A logistics network planning method and device

Abstract

A logistics network planning method and device, relating to the field of warehousing logistics technology. The method comprises: obtaining logistics node configuration information and neighborhood configuration information; establishing a target model according to the logistics node configuration information and setting constraints; the target model is used to characterize the correspondence between planning schemes and planning features, wherein any planning scheme includes part or all of a plurality of candidate logistics nodes, and the planning features include: investment cost and / or delivery delay; according to the domain configuration information, a variable neighborhood search algorithm is used to perform at least one local search calculation on the target model to obtain a target planning scheme; wherein the target planning scheme includes a target logistics node among a plurality of candidate logistics nodes, and the target investment cost in the target planning feature corresponding to the target planning scheme is lower than the cost threshold, and / or the target delivery delay is lower than the delay threshold. The method is helpful to reasonably plan the logistics network.

Description

Technical Field

[0001] The present application relates to the technical field of warehousing logistics, and in particular to a logistics network planning method and device. Background Art

[0002] With the rapid development of e-commerce, express delivery and other industries, the logistics industry has become a new industry with strong development momentum.

[0003] At present, the main competition in the logistics industry is reflected in the advantages and disadvantages of logistics network planning. The main factors of logistics network planning include the site selection of service target facilities, route arrangement, etc. Under the emphasis of the concept of "logistics integration", the logistics network system needs to be integrated and optimized, that is, the location and routing problem (LRP). In logistics network planning, decision makers are primarily concerned about the cost of the entire network system. For customers, what they care about is the response time of the entire system, that is, the timeliness issue. Therefore, it is necessary to comprehensively consider factors such as cost and timeliness, and make scientific plans for the site selection of facilities and route arrangement in the logistics network.

[0004] The current algorithms for solving LRP, including collaborative multi-objective algorithms, genetic algorithms, and exact solution algorithms, will encounter the problem of falling into local optimality too early in the problem-solving process, and it is difficult to solve the problem in a short time. Since the logistics network planning problem itself is a non-deterministic polynomial hard (NP-hard) problem, in a multi-level logistics network, the number of customer points is even larger and the network scale is also larger. When the above algorithms are actually applied to the planning of multi-level logistics networks, it is impossible to avoid the problem of the algorithm falling into local optimality, and it is impossible to achieve the purpose of solving complex multi-objective optimization problems.

[0005] Therefore, how to reasonably plan the logistics network remains an important issue that needs to be solved urgently. Summary of the invention

[0006] The embodiments of the present application provide a logistics network planning method and device, which are helpful for rationally planning a logistics network.

[0007] In a first aspect, an embodiment of the present application provides a logistics network planning method, which may include the following steps:

[0008] Obtaining logistics node configuration information and neighborhood configuration information; wherein the logistics node configuration information includes configuration data of multiple candidate logistics nodes of the logistics network to be planned, and the domain configuration information includes configuration data of multiple neighborhoods;

[0009] Establishing a target model according to the logistics node configuration information and set constraints; wherein the target model is used to characterize the correspondence between planning schemes and planning features, wherein any planning scheme includes some or all of the multiple candidate logistics nodes, and the planning features include: investment cost and / or delivery delay;

[0010] According to the domain configuration information, a variable neighborhood search algorithm is used to perform at least one local search calculation on the target model to obtain a target planning scheme; wherein the target planning scheme includes a target logistics node among the multiple alternative logistics nodes, and the target investment cost in the target planning characteristics corresponding to the target planning scheme is lower than a cost threshold, and / or the target delivery delay is lower than a delay threshold.

[0011] Through this solution, the investment cost and / or delivery delay of the logistics network can be used as optimization conditions, and the variable neighborhood search algorithm can be used to design the target model, so as to obtain the target planning solution according to the logistics node configuration information and neighborhood configuration information of the logistics network to be planned. Among them, the variable neighborhood search algorithm can adopt an improved VNS framework, add at least one local search calculation process, and avoid the algorithm from falling into the local optimum.

[0012] In one possible design, the logistics node configuration information and the neighborhood configuration information are obtained, including: obtaining alternative information of the target area; clustering the alternative information of the target area using a set clustering algorithm to obtain at least one clustering cluster, and any clustering cluster corresponds to an alternative logistics node; based on the at least one clustering cluster, the logistics node configuration information and the neighborhood configuration information are obtained.

[0013] Through this solution, a clustering algorithm can be used in advance to obtain configuration data of multiple candidate logistics nodes, so as to determine some or all of the candidate logistics nodes from the multiple candidate logistics nodes to establish the required logistics network. For example, the candidate information can be site information of the target area.

[0014] In a possible design, according to the domain configuration information, a variable neighborhood search algorithm is used to perform at least one variable neighborhood local search calculation on the target model to obtain a target planning scheme, including: obtaining a specified planning scheme; according to the domain configuration information, a variable neighborhood search algorithm is used to perform the following steps for a target neighborhood in a plurality of neighborhoods, wherein the target neighborhood traverses each neighborhood in the plurality of neighborhoods: according to the specified planning scheme and the configuration data of the target neighborhood, at least one perturbation operator is used to perform perturbation processing on the target model to obtain a first alternative planning scheme corresponding to the target neighborhood; based on the first An alternative planning scheme is to iterate the target model by performing at least one local search calculation to obtain a second alternative planning scheme corresponding to the target neighborhood; wherein, when the current planning scheme obtained by any local search calculation satisfies the set acceptance conditions, the current planning scheme obtained by this local search calculation is used as the first alternative planning scheme to iterate the target neighborhood for the next local search calculation until the second alternative planning scheme is obtained; when the set calculation termination conditions are met, the target planning scheme is output, wherein the target planning scheme is the alternative planning scheme with the smallest eigenvalue among the second alternative planning schemes of multiple neighborhoods.

[0015] Through this solution, the variable neighborhood search algorithm can adopt an improved VNS framework, add at least one local search calculation process, and avoid the algorithm falling into a local optimum.

[0016] In a possible design, the configuration data of the target neighborhood includes a neighborhood length; when calculating the target model, any candidate logistics node has an attribute value, and the attribute value is used to characterize the node type of the candidate logistics node in the corresponding planning scheme; according to the specified planning scheme and the configuration data of the target neighborhood, the target model is perturbed using at least one perturbation operator to obtain a first alternative planning scheme corresponding to the target neighborhood, including: according to the neighborhood length Km of the target neighborhood, randomly selecting Km first alternative logistics nodes from the multiple alternative logistics nodes, Among them, m represents the number of neighborhoods, and m and Km are integers greater than or equal to 1; according to the specified planning scheme, the attribute value of the corresponding first alternative logistics node is perturbed by using the disturbance operator of each of the Km first alternative logistics nodes, and then the target model is calculated to obtain the first alternative planning scheme corresponding to the target neighborhood; wherein, the disturbance operator of any first alternative logistics node is randomly selected from the first operator or the second operator, the first operator is used to reduce the attribute value of the alternative logistics node, and the second operator is used to increase the attribute value of the alternative logistics node.

