A multi-warehouse linkage-oriented intelligent spare parts warehousing scheduling control method and system

Abstract

The application discloses a kind of intelligent spare parts warehouse scheduling control methods and systems for multi-library linkage, and relates to the technical field of intelligent scheduling.The method comprises the following steps: constructing a multi-library digital model; performing multi-dimensional scheduling attention analysis on the multi-library digital model, and constructing a multi-dimensional scheduling attention graph network; performing multi-library linkage scheduling decision on the multi-library digital model to obtain a first linkage scheduling space; performing spare parts scheduling risk analysis optimization to obtain a second linkage scheduling space; performing inventory balancing evaluation optimization to obtain a third linkage scheduling space; performing neighborhood expansion optimization to obtain a linkage scheduling optimization strategy, and performing multi-library linkage scheduling control of multiple spare parts warehouses according to the linkage scheduling optimization strategy.The technical problem of isolated scheduling decision among multiple warehouses in the prior art, inability to dynamically optimize according to real-time demand, resulting in spare parts supply response lag and inventory structure imbalance, is solved, and the technical effects of improving the real-time, balance and intelligence of multi-library linkage scheduling are achieved.

Description

Technical Field

[0001] This invention relates to the field of intelligent scheduling technology, specifically to an intelligent spare parts storage scheduling and control method and system for multi-warehouse linkage. Background Technology

[0002] With the development of intelligent industrial equipment and large-scale production lines, the number of spare parts warehouses within enterprises and across regions continues to increase, forming a multi-warehouse collaborative support system. Existing spare parts warehouse scheduling typically optimizes localized operations on a single warehouse basis, lacking a global coordination mechanism across warehouses. Data silos exist between warehouses regarding inventory information, demand forecasting, and allocation plans. When production equipment experiences sudden spare parts demands, the scheduling system cannot dynamically adjust resources across multiple warehouses based on real-time demand, leading to delayed spare parts supply response, unbalanced inventory structures, and long-term stockpiling in some warehouses. Traditional warehouse management methods rely heavily on manual experience or static rule-based decision-making, making it difficult to balance the spatiotemporal linkages between multiple warehouses with the goal of inventory equilibrium, and thus failing to achieve intelligent collaborative control of complex supply networks. Summary of the Invention

[0003] This application provides an intelligent spare parts storage scheduling and control method and system for multi-warehouse linkage, which solves the technical problems in the prior art where scheduling decisions between multiple warehouses are isolated and cannot be dynamically linked and optimized according to real-time demand, resulting in delayed spare parts supply response and unbalanced inventory structure.

[0004] The first aspect of this application provides an intelligent spare parts storage scheduling and control method for multi-warehouse linkage, the method comprising: Multiple spare parts warehouses are monitored in real time to construct a multi-warehouse digital model. Based on real-time spare parts demand, the multi-warehouse digital model undergoes multi-dimensional scheduling attention analysis to construct a multi-dimensional scheduling attention graph network. Based on the real-time spare parts demand, multi-warehouse coordinated scheduling decisions are made using the multi-dimensional scheduling attention graph network to obtain a first coordinated scheduling space. The first coordinated scheduling space is then optimized using a spare parts scheduling risk analysis model to obtain a second coordinated scheduling space. The second coordinated scheduling space is then optimized using an inventory balance evaluation record set to obtain a third coordinated scheduling space. The third coordinated scheduling space is then optimized through neighborhood expansion using the spare parts scheduling risk analysis model to obtain a coordinated scheduling optimization strategy. Finally, multi-warehouse coordinated scheduling control of the multiple spare parts warehouses is executed according to the coordinated scheduling optimization strategy.

[0005] A second aspect of this application provides an intelligent spare parts warehousing and scheduling control system for multi-warehouse linkage, the system comprising: The system comprises the following modules: a digital model construction module for real-time monitoring of multiple spare parts warehouses and construction of a multi-warehouse digital model; an analysis module for multi-dimensional scheduling attention analysis of the multi-warehouse digital model based on real-time spare parts demand and construction of a multi-dimensional scheduling attention graph network; a scheduling decision module for multi-warehouse coordinated scheduling decisions based on the real-time spare parts demand and the multi-dimensional scheduling attention graph network, obtaining a first coordinated scheduling space; a first optimization module for optimizing the first coordinated scheduling space based on a spare parts scheduling risk analysis model, obtaining a second coordinated scheduling space; a second optimization module for optimizing the second coordinated scheduling space based on an inventory balance evaluation record set, obtaining a third coordinated scheduling space; and a scheduling control module for optimizing the third coordinated scheduling space through neighborhood expansion based on the spare parts scheduling risk analysis model, obtaining a coordinated scheduling optimization strategy, and executing multi-warehouse coordinated scheduling control of the multiple spare parts warehouses according to the coordinated scheduling optimization strategy.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, multiple spare parts warehouses are monitored in real time to construct a multi-warehouse digital model. Next, based on real-time spare parts demand, the multi-warehouse digital model undergoes multi-dimensional scheduling attention analysis to construct a multi-dimensional scheduling attention graph network. Then, based on real-time spare parts demand, multi-warehouse coordinated scheduling decisions are made using the multi-dimensional scheduling attention graph network to obtain a first coordinated scheduling space. The first coordinated scheduling space is then optimized using a spare parts scheduling risk analysis model to obtain a second coordinated scheduling space. The second coordinated scheduling space is then optimized using an inventory balance evaluation record set to obtain a third coordinated scheduling space. Finally, the third coordinated scheduling space is optimized through neighborhood expansion using the spare parts scheduling risk analysis model to obtain a coordinated scheduling optimization strategy. This strategy is then used to execute multi-warehouse coordinated scheduling control across multiple spare parts warehouses. This approach solves the technical problems of isolated scheduling decisions among multiple warehouses in existing technologies, which prevent dynamic optimization based on real-time demand, leading to delayed spare parts supply response and inventory imbalance. It achieves the technical effect of improving the real-time performance, balance, and intelligence of multi-warehouse coordinated scheduling. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1A schematic flowchart of an intelligent spare parts storage scheduling and control method for multi-warehouse linkage provided in an embodiment of this application; Figure 2 This is a schematic diagram of an intelligent spare parts storage and scheduling control system for multi-warehouse linkage, provided as an embodiment of this application.

