A warehouse unit selection method, system, electronic device and storage medium

Abstract

This invention relates to the field of intelligent warehousing technology, and discloses a method, system, electronic device, and storage medium for selecting warehousing units. The method includes: acquiring product inventory data, basic demand data, and shipping standard data; selecting multiple warehousing unit combinations that meet the basic demand data from the product inventory data; extracting product quality index values ​​and calculating statistical values; generating adaptive weights by combining them with shipping standards; calculating the difference between the overall quality index of the warehousing unit combination and the shipping standards; constructing an objective function; applying a preset optimization algorithm to the objective function; selecting the warehousing unit combination that minimizes the objective function value; determining the final selection result from multiple warehousing unit combinations; and outputting the result. This invention can automate the entire process of warehousing unit selection, dynamically adapt to inventory and customer needs, ensure globally optimal selection results, improve decision-making efficiency and accuracy, reduce labor costs, and meet the needs of high-precision products such as semiconductors.

Description

Technical Field

[0001] This invention relates to the field of intelligent warehousing technology, specifically to a method, system, electronic device, and storage medium for selecting warehousing units. Background Technology

[0002] In the product warehousing and shipping process, selecting the optimal combination of storage units from massive inventory for specific customer orders is a critical decision-making process that directly impacts product quality evaluation, customer satisfaction, and enterprise operational efficiency. Currently, the industry mainly offers two solutions, but both suffer from significant technical bottlenecks: 1. Currently, many manufacturing companies still rely on manual decision-making in their warehouse management systems. This depends entirely on the experience and subjective judgment of warehouse staff, and the comparison of multi-dimensional pallet parameters is time-consuming, limiting the speed of delivery response. The selection results lack objectivity and consistency, and are prone to ignoring key factors such as sales rate and quality indicators due to local convenience orientation (such as selecting pallets nearby), making it difficult to implement refined business strategies.

[0003] 2. Some companies have tried to introduce rule-based automated pallet selection systems, but their technical implementation has obvious limitations: they operate based on a single or a few pre-set rules (such as first-in-first-out or proximity principle), and can only consider limited dimensions such as entry time and warehouse location distance; there is no intelligent balancing mechanism when rules conflict, and they only mechanically execute the preset logic, which cannot maximize the comprehensive benefits under multiple constraints. Summary of the Invention

[0004] To address the technical problems of low decision-making efficiency and poor overall benefits of traditional warehouse selection methods, this invention provides a warehouse unit selection method, system, electronic device, and storage medium that can automatically, quickly, and accurately select the optimal combination of shipping warehouse units from massive inventories, thereby significantly reducing costs, improving selection efficiency, and maximizing the overall benefits of warehouse selection.

[0005] In a first aspect, the present invention provides a method for selecting storage units, comprising: Obtain product inventory data, basic demand data, and shipping standard data; Select multiple combinations of warehousing units that meet the basic requirements data from the product inventory data; Obtain the quality index values ​​and statistical values ​​of various products for each of the aforementioned warehousing unit combinations, and calculate the adaptive weights of various product quality indicators based on the shipping standard data; Obtain the overall product quality index value for each of the aforementioned storage unit combinations, calculate the degree of difference between the overall product quality index value and the shipping standard data, and construct an objective function based on the adaptive weights and the degree of difference; A preset optimization algorithm is applied to the objective function, and the combination of storage units corresponding to the minimum objective function value is taken as the final selection target.

[0006] This invention integrates three core data categories—product inventory, demand, and delivery standards—through a fully automated process. It achieves closed-loop decision-making, from candidate combination selection and adaptive weight calculation to objective function construction and algorithm optimization. This overcomes the bottlenecks of experience dependence and rigid rules. Dynamic weights precisely adapt to the real-time gap between inventory and customer standards, enabling intelligent balancing of multi-dimensional quality indicators. The objective function quantifies comprehensive benefits, and optimization algorithms ensure the global optimality of the selection results. The selection time is reduced from hours to minutes, significantly lowering labor costs while improving decision consistency and reliability. This effectively avoids problems such as local optima and omissions of quality shortcomings, adapting to the complex selection needs of high-end products such as semiconductor silicon rods.

[0007] In one optional implementation, the step of obtaining the various product quality index values ​​and their statistical values ​​for each of the warehousing unit combinations, and calculating the adaptive weights of the various product quality index values ​​based on the shipping standard data, includes: Product quality index B of the first category of indicators was obtained respectively. i And the second category of product quality indicators B i ; Calculate B respectively i and the B j The statistical average value AVG(B) in the product inventory data i ) and AVG(B j ); The shipping standard data includes shipping standard indicator values. Based on the shipping standard indicator values ​​and the statistical average, the B is calculated respectively. i and the B j Adaptive weights for various product quality indicators.

[0008] The screening method provided by this invention covers all valid combinations, providing sufficient candidate samples for subsequent quality optimization and avoiding local optima caused by missing combinations; it strictly controls the extent of over-shipment to prevent inventory waste and increased transportation costs caused by over-shipment, while also avoiding the fulfillment risk of insufficient total quantity; at the same time, the screening rules provided are standardized and quantifiable, replacing subjective human judgment and ensuring consistent screening logic under different orders and different inventory scenarios, laying a high-quality data foundation for subsequent dynamic weight calculation and algorithm optimization, and improving the efficiency of the entire process of selecting suppliers.

[0009] In one optional implementation, the first type of indicator is an indicator where the product quality indicator value is greater than the delivery standard indicator value, which is considered excellent; the second type of indicator is an indicator where the delivery standard indicator value is greater than the product quality indicator value, which is considered excellent.

[0010] This invention accurately depicts the overall quality level of inventory by distinguishing indicator types and calculating statistical values, avoiding evaluation bias caused by single pallet data; the weight calculation is directly related to the delivery standard, so that the importance of the indicator is strongly bound to the inventory-standard gap, automatically guiding subsequent optimization to prioritize making up for quality shortcomings, and solving the problem that fixed weights cannot adapt to dynamic inventory.

[0011] In one optional implementation, the adaptive weight degree[i] of the first type of index is calculated using the following formula:

[0012] The formula for calculating the adaptive weight degree[j] of the second type of indicator is:

[0013] Among them, A i This represents the standard delivery indicator value corresponding to the first indicator, A. j This indicates the standard delivery indicator value corresponding to the second type of indicator.

[0014] This invention provides a quantitative adaptive weight calculation process. The formulaic design frees weight calculation from subjective experience dependence and realizes full-process data-driven operation. This not only ensures the consistency and accuracy of weight generation in different scenarios, but also reduces the difficulty of engineering implementation, enhances the robustness of the technical solution, and ensures that reliable adaptive weights can still be output stably in complex quality index scenarios.

[0015] In one optional implementation, obtaining the overall product quality index value corresponding to each of the warehousing unit combinations and calculating the difference between its overall product quality index value and the shipping standard data includes: Based on the product quality index values ​​of each product in the pallet of the storage unit combination, calculate the overall product quality index value of each storage unit combination. Based on the aforementioned shipping standard index values ​​and the aforementioned overall product quality index values, calculate the relative deviation rates of various indicators for each combination of warehousing units; wherein: The relative deviation rate Var[i] of the first type of index is (C i -A i ) / A i ; The relative deviation rate Var[j] of the second type of index is (A i -C j ) / A j ; Among them, C i This represents the overall product quality index value corresponding to the first type of index in each combination of storage units; Cj This represents the overall product quality index value corresponding to the second type of index in each combination of storage units.

[0016] This invention achieves precise quantification of candidate combination quality deviations through overall index aggregation and relative deviation rate calculation. It aggregates fragmented indicators of a single pallet into a combined overall value, reflecting the overall quality level of batch products and avoiding the evaluation blind spot of individual pallets being qualified but the combination being substandard. The relative deviation rate formula distinguishes index types, accurately depicts the degree of deviation between the combination quality and the standard, and eliminates the misleading nature of absolute differences, so that multi-dimensional deviations have a unified evaluation scale.

[0017] In one optional implementation, the objective function is a weighted sum of the adaptive weights and the relative deviation rate, and before calculating the function value, it is verified that the relative deviation rates of all product quality indicators are not less than 0; wherein: The objective function for the first type of index is expressed as: ; The objective function for the second type of index is expressed as: .

[0018] This invention, through a function that uses a weighted sum of relative change rates and dynamic weights, reflects both the degree of conformity between each product quality indicator and customer standards, and highlights the dynamic importance and priority of different indicators. This function can comprehensively consider multi-dimensional quality requirements and inventory adaptability. The lower the function value, the better the overall quality of the combination. Before calculation, the deviation rate is verified to be no less than 0, which can filter out unqualified warehousing unit combinations in advance, avoid invalid calculations, and ensure that the final selection result meets the basic constraints of customer standards.