[0017] In a possible design, when calculating the target model, any alternative logistics node has an attribute value, and the attribute value is used to characterize the node type of the alternative logistics node in the corresponding planning scheme; based on the first alternative planning scheme, the target model is iterated at least once for local search calculation to obtain a second alternative planning scheme corresponding to the target neighborhood, including: performing the following steps for the target local search calculation in the at least one local search calculation, wherein the target local search calculation traverses any one of the at least one local search calculation: determining multiple second alternative logistics nodes from the multiple alternative logistics nodes; according to the first alternative planning scheme, using the perturbation operator of each second alternative logistics node in the multiple second alternative logistics nodes to perturb the attribute value of the corresponding second alternative logistics node, and then calculating the target model to obtain to the current planning scheme calculated by the local search of the target; wherein, the disturbance operator of each second alternative logistics node is at least one of the first operator, the second operator, and the third operator, the first operator is used to reduce the attribute value of the alternative logistics node, the second operator is used to increase the attribute value of the alternative logistics node, and the third operator is used to exchange the attribute values ​​of two alternative logistics nodes; wherein, if the planning feature value corresponding to the current planning scheme calculated by the local search of the target is less than the planning feature value corresponding to the first alternative planning scheme, the current planning scheme calculated by the local search of the target is used as the first alternative planning scheme until the local search calculation process of the target is terminated; when the first alternative planning scheme meets the set acceptance conditions, the current planning scheme obtained by the last local search calculation in the at least one local search calculation is used as the second alternative planning scheme corresponding to the target neighborhood.

[0018] In a second aspect, an embodiment of the present application provides a logistics network planning device, comprising: an information acquisition unit, used to acquire logistics node configuration information and neighborhood configuration information; wherein the logistics node configuration information includes configuration data of multiple alternative logistics nodes of the logistics network to be planned, and the domain configuration information includes configuration data of multiple neighborhoods; a modeling unit, used to establish a target model based on the logistics node configuration information and set constraints; wherein the target model is used to characterize the correspondence between planning schemes and planning characteristics, wherein any planning scheme includes some or all of the multiple alternative logistics nodes, and the planning characteristics include: investment cost and / or delivery delay; a calculation unit, used to perform at least one variable neighborhood local search calculation on the target model according to the domain configuration information using a variable neighborhood search algorithm to obtain a target planning scheme; wherein the target planning scheme includes a target logistics node among the multiple alternative logistics nodes, and the target investment cost in the target planning characteristics corresponding to the target planning scheme is lower than a cost threshold, and / or the target delivery delay is lower than a delay threshold.

[0019] In one possible design, the information acquisition unit is used to: obtain the site selection information of the target area; cluster the site selection information of the target area using a set clustering algorithm to obtain at least one clustering cluster, and any clustering cluster corresponds to an alternative logistics node; based on the at least one clustering cluster, obtain the logistics node configuration information and the neighborhood configuration information.

[0020] In a possible design, the computing unit is used to: obtain a designated planning scheme; according to the domain configuration information, use a variable neighborhood search algorithm to perform the following steps for a target neighborhood among multiple neighborhoods, wherein the target neighborhood traverses each of the multiple neighborhoods: according to the designated planning scheme and the configuration data of the target neighborhood, use at least one perturbation operator to perturb the target model to obtain a first alternative planning scheme corresponding to the target neighborhood; based on the first alternative planning scheme, iterate the target model at least once for local search calculation to obtain a second alternative planning scheme corresponding to the target neighborhood; wherein, when the current planning scheme obtained by any local search calculation meets the set acceptance conditions, the current planning scheme obtained by this local search calculation is used as the first alternative planning scheme to iterate the next local search calculation for the target neighborhood until the second alternative planning scheme is obtained; when the set calculation termination condition is reached, the target planning scheme is output, wherein the target planning scheme is the alternative planning scheme with the smallest eigenvalue among the second alternative planning schemes of the multiple neighborhoods.

[0021] In a possible design, the configuration data of the target neighborhood includes a neighborhood length; when calculating the target model, any alternative logistics node has an attribute value, and the attribute value is used to characterize the node type of the alternative logistics node in the corresponding planning scheme; the calculation unit is used to: randomly select Km first alternative logistics nodes from the multiple alternative logistics nodes according to the neighborhood length Km of the target neighborhood, wherein m represents the number of neighborhoods, and m and Km are integers greater than or equal to 1; according to the specified planning scheme, after perturbing the attribute value of the corresponding first alternative logistics node using the perturbation operator of each first alternative logistics node in the Km first alternative logistics nodes, the target model is calculated to obtain the first alternative planning scheme corresponding to the target neighborhood; wherein the perturbation operator of any first alternative logistics node is randomly selected from the first operator or the second operator, the first operator is used to reduce the attribute value of the alternative logistics node, and the second operator is used to increase the attribute value of the alternative logistics node.

[0022] In a possible design, when calculating the target model, any alternative logistics node has an attribute value, and the attribute value is used to characterize the node type of the alternative logistics node in the corresponding planning scheme; the calculation unit iterates the target model at least once based on the first alternative planning scheme to obtain a second alternative planning scheme corresponding to the target neighborhood, including: performing the following steps for the target local search calculation in the at least one local search calculation, wherein the target local search calculation traverses any one of the at least one local search calculation: determining multiple second alternative logistics nodes from the multiple alternative logistics nodes; according to the first alternative planning scheme, using the perturbation operator of each second alternative logistics node in the multiple second alternative logistics nodes to perturb the attribute value of the corresponding second alternative logistics node, and then calculating the target model. to obtain the current planning scheme of the target local search calculation; wherein, the disturbance operator of each second alternative logistics node is at least one of the first operator, the second operator, and the third operator, the first operator is used to reduce the attribute value of the alternative logistics node, the second operator is used to increase the attribute value of the alternative logistics node, and the third operator is used to exchange the attribute values ​​of two alternative logistics nodes; wherein, if the planning feature value corresponding to the current planning scheme of the target local search calculation is less than the planning feature value corresponding to the first alternative planning scheme, the current planning scheme of the target local search calculation is used as the first alternative planning scheme until the target local search calculation process is terminated; when the first alternative planning scheme meets the set acceptance conditions, the current planning scheme obtained by the last local search calculation in the at least one local search calculation is used as the second alternative planning scheme corresponding to the target neighborhood.

[0023] In a third aspect, an embodiment of the present application provides a computer-readable medium for storing a computer program, wherein the computer program includes instructions for executing the method in any optional implementation of the first aspect.

[0024] Based on the implementations provided in the above aspects, the present application can also be further combined to provide more implementations. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A flow chart of a logistics network planning method according to an embodiment of the present application;

[0026] Figure 2 This is a schematic diagram of the clustering algorithm of the embodiment of the present application;

[0027] Figure 3 A schematic diagram of the variable neighborhood local search principle of an embodiment of the present application;

[0028] Figure 4A flow chart of a logistics network planning method according to an embodiment of the present application;

[0029] Figure 5 A schematic diagram of the coding rules of an embodiment of the present application;

[0030] Figure 6 A schematic diagram of a logistics network planning device according to an embodiment of the present application. DETAILED DESCRIPTION

[0031] The embodiment of the present application provides a logistics network planning method and device, which is helpful for rationally planning a logistics network. The method and device are based on the same technical concept. Since the method and device solve the problem in a similar principle, the implementation of the device and the method can refer to each other, and the repeated parts will not be repeated.

