An integrated computing intelligent management and analysis system for enterprise resource data

Abstract

This invention relates to the field of industrial data processing technology and discloses an intelligent management and analysis system for integrated computing of enterprise resource data. The system includes a central processing server, a data storage unit, an interface interaction unit, and a terminal execution unit. The central processing server is equipped with topology construction, priority calculation, status monitoring, and resource arbitration modules. The system first constructs a dynamic dependency topology graph reflecting material dependencies across the entire chain and calculates the dynamic priority score of each task node. Simultaneously, it monitors the operational status of materials in the physical space and quantifies it into rollback cost values ​​based on preset mapping relationships. When a resource conflict is detected, the resource arbitration module calculates the priority difference between competing orders and compares it with the rollback cost value based on a damped inequality to perform a preemption determination. This invention introduces physical execution cost as scheduling damping, avoiding system jitter caused by frequent resource switching and improving the stability of production scheduling.

Description

Technical Field

[0001] This invention relates to the field of industrial data processing technology, specifically to an intelligent management and analysis system for the integrated calculation of enterprise resource data. Background Technology

[0002] In modern discrete manufacturing enterprises, Enterprise Resource Planning (ERP) systems and Manufacturing Execution Systems (MES) are core tools for managing production planning, material inventory, and execution processes. Enterprises typically generate corresponding procurement plans and production work orders by decomposing the Bill of Materials (BOM) based on the demand of sales orders, and use scheduling algorithms to allocate limited material, equipment capacity, and logistics resources to various task nodes.

[0003] With the widespread adoption of customized production models, concurrent processing of multi-variety, small-batch orders has become the norm, leading to increasingly fierce competition for resources in the production process. Existing resource management systems typically employ rule-based static priority scheduling strategies, such as determining the execution order of tasks solely based on contract amount, customer level, or order placement time. This traditional scheduling approach often assumes an ideal and static production environment, ignoring the complex real-time dependencies between tasks and the dynamic changes in the production site. In actual operation, when high-priority urgent orders are inserted, the scheduling algorithm usually adopts a greedy strategy, that is, as soon as it detects that resources are occupied by low-priority orders, it immediately triggers preemption logic to reallocate resources.

[0004] However, preemption strategies based purely on digital logic have significant limitations in the physical world. Existing scheduling systems generally lack the ability to perceive the real-time operational status of materials in the physical space, and cannot distinguish between different stages such as "logical pre-allocation," "logistics in transit," and "actual processing." From an algorithmic perspective, a material that has not yet been moved and a material being processed on a CNC machine tool may be considered the same inventory record. If the system forcibly executes resource preemption based solely on the difference in priority values, it is highly likely that tasks that have already entered the critical processing stage or are in the middle of the logistics process will be interrupted. Such interruptions not only cause the scrapping of semi-finished products, increased equipment changeover downtime, and increased reverse logistics costs, but also, in high-concurrency scenarios where frequent resource conflicts occur, trigger repeated switching and oscillations of task queues, leading to a decrease in the overall stability of the production system, and actual output efficiency falling below the theoretical planning value. Therefore, how to ensure the response speed of high-priority tasks while taking into account the continuity of physical operations and the stability of system operation is a technical challenge that urgently needs to be solved in the field of enterprise resource data integration and computing. Summary of the Invention

[0005] The technical problem addressed by this invention is that existing Enterprise Resource Planning (ERP) or Manufacturing Execution Systems (MES), when handling multiple concurrent orders and resource scheduling, typically rely solely on static priorities or single time parameters for scheduling, lacking awareness and consideration of the real-time operational status of materials in the physical space. When high-priority orders need to preempt resources, existing systems often ignore the current physical execution progress of the preempted task, causing tasks that have already entered critical processing or logistics stages to be forcibly interrupted. This leads to problems such as material scrapping, increased equipment wear and tear, and logistics system instability, reducing the overall stability and execution efficiency of the production system.

[0006] To address the aforementioned problems, the first aspect of this invention provides an integrated intelligent management and analysis system for enterprise resource data. This system includes a central processing server, a data storage unit, an interface interaction unit, and a terminal execution unit. The central processing server establishes bidirectional data communication connections with both the data storage unit and the interface interaction unit. The interface interaction unit is configured to connect to an external management system, and the terminal execution unit is configured to execute physical operation instructions.

[0007] The central processing server is internally configured with a topology building module, a priority calculation module, a status monitoring module, and a resource arbitration module.

[0008] The topology construction module is configured to decompose sales orders into multi-level sub-task nodes based on bill of materials and process route data, and construct a dynamic dependency topology graph reflecting the dependencies between material entities and processing tasks. This module does not process a single static list, but rather uses a graph theory model to transform discrete orders into a computational graph structure with pre-existing dependency logic.

[0009] The priority calculation module is configured to calculate the dynamic priority score of an order for a specific resource based on the dynamic dependency topology graph of the entire chain. This calculation process combines time and business value dimensions to quantify the urgency of tasks.

[0010] The status monitoring module is configured to acquire the physical operation status of materials occupied by pre-allocated orders in the physical space and convert the physical operation status into corresponding rollback cost values. This module establishes a mapping between digital logical status and physical execution actions, quantifying the time cost, material loss risk, and system operation cost required to cancel a task into calculable values.

[0011] The resource arbitration module is configured to, upon detecting a resource conflict, identify the pre-allocated order occupying the target resource and the request order applying for the target resource, calculate the difference in dynamic priority scores between the request order and the pre-allocated order, and compare this difference with the rollback cost of the pre-allocated order to determine whether to execute resource preemption. This module introduces a cost-based damping mechanism, triggering resource reallocation only when the value gained from preemption covers the physical rollback cost.

[0012] Furthermore, in the topology construction phase, the topology construction module reads the sales order data to be processed through the interface interaction unit, and decomposes the sales order into the smallest granularity of the work unit node based on the pre-stored bill of materials data and process route data; it establishes directed edges between work unit nodes that reflect the pre-process dependencies, and performs loop detection to ensure logical correctness; it establishes cross-order material reverse indexes for work unit nodes containing the same material code, generates a full-link dynamic dependency topology graph, and stores it in the data storage unit.

[0013] Furthermore, in the priority calculation stage, the priority calculation module performs a forward topology sorting traversal of the dynamic dependency topology graph of the entire link, and calculates the earliest start time of each node by combining standard working hour data and logistics turnaround time; it performs a reverse topology traversal of the dynamic dependency topology graph of the entire link according to the order delivery deadline, and calculates the latest start time of each node; it calculates the difference between the latest start time and the earliest start time to obtain the time slack, and uses the time slack as a parameter for calculating the dynamic priority score.

[0014] Furthermore, the priority calculation module employs a multi-dimensional quantification model to calculate the dynamic priority score. This model calculates the order's delay loss coefficient based on contract amount, penalty clauses, and customer credit rating; it constructs an exponential decay model based on time slack, ensuring that the dynamic priority score increases non-linearly as time slack decreases; simultaneously, it introduces a hierarchical decay factor to adjust the dynamic priority score according to the node's hierarchical depth in the bill of materials structure tree, thereby ensuring that nodes on the critical path receive higher resource competition weights.