[0019] In an optional embodiment, applying a preset optimization algorithm to the objective function, and selecting the combination of storage units corresponding to the minimum objective function value as the final selection target, includes: A preset optimization algorithm is run to search for combinations of storage units from all the storage units with the common optimization objective of minimizing the objective function value, thereby obtaining multiple candidate solutions and their corresponding objective function values. Compare the objective function values ​​corresponding to the multiple candidate solutions, and select the combination of storage units corresponding to the candidate solution with the smallest objective function value as the final selection result output.

[0020] The running preset optimization algorithm in this embodiment of the invention includes: running at least two different preset optimization algorithms in parallel; By running at least two optimization algorithms in parallel and optimizing the selection of storage unit combinations with minimizing the objective function value as the core objective, parallel search can cover a wider solution space, avoid the problem of a single algorithm getting stuck in a local optimum, significantly improve the global optimality of the selection results, and at the same time ensure the accuracy and reliability of the decision. By comparing the objective functions of different candidate solutions, the optimal storage unit combination is selected, achieving a balance of multiple business objectives, while ensuring the objectivity and consistency of the selection decision, and helping to implement refined business strategies.

[0021] In one optional implementation, the preset optimization algorithm includes: a genetic algorithm, a simulated annealing algorithm, and a greedy algorithm; wherein: The genetic algorithm uses binary encoding to represent the tray selection state, uses the objective function as the fitness function, and performs a global search through selection, crossover, and mutation operations. The simulated annealing algorithm uses the objective function as the energy function, and the energy minimization objective is iteratively searched by introducing a probabilistic jump mechanism to escape local optima. The greedy algorithm uses an iterative approach, selecting the pallet from the unselected pallets that causes the objective function value of the current partial combination to decrease the most in each step and adding it to the selected set.

[0022] This invention combines genetic algorithms, simulated annealing algorithms, and greedy algorithms to achieve complementary advantages of different optimization algorithms. It can adapt to scenarios with different inventory sizes, indicator complexity, and order urgency, ensuring that it can stably output high-quality and implementable selection solutions in various business scenarios.

[0023] In one alternative implementation, when multiple candidate solutions have completely identical objective function values, they are selected according to the following priority rules: First priority: Based on the algorithm's computational efficiency, prioritize the candidate solutions corresponding to the algorithms with the shortest computation time; Second priority: If multiple algorithms take the same amount of time to run, calculate the variance of the results of multiple runs of each algorithm as the index balance coefficient, and select the candidate solution with the smallest index balance coefficient. Third priority: If the balance coefficients of multiple algorithms are consistent, count the frequency of outputting this type of candidate solution in the last N iterations, and select the candidate solution corresponding to the algorithm with the highest frequency.

[0024] This invention employs a first priority based on computational time, prioritizing solutions from efficient algorithms to meet the rapid delivery needs of urgent orders. The second priority introduces a variance calculation index, the balance coefficient, where a smaller coefficient indicates stronger combination quality stability, ensuring the selection results balance overall benefits and index balance. The third priority relies on historical iteration frequency data to select solutions that better suit current inventory and customer needs, avoiding the blindness of random selection. This rule makes the selection decision in special scenarios more scientific and targeted, further enhancing the robustness and practicality of the technical solution.

[0025] In an optional implementation, the method further includes generating a warehouse unit selection result evaluation report.

[0026] This invention, through the generation of a standardized evaluation report on warehouse unit selection results, achieves transparency, traceability, and interpretability of the selection results, solving the problems of opaque decision-making processes and difficulty in verifying results in traditional warehouse unit selection processes. Simultaneously, the report data can support subsequent business optimization, such as analyzing inventory quality shortcomings through indicator comparison and optimizing algorithm parameters through changes in objective function values, helping enterprises continuously improve the accuracy of warehouse unit selection decisions and operational efficiency, and enhancing the practicality and implementation value of the technical solution.

[0027] In an optional implementation, the method further includes generating a logistics scheduling instruction based on the final selection result.

[0028] This invention establishes a closed-loop process from pallet selection to scheduling, skipping the manual secondary data entry step and enabling scheduling to start immediately after pallet selection, significantly shortening the order fulfillment cycle and solving the efficiency bottleneck of the disconnect between warehousing pallet selection and logistics in the traditional model.

[0029] Secondly, embodiments of the present invention provide a storage unit selection system, comprising: The data acquisition module is configured to acquire product inventory data, basic demand data, and shipping standard data. The warehouse unit combination filtering module is configured to filter out multiple warehouse unit combinations that meet the basic requirement data from the product inventory data; The adaptive weight calculation module is configured to obtain the various product quality index values ​​and their statistical values ​​for each of the warehousing unit combinations, and calculate the adaptive weights of various product quality indicators based on the shipping standard data. The objective function construction module is configured to obtain the overall product quality index value corresponding to each of the warehousing unit combinations, calculate the difference between its overall product quality index value and the shipping standard data, and construct an objective function based on the adaptive weights and the difference. The selection result generation module is configured to apply a preset optimization algorithm to the objective function, with the goal of selecting the combination of storage units that minimizes the objective function value, and to determine and output the final selection result from the multiple combinations of storage units.

[0030] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the storage unit selection method of the first aspect or any corresponding embodiment described above.

[0031] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions configured to cause a computer to perform the storage unit selection method of the first aspect or any corresponding embodiment thereof.

[0032] Fifthly, the present invention provides a computer program product including computer instructions configured to cause a computer to execute the storage unit selection method of the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0033] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0034] Figure 1 This is a flowchart illustrating the storage unit selection method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating another storage unit selection method according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating another storage unit selection method according to an embodiment of the present invention; Figure 4 This is a flowchart illustrating another storage unit selection method according to an embodiment of the present invention; Figure 5 This is a flowchart illustrating a method for selecting storage units using a greedy algorithm according to an embodiment of the present invention. Figure 6 This is a flowchart illustrating a method for selecting storage units using three parallel algorithms according to an embodiment of the present invention. Figure 7 This is a structural block diagram of a storage unit selection system according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0037] Based on this, this invention provides an embodiment of a warehousing unit selection method. Taking the warehousing and shipping of semiconductor material silicon rod products as an example, it selects the optimal combination of warehousing units from a massive inventory for a specific customer order. This abstracts a complex business problem into a computable mathematical model and constructs a complete, dynamic, multi-objective optimization decision-making method. It should be noted that the steps shown in the flowcharts in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions; and although a logical order is shown in the flowcharts, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0038] Figure 1 This is a flowchart of a storage unit selection method according to an embodiment of the present invention. The storage unit can be various types of cargo carriers such as pallets, boxes, bins, and racking units. This embodiment uses pallets as an example for illustration. Figure 1 As shown, the process includes the following steps: Step S101: Obtain product inventory data, basic demand data, and shipping standard data.

[0039] Specifically, the product inventory data, basic demand data, and shipping standard data in this embodiment can be retrieved from the warehouse management system (WMS). The product inventory data includes at least: pallet number, the number of each product within the pallet, the product quality index value of each product, real-time inventory quantity, inventory availability status, and real-time inventory data change information. Among them: The product quality index values ​​of each product in this embodiment of the invention are a set of independent indicators that characterize the product quality level from different core dimensions. Taking the selection of semiconductor material silicon rod storage units as an example, the multi-dimensional product quality indicators include: lifetime, resistivity, N-type oxygen content ratio, and N-type carbon content ratio, which correspond to different core evaluation dimensions of silicon rod quality. Lifetime reflects the durability and service life of the silicon rod and is a key indicator for measuring its service capability; resistivity determines the electrical performance of the silicon rod and directly affects the conductivity stability of subsequent semiconductor devices; the N-type oxygen content ratio and N-type carbon content ratio are related to the purity and crystal integrity of the silicon rod and are core parameters for determining whether the silicon rod meets the standards for high-end chip manufacturing. The multiple product quality indicators are not overlapping and are comprehensively covered from three independent dimensions of performance, lifetime, and purity to meet the evaluation needs of silicon rod quality.

[0040] Inventory availability status is configured to indicate whether a pallet has been locked by another order. Real-time inventory data changes are generated by continuous inbound and outbound operations, ensuring that subsequent pallet selection decisions are always based on the latest inventory snapshot, enabling application in a real, dynamic warehousing environment.

[0041] In practical applications, product inventory data also includes customer association information, which originates from historical shipping data. This information specifically covers core fields such as unique customer identifiers (e.g., customer code, company name), cooperation level, historical order preferences, customized quality thresholds, and past shipping feedback records. By mapping customer information with inventory data and shipping standard data, the system can quickly identify target customers based on their unique identifiers when receiving new orders. It can automatically retrieve their historical cooperation information, including quality requirement preferences (e.g., a customer's long-term preference for low-oxygen-content silicon rods), customized indicator thresholds (e.g., specific resistivity ranges), and service priority rules, eliminating the need for manual re-entry or filtering of customer standards.

