A spare parts warehouse location dynamic allocation method based on multi-objective optimization

Abstract

This invention relates to the field of intelligent warehousing and logistics automation technology, specifically to a method for dynamic allocation of spare parts storage locations based on multi-objective optimization. The method includes: constructing a digital twin model of the spare parts warehouse; acquiring multi-dimensional attribute information of spare parts to be received; generating initial storage location coordinates based on a multi-objective location allocation model, which comprehensively considers four objectives: inbound / outbound efficiency, clustered storage, shelf stability, and specification matching, and supports dynamic location splicing and stacking; continuously monitoring the warehouse status, triggering location optimization adjustments when dynamic thresholds are reached, and periodically updating spare parts frequency levels and storage hot zones using an ABC classification dynamic adjustment mechanism. This invention solves problems such as low space utilization in spare parts warehouses, remote storage of high-frequency spare parts, and difficulties in adapting storage for irregularly shaped parts, achieving adaptive dynamic allocation of storage locations and significantly improving the overall throughput of the warehouse.

Description

Technical Field

[0001] This invention relates to the field of intelligent warehousing and logistics automation technology, and in particular to a method for dynamic allocation of spare parts storage locations based on multi-objective optimization. Background Technology

[0002] In modern industrial production, sponge titanium production enterprises employ a wide variety of equipment and complex technologies, resulting in spare parts warehouses with significant industry-specific characteristics: First, the spare parts are diverse, including mechanical, electrical, instrument, and general spare parts, with varying specifications, sizes, and shapes; second, the frequency of use varies greatly, with some critical equipment spare parts being infrequently used but of high value and urgently needed; third, some spare parts have special requirements for storage environments, such as moisture-proof, dust-proof, and corrosion-proof; and fourth, spare parts warehouses have long faced the problem of "stagnant materials" occupying valuable storage space.

[0003] Traditional spare parts warehouse management often employs static location allocation or simple classification storage models, which present the following technical challenges in actual operation: First, spare parts are diverse in type and specifications, making it difficult for static locations to accommodate spare parts packaging of different sizes, resulting in low space utilization. Second, differences in spare parts usage frequency are not reflected in location allocation, with frequently used spare parts often stored in remote locations, affecting maintenance efficiency. Third, there is a lack of dynamic adjustment mechanisms; spare parts remain fixed in one location after entering the warehouse, and as the inventory structure changes, the original location allocation gradually deviates from the optimal state. Fourth, uneven weight distribution and haphazard stacking of heavy spare parts pose safety hazards.

[0004] Publication No.: CN118154110A A method and apparatus for allocating storage locations in an automated warehouse, and a warehouse management system, comprising: acquiring target cargo information for goods to be stored; optimizing a pre-constructed storage location allocation model based on the target cargo information to obtain the proposed storage location for the goods to be stored, wherein the storage location allocation model is determined based on an inbound / outbound efficiency model, a clustered storage model, and a shelf stability model. However, this technical solution cannot cope with complex industrial scenarios in spare parts warehouses, such as large differences in spare parts size, dynamic changes in usage frequency, the presence of irregularly shaped and small parts, and the need for continuous monitoring and adjustment.

[0005] Therefore, there is an urgent need for a dynamic allocation method for spare parts storage locations that can combine multi-objective optimization with the actual working conditions of spare parts warehouses in industrial enterprises, in order to solve the problems of low space utilization and low operational efficiency in existing spare parts warehouses. Summary of the Invention

[0006] In view of this, the present invention aims to propose a dynamic allocation method for spare parts storage locations based on multi-objective optimization, so as to solve the problems of low space utilization and low operational efficiency in existing spare parts warehouses.

[0007] The technical solution of this invention is implemented as follows:

[0008] This invention discloses a method for dynamic allocation of spare parts storage locations based on multi-objective optimization, comprising the following steps:

[0009] Establish a digital mapping model for the spare parts warehouse and keep it synchronized with the physical warehouse.

[0010] Obtain multi-dimensional attribute information of spare parts to be put into the warehouse;

[0011] Based on a multi-objective optimization strategy, the multi-dimensional attribute information is used as input to solve for the generation of storage locations; the multi-objective optimization strategy includes at least the optimization objectives of inbound and outbound efficiency, spare parts clustering and storage, shelf stability, and spare parts specification adaptation, and the weighted combination of each optimization objective is used as the overall optimization function;

[0012] During spare parts storage, the warehouse status is monitored. When the monitoring results meet the preset trigger conditions, the warehouse location is optimized and adjusted, and the category level of the spare parts and its corresponding storage area are updated according to the spare parts classification dynamic adjustment mechanism.

[0013] Optionally, the digital mapping model is a digital twin model; the establishment steps specifically include:

[0014] Collect physical structure data of the warehouse, geometric dimensions of storage locations, load-bearing parameters of shelves, and historical inbound and outbound operation data;

[0015] Import 3D modeling software to build a digital twin model that is synchronized with the physical warehouse in real time.