[0015] Furthermore, in the status monitoring phase, the status monitoring module defines discretized physical operation statuses, which include at least: virtual pre-occupancy status, indicating that the node has obtained the qualification for resource allocation but has not yet taken any action at the physical level; physical lock status, indicating that the execution command has been issued and the material has been locked by the physical equipment; operation execution status, indicating that the material is undergoing displacement or shape change; and irreversible processing status, indicating that the material has undergone a permanent change in physical shape or has been shipped.

[0016] Furthermore, the status monitoring module calculates the rollback cost value based on a preset mapping relationship: when the physical operation status is virtual pre-occupancy, the rollback cost value is set to zero, allowing lossless preemption; when the physical operation status is physically locked, the rollback cost value is calculated based on reverse logistics time and operation instruction cancellation cost; when the physical operation status is operation execution, a processing interruption penalty coefficient is introduced to increase the rollback cost value and suppress interruption of the processing task; when the physical operation status is irreversible processing, the rollback cost value is set to the system maximum threshold to prevent preemption.

[0017] Furthermore, during the resource arbitration phase, the resource arbitration module executes an arbitration decision based on the damped inequality. Specifically, it calculates the priority difference between the dynamic priority score of the requesting order and the dynamic priority score of the pre-allocated order; it then determines whether the priority difference is greater than the sum of the rollback cost and the preset system stability threshold. If the result is yes, resource preemption is deemed successful; if the result is no, the current allocation state is maintained, and the requesting order is added to the blocking waiting queue.

[0018] Furthermore, once the resource preemption is determined to be successful, the resource arbitration module generates a reverse logistics instruction and sends it to the terminal execution unit through the interface interaction unit to withdraw the physical operation task of the pre-allocated order; releases the lock mark of the pre-allocated order on the target resource in the database and relocks the target resource to the requesting order; rolls back the status of the pre-allocated order and reinserts it into the waiting queue.

[0019] Furthermore, the system performs closed-loop feedback and dynamic reordering. When it is determined that maintaining the current allocation state will cause the requested order to enter the blocking waiting queue, the priority calculation module recalculates the time parameters of the affected nodes in the full-link dynamic dependency topology graph based on the changed resource allocation results. If the recalculated estimated completion time exceeds the promised delivery date of the order, the delay loss coefficient of the order is automatically corrected to improve its dynamic priority score in the next round of calculation.

[0020] The second aspect of this invention provides a method for integrated calculation, intelligent management, and analysis of enterprise resource data. This method is based on the aforementioned system and includes: constructing a dynamic dependency topology diagram of the entire chain reflecting material dependencies; calculating the time-series parameters and dynamic priority scores of each node; monitoring the physical operation status of materials in real time and quantifying it into rollback cost values; and, when resource conflicts occur, performing resource arbitration decisions based on damping inequalities based on the comparison results of priority differences and rollback cost values.

[0021] This invention provides an intelligent management and analysis system for the integrated calculation of enterprise resource data. It has the following beneficial effects: 1. This invention constructs a dynamic dependency topology graph across the entire chain, transforming the traditional static bill of materials into a computable directed acyclic graph model. Combined with the calculation of time slack, it can accurately identify critical and non-critical paths in the production process, providing a data foundation for the parallelization and refinement of resource scheduling.

[0022] 2. This invention introduces a physical state awareness and rollback cost quantification mechanism, overcoming the shortcomings of traditional scheduling algorithms that only focus on logical priority while ignoring physical execution costs. By defining discrete states such as virtual pre-emption, physical locking, job execution, and irreversible processing, the system can accurately assess the overall cost of interrupting the current task.

[0023] 3. This invention employs a resource arbitration strategy based on damped inequality, incorporating a rollback cost positively correlated with physical state as a hysteresis term in the priority comparison. This mechanism effectively filters out frequent preemption requests with low returns, avoids repeated equipment start-ups and shutdowns and logistics path conflicts caused by minor priority differences, and significantly improves the operational stability of the production system.

[0024] 4. This invention establishes a closed-loop feedback mechanism based on delay risk. When a task is forced to wait due to the inability to seize resources, the system can automatically adjust its delay loss coefficient, so that the priority of the task increases non-linearly as the delivery deadline approaches. This ensures that urgent orders can eventually obtain the required resources, and achieves a dynamic balance between the overall efficiency of the system and local urgent needs. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the structure of an intelligent management and analysis system for integrated computing of enterprise resource data according to an embodiment of the present invention; Figure 2 This is a flowchart of an intelligent management and analysis method for the integrated calculation of enterprise resource data according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the full-link dynamic dependency topology graph construction process according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the time parameter calculation and critical path analysis process according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the dynamic priority quantization calculation logic according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the physical operation state discretization mapping and state machine transition according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the rollback cost function curve and the state damping threshold setting according to an embodiment of the present invention; Figure 8This is a flowchart of the resource conflict identification and lag comparison preemption logic according to an embodiment of the present invention; Figure 9 This is a schematic diagram of the reverse control flow and dynamic scheduling closed-loop feedback according to an embodiment of the present invention; Figure 10 This is a schematic diagram comparing the performance of the embodiments of the present invention and traditional algorithms in terms of task switching frequency and effective device utilization. Figure 11 This is a schematic diagram comparing the performance of the embodiments of the present invention with that of traditional algorithms in terms of overall delivery on-time rate and material loss cost.

[0026] Among them, 100 is the central processing server; 110 is the topology construction module; 120 is the priority calculation module; 130 is the status monitoring module; 140 is the resource arbitration module; 200 is the data storage unit; 300 is the interface interaction unit; and 400 is the terminal execution unit. Detailed Implementation

[0027] 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.

[0028] See attached document Figure 1 This invention provides an integrated intelligent management and analysis system for enterprise resource data. The system includes a central processing server 100, a data storage unit 200, an interface interaction unit 300, and a terminal execution unit 400. The central processing server 100 establishes bidirectional data communication connections with the data storage unit 200 and the interface interaction unit 300. The interface interaction unit 300 connects to external warehouse management systems, manufacturing execution systems, and enterprise resource planning systems. The terminal execution unit 400 includes a handheld PDA terminal, an automated guided vehicle (AGV) dispatcher, and an industrial PC terminal.

[0029] The central processing server 100 is internally configured with a topology building module 110, a priority calculation module 120, a status monitoring module 130, and a resource arbitration module 140. The topology building module 110 reads order data and constructs a directed acyclic graph reflecting material dependencies. The priority calculation module 120 calculates the dynamic priority value of an order for a specific resource. The status monitoring module 130 acquires the operational status of materials in the physical space. The resource arbitration module 140 performs resource conflict judgment and generates scheduling instructions.

[0030] See attached document Figure 2This invention provides a method for integrated computing, intelligent management, and analysis of enterprise resource data, comprising the following steps: S10, construct a dynamic dependency topology graph across the entire link; The topology building module 110 extracts undelivered sales order data from the enterprise resource planning system through the interface interaction unit 300. Based on the bill of materials and process route data, the sales order is decomposed into multi-level sub-task nodes. The topology building module 110 establishes dependency connections between nodes, generates a directed acyclic graph data model containing material entities and processing tasks, and stores the model in the data storage unit 200.