[0032] In the embodiment of the present application, the investment cost and / or delivery delay of the logistics network can be used as optimization conditions, and a variable neighborhood search algorithm (VNS) can be used to design a target model. The target model can be a mathematical model or a physical model, which can be used to characterize the correspondence between the planning scheme and the planning characteristics. Then, the target model is calculated according to the logistics node configuration information and the neighborhood configuration information of the logistics network to be planned, until the target planning scheme is obtained. The target planning scheme can include target planning information, including but not limited to logistics node addresses, logistics node types, logistics node numbers, logistics node service ranges, etc., so as to establish a relatively optimal logistics network based on the target planning scheme. Among them, the variable neighborhood search algorithm can adopt an improved VNS framework, add at least one local search calculation process, and design a disturbance operator to generate a neighborhood in combination with the relevant planning features of the logistics network planning to avoid the algorithm from falling into a local optimum.

[0033] It can be understood that the embodiments of the present application only take the logistics network as an example to illustrate the method of obtaining the planning results of the corresponding network based on the target model without any limitation. The embodiments of the present application can be applied to any network planning scenario with similar characteristics to the logistics network, and the present application does not limit this.

[0034] For ease of understanding, some terms in this application are explained below to facilitate understanding by those skilled in the art.

[0035] Logistics network planning takes the economic benefits of the logistics system and society as its goal. It uses system theory and system engineering methods to comprehensively consider factors such as the supply and demand of materials, transportation conditions, and the natural environment. It studies and designs the number, location, scale, supply range, and proportion of direct supply and transit supply of logistics nodes to establish an efficient logistics network to achieve the goals of low cost, good service, and high efficiency.

[0036] 1) A computing device is a device that provides business services and has data connectivity capabilities.

[0037] In the embodiment of the present application, the computing device may be a terminal device. The terminal device may also be referred to as a terminal device, or may be referred to as a user equipment (UE), a mobile station (MS), a mobile terminal (MT), etc. The terminal device may also be a chip or an application (APP) installed on the terminal device. The terminal device may be a handheld device with a wireless connection function, a vehicle-mounted device, etc. At present, some examples of terminal devices are: mobile phones, tablet computers, laptops, PDAs, mobile internet devices (MID), smart point of sale (POS), wearable devices, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control, wireless terminals in self driving, wireless terminals in remote medical surgery, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes, various smart meters (smart water meters, smart electricity meters, smart gas meters), etc.

[0038] The computing device may also be a server. A server is a device that provides data connectivity services. Since the server can respond to service requests from terminal devices and perform processing, the server should generally have the ability to undertake and guarantee services. In the present application, the server may be a server located in a data network (DN), such as an ordinary server or a server in a cloud platform; or a multi-access edge computing (MEC) server located in a core network, etc. This application does not limit this specific implementation.

[0039] 2) Variable neighborhood search (VNS): An improved local search algorithm, a heuristic approximate algorithm, which can avoid falling into the local optimal solution by jumping and searching in different neighborhood structures. The algorithm is mainly divided into two parts: one is local search, which is to find the local optimal solution in the same neighborhood structure; the other is to change the neighborhood, which is to change the neighborhood structure based on the local optimal solution. The above two steps are iterated alternately to achieve the global optimal solution.

[0040] In the embodiment of the present application, the variable neighborhood search algorithm used is implemented based on the improved VNS framework, and the target model designed by the variable neighborhood search algorithm can be used to characterize the correspondence between planning schemes and planning features. Among them, a planning scheme corresponds to a solution of the target model, the planning feature corresponds to the objective function value used to judge the quality of the solution in the target model, and the target planning scheme is the global optimal solution finally obtained by the target model. It can be understood that since the algorithm is an approximate algorithm, the target planning scheme obtained is only a relatively optimal scheme among multiple feasible planning schemes.

[0041] 3) Neighborhood: It is generally defined as the set of nodes on the problem domain obtained by transforming each node on the given problem domain according to a given transformation rule. In simple terms: Neighborhood refers to the set of all solutions obtained by performing an operation (which can be called neighborhood action) on the current solution. The essential difference between different neighborhoods lies in the difference in neighborhood actions. Neighborhood action is a function that generates the corresponding set of neighbor solutions for the current solution.

[0042] 4) Multiple refers to two or more.

[0043] 5) At least one means one or more.

[0044] 6) "and / or" describes the relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.

[0045] In addition, it should be understood that, in the description of this application, words such as "first" and "second" are only used for the purpose of distinguishing the description, and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying an order.

[0046] The logistics network planning method provided in the embodiment of the present application will be described in detail below with reference to the accompanying drawings.

[0047] Figure 1Schematic diagram of the flow of the logistics network planning method of the embodiment of the present application. The method can be implemented by a computing device, which can be a terminal device or a server, and the present application does not limit this. Figure 1 , the method comprises the following steps:

[0048] S1: The computing device obtains logistics node configuration information and neighborhood configuration information.

[0049] In an embodiment of the present application, the logistics node configuration information may include configuration data of multiple candidate logistics nodes of the logistics network to be planned.

[0050] In one implementation, the computing device may first determine the target area where the logistics node of the logistics network to be planned is located, and then obtain the logistics node configuration information based on the relevant information of the target area. The target area may be an area divided according to at least one granularity. For example, the target area may be a global area; it may also be an area divided by continents, such as Asia, America, Asia-Pacific, etc.; it may be a national area; it may also be one or more provinces / cities in a country, such as Beijing, China, the three provinces in northeast China, etc., and this application does not limit this.

[0051] For the target area, the computing device can select the location of the alternative logistics node in the target area to obtain the logistics node configuration information. For example, the computing device can obtain the alternative information of the target area, and then use the set clustering algorithm to cluster the alternative information of the target area to obtain at least one cluster cluster (referred to as a cluster). Any cluster can correspond to an alternative logistics node. For example, the center of each cluster is used as an alternative logistics node, that is, multiple alternative logistics nodes are obtained. Then, according to the at least one cluster cluster, the logistics node configuration data and the neighborhood configuration information are obtained.

[0052] It should be noted that, in the embodiment of the present application, the alternative information of the target area may include determining some alternative address information that may be set as logistics nodes based on factors such as geology, market, transportation, and environment through investigation and analysis of the target area. In practical applications, since the target area may include ports, warehouses (including existing warehouses / candidate warehouses), sites (also referred to as receiving points), etc., and the occurrence of a large number of logistics activities is closely related to the sites, considering the planning characteristics such as investment costs and delivery delays, the computing device may, for example, use the site information of the target area as the alternative information, and cluster the site information of the target area to determine multiple alternative logistics nodes. It should be understood that clustering the site information of the target area to determine the alternative logistics nodes is only an implementation scheme of the technical solution of the present application and not a limitation. According to actual application scenarios and business needs, the computing device may also choose to cluster other information as the site selection information of the target area to obtain logistics node configuration information, and the present application does not limit this.

[0053] like Figure 2 As shown, a rectangle 10 schematically represents the target area. By clustering the site information of the target area 10, M clusters 20 can be obtained, where M is an integer greater than 1, that is, the centers of the M clusters 20 can be used as the corresponding M alternative logistics nodes. Afterwards, the logistics node configuration information and the neighborhood configuration information can be obtained based on the M alternative logistics nodes. The logistics node configuration information may include any information required to obtain a logistics network planning scheme. For example, the logistics node configuration information may include configuration data of multiple alternative logistics nodes, such as the geographic location information of any alternative logistics node, the type of alternative node, whether it is a deployed logistics node, whether it is a reserved logistics node, etc., which is not limited in this application. Thus, the computing device can know which sites in the target area and their surrounding areas are suitable for setting up logistics nodes by clustering the site information of the target area, and obtain configuration data of multiple alternative logistics nodes.