[0016] Optionally, the multi-dimensional attribute information includes one or more of the following: spare parts category, specifications and dimensions, weight, storage environment requirements, usage frequency level dynamically calculated based on historical outbound data, and correlation between spare parts.

[0017] Optionally, the objective of optimizing inbound and outbound efficiency is: for all spare parts to be put into storage, calculate the sum of the products of the usage frequency weight of each spare part and the weighted distance from its target storage location to the warehouse entrance and exit, and minimize this sum.

[0018] The usage frequency weight is set to different values ​​according to the spare parts category level, and the weighted distance takes into account the actual driving path of the AGV and the real-time path congestion coefficient.

[0019] Optionally, the optimization objective of the spare parts clustering storage is: for all spare parts pairs and their corresponding storage location pairs, calculate the product of the correlation degree of each spare parts pair and the distance between the corresponding storage locations, and then minimize the sum after summing.

[0020] The correlation degree is determined based on the number of times two spare parts appear simultaneously in the same work task in historical data.

[0021] Optionally, the spare parts specification adaptation optimization objectives include: determining whether the spare parts packaging size exceeds the available storage capacity, and constraining that the excess must be zero to ensure that allocation is feasible;

[0022] When the size of a spare part is larger than a single storage location, it automatically searches for adjacent free storage locations and dynamically splices them together, logically merging multiple storage locations into a combined storage location.

[0023] When the size of a spare part is smaller than that of a single storage location, the remaining space in the storage location already occupied by spare parts of the same category will be retrieved first and then stacked.

[0024] Optionally, after the dynamic splicing is completed, the combined storage location is logically marked as a whole; when all spare parts in the combined storage location are taken out, the splicing is automatically de-spliced ​​and restored to an independent storage location.

[0025] Optionally, the preset triggering conditions include at least one of the following: the idle storage space rate is lower than a first threshold, the increase in the average storage distance of high-frequency spare parts exceeds a second threshold, or an outbound task order marked as urgent or major overhaul is received.

[0026] When any trigger condition is met, local or global storage location optimization adjustments will be automatically triggered.

[0027] Optionally, the spare parts classification dynamic adjustment mechanism includes: periodically recalculating the category level of all spare parts based on the latest historical outbound data;

[0028] Divide the physical area of ​​the warehouse into at least two heat zones;

[0029] When the category level of spare parts changes, a relocation suggestion is automatically generated to move the spare parts from the original area to a suitable storage location in the corresponding area, and the relocation operation is performed during off-peak hours.

[0030] Optionally, the process of solving for and generating the storage location specifically includes:

[0031] Calculate the distance from all storage locations to entrances and exits in advance, and store a list of available storage locations by category and level.

[0032] For each incoming spare part, candidate regions are determined based on its category level;

[0033] Within the candidate area, feasible storage locations are selected based on specification compatibility and shelf stability constraints.

[0034] The overall optimization function is used to evaluate and select the optimal storage location, generate an inbound instruction containing the coordinates of the target storage location, and send it to the execution terminal.

[0035] Compared with existing technologies, the present invention provides a method for dynamic allocation of spare parts storage locations based on multi-objective optimization, which has the following advantages:

[0036] 1. This invention generates storage locations through a digital twin model and a multi-objective optimization strategy. It combines four objectives—inbound / outbound efficiency, clustered storage, shelf stability, and specification adaptation—in a weighted combination. This allows high-frequency spare parts to be located near entrances / exits, related spare parts to be stored in clusters, and heavy spare parts to be placed on lower levels. Furthermore, it solves the storage problems of irregularly shaped and small parts through dynamic splicing and stacking. This addresses the issues of low space utilization, remote location of high-frequency spare parts, and potential safety hazards associated with traditional static allocation.

[0037] 2. This invention uses a matching threshold-triggered adjustment and ABC classification hot zone migration mechanism to automatically re-optimize when the idle storage space rate or the distance of high-frequency spare parts exceeds the threshold, and regularly updates the spare parts classification and migrates them to the corresponding hot zones. Storage space allocation can be automatically completed by scanning barcodes, increasing throughput by about 15%-19%, which is significantly better than the static allocation method. Attached Figure Description

[0038] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0039] Figure 1 Flowchart for a method to dynamically allocate storage locations for spare parts. Detailed Implementation

[0040] To make the technical means and objectives and effects of the present invention easier to understand, the embodiments of the present invention will be described in detail below with reference to specific illustrations.

[0041] Example 1

[0042] like Figure 1 As shown, this embodiment provides a method for dynamic allocation of spare parts storage locations based on multi-objective optimization. This method is particularly suitable for spare parts warehouses in industrial enterprises, such as, but not limited to, spare parts warehouses in sponge titanium production enterprises, machinery manufacturing enterprises, power enterprises, or chemical enterprises with a wide variety of types, specifications, and usage frequencies.

[0043] The method includes the following steps:

[0044] Establish a digital mapping model for the spare parts warehouse and keep it synchronized with the physical warehouse.

