Logistics network planning system and logistics network planning method

Abstract

To create optimal plans for logistics networks with minimal changes to entities such as warehouses and transporters (transportation means).SOLUTION: A logistics network planning system receives, as an input, existing logistics network data that is data relating to an existing logistics network constituted with a plurality of package routes including two or more warehouses from a first departure warehouse to a last arrival warehouse of a package as components. The system analyzes the existing logistics network data to identify a logistics network feature that is a feature of the existing logistics network. The system constructs an optimization problem that minimizes a target based on the logistics network feature and derives an optimum package route by solving the optimization problem. The system outputs logistics network plan data that is data representing a planned logistics network including the derived package route.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates generally to logistics network planning. [Background technology]

[0002] Regarding transportation planning, for example, a technique disclosed in Patent Document 1 is known. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-48586 Summary of the Invention [Problem to be solved by the invention]

[0004] With the expansion of the e-commerce market, the volume of small parcels handled, including home delivery, is increasing year by year. These small parcels require transportation and delivery to a variety of destinations, unlike in the past. In current logistics networks, vehicles, ferries, trains, etc. are used as the means of delivery (also known as transporters) from each warehouse (logistics network hub) to another warehouse, and these are generally scheduled services (means of transporting and delivering between warehouses at predetermined times). However, scheduled services are limited in terms of the time periods during which they are available and the destinations to which they are sent, making it difficult to transport and deliver each parcel directly to its destination.

[0005] In actual operation, packages generated in regions for delivery to individual destinations are first collected at a core warehouse in that region, from which they are sorted and delivered according to their destination. However, due to the nature of this operation, the core warehouse bears the entire load, and / or even if there are regular services that can deliver between warehouses in regions, the packages are collected at the core warehouse once, so there is a possibility that the load factor will decrease on regular services that directly connect regions.

[0006] As such, transport between warehouses in a logistics network that includes so-called TC (Transfer Center) warehouses can cause problems such as uneven loads between warehouses and / or reduced loading rates of transport means. Possible ways to solve these problems include changing the sorting capabilities or roles of warehouses or reducing the number of transport means, but frequent changes to the entities such as warehouses and transport means result in a large load. [Means for solving the problem]

[0007] The logistics network planning system inputs existing logistics network data, which is data regarding an existing logistics network consisting of multiple cargo routes, each of which consists of two or more warehouses, from the first departure warehouse to the last arrival warehouse. The system analyzes the existing logistics network data to identify logistics network features that are characteristic of the existing logistics network. The system formulates an optimization problem based on the logistics network features, which minimizes a target based on one or more indicators, and derives an optimal cargo route by solving the optimization problem. The system outputs logistics network plan data, which is data representing a planned logistics network including the derived cargo routes. [Effects of the Invention]

[0008] According to the present invention, it is possible to create an optimal plan for a logistics network with minimal changes to entities such as warehouses and transporters (transportation means). Objects, configurations, and effects other than those described above will become clear from the following description of the preferred embodiment of the invention. [Brief explanation of the drawings]

[0009] [Figure 1] 1 is a configuration diagram of an entire system including a logistics network planning system according to a first embodiment. [Figure 2] FIG. 2 is a diagram showing the structure of package data. [Figure 3] FIG. 2 is a diagram illustrating the configuration of vehicle data. [Figure 4] FIG. 10 is a diagram illustrating the structure of warehouse data. [Figure 5] FIG. 2 is a diagram illustrating the configuration of graph data. [Figure 6] 1 is a flowchart of a process executed by the logistics network planning system 100. [Figure 7] 10 is a flowchart of process 602. [Figure 8] FIG. 10 is a schematic diagram of the construction of clustering data in process 704. [Figure 9A] FIG. 1 is a schematic diagram of a first cluster. [Figure 9B] FIG. 10 is a schematic diagram of a second cluster. [Figure 10] FIG. 1 is a schematic diagram illustrating an example of a logistics network according to a first embodiment. [Figure 11] 10 is a flowchart of process 603. [Figure 12] 10 is a flowchart of process 604. [Figure 13] 12 is a flowchart of process 1204. [Figure 14] 12 is a flowchart of process 1206. [Figure 15] FIG. 10 is a schematic diagram showing an example of a logistics network screen. [Figure 16] FIG. 10 is a schematic diagram illustrating an example of path candidate generation. [Figure 17] FIG. 10 is a schematic diagram showing an example of the relationship between deviation from the warehouse role and costs. [Figure 18] FIG. 10 is a schematic diagram illustrating an example of suppressing the maximum warehouse load. [Figure 19] FIG. 10 is a schematic diagram showing an example of limiting path candidates. DETAILED DESCRIPTION OF THE INVENTION

[0010] In the following description, an "interface apparatus" may refer to one or more interface devices, which may be at least one of the following: One or more I / O (Input / Output) interface devices. The I / O (Input / Output) interface devices are interface devices for at least one of the I / O device and a remote display computer. The I / O interface device for the display computer may be a communications interface device. The at least one I / O device may be a user interface device, for example, either an input device such as a keyboard and a pointing device, or an output device such as a display device. One or more communication interface devices. The one or more communication interface devices may be one or more homogeneous communication interface devices (e.g., one or more NICs (Network Interface Cards)) or two or more heterogeneous communication interface devices (e.g., an NIC and an HBA (Host Bus Adapter)).

[0011] In the following description, "memory" refers to one or more memory devices, which are an example of one or more storage devices, and may typically be a primary storage device. At least one memory device in the memory may be a volatile memory device or a non-volatile memory device.

[0012] In the following description, a "persistent storage device" may refer to one or more persistent storage devices, which are an example of one or more storage devices. A persistent storage device may typically be a non-volatile storage device (e.g., an auxiliary storage device), and more specifically, may be, for example, a hard disk drive (HDD), a solid state drive (SSD), a non-volatile memory express (NVME) drive, or a storage class memory (SCM).

[0013] Also, in the following description, "storage" may be "memory" and / or "persistent storage."

[0014] Furthermore, in the following description, a "processor" may refer to one or more processor devices. The at least one processor device may typically be a microprocessor device such as a CPU (Central Processing Unit), but may also be another type of processor device such as a GPU (Graphics Processing Unit). The at least one processor device may be single-core or multi-core. The at least one processor device may also be a processor core. The at least one processor device may also be a processor device in a broader sense, such as a circuit that is a collection of gate arrays written in a hardware description language that performs some or all of the processing (for example, an FPGA (Field-Programmable Gate Array), a CPLD (Complex Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit)).

[0015] In the following description, information that provides an output for an input may be described using expressions such as "xxx table." However, this information may be data of any structure (for example, structured data or unstructured data), or may be a neural network that generates an output for an input, or a learning model such as a genetic algorithm or random forest. Therefore, "xxx table" may be referred to as "xxx data." In the following description, the structure of each table is an example, and one table may be divided into two or more tables, or all or part of two or more tables may be one table.