[0031] S20, calculates timing parameters and dynamic priorities; The priority calculation module 120 performs a topological sorting traversal of the directed acyclic graph and, combined with standard working time data, calculates the earliest start time, latest start time, and relaxation time of each node. Based on the relaxation time and a preset order delay loss weight, the priority calculation module 120 quantifies the dynamic priority score of an order for a specific material node.

[0032] S30 detects resource conflicts and initiates competition; When the available inventory of a specific material cannot meet the demand of all related orders, the resource arbitration module 140 identifies the pre-allocated orders currently occupying the inventory of that material, as well as the request orders that require the material but have not been allocated. The resource arbitration module 140 compares the dynamic priority scores of the request orders and the pre-allocated orders. If the score of the request order is higher than that of the pre-allocated order, the process proceeds to the physical status assessment stage.

[0033] S40, senses the physical operation status and quantifies the rollback cost; The status monitoring module 130 initiates a query request to the warehouse management system or manufacturing execution system through the interface interaction unit 300 to obtain the current physical operation status of the materials occupied by the pre-allocated order. The physical operation status includes discrete states such as virtual pre-occupancy, instruction issuance, physical locking, pre-processing, and irreversible processing. The resource arbitration module 140 converts the obtained physical operation status into corresponding rollback cost values ​​according to a preset mapping relationship.

[0034] S50 executes arbitration decisions based on damped inequalities; The resource arbitration module 140 calculates the priority difference between the request order and the pre-allocated order and compares this difference with the rollback cost value. The resource arbitration module 140 determines that resource preemption is successful only if the priority difference is greater than the rollback cost value; otherwise, it determines to maintain the current allocation status.

[0035] S60 performs closed-loop feedback and dynamic rearrangement; If resource preemption is deemed valid, the resource arbitration module 140 generates a reverse logistics instruction and sends it to the terminal execution unit 400 via the interface interaction unit 300, withdrawing the original physical operation task and re-locking the material resources to the requesting order. If the system determines to maintain the status quo, it adds the requesting order to the out-of-stock queue. The priority calculation module 120 recalculates the relevant order time parameters based on the changed resource allocation results.

[0036] See attached document Figure 3 In this embodiment, the topology construction module 110 functions to transform the static, hierarchical bill of materials data in the Enterprise Resource Planning (ERP) system into a graph theory model that can be traversed and computed by computer algorithms. Through this transformation, the system can abstract the complex production dependencies in the physical world into a mathematically directed acyclic graph (DAG), thereby providing a standardized data foundation for subsequent parallel computing and path optimization. Specifically, this includes the following steps: S110 performs raw data extraction and discretization processing; The topology building module 110 reads the sales order data to be processed from the ERP system through the interface interaction unit 300. In this embodiment, the system does not directly process the tree-like BOM structure, but flattens it into a linear task sequence.

[0037] For each sales order, the system breaks it down into the smallest granular work unit based on the pre-stored Bill of Materials (BOM) data and process route data. Specifically, the system traverses each level of the BOM tree, extracts the material code, required quantity, and parent item ID, and simultaneously associates the standard working hours, preparation time, and material attribute fields (self-made / purchased / outsourced) from the process route table.

[0038] Establishing connections to the ERP system database, executing SQL queries, and cleaning basic data are all conventional techniques well-known to those skilled in the art, and will not be elaborated upon here.

[0039] S120, Construct the node set of a directed acyclic graph; The system maps each job unit parsed above to a node in a graph structure. In this embodiment, a node represents not only physical material, but also the processing or transfer task of that material under a specific order.

[0040] For including The order set of each work unit, and the node set is defined as follows. Each node A set of attribute data tuples is assigned, which are represented as .in, This is the unique SKU code for the material in the ERP system; Identify whether the node is a production task node, a procurement and receiving node, or an outsourced processing node; This indicates the target quantity required to complete the task at this node, which is determined by multiplying the quantity of the top-level order by the BOM level usage coefficient; Used to identify the top-level sales order to which this node belongs; This indicates the node's hierarchical depth within the Bill of Materials (BOM) tree structure, and is used for subsequent calculations of hierarchical decay weights.

[0041] By introducing By leveraging attributes, the system can distinguish material nodes with the same physical specifications but belonging to different delivery targets in scenarios with multiple concurrent orders, thereby enabling independent tracking of the entire lifecycle of a single order.

[0042] S130, establish dependency edges between nodes; Based on the parent-child relationships in the BOM structure and the pre-process relationships in the process route, the system establishes directed edges between nodes. Define the edge set as Among them, ordered pairs Represents a node It is a node It is a direct prerequisite. In a physical sense, this means that it must be completed first. Only after the corresponding material production or procurement can it begin. The corresponding assembly or processing tasks.

[0043] To ensure the logical correctness of the graph structure, the system constructs edges... Simultaneously, a depth-first search (DFS) algorithm is used to detect loops in the generated graph structure. If a loop is detected from a node... After starting and traversing several edges, can you return again? If the closed-loop path is not found, the system will determine that there is a logical error in the BOM structure (e.g., part A is composed of part B, and part B is composed of part A), and output an exception log to block the process. After verification, a full-link dynamic dependency topology diagram for a single calculation cycle is generated. .

[0044] S140, perform graph fusion and index construction for shared materials across multiple orders; When handling multiple concurrent sales orders, the topology building module 110 maintains the logical independence of each order subgraph while establishing a cross-order material index. In this embodiment, a hash mapping table is used to construct the material inverse index.

[0045] The system traverses all node sets. Identify all For the same node, the material code ID is used as the key, and a list of all node numbers containing that material code is stored as the value in the index table.

[0046] This index table records specific material codes in the diagram. All location coordinates appearing in the table can be indexed using this structure, allowing the subsequent resource arbitration module to... This approach quickly locates all competing nodes vying for the same physical inventory within a short time complexity, without traversing the entire graph. This method ensures both the independence of the order perspective and supports global coordination from a resource perspective.

[0047] See attached document Figure 4 In this embodiment, the priority calculation module 120, based on the aforementioned constructed end-to-end dynamic dependency topology, determines the time window for each task node through forward and reverse traversal algorithms. The core logic of this process lies in utilizing the principles of the Critical Path Method (CPM) to combine static process data with dynamic delivery requirements, calculating the fluctuation range of each task node on the time axis. Through this calculation, the system can identify which tasks are rigid constraints (must be executed immediately) and which tasks have flexible space (can be preempted or postponed), thus providing a quantifiable time basis for subsequent resource arbitration. Specifically, it includes the following steps: S210, Perform a forward topological traversal to calculate the earliest start time; Priority calculation module 120 first processes the directed acyclic graph. Perform a topological sort to generate a linear sequence of nodes, ensuring that for any node in the sequence, all its predecessor nodes are placed before that node.

