Warehouse site selection method

Abstract

The present invention discloses a warehouse site selection method, comprising: selecting several candidate regions; determining customer value index weights for each candidate region; constructing a multi-objective function, including: constructing a total cost function; constructing a customer satisfaction function based on the customer value index weights; constructing a warehouse quantity function; and solving the multi-objective function using a non-dominated sorting genetic algorithm to obtain optimal warehouse site selection information. The warehouse site selection method of the present invention weights customer value based on different regions, introduces regional customer value weights into the site selection model, and establishes a regional customer differentiation module in the system, making the final site selection results more scientific and realistic.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing and analysis, and in particular relates to a warehouse site selection method. Background Art

[0002] The recent rise of online marketplaces worldwide has enabled the entire online process of developing, marketing, selling, delivering, and servicing products and services for consumers. While online transactions offer convenience and efficiency for both consumers and retailers, online retailers must enhance their supply chain networks to provide better service. The goal of supply chain management is to deliver goods at the lowest possible cost and in the shortest possible time, while ensuring optimal service quality. This is fundamental to the success of online retail operations. Designing effective warehouse location strategies is crucial to developing a successful supply chain system.

[0003] Based on this, how to scientifically and rationally select a site to minimize total cost, minimize the number of warehouses and maximize customer satisfaction is the main technical problem solved by the present invention.

[0004] The above information disclosed in this background technology is only used to increase the understanding of the background technology of this application. Therefore, it may contain information that does not constitute the prior art known to ordinary technicians in this field. Summary of the Invention

[0005] The present invention proposes a warehouse location selection method to achieve the problem of simultaneously optimizing total cost, number of warehouses and customer satisfaction.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical solutions:

[0007] A warehouse location selection method is proposed, including:

[0008] Select several candidate areas;

[0009] Determine the weight of customer value indicators for each candidate region;

[0010] Construct a multi-objective function, including:

[0011] Construct a customer satisfaction function based on the weights of the customer value indicators:

[0012] maxF1=∑ i∈I ∑ j∈J m j S(d ij )x ij ;

[0013] in,

[0014]

[0015] m j is the value weight of the alternative region customer of customer point j, x ij is the cargo transportation volume from alternative warehouse i to demand point j, d ij is the distance from alternative warehouse i to demand point j, D0 is the maximum distance at which customer satisfaction is 1, and D1 is the minimum distance at which customer satisfaction is 0;

[0016] Construct warehouse quantity function:

[0017]

[0018] I is the set of city warehouses i, N is the capacity level set of candidate warehouses i, The value of is: 1 if a warehouse with a capacity of n is built at point i, otherwise it is 0;

[0019] Construct the total cost function:

[0020]

[0021] δ i is the unit area rent of alternative warehouse i, T i is the number of orders for alternative warehouse i, τ is the purchase cost coefficient, α i is the safety factor of the inventory level of alternative warehouse i, L i is the order lead time of alternative warehouse i, h 0i is the distance from the industrial park to the alternative warehouse i, σ j is the standard deviation of customer j’s demand, H i is the inventory holding cost of units of goods in alternative warehouse i, J is the set of demand points j, M ij The value of is: customer j is assigned to warehouse i is 1, otherwise it is 0, B j is the average demand of customer j, γ is the transportation cost coefficient, y 0i is the transportation volume from the industrial park to the alternative warehouse i;

[0022] A non-dominated sorting genetic algorithm is used to solve the multi-objective function and obtain the optimal warehouse location information.

[0023] Furthermore, the method for determining the weight of the customer value index of each candidate area is as follows:

[0024] Determine the evaluation indicators, establish a data series matrix, collect historical data of the indicators, and form an evaluation indicator matrix:

[0025]

[0026] Among them, q is the number of evaluation indicators; l is the number of unit time; X qrepresents the qth evaluation index data sequence; X q (l) represents the data of the qth evaluation index in the lth unit time;

[0027] Determine the reference data series:

[0028] X0=(X0(1),X0(2),...,X0(1))

[0029] Selecting the optimal value of each evaluation index data sequence to form the reference data sequence;

[0030] Non-dimensionalize the reference data sequence:

[0031] (X0(l),X1(l),X2(l),...,X α (l))

[0032] Calculate the difference between the reference data series and the data series of each evaluation index:

[0033]

[0034] Find out The extreme value in , let

[0035]

[0036] Calculate the correlation coefficient between each evaluation index:

[0037]

[0038]

[0039] Determine the weight of each indicator:

[0040]

[0041] Determine the customer value index weight m of the candidate area j :

[0042]

[0043] w ij is the proportion of alternative areas.