[0016] In the following description, functions may be described using the expression "yyy unit." However, the functions may be realized by one or more computer programs executed by a processor, by one or more hardware circuits (e.g., FPGAs or ASICs), or by a combination thereof. When a function is realized by a program executed by a processor, the specified processing is performed using a storage device and / or an interface device, etc., as appropriate, and therefore the function may be considered to be at least a part of the processor. Processing described using a function as the subject may be processing performed by a processor or a device having the processor. A program may be installed from a program source. The program source may be, for example, a program distribution computer or a computer-readable storage medium (e.g., a non-transitory storage medium). The description of each function is merely an example; multiple functions may be combined into one function, or one function may be divided into multiple functions.

[0017] Furthermore, any information (for example, at least one of "ID," "name," and "number") may be employed as information for identifying an element (identification information, identifier).

[0018] In addition, in the following description, when describing elements of the same type without distinguishing between them, common reference symbols will be used, and when describing elements of the same type with distinction between them, reference symbols will be used.

[0019] In the following description, the unit of "time" may be a unit that is coarser or finer than the year, month, hour, and minute.

[0020] Hereinafter, an embodiment of a logistics network planning system according to the present invention will be described with reference to the drawings. [Example]

[0021] FIG. 1 is a configuration diagram of the entire system including a logistics network planning system 100 according to the first embodiment.

[0022] The logistics network planning system 100 creates a logistics network plan and outputs logistics network plan data, which is data representing the created logistics network plan. The logistics network plan created in this embodiment is a plan for distributing the load on each warehouse in an existing logistics network. In this embodiment, the logistics network planning system 100 is a physical computer system (one or more computers), but it may alternatively be a logical computer system based on a physical computer system (for example, a system as a cloud computer service based on a cloud infrastructure).

[0023] The logistics network planning system 100 includes an interface device 51, a storage device 52, and a processor 53 connected thereto.

[0024] The interface device 51 communicates with external systems, such as at least one of the user terminal 30 and the transportation planning system 40, via a communication network 70 (such as the Internet or a wide area network (WAN)).

[0025] The user terminal 30 is an information processing terminal (e.g., a personal computer or a smartphone) and is used by a user. The logistics network planning system 100 may be a server system, and the user terminal 30 may be a client system. The user terminal 30 receives logistics network plan data from the logistics network planning system 100 and displays information related to the logistics network plan. An example of a user is a user. A user is a person who creates a transportation plan based on the logistics network plan. The transportation plan is a plan that represents how transportation will be performed by multiple vehicles in the logistics network represented by the logistics network plan, and includes, for example, the arrival time and / or departure time for each of the multiple vehicles from two or more warehouses out of multiple warehouses in the logistics network represented by the logistics network plan. Instead of or in addition to the user terminal 30, an input device and an output device as a user interface device may be connected to the interface device 51.

[0026] The transportation planning system 40 is a physical or logical computer system. The transportation planning system 40 receives logistics network plan data from the logistics network planning system 100, formulates a transportation plan based on the logistics network plan data, and outputs transportation plan data representing the formulated transportation plan.

[0027] At least a portion of the logistics network plan data output by the logistics network planning system 100 and the transportation plan data output by the transportation planning system 40 or the user terminal 30 may be transmitted to at least one of a warehouse terminal, a vehicle, and a support system. The warehouse terminal is an information processing terminal installed in a warehouse. A vehicle is an example of a transportation means, and arrives at and / or departs from a warehouse. The support system is an example of a physical or logical computer system, and performs information processing using at least a portion of the logistics network plan data and the transportation plan data, and transmits information or instructions to an external device such as a warehouse terminal or a vehicle. For example, a control device installed in a vehicle may receive data or instructions from the logistics network planning system 100 or the support system, and control operation based on the received data or instructions, or display the received data or instructions on an in-vehicle display.

[0028] The storage device 52 stores data and programs. The data includes package data 101, vehicle data 102, and warehouse data 103. At least a part of the data 101 to 103 may be stored in advance or may be input from the user terminal 30.

[0029] The processor 53 executes the program to realize a logistics network analysis unit 120, a logistics network optimization unit 130, and an input / output unit 140. The logistics network analysis unit 120 analyzes the logistics network (e.g., the characteristics of the logistics network). The logistics network optimization unit 130 derives a package route based on the analysis results.

[0030] The logistics network analysis unit 120 includes a warehouse role estimation unit 121 and a similar warehouse classification unit 122. The logistics network optimization unit 130 includes a parcel route optimization unit 131. The input / output unit 140 allows the user to select data to be used in calculations, set calculation conditions, and display the calculation results of the system 100. Furthermore, "transportation" may also be broadly defined as "shipping and delivery." When it is necessary to distinguish between "transportation" and "delivery," for example, the movement of parcels between warehouses may be defined as "transportation" in the narrow sense, and the movement of parcels from a warehouse to a delivery destination (e.g., the parcel's destination address) may be defined as "delivery."

[0031] FIG. 10 is a schematic diagram showing an example of a logistics network according to this embodiment.

[0032] In FIG. 10, one or more warehouses 1001 are located in regions 1 to 6 (an example of multiple regions). When there is a means of transportation between warehouses 1001, it is represented by connecting each of them with a line. That is, the logistics network is represented by a graph such as a DG (Directed Graph). In the DG, vertices correspond to warehouses 1001, and edges connecting the vertices define which warehouse 1001 is the source and which warehouse 1001 is the destination between the warehouses 1001. Therefore, the direction of the edge is the direction of vehicle movement. In the logistics network, transport means move between warehouses in accordance with the structure of the DG representing the logistics network, and goods are unloaded and / or loaded at each warehouse.

[0033] 2 is a diagram showing the structure of the package data 101. In the following description, an element having a number or ID of "X" may be written as element "X".

[0034] The package data 101 indicates various information about packages transported on a logistics network. Specifically, for example, the package data 101 is made up of a first package table 201 and a second package table 202.

[0035] The first parcel table 201 has information about the parcel itself. For example, the first parcel table 201 has a row for each parcel, and the row has information such as a parcel number 211, departure location 212, arrival location 213, weight 214, arrival time 215, and delivery date 216. The parcel number 211 represents a parcel number for identifying the parcel. The departure location 212 represents the ID of the departure location (warehouse as the initial departure point). The arrival location 213 represents the ID of the arrival location (final arrival point). The weight 214 represents the weight of the parcel. The arrival time 215 represents the arrival time of the parcel. The delivery date 216 represents the delivery date of the parcel.