[0048] In this embodiment, the system calculates the earliest start time of all starting nodes with an in-degree of zero. Initialize to the current system time (for orders already placed) or the estimated arrival time of materials (for orders in transit). Then, the system calculates the earliest start time for each subsequent node according to the topology sequence. This calculation process simulates the earliest time a task can begin under ideal conditions of unlimited resources and no delays. For any node... Its earliest start time Depends on all its direct predecessor nodes The completion time. The calculation formula is as follows: ; in, Represents a node The standard operating cycle, which is obtained from the process route table stored in the ERP system, is specifically the sum of processing time and preparation time; Indicates from node Flow to node The required logistics turnaround time is determined by dividing the preset workshop logistics distance matrix by the average moving speed of the AGV; The function is used to select the longest-running path among all preceding paths to ensure that all dependency conditions are met.

[0049] S220, Perform a reverse topological traversal to calculate the latest start time; After completing the forward traversal, the priority calculation module 120 performs reverse derivation based on the final delivery deadline of the order. In this embodiment, the system sets the latest completion time of the terminating node with an out-degree of zero to the delivery date stipulated in the sales order contract. Next, the system calculates the latest start time for each node in reverse order of the topological sequence. The latest start time represents the time before the final delivery date is reached. The last moment that the task at this node must begin, assuming a delay.

[0050] For any node Its latest start time From all its direct successor nodes The constraints determine the calculation formula as follows: ; in, The function is used to select the most constrained path among all subsequent paths, ensuring that the timely completion of this node will not cause any subsequent nodes to be delayed.

[0051] The specific implementation of the topological sorting algorithm and the stack operations in computer memory are well-known technologies to those skilled in the art, and will not be elaborated here.

[0052] S230, calculate time slack and define the critical path; Based on the earliest and latest start times calculated above, the priority calculation module 120 calculates the time slack for each node. Time slack is a direct indicator of task urgency, quantifying the allowable time margin for delay at a given point without affecting the overall order's final delivery time. The calculation formula is: ; In this embodiment, time relaxation The numerical value has the following physical meaning: when When this occurs, it indicates that the node is on the critical path, there is no time buffer, and resources must be acquired immediately for production. when When the value is 0, it indicates that the node has time buffer resources. The larger the value, the higher the possibility of the resource being preempted. when When this occurs, it indicates that the order corresponding to that node is in a delayed state, and the system will raise the priority of such nodes to the highest level for processing.

[0053] The system will calculate The numerical values ​​are stored in the node attributes and serve as the core parameters for the subsequent resource arbitration module to determine the priority of resource allocation.

[0054] See attached document Figure 5 In this embodiment, the priority calculation module 120 does not rely solely on a single time parameter for scheduling, but introduces a business value dimension to construct a multi-dimensional dynamic priority quantification model. This model, based on the principle of multi-objective optimization, aims to resolve the contradiction in traditional scheduling algorithms where "high-value orders are shelved due to non-urgent nature" or "urgent low-value orders occupy core resources." By constructing a unified mathematical mapping function, the system integrates discrete contract amounts, customer credit ratings, and continuous time slack parameters into a dimensionless scalar value. This value directly reflects the rational weight of a specific task node acquiring competitive resources at the current moment. The specific calculation process includes the following steps: S240, construct a composite calculation model for the delay loss coefficient; The system first calculates the potential delay loss factor for each sales order. This coefficient reflects the overall losses that the company might suffer if the order is defaulted on. In this embodiment, the system extracts contract amount, penalty clauses, and customer credit rating data from the ERP and CRM systems, and calculates the coefficient using a weighted summation method. The calculation formula is as follows: ; in, Indicates the first One sales order; This indicates the total contract amount for the order; This indicates the daily penalty for delay or the amount of one-time compensation stipulated in the contract; Customer credit rating is assigned, which is generated by the CRM system based on the customer's historical transaction amount and payment cycle, with the value range normalized to [1, 10]. , , These are the weighting coefficients for the three dimensions mentioned above, satisfying the constraint α+β+γ=1. In this embodiment, the values ​​of the weighting coefficients are set by the company's management based on its business strategy, for example, setting α=0.4, β=0.4, and γ=0.2 to balance cash flow risk and customer relationship maintenance risk.

[0055] S250 calculates the dynamic priority score at the node level; Obtaining the order-level loss coefficient and the node-level time relaxation calculated in the aforementioned steps Then, the system uses a dynamic priority quantization formula to calculate the priority of each task node. Final score The formula is designed based on an exponential decay model using the natural logarithm. Its technical effect is that as the time slack decreases (i.e., the task becomes more urgent), the priority increases rapidly and non-linearly, thus forming absolute control over resources at the critical point. The specific mathematical model is as follows: ; in: Represents a node The dynamic priority score indicates that the higher the value, the higher the priority for acquiring resources.

[0056] This represents the delay loss coefficient of the order to which the node belongs, inherited from the calculation result of step S240.

[0057] This represents the temporal slack of a node. When... When it is positive, the exponent term Less than 1, lower the score; when When the value approaches 0 or becomes negative (after being postponed), the value of the exponent increases rapidly, thus significantly increasing its priority.

[0058] This represents the time sensitivity factor, used to eliminate the influence of time unit dimensions and control the rate at which priority changes with relaxation. In this embodiment, The value is a constant (e.g., 10). -5 (When the time unit is milliseconds), it is used to map huge time values ​​to a reasonable exponential range to prevent computational overflow.

[0059] Represents a node The hierarchy depth in the BOM structure tree. The finished product level is defined as 0, and the semi-finished product and raw material levels increase sequentially.

[0060] This represents the level decay factor, with a value range of (0,1). This parameter is used to fine-tune the resource competition weights of different BOM levels. In this embodiment, λ=0.95 is set, which means that under the same urgency and value, the closer the node is to the final delivery end (the shallower the level), the higher its priority, so as to ensure that semi-finished products are quickly transformed into finished products.

[0061] This represents the normalization constant. Because... This could involve a huge sum of money. This is used to scale the final score to a numerical range (such as 0 to 1000) that is easy for the system to process, so that it can be stored and compared in computer memory later.

[0062] S260, dynamic updating and storage of priority scores; The system calculates This is not a fixed value; the priority calculation module 120 will re-trigger the above calculation logic whenever the system clock advances, resource preemption occurs, or external input orders change. Specifically, for... The delay node becoming negative will cause the exponential term in the formula to result in... The numerical values ​​exhibit non-linear growth, thereby creating a numerical advantage in resource arbitration and ensuring that the system can automatically identify high-risk nodes and prioritize resource allocation.

[0063] The calculation results are written to the node attribute table of the in-memory database in real time, so that the subsequent resource arbitration module can directly call them for comparison.

