A digital management data analysis system for business information

Abstract

The application relates to the field of business data analysis, and discloses a digital management data analysis system for business information, which comprises a storage module, an acquisition module, a business module and a calculation module. The acquisition module acquires the residence time of a logistics node and calculates a real-time node blocking coefficient to form a node blocking state column vector. The calculation module operates a fluctuation conduction matrix with the column vector to extract a comprehensive expected lead time offset, and reconstructs a dynamic trigger threshold value by using the offset. When the real-time inventory decreases to the dynamic trigger threshold value, a target total demand is generated and an interruption mechanism is triggered, a soft constraint operational model containing a gap relaxation variable is constructed for the interruption mechanism, and an optimal quota allocation solution set is solved and output. The business module generates multi-source sub-purchase order data packets and abnormal supply interruption early warning queues according to the optimal quota allocation solution set. The application realizes dynamic self-adaptive adjustment of the inventory trigger water level, and outputs an optimal optional purchase quota strategy through operational solving when the logistics is blocked.

Description

Technical Field

[0001] This invention relates to the field of commercial data analysis technology, specifically a digital management data analysis system for commercial business information. Background Technology

[0002] In the digital management of commercial operations, inventory control and automatic generation of procurement plans are core functions of data analysis systems. Existing commercial business systems mainly operate based on preset fixed rules, typically extracting static delivery lead times and fixed safety stock values ​​as the underlying data basis for triggering procurement operations.

[0003] In actual commercial logistics and transportation networks, unpredictable delays frequently occur at various logistics nodes. Existing data analysis systems lack real-time collection and quantification mechanisms for the logistics congestion status of local nodes, making it difficult to calculate the impact of local node delays on the overall supply chain delivery cycle through mathematical models. This management approach, relying on static parameters, prevents the system from adaptively adjusting inventory trigger levels based on actual logistics conditions. When faced with logistics blockages, this often leads to delays in issuing procurement orders, resulting in inventory shortages at the end-user level. Furthermore, when the main supply chain is blocked and multi-source alternative procurement needs to be initiated, trading companies need to comprehensively consider multiple dimensions of attributes, including the physical capacity limits of each alternative supplier, overall logistics costs, and fulfillment timeliness. Existing digital systems lack the ability to coordinate and solve these multiple constraints, relying heavily on manual experience for order splitting. Moreover, existing basic allocation algorithms often stall and become unsolvable when faced with extreme situations where the total available alternative capacity across the entire network is still less than the target demand, due to the inability to meet the rigid constraint of total quantity conservation. The system is unable to output feasible partial allocation plans at the current stage, nor can it provide quantitative gap warning data to the business side, which restricts the decision-making efficiency of commercial enterprises in dealing with abnormal supply disruptions. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a digital management data analysis system for commercial business information. This system solves the problems of existing commercial business systems, which use static delivery parameters and fixed inventory thresholds, making it impossible to dynamically trigger procurement based on actual logistics disruptions, and lacking a multi-source alternative capacity allocation mechanism when facing supply chain anomalies.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a digital management and data analysis system for commercial business information, comprising a storage module, a data acquisition module, a business module, and a computing module.

[0006] The storage module is used to store the directed acyclic graph of the supply chain, the basic business attributes of the product objects, and the historical baseline time parameters.

[0007] The data acquisition module is used to obtain the dwell time of logistics nodes and calculate the real-time node stagnation coefficient, forming a column vector of node stagnation status.

[0008] The business module is used to obtain the real-time inventory of product objects, construct a directed acyclic graph of the supply chain using basic business attributes, and generate multi-source sub-purchase order data packages and abnormal supply interruption early warning queues based on the optimal quota allocation solution set.

[0009] The calculation module is used to extract the comprehensive expected lead time offset by operating the fluctuation transmission matrix and the column vector of the node stagnation state. The comprehensive expected lead time offset is used to reconstruct the dynamic trigger threshold. When the real-time inventory drops to the dynamic trigger threshold, the target total demand is generated and the interruption mechanism is triggered. A soft-constraint operations research model with gap relaxation variables is constructed for the interruption mechanism, and the optimal quota allocation solution set is output.

[0010] Through the above scheme, the system uses a fluctuation transmission matrix to quantify the obstruction state of local logistics nodes and transform it into the lead time offset of the entire link, realizing dynamic adaptive reconstruction of the trigger threshold. When the interruption mechanism is triggered, the system uses a soft-constraint operations research model to jointly solve the multi-source supply path, and outputs the quota allocation strategy with the lowest total system cost under the premise of ensuring the timeliness of fulfillment, thus solving the supply chain disruption problem caused by the obstruction of a single path.

[0011] Furthermore, the storage module includes graph data units and relational data units. Graph data units store the directed acyclic graph of the supply chain. The node set of each graph data unit corresponds to a fixed physical node, and the directed edges correspond to directional transportation links. A business graph association mapping is established using the unique identifier of the product as the root node attribute. Relational data units store basic business attributes, including average daily consumption, basic delivery lead time for the main supply path, basic safety stock value, and the physical capacity limit of alternative suppliers.

[0012] The above scheme utilizes graph data structures to perform topological mapping between logistics nodes and targeted transportation links, and associates them with basic business attributes of the commodity dimension, providing underlying data support for the subsequent reverse traversal of multi-source alternative supply paths and the construction of operations constraint equations.