[0036] The second parcel table 202 has information about the parcel route, which is the route taken by the parcel. The second parcel table 202 has a row for each parcel route element. For each parcel, a "parcel route element" is an element of the parcel route, specifically a pair of an edge and its two end vertices, in other words, a pair of a source store (the warehouse from which the parcel was moved) and a target store (the warehouse to which the parcel was moved). The row has information such as a parcel number 221, a vehicle number 222, a source store 223, a target store 224, a departure time 225, and an arrival time 226. The parcel number 221 represents the parcel number. The vehicle number 222 represents the number of the vehicle that transported the parcel. The source store 223 represents the number of the warehouse from which the vehicle was moved. The target store 224 represents the number of the warehouse to which the vehicle was moved. The departure time 225 represents the time when the vehicle departed from the warehouse from which the vehicle was moved. The arrival time 226 represents the time when the vehicle arrived at the warehouse to which the vehicle was moved. The warehouse numbers of the source store 223 and the target store 224 correspond to the warehouse number 411 in FIG. 4, which will be described later. For example, the first row of the second package table 202 indicates that package "100001" was transported by vehicle "v300001," and that the vehicle departed from source warehouse "415" at time "2022 / 6 / 7 18:15" and arrived at destination warehouse "435" at time "2022 / 6 / 7 19:50." This makes it possible to track the location (in the vehicle or in the warehouse) of each package at any time. Specifically, for example, the following tracking is possible for package "100001" according to the first row of the first package table and the first to fourth rows of the second package table 202: The route of package "100001" was warehouse (departure) "415" → warehouse "435" → warehouse "422" → warehouse "427" → warehouse (arrival) "425". The departure time and arrival time for each warehouse on the luggage route are as shown in the first to fourth rows of the second luggage table 202. The cargo was transported from warehouse (origin) “415” to warehouse “435” by vehicle “v300001”. The cargo was unloaded from vehicle “v300001” at warehouse “425” and loaded onto vehicle “v400002”. After that, the cargo was transported from warehouse "435" to warehouse (destination) "425" by vehicle "v400002".

[0037] FIG. 3 is a diagram showing the structure of the vehicle data 102.

[0038] Vehicle data 102 indicates various information about vehicles moving through a logistics network. Specifically, for example, vehicle data 102 is made up of a first vehicle table 301 and a second vehicle table 302. For convenience, in this embodiment, the means of transportation is a vehicle, but other means of transportation such as a ferry or a railway may also be used in the present invention.

[0039] The first vehicle table 301 has information about vehicle specifications. The first vehicle table 301 has a row for each vehicle. The row has information such as a vehicle number 311 and a maximum load capacity 312. The vehicle number 311 represents a vehicle number for identifying the vehicle. The maximum load capacity 312 represents the maximum load capacity, which is the upper limit of the amount of luggage that can be loaded onto the vehicle.

[0040] The second vehicle table 302 contains information about the warehouses and times at which vehicles depart and arrive. Hereinafter, a "vehicle diagram" refers to a diagram identified from the second vehicle table 302 (or a table of the same format). The second vehicle table 302 contains a row for each vehicle route element. A "vehicle route" is the route taken by a vehicle. A "vehicle route element" is an element of a vehicle route; specifically, a pair of an edge and its two vertices, in other words, a pair of a source store (the warehouse from which the vehicle is moved) and a target store (the warehouse to which the vehicle is moved). A row contains information such as a vehicle number 321, a source store 322, a target store 323, a departure time 324, and an arrival time 325. The vehicle number 321 represents the vehicle number. The source store 322 represents the number of the source store (the warehouse from which the vehicle is moved) that the vehicle traveled. The target store 323 represents the number of the target store (the warehouse to which the vehicle is moved) that the vehicle traveled. The departure time 324 represents the time when the vehicle departed from the source store. The arrival time 325 indicates the time when the vehicle arrived at the Target store. The interpretation of the second vehicle table 302 is similar to the second luggage table 202. From each row, the warehouse to which the vehicle traveled and the departure and arrival times can be read, making it possible to track the location of the vehicle at any time.

[0041] FIG. 4 is a diagram showing the structure of the warehouse data 103.

[0042] The warehouse data 103 consists of a warehouse table 401. The warehouse table 401 has a row for each warehouse. The row has information such as a warehouse number 411, latitude 412, longitude 413, area 414, maximum daily work volume 415, average daily work volume 416, and a pair of x-hour loading volume 417α and x-hour unloading volume 418α. "x" is an integer between 0 and 23. α is an alphabet such as A, B, ...

[0043] The warehouse number 411 represents the warehouse number that identifies the warehouse. The latitude 412 and longitude 413 represent the latitude and longitude as coordinates on a map of the warehouse. The region 414 represents the name of the region to which the warehouse belongs. The maximum daily workload 415 represents the maximum daily workload, which is the workload on the day with the highest workload within a specified period. The average daily workload 416 represents the average daily workload, which is the average workload for each day within a specified period. The loading volume at x hour 417α represents the loading volume at x hour (for example, the average loading volume for a specified period at x hour). The unloading volume at x hour 418α represents the unloading volume at x hour (for example, the average unloading volume for a specified period at x hour). The workload at x hour may be, for example, the sum of the unloading volume at x hour 418α and the loading volume at x hour 417α. The "unloading volume" is the volume of cargo unloaded. The "loading volume" is the volume of cargo loaded. The "amount of work" may be a value based on at least one of the amount of unloading and the amount of loading, for example, the sum of the amount of unloading and the amount of loading, as described above.

[0044] FIG. 5 is a diagram showing the structure of the graph data 104.

[0045] The graph data 104 is data that represents the logistics network as a graph network and is used when the logistics network optimization unit 130 derives a package route, and holds information about vertices and edges. The graph data 104 is made up of a first graph table 501 and a second graph table 502.

[0046] The first graph table 501 has information about vertices. The first graph table 501 has a row for each vertex. The row has information such as a vertex ID 511, a capacity limit 512, a cluster number 513, an inside-inside coefficient 514, an inside-outside coefficient 515, an outside-inside coefficient 516, and an outside-outside coefficient 517.

[0047] The second graph table 502 has information about edges. The second graph table 502 has a row for each edge. The row has information such as an edge ID 521, a source vertex 522, a target vertex 523, a capacity 524, and a cost 525. A "source vertex" may be called a parent vertex as one end of an edge. A "target vertex" may be called a child vertex as the other end of an edge.

[0048] The details of this data 104 will be explained in the explanation of process 1203 in FIG.

[0049] Hereinafter, the operation of the logistics network planning system 100 will be described, with reference to the process of deriving cargo routes in the logistics network (cargo routes for distributing the load on warehouses) using the data shown in FIGS. 2 to 5 described above.

[0050] FIG. 6 is a flowchart showing the processing performed by the logistics network planning system 100.

[0051] The input / output unit 140 executes the process 601. Specifically, the input / output unit 140 reads the package data 101, vehicle data 102, and warehouse data 103 required for the calculation from, for example, a persistent storage device into a memory.

[0052] The logistics network analysis unit 120 (warehouse role estimation unit 121) executes process 602 using the warehouse data 103. Specifically, the logistics network analysis unit 120 classifies warehouses with similar workload trends by clustering, with the workload per hour of existing warehouses as the main feature. Details of this process will be described with reference to FIG. 7.

[0053] The logistics network analysis unit 120 (similar warehouse classification unit 122) executes process 603 using the package data 101 and warehouse data 103. Specifically, the logistics network analysis unit 120 estimates the role of each warehouse in transporting packages within and outside the region. Details of this process will be described with reference to FIG. 11. Note that either or both of process 602 and process 603 may be executed.