[0042] The basic requirement data and delivery standard data in this embodiment of the invention originate from customer orders, contracts, or customer-sent requirement data, and are input or maintained by computer software system personnel or maintenance staff. The basic requirement data includes the quantity, weight, and tolerance range of the products required by the customer, with reasonable selections made based on the actual product characteristics (e.g., a customer needs 10 tons of silicon rods or 1000 standard parts). The delivery standard data includes multi-dimensional quantitative quality benchmarks set by the customer for the target product, as well as optional personalized additional requirements. It is the core basis for measuring whether the warehousing unit combination meets the customer's needs, covering the qualified thresholds or optimal ranges of key product performance parameters, such as the average lifespan, resistivity range, upper limit of N-type oxygen content, and upper limit of carbon content for silicon rods. In practical applications, this data can be sent by the client to the warehousing unit selection system through the order system, thereby automatically parsing structured data (such as tabular indicator standards) or extracting key requirements from unstructured text to form a standardized delivery standard dataset, providing a clear basis for subsequent quality indicator comparison, weight calculation, and warehousing unit combination optimization.

[0043] Step S102: Select multiple warehousing unit combinations that meet the basic requirements data from the product inventory data.

[0044] Specifically, this step uses methods such as permutation and combination to select warehouse unit combinations whose total product quantity meets the demand requirements. This avoids order non-fulfillment due to insufficient quantity or inventory waste due to over-shipment, ensuring a balance between supply and demand. Focusing on effective warehouse unit combinations that meet the quantity requirements, subsequent multi-dimensional quality optimization is carried out to avoid invalid combinations consuming computing resources. This solves the problem of low computational efficiency and difficulty in quickly finding the best solution when dealing with massive combinations in existing technologies.

[0045] Step S103: Obtain the quality index values ​​and statistical values ​​of various products for each of the warehousing unit combinations, and calculate the adaptive weights of various product quality indicators based on the shipping standard data.

[0046] Specifically, existing technologies generally use fixed weights or experience-based preset weights, which cannot adjust the optimization direction according to the dynamic gap between the actual inventory situation and customer standards. This results in a lack of adaptability in the selection of suppliers and makes it difficult to compensate for quality shortcomings. The embodiments of this invention calculate the inventory statistics of product quality indicators for each product, such as the average value (AVG), standard deviation (STD), range, and distribution (e.g., skewness, kurtosis), and then dynamically adjust the weights based on the real-time gap between the statistical values ​​and the shipping standard data. This automatically focuses on the product's quality shortcomings, solving the problem of the lack of flexibility of fixed weights and ensuring that the selection of suppliers aligns with actual business needs.

[0047] Step S104: Obtain the overall product quality index value corresponding to each of the warehousing unit combinations, calculate the difference between its overall product quality index value and the shipping standard data, and construct an objective function based on the adaptive weight and the difference.

[0048] Specifically, existing technologies cannot quantitatively integrate multi-dimensional product quality indicators, customer standards, and dynamic demands, resulting in a lack of unified evaluation criteria for multi-objective decision-making and difficulty in measuring the comprehensive adaptability of warehousing unit combinations. This embodiment solves the problem of the difficulty in uniformly evaluating multi-dimensional product quality indicators by calculating the difference between the overall product quality indicator value and the delivery standard, thus transforming the abstract concept of "whether it meets the requirements" into calculable quantitative data.

[0049] The objective function in this invention is a function for calculating comprehensive benefits, which is the weighted sum of the differences between the product quality index values ​​and delivery standards of each product and their corresponding adaptive weights. Comprehensive benefits are precisely measured by quantitatively evaluating the deviation of each product's quality index. Specifically, the degree of deviation is defined by the difference between the overall statistical value of each product's quality index and the delivery standard value. Then, a weighted average of the deviations is calculated using dynamic weights to construct the comprehensive benefit function. The magnitude of this function value directly reflects the overall quality level of the warehousing unit combination. The smaller the function value, the lower the degree of deviation in the multi-dimensional quality indicators of the warehousing unit combination, the closer the overall quality is to or even better than the customer-set delivery standards, and the better the comprehensive benefits.

[0050] Step S105: Apply a preset optimization algorithm to the objective function, and select the combination of storage units corresponding to the minimum objective function value as the final selection target.

[0051] This invention utilizes automated optimization algorithms to perform optimization calculations for massive combinations of storage units, replacing existing manual screening and single-rule matching. This significantly reduces the time required for selection decisions, adapts to the high-inventory, high-order, and high-efficiency delivery needs, and solves the problem of low efficiency in existing technologies. It ensures the objectivity and consistency of selection results and uses a quantified objective function value as the criterion for judging merit, eliminating decision-making fluctuations caused by differences in human experience. This helps enterprises to stably implement refined warehouse management strategies, improve customer satisfaction, and enhance overall operational efficiency.

[0052] The storage unit selection method provided in this invention achieves intelligent, efficient, and precise storage unit selection decisions through a full-process design encompassing multi-source data acquisition and processing, dynamic weight calculation, objective function construction, and algorithm optimization. It completely eliminates reliance on human experience, significantly reducing labor costs and improving decision-making speed and consistency through automated processes. By integrating multi-dimensional data on product quality, customer standards, and dynamic inventory, it addresses the limitations of traditional single-dimensional decision-making methods. Dynamic weights and objective functions enable intelligent multi-objective trade-offs, prioritizing quality shortcomings and maximizing the overall benefits of storage unit selection. Algorithm optimization ensures globally optimal results and robustness. Adaptable to complex warehousing scenarios, it seamlessly integrates with existing systems, achieving full-process automation from order to storage unit selection, effectively reducing return and exchange risks and operating costs, and significantly improving customer satisfaction and inventory management efficiency.

[0053] This invention also provides a method for selecting storage units, such as... Figure 2 As shown, it includes the following steps: Step S201: Obtain product inventory data, basic demand data, and shipping standard data. For details, please refer to step S101, which will not be repeated here.

[0054] Step S202 involves selecting multiple combinations of warehousing units from the product inventory data that meet the basic demand data. Specifically, this step includes: Step S2021: The basic demand data includes demand quantity index values. Based on the demand quantity index values, all selectable storage units in the product inventory data are combined in a permutation and combination manner.

[0055] Specifically, by traversing all permutations and combinations of storage units in the inventory, we ensure that no potential optimal combination that meets the demand requirements is overlooked, providing a comprehensive selection basis for subsequent multi-dimensional quality optimization and guaranteeing the global optimality of the selection results.

[0056] Step S2022: Select warehouse unit combinations whose total product quantity is not less than the demand index value and whose difference between the total product quantity and the demand index value is within a preset reasonable range.

[0057] In this embodiment of the invention, through dual screening of total quantity compliance and controllable difference, it ensures that the combination of warehousing units can meet the core needs of customers, avoiding order fulfillment failures due to insufficient quantity; at the same time, it strictly controls the extent of over-shipment to prevent inventory waste, increased transportation costs, and idle resources caused by over-shipment, thus achieving a precise balance between supply and demand. The preset reasonable difference range can be flexibly adjusted according to product characteristics, customer needs, and inventory status to adapt to the personalized needs of different business scenarios, enhancing the versatility and flexibility of the method. This ensures that all candidate combinations entering the subsequent optimization process meet the basic total quantity requirements, making subsequent multi-dimensional quality optimization practically meaningful and enhancing the practical value of the final selection result.

[0058] In one application scenario, a semiconductor silicon rod manufacturer receives a customer order for 1 ton, with a pre-defined reasonable tolerance range of ≤0.5 tons (i.e., total weight 10-10.5 tons). Since the loading capacity of a single pallet for silicon rods needs to be controlled at around 200 kg (standardized warehousing requirements), the theoretically required number of pallets is: 1000 kg ÷ 200 kg / pallet = 5 pallets. There are 8 available pallets in the product inventory (not locked by other orders), with the following loading weights: T1 (198 kg), T2 (203 kg), T3 (195 kg), T4 (201 kg), T5 (197 kg), T6 (202 kg), T7 (199 kg), and T8 (204 kg).

[0059] Based on the loading weight of 8 available pallets (195kg-204kg), the total weight of products on 5 pallets ranges from approximately 985kg-1020kg, and the total weight of products on 6 pallets ranges from approximately 1188kg-1224kg, both within a reasonable range. Therefore, the actual number of pallets required is 5-6. The following example illustrates this using 5 storage units: First, select 5 pallets from 8 available pallets and arrange them in a combination. This generates all possible combinations (such as [T1+T3+T4+T5+T7], [T2+T3+T6+T7+T8], etc.); further, it calculates and filters the total weight of each combination, for example: Combination [T3+T5+T7+T1+T4]: 195+197+199+198+201=1000kg (=1 ton, difference 0 tons, within range); Combination [T2+T4+T6+T7+T8]: 203+201+202+199+204=1009kg (≈1.009 tons, difference 0.009 tons, within the range). Combination [T3+T4+T5+T6+T8]: 195+201+197+202+204=999kg (<1 ton, eliminated).