[0045] By establishing a digital mapping model that is synchronized with the physical warehouse in real time, the optimization algorithm always makes decisions based on the latest status, eliminating the problems of information lag and discrepancies between drawings and the actual situation in traditional management, and ensuring the timeliness and accuracy of warehouse location allocation.

[0046] Obtain multi-dimensional attribute information of spare parts to be put into the warehouse;

[0047] By acquiring multi-dimensional attribute information of spare parts, features such as category, size, weight, frequency of use, and relevance are quantified into model inputs. This makes warehouse location allocation no longer a "one-size-fits-all" general rule, but a personalized match based on the characteristics of each spare part. High-frequency spare parts, related spare parts, heavy spare parts, and irregularly shaped spare parts can all obtain warehouse locations that are suitable for their characteristics.

[0048] Based on a multi-objective optimization strategy, and using multi-dimensional attribute information as input, the solution is used to generate the storage location.

[0049] A multi-objective optimization strategy is employed to weight and combine four objectives—inbound / outbound efficiency, clustered storage, shelf stability, and specification adaptation—to find the optimal storage location while satisfying all constraints. Specifically, the inbound / outbound efficiency objective places frequently used spare parts near entrances and exits; the clustered storage objective physically groups related spare parts together; the shelf stability objective places heavy spare parts on lower shelves to eliminate safety hazards; and the specification adaptation objective solves the storage challenges of irregularly shaped and small parts through dynamic splicing and stacking. The synergistic optimization of these four objectives avoids the "one-sided" problems of traditional single-objective or asymmetric solutions.

[0050] During spare parts storage, the warehouse status is monitored. When the monitoring results meet the preset trigger conditions, the warehouse location is optimized and adjusted, and the category level of the spare parts and its corresponding storage area are updated according to the spare parts classification dynamic adjustment mechanism.

[0051] Through a threshold-triggered dynamic adjustment mechanism, re-optimization is automatically triggered when the idle storage space rate is too low, the distance of high-frequency spare parts is too large, or an emergency task arrives, thus avoiding the waste of computing resources caused by over-optimization. Through an ABC classification hot zone migration mechanism, the spare parts classification level is updated regularly and automatically migrated to the corresponding hot zone, so that the storage space allocation can adaptively adjust with the changes in the frequency of spare parts use.

[0052] The following section provides a detailed explanation of the specific implementation method.

[0053] S1: Construct a digital mapping model;

[0054] In this step, a digital model that is mapped in real time to the physical warehouse is first established as the basis for subsequent optimization calculations. The preferred digital mapping model is a digital twin model, but other forms of real-time synchronization models, such as database mirroring models or real-time simulation models, can also be used.

[0055] Specifically, at the spare parts warehouse, data is collected in the following ways: A 3D laser scanner is used to scan the warehouse building structure, obtaining physical structural data such as the warehouse's length, width, and height, column locations, entrance / exit coordinates, and aisle width; the layout of the shelving is measured, including the number of rows, columns, and layers, as well as the spacing between each row; each standard storage location is precisely measured, recording its length, width, height, and load-bearing capacity; for adjustable shelving, its adjustment range is recorded; for high-density intelligent shelving, its movement speed, acceleration, and energy consumption are also recorded; at least one year of historical inbound / outbound operation data is exported from the enterprise resource planning system or warehouse management system, including the spare parts code, operation time, operator, and storage location changes for each operation. This data can be used to analyze the frequency of spare parts usage and correlate outbound patterns.

[0056] After being cleaned and formatted, the collected data is imported into 3D modeling software to construct a digital twin model of the warehouse. This model includes not only static geometric information but also dynamic attributes, such as, but not limited to, current inventory status, AGV real-time position, and task queues. The digital twin model interacts with the field control system in real time via industrial Ethernet, ensuring that the model's status is synchronized with the physical warehouse, serving as the real-time mapping basis for subsequent optimization calculations.

[0057] Traditional static warehouse models fail to reflect real-time changes in inventory status, leading to a disconnect between optimization results and actual operating conditions. This new setup, by constructing a digital model synchronized with the physical warehouse in real time, ensures that subsequent optimization algorithms are always calculated based on the latest warehouse status, thereby significantly improving the accuracy and timeliness of warehouse location allocation.

[0058] S2: Obtain multi-dimensional attribute information of spare parts to be put into the warehouse;

[0059] When a new batch of spare parts arrives at the warehouse receiving area, its multi-dimensional attribute information is obtained through the following methods. This multi-dimensional attribute information includes, but is not limited to, one or more of the following: spare part category, specifications and dimensions, weight, storage environment requirements, usage frequency level dynamically calculated based on historical outbound data, and the correlation between spare parts.

[0060] Specifically, operators use handheld terminals to scan barcodes or RFID tags on spare parts packaging, and the system automatically parses the spare parts code. Then, based on the spare parts code, the system retrieves the static attributes of the spare parts from the ERP management system, including: spare parts category (e.g., mechanical spare parts, electrical spare parts, instrument spare parts, general spare parts, etc.), dimensions (length, width, height, units can be selected as millimeters or centimeters), weight (units can be selected as kilograms or grams), storage environment requirements (e.g., moisture-proof, dust-proof, corrosion-proof, temperature and humidity range, etc.), and batch information. Batch information includes production batch, warehousing date, supplier information, etc.