[0009] Figure labeling: Digital model construction module 11, analysis module 12, scheduling decision module 13, first optimization module 14, second optimization module 15, scheduling control module 16. Detailed Implementation

[0010] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0011] Example 1, as Figure 1 As shown, this application provides an intelligent spare parts warehousing scheduling and control method for multi-warehouse linkage, wherein the method includes: Real-time monitoring of multiple spare parts warehouses is conducted to build a multi-warehouse digital model.

[0012] The operational status of each spare parts warehouse is monitored in real time by deploying multiple types of sensing nodes. These sensing nodes include environmental sensors, inventory identification terminals, intelligent access control systems, electronic tag systems, and e-ink screen location identification modules. Environmental sensors collect data on temperature, humidity, illuminance, and vibration to reflect the warehouse environment. Inventory identification terminals and electronic tag systems collect data on spare parts entering, leaving, and remaining in stock, as well as location data. The intelligent access control system collects data on personnel and vehicle entry and exit to record allocation and delivery activities. Data collected from each warehouse is aggregated to a cloud-based scheduling platform via an edge computing gateway, where data format standardization and timestamp alignment are performed to form a multi-warehouse real-time dataset. Based on this dataset, a multi-warehouse digital model is established, consisting of multiple warehouse sub-models. Each sub-model includes an inventory status sub-layer, an allocation capacity sub-layer, and a transportation timeliness sub-layer. By mapping these sub-models together in the cloud, a global warehouse digital twin containing information on spatial distribution, inventory structure, and scheduling capabilities is constructed. Based on this, the system takes the inventory capacity, spare parts type, turnover rate and geographical coordinates of each warehouse as input parameters, performs multi-dimensional parameter normalization and vectorization encoding, and generates a multi-database digital model that can be used for scheduling and parsing.

[0013] Furthermore, each spare parts warehouse is equipped with an intelligent access control system and an electronic tag system.

[0014] The intelligent access control system is used to realize unmanned management and security control of the warehouse. The system is linked with the document management platform through facial recognition terminals, which restricts the entry of the corresponding document applicant to the warehouse only when the business documents such as transfer, requisition, and return are in the "approved" or "in execution" status. The system automatically records the entry and exit time, operation identity and associated document number, so as to realize the traceability and controllable access of the entire process of entry and exit behavior.

[0015] The electronic tag system is applied to location management and material identification, achieving dynamic binding of "one item per location." Each location is equipped with an electronic tag display terminal, which communicates in real time with the warehouse management system. It automatically displays the name, specifications, quantity, unit, entry / exit time, supplier, and batch information of the materials stored in that location. When the status of materials changes, the electronic tag automatically updates its display and provides location guidance when new materials are placed on the shelves, prompting operators to accurately locate the target location and preventing misplacement or omission. Through the collaboration of the intelligent access control system and the electronic tag system, unmanned warehouse management, real-time visualization of material information, and efficient traceability management of spare parts entry / exit operations are achieved.

[0016] Furthermore, electronic ink bottles are installed in each storage location within multiple spare parts warehouses.

[0017] For standardized warehouse racking structures, an electronic ink bottle is installed at each storage location. This bottle wirelessly connects to the warehouse management system to provide real-time indication and information display of the storage location status. When an outbound or inbound operation is performed, the system automatically activates the electronic ink bottle at the target storage location based on the corresponding document instruction, causing it to illuminate as a warning signal. This guides staff or warehouse managers to quickly locate the required storage location, enabling rapid storage and accurate retrieval of spare parts, significantly reducing inbound and outbound operation time and preventing misplacement of materials. The electronic ink bottle's display interface can be dynamically updated according to system instructions, displaying in real-time the name, specifications, quantity, batch number, supplier information, and the most recent inbound / outbound time of the materials stored at the current location. When the material status changes or the storage location becomes vacant, the electronic ink bottle automatically refreshes its display and enters standby mode. The application of this electronic ink bottle system enables precise and visual management at the storage location level, improving warehouse operational efficiency and the accuracy of material information, and providing real-time storage location-level data support for multi-warehouse digital models.

[0018] Based on the real-time demand for spare parts, the multi-database digital model is subjected to multi-dimensional scheduling attention analysis to construct a multi-dimensional scheduling attention graph network.

[0019] Furthermore, based on the real-time demand for spare parts, the multi-database digital model is subjected to multi-dimensional scheduling attention analysis to construct a multi-dimensional scheduling attention graph network, including: Based on the real-time spare parts demand, the multi-database digital model is subjected to inventory matching attention analysis to establish an inventory matching attention graph network; based on the real-time spare parts demand, the multi-database digital model is subjected to scheduling distance attention analysis to establish a scheduling distance attention graph network; based on the real-time spare parts demand, the multi-database digital model is subjected to emergency matching degree attention analysis to establish an emergency matching degree attention graph network; attention is aggregated based on the inventory matching attention graph network, the scheduling distance attention graph network, and the emergency matching degree attention graph network to obtain a comprehensive scheduling attention graph network; based on the inventory matching attention graph network, the scheduling distance attention graph network, the emergency matching degree attention graph network, and the comprehensive scheduling attention graph network, the multi-dimensional scheduling attention graph network is generated.