[0054] In an embodiment of the present application, the domain configuration information may include configuration data of multiple neighborhoods. Wherein, a neighborhood refers to a set of all solutions obtained by performing an operation (which may be referred to as a neighborhood action) on the current solution of the target model. The computing device may be configured according to the logistics node configuration information and / or experience based on the variable neighborhood search algorithm to be adopted to obtain the domain configuration information. The domain configuration information may be used to subsequently calculate the target model using the corresponding variable neighborhood search algorithm to obtain feasible alternative planning schemes corresponding to multiple neighborhoods, until the target planning scheme is obtained.

[0055] For example, the neighborhood configuration information may include an initial neighborhood length and a neighborhood length change rule, wherein the neighborhood length may also be referred to as the neighborhood search step, and the neighborhood length change rule may also be referred to as a neighborhood structure change rule. The computing device may define M neighborhoods, and the neighborhood structure of the mth neighborhood is denoted as Nm, where m=1,2,3...M, and N and M are integers; Km represents the neighborhood length of neighborhood Nm, where Km is an integer greater than or equal to 1, and the neighborhood length of N1 (i.e., the initial neighborhood length) K1 may be configured to be 1 by default. The neighborhood length change rule may be expressed as the following expression (1):

[0056] Km+1=Km%Kmax+K1 (1)

[0057] Among them, Km represents the neighborhood length of neighborhood Nm; Km+1 represents the neighborhood length of neighborhood Nm+1 (i.e., the next neighborhood of neighborhood Nm); Kmax represents the maximum neighborhood length; and K1 represents the initial neighborhood length.

[0058] Through the above expression (1), in the process of subsequently using the variable neighborhood search algorithm to iteratively calculate the target model to obtain the target planning solution, the computing device can search in the neighborhood based on the neighborhood length Km of any neighborhood to obtain the local optimal solution corresponding to the neighborhood, that is, the optimal alternative planning solution corresponding to the neighborhood. After the search of the current neighborhood is completed, the neighborhood structure is changed according to the above expression (1), the neighborhood length of the next neighborhood is obtained, and the search is jumped to the next neighborhood. Thus, different neighborhoods can be iterated by combining the neighborhood length change rule until the target planning solution is obtained.

[0059] It should be noted that in the embodiment of the present application, when the computing device searches for a local optimal solution in the same neighborhood structure, the current solution can be perturbed by a variety of operations, and accordingly, different neighborhood solutions may exist, corresponding to different planning schemes. The purpose of the present application is to search among different alternative planning schemes through iteration based on an improved VNS framework, so as to obtain a target planning scheme. This will be described in detail below in conjunction with the embodiments, and will not be repeated here.

[0060] S2: The computing device establishes a target model according to the logistics node configuration information and the set constraints. The target model is used to characterize the correspondence between the planning scheme and the planning features, wherein any planning scheme includes some or all of the multiple candidate logistics nodes, and the planning features include: investment cost and / or delivery delay.

[0061] In the embodiment of the present application, the computing device can, for example, design at least one constraint condition based on the investment cost and / or delivery delay of the logistics network as the optimization condition, and any constraint condition can be used to characterize the correspondence between the logistics node and the corresponding planning feature. The computing device can establish a target model based on the logistics node configuration information and the set constraints, and the target model can be calculated based on the improved variable neighborhood search algorithm to output the target planning scheme, so that it can be known which of the multiple alternative logistics nodes can be used as target logistics nodes to establish the desired optimal logistics network, as well as the node type of each target logistics node in the logistics network to be established or other planning result information.

[0062] In the application, the constraints may include constraints designed at various levels, including but not limited to various cost constraints, time constraints, variable relationship constraints, and node type constraints, quantity constraints, service object constraints, etc. of logistics nodes, so as to obtain a better planning scheme according to various constraints to achieve the optimization goals such as investment cost and / or delivery delay of the established logistics network. This application does not limit the specific implementation of these constraints.

[0063] In one implementation, all constraints can be implemented as one or more objective functions. In different application scenarios, a target model can be established based on the actual logistics node configuration information and the objective function. Alternatively, the required target model can be obtained by adaptively adjusting the relevant parameters of the objective function based on the actual logistics node configuration information.

[0064] Taking the design of the objective function based on the investment cost as an example, the investment cost may include, for example, transportation cost, storage cost, and fixed cost, etc. Accordingly, the designed objective function may include sub-functions corresponding to each type of cost, as shown in the following expression (2):

[0065] min f=ftransport+fwarehousing+ffixed(2)

[0066] Wherein, min f represents the objective function value, corresponding to the solution of the objective model; ftransport represents the sub-function corresponding to the transportation cost; fwarehousing represents the sub-function corresponding to the warehousing cost; and ffixed represents the sub-function corresponding to the fixed cost. Wherein, when the computing device uses the variable neighborhood search algorithm to calculate the objective model, min f represents the planning characteristic value of the planning characteristic, and the corresponding solution used to obtain the min f is the corresponding planning scheme. For different planning schemes and min f that may appear in the calculation process, it can be determined whether the planning scheme is better by comparing the different min fs. For example, if minf is smaller, the corresponding planning scheme for obtaining the min f is better, and accordingly, the better planning scheme is more suitable to be selected as the target planning scheme to establish the desired logistics network.

[0067] It is understandable that the fixed cost may be, for example, the cost corresponding to several deployed and to-be-retained logistics nodes in the target area. Since the cost corresponding to the deployed and to-be-retained logistics nodes is a definite item, it is called a fixed cost. The cost corresponding to the candidate logistics nodes that have not yet been deployed is an uncertain item and is not included in the fixed cost.

[0068] S3: According to the domain configuration information, a variable neighborhood search algorithm is used to perform at least one local search calculation on the target model to obtain a target planning scheme. The target planning scheme includes a target logistics node among the multiple candidate logistics nodes, and the target investment cost in the target planning feature corresponding to the target planning scheme is lower than a cost threshold, and / or the target delivery delay is lower than a delay threshold.

[0069] The variable neighborhood search algorithm used in the embodiment of the present application is based on an improved VNS framework, adds at least one local search calculation process, and designs a perturbation operator in combination with relevant planning features of logistics network planning to prevent the algorithm from falling into local optimality.

[0070] In an embodiment of the present application, the computing device can perform initialization calculations on the target model to obtain an initial planning scheme. Based on the initial planning scheme, the computing device can use different neighborhood structures to search until a target planning scheme is determined. In the process of searching and calculating for any neighborhood Nm, the computing device uses the neighborhood length Km of Nm to perform at least one local search in the neighborhood. If the current solution obtained by any local search in the neighborhood is improved and acceptable compared to the solution obtained by the previous local search, the next local search is performed based on the latest current solution to further improve the solution, until the last local search in the neighborhood obtains the local optimal solution in this field. If the solution is unacceptable, the next neighborhood Nm+1 is searched until the target solution, that is, the target planning scheme, is obtained. Figure 3As shown, the search is performed in sequence from neighborhood N1 to area N2, and finally to neighborhood Nm to find a better planning solution. The dotted arrow indicates that the neighborhood solution is not improved and the search is skipped to the next neighborhood, and the solid arrow indicates that the neighborhood solution is improved and the search is continued in this area.