[0060] Finally, combinations that meet the weight requirements, such as [T3+T5+T7+T1+T4] and [T2+T4+T6+T7+T8], are selected as storage unit combinations for subsequent processes. Further, by combining quality indicators such as silicon rod resistivity and oxygen content, dynamic weight calculation and multi-algorithm optimization are used to determine the optimal overall shipping combination.

[0061] Step S203 involves obtaining the quality index values ​​and statistical values ​​of various products for each of the warehousing unit combinations, and calculating the adaptive weights of various product quality indicators based on the shipping standard data. Specifically, this includes the following steps: Step S2031: Obtain the product quality index B of the first category of indicators. i And the second category of product quality indicators B j .

[0062] In practical applications, when determining whether delivery standards are met, product quality indicators are categorized into two types based on the product's characteristics: The first type is suitable for using the difference between the product quality indicator value and the delivery standard indicator value as the indicator of difference (e.g., silicon rod lifespan, purity, etc., where higher values ​​indicate better quality); the second type is suitable for using the difference between the delivery standard indicator value and the product quality indicator as the indicator of difference (e.g., silicon rod resistivity, whose value must be controlled within a standard threshold). This invention's embodiments design differentiated calculation logic for the characteristics of different quality indicators, avoiding misjudgments of different types of indicators by a single difference rule, making the difference assessment more closely aligned with the actual product quality requirements.

[0063] In this embodiment of the invention, the first type of indicator (type i) is an indicator where the product quality indicator value is greater than the delivery standard indicator value, which is considered excellent; the second type of indicator (type j) is an indicator where the delivery standard indicator value is greater than the product quality indicator value, which is also considered excellent. Here, i + j = n (where n is the total number of product quality indicators). By clearly distinguishing the attributes and difference calculation logic of the two types of indicators, the subsequent calculation of the statistical average value is made more consistent with the actual quality requirements of different indicators, avoiding evaluation bias caused by a single difference calculation logic, and ensuring the accuracy of quality assessment. In practical applications, the classification of type i and type j indicators can be flexibly adjusted according to the indicator characteristics of products in different industries (e.g., high-end wines are classified as type i, and cosmetics are classified as type j), adapting to the selection needs of multiple fields such as semiconductors, high-end manufacturing, and consumer goods.

[0064] Step S2032, calculate B respectively i and the B j The statistical average value AVG(B) in the product inventory data i ) and AVG(B j ).

[0065] Specifically, this step, by calculating the statistical averages of the two types of indicators separately, accurately depicts the overall level of the current inventory products across different attributes. This ensures that subsequent weight adjustments align with the actual quality status of the inventory, avoiding decision-making imbalances caused by biases in a single indicator. Taking silicon rod products as an example, based on their core quality indicator characteristics, indicators are divided into categories i and j, and their statistical averages are calculated, as follows: 1. Indicator Classification and Basic Data: Assuming there are 8 available pallets (T1-T8) in the product inventory, and based on the customer's shipping standards, define the indicator categories and quality indicator values ​​for each pallet: Category i indicators (higher values ​​are better, such as lifetime and minority carrier lifetime): Customer shipping standard A i =120, lifespan value B for each pallet i (Unit: μs) are: T1(118), T2(123), T3(119), T4(125), T5(121), T6(124), T7(120), T8(122); Class j indicators (values ​​must be lower than the standard, such as oxygen content): Customer delivery standard A j =1.0×10¹ 8 atoms / cm³, oxygen content value of each tray B j (Unit: ×10¹) 8 atoms / cm³) are: T1(0.92), T2(0.88), T3(0.95), T4(0.85), T5(0.90), T6(0.87), T7(0.93), T8(0.89).

[0066] 2. Calculation process of statistical average: Statistical average of i-type indicators AVG(B) i ), according to all available trays B i The values ​​are summed and then averaged. The formula is: AVG(B i ) = (118 + 123 + 119 + 125 + 121 + 124 + 120 + 122) ÷ 8 = 972 ÷ 8 = 121.5 This value reflects the overall level of the current inventory of silicon rods' lifespan. It will be compared with the customer's standard delivery value Ai, and the difference will be calculated to dynamically adjust the weights.

[0067] Statistical average of category j indicators AVG(B) j )B by all available trays j The values ​​are summed and then averaged. The formula is: AVG(B j = (0.92 + 0.88 + 0.95 + 0.85 + 0.90 + 0.87 + 0.93 + 0.89) ÷ 8 = 7.29 ÷ 8 = 0.91125 × 10¹8 This value reflects the overall oxygen content of the current inventory of silicon rods and will be configured to match the customer's standard shipping value A. j Compare and quantify the gap between inventory and standards.

[0068] Step S2033: The shipping standard data includes shipping standard indicator values. Based on the shipping standard indicator values ​​and the statistical average, the B is calculated respectively. i and the B j The quality of all kinds of products.

[0069] This invention calculates weights based on the difference between the standard shipping indicator value and the average inventory value; the larger the difference, the higher the weight. The automatic guidance algorithm prioritizes addressing quality weaknesses, solving the problem of fixed weights failing to adapt to dynamic inventory. Furthermore, it directly links the weights to customer standards, making multi-dimensional quality assessments more aligned with core customer needs, avoiding irrelevant indicators from interfering with decision-making, and ensuring that the selection results accurately match the shipping standards across key quality dimensions. Specifically: The formula for calculating the adaptive weight degree[i] of the first type of indicator is as follows:

[0070] The formula for calculating the adaptive weight degree[j] of the second type of indicator is:

[0071] Among them, A i A represents the standard shipping indicator value corresponding to the first type of indicator. j This refers to the standard delivery indicator value corresponding to the second type of indicator.

[0072] The numerator of the adaptive weighting formula for the two types of indicators mentioned above is the relative difference between a certain type of indicator (type i / type j), i.e., "the difference between the average inventory value and the customer's standard delivery value ÷ the standard delivery value," quantifying the degree of deviation between the current inventory and customer requirements. The denominator is the sum of the relative differences between all type i indicators and type j indicators, which is configured to be normalized to ensure that the sum of the weights of all indicators is 1, satisfying the mathematical logic of weight allocation. The weights of each quality indicator of the product are automatically generated from the difference between the actual inventory situation and the customer's standard through the above formula, replacing fixed weights or experience-based presets. This solves the problem that traditional methods cannot adapt to dynamic inventory, ensuring that the optimization direction always aligns with actual quality shortcomings.

[0073] Step S204: Obtain the overall product quality index value corresponding to each of the warehousing unit combinations, calculate the difference between its overall product quality index value and the shipping standard data, and construct an objective function based on the adaptive weights and the difference. Specifically, this includes the following steps: S2041, Calculate the overall product quality index value C of each storage unit combination based on the product quality index values ​​of each product in the pallet within the storage unit combination. i C j .

[0074] Specifically, this embodiment of the invention aggregates fragmented quality data of a single pallet into an overall index value for candidate combinations, solving the evaluation blind spot where a single pallet is qualified but the combination as a whole fails to meet the standard, and ensuring that pallet selection decisions are based on the true overall quality of batch products.

[0075] In one example, the overall product quality index value of each storage unit combination is calculated by averaging the product quality index values ​​of the products in each pallet, i.e.: C i =(B1+B2+B……+B i ) / i C j =(B1+B2+B……++B j ) / j The overall product quality index (C) of the storage unit combination is calculated using the average value. i / C j To better meet the actual needs of bulk shipments, shipments of high-precision products such as semiconductors require consideration of the overall quality of the entire assembly, rather than the performance of a single pallet. Average values ​​accurately reflect the overall quality level of the assembly, avoiding the problem of high-quality individual pallets but substandard overall quality. For example: Customers have specific shipment standards A for product lifespan. i =120μs, lifespan value B for each tray product i Given T1(118), T2(123), T3(119), T4(125), and T5(121), the total i-type index value C is... i =(118+123+119+125+121)÷5=606÷5=121.2, this value reflects the overall lifespan level of this pallet product portfolio, which is higher than customer standard A. i =120μs, which meets the requirement that the higher the value, the better.

[0076] It should be noted that the overall product quality index values ​​can be calculated using the above formula or retrieved from the warehouse management system (WMS).