[0020] The system acquires real-time spare parts demand data within a target time window. This data originates from production equipment maintenance work orders, predictive maintenance systems, or emergency spare parts dispatch tasks. It extracts a set of demand feature parameters, including spare parts code, required quantity, urgency level, required location, and expected delivery time. This set of parameters is input into a multi-database digital model to trigger a multi-dimensional attention parsing process. Based on the inventory status sub-layer of the multi-database digital model, the system extracts available inventory information from each spare parts warehouse and quantifies it into an inventory feature vector set. Simultaneously, it vectorizes real-time demand parameters into demand feature vectors. By calculating the similarity between the demand feature vectors and each inventory feature vector, the system obtains the demand-inventory similarity distribution, thereby generating an inventory matching attention weight matrix and representing the relationship between warehouses in the inventory satisfaction dimension in a graph structure, forming an inventory matching attention graph network. Furthermore, based on geographical coordinates, traffic accessibility, and historical transportation timeliness parameters in the multi-database digital model, the system calculates the transportation distance and expected response time from each warehouse to the target demand location, obtaining scheduling distance weight coefficients and generating a scheduling distance attention graph network reflecting the spatiotemporal response priority relationship. Simultaneously, based on the urgency level parameters of the demand, combined with the current warehouse's operational load rate, inbound / outbound frequency, and in-transit order status, the system calculates an emergency response score and constructs an emergency matching degree attention graph network to characterize each warehouse's responsiveness to high-priority demands. The system inputs the inventory matching attention graph network, scheduling distance attention graph network, and emergency matching degree attention graph network into the multi-head attention fusion module, performs aggregation calculations through a weighted adaptive mechanism, and generates a comprehensive scheduling attention graph network. This comprehensive attention graph network is then fused with the aforementioned three sub-graph networks in a layer-level structure, ultimately forming a multi-dimensional scheduling attention graph network.

[0021] Furthermore, based on the real-time demand for spare parts, inventory matching attention analysis is performed on the multi-database digital model to establish an inventory matching attention graph network, including: Based on the multi-database digital model, the available spare parts inventory features of the multiple spare parts warehouses are analyzed to obtain multiple available spare parts inventory vectors; based on the real-time spare parts demand, spare parts demand features are analyzed to obtain spare parts demand feature vectors; based on the spare parts demand feature vectors, the multiple available spare parts inventory vectors are evaluated for similarity to obtain a demand-inventory similarity evaluation sequence; based on the demand-inventory similarity evaluation sequence, inventory matching attention is allocated to the multi-database digital model to generate the inventory matching attention graph network.

[0022] The system first calls the inventory status sublayer of each warehouse in the multi-database digital model to perform feature analysis on the available inventory information of multiple spare parts warehouses, extracting key parameters such as inventory quantity, spare parts model, batch, quality grade, and inventory update time. These parameters are then normalized and vectorized to form multiple available spare parts inventory vectors. Simultaneously, the system performs demand feature analysis on the input real-time spare parts demand information, extracting attributes such as spare parts code, demand quantity, demand time window, and demand grade, and mapping them to spare parts demand feature vectors through a feature embedding algorithm. Subsequently, the system uses the spare parts demand feature vectors as references to perform similarity evaluation calculations with each available spare parts inventory vector, employing cosine similarity or weighted Euclidean distance methods to measure the degree of matching between demand and inventory, obtaining a demand-inventory similarity evaluation sequence. Based on the distribution results of this similarity evaluation sequence, the system performs inventory matching attention allocation at the node layer of the multi-database digital model, assigning higher attention weights to warehouse nodes with higher similarity, thereby generating an inventory matching attention weight matrix. The system establishes an inventory matching attention graph network based on this weight matrix to reflect the relative advantages of each warehouse in meeting the target spare parts demand, and realizes dynamic correlation modeling and refined scheduling weight output at the inventory level.

[0023] Based on the real-time demand for spare parts, the multi-database digital model is subjected to multi-database linkage scheduling decisions according to the multi-dimensional scheduling attention graph network to obtain the first linkage scheduling space.

[0024] After receiving the parsed real-time spare parts demand parameters, the system injects them as input signals into a multi-dimensional scheduling attention graph network. Guided by attention weights across multiple dimensions such as inventory matching, scheduling distance, and urgency matching degree, the system performs global scheduling decision calculations on the multi-database digital model. First, the system uses the inventory matching attention graph network as the primary tool to identify the warehouse node group with the highest inventory sufficiency to meet demand, generating the first linked scheduling decision group. Then, based on the scheduling distance attention graph network, it performs weighted calculations on the distance from each warehouse node to the target demand location, transportation timeliness, and available transportation capacity, generating the second linked scheduling decision group. Next, based on the urgency matching degree attention graph network, combined with the demand urgency level and warehouse workload rate, it determines the priority response warehouse nodes, generating the third linked scheduling decision group. Simultaneously, based on the aggregation weights of the comprehensive attention graph network, it performs multi-dimensional fusion calculations on the aforementioned decision groups to obtain the fourth linked scheduling decision group. The system merges these four types of linked scheduling decision groups and performs node aggregation and edge weight optimization in the spatial topology of the multi-database digital model, forming a first linked scheduling space containing candidate solution sets for different scheduling strategies.

[0025] Furthermore, based on the real-time demand for spare parts, a first coordinated scheduling space is obtained by performing multi-database coordinated scheduling decisions on the multi-database digital model according to the multi-dimensional scheduling attention graph network, including: Based on the real-time spare parts demand, a first coordinated scheduling decision group is obtained by performing multi-database coordinated scheduling decisions on the multi-database digital model according to the inventory matching attention graph network; a second coordinated scheduling decision group is obtained by performing multi-database coordinated scheduling decisions on the multi-database digital model according to the scheduling distance attention graph network; a third coordinated scheduling decision group is obtained by performing multi-database coordinated scheduling decisions on the multi-database digital model according to the emergency matching degree attention graph network; a fourth coordinated scheduling decision group is obtained by performing multi-database coordinated scheduling decisions on the multi-database digital model according to the real-time spare parts demand and the scheduling comprehensive attention graph network; the first, second, third, and fourth coordinated scheduling decision groups are added to the first coordinated scheduling space.