[0054] The logistics network optimization unit 130 (luggage route optimization unit 131) executes process 604 using the luggage data 101, vehicle data 102, and warehouse data 103. Specifically, the luggage route optimization unit 131 derives luggage routes that distribute the loads on each warehouse. Details of this process will be described with reference to FIG. 12.

[0055] Finally, the input / output unit 140 executes process 605. Specifically, the input / output unit 140 displays on a screen the logistics network after the change based on the derived package route, the similar warehouse classification results obtained by calculation, the warehouse role estimation results, and the like.

[0056] First, the details of process 602 will be described. In process 602, the logistics network analysis unit 120 performs clustering for multiple warehouses using the workload (the sum of the unloading and loading amounts) at each time and the peak times and peak workloads in a day as feature quantities, and classifies warehouses with similar workload trends. Process 602 follows the flowchart shown in Figure 7. Note that process 602 does not have to be executed. If process 602 is not executed, processes 1304 and 1305 in the flowchart of Figure 13, which will be described later, do not have to be executed.

[0057] In step 701, the logistics network analysis unit 120 copies the warehouse data 103 already read in step 601 into memory and stores it as temporary data for calculation. Then, in step 702, the logistics network analysis unit 120 divides the warehouse data 103 by region. Specifically, a set of rows in the warehouse table 401 with the same value for region 414 is obtained. For example, if the warehouse table 401 has "Kanto" and "Kinki" as regions 414, the warehouse table 401 is divided into a regional warehouse table 401_1 consisting only of rows with region 414 "Kanto," and a regional warehouse table 401_2 consisting only of rows with region 414 "Kinki" (illustration omitted). Furthermore, for the regional warehouse tables obtained for each region (i.e., for each regional warehouse table), the logistics network analysis unit 120 executes steps 703 to 705.

[0058] In process 703, the logistics network analysis unit 120 selects the divided regional warehouse table (for example, the aforementioned regional warehouse table 401_1) to be used for calculation. Next, in process 704, the logistics network analysis unit 120 constructs clustering data for each warehouse. To avoid complexity, hereinafter, the regional warehouse table will be simply referred to as warehouse data 103, assuming that there is only a single region.

[0059] FIG. 8 is a schematic diagram of the construction of clustering data.

[0060] The logistics network analysis unit 120 creates a first table 801 by converting the workload at each time held in the warehouse data 103 into a time-series format for each warehouse. In this embodiment, the workload is the sum of the unloading and loading workloads, but it may be either the unloading or loading workload alone. The logistics network analysis unit 120 creates a second table 802 by converting the first table 801 into an hourly average value from midnight to 11:00 pm for each warehouse over a predetermined period (e.g., one week). For example, the first row of the second table 802 represents the average workload value at midnight for the predetermined period.

[0061] Additionally, the logistics network analysis unit 120 extracts, as feature quantities for clustering unloading and loading amounts, the time of day with the highest workload (peak time) and the workload at that peak time (peak workload) from the second table 802. From the data extracted in this way, the logistics network analysis unit 120 creates a third table 803 for each warehouse, with the total workload at each time, the peak time, and the peak workload values ​​as elements.

[0062] 7, the logistics network analysis unit 120 constructs a two-dimensional matrix using the values ​​of each row of this table 803 as elements, and performs clustering to classify the warehouses. As the clustering method, for example, an existing method such as the K-means method may be applied. Furthermore, the number of clusters when applying the K-means method may be stored in advance in a memory area of ​​the logistics network planning system 100 and read from the memory area, or may be provided from the user terminal 30.

[0063] Finally, in process 706, the logistics network analysis unit 120 assigns cluster numbers obtained as a result of clustering to each warehouse. Specifically, for example, the logistics network analysis unit 120 adds a column called "cluster number" to the warehouse data 103 and stores the corresponding numbers. The numbers used here are unique numbers so as not to overlap with clustering results in other regional warehouse tables.

[0064] Figure 9A is a schematic diagram of the first cluster obtained in process 602. Figure 9B is a schematic diagram of the second cluster obtained in process 602.

[0065] A cluster is a cluster of warehouses with the same cluster number. Regarding the cluster trend graph 901, the horizontal axis corresponds to time, the vertical axis corresponds to the amount of work, and the polygonal line corresponds to the warehouse.

[0066] According to the trend graph 901A of the first cluster and the trend graph 901B of the second cluster, the warehouses constituting the first cluster have peak workloads between 7:00 PM and 8:00 PM, and the warehouses constituting the second cluster have peak workloads between 8:00 PM and 10:00 PM. This allows the user to interpret the trend graphs to obtain information about warehouse combinations for which load balancing is appropriate. For example, the second cluster has a surplus of workloads between 7:00 PM and 10:00 PM. Therefore, the user or the parcel route optimization unit 131 can recognize that it is possible to transfer some of the workload in the first cluster during the 7:00 PM period to the second cluster. For example, the parcel route optimization unit 131 may preliminarily eliminate warehouse combinations that are ineffective for load balancing (e.g., warehouse combinations with peak workloads at the same time). The input / output unit 140 may also display these trend graphs 901A and 901B on the user terminal 30.

[0067] As shown in Fig. 6, in parallel with (following) process 602, the logistics network analysis unit 120 estimates the role of each warehouse in the existing logistics network in process 603. Note that process 603 does not have to be executed. In that case, process 1205 in the flowchart of Fig. 12, which will be described later, does not have to be executed.

[0068] As mentioned above, FIG. 10 is a schematic diagram showing an example of a logistics network.

[0069] This diagram shows the roles of warehouses. Specifically, the warehouses that make up the logistics network consist of warehouses (called GWs (gateways) in this example) that have the role of appropriately sorting packages for their destination within and outside the region, and warehouses that are the ends of the logistics network. In other words, a warehouse that corresponds to a vertex that has intermediate vertices (a source vertex and a target vertex) is a GW, and a warehouse that corresponds to either a root vertex (a vertex that does not have a source vertex) or a leaf vertex (a vertex that does not have a target vertex) is an end warehouse.

[0070] Take region 4 as an example. Warehouse 1001-4A is a warehouse that has the role of sorting luggage transported from regions outside region 4 into region 4 (referred to as an external / internal gateway). There are also warehouses like warehouse 1001-4B that have the role of sorting luggage transported from within region 4 into region 4 (referred to as an internal / internal gateway), and warehouse 1001-4F that has the role of sorting luggage transported from within region 4 into regions outside region 4 (referred to as an internal / external gateway). There may also be warehouses that have the role of sorting luggage transported from regions outside region 4 into other regions outside region 4 (referred to as an external / external gateway). Each warehouse may not only serve as a single gateway, but may also have multiple roles, such as being an external / internal gateway during certain time periods and an internal / internal gateway during other time periods.

[0071] FIG. 11 is a flowchart of the process 603.

[0072] In process 603, the logistics network analysis unit 120 estimates the role of each warehouse whose operation (role) has become unclear over the years.