[0064] See attached document Figure 6 In this embodiment, the function of the state monitoring module 130 is to establish a mapping relationship between continuously changing logistics actions in the physical world and discrete control states in the digital system. Since the actions of physical equipment are continuous and span a time span, while computer scheduling algorithms are typically based on discrete time points and state values, this embodiment employs a finite state machine (FSM) model to abstract the complex on-site operation process into computable discrete state variables. This modeling approach enables the system to accurately define the attributes and permissions of materials at different operation stages, thereby providing definite logical boundaries for subsequent resource preemption and locking. Specifically, the following steps are included: S310 defines the discretized set of job states; The status monitoring module 130 has a set of standardized status presets. Used to describe any job node Current execution progress. In this embodiment, the state set is defined as follows: The system stores the state value by maintaining an enumerated type field, CurrentState, in the database node table. The physical meaning and technical definition of each state are as follows: (Virtual pre-occupancy status): This indicates that the node has been granted allocation eligibility through resource arbitration, but no physical action has yet occurred. At this time, the system has only marked the logical occupancy flag of the inventory record at the database level; the material entity is still in a free storage state and has not yet been bound to any handling equipment.

[0065] (Physical Lock Status): This indicates that a specific execution command has been issued. At this time, the WMS system has locked the specific cargo location or container, the AGV scheduling system has generated a handling task queue, and the material is locked by the physical equipment, which is in a critical state of "task issued but not executed".

[0066] (Work Execution Status): Indicates that the material is undergoing displacement or shape change. For example, an AGV is transporting materials, or a CNC machine tool is performing cutting operations. This status corresponds to a continuous process of physical action and has a time continuity.

[0067] (Buffering and waiting for inspection status): This indicates that the current process has been completed and the material has entered the line-side warehouse or quality inspection buffer zone, waiting to be transferred to the next process.

[0068] (Irreversible / Consumable State): This indicates that the material has undergone a permanent physical change (such as cutting or welding) or has been shipped out. This state marks the end of the task at this node and cannot be physically reversed to the initial state.

[0069] S320 establishes a mapping relationship between status and underlying control commands; To enable automatic state transitions, the system maintains a "state-instruction mapping table" in memory, establishing a one-to-one correspondence between the aforementioned discrete states and the specific operation instructions of the WMS (Warehouse Management System) and MES (Manufacturing Execution System) for triggering and feedback mechanisms.

[0070] When the status is marked as At that time, the system executes database-level transaction operations to update the "Available Quantity" and "Frozen Quantity" fields in the inventory table.

[0071] When the state transitions to At that time, the status monitoring module 130 sends picking instructions to the WMS through the API interface. The WMS responds and generates a picking order number, while simultaneously driving the electronic tag (PTL) to light up or locking the automated warehouse stacker crane.

[0072] When the state enters At this time, the system subscribes to the equipment operation data of the MES system or the real-time position data of the AGV control system. In specific implementation, the system listens for the "CycleStart" signal sent by the PLC controller or the "TaskStarted" message uploaded by the AGV. Once an acknowledgment signal is received, the CurrentState field is updated immediately.

[0073] When the state is reached When the system triggers the inventory deduction logic of the ERP system, it generates inbound and outbound documents and completes the cost transfer at the financial level.

[0074] The implementation of underlying device communication protocols (such as OPC UA, Modbus TCP, etc.) is a well-known technology in the art and will not be elaborated here.

[0075] S330, set the status acquisition frequency and flow constraint rules; The status monitoring module 130 uses a combination of event-driven and polling mechanisms for data collection. For critical nodes (such as...) to (For switching), the system uses a real-time message queue to listen for state change events reported by the underlying device, and the response latency is controlled within 100ms; for non-critical intermediate states, the system sets a fixed polling period to read the device register values. In this embodiment, the fixed polling period is set to be adjustable from 500ms to 2000ms.

[0076] Meanwhile, the system defines flow direction constraints in the state machine, which are determined by comparing the current state index value. With target state index value To control the flow logic: Bidirectional invertible interval: and The state transitions between these processes are bidirectional and reversible. If a high-priority order triggers the preemption mechanism, the system can send a "cancel command" to cancel it. For AGV tasks in the current state, release the physical lock and revert the node state to the previous state. Alternatively, it can be reset to idle, thereby achieving lossless resource reallocation.

[0077] One-way irreversible region: Once the state enters (Right now If the index value increases, the state flow becomes a unidirectional constraint, meaning it can only flow in the direction of increasing index value. This is determined by the continuity of physical operations; for example, a cut steel plate cannot be restored. The system will... The threshold is set to be irreversible. Once this threshold is exceeded, the node is marked as high-cost occupation. The system will significantly raise the preemption threshold for this node (i.e., the rollback cost increases sharply). The current task will only be interrupted if the preempting party has an extremely high priority advantage, until the task is completed.

[0078] See attached document Figure 7 In this embodiment, to address the system jitter caused by frequent resource contention in a high-concurrency environment, the state monitoring module 130 introduces a rollback cost function based on physical state. The function is constructed based on the hysteresis comparison mechanism in control theory, which introduces a damping term related to the current physical state into the resource arbitration equation. When a high-priority task attempts to preempt resources held by a low-priority task, not only are their priority scores compared, but the additional cost of physical state rollback must also be paid. This mechanism ensures that resource reallocation is triggered only when the value of the new task is significantly higher than the sum of the value of the original task and the rollback cost, thus guaranteeing the overall stability of the production system. The specific calculation and setting process includes the following steps: S340, construct the segmented rollback cost function; The system constructs a piecewise function to compute the current node based on the previously defined discrete state set SS. The cost of rollback This function maps the time, logistics, and potential risk costs of canceling the current task to numerical values ​​with the same dimensions as the priority score. The calculation formula is as follows: ; in: Represents a node The current physical state.

[0079] This represents the logistics damping coefficient, used to quantify the degree of disturbance caused by the removal of physical locks, and has a value greater than 1.

[0080] This represents the reverse logistics time, which is the estimated time required for the AGV to transport materials from its current location back to its original storage location or intermediate buffer area. This value is calculated by the AGV scheduling system based on the path planning distance between the current coordinates and the origin coordinates, divided by the average operating speed of the AGV.

[0081] This represents the cost of canceling an operation instruction. It is a preset constant and represents the average time-consuming value of system operations such as canceling documents, updating database transactions, and extinguishing electronic tags in the WMS system.

[0082] This represents the processing interruption penalty coefficient, used to amplify the weight of the currently executing task in order to suppress interruptions to the processing task.

[0083] Indicates the current node The dynamic priority score is used here to indicate that the higher the task priority, the greater the potential loss from its interruption.

[0084] This represents the maximum cost threshold set by the system. In computer implementations, it is typically taken as the maximum value of an integer variable (such as INT_MAX) or an order of magnitude exceeding the system's theoretical maximum priority score (e.g., 10). 9 ).

[0085] S350, set the damping threshold parameters under different conditions; In this embodiment, regarding the damping parameter in the formula And related variables, the system sets them as follows based on the actual characteristics of the physical operation: against In the (virtual pre-emption) state, since it only involves logical flag clearing at the database level and does not involve physical device actions, the rollback cost is set to 0. This means that before physical actions occur, the system allows high-priority orders to pre-empt resources at zero marginal cost to achieve optimal allocation.

[0086] against (Physically locked) state, at which point the AGV may already be en route to the storage location. This embodiment sets... The value range is [1.2, 1.5]. The design intention of this coefficient being greater than 1 is to introduce bias protection, that is, unless the benefits brought by preemption exceed 1.2 to 1.5 times the original task rollback cost, the system tends to maintain the status quo.