[0026] After receiving the parsed real-time spare parts demand information, the system imports the demand characteristic parameters as input signals into a multi-dimensional scheduling attention graph network. Different dimensions of attention weights guide the coordinated decision-making process of the multi-database digital model. First, based on the inventory matching attention graph network, the system calculates the inventory sufficiency, inventory type consistency, and inventory update timeliness of each warehouse node, selecting the combination of warehouse nodes with the highest demand characteristic matching degree to generate the first coordinated scheduling decision group. Then, based on the scheduling distance attention graph network, the system comprehensively evaluates the transportation distance, transportation timeliness, and accessibility between each warehouse node and the target demand location, determining the optimal allocation path combination to generate the second coordinated scheduling decision group. Next, based on the urgency matching attention graph network, combined with the urgency level of the demand, warehouse operation load rate, and current inbound / outbound task status, the system determines the set of warehouse nodes with priority response capabilities to generate the third coordinated scheduling decision group. Finally, based on the comprehensive scheduling attention graph network, the system performs multi-dimensional weight aggregation and strategy fusion on the basis of the aforementioned multi-dimensional decision groups to form a comprehensive scheduling scheme and generate the fourth coordinated scheduling decision group. The four interconnected scheduling decision groups with different dimensions are collectively incorporated into the first interconnected scheduling space. Structural mapping and weight normalization are then performed on the node topology of the multi-database digital model to form a preliminary scheduling solution space containing multi-dimensional policy candidate solutions. This first interconnected scheduling space characterizes the multi-dimensional scheduling combinations and execution possibilities of the multi-database system under current real-time requirements, providing computational basis and strategy input for subsequent risk analysis optimization, inventory balancing assessment, and neighborhood expansion optimization.

[0027] Based on the spare parts scheduling risk analysis model, the first linkage scheduling space is optimized by performing spare parts scheduling risk analysis to obtain the second linkage scheduling space.

[0028] After generating the first coordinated scheduling space, the system simulates and schedules each coordinated scheduling decision scheme within it. Using a multi-database digital model, it dynamically simulates warehouse status, transportation routes, allocation delays, and operational resource occupancy, obtaining scheduling simulation data corresponding to each scheduling scheme. Subsequently, the system activates a pre-trained spare parts scheduling risk analysis model and performs risk quantification calculations on each simulation data. The spare parts scheduling risk analysis model includes multiple scheduling risk indicator dimensions, specifically scheduling conflict risk, scheduling safety risk, and scheduling delay risk. Scheduling conflict risk characterizes the probability of resource competition or task overlap when multiple warehouses respond simultaneously; scheduling safety risk characterizes the risk of scheduling failure due to environmental, equipment, or human factors during transportation and operations; and scheduling delay risk reflects the degree of timeliness deviation when allocation tasks cannot be completed within the demand time window. The system outputs the above risk indicators to form a scheduling risk analysis sequence and applies constraints to each risk indicator based on set scheduling risk optimization constraints. These constraints include scheduling conflict risk constraints, scheduling safety risk constraints, and scheduling delay risk constraints. Based on this, the system adopts a multi-objective optimization algorithm (such as the Pareto optimal solution screening mechanism based on weight adaptation) to identify the scheduling decision set with the lowest risk comprehensive value in the first linkage scheduling space, eliminate inefficient solutions that exceed risk constraints, and generate a second linkage scheduling space that meets the risk analysis optimization conditions.

[0029] Furthermore, based on the spare parts scheduling risk analysis model, the first linkage scheduling space is optimized through spare parts scheduling risk analysis to obtain the second linkage scheduling space, including: Simulation scheduling is performed on the multi-database digital model based on each linkage scheduling decision within the first linkage scheduling space to obtain simulation data for each decision scheduling; the spare parts scheduling risk analysis model is activated, which includes multi-dimensional scheduling risk indicators, including scheduling conflict risk, scheduling safety risk, and scheduling delay risk; the simulation data for each decision scheduling is input into the spare parts scheduling risk analysis model to obtain multiple scheduling risk analysis sequences; scheduling risk optimization constraints are obtained, including scheduling conflict risk constraints, scheduling safety risk constraints, and scheduling delay risk constraints; the first linkage scheduling space is identified through the multiple scheduling risk analysis sequences and the scheduling risk optimization constraints to generate the second linkage scheduling space.

[0030] After obtaining the first coordinated scheduling space, the system simulates the scheduling of each coordinated scheduling decision scheme within it in a multi-database digital model. This simulates the inventory turnover, transportation route occupancy, workload allocation, and response latency of each scheme during actual execution, generating a corresponding scheduling simulation dataset. Subsequently, the system activates a pre-trained spare parts scheduling risk analysis model. This model is a multi-dimensional risk identification model built based on machine learning algorithms, used to perform multi-dimensional risk quantification assessment of each scheduling simulation data. The spare parts scheduling risk analysis model includes multiple scheduling risk indicator dimensions, specifically three categories: scheduling conflict risk, scheduling safety risk, and scheduling delay risk. These indicators respectively characterize the probability of resource competition conflicts between allocation tasks, the probability of safety failures during operation and transportation, and the degree of time deviation during scheduling execution.

[0031] The system inputs simulation data corresponding to each coordinated scheduling decision into the spare parts scheduling risk analysis model for inference calculation, outputting multiple scheduling risk analysis sequences. Each scheduling risk analysis sequence contains risk coefficients corresponding to the three types of risk indicators mentioned above. To ensure the reliability of the scheduling scheme execution, the system sets scheduling risk optimization constraints, including scheduling conflict risk constraints, scheduling safety risk constraints, and scheduling delay risk constraints, which are used to limit the maximum allowable threshold range of risk indicators. Based on multiple scheduling risk analysis sequences, the system performs multi-objective constraint screening and risk optimization analysis in conjunction with the above optimization constraints, identifying the optimal scheduling scheme set that meets the risk constraint conditions, and eliminating inefficient schemes that do not meet the constraints or exceed the risk limits. Through this identification process, the system extracts the effective scheduling solution set that meets the risk optimization conditions from the first coordinated scheduling space, and then generates the second coordinated scheduling space.

[0032] Based on the inventory balance evaluation record set, the second linkage scheduling space is evaluated and optimized to obtain the third linkage scheduling space.