[0073] Specifically, in step 1101, the logistics network analysis unit 120 copies the parcel data 101 and warehouse data 103 already read in step 601 into memory and stores them as temporary data for calculation. Next, the logistics network analysis unit 120 references the second parcel table 202 related to the parcel route in the parcel data 101, and performs steps 1102 to 1106 for each row. The row number of the referenced row is set to i (for program convenience, i is set to 0 or greater and equal to the total number of rows in the parcel route minus 1). In the following explanation of FIG. 11, if the referenced information is information on the xth row, x may be added to the end of the information for ease of explanation.

[0074] In process 1102, the logistics network analysis unit 120 refers to the columns of the parcel number 221 and the vehicle number 222 in the i-th and i+1-th rows of the second parcel table 202. Here, if the parcel numbers 221i and 221(i+1) are the same and the vehicle numbers 222i and 222(i+1) are different, the logistics network analysis unit 120 determines that a change of vehicles has occurred (process 1102: YES).

[0075] If the determination result in process 1102 is YES, the logistics network analysis unit 120 further determines whether the target store and source store are in the same region for the i-th row in process 1103. Specifically, for example, the logistics network analysis unit 120 refers to the warehouse data 103 and determines whether the region 414 corresponding to the warehouse number 411 that matches the target store 224i is the same as the region 414 corresponding to the warehouse number 411 that matches the source store 223i.

[0076] If the target store i and the source store i are in the same region (process 1103: YES), that is, if the parcel route element corresponding to the i-th row is a branch line, the logistics network analysis unit 120 determines in process 1104A whether the region 414 corresponding to the target store 224i+1 is the same as the region 414 corresponding to the source store 223(i+1).

[0077] If target store i (source store i+1) and target store i+1 are in the same region (process 1104A: YES), that is, if the parcel route element corresponding to row i+1 is also a branch line, the logistics network analysis unit 120 assigns the label "inside" to row i in process 1105A. This indicates that the warehouse corresponding to the target store 224 in row i plays the role of an inside-out gateway (see reference numeral 1001-4B in Figure 10) that sorts parcels arriving from within the region into vehicles heading to warehouses within the same region. Note that the label can be assigned by writing a value as a label in the "Label" column added to the second parcel table 202.

[0078] If target store i (source store i+1) and target store i+1 are in different areas (process 1104A: NO), that is, if the parcel route element corresponding to row i+1 is a trunk line, the logistics network analysis unit 120 assigns the label "inside / outside" to row i in process 1105B.

[0079] If the target store i and the source store i are in different regions (step 1103: NO), that is, if the parcel route element corresponding to the i-th row is a trunk line, the logistics network analysis unit 120 determines in step 1104B whether the region 414 corresponding to the target store 224i+1 is the same as the region 414 corresponding to the source store 223(i+1).

[0080] If target store i (source store i+1) and target store i+1 are in the same area (process 1104B: YES), that is, if the parcel route element corresponding to row i+1 is a branch line, the logistics network analysis unit 120 assigns the label "outside inside" to row i in process 1105C.

[0081] If target store i (source store i+1) and target store i+1 are in different areas (process 1104B: NO), that is, if the parcel route element corresponding to row i+1 is also a trunk line, the logistics network analysis unit 120 assigns the label "outside outside" to row i in process 1105D.

[0082] If the judgment result of process 1102 is NO, that is, if the parcel numbers 221i and 221(i+1) are different (i.e., the parcel corresponding to row i has arrived at the destination store), or if the vehicle numbers 222i and 222(i+1) are the same, the target store i does not play the role of GW, and the logistics network analysis unit 120 assigns the label "other", which means a role other than GW, to the i-th row in process 1106.

[0083] Finally, in step 1107, the logistics network analysis unit 120 calculates the proportion of the number of labels for each warehouse (illustration omitted). This process is performed, for example, by the following procedure. First, the logistics network analysis unit 120 adds the label columns "internal internal," "internal external," "external internal," "external external," and "other" to the warehouse table 401. Next, for each row in the second parcel table 202, the logistics network analysis unit 120 references the target store 224 and, for the warehouse with a warehouse number that matches the target store 224, increments the number of labels in the label column corresponding to the label assigned to that row among the five added label columns. Thereafter, the logistics network analysis unit 120 calculates the total number of labels for the five labels for each warehouse. Furthermore, the logistics network analysis unit 120 calculates "label proportion = number of labels ÷ total number of five labels" for each of the five labels and stores the label proportion corresponding to that label column in each of the five label columns in the warehouse table 401. Through the above process, the label ratio (ratio of the number of labels) is calculated for each label for each warehouse.

[0084] The label ratios obtained by the logistics network analysis unit 120 executing process 603 can be confirmed, for example, in a graph of "Role Breakdown" in the second screen element 1502 of the logistics network screen 1500 shown in FIG. 15. In this graph, the horizontal axis corresponds to the warehouse number, and the vertical axis corresponds to the warehouse label ratio, and a stacked bar graph is displayed. For warehouse 5, the role as "internal internal gateway" is 15%, the role as "internal external gateway" is 60%, and "other" is 25%.

[0085] Returning to the explanation of the flowchart in FIG.

[0086] After the above-described steps 602 and 603, the logistics network optimization unit 130 executes step 604, i.e., derives a package route. Details of step 604 will be explained using the flowchart shown in Fig. 12. In this flowchart, a process is executed to derive a package route that distributes the load at warehouses in the logistics network using optimization technology.

[0087] First, in process 1201, the logistics network optimization unit 130 sets the calculation conditions to be used for planning. The calculation conditions include the "start date and time" and "end date and time" of the logistics network to be planned. The "start date and time" and "end date and time" may be input from a screen by a user of the system when starting the logistics network planning system 100, or may be stored in advance as a file in the storage area of ​​the logistics network planning system 100 and set by the logistics network optimization unit 130 by reading the file. In addition, information on the cost of moving between each warehouse is also set. Similarly, the cost may be input from a screen by the user or set by reading it from a file saved in the storage area. For example, a table may be set that represents the correspondence between pairs of warehouse numbers of two warehouses and real numerical values ​​as costs.

[0088] Next, the logistics network optimization unit 130 reads the data to be used for planning in process 1202. Specifically, the logistics network optimization unit 130 copies the cargo data 101, vehicle data 102, and warehouse data 103 that have been read in process 601 onto memory and uses them as temporary data for calculation.

[0089] Furthermore, in process 1202, the logistics network optimization unit 130 selects from the parcel data 101 parcels that have already occurred as of the start date and time or that will occur between the start date and time and the end date and time, and sets these as target parcel data 101T (first parcel table 201T and second parcel table 202T). Whether a parcel has already occurred as of the start date and time may be determined by whether the occurrence time 215 of a parcel in the first parcel table 201 is before the start date and time, and whether the last arrival time 226 of the parcel listed in the second parcel table 202 (i.e., the time when the parcel arrives at the destination store and is no longer subject to transportation in the logistics network) is after the start date and time. Whether a parcel is occurring between the start date and time and the end date and time may be determined by whether the occurrence time 215 of a parcel in the first parcel table 201 is after the start date and time and before the end date and time.