[0044] Furthermore, the constraints of the multi-objective function are:

[0045]

[0046]

[0047]

[0048]

[0049]

[0050]

[0051] in, M ij ={0,1},x ij ≥0,y 0i ≥0,v ij ≥0,T i ≥0, is the warehouse available area coefficient.

[0052] Furthermore, it also includes improving the non-dominated sorting genetic algorithm to solve the multi-objective function:

[0053] S1: Set parameters and generate the initial population P0;

[0054] S2: The non-dominated sorting algorithm stratifies the initial population P0, performs selection, crossover, and mutation operations on the initial population P0, and obtains the first-generation offspring population Q0;

[0055] S3: The parent population P t and the offspring population Q t Merge and obtain a new population R through the elite retention strategy t =P t ∪Q t , then R t Use the non-dominated sorting algorithm to stratify and calculate the crowding degree, select and sort all individuals, and select some individuals to obtain the new population P t+1 ;

[0056] S4: For the new population P t+1 Select individuals for crossover and mutation operations to obtain the offspring population Q t+1 ;

[0057] S5: Let i=i+1. If the maximum number of iterations is reached, end the process; otherwise, return to S3.

[0058] Furthermore, in step S2, the crossover operation is performed on the initial population P0, and the change of the crossover probability with generations is described as a linear decreasing model, setting is the range of the crossover operator, R c is the level value of the individual with higher non-dominated ranking level in the crossover operation, is the highest level value of the non-dominated sorting level in the population, is the lowest level value of the non-dominated sorting level in the population, Tmax is the total number of iterations, i is the current number of iterations, and the crossover rate is calculated as follows:

[0059]

[0060] Furthermore, in step S2, the variation of mutation rate with the number of iterations is described as a normal distribution model, setting is the variation range of the mutation operator, and the mean μ in the normal distribution is Variance σ 2 for R m It is the level value of the individual with higher non-dominated ranking level in the mutation operation. The mutation rate calculation formula is:

[0061]

[0062] Furthermore, in step S2, the method for stratifying the initial population P0 is:

[0063] The initial population of size n is stratified by a non-dominated sorting algorithm, including:

[0064] S11: let i=1;

[0065] S12: For all j=1,2.......n, and j≠i, compare individual x i and individual x j the relationship of domination and non-domination between them;

[0066] S13: If there is no individual x j Better than x i , then x i Labeled as nondominant individuals;

[0067] S14: Let i = i + 1, go to step S12, until all non-dominated individuals are found;

[0068] The set of non-dominated individuals obtained through the above steps is used as the first non-dominated layer of the initial population. Then, steps S11-S14 are performed again on the unlabeled individuals to obtain the second non-dominated layer, until all individuals are labeled and all strata of the initial population are obtained.

[0069] Furthermore, in step S3, R t The method for calculating congestion is:

[0070] S31: Initialize the distance between individuals in the same layer, let d i =0, i=1,2,3…n;

[0071] S32: Assign an infinite number to the two individuals at the edge, d1=d n =∞, to ensure that the next iteration can be entered in each loop;

[0072] S33: For individual i in the middle, Among them, d i represents the crowding degree of individual i, f j i+1 represents the mth objective function value of individual (i+1), f j i-1 Represents the mth objective function value of individual (i-1).

[0073] Furthermore, in step S3, in the selection sorting step, the selection operator selects individuals based on two indicators: non-dominated hierarchy and congestion. If x i The non-dominated level of is less than x j , or both have the same non-dominated level, x i The congestion degree is greater than x j The congestion degree of x i Better than x j , and sort all individuals.

[0074] Furthermore, in step S3, all individuals are sorted in descending order of performance, and the top N individuals are selected from the sorting to form a new population P. t+1 .

[0075] Compared with the prior art, the advantages and positive effects of the present invention are:

[0076] The warehouse site selection method of the present invention divides the weights according to the customer values ​​of different regions, introduces the regional customer value weights into the site selection model, and sets it as a regional customer differentiation module in the system, so that the final site selection result is more scientific and realistic.