[0087] against The (job execution) status indicates that materials are being processed or transported at high speed. Forced interruption may result in tool damage, material scrap, or equipment malfunction. Therefore, this embodiment will... A high sensitivity threshold is set, with a value range of [5.0, 10.0]. This ensures that unless the preemptive party has an extremely high delay loss coefficient... (For example, emergency shutdown for spare parts maintenance), otherwise the rollback cost cannot be paid, thus protecting ongoing physical operations from being easily interrupted.

[0088] S360 provides logic locking and stability assurance for irreversible states; When the node state enters (Buffer pending inspection) or When the situation is irreversible / consumable, the physical state has changed or the task has been substantially completed. At this point, rollback is physically impossible. Therefore, the system will... Set as In the specific algorithm implementation, when calculating net profit, the resource arbitration module will first determine... Is it equal to If the condition is met, the system marks the node as rigidly occupied, skips the preemption decision logic, and keeps the original task queue unchanged. This mechanism prevents illogical device command issuance due to over-optimization of the algorithm at the algorithm level.

[0089] See attached document Figure 8 In this embodiment, the resource arbitration module 140, as the core decision-making unit of the scheduling system, is responsible for executing a deterministic allocation strategy when limited production resources (such as inventory materials, machine capacity, and AGV transport capacity) face multi-task competition. To address the problem of frequent resource switching and system oscillations caused by small priority fluctuations in traditional greedy algorithms in dynamic environments, this embodiment draws on the Schmitt trigger principle in control theory to construct a hysteresis comparator model based on damped inequalities. This model introduces an inertial component related to the physical state into the decision logic, artificially constructing a hysteresis interval between the trigger point and release point of resource preemption, thereby filtering out high-frequency, low-value preemption requests. The specific processing includes the following steps: S410, real-time identification of resource conflict scenarios; The system identifies potential conflicts by monitoring the resource request queue through polling or event triggering. When a new task node... Initiate a resource request to the system At that time, the resource arbitration module 140 first retrieves the target resource. The current state record.

[0090] Specifically, the system checks the AllocationStatus and OccupiedBy fields in the resource table. If If the available resources meet the request volume, allocate directly; if the available resources are insufficient and the resource has already been used by other task nodes. Pre-occupancy or locking (i.e.) In or (Status) The system identifies the set of all candidate preempted tasks currently occupying the inventory of that material (usually selecting the task with the lowest priority or the lowest rollback cost as the potential preempted object). If a conflict occurs, the system determines that a resource conflict is currently in effect. At this point, the system triggers the preemption logic and... Defined as the initiator of the preemption, Defined as the current holder.

[0091] S420, obtain quantitative evaluation parameters for both competitors; After a conflict is detected, the module calls the aforementioned priority calculation module 120 and status monitoring module 130 to obtain the following key parameters: Seize the dynamic priority score of the initiator This value reflects the urgency and business value of the new task at the current moment.

[0092] Dynamic priority score of the current holder This value reflects the rationale for the original task to continue holding the resource.

[0093] Rollback cost for current holders This value is determined by the status monitoring module based on... Current physical state ( to The value is calculated in real time and represents the total system cost required to forcibly remove the resource.

[0094] S430, hysteresis comparison determination based on damping inequality; The system employs a nonlinear hysteresis comparison algorithm to determine whether to perform preemption. The core of this embodiment lies in the introduction of a physical state damping term, making the preemption threshold no longer a simple numerical comparison, but a dynamically changing decision boundary. The specific preemption determination formula is as follows: ; in, The priority net difference represents the value advantage of the new task relative to the old task. As the rollback cost, and as the damping term on the right-hand side of the decision inequality, when From virtual pre-registration Enter physical lock When the state is in a certain condition, this value increases dramatically, causing the threshold on the right side of the inequality to rise, thus increasing the difficulty of preemption. The system stability threshold is a preset positive real constant used to set the minimum preemption benefit, filtering out priority jitter caused by fluctuations in calculation accuracy or minor parameters. In this embodiment, The value range is set to [0.5, 2.0], and the specific value is calibrated based on 5% to 10% of the system's historical average priority score.

[0095] S440, Execute arbitration results and queue adjustments; Based on the truth value of the above inequality, resource arbitration module 140 executes the following branch logic: If the inequality is true, it means that the value advantage of the new task is sufficient to cover the rollback cost and system disturbance risk of the old task. In this case, the system executes a "forced preemption" process: first, it sends an Abort command to the underlying control system to terminate the process. The physical action is followed by a database transaction rollback to release the data. The lock mark, and Reinsert into the head of the waiting queue; finally, allocate the resources. Ownership ID updated to and will Status updated to .

[0096] If the inequality is false, it means the preemption condition has not been met. This could be because the new task's priority is not high enough, or the old task has already entered the high-cost physical execution phase (i.e., (Too large). At this time, the system remains... The resource holding status remains unchanged, and Add resources In the blocking waiting queue, and according to its The values ​​are being sorted and are waiting.

[0097] Based on the above logic, this embodiment constructs a "hysteresis loop" at the algorithm level. With... Progress of physics assignments (from state to state) Towards (evolution), its The value gradually increases, which is equivalent to continuously raising the threshold for seizing tasks. This mechanism effectively prevents resources from repeatedly switching between two tasks with similar priorities, ensuring that once the physical production process enters a critical stage, it can be executed stably until completion unless an extremely high-priority emergency task is encountered.

[0098] See attached document Figure 9 In this embodiment, once the hysteresis comparator outputs a definite arbitration result, the system needs to perform physical-level reverse control and update the global scheduling parameters. This process is based on an event-driven architecture, triggering corresponding low-level control instructions through state machine state transitions to ensure the synchronization of information flow and logistics. Specifically, it includes the following steps: S450 executes reverse logistics control after the preemption is established; When the arbitration result allows for the new task Seize the old mission When resources are unavailable, the system immediately initiates the reverse control subroutine. This subroutine performs the following operations in parallel by calling the standard interface of WCS (Warehouse Control System): Physical action cancellation: The system sends a Task_Abort command to the AGV scheduling service. If When the corresponding AGV is in an unloaded moving state, the system issues an Update_Target command to reset its target coordinates (x, y) to the coordinates of the nearest standby point; if the AGV is in a loaded state (i.e., in a loaded state...), the system will issue an Update_Target command to reset its target coordinates (x, y) to the coordinates of the nearest standby point. and (At the critical point), the system generates the Reverse_Logistics instruction, which drives the AGV to transport the material back to the original storage location or the specified intermediate buffer area.

[0099] Logical lock release and reallocation: The system executes atomic transactions in the database. Specifically, the system utilizes the database's row-level locking mechanism to first release and reallocate logical locks. Set the Occupied_By field in the resource table to null and roll back its status to Pending; then, in the same transaction, update the field to... ID, and Set the state to (Virtual pre-occupancy).