[0013] Furthermore, when the acquisition module calculates the real-time node obstruction coefficient, it extracts the actual logistics dwell time of the logistics node and compares it with the historical benchmark dwell time in the historical benchmark time parameter.

[0014] If the actual logistics dwell time is less than or equal to the historical baseline dwell time, the real-time node stagnation coefficient of the logistics node will be set to zero. If the actual logistics dwell time is greater than the historical benchmark dwell time, calculate the relative time overflow difference between the actual logistics dwell time and the historical benchmark dwell time, and then calculate the ratio of the relative time overflow difference to the historical benchmark dwell time with zero overflow compensation to obtain the real-time node stall coefficient.

[0015] The acquisition module also performs fault-tolerant degradation extraction operations. When the external network interface experiences a response timeout or returns a service error code, it terminates the external network waiting process, reads the time-weighted cache value of the stagnation coefficient that was overwritten and saved in the previous cycle from the storage module, and uses the time-weighted cache value of the stagnation coefficient as the real-time node stagnation coefficient for the current calculation cycle.

[0016] The storage module performs overwrite update tasks based on time series. It uses the system's preset trust weight to perform time-weighted fusion calculation on the currently calculated real-time node stall coefficient and the historical cache value retained in the previous period to obtain the stall coefficient time-weighted cache value to be written in the current period.

[0017] Through the above scheme, the system constructs a node stall quantization mechanism based on time ratio, and combines it with fault tolerance degradation strategy and time-weighted fusion calculation when the network is abnormal to ensure that the stall coefficient can still be output smoothly and continuously when the external interface fails, so as to avoid the overall link calculation process from stalling due to the lack of data from a single node.

[0018] Furthermore, the calculation module performs algebraic multiplication on the fluctuation transmission matrix and the column vector of the node stagnation state, extracts the numerical scalar of the corresponding dimension of the target core warehouse as the comprehensive expected lead time offset; uses the comprehensive expected lead time offset to superimpose compensation on the basic delivery lead time, calculates the expected consumption in combination with the average daily consumption, and superimposes the expected consumption with the basic safety stock value to reconstruct and generate a dynamic trigger threshold. When the real-time inventory drops to or below the dynamic trigger threshold, the difference between the highest target inventory level parameter and the real-time available inventory quantity is calculated as the target total demand.

[0019] When the calculation module triggers the interrupt mechanism, it traverses the node identifier set in reverse along the set main supply path and compares the real-time node blocking coefficient of each node with the preset blocking circuit breaker threshold. If the real-time node stall coefficients of all nodes do not exceed the corresponding stall circuit breaker threshold, then a regular procurement instruction will be issued. If the real-time node stall coefficient of any node is greater than the stall circuit breaker threshold corresponding to that node, the underlying interruption mechanism of the system is triggered, the original document generation process is intercepted and the process is switched to building a soft constraint operation model.

[0020] The above solution converts actual logistics delays into incremental increases in expected consumption, allowing the system's inventory trigger level to dynamically rise based on logistics status. Simultaneously, setting a circuit breaker threshold for conditional judgment enables precise separation between routine procurement processes and emergency response procedures.

[0021] Furthermore, the soft-constraint operations research model establishes an objective function oriented towards minimizing the total system cost, which includes the sum of two costs.

[0022] The first part is the sum of the products of the individual procurement logistics cost parameters corresponding to all available alternative supply paths and the procurement quota decision variables allocated to the alternative supply paths; The second part is the product of the gap slack variable and the preset gap penalty constant.

[0023] The gap slack variable is used to characterize the difference between the upper limit of the total physical capacity of the entire network and the total target demand. The gap penalty constant is set to be greater than the comprehensive procurement and logistics cost parameter of any single entity.

[0024] The soft-constraint operations research model also includes a set of boundary constraint equations that limit the solution space, including quota conservation constraints, physical capacity boundary constraints, non-negativity constraints of slack variables, and hard constraints on performance timeliness.

[0025] The quota conservation constraint requires that the sum of all procurement quota decision variables and the cumulative value of the gap slack variable equal the total target demand. When the hard constraint of performance time limits the procurement quota decision variable to be greater than zero, the sum of the offsets of the basic delivery lead time and the comprehensive expected lead time shall not be greater than the supply interruption deadline time parameter.

[0026] For procurement quota decision variables with values ​​greater than zero, the business module extracts the corresponding alternative supply node identifiers and assembles a multi-source sub-procurement order data package containing the alternative supply node identifiers and the procurement quantity. For gap slack variables with values ​​greater than zero, extract the specific values ​​of the gap slack variables, generate an abnormal state data object with a unique identifier for the product and the number of gaps attached to the product object, and push it into the abnormal supply interruption early warning queue based on the message publish-subscribe model.

[0027] By introducing gap slack variables and high-weight penalty constants into the operations research model, the algorithm remains feasible even when the total physical capacity of the network is insufficient to fully cover the total demand. Under the premise of satisfying the hard constraints of delivery timeliness and capacity boundary constraints, the system outputs procurement allocation results distributed across multiple effective paths. Simultaneously, it transforms unbridgeable gaps into quantified abnormal supply disruption warning data and pushes it into a queue, providing accurate data for subsequent business decisions.