[0071] In the embodiment of the present application, the computing device can design a variety of shaking operators for implementing various shaking operations, and can also design the use rules of each shaking operator. In the process of calculating the target model using the variable neighborhood search algorithm, the computing device can use at least one shaking operator to implement the shaking process of the target model within the same neighborhood structure, so as to adjust and update the corresponding neighborhood solution calculated by the target model, obtain more neighborhood solutions, and facilitate iterative calculations in the neighborhood until the local optimal solution of the neighborhood is found. After changing the neighborhood structure, the target model is disturbed by the same or similar method to obtain the local optimal solution of different neighborhoods until the global optimal solution, that is, the target planning solution, is obtained.

[0072] For example, the computing device designs a decision parameter for the candidate logistics node, and the decision parameter is a variable that can be used to judge the pros and cons of the planning scheme. For example, the computing device can use the attribute value as the decision variable, and the attribute value can characterize the node type of the candidate logistics node. The disturbance operator can include, for example, a first operator, a second operator, and a third operator. Among them, the first operator can be used to implement a first disturbance operation, and the first disturbance operation can be used to reduce the attribute value of the candidate logistics node. The second operator is used to implement a second disturbance operation, and the second disturbance operation is used to increase the attribute value of the candidate logistics node. The third operator can be used to implement a third disturbance operation, and the third disturbance operation can be used to exchange the attribute values ​​of two candidate logistics nodes. In the process of calculating the target model using the variable neighborhood search algorithm, the computing device can use the corresponding disturbance operator (for example, random selection) to perform disturbance processing on the target model according to the algorithm configuration, so as to adjust and update the relevant planning schemes calculated based on the target model, so as to obtain more planning schemes, so as to determine the target planning scheme from the more planning schemes, and improve the accuracy of the target model.

[0073] To facilitate understanding, the following takes two local searches within the same neighborhood structure as an example. Figure 4 The flowchart shown introduces the specific implementation steps of S3.

[0074] See also Figure 4 As shown, S3 implemented by the computing device may specifically include the following steps:

[0075] S31: performing initialization calculation on the target model to obtain an initial solution and initialization parameters of the algorithm, that is, an initial planning scheme.

[0076] S32: Determine whether the current solution reaches the calculation termination condition. If so, end the calculation process; otherwise, proceed to S331.

[0077] In the embodiments of the present application, the calculation termination condition can be preset, which can be a time condition or an iteration number condition, etc. The present application does not limit this. Among them, when performing different local searches within the same neighborhood structure or when jumping to the next neighborhood, it is necessary to return to S32 first to determine whether the current solution reaches the calculation termination condition, so as to avoid infinite iterative calculations of the target model.

[0078] S33: According to the domain configuration information, adopt the variable neighborhood search algorithm to perform the following steps for the target neighborhood among multiple neighborhoods, where the target neighborhood traverses each neighborhood among the multiple neighborhoods:

[0079] Refer to Figure 4 As shown, S331: According to the specified planning scheme and the configuration data of the target neighborhood, use at least one perturbation operator to perform perturbation processing on the target model to obtain the first alternative planning scheme corresponding to the target neighborhood. Among them, the specified planning scheme can be the planning scheme corresponding to the improved solution obtained each time during the calculation. For example, if the target neighborhood is neighborhood N1, the specified planning scheme in S331 is the initial planning scheme obtained by S31; if the target neighborhood is neighborhood Ni, 1 < i ≤ m, the specified planning scheme in S331 is the planning scheme corresponding to the optimal solution found within neighborhood Ni-1.

[0080] Specifically, taking neighborhood Nm as the target neighborhood, the configuration data of this target neighborhood Nm can include the neighborhood length Km. When calculating the target model, any alternative logistics node has an attribute value, and the attribute value is used to characterize the node type of the alternative logistics node in the corresponding planning scheme. S331 can include the following steps:

[0081] S3311: Randomly select Km first alternative logistics nodes from the multiple alternative logistics nodes according to the neighborhood length Km of the target neighborhood Nm, where Km is an integer greater than or equal to 1.

[0082] S3312: According to the initial planning scheme, after perturbing the attribute value of the corresponding first candidate logistics node using the perturbation operator of each of the Km first candidate logistics nodes, the target model is calculated to obtain the first candidate planning scheme corresponding to the target neighborhood. Wherein, the perturbation operator of any first candidate logistics node is randomly selected from the first operator or the second operator, the first operator is used to reduce the attribute value of the candidate logistics node, and the second operator is used to increase the attribute value of the candidate logistics node.

[0083] Thus, the computing device randomly selects a specified number of candidate logistics nodes in the target neighborhood Nm, and randomly increases or decreases the current attribute values ​​thereof, thereby realizing the perturbation processing of the target model and updating the neighborhood solution, so as to obtain as many solutions as possible, thereby facilitating the search for the local optimal solution of the target neighborhood. It should be understood that the above is only an example of the perturbation operator and not any limitation. In practical applications, perturbation operators with other functions can also be designed according to application scenarios or business requirements to realize the perturbation processing of the target model, and this application does not limit this.

[0084] S332: Based on the first alternative planning scheme, perform at least one local search calculation on the target model iteration to obtain a second alternative planning scheme corresponding to the target neighborhood. Wherein, when the current planning scheme obtained by any local search calculation satisfies the set acceptance condition, the current planning scheme obtained by this local search calculation is used as the first alternative planning scheme to iterate the next local search calculation for the target neighborhood until the second alternative planning scheme is obtained;

[0085] Specifically, S332 may include the following steps:

[0086] S3321: Execute the following steps for a target local search calculation in the at least one local search calculation, wherein the target local search calculation traverses any one of the at least one local search calculation:

[0087] Determining a plurality of second candidate logistics nodes among the plurality of candidate logistics nodes;

[0088] According to the first alternative planning scheme, after perturbing the attribute value of the corresponding second alternative logistics node using the perturbation operator of each of the plurality of second alternative logistics nodes, the target model is calculated to obtain the current planning scheme of the target local search calculation. Wherein, the perturbation operator of each second alternative logistics node is at least one of the first operator, the second operator, and the third operator, the first operator is used to reduce the attribute value of the alternative logistics node, the second operator is used to increase the attribute value of the alternative logistics node, and the third operator is used to exchange the attribute values ​​of two alternative logistics nodes.

[0089] If the current planning scheme calculated by the target local search is improved, that is, the planning characteristic value corresponding to the current planning scheme calculated by the target local search is less than the planning characteristic value corresponding to the first alternative planning scheme, then the current planning scheme calculated by the target local search is updated and used as the first alternative planning scheme.

[0090] S3322: Determine whether the current planning scheme obtained by the local search calculation of the target meets the set acceptance conditions. If so, perform the next local search calculation on the target model iteration based on the first alternative planning scheme, otherwise end the local search calculation process for the target neighborhood, and enter S34: update the neighborhood length according to the neighborhood length change rule, and search the next neighborhood according to the updated neighborhood length, wherein, when the updated neighborhood length exceeds the maximum neighborhood length, return to the initial neighborhood for search.

[0091] S3323: When the first alternative planning scheme meets the set acceptance condition, the current planning scheme obtained by the last local search calculation in the at least one local search calculation is used as the second alternative planning scheme corresponding to the target neighborhood.

[0092] S3324: Return to the initial neighborhood N1 for searching, and the neighborhood length is K1=1.