[0100] On-site visual update: The system sends a signal to the field terminal via an industrial bus (such as Modbus TCP) to turn off the electronic tag (PTL) corresponding to the original task and light up the indicator light corresponding to the new task, so as to realize the synchronous refresh of the human-machine interface.

[0101] S460, executes out-of-stock queuing and procurement triggering after preemption failure; When the arbitration result determines that preemption is not allowed, it indicates that the current resource is occupied by a high-value task or is physically irreversible. The system executes the following resource buffering logic: Blocking queue management: Insert target resource The "blocking wait queue" is based on dynamic priority scores. A priority queue is constructed. The system refreshes the scores of each task in the queue at preset intervals to ensure that the head of the queue can be dynamically adjusted if a higher-value task is added.

[0102] Dynamic stockout alerts: The system verifies resources in real time. The expected release time. Among them, the expected release time... Equal to the task currently holding the resource Earliest expected completion time If detected Exceeded Latest start time If the total inventory is insufficient to meet the needs of all tasks in the queue, the system will trigger a shortage exception. In this case, the module will push a replenishment request to the ERP system via the API interface. The document will include the quantity in shortfall and the suggested latest arrival time.

[0103] S470, global time parameter recalculation and closed-loop feedback; Whether preemption causes an old task to be delayed or a new task enters the waiting queue, it will change the time topology of the production plan. In order to maintain the accuracy of scheduling, the system triggers a time parameter recalculation mechanism based on a directed acyclic graph (DAG).

[0104] The system starts from the node where the change occurred, traverses downstream along the process route, and updates all subsequent related nodes. The earliest expected start time Compared with the earliest expected completion time The calculation formula is as follows: ; ; in, For the affected downstream task nodes; For nodes The set of all preceding immediate tasks; The current system time is used to constrain the earliest start time of tasks to be no earlier than the current time. For nodes The standard operating cycle, this value is read from the process database of the MES system; This is the time for switching or preparation. If the tasks before and after belong to different product families, this value is obtained by the system from a table lookup table based on the switching matrix.

[0105] Based on the above calculations, the system updates the timeline for the entire task chain. To achieve closed-loop feedback control, the recalculated final delivery time... The promised delivery date of the order has been exceeded. The system will automatically adjust the delay loss coefficient for this order according to the following formula. : ; in, The feedback gain coefficient ranges from [0.1, 0.5]. This formula's function is to automatically amplify the delay weight of a task as the delay risk increases, thereby increasing its weight in the next round of resource arbitration. This value makes it easier for them to seize resources, forming a closed-loop control of "deviation perception - parameter correction - execution optimization". Example

[0106] To more intuitively illustrate the operational logic and technical effects of this invention in a real industrial production environment, the following detailed explanation is provided using a typical precision automotive parts manufacturing scenario.

[0107] Scene background: On a high-precision gear assembly line in an automotive parts manufacturing workshop, the core raw material, aerospace-grade bearings (material code: M-Ti-008), is currently in short supply, with only 50 units remaining. Multiple sales orders within the system are competing for this limited resource.

[0108] Step 1: Initial State and Resource Locking The system has currently allocated resources to ordinary order A according to the time sequence.

[0109] Order A Attributes: The associated customer is a general channel distributor, the total order amount is 50,000 yuan, the contract delivery period is 5 days, and the time slack is... Relatively large.

[0110] System calculation: Priority calculation module 120 calculates the dynamic priority score for order A. =320.

[0111] Physical Status: Resource arbitration module 140 has locked 50 bearings to order A. The warehouse AGV (AGV-05) has received the instruction and is en route to the automated warehouse to retrieve the goods, but has not yet loaded them. At this time, status monitoring module 130 determines that the material is in S1 (physically locked state) and calculates its rollback cost (logistics return cost) based on the distance. ≈50.

[0112] Step 2: High-priority task emergency intervention At this moment, the system receives an urgent, expedited order B.

[0113] Order B attributes: The associated customer is a strategic OEM, the total order amount is 150,000 yuan (approximately three times that of a regular order), and it is a downtime repair order requiring delivery within 12 hours. It approaches 0 and has extremely high time sensitivity.

[0114] Priority Update: The system immediately triggers the calculation logic. This is due to the delay loss coefficient of order B. The requirements are high and time is tight, based on the formula The dynamic priority score of order B is calculated. =850.

[0115] Step 3: Conflict arbitration based on damping inequality The resource arbitration module 140 detected insufficient inventory and triggered a competition determination for the M-Ti-008 resource.

[0116] Calculate the net difference in priority: =850−320=530.

[0117] Calculate the preemption threshold: The system sets a basic stability threshold. =20. At this point, the right side of the inequality is determined to be... =50+20=70.

[0118] Hysteresis comparison determination: because The inequality holds.

[0119] Judgment result: Preemption is permitted. The system determines that the cost of withdrawing the AGV at this point is far less than the business loss caused by the delay of order B.

[0120] Step 4: Perform reverse control and resource rearrangement The system performs the following closed-loop operation: 1. Command Revocation: Send a Task_Redirect command to the WCS system, and AGV-05 will stop executing the task for Order A. The task objective will be changed to "collect materials for Order B".

[0121] 2. Data update: The inventory of the batch of materials held by order A in the database is released, the status is rolled back to Pending, and it is placed at the top of the out-of-stock queue; the batch of 50 bearings is re-locked to order B.

[0122] Step 5 (Comparison Scenario Demonstration): If it enters an irreversible, high-cost stage Assume that when step 2 occurs, the AGV has already completed picking up the goods and sending them to the machine to start cutting (the state transitions to the S2 job execution state).

[0123] Parameter changes: At this point, the rollback cost... The value surged to 600 due to the introduction of a processing interruption penalty coefficient and potential scrap costs.

[0124] Decision logic: The inequality then becomes .

[0125] Result: 530 < 620, the inequality is not true.

[0126] Technical effect: Although order B has a higher priority, it is insufficient to cover the high cost of interrupting production. The system refused to preempt, maintaining order A's production and avoiding the risk of expensive semi-finished product scrapping and emergency production line shutdown, demonstrating the system's ability to make optimal decisions on global costs. Example

[0127] To verify the performance advantages of the hysteresis arbitration algorithm based on damped inequality proposed in this invention under complex dynamic environments, this embodiment constructs a discrete event simulation model for verification.

[0128] 1. Experimental Environment Setup Dataset Construction: Simulates the production flow of a discrete manufacturing workshop over 30 working days. The input data contains 1000 randomly generated orders. Order arrival times follow a Poisson distribution, while order values ​​and urgency follow a normal distribution. Control Group: Employs a traditional absolute priority preemption algorithm, i.e., as long as... Seize it immediately.

[0129] Experimental group: The damped hysteresis arbitration algorithm of this invention is used.

[0130] 2. Definition of Evaluation Indicators Task switching frequency: The number of physical task interruptions and reschedulings that occur per unit of time, reflecting production stability.

[0131] Overall On-Time Delivery Rate: The percentage of on-time deliveries weighted by order value.

[0132] Material loss cost: The total amount of material scrap caused by forced interruption of processing (S2 state is interrupted).