[0077] Other features and advantages of the present invention will become more apparent after reading the detailed description of the present invention in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0079] Figure 1 It is a flow chart of an embodiment of the warehouse site selection method proposed by the present invention. DETAILED DESCRIPTION

[0080] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0081] It should be noted that, in the description of the present invention, the terms "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside" and the like indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is merely for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0082] In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "connect," "fixed," etc. should be understood broadly. For example, they may refer to fixed connection, detachable connection, or integration; mechanical connection or electrical connection; direct connection or indirect connection through an intermediate medium; internal communication between two components or interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0083] Example 1

[0084] This embodiment proposes a warehouse location selection method, such as Figure 1 As shown, including:

[0085] Select several candidate areas.

[0086] Determine the weight of customer value indicators for each candidate area.

[0087] Construct a multi-objective function, including:

[0088] Construct a customer satisfaction function based on the weights of customer value indicators:

[0089] maxF1=∑ i∈I ∑ j∈J m j S(d ij )x ij; This objective function takes into account the delivery time of orders and the customer value in different regions, making it more applicable to the actual situation of the enterprise and more realistic for the construction of the site selection model.

[0090] in,

[0091]

[0092] m j is the value weight of the alternative region customer of customer point j, x ij is the cargo transportation volume from alternative warehouse i to demand point j, d ij is the distance from alternative warehouse i to demand point j, D0 is the maximum distance at which customer satisfaction is 1, and D1 is the minimum distance at which customer satisfaction is 0.

[0093] In real life, the value that each customer brings to the enterprise is different, and the enterprise's resources cannot be consumed indefinitely in pursuit of absolute satisfaction of all customers. Therefore, in order to maximize the interests of the enterprise more efficiently, this method introduces regional customer weights to help the enterprise make more reasonable and reliable site selection decisions.

[0094] Excessive pursuit of customer satisfaction and the construction of too many city warehouses will bring greater difficulties to the company's subsequent management work. Therefore, the number of city warehouses built should be as small as possible. Construct a warehouse quantity function:

[0095]

[0096] I is the set of city warehouses i, N is the capacity level set of candidate warehouses i, The value of is: 1 if a warehouse with a capacity of n is built at point i, otherwise it is 0.

[0097] When selecting a warehouse location, a company shouldn't simply consider customer satisfaction and the number of warehouses. Controlling the total cost of adding new warehouses should also be a primary objective. Therefore, understanding the cost structure of warehouse location selection can help provide a more scientific theoretical basis for the final location decision. Based on the actual situation of Gooday Mart, the main cost components of warehouse location selection are: warehousing costs, ordering costs, inventory costs, and transportation costs. Therefore, a total cost function is constructed:

[0098]

[0099] δ i is the unit area rent of alternative warehouse i, T i is the number of orders for alternative warehouse i, τ is the purchase cost coefficient, α i is the safety factor of the inventory level of alternative warehouse i, L iis the order lead time of alternative warehouse i, h 0i is the distance from the industrial park to the alternative warehouse i, σ j is the standard deviation of customer j’s demand, H i is the inventory holding cost of units of goods in alternative warehouse i, J is the set of demand points j, M ij The value of is: customer j is assigned to warehouse i is 1, otherwise it is 0, B j is the average demand of customer j, γ is the transportation cost coefficient, y 0i is the transportation volume from the industrial park to the alternative warehouse i.

[0100] Since Gooday Logistics mainly chooses to rent warehouses, the warehousing costs calculated in this article mainly include warehouse rent, management fees, and other fixed costs related to warehouse operations, which are related to the warehouse area. The calculation formula is:

[0101]

[0102] (ii) Ordering cost: Ordering cost refers to the cost incurred from the time the order is placed to the time the inventory is received.

[0103] Such as order processing fees (including office expenses and clerical expenses), insurance premiums, and loading and unloading fees. The calculation formula is:

[0104] Inventory costs mainly include safety stock costs and average inventory costs.

[0105] During the planning period, the safety stock of city warehouse i is:

[0106] Therefore, it can be concluded that the safety inventory cost of all city warehouses during the planning period is:

[0107]

[0108] The expected delivery quantity of each order of city warehouse i during the entire planning period is: Q i =∑ j∈J M ij B j / T i .

[0109] Therefore, it can be concluded that the average inventory cost of all city warehouses during the planning period is:

[0110] ∑ i∈I ∑ j∈J M ij B j / 2T i .