[0028] This invention provides a digital management and data analysis system for commercial business information. It has the following beneficial effects: 1. This invention calculates the real-time node stall coefficient by extracting the residence time of logistics nodes and combines it with the fluctuation transmission matrix to obtain the comprehensive expected lead time offset. This offset is then used to reconstruct the dynamic trigger threshold. This mechanism changes the existing system's reliance on static delivery parameters, enabling the system to adaptively adjust the inventory trigger level based on the actual logistics delay status. This allows for the early issuance of procurement orders when facing logistics congestion, preventing inventory shortages at the end of the supply chain.

[0029] 2. This invention addresses situations where the main supply chain is disrupted by constructing an operations research model that comprehensively considers constraints such as physical capacity boundaries, overall logistics costs, and delivery timelines. This overcomes the shortcomings of traditional digital systems, such as insufficient multi-constraint processing capabilities and reliance on manual order splitting. Through algorithmic iteration, it can automatically output the solution set of multi-source alternative procurement quotas that minimizes the system's total cost, thereby improving the calculation and execution efficiency of order reallocation.

[0030] 3. This invention introduces gap relaxation variables and gap penalty constants into the early soft constraint operation model. This solves the problem that the basic allocation algorithm stagnates and becomes unsolvable when the total available capacity of the entire network is less than the target demand due to the failure to satisfy the total quantity conservation. It ensures that the system can still output the currently feasible allocation scheme under extreme conditions and automatically extracts gap values ​​to generate an abnormal supply interruption early warning queue, providing quantitative data support for the enterprise's subsequent anomaly handling. Attached Figure Description

[0031] Figure 1 This is a system module architecture diagram of the present invention; Figure 2 This is a flowchart illustrating the system operation of the present invention. Figure 3 This is a cross-comparison diagram of the dynamic threshold pre-response and inventory status of the present invention; Figure 4 This is a bar chart illustrating the multi-source quota allocation strategy under the limit capacity constraint of this invention. Detailed Implementation

[0032] 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.

[0033] Please see the appendix Figure 1This invention provides a digital management and data analysis system for commercial business information, comprising: a business module, a data acquisition module, a storage module, and a computing module.

[0034] The business module is used to maintain basic business attributes and generate final procurement instructions. The business module establishes a bidirectional data communication connection with the storage module.

[0035] The storage module includes graph data units and relational data units. Graph data units store the instantiated directed acyclic graph data structure of the supply chain. Relational data units store the basic business attributes of product objects and historical baseline time parameters.

[0036] The acquisition module interfaces with an external network to retrieve physical status data from external supply chain nodes. The acquisition module then writes the calculated node impedance parameters into the storage module.

[0037] The computation module has a built-in matrix operation instruction set and a linear programming solution library. The computation module reads node status data and business parameters from the storage module, performs spatial propagation time mapping and constraint solving operations, and returns the redistribution results to the business module.

[0038] Please see the appendix Figure 2 The above system module operation process specifically includes the following steps: S100, the business module extracts the basic business attributes from the main database and instantiates and constructs a directed acyclic graph of the supply chain for a specific product object in the graph data unit; S200: The acquisition module calls an external interface to obtain the dwell time of logistics nodes, calculates the stagnation coefficient of each node in the current cycle based on the normal branch or the fault-tolerant degradation branch, and forms a column vector of node stagnation status. S300, the calculation module extracts the pre-configured fluctuation transmission matrix, multiplies it with the column vector of node stagnation state, and extracts the comprehensive expected lead time offset corresponding to the target core warehouse; S400, the calculation module reconstructs the dynamic trigger threshold of the commodity object using the comprehensive expected lead time offset. When the real-time inventory quantity drops to or below the dynamic trigger threshold, it generates a corresponding replenishment demand and triggers the subsequent verification mechanism. S500, the calculation module reverses the physical nodes on the current main supply path of the product object, compares the real-time blocking coefficient of each node with the preset blocking circuit breaker threshold. If all nodes have not reached the blocking circuit breaker threshold, the business module generates a purchase order according to the default path. If any node reaches the blocking circuit breaker threshold, the underlying interruption mechanism of the system is triggered. S600, targeting the system's underlying interruption mechanism, the calculation module extracts the multi-source supply configuration table of the commodity object, constructs a soft-constraint operations research model containing gap slack variables, calls the linear programming solution library to output the optimal quota allocation solution set, and the business module generates multiple sub-purchase order data packets pointing to each alternative supply node and an abnormal supply interruption early warning queue based on the optimal quota allocation solution set.

[0039] In this embodiment, the storage module includes graph data units and relational data units. The storage module constitutes the underlying data persistence component of the data analysis system, providing support for the categorized storage and indexing of multimodal data. During data read / write and storage organization, the module's execution logic includes a data topology definition operation S131, a relational field mapping operation S132, and a cache overwrite operation S133.

[0040] S131, in the data topology definition operation, the graph data unit is implemented using a graph storage medium that supports the attribute graph model. Based on the basic principles of graph theory, the system abstracts various entities in the business environment into graph data structures.

[0041] This involves a directed acyclic graph (DAG) in the supply chain, which is implemented at the system level as follows: The node set data entities in the DAG correspond to fixed physical nodes in the real logistics network, such as supplier factories, customs ports, transit distribution centers, and destination warehouses. Directed edge data entities store the directional transportation links between the aforementioned physical nodes, such as highway trunk lines, railway trains, or sea routes. Graph data units use unique product identifiers as root node attributes to establish graph association mappings between specific product objects and filtered physical nodes and directional transportation links. Through this mapping mechanism, upper-layer computing components can directly traverse all upstream supply links of a specific product object using graph query statements.