[0061] Meanwhile, the system dynamically calculates the usage frequency level of spare parts based on historical data. In one specific implementation, by analyzing the outbound records of the past 12 months, spare parts are divided into Category A (high frequency, top 20% of outbound frequency), Category B (medium frequency, middle 30%), and Category C (low frequency, bottom 50%). This level can be set to be updated weekly, monthly, or quarterly. For new spare parts entering the warehouse for the first time, the system assigns an initial frequency level based on its category and the usage of similar spare parts, for example, setting it to Category B by default, and then dynamically adjusting it after accumulating sufficient data.

[0062] In addition, the system also calculates the correlation between spare parts. For example, it counts the number of times two spare parts appear simultaneously in the same outbound order over the past year, normalizes the count to the range of 0 to 1, and uses this as the correlation between the two spare parts. If two spare parts have never been outbound at the same time but belong to the same category, a basic correlation score is assigned, such as 0.2; if they belong to different categories and have no simultaneous outbound records, the correlation score is 0.

[0063] The multi-dimensional attribute information obtained above is combined into a data vector, which serves as the input for the subsequent optimization model.

[0064] S3: Solving for storage locations based on a multi-objective optimization strategy;

[0065] This step is the core optimization process. Based on the acquired multi-dimensional attribute information of the spare parts, the system generates the coordinates of the storage location for inbound goods using a multi-objective optimization strategy. The multi-objective optimization strategy includes at least the following objectives: inbound / outbound efficiency optimization, spare parts clustering and storage optimization, shelf stability optimization, and spare parts specification adaptation optimization. The weighted combination of these objectives serves as the overall optimization function.

[0066] in:

[0067] The goal is to optimize inbound and outbound efficiency.

[0068] The goal of optimizing inbound and outbound efficiency is to calculate the sum of the products of the usage frequency weight of each spare part and the weighted distance from its target storage location to the warehouse entrance and exit for all spare parts to be put into storage, and minimize this sum.

[0069] The usage frequency weight is set to different values ​​according to the spare parts category level. For example, but not limited to: a weight of 3 for category A spare parts, a weight of 2 for category B spare parts, and a weight of 1 for category C spare parts. The weighted distance calculation considers the actual travel path of the AGV rather than a simple Euclidean distance, and also introduces a path congestion coefficient. The path congestion coefficient can be dynamically adjusted according to the real-time number of AGV tasks. For example, when the number of tasks being executed on a certain path exceeds 3, the congestion coefficient increases by 0.2.

[0070] Preferably, the inbound / outbound efficiency target f1 is:

[0071] ;

[0072] Where w i Assign a usage frequency weight to spare part i (3 for class A, 2 for class B, and 1 for class C), d j This is the weighted distance from storage location j to the main entrance / exit. The distance calculation considers the actual travel path of the AGV, rather than the Euclidean distance, and also incorporates a path congestion coefficient (dynamically adjusted based on the real-time number of AGV tasks). Minimizing f1 allows high-frequency spare parts to be placed closer to the entrance / exit, improving outbound efficiency.

[0073] Optimization goal for spare parts clustering and storage;

[0074] The optimization objective of spare parts clustering storage is to calculate the product of the correlation degree of each spare parts pair and the distance between the corresponding storage locations for all spare parts pairs and their corresponding storage locations, and then minimize this sum.

[0075] The Manhattan distance is preferred for the distance between storage locations, but Euclidean distance or other distance metrics can also be used. The correlation degree is determined based on the number of times two spare parts appear simultaneously in the same task in historical data, calculated as described above. Minimizing this objective allows spare parts with high correlation to be physically close together, thereby reducing picking path length and improving outbound efficiency.

[0076] Preferably, the spare parts cluster storage target f2:

[0077] ;

[0078] Where sim(i,k) is the correlation between spare part i and spare part k, calculated as follows: based on historical data, the number of times they appear simultaneously in the same outbound task order is counted and normalized to the interval [0,1]. If they have never been outbound simultaneously but are of the same category, a basic correlation score (e.g., 0.2) is assigned; if they are of different categories and have no simultaneous outbound records, the score is 0. dist(j,l) is the Manhattan distance between storage location j and storage location l. Minimizing f2 can make spare parts with high correlation physically close together, reducing picking paths.

[0079] Shelf stability optimization objectives;

[0080] The optimization objective for shelf stability is as follows: For each storage location, calculate the sum of the products of the weight of all spare parts stored in that location and the height level of the storage location, then square this sum, sum the squared values ​​of all storage locations, and minimize this total sum.

[0081] The storage location height levels increase progressively from the bottom, for example, the bottom level is 0, the second level is 1, the third level is 2, and so on. This goal encourages the placement of heavy spare parts on lower levels to avoid excessively high center of gravity on the shelves, thereby improving the safety of warehousing operations.