[0111] Transportation costs: Transportation costs are a cost that companies must consider when selecting a location. They include both the transportation cost from the logistics park to the city warehouse and the transportation cost from the city warehouse to the demand point. Transportation costs are related to the transportation distance, unit transportation rate, and transportation volume. The calculation formula is:

[0112]

[0113] The non-dominated sorting genetic algorithm is used to solve the multi-objective function and obtain the optimal warehouse location information.

[0114] In this section, we will construct a location selection model based on two aspects: customer time satisfaction and regional customer value.

[0115] The site selection method of this plan can be described as: establishing a multi-objective model with minimization of total cost, minimization of the number of warehouses and maximization of customer satisfaction as the objective functions, solving the model through the NSGA-Ⅱ heuristic algorithm, and outputting a site selection plan.

[0116] Principles of warehouse site selection

[0117] (1) The principle of economy.

[0118] Costs associated with site selection primarily include construction and logistics costs. Economic efficiency is the primary principle in site selection, aiming to minimize the sum of operating and logistics costs. A well-chosen site can reduce costs and improve service levels. Therefore, this article develops a multi-objective model for warehouse site selection, using the objective functions of minimizing total costs, minimizing the number of warehouses, and maximizing customer satisfaction. Total costs include storage costs, ordering costs, inventory costs, and transportation costs.

[0119] (2) The principle of coordination.

[0120] From a holistic perspective, warehouse location selection is contextualized within the larger logistics network, ensuring that warehouse facilities and equipment are rationally distributed across regions. Logistics productivity and technical capabilities must be coordinated with the logistics network to promote the coordinated development of regional logistics systems. This article explores warehouse location selection by balancing workloads across regions through the addition of new warehouses, avoiding imbalances in workloads across selected warehouses, where some warehouses have minimal workloads, resulting in waste, while others have excessive workloads, leading to decreased efficiency.

[0121] (3) Strategic principles.

[0122] Strategic planning requires a long-term perspective when selecting a site, approaching the issue from a developmental perspective. This requires considering both the current situation and future development possibilities. We must consider the overall situation, prioritizing both immediate and long-term interests. Urban warehouses are built to meet future sales needs, and establishing a warehouse is a long-term decision. Therefore, the modeling solution in this article conducts warehouse site selection research based on future sales forecasts.

[0123] 2. Determine the alternative point

[0124] Warehouse site selection involves many factors that influence and restrict the location of the warehouse. Different types of sites have different focuses, so when selecting a site, various factors should be considered based on the characteristics of the company's warehouse. When a company builds a city warehouse, it must fully consider factors such as the customer's location and the surrounding environment to determine the candidate site. The specific factors are as follows:

[0125] (1) Traffic conditions

[0126] Convenient transportation is a condition that must be considered when selecting a warehouse site. It is best to locate the warehouse near a city transportation hub to ensure the smooth progress of distribution operations.

[0127] (2) Warehouse availability

[0128] Since the company plans to lease urban warehouses, the selection of alternative warehouse sites needs to give priority to warehouses that are already in operation locally and open up the warehouse leasing business to the outside world.

[0129] (3) Customer distribution and demand

[0130] The warehouse serves customers. When selecting a location for a warehouse, one should not only consider the distance to the customers, but also the concentration of customers. The density of customers is related to the density of the warehouse, and the service scope of the warehouse is determined based on distance. When selecting a location, one should also consider the demand for stores. The greater the demand, the greater the average cost of the warehouse that can be shared.

[0131] Based on the above three main factors, this plan should select several alternative warehouse sites as selection targets and enter them into the alternative warehouse database to provide a reference for future warehouse network planning and layout.

[0132] When selecting urban warehouse locations, logistics companies aim to improve the timeliness of goods delivery without increasing investment to meet all customer requirements. Customer satisfaction with timeliness varies. In real-world scenarios, the value of customers in different regions varies, and companies categorize customers and provide differentiated services for each type. Therefore, this module weights customer value based on region, introduces regional customer value weights into the location selection model, and establishes a regional customer differentiation module within the system, making the final location selection more scientific and realistic.