[0093] It should be noted that in the embodiments of the present application, the specific implementation of each local search calculation in at least one local search calculation may be different, and the present application does not limit this. Figure 4 As shown, taking the variable neighborhood search calculation including two local searches as an example, for the sake of distinction, S3321-1 represents the first local search calculation, S3322-1 represents the judgment process of the result of the first local search calculation, S3321-2 represents the second local search calculation, and S3322-2 represents the judgment process of the result of the second local search calculation. In the application, three or more local search calculations can also be designed as the upper limit of the number of local search calculations according to the application scenario or business needs, so as to avoid the algorithm from falling into the local optimum, which is not limited in this application. Accordingly, the specific implementation of any local search algorithm can include any combination of at least one of the above steps, which is not limited in this application.

[0094] S32: When the set calculation termination condition is reached, the target planning scheme is output, wherein the target planning scheme is the alternative planning scheme with the smallest eigenvalue among the second alternative planning schemes in the multiple neighborhoods.

[0095] Thus, through Figure 4The flowchart shown adopts the variable neighborhood search algorithm to calculate the target model. Through at least one local search calculation, it is possible to avoid falling into the local optimum and finally obtain a relatively better target planning solution.

[0096] For ease of understanding, the implementation of the logistics network planning method of the present application is explained below by taking the establishment of a secondary logistics network as an example.

[0097] Specifically, the logistics node configuration information may include coding rules for multiple candidate logistics nodes. Figure 5 As shown in the figure, for example, for the secondary logistics network to be established, the attribute values ​​0, 1, and 2 can be used to represent the node type of the logistics node, which respectively represent not building a warehouse, building a primary logistics node, and building a secondary logistics node. The attribute value is used as a decision variable to input the target model, so as to calculate the target model to obtain the corresponding planning scheme. Among them, the initialized attribute values ​​of multiple candidate logistics nodes can be obtained by initializing the target model. Then, during the calculation process, the attribute values ​​of the candidate logistics nodes may change in the planning schemes corresponding to different solutions.

[0098] It should be understood that this is only an example and not a limitation. In practical applications, if you want to establish a three-level or more level logistics network, you can use the same or similar method to adjust the coding rules of the alternative logistics nodes in the multi-level logistics network and the values ​​of the decision variables to adapt to the target model of this application and obtain the target planning scheme of the multi-level logistics network. This application does not limit this. Alternatively, when planning other types of networks or using other planning features, you can adaptively modify parameters such as decision variables. This application does not limit this.

[0099] In the process of calculating the target model using the variable neighborhood search algorithm, the following steps may be specifically included:

[0100] (1) Generate an initial solution to the target model, i.e., an initial planning scheme;

[0101] (2) Find the target solution based on the initial solution, that is, the target planning solution;

[0102] (3) Based on the neighborhood length Km of the target neighborhood (e.g., Km=1), the target neighborhood is perturbed to generate a new neighborhood solution S;

[0103] (4) Determine whether the set calculation termination condition is met. If the calculation termination condition is not met, proceed to step (5):

[0104] (5) According to the neighborhood solution S, the target model is perturbed using at least one perturbation operator to obtain a neighborhood solution S', that is, a first alternative planning scheme corresponding to the target neighborhood;

[0105] For example, the at least one perturbation operator may include a first operator or a second operator.

[0106] In the secondary logistics network to be established, the first operator can also be called the ToMin operator, which is used to randomly select an alternative logistics node and adjust its current attribute value to the minimum value of two convertible situations. For example, the currently selected alternative logistics node is a primary logistics node, that is, the attribute value is 1, then the ToMin operator will adjust its attribute value to 0, that is, not to establish a primary logistics node there. The second operator can also be called the ToMax operator, which is used to randomly select an alternative logistics node and adjust its current attribute value to the maximum value of two convertible situations. For example: the currently selected alternative logistics node is a primary logistics node, that is, the attribute value is 1, then the ToMax operator will adjust its attribute value to 2, that is, to establish a secondary logistics node there.

[0107] When the target model is perturbed using at least one perturbation operator in step (5), the following steps may be included, for example:

[0108] (5-1) According to the neighborhood length Km of the target neighborhood, randomly select Km candidate logistics nodes, i.e., the first candidate logistics node;

[0109] (5-2) For each selected first candidate logistics node, randomly select one of the ToMin operator and the ToMax operator to operate on it;

[0110] (5-3) Update the current solution S', that is, obtain the first alternative planning scheme.

[0111] (6) Perform the first local search based on S' to obtain S", S" is the current solution obtained after the first local search calculation is completed.

[0112] The first local search may specifically include the following:

[0113] (6-1) Select multiple second candidate logistics nodes from multiple candidate logistics nodes, and use a third operator (also called an exchange operator) to exchange attribute values ​​of each candidate logistics node in the dictionary order to search. Each exchange operation corresponds to a round of calculation process, and the corresponding current solution will be obtained. If the current solution is improved, the current solution is updated and the next round of iteration is performed;

[0114] (6-2) Perform ToMin operation on each second candidate logistics node. Each ToMin operation corresponds to a round of calculation process, and the corresponding current solution will be obtained. If the current solution is improved, the current solution is updated and the next round of iteration is performed;

[0115] (6-3) Perform ToMax operation on each second candidate logistics node. Any ToMax operation corresponds to a round of calculation process, and the corresponding current solution will be obtained. If the current solution is improved, the current solution is updated and the next round of iteration is performed;

[0116] Therefore, in the first local search, multiple rounds of iterations are performed through the above steps (6-1), (6-2), and (6-3) to find the current solution S".

[0117] (7) If f(S”)≤f(S), where S is the relative global optimal solution of the target model, and S” satisfies the acceptance condition, then proceed to step (8); otherwise, update the neighborhood length based on the neighborhood length change rule Km+1=Km%Kmax+K1, and return to (3).

[0118] (8) Perform a second local search based on S' to obtain S'';

[0119] The second local search may specifically include the following:

[0120] A plurality of second candidate logistics nodes are selected from the plurality of candidate logistics nodes, and according to the current attribute value of each candidate logistics node, the third operator is used to search for all the candidate logistics nodes by exchanging the attribute values ​​two by two, wherein any exchange operation corresponds to a round of calculation process, and the corresponding current solution will be obtained. If the current solution is improved, the current solution is updated and the next round of iteration is performed.

[0121] Therefore, in the second local search, the current solution S'' is found through multiple rounds of iterations.

[0122] (9) If f(S'') ≤ f(S), where S is the relative global optimal solution of the target model, then S'' is taken as S and return to (3).

[0123] Therefore, through the above scheme, when establishing a two-level logistics network, based on the variable neighborhood search algorithm adopted, at least one local search is added, thereby enhancing the local optimization ability of the algorithm, which helps to avoid falling into the local optimum. It should be understood that the above scheme is also applicable to three-level or multi-level logistics networks, as well as other networks with the same or similar structure, and this application does not limit this.

[0124] Based on the same technical concept, the embodiment of the present application also provides a logistics network planning device, such as Figure 6 As shown, the logistics network planning device 600 may include: an information acquisition unit 610, a modeling unit 620 and a calculation unit 630.