[0033] Furthermore, based on the inventory balance evaluation record set, the second linkage scheduling space is optimized through inventory balance evaluation to obtain the third linkage scheduling space, including: Based on the second linkage scheduling space, the Pth linkage scheduling decision is extracted, and the Pth inventory feedback simulation data corresponding to the Pth linkage scheduling decision is loaded, where P is a positive integer; supervised training is performed based on the inventory balance evaluation record set to obtain the first inventory balance evaluation model; decoupled parameter optimization is performed on the first inventory balance evaluation model using the AdamW optimizer to obtain the second inventory balance evaluation model; the Pth inventory feedback simulation data is input into the second inventory balance evaluation model to obtain the Pth inventory balance evaluation coefficient; if the Pth inventory balance evaluation coefficient is greater than or equal to the inventory balance evaluation threshold, the Pth linkage scheduling decision is added to the third linkage scheduling space.

[0034] After obtaining the second linkage scheduling space, the system evaluates the inventory balance of each linkage scheduling decision scheme. Specifically, the system sequentially extracts the P-th linkage scheduling decision scheme from the second linkage scheduling space and loads the corresponding P-th inventory feedback simulation data, where P is a positive integer. The inventory feedback simulation data reflects the inventory changes generated by each warehouse after executing the scheduling scheme, including dynamic indicators such as inventory occupancy rate, inventory turnover rate, spare parts replenishment frequency, and safety stock deviation. The system uses a historical inventory balance evaluation record set as a supervised sample to train the model, obtaining the first inventory balance evaluation model. The historical inventory balance evaluation record set contains inventory status and balance score labels of multi-time period and multi-regional warehousing networks, which are used to guide the model to learn the balance characteristics of multi-warehouse inventory distribution.

[0035] After initial training, the system uses the AdamW optimizer to perform decoupled parameter optimization on the first inventory balance evaluation model. Overfitting is eliminated through adaptive weight decay and gradient correction mechanisms, improving the model's generalization stability and nonlinear feature capture capability, thus obtaining the second inventory balance evaluation model. Subsequently, the P-th inventory feedback simulation data is input into the optimized second inventory balance evaluation model, and the P-th inventory balance evaluation coefficient is calculated and output, representing the overall balance of the multi-database system after the scheduling scheme is implemented. The system compares this coefficient with a preset inventory balance evaluation threshold. If the P-th inventory balance evaluation coefficient is greater than or equal to the threshold, the scheduling scheme is determined to maintain a reasonable inventory balance and is added to the third linked scheduling space; if it is lower than the threshold, the scheme is discarded and does not participate in subsequent neighborhood expansion and global optimization calculations.

[0036] The third linkage scheduling space is expanded and optimized according to the spare parts scheduling risk analysis model to obtain the linkage scheduling optimization strategy, and the multi-warehouse linkage scheduling control of the multiple spare parts warehouses is executed according to the linkage scheduling optimization strategy.

[0037] After obtaining the third linkage scheduling space, the system first performs random perturbation and parameter variation on the scheduling scheme set within the third linkage scheduling space based on the current real-time demand for spare parts, constructing multiple adjacent scheduling solution sets to form the first linkage scheduling neighborhood. The system then invokes the spare parts scheduling risk analysis model to perform multi-dimensional risk calculation and analysis on each scheme within the first linkage scheduling neighborhood. Based on indicators such as scheduling conflict risk, scheduling security risk, and scheduling delay risk, risk optimization is performed to select the second linkage scheduling neighborhood that meets the security and timeliness constraints. Subsequently, based on the inventory balance evaluation threshold, the system uses an optimized inventory balance evaluation model to evaluate the balance of the second linkage scheduling neighborhood, identifying the optimal scheme set in terms of inventory distribution, resource utilization, and spare parts occupancy rate, forming the third linkage scheduling neighborhood. The system merges and expands this third linkage scheduling neighborhood with the original third linkage scheduling space to obtain a fourth linkage scheduling space containing multi-dimensional optimal solution candidates. Based on this, the system uses a multi-objective optimization algorithm to optimize the scheduling efficiency of the fourth linkage scheduling space, comprehensively considering objectives such as shortest response time, highest inventory utilization, and lowest risk cost, to generate the final linkage scheduling optimization strategy. Based on this linkage scheduling optimization strategy, the system performs automated scheduling control on multiple spare parts warehouses, dynamically allocates outbound, transfer and replenishment tasks, realizes real-time linkage and optimal resource allocation across warehouses, and ensures the timeliness of spare parts supply and the stability of the overall system operation.

[0038] Furthermore, based on the aforementioned spare parts scheduling risk analysis model, the third linkage scheduling space is expanded and optimized to obtain a linkage scheduling optimization strategy, including: Based on the real-time demand for spare parts, the third linkage scheduling space is randomly mutated to obtain a first linkage scheduling neighborhood; the first linkage scheduling neighborhood is optimized for spare parts scheduling risk analysis according to the spare parts scheduling risk analysis model to obtain a second linkage scheduling neighborhood; based on the inventory balance evaluation threshold, the second linkage scheduling neighborhood is optimized for inventory balance evaluation according to the second inventory balance evaluation model to obtain a third linkage scheduling neighborhood; the third linkage scheduling space is expanded according to the third linkage scheduling neighborhood to obtain a fourth linkage scheduling space; the fourth linkage scheduling space is optimized for maximizing spare parts scheduling efficiency to generate the linkage scheduling optimization strategy.