[0042] S132, in the relational field mapping operation, the relational data unit uses a two-dimensional relational data table structure to store business configurations and numerical parameters. The table structure of the relational data unit includes a field for recording the average daily consumption of the product object, a field for recording the basic delivery lead time of the main supply path, a field for recording the basic safety stock value, and a field for recording the physical capacity limit of the alternative suppliers.

[0043] An independent baseline time parameter table is established within the relational data unit to specifically record the historical baseline dwell time parameters of each physical node under normal operating conditions. As a preferred method, the historical baseline dwell time parameter is determined by extracting the median of the actual dwell time of the physical node over the past thirty calendar days, thereby filtering out the interference of single abnormal fluctuations. This historical baseline dwell time parameter is used as a baseline for subsequent evaluation of whether abnormal congestion has occurred at the logistics node.

[0044] In S133, during the cache overwrite operation, the local persistence layer of the storage module receives real-time node stall status data pulled by the acquisition module and executes a time-series-based overwrite update task. To address the inherent risk of downtime due to external network interface failures, the storage module allocates a key-value pair storage area within the relational data unit to independently maintain a time-weighted cache of the stall coefficients for each physical node. To ensure the timeliness of data fusion, the system verifies the timestamps of the current real-time state and the cached state before updating. Only when the timestamp difference is within one calculation cycle does the system trigger the time-weighted algorithm to update the corresponding value in that storage area. The specific update calculation model is as follows: ; In the formula, For a specific physical node In the current calculation cycle Write the time-weighted cache value of the hindrance coefficient to the persistent layer; The real-time node impedance coefficient of this specific physical node is calculated based on externally collected data. The time-weighted cache value of the historical stall coefficient retained for this specific physical node in the previous calculation cycle; The value range is configured as [0.6, 0.8], and its specific value is determined based on the positive dynamic mapping of the historical fluctuation variance of the node. The physical meaning is to give a higher trust weight to the recent real observation state.

[0045] This update mechanism ensures that the system can directly read historical cached data during a smooth transition when the external data source is disconnected, maintaining the continuity of input variables for subsequent operations research modules and avoiding null pointer exceptions.

[0046] For the read / write separation control mechanism and high availability disaster recovery backup mechanism at the underlying level of the storage module, those skilled in the art can configure conventional database connection pool components and master-slave synchronization architecture to implement them. The corresponding database concurrent transaction control and locking mechanism are well-known technologies in this field and will not be described in detail here.

[0047] Based on the aforementioned underlying data structure, the business module is responsible for the interactive input of external business data and the encapsulation and dispatch of final business decision instructions. The module's operational logic specifically includes parameter timed synchronization operation S111, graph instantiation control operation S112, and instruction data assembly and distribution operation S113.

[0048] In S111, during the timed parameter synchronization operation, the business module establishes a data interface connection with the enterprise's existing business master database. The system is configured with time-period-based timed triggering tasks that retrieve inventory and consumption indicators for specified product objects at fixed time intervals.

[0049] The core business attributes include, at the implementation level, parameters such as the average daily consumption of goods within a specific time window, the set minimum safety stock level, and the maximum target inventory level to guide replenishment. After reading these parameters through the data interface, the business module performs alignment verification along the time window dimension, formats them, and temporarily stores them in memory for subsequent system calculations.

[0050] S112, in the graph instantiation control operation, the business module sends a graph construction association instruction to the storage module based on the unique identifier of the product object extracted synchronously. This control logic is used to establish data binding between specific business entities and the underlying spatial topology network within the system. The specific lower-level technical features involved in establishing the mapping relationship between the target product object and the corresponding directed acyclic graph of the supply chain are as follows: The business module queries the supplier code of the product object recorded in the relational data unit to locate the system's default main supply path and all alternative supply paths; The business module sends an attribute update request to the graph data unit, using the unique identifier of the product object as a business attribute tag, and injects it into the physical node and directed edge data structure covered by the corresponding path.

[0051] This mapping process enables the underlying physical topology network, which originally had no business attributes, to be instantiated by the business parameters of specific product objects. This allows subsequent computing components to directly use the product object identifier as the query index to initiate graph data traversal and state tracing.

[0052] In step S113, during the instruction data assembly and distribution operation, the business module receives the optimal quota allocation solution set output by the calculation module and executes the instantiation and distribution of the corresponding trade documents. When the system triggers the multi-source quota reallocation mechanism and outputs the optimal quota allocation solution set containing the supply shares of multiple alternative suppliers, the business module parses the values ​​of various decision variables in the solution set.

[0053] For allocation quota variables whose values ​​in the solution set are strictly greater than zero, the business module extracts the corresponding alternative supply node identifiers and assembles multi-source sub-purchase order data packets according to the standard business data exchange format. The specific data structure of the multi-source sub-purchase order data packet is configured as a data set carrier containing supplier identifier codes, purchase quantities converted from allocation quota variables, expected delivery dates calculated by combining expected time offsets, and specified transit logistics node codes.

[0054] The business module pushes multiple independently generated sub-purchase order data packages to the downstream procurement execution system for document entry operations through the communication interface.