[0077] S2042, based on the aforementioned shipping standard index value and the aforementioned overall product quality index value, calculate the relative deviation rate of various indicators for each combination of warehousing units; wherein: The relative deviation rate Var[i] of the first type of indicator = (C i -A i ) / A i ; The relative deviation rate Var[j] of the second type of index = (A i -C j ) / A j ; Among them, the corresponding standard value A for each type of indicator is the shipping benchmark value. i A j The calculation method is as follows: A i =(A1+A2+A……+Ai) / i A j =(A1+A2+A……+Aj) / j It should be noted that the standard value for shipment can be calculated using the formula above, or it can be retrieved from the warehouse management system (WMS) or the customer's latest requirement standards.

[0078] This invention, through differentiated formulas for categories i and j of indicators, transforms the difference between the overall product quality indicator value and the delivery standard indicator value into a calculable relative deviation rate, thus eliminating the differences in the dimensions of different indicators (such as lifespan μs, oxygen content × 10¹). 8 (atoms / cm³), allowing for a unified evaluation scale for multi-dimensional quality deviations.

[0079] In this embodiment of the invention, the objective function is the weighted sum of the adaptive weights and the relative deviation rate, and before calculating the function value, it is verified that the relative deviation rates of all product quality indicators are not less than 0. The objective function for the first type of indicator is expressed as: ; The objective function for the second type of indicator is expressed as: .

[0080] In this embodiment of the invention, the relative deviation rate of various product quality indicators is ≥0 before the function value is calculated and verified. This can filter out unqualified warehousing unit combinations in advance, avoiding invalid calculations. At the same time, it ensures that the final selection result meets the basic constraints of customer standards. This reflects the degree of compliance of the indicators and highlights the priority of the weakest indicators. The rule that the lower the function value corresponds to the better the overall quality makes the optimization target of warehousing unit selection clear and quantifiable, improving the accuracy and efficiency of the algorithm search.

[0081] To achieve a comprehensive quantitative assessment of the overall quality of the combined storage units, this invention integrates the objective function values ​​of two types of indicators to calculate a final objective function value. This final objective function value serves as the core criterion for determining the overall quality of the combined storage units. The formula for calculating the final objective function value based on the objective function values ​​of the two types of indicators is as follows:

[0082] The design of calculating the objective function by sub-indicators and then merging them into the final objective function not only reflects the degree of compliance of different types of indicators, but also highlights the priority of the quality weakness indicators through adaptive weights. The lower the value of the final objective function, the higher the degree of conformity between the warehousing unit combination and the shipping standards in terms of multi-dimensional quality indicators, the better the overall quality, and the more accurate and efficient the algorithm search is.

[0083] Step S205: Apply a preset optimization algorithm to the objective function, aiming to select the combination of storage units that minimizes the objective function value, and determine and output the final selection result from the multiple combinations of storage units. Specifically, step S205 includes the following steps: Step S2051: At least two different preset optimization algorithms are run in parallel to search for storage unit combinations from all the storage units with the common optimization objective of minimizing the objective function value, thereby obtaining multiple candidate solutions and their corresponding objective function values.

[0084] In practical applications, at least one of several pre-defined optimization algorithms, such as genetic algorithms, simulated annealing algorithms, and greedy algorithms, can be used. The advantages of greedy algorithms are fast computation and convergence, quickly finding local optima, making them suitable as a rapid initial screening tool; their disadvantage is that they are prone to getting trapped in local optima and cannot find global optimization. Genetic algorithms have strong global search capabilities, traversing multi-dimensional solution spaces, making them suitable for discovering global optima; their disadvantage is slower convergence and slightly higher computational cost. Simulated annealing algorithms have the advantage of escaping local optima, compensating for the shortcomings of greedy algorithms, and their computational complexity is between that of greedy and genetic algorithms. When configured for intelligent selection, the specific process is as follows: 1. The genetic algorithm uses binary encoding to represent the pallet selection state, and uses the objective function as the fitness function. It performs a global search through selection, crossover, and mutation operations. The parameters involved, such as population size, crossover probability, and mutation probability, are adapted to both the inventory size (total number of pallets) and the indicator dimension. Taking the assumption that an order needs to select 3 pallets from 10 candidate pallets (K=3) as an example, the steps include: A1 uses a 10-bit binary number to represent the tray selection status, where "1" represents selected and "0" represents unselected (e.g., "1001000010" means trays 1, 4, and 9 are selected). A2, Parameter settings: Inventory size 10 pallets, Population size 50, Crossover probability 0.8, Mutation probability 0.05; A3, Execution process: Initialize a population of 50 individuals with random binary codes, calculate the objective function value of each individual (combination of storage units) as fitness; select individuals with high fitness through roulette wheel selection, exchange some code segments according to the crossover probability (e.g., "10010|00010" crosses with "01100|11000" to get "1001011000"), and randomly flip individual code bits according to the mutation probability (e.g., "1001000010" mutates to "1001000110"); after 30 iterations, output the individual with the highest fitness (lowest objective function value) as a candidate solution.

[0085] A genetic algorithm is employed to adapt the selection scenario for storage unit combinations using binary encoding. The population size is dynamically adjusted according to the inventory size to ensure that a sufficient number of combination possibilities are covered from a massive number of candidate pallets. The collaborative operations of selection, crossover, and mutation can break through the constraints of local optima, making it particularly suitable for scenarios with large inventory sizes and complex indicator constraints, providing high-quality globally optimal candidate solutions for the final pallet selection.

[0086] 2. Simulated annealing uses the objective function as the energy function. The goal of minimizing energy is iteratively searched by introducing a probabilistic jump mechanism to escape local optima. This probabilistic jump mechanism is a core advantage, allowing for a certain degree of degradation in the early stages of iteration, preventing premature convergence to local optima. Even with poor initial combinations, the algorithm can gradually focus on the optimal region through temperature decay, complementing the genetic algorithm and effectively solving the problem of single algorithms easily getting trapped in local optima. This ensures stable output of high-quality solutions under different inventory distribution scenarios. In one embodiment, the following process is included: B1, Initial state: Randomly select trays 2, 5, and 7, and calculate the objective function value (energy) E0 = 0.35; B2, Iteration process: Initial temperature T0=1.0, cooling coefficient 0.95; randomly replace one tray (e.g., replace tray number 5 with tray number 8), generate a new combination, calculate the new energy E1=0.42 (ΔE=0.07>0); calculate the acceptance probability ≈0.93 according to the probability formula P=exp(-ΔE / T), generate a random number 0.78<0.93, accept the deteriorated state; continue cooling iteration, when T drops to 0.001, stop the search, output the combination with the lowest energy (E=0.28) during the iteration process.

[0087] 3. The greedy algorithm uses an iterative approach, selecting the pallet from the unselected pallets that maximizes the decrease in the objective function value of the current combination and adding it to the selected set at each step. The greedy algorithm selects the current optimal pallet at each step during iteration, eliminating the need to traverse all combinations. This results in significantly higher efficiency than the other two algorithms, providing an approximate optimal solution within milliseconds. For urgent orders and scenarios with frequent inventory turnover, it can quickly respond to pallet selection needs, avoiding the impact of excessive algorithm processing time on shipping efficiency and achieving a balance between efficiency and quality. An example includes the following process: C1, Initialization: The selected set is empty, and the remaining set consists of 10 candidate trays; C2, First iteration: Traverse all remaining trays, simulate adding each tray to the selected set, calculate the objective function value, and find that the function value is the lowest (0.12) when tray number 3 is added, so add it to the selected set; C3, Second iteration: Traverse the remaining 9 trays, simulate adding them and calculate the function value, and select the 6th tray that causes the current combined function value to decrease the most (the function value drops to 0.08). C4, Third iteration: Repeat the above operation, select tray number 10, the objective function value of the final combination (3, 6, 10) is 0.05, output the combination.

[0088] In practical applications, multiple combinations can be selected: Option 1: Choose one of the greedy algorithm, genetic algorithm, or simulated annealing algorithm for optimization. This can quickly provide high-quality initial candidate solutions, which can be directly used as the selection result for urgent orders.

[0089] Option 2: Greedy Algorithm + Genetic Algorithm (balancing efficiency and global optimum). The objective function values ​​of all candidate solutions output by the greedy algorithm and the genetic algorithm are compared, and the combination with the smallest value is selected as the final result. The greedy algorithm ensures fast response time, while the genetic algorithm avoids substandard quality due to local optima. The combination of the two satisfies both delivery timeliness and global optimum.

[0090] Option 3: Simulated annealing algorithm + genetic algorithm (high-precision global optimization). Compare the objective function values ​​of the candidate solutions of the two algorithms and select the combination corresponding to the minimum value. If the optimal values ​​of the two are close, the result of the simulated annealing algorithm is preferred (its characteristic of escaping local optima is more suitable for balancing multi-dimensional quality indicators). Suitable for customer orders with extremely high requirements for silicon rod quality, this option improves the selection accuracy by using a dual global optimization algorithm and avoids the blind spots of a single algorithm.