[0039] After obtaining the third linkage scheduling space, the system first performs a random mutation operation on the scheduling scheme set within this space based on the current real-time demand for spare parts. This simulates scheme disturbances under different scheduling conditions, forming multiple adjacent scheduling solution sets, thus generating the first linkage scheduling neighborhood. The system calls the spare parts scheduling risk analysis model to perform multi-dimensional risk analysis and evaluation on each scheduling scheme in the first linkage scheduling neighborhood, calculating its scheduling conflict risk, scheduling safety risk, and scheduling delay risk respectively. Based on risk constraints, it performs optimization screening to obtain the second linkage scheduling neighborhood after risk constraint optimization. Subsequently, based on the inventory balance evaluation threshold, the system uses the second inventory balance evaluation model optimized by AdamW to evaluate the inventory balance of the second linkage scheduling neighborhood, identifying the set of schemes whose inventory balance evaluation coefficient is greater than or equal to the preset threshold, forming the third linkage scheduling neighborhood. The system then expands and merges the effective schemes of this third linkage scheduling neighborhood with the original third linkage scheduling space, constructing a fourth linkage scheduling space containing more optimal solution candidates. Based on this, the system takes scheduling time response efficiency, inventory resource utilization and multi-warehouse coordination as objective functions, optimizes the spare parts scheduling efficiency in the fourth linkage scheduling space, and comprehensively adopts a multi-objective convergence algorithm to determine the optimal scheduling path and task allocation scheme, and finally generates a linkage scheduling optimization strategy.

[0040] In summary, the embodiments of this application have at least the following technical effects: First, multiple spare parts warehouses are monitored in real time to construct a multi-warehouse digital model. Next, based on real-time spare parts demand, the multi-warehouse digital model undergoes multi-dimensional scheduling attention analysis to construct a multi-dimensional scheduling attention graph network. Then, based on real-time spare parts demand, multi-warehouse coordinated scheduling decisions are made using the multi-dimensional scheduling attention graph network to obtain a first coordinated scheduling space. The first coordinated scheduling space is then optimized using a spare parts scheduling risk analysis model to obtain a second coordinated scheduling space. The second coordinated scheduling space is then optimized using an inventory balance evaluation record set to obtain a third coordinated scheduling space. Finally, the third coordinated scheduling space is optimized through neighborhood expansion using the spare parts scheduling risk analysis model to obtain a coordinated scheduling optimization strategy. This strategy is then used to execute multi-warehouse coordinated scheduling control across multiple spare parts warehouses. This approach solves the technical problems of isolated scheduling decisions among multiple warehouses in existing technologies, which prevent dynamic optimization based on real-time demand, leading to delayed spare parts supply response and inventory imbalance. It achieves the technical effect of improving the real-time performance, balance, and intelligence of multi-warehouse coordinated scheduling.

[0041] Example 2 is based on the same inventive concept as the intelligent spare parts storage scheduling and control method for multi-warehouse linkage in the previous example, such as... Figure 2 As shown, this application provides an intelligent spare parts warehousing and scheduling control system for multi-warehouse linkage, wherein the system includes: Digital model construction module 11: Real-time monitoring of multiple spare parts warehouses to construct a multi-warehouse digital model; Analysis module 12: Multi-dimensional scheduling attention analysis of the multi-warehouse digital model based on real-time spare parts demand to construct a multi-dimensional scheduling attention graph network; Scheduling decision module 13: Based on the real-time spare parts demand, multi-warehouse linkage scheduling decision of the multi-warehouse digital model according to the multi-dimensional scheduling attention graph network to obtain a first linkage scheduling space; First optimization module 14: Optimization of the first linkage scheduling space based on the spare parts scheduling risk analysis model to obtain a second linkage scheduling space; Second optimization module 15: Optimization of the second linkage scheduling space based on the inventory balance evaluation record set to obtain a third linkage scheduling space; Scheduling control module 16: Neighborhood expansion optimization of the third linkage scheduling space based on the spare parts scheduling risk analysis model to obtain a linkage scheduling optimization strategy, and execution of multi-warehouse linkage scheduling control of the multiple spare parts warehouses according to the linkage scheduling optimization strategy.

[0042] Furthermore, the parsing module 12 is used to perform the following methods: Based on the real-time spare parts demand, the multi-database digital model is subjected to inventory matching attention analysis to establish an inventory matching attention graph network; based on the real-time spare parts demand, the multi-database digital model is subjected to scheduling distance attention analysis to establish a scheduling distance attention graph network; based on the real-time spare parts demand, the multi-database digital model is subjected to emergency matching degree attention analysis to establish an emergency matching degree attention graph network; attention is aggregated based on the inventory matching attention graph network, the scheduling distance attention graph network, and the emergency matching degree attention graph network to obtain a comprehensive scheduling attention graph network; based on the inventory matching attention graph network, the scheduling distance attention graph network, the emergency matching degree attention graph network, and the comprehensive scheduling attention graph network, the multi-dimensional scheduling attention graph network is generated.

[0043] Furthermore, the parsing module 12 is used to perform the following methods: Based on the multi-database digital model, the available spare parts inventory features of the multiple spare parts warehouses are analyzed to obtain multiple available spare parts inventory vectors; based on the real-time spare parts demand, spare parts demand features are analyzed to obtain spare parts demand feature vectors; based on the spare parts demand feature vectors, the multiple available spare parts inventory vectors are evaluated for similarity to obtain a demand-inventory similarity evaluation sequence; based on the demand-inventory similarity evaluation sequence, inventory matching attention is allocated to the multi-database digital model to generate the inventory matching attention graph network.

[0044] Furthermore, the scheduling decision module 13 is used to execute the following method: Based on the real-time spare parts demand, a first coordinated scheduling decision group is obtained by performing multi-database coordinated scheduling decisions on the multi-database digital model according to the inventory matching attention graph network; a second coordinated scheduling decision group is obtained by performing multi-database coordinated scheduling decisions on the multi-database digital model according to the scheduling distance attention graph network; a third coordinated scheduling decision group is obtained by performing multi-database coordinated scheduling decisions on the multi-database digital model according to the emergency matching degree attention graph network; a fourth coordinated scheduling decision group is obtained by performing multi-database coordinated scheduling decisions on the multi-database digital model according to the real-time spare parts demand and the scheduling comprehensive attention graph network; the first, second, third, and fourth coordinated scheduling decision groups are added to the first coordinated scheduling space.