[0055] When a gap slack variable with a strictly positive value exists in the optimal quota allocation solution set, it indicates that the total available physical capacity of the entire network has reached its limit and cannot meet the target total demand. For this situation where the network's physical capacity is exceeded, the business module extracts the specific value of the gap slack variable, generates an abnormal state data object with a product code and the quantity of the gap, and pushes the abnormal state data object into the abnormal supply disruption warning queue. The abnormal supply disruption warning queue operates in the system using a message publish-subscribe model. The system uses this queue to trigger manual intervention tasks in the front-end application layer, thereby achieving a collaborative operation between automated order allocation and manual backup under physical capacity limits.

[0056] The storage and retrieval mechanism of parameters in the enterprise's main business database and the filtering and cleaning process of dirty data can be implemented by conventional extraction, transformation and loading tools. The specific data extraction and outlier cleaning logic is a well-known technology in this field and will not be described in detail here.

[0057] To provide the system with real-world physical perception input, the acquisition module is responsible for network request tasks related to the system's perception of the physical logistics status, and executes a data degradation processing model when the external network interface fails. The module's operational logic specifically includes network request and time window control operations S121, real-time node congestion calculation operations S122, and fault-tolerant degradation extraction operations S123.

[0058] In S121, during the network request and time window control operation, the data acquisition module generates periodic data retrieval tasks based on a preset time step. The system initiates network status query requests to external logistics carriers or port information systems and sets strict interface response timeout thresholds.

[0059] The specific technical features of the call to external interface to obtain the dwell time of logistics nodes are as follows: the acquisition module parses the node operation data message returned by the external system, extracts the time difference between the entry scan timestamp and the exit scan timestamp of a specific logistics object at a specific logistics node, and after aligning and compensating the time difference with the system's global clock, converts and records it as the actual dwell time of the physical node in the current calculation cycle.

[0060] S122, in the real-time node congestion calculation operation, when the acquisition module receives normal external interface response data within the set interface response timeout threshold, the system extracts the corresponding historical baseline dwell time parameter from the relational data unit. The acquisition module compares the actual logistics dwell time with the historical baseline dwell time. If the actual logistics dwell time is less than or equal to the historical baseline dwell time, it indicates that the physical node is in normal flow, and the system directly assigns the real-time node congestion coefficient of the node to zero. If the actual logistics dwell time is significantly greater than the historical baseline dwell time, it indicates that the physical node is experiencing logistics congestion. The system calculates the ratio of the relative time overflow difference between the two to the baseline time to generate the real-time node congestion coefficient. To ensure the rigor of the calculation and avoid the abnormal zeroing of the division denominator due to missing baseline time, a very small positive number is introduced into the model as denominator compensation. The specific mathematical calculation model is as follows: ; In the formula, For a specific physical node The real-time node congestion coefficient generated in the current calculation cycle represents the percentage degree to which node congestion deviates from the normal state. The actual logistics dwell time of this specific physical node in the current cycle, as obtained by the acquisition module; The historical baseline residence time of the specific physical node pre-recorded in the relational data unit; This is a preset minimum constant for preventing zero overflow in the system.

[0061] S123, in the fault-tolerant degradation extraction operation, when the external network interface experiences a response timeout or returns a service error code, the acquisition module triggers the data degradation rollback logic.

[0062] The summary is based on the calculation of the stall coefficient according to the fault tolerance and degradation branch. Its specific lower-level feature is as follows: the system terminates the current external network waiting process and instead sends an internal query command to the local persistence layer of the storage module to directly read the stall coefficient time-weighted cache value corresponding to the physical node that was overwritten and saved in the previous cycle.

[0063] The acquisition module uses the historical cached values ​​it reads as the real-time node stall coefficient for the current calculation cycle. Output. This fallback mechanism ensures that even in the event of a temporary failure of the external physical monitoring data source, the system can still output stall status parameters with practical business significance, preventing subsequent matrix mapping operations from generating calculation anomalies due to missing input variables.

[0064] The acquisition module traverses all instantiated physical nodes and performs the above-mentioned normal branch or fault-tolerant degradation branch operations. Finally, it combines the real-time blocking coefficients of all nodes according to the network topology index order to generate a node blocking state column vector, and writes the node blocking state column vector into memory for the calculation module to call.

[0065] For the communication protocol and data packet parsing process of the external network interface, those skilled in the art can configure standard hypertext transfer protocol or remote procedure call protocol to implement it. Its network communication interaction mechanism and data serialization parsing technology are well known in the field and will not be described in detail here.

[0066] As the core hub of this system, the computation module is responsible for cross-domain data analysis, time dimension compensation, spatial topology diagnosis, and multi-constraint optimization. The underlying computational logic of this module specifically includes the wave propagation mapping operation S141, the dynamic trigger reconstruction operation S142, the stagnation tracing circuit breaker verification operation S143, the operations research model construction operation S144, and the constraint solution invocation operation S145.