[0091] Option 4: Run the three algorithms mentioned above in parallel, and collect the three candidate solutions output by each algorithm, along with the objective function value as their score. Directly compare the scores of the three algorithms, and use an optimal solution selector to select the combination of storage units corresponding to the candidate solution with the smallest score as the final result. By running the three algorithms in parallel, and outputting candidate solutions from the three dimensions of global optimum, robustness, and efficiency, respectively, and then selecting the combination with the lowest objective function value, the selection result is ensured to have the characteristics of meeting quality standards, being feasible to execute, and being optimally efficient under different business scenarios.

[0092] Step S2052: Compare the objective function values ​​of each candidate solution, and select the combination of storage units corresponding to the candidate solution with the smallest objective function value as the final selection result output.

[0093] Specifically, embodiments of the present invention apply a preset optimization algorithm to minimize the objective function value (i.e. To achieve a unified goal, the system automatically weighs multiple quality indicators (such as silicon rod lifespan and oxygen content), customer standards, and inventory dynamics based on dynamic weights. This ensures that the decision-making logic prioritizes addressing quality shortcomings, avoids bias caused by subjective human error or single rules, and ensures that the selection results align with the core needs of customers.

[0094] In practical applications, when multiple candidate solutions have completely identical objective function values, the following priority rules should be followed for selection: First priority: Based on algorithmic efficiency, the candidate solution corresponding to the algorithm with the shortest computation time is selected first. Specifically, the core criterion is the "effective computation time" of the algorithm's output candidate solution, prioritizing the candidate solution corresponding to the algorithm with the shortest computation time. Effective computation time is defined as the actual computation time from when the algorithm receives input parameters (such as pallet basic data and constraints) until it outputs the candidate solution (excluding time consumed by hardware delays, system resource scheduling, and other factors not caused by the algorithm itself). This first priority focuses on computational efficiency to meet the rapid delivery needs of urgent orders.

[0095] Second priority: If multiple algorithms have the same computation time, calculate the variance of the results of multiple runs of each algorithm as the index balance coefficient, and select the candidate solution with the smallest coefficient. The second priority measures the index balance through variance to ensure the quality and stability of the selected combination (the smaller the variance, the smaller the index fluctuation).

[0096] Third priority: If the balance coefficients of multiple algorithm indicators are consistent, count the frequency of the candidate solutions of this type in the last N iterations (N can be configured according to actual needs, for example, N=5), and select the candidate solution corresponding to the algorithm with the highest frequency. The third priority relies on historical frequency data to select the algorithm solution that is more suitable for the current inventory and customer needs.

[0097] This invention solves the problem of choosing candidate solutions when the objective function values ​​are consistent by using a multi-priority screening rule based on computational efficiency, indicator balance, and output frequency. This significantly improves the completeness and adaptability of the selection decision. The rule takes into account both efficiency and quality, and enhances the reliability of the decision through historical data, ensuring that even in special scenarios where the objective function values ​​are consistent, the optimal selection result that best suits the business needs can still be output.

[0098] In one embodiment, such as Figure 3 As shown, the above-mentioned method for selecting storage units also includes: Step S206: Generate a warehouse unit selection result evaluation report.

[0099] In one example, the generated warehouse unit selection result evaluation report includes at least the final selected warehouse unit combination list, comparison data of the actual values ​​of each product quality indicator and the shipping standard values, and the final objective function value. For example, the warehouse unit combination list includes pallet number, product quality indicators of the products in the pallet, selection algorithm, etc. By statistically analyzing the selection frequency of different algorithms and the subsequent fulfillment feedback of the corresponding warehouse unit combinations (such as customer satisfaction and return rate), the performance of each algorithm under different inventory scenarios and customer needs can be analyzed, providing real business data support for algorithm parameter tuning (such as the population size of the genetic algorithm and the cooling coefficient of the simulated annealing algorithm).

[0100] Simultaneously, by combining the records of warehouse location distribution and quality status, inventory layout can be optimized (such as storing high-frequency selected high-quality pallets in a centralized manner), further improving overall operational efficiency; by comparing the actual values ​​of product quality indicators with the standard values ​​of delivery item by item, the compliance status of the combination in each dimension can be clearly displayed, ensuring that the delivery quality meets customer needs and reducing the risk of returns; the objective function value provides data basis for subsequent business optimization, for example, for indicators with relatively high deviation rates, the quality control in the production process can be optimized to improve the overall inventory's fit with customer standards.

[0101] The method provided in this invention can be directly configured as an advanced decision-making module in a WMS (Warehouse Management System) or ERP (Enterprise Resource Planning) system. For example, in the semiconductor silicon rod manufacturing industry, upon receiving a customer order, the system automatically invokes the storage unit selection method provided by this invention to select the optimal silicon rod storage unit combination from the inventory based on multiple indicators such as lifetime, resistivity, oxygen content, and carbon content. This combination is then directly configured to guide the shipping operation, ensuring the quality of finished products and optimizing the inventory structure. Besides being configured for the semiconductor silicon rod manufacturing industry, it can also be configured for other high-end material manufacturing industries (such as photovoltaic silicon wafers, electronic glass, and special ceramics), whose products also have multi-dimensional and stringent quality indicators. This invention can be adapted to modify the input quality indicators (such as conversion efficiency, transmittance, and intensity) and customer standards, and configured for pallet or box selection for finished product warehouses in these industries, achieving quality-oriented intelligent shipping.

[0102] In one embodiment, the above-described storage unit selection method further includes: Step S207: Generate logistics scheduling instructions based on the final selection result.

[0103] In one example, the generated logistics scheduling instructions include: vehicle model, scheduling time, transportation route, and target delivery location. These instructions are sent to the corresponding logistics scheduling system and configured to schedule the transportation vehicles to complete the transport of the warehouse unit combination. This embodiment of the invention directly converts the selection result into standardized logistics scheduling instructions, skipping the manual secondary input and processing stages. It enables scheduling to begin immediately upon completion of pallet selection, solving the efficiency bottleneck of the disconnect between pallet selection and logistics in the traditional model, and significantly shortening the order fulfillment cycle. In practice, scheduling instruction parameters can be flexibly adjusted according to order size (e.g., ton-level silicon rod orders) and customer requirements (e.g., expedited delivery, specified transportation methods). It also supports the generation of batch scheduling instructions for bulk orders, adapting to large-scale shipping scenarios while also meeting personalized logistics needs, enhancing the practicality of the solution.

[0104] It should be noted that, as Figure 4 The step S207 shown is executed after step S206. In other alternative embodiments, step S207 may be executed after step S205.

[0105] In one embodiment, such as Figure 5 As shown, the process of using a greedy algorithm to determine the final selection result from all candidate pallet combinations and generating a pallet combination selection result evaluation report and logistics scheduling instructions includes: 1. Input core data: The system synchronously obtains three types of key information: product inventory data (weight and quality indicators of each pallet), basic demand data (such as customer order tonnage and error range), and delivery standard data (customer's hard requirements for product quality).

[0106] 2. Screening candidate combinations: Based on basic demand data, filter out all storage unit combinations that meet the weight requirements from the inventory (i.e., the total weight is within a reasonable range), and remove invalid pallets with inconsistent weights to narrow down the calculation scope for subsequent optimization.

[0107] 3. Quality Indicator Preprocessing: Calculate the statistical values ​​(such as average values) of the quality indicators of inventory products, and generate adaptive weights by combining them with the delivery standards; at the same time, measure the difference between the overall quality of each candidate combination and the standard, which is configured as the objective function to provide a quantitative basis for subsequent scoring.

[0108] 4. Algorithm Optimization: Employing the optimal strategy at each step of a greedy algorithm, the final combination is constructed through iterative iteration. Initialization and Startup: Enter the "Greedy Algorithm Optimization" module and begin selecting tray icons step by step.

[0109] Iterative judgment (loop entry): Determine if K pallets have been selected (K is the theoretical number of pallets required to meet the order weight, such as 5 200kg pallets for a 1-ton order). If "Yes (Y)": The combination construction is complete, and the result is output directly. If "No (N)": Proceed to the subsequent filtering logic.

[0110] Traverse remaining pallets: Scan all inventory pallets that are not yet selected and locked.

[0111] Simulated addition: Each remaining tray is "virtually added" to the currently selected combination in turn to form a temporary combination.

[0112] Calculate the temporary score: Based on the previously generated adaptive weights and differences, calculate the weighted comprehensive score (objective function value) of the temporary combination. The lower the score, the smaller the quality deviation and the better the overall benefit.

[0113] Local optimal selection: From all temporary combinations, select the one with the best (lowest) score and add it to the final combination. Then the process jumps back to the iterative judgment step and repeats the above process until K trays are selected.