[0133] Regional customer differentiation indicators categorize customer value over time into two dimensions: current value and potential value. Regarding current value, the system's regional customer differentiation module uses shipping order volume as a measure of current customer value. It also develops additional indicators tailored to the company's specific characteristics. For example, since the company's primary business is delivering large household goods like home appliances, the average housing price in the region where its customers are located can, to a certain extent, reflect customer demand for appliance delivery quality. Therefore, potential customer consumption is represented by indicators such as the average unit price of housing in the region and population growth rate. Based on the conclusions of scholarly research on customer value, indicators such as the level of local economic development can be used to measure a region's current customer value. Regarding potential customer value, the module evaluates potential customer consumption and resident consumption level, based on the logistics target group and the living standards of local residents. Furthermore, this module utilizes a more detailed and quantifiable indicator hierarchy to measure regional customer value.

[0134] Customer current value

[0135] Current customer value refers to the value of a customer to a business based on current customer behavior and the local economic level. This article primarily measures this from two perspectives: order volume and the local economic level.

[0136] (1) Order volume

[0137] Warehouse goods are shipped according to orders. A region's order volume reflects the frequency of future business interactions between the warehouse and that region and is a key indicator of the region's customer value. This metric is composed of order quantity and order value: order volume, representing the total number of orders placed in the region over the past six months, directly reflects the relationship between warehouse workload and the region; order value, representing the cumulative total value of customer orders placed in the region over the past six months, directly reflects the relationship between the company's revenue and the region. These are the two indicators that best reflect a region's customer value from a company's business perspective.

[0138] (2) Local economic level

[0139] The local economic development level is an important indicator for measuring the local economic development status. From the perspectives of scale and speed, it can be divided into two aspects: GDP total and GDP growth rate. The GDP total can directly reflect the overall economic capacity of the local area and can indirectly understand the living standards of local residents. The GDP growth rate can reflect the trend of local economic development, whether the local economic development momentum is sufficient, whether local residents can create long-term value for the enterprise, whether they can form a long-term buying and selling relationship with the enterprise, and improve the retention rate of local residents in the enterprise.

[0140] 2. Potential customer value

[0141] Customer potential value helps companies decide whether to continue investing in a particular customer. It reflects whether users in that region will continue to create value and drive profit growth for the company over the long term, significantly impacting long-term strategic decision-making. Considering the potential value of a region's customers is crucial for warehouse location selection, a crucial decision that can impact future growth. This article analyzes this from the perspectives of current potential customer spending and household consumption levels.

[0142] (1) Potential customer consumption

[0143] Given the company's business characteristics, which primarily target consumers with housing needs, the average unit price of housing in a region can reflect the area's potential consumption level. Furthermore, the local population growth rate can also reflect the company's potential customer base. The more potential customers a region has, the more value it can create for the company in its future development, generating more orders and generating more value, thereby increasing the customer value of the region.

[0144] (2) Residents’ consumption level

[0145] The level of resident consumption reflects the extent to which local residents meet their daily needs for survival, development, and enjoyment, and is also a reflection of their spending power. This indicator is measured based on both population size and per capita disposable income. Population size reflects the total number of consumers in a region, while per capita disposable income is the primary driver of social consumption demand and a major factor driving socioeconomic growth. This indicator provides a valuable reference for companies' future business development.

[0146] Grey correlation analysis combines incomplete information with system behavior characteristics and other relevant data. It processes the factors to be analyzed and studied to a certain degree, finding the correlation of target elements in a random sequence of factors. It can also quantify the relationships between the various elements in the grey system. By grasping the main contradictions of things, it can find their main characteristics and main influencing factors. Then, it can rank the correlation of the various influencing factors of the grey system based on their size. This section will use the grey correlation analysis method to calculate the weights of the regional customer value indicators in the previous section.

[0147] Using the grey correlation analysis method, companies can analyze the customer characteristics, behaviors, and regional conditions of the region, find out the correlation between various indicators, and determine the weight of each indicator based on the degree of correlation between the indicators. In this way, the weight of customer value in each region can be determined based on the weight of each indicator, helping companies understand the areas where corporate services should be focused and making more scientific site selection decisions.

[0148] In this embodiment, the method for determining the weight of the customer value index of each candidate area is as follows:

[0149] Determine the evaluation indicators, establish a data series matrix, collect historical data of the indicators, and form an evaluation indicator matrix:

[0150]

[0151] Among them, q is the number of evaluation indicators; l is the number of unit time; X q represents the qth evaluation index data sequence; X q (l) represents the data of the qth evaluation index in the lth unit time;

[0152] Determine the reference data series:

[0153] X0=(X0(1),X0(2),...,X0(1))

[0154] Selecting the optimal value of each evaluation index data sequence to form the reference data sequence;

[0155] Non-dimensionalize the reference data sequence:

[0156] (X0(l),X1(l),X2(l),...,X α (l))

[0157] Calculate the difference between the reference data series and the data series of each evaluation index:

[0158]

[0159] Find out The extreme value in , let

[0160]

[0161] Calculate the correlation coefficient between each evaluation index:

[0162]

[0163]

[0164] Determine the weight of each indicator:

[0165]

[0166] Determine the customer value index weight m of the candidate area j :

[0167]

[0168] w ij is the proportion of alternative areas.