[0082] Preferably, the shelf stability target f3 is:

[0083] ;

[0084] height j This represents the height level of storage location j (0 for the bottom level, increasing with each subsequent level). This objective encourages the placement of heavy spare parts on lower levels to avoid excessively high center of gravity on the shelving.

[0085] Optimization goals for spare parts specification adaptation;

[0086] The optimization goals for spare parts specification adaptation include: determining whether the spare parts packaging size exceeds the available capacity of the storage location, and constraining that the excess must be zero to ensure that allocation is feasible.

[0087] To address the mismatch between spare parts specifications and storage location dimensions, this method introduces a dynamic splicing and stacking strategy. Specifically:

[0088] When a spare part's size exceeds the size of a single storage location, the system automatically searches for adjacent available storage locations and dynamically combines them into a single combined storage location. For example, when the length of a spare part exceeds the length of a single standard storage location, the system searches for two or more adjacent available storage locations in the horizontal direction and combines them into a single combined storage location. The total length of the combined location is equal to the sum of the lengths of its individual sub-locations. The combined storage location is logically marked as a single unit, and its available capacity is the sum of the capacities of its individual sub-locations.

[0089] When a spare part is smaller than a single storage location, the system prioritizes searching for remaining space in storage locations already occupied by spare parts of the same type and stacking them. For example, when a small electrical spare part is received, the system first checks if there is remaining space in storage locations that already contain similar spare parts. If so, and if the weight-bearing capacity allows, the spare part is stacked and stored in that storage location, thus avoiding wasting storage space.

[0090] Preferably, the spare parts specifications match target f4:

[0091] ;

[0092] Where size i Spare parts packaging dimensions (vector), capacity j Let f4 be the available capacity of storage location j. If spare parts cannot be placed in the storage location, f4 is positive infinity, indicating that the allocation is not feasible. Simultaneously, the model allows dynamic concatenation; that is, when the size of a spare part is larger than a single storage location, the system automatically searches for adjacent free storage locations and logically merges multiple storage locations into a single combined storage location. In this case, the capacity... j This represents the total capacity after merging; if splicing is not enabled, the corresponding penalty is extremely high.

[0093] Overall optimization function;

[0094] The overall optimization function is a weighted combination of the four optimization objectives mentioned above. In one specific implementation, a linear weighted summation form is used:

[0095] The total optimization function value = α × inbound / outbound efficiency optimization target value + β × clustered storage optimization target value + γ × shelf stability optimization target value + δ × specification adaptation optimization target value. That is:

[0096] ;

[0097] The weighting coefficients α, β, γ, and δ can be dynamically adjusted according to management priorities. For example, but not limited to: during normal periods, α can be set to 0.4, β to 0.3, γ to 0.2, and δ to 0.1; when receiving an emergency maintenance task, α can be temporarily increased to 0.7 to improve outbound efficiency; when warehouse stability issues are prominent, γ can be increased to 0.4. The preferred range for the weighting coefficients is between 0 and 1, and the sum of the four weighting coefficients is preferably 1.

[0098] Solve for generating the storage location;

[0099] The specific steps for generating the storage location are as follows:

[0100] The model solver employs a heuristic algorithm combined with a greedy strategy. Considering real-time requirements, the distance from all storage locations to the entrances and exits is pre-calculated, and a list of available storage locations is stored in partitions according to category level. For example, areas near the entrances and exits are designated as high-frequency zones (primarily storing Class A spare parts), intermediate areas as medium-frequency zones (primarily storing Class B spare parts), and distant areas as low-frequency zones (primarily storing Class C spare parts).

[0101] For each incoming spare part, candidate storage areas are first determined based on its category level. For example, Category A spare parts are preferentially screened for feasible storage locations in the high-frequency zone.

[0102] Then, feasible storage locations are selected within the candidate area based on specification compatibility and rack stability constraints. Specification compatibility constraints require that the spare parts can be placed in the storage location (for assembled storage locations, it is required that they can be placed in the combined space); rack stability constraints require that the total weight of the storage location after placement does not exceed the load-bearing limit.

[0103] Finally, the selected feasible storage locations are evaluated using the overall optimization function. The storage location that minimizes the overall optimization function value is selected as the optimal storage location. An inbound instruction containing the coordinates of the target storage location is generated and sent to the AGV scheduling system or the operator's handheld terminal to complete the inbound operation.

[0104] This multi-objective optimization strategy addresses the shortcomings of traditional single-objective optimization methods that only consider inbound and outbound efficiency, which can lead to neglect of clustered storage needs, potential risks to shelf stability, or frequent specification compatibility issues. By weighting and combining four optimization objectives and employing a partitioning and filtering strategy during the solution process, a globally optimal solution can be obtained within an acceptable computation time, thereby achieving a dual improvement in space utilization and operational efficiency.

[0105] S4: Dynamic monitoring and classification adjustment;

[0106] During the spare parts storage period, the system continuously monitors the warehouse status and determines whether to initiate warehouse location optimization adjustments based on preset trigger conditions.