[0090] Using a similar procedure, in process 1202, the logistics network optimization unit 130 also selects vehicles from the vehicle data 102 that have already occurred at the start date and time (the vehicle has already departed from the first departure store and has not yet arrived at the last arrival store) or that will occur between the start date and time and the end date and time, and sets these as the target vehicle data 102T (first vehicle table 301T and second vehicle table 302T).

[0091] The warehouse data 103 is used as is. That is, for the warehouse data 103, the processing performed in step 1202 regarding the package data 101 and vehicle data 102 is not necessary.

[0092] Next, in process 1203, the logistics network optimization unit 130 constructs logistics network data to be used in subsequent processes. The logistics network data is graph data 104 consisting of a first graph table 501 and a second graph table 502 shown in FIG. 5. The graph data 104 is stored in the storage device 52.

[0093] In process 1203, first, the logistics network optimization unit 130 constructs a first graph table 501 corresponding to the vertex. Specifically, the logistics network optimization unit 130 acquires a warehouse number 411 from the warehouse table 401 of the warehouse data 103 and sets a vertex ID 511 in the first graph table 501 based on the warehouse number 411. The logistics network optimization unit 130 acquires a maximum daily workload 415 for each warehouse from the warehouse table 401 and sets a capacity upper limit 512 representing the upper limit of capacity for the vertex ID 511 corresponding to the warehouse based on the maximum daily workload 415. The capacity upper limit 512 may be modified to a larger value by multiplying the maximum daily workload 415 by an appropriate coefficient, rather than simply using the maximum daily workload 415. Furthermore, the logistics network optimization unit 130 sets the cluster number of the warehouse calculated in process 602 as the cluster number 513 in the first graph table 501. Finally, the logistics network optimization unit 130 sets an inside-inside coefficient 514, an inside-outside coefficient 515, an outside-inside coefficient 516, and an outside-outside coefficient 517 from the label proportions calculated for each label (warehouse role) in process 603. Each of the coefficients 514 to 517 is based on the label proportion corresponding to the label coefficient. Specifically, for example, for a vertex corresponding to a certain warehouse, the inside-inside coefficient 514 is obtained by subtracting the label proportion (a real number between 0.0 and 1.0) of the label "inside-inside" of the warehouse from 1. More specifically, for example, for the warehouse with vertex ID 511 "d0" shown in FIG. 5, if the label proportion of "inside-inside" is 0.50, the inside-inside coefficient 514 can be calculated as "1 - 0.50 = 0.50." The same applies to other labels. In the embodiment, for simplicity, a value subtracted from 1 is set, but other calculation methods may be adopted as long as the label coefficient is set so that the cost decreases as the label ratio (or frequency) increases (in other words, the label coefficient is set so that the cost increases as the label ratio (or frequency) decreases).

[0094] In process 1203, the logistics network optimization unit 130 calculates the edges and edge attributes connecting each vertex constructed in the first graph table 501 to construct a second graph table 502. The logistics network optimization unit 130 determines whether to draw an edge by referencing the second vehicle table 302T of the selected vehicle data 102T and determining whether a vehicle exists between any two vertices. If a vehicle exists, the logistics network optimization unit 130 sets an edge ID 521 (for example, assigning consecutive numbers starting from "e00001") and sets a source vertex 522 and a target vertex 523 representing the vertex IDs corresponding to the warehouses (source store and target store) to which the vehicle moved for the source vertex (starting point) and target vertex (ending point) of the edge. Next, the logistics network optimization unit 130 calculates the capacity as an attribute of the edge and sets the calculated capacity 524. For each edge, the logistics network optimization unit 130 extracts from the second vehicle table 302T one or more vehicles that travel between the source store and the target store corresponding to the source vertex and the target vertex of that edge, and the capacity 524 may be the sum of the maximum load capacities 312 of the one or more vehicles. Finally, the logistics network optimization unit 130 sets the cost of traveling along the edge based on the calculation conditions set in the above-mentioned process 1201. The "cost" here is set, for example, based on the fuel cost and labor cost required for traveling between warehouses. Although the term "cost" is used for convenience, the cost is merely an example of cost, and the cost may be set based on any indicator such as the travel distance between warehouses, travel time, and environmental impact (e.g., CO2 emissions) instead of or in addition to the cost.

[0095] Next, the logistics network optimization unit 130 generates path candidates that will be one of the inputs for optimization in process 1204. Details of process 1204 will be explained using FIG.

[0096] First, the logistics network optimization unit 130 reads data in step 1301. The data read here is the first graph table 501 and the second graph table 502 that make up the graph data 104 of the logistics network.

[0097] Next, the logistics network optimization unit 130 sets search conditions for generating path candidates in process 1303. The search conditions may be input by the user via a screen, or may be specified from a pre-stored external file, etc. The search conditions set here may be the "path candidate depth limit" (e.g., an integer value) and the "path candidate cost limit" used in process 1306, which will be described later.

[0098] After setting the search conditions, the logistics network optimization unit 130 searches for path candidates by executing steps 1303 to 1307 for each package in the package data 101T. For convenience of explanation, the package currently being selected is assumed to be package k.

[0099] In the search, first, in step 1303, the logistics network optimization unit 130 extracts the actual route of package k. This is performed by extracting all rows with package number 221 for package k in the second package table 202T of package data 101T, and then extracting the source store 223 where the vehicle number 222 has changed (i.e., the package has changed vehicles) and the target store 224 for the final move. For example, using the second package table 202 of FIG. 2, the warehouses visited by package "100001" are "415 → 435 → 422 → 427 → 425." Of these, if the source store 223 where the vehicle number 222 has changed and the target store 224 for the final move are extracted, the result is "415 → 435 → 425." Based on the visit order at this time, the initial path p_k_init is set to (415, 435, 425).

[0100] Next, in process 1304, the logistics network optimization unit 130 extracts the warehouse with the highest daily workload in the initial path. This process is performed by referencing the warehouse table 401 in the warehouse data 103 for each warehouse excluding the first and last warehouses in the initial path, and extracting the warehouse with the highest "average daily workload 416 ÷ maximum daily workload 415." The warehouse obtained here is designated as the maximum-load warehouse d_k_max.

[0101] Furthermore, in process 1305, the logistics network optimization unit 130 acquires warehouses similar to the maximum-load warehouse d_k_max. This is acquired by referencing the cluster numbers assigned after the logistics network analysis unit 120 executes the similar warehouse classification unit 122, and extracting the warehouse numbers of warehouses that have the same cluster number as the maximum-load warehouse d_k_max. The acquired warehouse numbers are stored in memory as a set with an element count of 0 or more.