[0125] The information acquisition unit 610 is used to acquire logistics node configuration information and neighborhood configuration information; wherein the logistics node configuration information includes configuration data of multiple candidate logistics nodes of the logistics network to be planned, and the domain configuration information includes configuration data of multiple neighborhoods;

[0126] A modeling unit 620 is used to establish a target model according to the logistics node configuration information and the set constraint conditions; wherein the target model is used to characterize the corresponding relationship between the planning scheme and the planning characteristics, wherein any planning scheme includes some or all of the multiple candidate logistics nodes, and the planning characteristics include: investment cost and / or delivery delay;

[0127] The calculation unit 630 is used to perform at least one local search calculation on the target model using a variable neighborhood search algorithm according to the domain configuration information to obtain a target planning scheme; wherein the target planning scheme includes a target logistics node among the multiple alternative logistics nodes, and the target investment cost in the target planning feature corresponding to the target planning scheme is lower than a cost threshold, and / or the target delivery delay is lower than a delay threshold.

[0128] In a possible design, the information acquisition unit 610 is used to: acquire candidate information of the target area; cluster the candidate information of the target area using a set clustering algorithm to obtain at least one cluster, and any cluster corresponds to an alternative logistics node; and acquire the logistics node configuration information and the neighborhood configuration information according to the at least one cluster. For example, the candidate information may be site information of the target area.

[0129] In a possible design, the calculation unit 630 is used to: calculate the target model according to the domain configuration information to obtain an initial planning scheme; use a variable neighborhood search algorithm to perform the following steps for a target neighborhood in multiple neighborhoods, wherein the target neighborhood traverses each of the multiple neighborhoods: according to the initial planning scheme and the configuration data of the target neighborhood, use at least one perturbation operator to perturb the target model to obtain a first alternative planning scheme corresponding to the target neighborhood; based on the first alternative planning scheme, iterate the target model at least once for local search calculation to obtain a second alternative planning scheme corresponding to the target neighborhood; wherein, when the current planning scheme obtained by any local search calculation meets the set acceptance condition, the current planning scheme obtained by this local search calculation is used as the first alternative planning scheme to iterate the next local search calculation for the target neighborhood until the second alternative planning scheme is obtained; when the set calculation termination condition is reached, the target planning scheme is output, wherein the target planning scheme is the alternative planning scheme with the smallest eigenvalue among the second alternative planning schemes of the multiple neighborhoods.

[0130] In a possible design, the configuration data of the target neighborhood includes a neighborhood length; when calculating the target model, any alternative logistics node has an attribute value, and the attribute value is used to characterize the node type of the alternative logistics node in the corresponding planning scheme; the calculation unit is used to: randomly select Km first alternative logistics nodes from the multiple alternative logistics nodes according to the neighborhood length Km of the target neighborhood, wherein m represents the number of neighborhoods, and m and Km are integers greater than or equal to 1; according to the initial planning scheme, after perturbing the attribute value of the corresponding first alternative logistics node using the perturbation operator of each first alternative logistics node in the Km first alternative logistics nodes, the target model is calculated to obtain the first alternative planning scheme corresponding to the target neighborhood; wherein the perturbation operator of any first alternative logistics node is randomly selected from the first operator or the second operator, the first operator is used to reduce the attribute value of the alternative logistics node, and the second operator is used to increase the attribute value of the alternative logistics node.

[0131] In a possible design, when calculating the target model, any alternative logistics node has an attribute value, and the attribute value is used to characterize the node type of the alternative logistics node in the corresponding planning scheme; the calculation unit iterates the target model at least once based on the first alternative planning scheme to obtain a second alternative planning scheme corresponding to the target neighborhood, including: performing the following steps for the target local search calculation in the at least one local search calculation, wherein the target local search calculation traverses any one of the at least one local search calculation: determining multiple second alternative logistics nodes from the multiple alternative logistics nodes; according to the first alternative planning scheme, using the perturbation operator of each second alternative logistics node in the multiple second alternative logistics nodes to perturb the attribute value of the corresponding second alternative logistics node, and then calculating the target model. to obtain the current planning scheme of the target local search calculation; wherein, the disturbance operator of each second alternative logistics node is at least one of the first operator, the second operator, and the third operator, the first operator is used to reduce the attribute value of the alternative logistics node, the second operator is used to increase the attribute value of the alternative logistics node, and the third operator is used to exchange the attribute values ​​of two alternative logistics nodes; wherein, if the planning feature value corresponding to the current planning scheme of the target local search calculation is less than the planning feature value corresponding to the first alternative planning scheme, the current planning scheme of the target local search calculation is used as the first alternative planning scheme until the target local search calculation process is terminated; when the first alternative planning scheme meets the set acceptance conditions, the current planning scheme obtained by the last local search calculation in the at least one local search calculation is used as the second alternative planning scheme corresponding to the target neighborhood.

[0132] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) that contain computer-usable program code.

[0133] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0134] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0135] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A logistics network planning method, characterized in that: Methods include: Obtaining logistics node configuration information and neighborhood configuration information; wherein the logistics node configuration information includes configuration data of multiple candidate logistics nodes of the logistics network to be planned, and the neighborhood configuration information includes configuration data of multiple neighborhoods; Establishing a target model according to the logistics node configuration information and set constraints; wherein the target model is used to characterize the correspondence between planning schemes and planning features, wherein any planning scheme includes some or all of the multiple candidate logistics nodes, and the planning features include: investment cost and / or delivery delay; According to the neighborhood configuration information, a variable neighborhood search algorithm is adopted, and at least one disturbance operator is used to perform at least one variable neighborhood local search calculation on the target model to obtain a target planning scheme; wherein the target planning scheme includes a target logistics node among the multiple candidate logistics nodes, and the target investment cost in the target planning feature corresponding to the target planning scheme is lower than a cost threshold, and / or the target delivery delay is lower than a delay threshold; According to the neighborhood configuration information, a variable neighborhood search algorithm is adopted, and at least one disturbance operator is used to perform at least one variable neighborhood local search calculation on the target model to obtain a target planning solution, including: Obtain designated planning scheme; According to the neighborhood configuration information, a variable neighborhood search algorithm is used to perform the following steps for a target neighborhood in multiple neighborhoods: According to the specified planning scheme and the configuration data of the target neighborhood, the target model is perturbed by using at least one perturbation operator to obtain a first candidate planning scheme corresponding to the target neighborhood; Based on the first candidate planning scheme, iteratively perform at least one local search calculation on the target model to obtain a second candidate planning scheme corresponding to the target neighborhood; wherein, when the current planning scheme obtained by any local search calculation satisfies the set acceptance condition, the current planning scheme obtained by this local search calculation is used as the first candidate planning scheme to iteratively perform the next local search calculation for the target neighborhood until the second candidate planning scheme is obtained; When the set calculation termination condition is reached, the target planning scheme is output, wherein the target planning scheme is the alternative planning scheme with the smallest characteristic value among the second alternative planning schemes in multiple neighborhoods.

2. The method according to claim 1, characterized in that Acquiring the logistics node configuration information and the neighborhood configuration information includes: Obtain alternative information of the target area; Clustering the candidate information of the target area using a set clustering algorithm to obtain at least one cluster, and any cluster corresponds to a candidate logistics node; The logistics node configuration information and the neighborhood configuration information are acquired according to the at least one cluster.