[0169] The constraints of the multi-objective function are:

[0170] Indicates that each selected warehouse can only have one capacity level.

[0171] This means that each customer point can only be served by one warehouse.

[0172] It means that the capacity of the warehouse must be greater than or equal to the capacity of the product demand of the customer points it serves.

[0173] Indicates the service level of the warehouse.

[0174] It means that the warehouse supply is greater than the customer demand.

[0175] The volume shipped in to each warehouse is greater than the volume shipped out.

[0176] in, M ij ={0,1},x ij ≥0,y 0i ≥0,v ij ≥0,T i ≥0, is the warehouse available area coefficient.

[0177] That is, under the above constraints, the total customer satisfaction is maximized, the total number of warehouses is minimized, and the total cost is minimized.

[0178] The NSGA2 algorithm (non-dominated sorting genetic algorithm) has relatively relaxed requirements on the mathematical properties of the problem being solved, making it suitable for complex multi-objective programming problems. However, as the problem size and the number of objective functions increase, the NSGA2 algorithm also exhibits some drawbacks, such as decreased convergence accuracy and increased convergence time. Therefore, this paper uses an improved NSGA2 algorithm based on the characteristics of the problem.

[0179] Therefore, this embodiment also includes improving the non-dominated sorting genetic algorithm to solve the multi-objective function:

[0180] S1: Set parameters and generate the initial population P0;

[0181] S2: The non-dominated sorting algorithm stratifies the initial population P0, performs selection, crossover, and mutation operations on the initial population P0, and obtains the first-generation offspring population Q0;

[0182] S3: The parent population P t and the offspring population Q t Merge and obtain a new population R through the elite retention strategy t =P t ∪Q t , then R t Use the non-dominated sorting algorithm to stratify and calculate the crowding degree, select and sort all individuals, and select some individuals to obtain the new population P t+1 ;

[0183] S4: For the new population P t+1 Select individuals for crossover and mutation operations to obtain the offspring population Q t+1 ;

[0184] S5: Let i=i+1. If the maximum number of iterations is reached, end the process; otherwise, return to S3.

[0185] The traditional NSGA2 algorithm, due to its fixed parameters, is sensitive to initial conditions, making it prone to premature population maturation and trapping in local optima. To improve algorithm efficiency, this paper proposes an adaptive genetic operator. If the mutation and crossover rates of the genetic operator are fixed, if they are too high, the rate of new individuals generated will increase, potentially damaging individuals with high fitness; if they are too low, the search efficiency will be too low. This paper adopts a method that automatically adjusts the crossover and mutation rates based on the Pareto level of excellence and the number of evolutionary generations. Specifically, the crossover and mutation rates decrease as fitness and the number of evolutionary generations increase.

[0186] In the early stages of an iteration, individuals are less similar. Increasing the crossover probability will accelerate the evolutionary process. At the end of the iteration, individuals are more similar. To prevent significant changes in the solution set, the crossover probability should be reduced, allowing the algorithm to gradually converge. Therefore, the change in crossover probability over generations can be described as a linearly decreasing model.

[0187] In step S2, the crossover operation is performed on the initial population P0, and the change of the crossover probability with generations is described as a linear decreasing model, setting is the range of the crossover operator, R c is the level value of the individual with higher non-dominated ranking level in the crossover operation, is the highest level value of the non-dominated sorting level in the population, is the lowest level value of the non-dominated sorting level in the population, T max is the total number of iterations, i is the current number of iterations, and the crossover rate is calculated as follows:

[0188]

[0189] In the early stages of iteration, when the mutation rate is small, the calculation process can be reduced and the population can maintain a rich solution pattern; when the intermediate algebraic algorithm tends to be "premature", increasing the mutation rate can effectively make the population jump out of the local optimum and facilitate the search for new individual patterns; in the later stages of iteration, in order to make the algorithm converge, the mutation operation should be reduced or even stopped. In step S2, the change of the mutation rate with the number of iterations is described as a normal distribution model, and the setting is the variation range of the mutation operator, and the mean μ in the normal distribution is Variance σ 2 for R m It is the level value of the individual with higher non-dominated ranking level in the mutation operation. The mutation rate calculation formula is:

[0190]

[0191] To solve multi-objective problems, conventional genetic algorithms are incapable of solving them. Therefore, this method utilizes a non-dominated sorting genetic algorithm (NSGA), distinguished by three key features: ① a fast non-dominated sorting method; ② a crowding comparison method; and ③ the introduction of an elitist strategy to expand the sampling space. This allows the parent and offspring populations to combine and compete to produce the next generation, facilitating the production of a superior next generation.

[0192] In step S2, the method for stratifying the initial population P0 is:

[0193] The initial population of size n is stratified by a non-dominated sorting algorithm, including:

[0194] S11: let i=1;

[0195] S12: For all j=1,2.......n, and j≠i, compare individual x i and individual x j the relationship of domination and non-domination between them;

[0196] S13: If there is no individual x j Better than x i , then x i Labeled as nondominant individuals;

[0197] S14: Let i = i + 1, go to step S12, until all non-dominated individuals are found;

[0198] The set of non-dominated individuals obtained through the above steps is used as the first non-dominated layer of the initial population. Then, steps S11-S14 are performed again on the unlabeled individuals to obtain the second non-dominated layer, until all individuals are labeled and all strata of the initial population are obtained.

[0199] In step S3, R t The method for calculating congestion is:

[0200] S31: Initialize the distance between individuals in the same layer, let d i =0, i=1,2,3…n;

[0201] S32: Assign an infinite number to the two individuals at the edge, d1=d n =∞, to ensure that the next iteration can be entered in each loop;

[0202] S33: For individual i in the middle, Among them, d i represents the crowding degree of individual i, f j i+1 represents the mth objective function value of individual (i+1), f j i-1 Represents the mth objective function value of individual (i-1).

[0203] In step S3, in the selection sorting step, the selection operator selects individuals based on the two indicators of non-dominated hierarchy and congestion. If x i The non-dominated level of is less than x j , or both have the same non-dominated level, x i The congestion degree is greater than x j The congestion degree of x i Better than x j , and sort all individuals.

[0204] In step S3, all individuals are sorted in descending order of performance, and the top N individuals are selected from the sorting to form a new population P. t+1 .

[0205] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for a person skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to replace some of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions claimed to be protected by the present invention.

Claims

1. A warehouse site selection method, characterized in that: include: Select several candidate areas; Determine the weight of customer value indicators for each candidate region; Construct a multi-objective function, including: Construct a customer satisfaction function based on the weights of the customer value indicators: maxF1=∑ i∈I ∑ j∈J m j S(d ij )x ij ; in, m j is the value weight of the alternative region customer of customer point j, x ij is the cargo transportation volume from alternative warehouse i to demand point j, d ij is the distance from alternative warehouse i to demand point j, D0 is the maximum distance at which customer satisfaction is 1, and D1 is the minimum distance at which customer satisfaction is 0; Construct warehouse quantity function: I is the set of city warehouses i, N is the capacity level set of candidate warehouses i, The value of is: 1 if a warehouse with a capacity of n is built at point i, otherwise it is 0; Construct the total cost function: δ i is the unit area rent of alternative warehouse i, T i is the number of orders for alternative warehouse i, τ is the purchase cost coefficient, α i is the safety factor of the inventory level of alternative warehouse i, L i is the order lead time of alternative warehouse i, h 0i is the distance from the industrial park to the alternative warehouse i, σ j is the standard deviation of customer j’s demand, H i is the inventory holding cost of units of goods in alternative warehouse i, J is the set of demand points j, M ij The value of is: customer j is assigned to warehouse i is 1, otherwise it is 0, B j is the average demand of customer j, γ is the transportation cost coefficient, y 0i is the transportation volume from the industrial park to the alternative warehouse i; A non-dominated sorting genetic algorithm is used to solve the multi-objective function to obtain optimal warehouse location information; The method for determining the weight of customer value indicators for each candidate region is as follows: Determine the evaluation indicators, establish a data series matrix, collect historical data of the indicators, and form an evaluation indicator matrix: Among them, q is the number of evaluation indicators; l is the number of unit time; X q represents the qth evaluation index data sequence; X q (l) represents the data of the qth evaluation index in the lth unit time; Determine the reference data series: X0=(X0(1),X0(2),...,X0(1)) Selecting the optimal value of each evaluation index data sequence to form the reference data sequence; Non-dimensionalize the reference data sequence: (X0(l),X1(l),X2(l),…,X α (l)) Calculate the difference between the reference data series and the data series of each evaluation index: Find out The extreme value in , let (k=1,2,...,l)(i=1,2,...,q) Calculate the correlation coefficient between each evaluation index: Determine the weight of each indicator: Determine the customer value index weight m of the candidate area j : W ij is the proportion of alternative areas.