[0102] Based on the conditions obtained by the above processing, in processing 1306, the logistics network optimization unit 130 lists path candidates from the departure store 212 to the arrival store 213 of the package k. The list of path candidates can be realized by using an existing graph search algorithm such as depth-first search or breadth-first search. Alternatively, all paths from the departure store 212 to the arrival store 213 of the package k may be listed. At this time, the logistics network optimization unit 130 excludes path candidates that satisfy any of the following three conditions from the final path candidates. (Condition 1) The depth of the path candidates exceeds the “upper limit of the depth of the path candidates” held by the logistics network optimization unit 130. (Condition 2) The sum of the costs of the edges that make up the path candidate exceeds the "upper cost limit of the path candidate." (Condition 3) During the search, the path candidate passes through a warehouse similar to the maximum loaded warehouse d_k_max.

[0103] Finally, in process 1307, the logistics network optimization unit 130 stores data in memory that represents path candidates (a set of one or more paths) corresponding to the package k. This data may be, for example, a table in which the package number of the package k is the key and the path candidates are the values.

[0104] Returning to the explanation of Figure 12, after generating the path candidates, the logistics network optimization unit 130 then, in process 1205, sets a coefficient according to the warehouse role (label ratio) for the edges that make up each path candidate. When an edge is drawn into a warehouse, the warehouse role in that path changes depending on whether it is drawn from a warehouse outside the region or from within the region. In process 1206 (optimization process) described below, this process 1205 is executed to set costs according to the warehouse role in the existing logistics network.

[0105] For convenience, the logistics network optimization unit 130 selects one path candidate for a certain package k and sets p_k = (d1, d2, d3), where d1, d2, and d3 are warehouse numbers, respectively, and the edges that make up the path candidate are (d1, d2) and (d2, d3).

[0106] The logistics network optimization unit 130 first determines the movement type of cargo moving along each edge of each path candidate. Here, the "movement type" refers to either a branch line (where the two warehouses corresponding to the source and target vertices of the edge are in the same region) or a trunk line (where the two warehouses corresponding to the source and target vertices of the edge are in different regions). Furthermore, the logistics network optimization unit 130 acquires a combination of the movement type of the inbound edge and the movement type of the outbound edge for each warehouse listed in the path candidate. For example, in the path candidate p_k, if d1 and d2 are in the same region and d2 and d3 are in different regions, then edge (d1, d2) is a branch line and edge (d2, d3) is a trunk line. Considering d2 as the base point, in path candidate p_k, d2 sorts cargo transported from within the region to outside the region, i.e., acts as an inbound / outbound gateway. In this way, the logistics network optimization unit 130 determines the role of the warehouses that make up the path candidate, and identifies coefficients according to the actual warehouse roles from the first graph table 501, thereby making it possible to set a coefficient (γ^{path candidate}_{start warehouse number, end warehouse number}) in advance for each edge that makes up the path. Explaining the above example, edge (d1, d2) is an edge that passes through the internal and external gateways in this path candidate p_k, so γ^{p_k}_{d1, d2}=0.70 (equal to the internal and external coefficient of vertex d2).

[0107] Furthermore, as a constant to be used in subsequent optimization problems, the logistics network optimization unit 130 generates a constant δ^{p}_{i,j} indicating whether each edge is included in the path candidate. This constant is 1 when edge (i,j) is included in path candidate p, and 0 when it is not. Since edges are checked one by one in the process of determining the role coefficients described above, they can be calculated simultaneously.

[0108] In this way, the logistics network optimization unit 130 sets the coefficients and constants of each edge that makes up the path candidate for all path candidates, and then in process 1206, constructs an optimization problem and solves the optimization problem to derive the optimal cargo route.

[0109] The process 1206 will be described with reference to FIG.

[0110] First, the logistics network optimization unit 130 reads a set of data to be used in the optimization process in process 1401. The data to be read includes the first graph table 501 and the second graph table 502 of the logistics network data, data of all the path candidates generated in process 1204, and coefficients γ^{p_k}_{d i ,d j} and constants δ^{p}_{i,j}.

[0111] When constructing the optimization problem, the logistics network optimization unit 130 generates decision variables in process 1402. There are two types of decision variables: those related to path candidates and those that reduce the load on the warehouse.

[0112] First, we will discuss the decision variables related to path candidates. Regarding the path candidates generated in the above-mentioned process 1204, the set of path candidates related to package k is defined as P^{k}={p_k|path candidates for package k}. Each path candidate is a route connecting the package's departure store to its arrival store. To derive a route for a package to reach its arrival store, it is necessary to select at least one path candidate from the set of path candidates P^{k}. Therefore, the logistics network optimization unit 130 sets the decision variable for selecting a path candidate as shown in equation (1).

number

[0113] When the value of z^{k}_{p} is 1, it means that candidate path p is used to transport package k, where K indicates all packages held by package data 101T.

[0114] Furthermore, in order to distribute the workload among the warehouses, the logistics network optimization unit 130 sets equation (2) as a decision variable for suppressing the inflow amount to each warehouse.

number

[0115] where R^{+} is the set of positive real values.

[0116] Next, the logistics network optimization unit 130 sets constraints in steps 1403 to 1405. First, in step 1403, the logistics network optimization unit 130 sets constraints related to paths.

[0117] As mentioned above, at least one path candidate must be selected for the package to reach the destination from the departure store. Furthermore, to determine a unique solution, exactly one path candidate must be selected, not multiple paths. Such a constraint is, for example, as shown in equation (3).

number

[0118] Next, in process 1404, the logistics network optimization unit 130 sets constraints on the edges. A capacity obtained from the performance of the vehicle diagram is set for each edge (i, j). This means an upper limit on the weight of cargo that can be transported for edge (i, j) during a specified planning period. This is not necessarily set for a single vehicle, but may be set by adding together the maximum load capacities of multiple vehicles. To ensure that the planned amount of cargo does not exceed this upper limit, the logistics network optimization unit 130 sets the constraint of equation (4) as follows:

number

[0119] Here, w_{k} is the weight of baggage k, E is the set of all edges represented by the second graph table 502, and Cap^{e}_{ij} is the capacity of edge (i, j).

[0120] Next, in process 1405, the logistics network optimization unit 130 sets a constraint on the vertices. Specifically, this constraint is a constraint that the amount of cargo flowing into each warehouse must not exceed the upper capacity limit. This can be achieved by setting a constraint that, for edges flowing into a warehouse, when the warehouse is selected as a path for the cargo route, the sum of the weight of the cargo flowing through the warehouse must be equal to or less than the upper capacity limit. Specifically, the logistics network optimization unit 130 adds the constraint of equation (5).

number

[0121] Here, D is the set of all warehouses, and Cap^{d}_{j} is the upper capacity limit of a certain warehouse j.

[0122] In addition, from the viewpoint of load distribution, the logistics network optimization unit 130 uses the decision variable L to add equation (6) as a constraint for suppressing the load on the warehouse with the highest load.

number

[0123] Note that the left-hand sides of equations (5) and (6) are identical, and only the right-hand sides differ. For a given warehouse j, it is unclear which of L or Cap_{d}_{j} will be smaller until L is determined in the optimization process, so both constraints are required.