[0107] Preset trigger conditions;

[0108] Preset trigger conditions include, but are not limited to, at least one of the following:

[0109] First, the vacancy rate of storage locations is below a first threshold. For example, when the vacancy rate is below 15%, it indicates that warehouse space is tight, and subsequent warehousing may be difficult due to storage location fragmentation. The system will automatically trigger storage location consolidation, integrating vacant areas by merging fragmented spaces or relocating low-frequency spare parts. The first threshold can be set within the range of 10% to 25%, preferably 15%.

[0110] Second, the increase in the average storage distance of high-frequency spare parts exceeds the second threshold. The system periodically (e.g., daily or weekly) calculates the average distance from the current storage location to the entrance / exit for all Class A spare parts and compares it with the initial baseline value or the baseline value after the last optimization. If the increase exceeds 30%, it indicates that high-frequency spare parts are being pushed to remote locations due to subsequent warehousing, triggering local re-optimization. The second threshold can be set in the range of 20% to 40%, preferably 30%.

[0111] Third, the system receives outbound task orders marked "urgent" or "major overhaul". When the system receives such task orders, it automatically triggers local optimization adjustments to optimize the picking order and path of spare parts involved in the task, ensuring that urgent tasks can be completed quickly.

[0112] When any trigger condition is met, the system automatically initiates local or global storage space optimization adjustments. The magnitude and scope of the adjustment can be determined based on the severity of the trigger condition: for example, global optimization is triggered when the idle storage space rate is below 10%, and local optimization is triggered when it is between 10% and 15%.

[0113] Spare parts classification dynamic adjustment mechanism;

[0114] The spare parts classification dynamic adjustment mechanism includes the following:

[0115] Regularly (e.g., monthly or quarterly), recalculate the category levels of all spare parts based on the latest historical outbound data. For example, recalculate the outbound frequency over the past 12 months and update the A / B / C categories.

[0116] The physical area of ​​the warehouse is divided into at least two heat zones. In one specific implementation, it is divided into three heat zones: a high-frequency zone, a mid-frequency zone, and a low-frequency zone. Preferably, the high-frequency zone is the area near the entrance / exit, within 0 to 20 meters of the entrance / exit; the mid-frequency zone is the middle area, within 20 to 50 meters of the entrance / exit; and the low-frequency zone is the far-end area, more than 50 meters away from the entrance / exit.

[0117] When the category level of a spare part changes, the system automatically generates a relocation suggestion. For example, if a spare part was originally classified as Class C (low frequency) but has been upgraded to Class A due to increased usage recently, the system suggests moving it from the low-frequency zone to a suitable storage location in the high-frequency zone. Conversely, if a Class A spare part has not been used for a long time and has been downgraded to Class C, the system suggests moving it to the low-frequency zone, freeing up prime storage space for other high-frequency spare parts.

[0118] Inventory transfer tasks are typically performed during off-peak hours, such as at night or on weekends, to avoid disrupting normal inbound and outbound operations. After the transfer is complete, the system updates the inventory status and spare parts attribute information in the digital twin model to ensure that subsequent inbound allocations are based on the latest classification results.

[0119] This dynamic adjustment mechanism addresses the issue that the frequency of spare parts usage changes dynamically with production plans and equipment maintenance cycles, which traditional static classification methods cannot adapt to. This leads to the problem that high-frequency spare parts may occupy remote storage locations for a long time, or low-frequency spare parts may occupy prime storage locations for a long time. By periodically recalculating classification levels and automatically migrating spare parts, the allocation of storage locations is always kept in line with the current usage frequency, thereby maintaining high operational efficiency.

[0120] Beneficial effects:

[0121] First, by using a digital mapping model (preferably a digital twin model) to synchronize the physical warehouse status in real time, the optimization algorithm is always based on the latest data, avoiding the lag of traditional static allocation schemes, thereby improving the timeliness and accuracy of warehouse location allocation decisions.

[0122] Second, the multi-objective optimization strategy considers four dimensions simultaneously: inbound / outbound efficiency, clustered storage, shelf stability, and specification adaptation. Among these, the dynamic splicing and stacking strategy within the specification adaptation objective solves the storage adaptation problem for irregularly shaped and small spare parts for the first time. By logically merging adjacent storage locations to form combined storage locations and stacking small parts of the same category, storage location utilization is significantly improved, and tests have shown that it can improve storage capacity utilization.

[0123] Third, the threshold-triggered dynamic adjustment mechanism avoids the waste of computing resources caused by "over-optimization". The system only initiates re-optimization when space is tight (low idle storage space rate), efficiency declines (excessive average storage distance of high-frequency spare parts), or when an emergency task arrives. This achieves a balance between efficiency and computing cost while ensuring warehousing and delivery capabilities, avoiding the system burden caused by continuous optimization.

[0124] Fourth, the dynamic adjustment mechanism for spare parts classification ensures that high-frequency spare parts are always located in prime areas near entrances and exits, changing the passive situation of traditional spare parts warehouses characterized by "fuzzy classification and time-consuming searches." By periodically recalculating frequency levels and automatically migrating spare parts, the allocation of storage locations can adaptively adjust to changes in spare parts usage frequency, thereby maintaining high outbound efficiency in the long term.