[0067] S141, in the fluctuation propagation mapping operation, based on the state-space mapping principle in linear algebra, the calculation module constructs a fluctuation propagation matrix representing the hierarchical depth attenuation weights. Let the total number of target core warehouses monitored by the system be... The total number of underlying logistics nodes instantiated in the directed acyclic graph of the supply chain is The calculation module extracts dimensions as follows: The wave propagation matrix is ​​compared with the dimension generated by the acquisition module. The column vector of nodal impediment states is subjected to algebraic multiplication. The expression for the spatial state to time expectation mapping is as follows: ; In the formula, This is a column vector of the comprehensive expected lead time offset with dimensions m×1, output by the operation. The pre-configured dimensions are The fluctuation transmission matrix contains elements that record the attenuation weights when anomalies from upstream physical nodes are transmitted to a specific terminal core warehouse. The dimension is calculated from the actual physical state. The column vector of node stall states. The calculation module calculates the value based on the dimension index of the target core warehouse associated with the current product object, from... The corresponding numerical scalar is extracted as the overall expected lead time offset.

[0068] S142, In the dynamic triggering reconstruction operation, the calculation module substitutes the aforementioned comprehensive expected lead time offset into the business parameter configuration and reconstructs the dynamic triggering threshold in memory. The specific reconstruction calculation formula is as follows: ; In the formula, The dynamic trigger threshold generated for the product object in the current calculation cycle; This represents the average daily consumption of this commodity. This is the basic delivery lead time for the main supply path corresponding to this product. This is the scalar value for the overall expected lead time offset extracted from the matrix multiplication result; This is the set baseline safety stock value for the product object. The calculation module compares the current real-time inventory quantity of the product object with the dynamic trigger threshold. When the real-time inventory quantity drops to or below the dynamic trigger threshold, the calculation module generates a target total demand parameter to trigger the subsequent verification process. The formula for calculating the target total demand parameter is as follows: ; In the formula, The total target demand to trigger the system's replenishment process; This refers to the highest target inventory level parameter for this product object recorded in the relational data unit; This represents the real-time available inventory quantity of the currently retrieved product object.

[0069] In step S143, during the circuit breaker verification operation for source tracing, the calculation module traverses the main supply path currently set for the product object in reverse order, extracting the identifier set of all physical nodes contained in that path. The calculation module reads the pre-configured circuit breaker threshold for each node from the relational data unit. For each node in the path set, the calculation module compares its corresponding real-time node blocking coefficient with the circuit breaker threshold one by one. To avoid biased misjudgment caused by a single extreme value, the system also introduces secondary verification logic for the abnormal duration of the node.

[0070] If the real-time node stall coefficients of all nodes within the path set are less than or equal to their corresponding stall thresholds, and this abnormal state persists for more than the set time window, the calculation module issues a regular procurement instruction to the business module, generating a demand quantity based on the default main path equal to the target total demand. If the real-time node stall coefficient of any node within the path set is strictly greater than its corresponding stall threshold, the calculation module immediately triggers the system's underlying interruption mechanism, intercepting the original document generation process and switching to multi-source quota reallocation.

[0071] S144, in the operations research model construction operation, for congestion events that trigger the underlying interruption mechanism, the calculation module extracts a set of available alternative paths from the multi-source supply configuration table. Assume that after filtering out failed nodes, there are... There are several available alternative supply paths. Based on the principle of linear programming in operations research, the computation module constructs an objective function oriented towards minimizing the total system cost penalty. The specific linear programming objective function is as follows: ; In the formula, The objective function value for operations research optimization; This represents the total number of available alternative supply routes; Index variables for alternative supply paths; To be assigned to the Procurement quota decision variables for alternative supply routes; For the first Individual procurement and logistics cost parameters for each alternative supply route; The introduced gap slack variable is used to characterize the difference between the upper limit of physical production capacity and the total target demand. This is a preset gap penalty constant. Wherein, The value is configured to be much larger than any The numerical values ​​force the solver engine to prioritize available production capacity variables with actual physical significance during the iteration process. .

[0072] S145, in the constraint solver call operation, the calculation module loads the boundary constraint equations converted from business parameters into the linear programming solver library to limit the feasible solution space of the objective function. The specific constraint equations are constructed as follows: Quota conservation constraints including slack variables: ; Physical capacity boundary constraints for a single supply path: ; Nonnegativity mathematical constraints on slack variables: ; Hard constraints on the time limit for performance: ; In the formula, For the record in the relational data unit The upper limit parameter of the physical capacity of the suppliers corresponding to each alternative supply route; For the first Basic lead time parameters for alternative supply routes; The first result obtained through matrix operations The expected time offset corresponding to each alternative supply route; This refers to the cutoff deadline time parameter calculated by the system based on the minimum safety stock. The calculation module iteratively solves the above model by calling the simplex method or interior point method algorithm module. The system finally outputs the optimal quota allocation solution set that minimizes the objective function value. This solution set contains various... Specific numerical results and The specific numerical results are then sent back to the business module by the calculation module to generate subsequent supply interruption early warning queues and business documents.

[0073] For the simplex method core algorithm or interior point method iterative calculation mechanism used in the underlying linear programming solution library of the operations research model, those skilled in the art can call existing commercial operations research solver components to implement it. The underlying mathematical iterative solution logic is a well-known technology in this field and will not be elaborated here.

[0074] Specific test example: This test aims to verify the correctness of the underlying soft-constraint operations research model and its robustness against overflow under the dual extreme conditions of sudden congestion at logistics nodes and insufficient alternative capacity across the entire network. Specific test sequences include: minimum constant compensation to prevent division by zero test, dynamic threshold time dimension mapping test, and multi-source quota allocation solution test with gap relaxation variables.

