain, characterized in that, include: The business data stream sampling module is used to periodically acquire the physical outbound logs of the same-level front warehouse nodes in the supply chain network at a preset discrete time step, and calculate the first derivative sequence of the outbound rate of each front warehouse node within the sliding time window. The resonance matrix construction module is used to calculate the Pearson cross-correlation coefficient of the first derivative sequence of the outbound rate of any two peer front warehouse nodes, and combine the Pearson cross-correlation coefficient with the physical distance of logistics transportation between the two nodes to generate the spatiotemporal consumption resonance index, and then construct a spatiotemporal consumption resonance weighted matrix containing the correlation of all peer front warehouse nodes. The network topology reconstruction module is used to modify the transfer edge impedance between the two nodes to infinity in the adjacency matrix of the graph theory pathfinding algorithm when the spatiotemporal consumption resonance index between the two nodes is greater than the resonance judgment threshold, so as to cut off the lateral transfer execution link, and aggregate multiple front warehouse nodes that are interconnected and whose spatiotemporal consumption resonance index is greater than the resonance judgment threshold into a resonance cluster set. The master node election and replenishment interception module is used to elect a unique node as the master node warehouse in the resonant cluster set according to a preset weight formula, and update the remaining nodes to a controlled suspended state. When a node in the resonance cluster triggers a stockout alarm, the independent replenishment request of the controlled suspended node is intercepted, and the upper-level central warehouse is controlled to generate a single aggregated order and send physical goods to the master control node warehouse. The fulfillment status decoupling module is used to virtually map the corresponding share of the aggregated order to a controlled suspended status node in the form of available commitment amount in the order management system while the physical goods are in transit; and after the master control node warehouse receives the physical goods into the warehouse, it starts a delay waiting window, and within the delay waiting window, the master control node warehouse executes the terminal business orders within the jurisdiction of the controlled suspended status node on behalf of the master control node through cross-node cross-database delivery rules. The steady-state feedback recovery module is used to terminate the cross-node cross-database distribution rule and clear the virtual mapping of available commitments when the delay waiting window expires, restore the transfer edge impedance between the master control node warehouse and the controlled suspended state node, and control the master control node warehouse to transfer the remaining physical inventory to the controlled suspended state node.
2. The intelligent inventory scheduling management system and method for a commercial supply chain according to claim 1, characterized in that, The physical outbound log obtained in the business data stream sampling module is limited to the actual physical picking and outbound movement data recorded by the warehouse execution system; the calculation logic of the first derivative of the outbound rate is as follows: extract the difference between the cumulative physical outbound quantity at the current time point and the cumulative physical outbound quantity at the previous discrete time step, and divide the difference by the discrete time step.
3. The intelligent inventory scheduling management system and method for a commercial supply chain according to claim 1, characterized in that, The logic for generating the spatiotemporal consumption resonance index in the resonance matrix construction module is as follows: extract nodes. With nodes Within the sliding time window, for discrete difference time series, calculate the Pearson cross-correlation coefficients of the two sets of series; obtain the nodes. With nodes The spatiotemporal consumption resonance index is obtained by multiplying the Pearson cross-correlation coefficient by the reciprocal of the physical distance of logistics transportation between them.
4. The intelligent inventory scheduling management system and method for a commercial supply chain according to claim 3, characterized in that, The network topology reconstruction module calls the graph connectivity algorithm to traverse the spatiotemporal consumption resonance weighted matrix, identifies closed-loop subgraphs in the network topology where the spatiotemporal consumption resonance index between each pair is greater than the resonance judgment threshold, and defines the closed-loop subgraphs as the resonance cluster set in the control logic layer.
5. The intelligent inventory scheduling management system and method for a commercial supply chain according to claim 1, characterized in that, The preset weight formula in the master control election and replenishment interception module is as follows: obtain the historical maximum daily throughput limit of each node in the resonance cluster set and the standard vertical trunk transportation cost from the upper central warehouse to the corresponding node; divide the historical maximum daily throughput limit by the standard vertical trunk transportation cost to obtain the anchor point weight value of the corresponding node, and select the node with the largest anchor point weight value as the master control node warehouse.