[0125] In summary, this invention achieves adaptive dynamic allocation of spare parts storage locations through the synergistic effect of multiple technologies, including digital mapping, multi-objective optimization, dynamic splicing and stacking, threshold triggering adjustment, and dynamic classification migration, providing strong support for equipment maintenance and continuous production in industrial enterprises.

[0126] Example 2

[0127] Compared with Example 1, the difference is that, in addition to 3D laser scanning, at least one of the following methods can be used when establishing the digital mapping model: obtaining 3D point cloud data of the warehouse through drone aerial photography, which is especially suitable for large or high-rise warehouses; obtaining the warehouse structure through a fixed camera array combined with a visual reconstruction algorithm; and manually modeling after manual measurement, which is suitable for small warehouses or renovation projects.

[0128] Furthermore, the data collection period for historical inbound and outbound data is not limited to one year. Different periods, such as 6 months, 18 months, or 24 months, can be selected based on the replacement frequency of spare parts and the operational characteristics of the warehouse. For spare parts with significant seasonal fluctuations, it is preferable to collect data for at least one full year to cover all seasons.

[0129] Example 3

[0130] Compared to Example 1, the difference lies in that, when calculating the spare parts usage frequency level, in addition to the number of outbound shipments, a weighted calculation can also be performed based on the quantity outbound. For example, a single outbound shipment exceeding 10 pieces is defined as a large-value outbound shipment, and a higher weighting coefficient is assigned to it. Specifically, the usage frequency score = number of outbound shipments × basic weight + number of large-value outbound shipments × additional weight, and then the parts are divided into three categories, A, B, and C, based on the scores.

[0131] For new spare parts entering the warehouse for the first time, the initial frequency level can be determined not only by default assignment, but also by prediction based on the similarity of attributes with existing spare parts. For example, features such as category, size range, and supplier of the new spare part can be extracted, and the K nearest neighbor algorithm can be used to find the K most similar spare parts among existing spare parts. The mode or average of the frequency levels of these spare parts can then be taken as the initial frequency level of the new spare part.

[0132] Example 4

[0133] Compared to Example 1, the difference lies in that the dynamic adjustment of weight coefficients in the overall optimization function of the allocation steps can further incorporate machine learning methods. Specifically, the system records operational efficiency indicators after each optimization, including but not limited to average outbound time, total AGV travel distance, and warehouse space utilization. The four weight coefficients are automatically adjusted using a reinforcement learning model (e.g., Q-learning or deep Q-networks), allowing the model to gradually adapt to the actual operational patterns of the warehouse.

[0134] For example, when the system detects multiple consecutive emergency tasks, the model can automatically increase the weight of the inbound / outbound efficiency optimization objective and decrease the weight of the clustered storage optimization objective; when the system detects an increase in shelf stability alarm events, it can automatically increase the weight of the shelf stability optimization objective.

[0135] Example 5

[0136] Compared with Example 1, the difference is that in the dynamic splicing strategy for optimizing spare parts specifications, for irregularly shaped spare parts (such as irregularly shaped mechanical parts), the following improvement scheme can be adopted: the system first obtains the actual outline size and shape of the spare parts through three-dimensional vision scanning, and then automatically matches adjacent storage locations to form L-shaped, U-shaped or more complex combined shapes, rather than being limited to straight-line splicing.

[0137] Meanwhile, within a combined storage location, some sub-locations are allowed to have different heights. For example, a high-level storage location (height level 2) can be combined with a low-level storage location (height level 0), as long as the combined space can accommodate the overall outline of the spare parts. The combined storage location is logically marked as a whole, which can span different physical shelving areas.

[0138] Example 6

[0139] Compared with Example 1, the difference is that in the preset triggering conditions of the adjustment step, the first threshold (idle storage space rate) and the second threshold (average storage distance increase of high-frequency spare parts) can be dynamically adjusted according to the historical operation data of the warehouse, rather than being set as fixed values.

[0140] For example, the first threshold can be set to 70% of the average idle storage space rate over the past week; the second threshold can be set to 1.5 times the standard deviation of the average storage distance of high-frequency spare parts over the past month. This adaptive threshold setting method can better adapt to the operating characteristics of different warehouses and different seasons.

[0141] When an emergency task is triggered, the system can temporarily bypass some constraints (such as temporarily ignoring the clustering storage optimization objective or relaxing the shelf stability constraint) to prioritize the outbound efficiency objective, and then restore the original strategy after the emergency task is completed.

[0142] The multi-objective optimization-based dynamic allocation method for spare parts storage locations of this invention can be widely applied to spare parts warehouses in various industrial enterprises, especially suitable for, but not limited to, the following industries: sponge titanium and non-ferrous metal smelting, petrochemical industry, power energy industry, heavy machinery manufacturing industry, iron and steel metallurgy industry, and automobile manufacturing industry, where spare parts are diverse in type, specifications, and usage frequency. By combining AGV automated guided vehicles, WMS warehouse management system, ERP enterprise resource planning system, and digital twin platform, fully automated intelligent warehouse management can be achieved, significantly reducing the degree of manual intervention and improving the intelligence level of warehouse operations.