[0075] Test steps: Step 1: Test Environment Initialization and Boundary Parameter Injection. Inject simulated business parameters, such as average daily consumption, into the system's relational data unit. Basic safety stock The highest target inventory level The main supply route basic delivery time Minimum constant for preventing zero overflow Inject alternative supplier network parameters: Alternative A's maximum capacity is 5000 units, Alternative B's maximum capacity is 3000 units, and there is a gap penalty constant. .

[0076] Step 2: Network State Simulation and Impedance Sensing Calculation. Inject the current period's external interface data into the acquisition module to simulate the data transmission back: the actual dwell time of the Wuhan transit node. The calculation module is triggered to call the formula: .

[0077] Step 3: Threshold Reconstruction and Demand-Triggered Interception. Inject the system's preset fluctuation propagation attenuation weight of 2.0. The calculation module performs matrix mapping, converting the hysteresis coefficient into a lead time offset. And substitute it into the formula to calculate the dynamic trigger threshold for the current cycle. Then, the current real-time available inventory is injected into the system. This triggers the underlying interception and verification, and calculates the total target demand. .

[0078] Step 4: For the circuit breaker mechanism triggered by the main supply path exceeding the obstruction coefficient, the calculation module calls the system's built-in linear programming solver engine, loads the capacity boundary and quota conservation constraints, and iteratively calculates the soft-constraint model with the goal of minimizing total cost, outputting the optimal decision solution set. The system background log independently captures all intermediate variables and the final solution results, and formats the output as the following test data table.

[0079] The tabular data generated by the test steps is based on the execution of the above test steps. The system background automatically captures and generates the calculation process monitoring data. For details, please refer to Tables 1 and 2: Table 1 Table 2 Summarize: In summary, this test case successfully simulated the dual extreme conditions of a sudden severe congestion in logistics transit within the commercial supply chain, coupled with the depletion of all available production capacity across the network. It also fully verified the technological advancement and system robustness of the data processing method of this invention in dealing with complex crises.

[0080] First, in the perception and interception phase, this test verified the system's proactive defense capabilities against supply disruption risks. Faced with sudden congestion at logistics nodes, the system did not passively wait for inventory depletion alerts, but instead used underlying algorithms to accurately quantify the obstruction state of 0.7496 and converted it across dimensions into an expected time compensation of 1.50 days. Based on this, the system adaptively raised the safety defense line (dynamic threshold) to 11,500 units. As shown in Table 1 and Appendix Figure 3 This dynamic compensation mechanism, as further evidenced, enables the system to proactively trigger the interception mechanism and establish a replenishment demand of 9,000 units even when the current available inventory (11,000 units) is still at a relatively high level. This demonstrates the accuracy of the proposed solution in spatiotemporal mapping compensation, effectively eliminating the lag and blind spots inherent in traditional static inventory management.

[0081] Secondly, during the operation and redistribution phase, this test verified the overflow resistance and flexible decision-making value of the system's underlying core algorithm. Faced with 9000 generated urgent demands, the system encountered a deadlock where the total available physical capacity of the entire network was only 8000 units (the sum of alternatives A and B). Under such an absolute constraint of demand exceeding supply, traditional linear programming solvers are highly prone to throwing unsolvable anomalies, leading to system crashes and deadlocks.

[0082] However, this test case conclusively demonstrates that, thanks to the soft constraint model and gap penalty mechanism constructed in this invention, the computing module not only maintained stable operation of computing power but also automatically completed full-load production scheduling of physical capacity (5000 units allocated to Alternate A and 3000 units allocated to Alternate B) guided by global cost optimization. Meanwhile, as shown in Table 2 and Appendix... Figure 4 As shown, the system successfully and accurately separated 1,000 demands that exceeded physical limits into slack variables and transformed them into an independent anomaly warning queue.

[0083] Therefore, this test case fully demonstrates that the system has successfully completed the entire process from anomaly physical perception and dynamic defense reconstruction to extreme order splitting. Without causing computing power failure, the system achieved coordinated operation of automatic optimal production scheduling and anomaly pre-emptive isolation, providing valuable buffer time for manual intervention and demonstrating excellent logical completeness and high engineering feasibility.

Claims

1. A digital management and data analysis system for commercial business information, characterized in that, include: The storage module is used to store the directed acyclic graph of the supply chain, the basic business attributes of the product objects, and the historical baseline time parameters. The data acquisition module is used to obtain the dwell time of logistics nodes and calculate the real-time node stall coefficient, forming a column vector of node stall status. The business module is used to obtain the real-time inventory of the product object, construct the directed acyclic graph of the supply chain using the basic business attributes, and generate multi-source sub-purchase order data packets and abnormal supply interruption early warning queues based on the optimal quota allocation solution set. The calculation module is used to extract the comprehensive expected lead time offset by operating the fluctuation transmission matrix and the column vector of the node stagnation state. At the same time, the comprehensive expected lead time offset is used to reconstruct the dynamic trigger threshold. When the real-time inventory drops to the dynamic trigger threshold, the target total demand is generated and the interruption mechanism is triggered. Finally, a soft-constraint operations research model with gap relaxation variables is constructed for the interruption mechanism, and the optimal quota allocation solution set is output.