3. The method according to claim 1, characterized in that The configuration data of the target neighborhood includes the neighborhood length; when calculating the target model, any candidate logistics node has an attribute value, and the attribute value is used to characterize the node type of the candidate logistics node in the corresponding planning scheme; According to the specified planning scheme and the configuration data of the target neighborhood, the target model is perturbed by using at least one perturbation operator to obtain a first candidate planning scheme corresponding to the target neighborhood, including: According to the neighborhood length Km of the target neighborhood, randomly select Km first candidate logistics nodes from the multiple candidate logistics nodes, where m represents the number of neighborhoods, and m and Km are integers greater than or equal to 1; According to the specified planning scheme, after the attribute value of the corresponding first alternative logistics node is disturbed by using the disturbance operator of each first alternative logistics node in the Km first alternative logistics nodes, the target model is calculated to obtain the first alternative planning scheme corresponding to the target neighborhood; wherein, the disturbance operator of any first alternative logistics node is randomly selected from the first operator or the second operator, the first operator is used to reduce the attribute value of the alternative logistics node, and the second operator is used to increase the attribute value of the alternative logistics node.

4. The method according to claim 1 or 3, characterized in that: When calculating the target model, any candidate logistics node has an attribute value, and the attribute value is used to characterize the node type of the candidate logistics node in the corresponding planning scheme; Based on the first alternative planning scheme, iteratively performing at least one local search calculation on the target model to obtain a second alternative planning scheme corresponding to the target neighborhood, including: The following steps are performed for a target local search calculation in the at least one local search calculation, wherein the target local search calculation traverses any one of the at least one local search calculation: Determining a plurality of second candidate logistics nodes among the plurality of candidate logistics nodes; According to the first alternative planning scheme, after perturbing the attribute value of the corresponding second alternative logistics node using the perturbation operator of each of the multiple second alternative logistics nodes, the target model is calculated to obtain the current planning scheme of the target local search calculation; wherein the perturbation operator of each second alternative logistics node is at least one of the first operator, the second operator, and the third operator, the first operator is used to reduce the attribute value of the alternative logistics node, the second operator is used to increase the attribute value of the alternative logistics node, and the third operator is used to exchange the attribute values ​​of two alternative logistics nodes; If the planning characteristic value corresponding to the current planning scheme calculated by the target local search is less than the planning characteristic value corresponding to the first alternative planning scheme, the current planning scheme calculated by the target local search is used as the first alternative planning scheme until the target local search calculation process is terminated; When the first alternative planning scheme meets the set acceptance condition, the current planning scheme obtained by the last local search calculation in the at least one local search calculation is used as the second alternative planning scheme corresponding to the target neighborhood.

5. A logistics network planning device, characterized in that: include: An information acquisition unit, used to acquire logistics node configuration information and neighborhood configuration information; wherein the logistics node configuration information includes configuration data of multiple candidate logistics nodes of the logistics network to be planned, and the neighborhood configuration information includes configuration data of multiple neighborhoods; A modeling unit, used to establish a target model according to the logistics node configuration information and set constraints; wherein the target model is used to characterize the corresponding relationship between the planning scheme and the planning characteristics, wherein any planning scheme includes some or all of the multiple candidate logistics nodes, and the planning characteristics include: investment cost and / or delivery delay; A calculation unit is used to use a variable neighborhood search algorithm according to the neighborhood configuration information, and to use at least one disturbance operator to perform at least one variable neighborhood local search calculation on the target model to obtain a target planning scheme; wherein the target planning scheme includes a target logistics node among the multiple candidate logistics nodes, and the target investment cost in the target planning feature corresponding to the target planning scheme is lower than a cost threshold, and / or the target delivery delay is lower than a delay threshold; Wherein, the computing unit is used for: Obtain designated planning scheme; According to the neighborhood configuration information, a variable neighborhood search algorithm is used to perform the following steps for a target neighborhood in multiple neighborhoods: According to the specified planning scheme and the configuration data of the target neighborhood, the target model is perturbed by using at least one perturbation operator to obtain a first candidate planning scheme corresponding to the target neighborhood; Based on the first candidate planning scheme, iteratively perform at least one local search calculation on the target model to obtain a second candidate planning scheme corresponding to the target neighborhood; wherein, when the current planning scheme obtained by any local search calculation satisfies the set acceptance condition, the current planning scheme obtained by this local search calculation is used as the first candidate planning scheme to iteratively perform the next local search calculation for the target neighborhood until the second candidate planning scheme is obtained; When the set calculation termination condition is reached, the target planning scheme is output, wherein the target planning scheme is the alternative planning scheme with the smallest characteristic value among the second alternative planning schemes in multiple neighborhoods.

6. The device according to claim 5, characterized in that The information acquisition unit is used for: Obtain alternative information of the target area; Clustering the candidate information of the target area using a set clustering algorithm to obtain at least one cluster, and any cluster corresponds to a candidate logistics node; The logistics node configuration information and the neighborhood configuration information are acquired according to the at least one cluster.

7. The device according to claim 5, characterized in that The configuration data of the target neighborhood includes the neighborhood length; when calculating the target model, any candidate logistics node has an attribute value, and the attribute value is used to characterize the node type of the candidate logistics node in the corresponding planning scheme; the calculation unit is used to: According to the neighborhood length Km of the target neighborhood, randomly select Km first candidate logistics nodes from the multiple candidate logistics nodes, where m represents the number of neighborhoods, and m and Km are integers greater than or equal to 1; According to the specified planning scheme, after the attribute value of the corresponding first alternative logistics node is disturbed by using the disturbance operator of each first alternative logistics node in the Km first alternative logistics nodes, the target model is calculated to obtain the first alternative planning scheme corresponding to the target neighborhood; wherein, the disturbance operator of any first alternative logistics node is randomly selected from the first operator or the second operator, the first operator is used to reduce the attribute value of the alternative logistics node, and the second operator is used to increase the attribute value of the alternative logistics node.

8. The device according to claim 5 or 7, characterized in that When calculating the target model, any candidate logistics node has an attribute value, and the attribute value is used to characterize the node type of the candidate logistics node in the corresponding planning scheme; The computing unit iteratively performs at least one local search calculation on the target model based on the first candidate planning scheme to obtain a second candidate planning scheme corresponding to the target neighborhood, including: The following steps are performed for a target local search calculation in the at least one local search calculation, wherein the target local search calculation traverses any one of the at least one local search calculation: Determining a plurality of second candidate logistics nodes among the plurality of candidate logistics nodes; According to the first alternative planning scheme, after perturbing the attribute value of the corresponding second alternative logistics node using the perturbation operator of each of the multiple second alternative logistics nodes, the target model is calculated to obtain the current planning scheme of the target local search calculation; wherein the perturbation operator of each second alternative logistics node is at least one of the first operator, the second operator, and the third operator, the first operator is used to reduce the attribute value of the alternative logistics node, the second operator is used to increase the attribute value of the alternative logistics node, and the third operator is used to exchange the attribute values ​​of two alternative logistics nodes; Among them, if the planning characteristic value corresponding to the current planning scheme calculated by the target local search is less than the planning characteristic value corresponding to the first alternative planning scheme, then the current planning scheme calculated by the target local search is used as the first alternative planning scheme, wherein, if the planning characteristic value corresponding to the current planning scheme calculated by the target local search is less than the planning characteristic value corresponding to the first alternative planning scheme, then the current planning scheme calculated by the target local search is used as the first alternative planning scheme, until the target local search calculation process is terminated; When the first alternative planning scheme meets the set acceptance condition, the current planning scheme obtained by the last local search calculation in the at least one local search calculation is used as the second alternative planning scheme corresponding to the target neighborhood.

9. A computer-readable storage medium, characterized in that: The method comprises a program or an instruction. When the program or the instruction is executed, the method according to any one of claims 1 to 4 is performed.

Images