[0045] Furthermore, the first optimization module 14 is used to perform the following method: Simulation scheduling is performed on the multi-database digital model based on each linkage scheduling decision within the first linkage scheduling space to obtain simulation data for each decision scheduling; the spare parts scheduling risk analysis model is activated, which includes multi-dimensional scheduling risk indicators, including scheduling conflict risk, scheduling safety risk, and scheduling delay risk; the simulation data for each decision scheduling is input into the spare parts scheduling risk analysis model to obtain multiple scheduling risk analysis sequences; scheduling risk optimization constraints are obtained, including scheduling conflict risk constraints, scheduling safety risk constraints, and scheduling delay risk constraints; the first linkage scheduling space is identified through the multiple scheduling risk analysis sequences and the scheduling risk optimization constraints to generate the second linkage scheduling space.

[0046] Furthermore, the second optimization module 15 is used to perform the following method: Based on the second linkage scheduling space, the Pth linkage scheduling decision is extracted, and the Pth inventory feedback simulation data corresponding to the Pth linkage scheduling decision is loaded, where P is a positive integer; supervised training is performed based on the inventory balance evaluation record set to obtain the first inventory balance evaluation model; decoupled parameter optimization is performed on the first inventory balance evaluation model using the AdamW optimizer to obtain the second inventory balance evaluation model; the Pth inventory feedback simulation data is input into the second inventory balance evaluation model to obtain the Pth inventory balance evaluation coefficient; if the Pth inventory balance evaluation coefficient is greater than or equal to the inventory balance evaluation threshold, the Pth linkage scheduling decision is added to the third linkage scheduling space.

[0047] Furthermore, the scheduling control module 16 is used to execute the following methods: Based on the real-time demand for spare parts, the third linkage scheduling space is randomly mutated to obtain a first linkage scheduling neighborhood; the first linkage scheduling neighborhood is optimized for spare parts scheduling risk analysis according to the spare parts scheduling risk analysis model to obtain a second linkage scheduling neighborhood; based on the inventory balance evaluation threshold, the second linkage scheduling neighborhood is optimized for inventory balance evaluation according to the second inventory balance evaluation model to obtain a third linkage scheduling neighborhood; the third linkage scheduling space is expanded according to the third linkage scheduling neighborhood to obtain a fourth linkage scheduling space; the fourth linkage scheduling space is optimized for maximizing spare parts scheduling efficiency to generate the linkage scheduling optimization strategy.

[0048] Furthermore, the scheduling control module 16 is used to execute the following methods: Each spare parts warehouse is equipped with an intelligent access control system and an electronic tag system.

[0049] Furthermore, the scheduling control module 16 is used to execute the following methods: Each storage location in the multiple spare parts warehouses is equipped with an electronic ink bottle.

[0050] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for intelligent spare parts warehousing scheduling and control oriented towards multi-warehouse linkage, characterized in that, The method includes: Real-time monitoring of multiple spare parts warehouses to build a multi-warehouse digital model; The multi-warehouse digital model consists of multiple warehouse sub-models, each of which includes an inventory status sub-layer, a transfer capacity sub-layer, and a transportation timeliness sub-layer. By associating and mapping the various warehouse sub-models in the cloud, a global warehouse digital twin containing information on spatial distribution, inventory structure, and scheduling capabilities is constructed. Based on the real-time demand for spare parts, the multi-database digital model is subjected to multi-dimensional scheduling attention analysis to construct a multi-dimensional scheduling attention graph network. Based on the real-time demand for spare parts, the multi-database linkage scheduling decision is made on the multi-database digital model according to the multi-dimensional scheduling attention graph network to obtain the first linkage scheduling space. Based on the spare parts scheduling risk analysis model, the first linkage scheduling space is optimized by performing spare parts scheduling risk analysis to obtain the second linkage scheduling space; Based on the inventory balance evaluation record set, the second linkage scheduling space is evaluated for inventory balance and optimized to obtain the third linkage scheduling space; The third linkage scheduling space is expanded and optimized according to the spare parts scheduling risk analysis model to obtain the linkage scheduling optimization strategy, and the multi-warehouse linkage scheduling control of the multiple spare parts warehouses is executed according to the linkage scheduling optimization strategy.

2. The intelligent spare parts storage scheduling and control method for multi-warehouse linkage as described in claim 1, characterized in that, Based on the real-time demand for spare parts, the multi-database digital model is subjected to multi-dimensional scheduling attention analysis to construct a multi-dimensional scheduling attention graph network, including: Based on the real-time demand for spare parts, inventory matching attention analysis is performed on the multi-database digital model to establish an inventory matching attention graph network. Based on the real-time demand for spare parts, the scheduling distance attention analysis is performed on the multi-database digital model to establish a scheduling distance attention graph network; Based on the real-time demand for spare parts, perform emergency matching degree attention analysis on the multi-database digital model and establish an emergency matching degree attention graph network. Attention is aggregated based on the inventory matching attention graph, the scheduling distance attention graph, and the emergency matching degree attention graph to obtain a comprehensive scheduling attention graph. The multidimensional scheduling attention graph is generated based on the inventory matching attention graph, the scheduling distance attention graph, the emergency matching degree attention graph, and the scheduling comprehensive attention graph.

3. The intelligent spare parts storage scheduling and control method for multi-warehouse linkage as described in claim 2, characterized in that, Based on the real-time demand for spare parts, inventory matching attention analysis is performed on the multi-database digital model to establish an inventory matching attention graph network, including: Based on the multi-database digital model, the available spare parts inventory features of the multiple spare parts warehouses are analyzed to obtain multiple available spare parts inventory vectors. Based on the real-time spare parts demand, spare parts demand feature analysis is performed to obtain a spare parts demand feature vector; Based on the spare parts demand feature vector, the similarity evaluation of the multiple available spare parts inventory vectors is performed to obtain a demand inventory similarity evaluation sequence. Based on the demand-inventory similarity evaluation sequence, inventory matching attention is allocated to the multi-database digital model to generate the inventory matching attention graph network.