[0114] 5. After optimization, the process is extended to actual business implementation, achieving a seamless connection between "selection and scheduling": Multi-dimensional output results: The system not only outputs the optimal tray combination, but also includes an evaluation report on the tray combination selection results (including weight allocation, indicator deviation rate, algorithm optimization process, etc., solving the pain point of "black box decision-making"). Logistics instruction generation: Based on the final selection result, logistics dispatch instructions (including transport vehicle type, dispatch time, route, etc.) are automatically generated and directly connected to the logistics dispatch system. No manual secondary input is required, completing the entire closed loop from "cargo selection decision" to "logistics execution".

[0115] 6. Process completion: All selection and scheduling instructions have been generated.

[0116] In another embodiment, such as Figure 6 As shown, the process of using three optimization algorithms—genetic algorithm, simulated annealing algorithm, and greedy algorithm—in parallel execution to determine the final selection result from all candidate pallet combinations and generate a pallet combination selection result evaluation report and logistics scheduling instructions includes: 1. Input core data: The system synchronously obtains three types of key information: product inventory data (weight and quality indicators of each pallet), basic demand data (such as customer order tonnage and error range), and delivery standard data (customer's hard requirements for product quality).

[0117] 2. Screening candidate combinations: Based on basic demand data, filter out all pallet combinations that meet the weight requirements from the inventory (i.e., the total weight is within a reasonable range), and remove invalid pallets that do not meet the weight requirements to narrow down the calculation scope for subsequent optimization.

[0118] 3. Quality Indicator Preprocessing: Calculate the statistical values ​​(such as average values) of the quality indicators of inventory products, and generate adaptive weights by combining them with the delivery standards; at the same time, measure the difference between the overall quality of each candidate combination and the standard, which is configured as the objective function to provide a quantitative basis for subsequent scoring.

[0119] 4. Algorithm Optimization: Three optimization algorithms with different characteristics are invoked simultaneously to perform independent optimization on the same candidate combination pool without interference. Genetic Algorithm: It uses binary encoding to simulate the tray selection state. Through iterative operations of selection, crossover, and mutation, it performs a global breadth search of the combination space and outputs candidate solutions P1 and corresponding scores S1 (objective function values, the lower the better). Simulated annealing algorithm: using the objective function as the energy function, it escapes the local optimum trap through a probabilistic jump mechanism, performs fine mining of high-potential combination regions, and outputs candidate solutions P2 and scores S2; Greedy Algorithm: Iterates rapidly using a "stepwise optimization" strategy. At each step, select the pallet that minimizes the score of the current combination, quickly converge to generate a feasible solution, and output the candidate solution P3 and the score S3 (while taking into account the efficiency requirements of urgent orders).

[0120] Multiple-solution selection decision: The candidate solutions (P1, P2, P3) and their corresponding scores (S1, S2, S3) of the three algorithms are summarized. By "comparing the scores of all candidate solutions", the candidate solution with the lowest score is strictly selected as the final selection scheme to ensure the overall optimality of the result.

[0121] 5. After optimization, the process is extended to actual business implementation, achieving a seamless connection between "selection and scheduling": Multi-dimensional output results: The system not only outputs the optimal tray combination, but also includes an evaluation report on the tray combination selection results (including weight allocation, indicator deviation rate, algorithm optimization process, etc., solving the pain point of "black box decision-making"). Logistics instruction generation: Based on the final selection result, logistics dispatch instructions (including transport vehicle type, dispatch time, route, etc.) are automatically generated and directly connected to the logistics dispatch system. No manual secondary input is required, completing the entire closed loop from "cargo selection decision" to "logistics execution".

[0122] 6. Process completion: All selection and scheduling instructions have been generated.

[0123] This invention also provides a storage unit selection system, such as... Figure 7 As shown, it includes: The data acquisition module 71 is configured to acquire product inventory data, basic demand data, and shipping standard data. The storage unit combination filtering module 72 is configured to filter out multiple storage unit combinations that meet the basic requirement data from the product inventory data; The adaptive weight calculation module 73 is configured to obtain the values ​​of various product quality indicators and their statistical values ​​for each of the warehousing unit combinations, and to calculate the adaptive weights of various product quality indicators based on the shipping standard data. The objective function construction module 74 is configured to obtain the overall product quality index value of each of the warehousing unit combinations, calculate the difference between the overall product quality index value and the shipping standard data, and construct an objective function based on the adaptive weights and the difference. The selection result generation module 75 is configured to apply a preset optimization algorithm to the objective function and select the combination of storage units corresponding to the minimum objective function value as the final selection target.

[0124] In an optional embodiment, the storage unit combination screening module 72 includes: The storage unit combination unit is configured such that the basic demand data includes demand quantity index values, and the unit combines all selectable storage units in the product inventory data in a permutation and combination manner according to the demand quantity index values. The storage unit combination screening unit is configured to screen storage unit combinations in which the sum of the total number of products is not less than the demand index value, and the difference between the sum of the total number of products and the demand index value is within a preset reasonable range.

[0125] In an optional embodiment, the adaptive weight calculation module 73 includes: The product quality index value calculation unit is configured to obtain the product quality index value B of the first type of index. i And the product quality index value B of the second category of indicators.j ; The statistical average calculation unit is configured to calculate the B respectively. i and the B j The statistical average value AVG(B) in the product inventory data i ) and AVG(B j ); The adaptive weight calculation unit is configured to calculate B based on the shipping standard indicator value and the statistical average value, respectively. i and the B j Adaptive weights for various product quality indicators.

[0126] In one optional embodiment, the first type of indicator is an indicator where the product quality indicator value is greater than the delivery standard indicator value, which is considered excellent; the second type of indicator is an indicator where the delivery standard indicator value is greater than the product quality indicator value, which is considered excellent.

[0127] In an optional embodiment, the formula for calculating the adaptive weight degree[i] of the first type of index is: ; The formula for calculating the adaptive weight degree[j] of the second type of index is: ; Among them, A i This represents the standard shipping indicator value corresponding to the first type of indicator, A. j This indicates the standard delivery indicator value corresponding to the second type of indicator.

[0128] In an optional embodiment, the objective function construction module 74 includes: The overall product quality index value calculation unit is configured to calculate the overall product quality index value of each storage unit combination based on the product quality index values ​​of each product in the pallet in the storage unit combination. The relative deviation rate calculation unit is configured to calculate the relative deviation rates of various indicators for each combination of warehousing units based on the shipping standard indicator value and the overall product quality indicator value; wherein: The relative deviation rate Var[i] of the first type of indicator = (C i -A i ) / A i ; The relative deviation rate Var[j] of the second type of index = (A i -C j ) / A j ; Among them, C i This represents the overall product quality index value corresponding to category i in each combination of storage units; C jThis represents the overall product quality index value corresponding to category j in each combination of storage units.

[0129] In an optional embodiment, the formula for calculating the adaptive weight degree[i] of the first type of index is: ; The formula for calculating the adaptive weight degree[j] of the second type of index is: ; Among them, A i A represents the standard shipping indicator value corresponding to the first type of indicator. j This refers to the standard delivery indicator value corresponding to the second type of indicator.

[0130] In an optional embodiment, the objective function construction module 74 includes: The overall product quality index value calculation unit is configured to calculate the overall product quality index value of each storage unit combination based on the individual product quality index values ​​in the storage unit combination. The relative deviation rate calculation unit is configured to calculate the relative deviation rates of various indicators for each combination of warehousing units based on the shipping standard indicator value and the overall product quality indicator value; wherein: The relative deviation rate Var[i] of the first type of index is (C i -A i ) / A i ; The relative deviation rate Var[j] of the second type of index is (A i -C j ) / A j ; Among them, C i This represents the overall product quality index value corresponding to the first type of index in each combination of storage units; C j This represents the overall product quality index value corresponding to the second type of index in each combination of storage units.

[0131] In an optional embodiment, the selection result output module 75 includes: The multi-algorithm parallel optimization unit is configured to run a preset optimization algorithm to search for storage unit combinations from all the storage units with the common optimization objective of minimizing the objective function value, thereby obtaining multiple candidate solutions and their corresponding objective function values. The optimal solution selector unit is configured to compare the objective function values ​​of each candidate solution and select the combination of storage units corresponding to the candidate solution with the smallest objective function value as the final selection result output.

[0132] In an optional embodiment, running a preset optimization algorithm in the multi-algorithm parallel optimization unit includes: running at least two different preset optimization algorithms in parallel, the preset optimization algorithms including: genetic algorithm, simulated annealing algorithm, and greedy algorithm, wherein: The genetic algorithm uses binary encoding to represent the tray selection state, uses the objective function as the fitness function, and performs a global search through selection, crossover, and mutation operations. The simulated annealing algorithm uses the objective function as the energy function, and the energy minimization objective is iteratively searched by introducing a probabilistic jump mechanism to escape local optima. The greedy algorithm uses an iterative approach, selecting the pallet from the unselected pallets that causes the objective function value of the current partial combination to decrease the most in each step and adding it to the selected set.