[0143] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for dynamic allocation of spare parts storage locations based on multi-objective optimization, characterized in that, Includes the following steps: Establish a digital mapping model for the spare parts warehouse and keep it synchronized with the physical warehouse. Obtain multi-dimensional attribute information of spare parts to be put into the warehouse; Based on a multi-objective optimization strategy, the multi-dimensional attribute information is used as input to solve for the generation of storage locations; The multi-objective optimization strategy includes at least the objectives of inbound and outbound efficiency optimization, spare parts clustering and storage optimization, shelf stability optimization, and spare parts specification adaptation optimization, and uses the weighted combination of each optimization objective as the overall optimization function. During spare parts storage, the warehouse status is monitored. When the monitoring results meet the preset trigger conditions, the warehouse location is optimized and adjusted, and the category level of the spare parts and its corresponding storage area are updated according to the spare parts classification dynamic adjustment mechanism.

2. The method for dynamic allocation of spare parts storage locations based on multi-objective optimization according to claim 1, characterized in that, The digital mapping model is a digital twin model; the establishment steps specifically include: Collect physical structure data of the warehouse, geometric dimensions of storage locations, load-bearing parameters of shelves, and historical inbound and outbound operation data; Import 3D modeling software to build a digital twin model that is synchronized with the physical warehouse in real time.

3. The method for dynamic allocation of spare parts storage locations based on multi-objective optimization according to claim 1, characterized in that, The multi-dimensional attribute information includes one or more of the following: spare parts category, specifications and dimensions, weight, storage environment requirements, usage frequency level dynamically calculated based on historical outbound data, and correlation between spare parts.

4. The method for dynamic allocation of spare parts storage locations based on multi-objective optimization according to claim 3, characterized in that, The objective of optimizing inbound and outbound efficiency is to calculate the sum of the products of the usage frequency weight of each spare part and the weighted distance from its target storage location to the warehouse entrance and exit for all spare parts to be put into storage, and to minimize this sum. The usage frequency weight is set to different values ​​according to the spare parts category level, and the weighted distance takes into account the actual driving path of the AGV and the real-time path congestion coefficient.

5. The method for dynamic allocation of spare parts storage locations based on multi-objective optimization according to claim 1, characterized in that, The optimization objective for the clustering and storage of spare parts is to calculate the product of the correlation degree of each spare part pair and the distance between the corresponding storage locations for all spare part pairs and their corresponding storage location pairs, and then minimize the sum after summing the products. The correlation degree is determined based on the number of times two spare parts appear simultaneously in the same work task in historical data.

6. The method for dynamic allocation of spare parts storage locations based on multi-objective optimization according to claim 1, characterized in that, The optimization objectives for spare parts specification adaptation include: determining whether the spare parts packaging size exceeds the available capacity of the storage location, and constraining that the excess amount must be zero to ensure that allocation is feasible; When the size of a spare part is larger than a single storage location, it automatically searches for adjacent free storage locations and dynamically splices them together, logically merging multiple storage locations into a combined storage location. When the size of a spare part is smaller than that of a single storage location, the remaining space in the storage location already occupied by spare parts of the same category will be retrieved first and then stacked.

7. The method for dynamic allocation of spare parts storage locations based on multi-objective optimization according to claim 6, characterized in that, After the dynamic splicing is completed, the combined storage location is logically marked as a whole; when all spare parts in the combined storage location are taken out, the splicing is automatically de-spliced ​​and restored to an independent storage location.

8. The method for dynamic allocation of spare parts storage locations based on multi-objective optimization according to claim 1, characterized in that, The preset triggering conditions include at least one of the following: the idle storage space rate is lower than the first threshold, the increase in the average storage distance of high-frequency spare parts exceeds the second threshold, or an outbound task order marked as emergency or major repair is received. When any trigger condition is met, local or global storage location optimization adjustments will be automatically triggered.

9. The method for dynamic allocation of spare parts storage locations based on multi-objective optimization according to claim 1, characterized in that, The spare parts classification dynamic adjustment mechanism includes: periodically recalculating the category level of all spare parts based on the latest historical outbound data; Divide the physical area of ​​the warehouse into at least two heat zones; When the category level of spare parts changes, a relocation suggestion is automatically generated to move the spare parts from the original area to a suitable storage location in the corresponding area, and the relocation operation is performed during off-peak hours.

10. The method for dynamic allocation of spare parts storage locations based on multi-objective optimization according to claim 1, characterized in that, The specific steps involved in generating the storage location are as follows: Calculate the distance from all storage locations to entrances and exits in advance, and store a list of available storage locations by category and level. For each incoming spare part, candidate regions are determined based on its category level; Within the candidate area, feasible storage locations are selected based on specification compatibility and shelf stability constraints. The overall optimization function is used to evaluate and select the optimal storage location, generate an inbound instruction containing the coordinates of the target storage location, and send it to the execution terminal.

Images