4. The intelligent spare parts warehousing scheduling and control method for multi-warehouse linkage as described in claim 1, characterized in that, Based on the real-time demand for spare parts, a first coordinated scheduling space is obtained by performing multi-database coordinated scheduling decisions on the multi-database digital model according to the multi-dimensional scheduling attention graph network, including: Based on the real-time demand for spare parts, the multi-database linkage scheduling decision is made on the multi-database digital model according to the inventory matching attention graph network to obtain the first linkage scheduling decision group. Based on the real-time demand for spare parts, a second linkage scheduling decision group is obtained by performing multi-database linkage scheduling decisions on the multi-database digital model according to the scheduling distance attention graph network. Based on the real-time demand for spare parts, the multi-database linkage scheduling decision is made on the multi-database digital model according to the emergency matching degree attention graph network to obtain the third linkage scheduling decision group. Based on the real-time demand for spare parts, a fourth linkage scheduling decision group is obtained by performing multi-database linkage scheduling decisions on the multi-database digital model according to the scheduling comprehensive attention graph network. Add the first joint scheduling decision group, the second joint scheduling decision group, the third joint scheduling decision group, and the fourth joint scheduling decision group to the first joint scheduling space.

5. The intelligent spare parts storage scheduling and control method for multi-warehouse linkage as described in claim 1, characterized in that, Based on the spare parts scheduling risk analysis model, the first linkage scheduling space is optimized through spare parts scheduling risk analysis to obtain the second linkage scheduling space, including: Based on each linkage scheduling decision within the first linkage scheduling space, the multi-database digital model is simulated and scheduled to obtain simulation data for each decision scheduling; Activate the spare parts scheduling risk analysis model, which includes multi-dimensional scheduling risk indicators, including scheduling conflict risk, scheduling security risk, and scheduling delay risk; The decision-making and scheduling simulation data are input into the spare parts scheduling risk analysis model to obtain multiple scheduling risk analysis sequences. Obtain scheduling risk optimization constraints, which include scheduling conflict risk constraints, scheduling security risk constraints, and scheduling delay risk constraints; The first linked scheduling space is identified by the multiple scheduling risk parsing sequences and the scheduling risk optimization constraint, and the second linked scheduling space is generated.

6. The intelligent spare parts warehousing scheduling and control method for multi-warehouse linkage as described in claim 1, characterized in that, Based on the inventory balance evaluation record set, the second linkage scheduling space is optimized through inventory balance evaluation to obtain the third linkage scheduling space, including: Based on the second linkage scheduling space, extract the Pth linkage scheduling decision and load the Pth inventory feedback simulation data corresponding to the Pth linkage scheduling decision, where P is a positive integer; Supervised training is performed based on the aforementioned inventory balance evaluation record set to obtain the first inventory balance evaluation model; The first inventory balance evaluation model is decoupled and optimized using the AdamW optimizer to obtain the second inventory balance evaluation model. Input the Pth inventory feedback simulation data into the second inventory equilibrium evaluation model to obtain the Pth inventory equilibrium evaluation coefficient. If the Pth inventory balance evaluation coefficient is greater than or equal to the inventory balance evaluation threshold, the Pth linkage scheduling decision is added to the third linkage scheduling space.

7. The intelligent spare parts storage scheduling and control method for multi-warehouse linkage as described in claim 1, characterized in that, Based on the aforementioned spare parts scheduling risk analysis model, the third coordinated scheduling space is expanded and optimized to obtain a coordinated scheduling optimization strategy, including: Based on the real-time demand for spare parts, the third linkage scheduling space is randomly mutated to obtain the first linkage scheduling neighborhood. Based on the spare parts scheduling risk analysis model, the first joint scheduling neighborhood is analyzed and optimized to obtain the second joint scheduling neighborhood. Based on the inventory balance evaluation threshold, the second linkage scheduling neighborhood is optimized by performing inventory balance evaluation on the second linkage scheduling neighborhood according to the second inventory balance evaluation model to obtain the third linkage scheduling neighborhood. The third linkage scheduling space is expanded according to the third linkage scheduling neighborhood to obtain the fourth linkage scheduling space; The fourth linkage scheduling space is optimized to maximize the efficiency of spare parts scheduling, and the linkage scheduling optimization strategy is generated.

8. The intelligent spare parts storage scheduling and control method for multi-warehouse linkage as described in claim 1, characterized in that, Each spare parts warehouse is equipped with an intelligent access control system and an electronic tag system.

9. The intelligent spare parts storage scheduling and control method for multi-warehouse linkage as described in claim 1, characterized in that, Each storage location in the multiple spare parts warehouses is equipped with an e-ink screen.

10. An intelligent spare parts warehousing and scheduling control system for multi-warehouse linkage, characterized in that, The system is used to implement the intelligent spare parts storage scheduling and control method for multi-warehouse linkage as described in any one of claims 1-9, the system comprising: Digital model building module: Real-time monitoring of multiple spare parts warehouses to build multi-warehouse digital models; Analysis module: Performs multi-dimensional scheduling attention analysis on the multi-database digital model based on real-time spare parts requirements, and constructs a multi-dimensional scheduling attention graph network; Scheduling decision module: Based on the real-time demand for spare parts, the module performs multi-database linkage scheduling decisions on the multi-database digital model according to the multi-dimensional scheduling attention graph network to obtain the first linkage scheduling space; First optimization module: Based on the spare parts scheduling risk analysis model, perform spare parts scheduling risk analysis and optimization on the first linkage scheduling space to obtain the second linkage scheduling space; The second optimization module: Based on the inventory balance evaluation record set, the second linkage scheduling space is evaluated and optimized to obtain the third linkage scheduling space; Scheduling control module: Based on the spare parts scheduling risk analysis model, the module performs neighborhood expansion optimization on the third linkage scheduling space to obtain a linkage scheduling optimization strategy, and executes multi-warehouse linkage scheduling control of the multiple spare parts warehouses according to the linkage scheduling optimization strategy.

Images