[0133] In an optional embodiment, the optimal solution selector unit filters solutions according to the following priority rules when the objective function values ​​of multiple candidate solutions are completely identical: First priority: Based on the algorithm's computational efficiency, prioritize the candidate solutions corresponding to the algorithms with the shortest computation time; Second priority: If multiple algorithms take the same amount of time to run, calculate the variance of the results of multiple runs of each algorithm as the index balance coefficient, and select the candidate solution with the smallest coefficient. Third priority: If the balance coefficients of multiple algorithms are consistent, count the frequency of outputting this type of candidate solution in the last N iterations, and select the candidate solution corresponding to the algorithm with the highest frequency.

[0134] In an optional embodiment, an evaluation report output module is also included, configured to generate an evaluation report of the storage unit selection results.

[0135] In an optional embodiment, a scheduling instruction generation module is also included, configured to generate logistics scheduling instructions based on the final selection result.

[0136] The storage unit selection system provided in this embodiment of the invention can execute the storage unit selection method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method. Further functional descriptions of the various modules and units are the same as in the corresponding embodiments described above, and will not be repeated here.

[0137] The system provided in this invention embodiment can be encapsulated as an independent decision service (such as a microservice) and seamlessly integrated with existing WMS, ERP and other systems through API. As a value-added service or a standardized software product, it serves customers with complex product selection needs, realizing intelligent and lean management of the entire process from order placement to pallet recommendation.

[0138] Figure 8This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0139] The following is a detailed reference. Figure 8 This diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 801, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 802 or a program loaded from memory 808 into random access memory (RAM) 803. RAM 803 also stores various programs and data required for the operation of the electronic device. The processor 801, ROM 802, and RAM 803 are interconnected via bus 804. Input / output (I / O) interface 805 is also connected to bus 804.

[0140] Typically, the following devices can be connected to I / O interface 805: input devices 806 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 807 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 808 including, for example, magnetic tapes, hard disks, etc.; and communication devices 809. Communication device 809 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 8 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0141] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code configured to perform the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 809, or installed from a memory 808, or installed from a ROM 802. When the computer program is executed by the processor 801, it performs the functions defined in the storage unit selection method of the embodiments of the present invention.

[0142] Figure 8 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0143] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the storage unit selection method shown in the above embodiments is implemented.

[0144] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0145] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for selecting storage units, characterized in that, include: Obtain product inventory data, basic demand data, and shipping standard data; Select multiple combinations of warehousing units that meet the basic requirements data from the product inventory data; Obtain the quality index values ​​and statistical values ​​of various products for each of the aforementioned warehousing unit combinations, and calculate the adaptive weights of various product quality indicators based on the shipping standard data; Obtain the overall product quality index value for each of the aforementioned storage unit combinations, calculate the degree of difference between the overall product quality index value and the shipping standard data, and construct an objective function based on the adaptive weights and the degree of difference; A preset optimization algorithm is applied to the objective function, and the combination of storage units corresponding to the minimum objective function value is taken as the final selection target.

2. The storage unit selection method according to claim 1, characterized in that, Obtain the quality index values ​​and statistical values ​​of various products for each of the aforementioned warehousing unit combinations, and calculate the adaptive weights of various product quality indicators based on the aforementioned shipping standard data, including: Product quality index B of the first category of indicators was obtained respectively. i And the second category of product quality indicators B j ; Calculate B respectively i and the B j The statistical average value AVG(B) in the product inventory data i ) and AVG(B j ); The shipping standard data includes shipping standard indicator values. Based on the shipping standard indicator values ​​and the statistical average, the B is calculated respectively. i and the B j Adaptive weights for various product quality indicators.

3. The storage unit selection method according to claim 2, characterized in that, The first category of indicators refers to indicators where the product quality indicator value is greater than the delivery standard indicator value, indicating that the product is of good quality; the second category of indicators refers to indicators where the delivery standard indicator value is greater than the product quality indicator value, indicating that the product is of good quality.

4. The storage unit selection method according to claim 3, characterized in that, The formula for calculating the adaptive weight degree[i] of the first type of index is: ； The formula for calculating the adaptive weight degree[j] of the second type of indicator is: ； Among them, A i This represents the standard shipping indicator value corresponding to the first type of indicator, A. j This indicates the standard delivery indicator value corresponding to the second type of indicator.

5. The method for selecting storage units according to any one of claims 2-4, characterized in that, The step of obtaining the overall product quality index value for each of the aforementioned storage unit combinations and calculating the difference between the overall product quality index value and the shipping standard data includes: Calculate the overall product quality index value of each warehouse unit combination based on the quality index values ​​of each product in the warehouse unit combination; Based on the aforementioned shipping standard index values ​​and the aforementioned overall product quality index values, calculate the relative deviation rates of various indicators for each combination of warehousing units; wherein: The relative deviation rate Var[i] of the first type of index is (C i -A i ) / A i ; The relative deviation rate Var[j] of the second type of index is (A i -C j ) / A j ; Among them, C i This represents the overall product quality index value corresponding to the first type of index in each combination of storage units; C j This represents the overall product quality index value corresponding to the second type of index in each combination of storage units.

6. The method for selecting storage units according to claim 5, characterized in that, The objective function is a weighted sum of the adaptive weights and the relative deviation rate; where: The objective function for the first type of index is expressed as: ； The objective function for the second type of index is expressed as: 。 7. The method for selecting storage units according to any one of claims 1, 2, 3, 4, and 6, characterized in that, The step of applying a preset optimization algorithm to the objective function, with the combination of storage units corresponding to the minimum objective function value as the final selection target, includes: Run a preset optimization algorithm to search for combinations of storage units from all the storage units with the objective function value as the goal, and obtain multiple candidate solutions and their corresponding objective function values; Compare the objective function values ​​corresponding to the multiple candidate solutions, and select the combination of storage units corresponding to the candidate solution with the smallest objective function value as the final selection result output.

8. The method for selecting storage units according to claim 7, characterized in that, The operation of the preset optimization algorithm includes: running at least two different preset optimization algorithms in parallel; The preset optimization algorithms include: genetic algorithm, simulated annealing algorithm, and greedy algorithm; wherein: The genetic algorithm uses binary encoding to represent the tray selection state, uses the objective function as the fitness function, and performs a global search through selection, crossover, and mutation operations. The simulated annealing algorithm uses the objective function as the energy function, and the energy minimization objective is iteratively searched by introducing a probabilistic jump mechanism to escape local optima. The greedy algorithm uses an iterative approach, selecting the pallet from the unselected pallets that causes the objective function value of the current partial combination to decrease the most in each step and adding it to the selected set.

9. The method for selecting storage units according to any one of claims 7 or 8, characterized in that, When the objective function values ​​are completely identical, the following priority rules apply: First priority: Based on the algorithm's computational efficiency, prioritize the candidate solutions corresponding to the algorithms with the shortest computation time; Second priority: If multiple algorithms take the same amount of time to run, calculate the variance of the results of multiple runs of each algorithm as the index balance coefficient, and select the candidate solution with the smallest index balance coefficient. Third priority: If the balance coefficients of multiple algorithms are consistent, count the frequency of outputting this type of candidate solution in the last N iterations, and select the candidate solution corresponding to the algorithm with the highest frequency.

10. A storage unit selection system, characterized in that, include: The data acquisition module is configured to acquire product inventory data, basic demand data, and shipping standard data. The warehouse unit combination filtering module is configured to filter out multiple warehouse unit combinations that meet the basic requirement data from the product inventory data; The adaptive weight calculation module is configured to obtain the various product quality index values ​​and their statistical values ​​for each of the warehousing unit combinations, and calculate the adaptive weights of various product quality indicators based on the shipping standard data. The objective function construction module is configured to obtain the overall product quality index value corresponding to each of the warehousing unit combinations, calculate the difference between its overall product quality index value and the shipping standard data, and construct an objective function based on the adaptive weights and the difference. The selection result generation module is configured to apply a preset optimization algorithm to the objective function, with the goal of selecting the combination of storage units that minimizes the objective function value, and to determine and output the final selection result from the multiple combinations of storage units.

11. An electronic device, characterized in that, include: Memory and processor; The memory and the processor are communicatively connected. The memory stores computer instructions, and the processor executes the computer instructions to perform the storage unit selection method according to any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions configured to cause a computer to perform the storage unit selection method according to any one of claims 1 to 9.

13. A computer program product, characterized in that, Includes computer instructions configured to cause a computer to perform the storage unit selection method according to any one of claims 1 to 9.

Images