[0133] 3. Analysis of Experimental Results See attached document Figure 10 With appendix Figure 11 : Performance indicators Control group (traditional algorithm) Experimental group (algorithm of this invention) Optimization effect Task switching frequency 1245 times / month 186 times / month Reduced by 85.1% Equipment Effectiveness (OEE) 0.724 0.918 An increase of 19.4% Overall delivery on-time rate 0.865 0.942 Increased by 7.7% Material loss cost 125,000 yuan 12,000 yuan Reduced by 90.4% like Figure 10 and Figure 11 As shown, traditional algorithms are prone to the ping-pong effect when faced with a group of orders with similar priorities, such as orders with scores of 300 and 305. This causes the device to switch frequently between different tasks, resulting in a large amount of invalid setup time. This invention, however, introduces a rollback cost related to physical state. At the algorithm level, a preemption protection zone was constructed. Experimental data shows that this invention achieves an order-of-magnitude improvement in production line stability and reduces material scrap costs caused by unreasonable order insertions by more than 90% by sacrificing only a very small number of edge-priority orders (with minimal differences) for preemption opportunities.

[0134] 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 unified computing intelligent management and analysis system for enterprise resource data, characterized in that, Includes the following steps: The system includes a central processing server, a data storage unit, an interface interaction unit, and a terminal execution unit; The central processing server establishes bidirectional data communication connections with the data storage unit and the interface interaction unit respectively. The interface interaction unit is configured to connect to an external management system, and the terminal execution unit is configured to execute physical operation instructions. The central processing server is internally configured with a topology building module, a priority calculation module, a status monitoring module, and a resource arbitration module. The topology construction module is configured to decompose sales orders into multi-level sub-task nodes based on bill of materials and process route data, and construct a dynamic dependency topology diagram of the entire link reflecting the dependency relationship between material entities and processing tasks. The priority calculation module is configured to calculate the dynamic priority score of an order for a specific resource based on the full-link dynamic dependency topology graph. The status monitoring module is configured to obtain the physical operation status of the materials occupied by the pre-allocated order in the physical space, and convert the physical operation status into the corresponding rollback cost value. The resource arbitration module is configured to, when a resource conflict is detected, identify the pre-allocated order occupying the target resource and the request order applying for the target resource, calculate the difference in dynamic priority score between the request order and the pre-allocated order, and compare the difference with the rollback cost value of the pre-allocated order to determine whether to perform resource preemption.

2. The enterprise resource data integrated computing intelligent management and analysis system according to claim 1, characterized in that, The topology building module performs the following operations: The interface interaction unit reads the sales order data to be processed and decomposes the sales order into the smallest granularity operation unit node based on the pre-stored bill of materials data and process route data. Directed edges reflecting the pre-process dependencies are established between the work unit nodes, and loop detection is performed; For work unit nodes containing the same material code, a cross-order material reverse index is established, the full-link dynamic dependency topology is generated, and stored in the data storage unit.

3. The enterprise resource data integrated computing intelligent management and analysis system according to claim 1, characterized in that, The priority calculation module performs the following operations: The entire dynamic dependency topology graph is traversed in a forward topology sorting manner. The earliest start time of each node is calculated by combining standard working hour data and logistics turnaround time. Based on the order delivery deadline, perform a reverse topology traversal of the full-link dynamic dependency topology graph and calculate the latest start time of each node; The difference between the latest start time and the earliest start time is calculated to obtain the time slack, and the time slack is used as a parameter to calculate the dynamic priority score.

4. The enterprise resource data integrated computing intelligent management and analysis system according to claim 3, characterized in that, The priority calculation module uses a multidimensional quantization model to calculate the dynamic priority score. The multidimensional quantization model is configured as follows: The delay loss factor for an order is calculated based on the contract amount, penalty clauses, and customer credit rating. An exponential decay model is constructed based on the time relaxation, such that the dynamic priority score increases non-linearly as the time relaxation decreases. A hierarchical decay factor is introduced to adjust the dynamic priority score based on the hierarchical depth of the node in the bill of materials structure tree.

5. The intelligent management and analysis system for integrated computing of enterprise resource data according to claim 1, characterized in that, The status monitoring module defines the discretized physical job status, which includes at least the following: The virtual pre-emptive state indicates that the node has been granted the right to allocate resources but has not yet taken any action at the physical level; The physical lock status indicates that the execution command has been issued and the material has been physically locked by the equipment; The operation status indicates that the material is undergoing displacement or shape change; Irreversible processing status indicates that the material has undergone a permanent change in physical form or has been shipped.

6. The intelligent management and analysis system for integrated computing of enterprise resource data according to claim 5, characterized in that, The status monitoring module calculates the rollback cost value based on a preset mapping relationship: When the physical job status is virtual pre-occupancy, the rollback cost value is set to zero; When the physical operation status is physically locked, the rollback cost value is calculated based on the reverse logistics time and the cost of canceling the operation instruction. When the physical job state is job execution state, a processing interruption penalty coefficient is introduced to increase the rollback cost value. When the physical operation is in an irreversible state, the rollback cost is set to the system's maximum threshold to prevent preemption.

7. The intelligent management and analysis system for integrated computing of enterprise resource data according to claim 1, characterized in that, The resource arbitration module performs arbitration decisions based on damped inequalities: Calculate the priority difference between the dynamic priority score of the requested order and the dynamic priority score of the pre-allocated order; Determine whether the priority difference is greater than the sum of the rollback cost value and the preset system stability threshold; If the judgment result is yes, then the resource preemption is determined to be valid; If the judgment result is negative, the current allocation state is maintained, and the requested order is added to the blocking waiting queue.

8. The intelligent management and analysis system for integrated computing of enterprise resource data according to claim 7, characterized in that, After determining that resource preemption has occurred, the resource arbitration module is configured to perform the following operations: A reverse logistics instruction is generated and sent to the terminal execution unit through the interface interaction unit to withdraw the physical operation task of the pre-allocated order; Release the lock flag of the pre-allocated order on the target resource in the database, and relock the target resource to the requesting order; The status of the pre-assigned order is rolled back and it is reinserted into the waiting queue.

9. The intelligent management and analysis system for integrated computing of enterprise resource data according to claim 7, characterized in that, The priority calculation module is also configured to perform closed-loop feedback and dynamic rearrangement: When it is determined that maintaining the current allocation state will cause the requested order to enter the blocking waiting queue, the time parameters of the affected nodes in the full-link dynamic dependency topology graph are recalculated based on the changed resource allocation results. If the recalculated estimated completion time exceeds the promised delivery date of the order, the delay loss coefficient of the order will be automatically adjusted to improve its dynamic priority score in the next round of calculation.

10. The intelligent management and analysis system for integrated computing of enterprise resource data according to claim 1, characterized in that, The terminal execution unit includes an automated guided vehicle scheduling controller and an industrial PC terminal; The terminal execution unit is configured to receive scheduling instructions or reverse logistics instructions from the resource arbitration module, and control physical equipment to perform material handling, processing or task cancellation operations. At the same time, it feeds back real-time equipment operation data to the interface interaction unit for the status monitoring module to collect.

Images