2. The digital management data analysis system for commercial business information according to claim 1, characterized in that, The storage module includes graph data units and relational data units; The graph data unit is used to store the directed acyclic graph of the supply chain. The node set of the graph data unit corresponds to a fixed physical node, and the directed edge corresponds to a directional transportation link. The business graph association mapping is established using the unique identifier of the product as the root node attribute. The relational data unit is used to store the basic business attributes, which include average daily consumption, basic delivery lead time of the main supply path, basic safety stock value, and physical capacity limit of alternative suppliers.

3. The digital management and data analysis system for commercial business information according to claim 1, characterized in that, The calculation of the real-time node stall coefficient specifically includes: The actual logistics dwell time of the extracted logistics node is compared with the historical benchmark dwell time in the historical benchmark time parameter; If the actual logistics dwell time is less than or equal to the historical baseline dwell time, the real-time node stagnation coefficient of the logistics node will be set to zero. If the actual logistics dwell time is greater than the historical benchmark dwell time, calculate the relative time overflow difference between the actual logistics dwell time and the historical benchmark dwell time, and then calculate the ratio of the relative time overflow difference to the historical benchmark dwell time with zero overflow compensation to obtain the real-time node stall coefficient.

4. The digital management data analysis system for commercial business information according to claim 1, characterized in that, The acquisition module is also used to perform fault-tolerant degradation extraction operations, specifically including: When the external network interface experiences a response timeout or returns a service error code, the acquisition module terminates the external network waiting process, reads the time-weighted cache value of the stall coefficient that was overwritten and saved in the previous cycle from the storage module, and uses the time-weighted cache value of the stall coefficient as the real-time node stall coefficient for the current calculation cycle.

5. The digital management data analysis system for commercial business information according to claim 4, characterized in that, The storage module is also used to perform time-series-based overwrite update tasks to generate the time-weighted cache values ​​of the hysteresis coefficient, specifically including: Based on the time series, and using the system's preset trust weight, the currently calculated real-time node stall coefficient is fused with the historical cache value retained in the previous period to obtain the stall coefficient time-weighted cache value written in the current period.

6. The digital management data analysis system for commercial business information according to claim 1, characterized in that, Extract the comprehensive expected lead time offset, and simultaneously reconstruct the dynamic trigger threshold using the comprehensive expected lead time offset, specifically for: Perform algebraic multiplication on the fluctuation transmission matrix and the column vector of the node stagnation state, and extract the numerical scalar of the dimension corresponding to the target core warehouse as the comprehensive expected lead time offset; The basic delivery lead time is compensated by superimposing the comprehensive expected lead time offset, and the expected consumption is calculated by combining the average daily consumption. The expected consumption is superimposed with the basic safety stock value to reconstruct and generate the dynamic trigger threshold. When the real-time inventory drops to or below the dynamic trigger threshold, the difference between the highest target inventory level parameter and the real-time available inventory quantity is calculated as the target total demand.

7. The digital management data analysis system for commercial business information according to claim 1, characterized in that, The aforementioned interruption triggering mechanism is specifically implemented by performing path circuit breaker verification after generating the target total demand, specifically including: Traverse the node identifier set in reverse along the set main supply path, and compare the real-time node blocking coefficient of each node with the preset blocking and circuit breaker threshold. If the real-time node stall coefficients of all nodes do not exceed the corresponding stall circuit breaker threshold, then a regular procurement instruction will be issued. If the real-time node stall coefficient of any node is greater than the stall circuit breaker threshold corresponding to that node, the underlying interruption mechanism of the system is triggered, the original document generation process is intercepted and the process is switched to building the soft constraint operation model.

8. The digital management data analysis system for commercial business information according to claim 1, characterized in that, The soft-constraint operations research model is used to establish an objective function oriented towards minimizing the total system cost. This objective function comprises two parts: a total cost. The first part is the sum of the products of the individual procurement logistics comprehensive cost parameters corresponding to all available alternative supply paths and the procurement quota decision variables allocated to that path; The second part is the product of the gap relaxation variable and the preset gap penalty constant; The gap slack variable is used to characterize the difference between the upper limit of the total physical capacity of the entire network and the total target demand, and the gap penalty constant is greater than any of the individual procurement and logistics comprehensive cost parameters.

9. The digital management data analysis system for commercial business information according to claim 8, characterized in that, The soft-constraint operations research model also includes a set of boundary constraint equations that limit the solution space, specifically including: Quota conservation constraint: The sum of all procurement quota decision variables and the accumulated value of the gap slack variable shall be equal to the target total demand. Physical capacity boundary constraints: limiting the procurement quota decision variable for a single alternative supply path to between zero and the corresponding physical capacity limit; Non-negativity constraint for slack variables: Requires the value of the gap slack variables to be no less than zero; Hard constraint on delivery time: When the procurement quota decision variable is greater than zero, the sum of the offset of the basic delivery lead time and the comprehensive expected lead time shall not be greater than the supply cutoff deadline time parameter.

10. The digital management data analysis system for commercial business information according to claim 1, characterized in that, The optimal quota allocation solution set is used to generate the multi-source sub-purchase order data package and the abnormal supply interruption early warning queue, specifically including: For the procurement quota decision variables with values ​​greater than zero, extract the corresponding alternative supply node identifiers and assemble the multi-source sub-procurement order data package containing the alternative supply node identifiers and the procurement quantity; For the gap relaxation variable with a value greater than zero, extract the specific value, generate an abnormal state data object with a product code and gap quantity, and push the abnormal state data object into the abnormal supply interruption early warning queue based on the message publish-subscribe model.

Images