2. The warehouse site selection method according to claim 1, characterized in that: The constraints of the multi-objective function are: in, M ij ={0,1},x ij ≥0,y 0i ≥0,v ij ≥0,T i ≥0, θ is the warehouse available area coefficient.

3. The warehouse site selection method according to claim 1 or 2, characterized in that: It also includes improving the non-dominated sorting genetic algorithm to solve the multi-objective function: S1: Set parameters and generate the initial population P0; S2: The non-dominated sorting algorithm stratifies the initial population P0, performs selection, crossover, and mutation operations on the initial population P0, and obtains the first-generation offspring population Q0; S3: The parent population P t and the offspring population Q t Merge and obtain a new population R through the elite retention strategy t =P t ∪Q t , then R t Use the non-dominated sorting algorithm to stratify and calculate the crowding degree, select and sort all individuals, and select some individuals to obtain the new population P t+1 ; S4: For the new population P t+1 Select individuals for crossover and mutation operations to obtain the offspring population Q t+1 ; S5: Let i=i+1. If the maximum number of iterations is reached, end the process; otherwise, return to S3.

4. The warehouse site selection method according to claim 3, characterized in that: In step S2, the crossover operation is performed on the initial population P0, and the change of the crossover probability with generations is described as a linear decreasing model, setting is the range of the crossover operator, R c is the level value of the individual with higher non-dominated ranking level in the crossover operation, is the highest level value of the non-dominated sorting level in the population, is the lowest level value of the non-dominated sorting level in the population, T max is the total number of iterations, i is the current number of iterations, and the crossover rate is calculated as follows:

5. The warehouse site selection method according to claim 3, characterized in that: In step S2, the variation of mutation rate with the number of iterations is described as a normal distribution model, setting is the variation range of the mutation operator, and the mean μ in the normal distribution is Variance σ 2 for R m It is the level value of the individual with higher non-dominated ranking level in the mutation operation. The mutation rate calculation formula is:

6. The warehouse site selection method according to claim 3, characterized in that: In step S2, the method for stratifying the initial population P0 is: The initial population of size n is stratified by a non-dominated sorting algorithm, including: S11: let i=1; S12: For all j=1,2.......n, and j≠i, compare individual x i and individual x j the relationship of domination and non-domination between them; S13: If there is no individual x j Better than x i , then x i Labeled as nondominant individuals; S14: Let i = i + 1, go to step S12, until all non-dominated individuals are found; The set of non-dominated individuals obtained through the above steps is used as the first non-dominated layer of the initial population. Then, steps S11-S14 are performed again on the unlabeled individuals to obtain the second non-dominated layer, until all individuals are labeled and all strata of the initial population are obtained.

7. The warehouse site selection method according to claim 3, characterized in that: In step S3, R t The method for calculating congestion is: S31: Initialize the distance between individuals in the same layer, let d i =0, i=1,2,3…n; S32: Assign an infinite number to the two individuals at the edge, d1=d n =∞, to ensure that the next iteration can be entered in each loop; S33: For individual i in the middle, Among them, d i represents the crowding degree of individual i, represents the mth objective function value of individual (i+1), Represents the mth objective function value of individual (i-1).

8. The warehouse site selection method according to claim 7, characterized in that: In step S3, in the selection sorting step, the selection operator selects individuals based on the two indicators of non-dominated hierarchy and congestion. x j ∈x, if x i The non-dominated level of is less than x j , or both have the same non-dominated level, x i The congestion degree is greater than x j The congestion degree of x i Better than x j , and sort all individuals.

9. The warehouse site selection method according to claim 8, characterized in that: In step S3, all individuals are sorted in descending order of performance, and the top N individuals are selected from the sorting to form a new population P. t+1 .

Images