6. The intelligent inventory scheduling management system and method for a commercial supply chain according to claim 1, characterized in that, When executing cross-node cross-database delivery rules, the fulfillment status decoupling module intercepts terminal business orders belonging to the service area of controlled suspended nodes in the order management system, modifies the data structure of the terminal business orders, and redirects the underlying physical fulfillment execution node identifier to the identifier of the master control node warehouse. The warehouse execution system of the master control node warehouse then performs physical picking and outbound operations based on the redirected order details.
7. The intelligent inventory scheduling management system and method for a commercial supply chain according to claim 6, characterized in that, During the delay waiting window, due to the execution of the cross-node cross-warehouse delivery rule, the local warehousing execution system of the controlled suspended node is in a state of no physical operation, and its cumulative physical outbound volume remains constant, so that the first derivative value of the outbound rate of the controlled suspended node is forcibly maintained at zero. By suspending physical operations, the cross-correlation of the first derivative sequence of the outbound rate between the controlled suspended node and the master control node warehouse is cut off.
8. The intelligent inventory scheduling management system and method for a commercial supply chain according to claim 1, characterized in that, The system also includes a constrained pathfinding module. For independent front-end warehouse nodes that do not belong to any resonance cluster set and trigger a stockout alarm, the constrained pathfinding module calls a graph theory pathfinding algorithm to search for lateral allocation supply nodes. The objective function of the graph theory pathfinding algorithm is calculated as follows: the baseline allocation cost between the potential supply node and the stockout node, multiplied by a natural constant as the base and the exponent of the penalty coefficient, and the positive power of the product of the spatiotemporal consumption resonance exponent between the two nodes. The system selects the potential supply node with the smallest output value of the objective function as the allocation execution node.
9. The intelligent inventory scheduling management system and method for a commercial supply chain according to claim 1, characterized in that, The order management system and warehouse execution system of the same-level front warehouse node are deployed separately at the data layer. The available commitment quantity only modifies the available quantity field of the order management system and does not generate physical inventory deduction flow in the warehouse execution system.
10. A method for intelligent inventory scheduling and management in a commercial supply chain, used in the intelligent inventory scheduling and management system and method for a commercial supply chain according to any one of claims 1-9, characterized in that, Includes the following steps: S1: Periodically acquire the physical outbound logs of the same-level front warehouse nodes in the supply chain network at a preset discrete time step, and calculate the first derivative sequence of the outbound rate within the sliding time window. S2: Calculate the Pearson cross-correlation coefficient of the first derivative sequence of the outbound rate of any two front warehouse nodes of the same level and combine it with the reciprocal of the physical distance of logistics transportation to generate the spatiotemporal consumption resonance index and construct the spatiotemporal consumption resonance weighted matrix. S3: When the spatiotemporal consumption resonance index between two nodes is greater than the resonance judgment threshold, the transfer edge impedance between the two nodes is modified to infinity in the adjacency matrix of the graph theory pathfinding algorithm, and multiple front warehouse nodes that are interconnected and whose spatiotemporal consumption resonance index is greater than the resonance judgment threshold are aggregated into a resonance cluster set. S4: Select the only node in the resonance cluster set as the master node based on the preset weight formula, and update the remaining nodes to the controlled suspended state. S5: Intercept replenishment requests from nodes in a controlled suspended state, and control the upper-level central warehouse to generate aggregated orders and send physical goods to the main control node warehouse; S6: During the period when the physical goods are in transit, the corresponding share of the aggregated order is virtually mapped to the controlled suspended state node in the form of available commitment in the order management system; S7: After the main control node warehouse receives the physical goods into the warehouse, a delay waiting window is started. Within the delay waiting window, the main control node warehouse executes the terminal business orders within the jurisdiction of the controlled suspended node by means of the cross-node cross-database delivery rule. S8: When the delay waiting window timer expires, terminate the cross-node cross-database distribution rule and clear the virtual mapping, restore the allocation edge impedance, and control the master control node warehouse to allocate the remaining physical inventory to the controlled suspended state node.