[0124] Finally, in process 1406, the logistics network optimization unit 130 sets an objective function. First, from the perspective of load balancing, the logistics network optimization unit 130 adds L, which is the right-hand side of the constraint that suppresses the load on each warehouse, as the objective function. Furthermore, since the logistics network optimization unit 130 is required to transport goods at the minimum cost while balancing the load, it adds the total value of the cost for goods to pass through each edge. At this time, if an edge included in the selected path candidate enters a warehouse for a purpose different from the existing warehouse role, it will be costly because it disrupts existing operations. However, if the edge enters the warehouse for the same purpose as the existing role, the logistics network optimization unit 130 multiplies the coefficient γ above by the objective function to reduce costs. Minimizing the above two items can be expressed as a formula, as shown in Equation (7).

number

[0125] Here, c_{i,j} is the cost required to move along edge (i,j). Furthermore, f_1 and f_2 are weighting parameters that determine the importance of each term, and may be entered by the user on the screen or may be specified in advance from the storage area of ​​the logistics network planning system 100.

[0126] The logistics network optimization unit 130 solves the constructed optimization problem in process 1407. Any existing solution method for integer programming, such as the branch and bound method, may be used as the solution method. A commercially available solver for integer programming may also be used.

[0127] The resulting decision variable z^{k}_{p} is used as the return value, and processing 1206 ends.

[0128] Returning to the explanation of Figure 12, in process 1207, the logistics network optimization unit 130 uses the decision variables obtained in the optimization process to output logistics network plan data representing the package route, i.e., the proposed logistics network plan. The logistics network plan data may include, for example, a table in which the package number of each package is used as a key and a list of warehouses where the package is sorted (i.e., warehouses where vehicles will be transferred) is used as a value. Once the package route from the departure store to the arrival store is determined, the vehicle to transport each package can be determined by a method for solving an existing delivery planning problem, given a known vehicle diagram. The logistics network optimization unit 130 saves the final solution in the same format as the second package table 202 in Figure 2, and then terminates the process. Data containing data in this same format may be the logistics network plan data.

[0129] An example of a logistics network screen 1500 showing a planned logistics network will be described with reference to FIG.

[0130] The input / output unit 140 displays a logistics network screen on the user terminal 30. The logistics network screen 1500 includes, for example, a first screen element 1501 that displays the package route output in process 1207, a second screen element 1502 that displays information obtained by processes 602 and 603, and a third screen element 1503 that displays information about packages moving on the edges that make up the logistics network.

[0131] For example, the input / output unit 140 displays the logistics network in the first screen element 1501 by drawing lines between warehouses that form the logistics network and between warehouses when there is cargo moving between warehouses. At this time, the display position of the warehouse is determined based on coordinate information such as the latitude 412 and longitude 413 contained in the warehouse data 103. The mark of the warehouse (vertex) may be a mark in a display mode corresponding to the warehouse role calculated by process 603 (for example, a square if the role is an internal / external gateway). The display mode of the mark may be a shape, color, etc.

[0132] A second screen element 1502 displays information about the classification and role of each warehouse calculated by processes 602 and 603. This visualizes the analysis results of the characteristics of each warehouse in the existing logistics network.

[0133] The third screen element 1503 displays the cargo moving on each side and the role of the warehouse on the arrival side at that time, based on the cargo route obtained from process 1206 (optimization process). Data to be used for planning may be selected from the logistics network screen 1500, and the processing of the logistics network planning system 100 may be executed again by manual operation of the user. As a result, a new plan is made by the logistics network planning system 100, and the analysis results of the logistics network and warehouses based on the calculation results are redisplayed on the screen.

[0134] Furthermore, a user may input today's parcel data representing parcels that need to be transported that day into screen 1500 shown in FIG. 15, and a transportation planning unit (not shown) may derive the parcel route for the parcel for that day based on the data of the planned logistics network plan and the input today's parcel data, and create a transportation plan. In addition to the today's parcel data, a transportation plan may be created based on today's vehicle data (data containing information about each of multiple vehicles that can operate that day). This plan can also be used for daily operations by issuing transportation instructions to each vehicle and warehouse in the logistics network. Note that the "transportation planning unit" may be realized by processor 53 or may be realized in transportation planning system 40. The configuration of today's parcel data may be the same as that of first parcel table 201 for parcels that need to be transported that day (for example, the occurrence time 215 is the same day). [Example]

[0135] In the first embodiment, a method for load distribution was described from the viewpoint of distributing the load of warehouses in a logistics network. In the present embodiment, a method for concentrating the load on a specific warehouse will be described, which is the opposite of the first embodiment. Note that in the second embodiment, differences from the first embodiment will be mainly described, and explanations of commonalities with the first embodiment will be omitted or simplified.

[0136] With this method, an increase in the number of packages passing through a specific warehouse is expected to improve the loading rate of vehicles passing through that warehouse, leading to improved efficiency. In addition, since some vehicles will no longer carry packages, eliminating the relevant vehicle diagrams is expected to reduce vehicle maintenance costs.

[0137] However, the components in the second embodiment are the same as those in Fig. 1, and the logistics network planning system 100 executes the same processes as those in Figs. 1 to 15.

[0138] The difference from the first embodiment is the setting of vertex constraints in process 1405 in Fig. 14, which are executed by the logistics network optimization unit 130, and the setting of an objective function in process 1406. Other explanations will be omitted.

[0139] Assume that the process has been executed up to process 1404. In process 1405, the logistics network optimization unit 130 sets a constraint on the vertices. Specifically, this constraint is a constraint on whether the amount of cargo flowing into each warehouse does not exceed the upper limit value of the upper capacity limit 512. This can be realized by setting a constraint on the edges flowing into a warehouse such that when the warehouse is selected as a path for the cargo route, the sum of the weight of the cargo flowing through the warehouse is equal to or less than the upper capacity limit. Specifically, the logistics network optimization unit 130 adds the constraint of equation (8).

number

[0140] The difference from the first embodiment is that the constraint equation that limits the right-hand side by L (equation (6) in the first embodiment) is not added, and equations (5) and (8) are the same.

[0141] Furthermore, in step 1406, the logistics network optimization unit 130 sets an objective function.

[0142] First, the logistics network optimization unit 130 adds an equation to the objective function that maximizes the difference in workload between two warehouses in each region from the perspective of load concentration. This equation maximizes the difference between the amount of cargo flowing into warehouse j and the amount of cargo flowing into warehouse j'. While the overall optimization problem aims to minimize this, the logistics network optimization unit 130 adds an equation representing the difference in cargo volume as a negative value to the objective function to express this. Note that in this case, the targets are limited to warehouses that serve as gateways in order to exclude warehouses that are unlikely to experience load concentration, such as terminal warehouses. Such warehouses may be identified by determining whether the proportion of gateways recorded in the warehouse table 401 exceeds a predetermined value (specified by the user or recorded in advance as a file in the storage area of ​​the logistics network planning system 100).

[0143] Furthermore, since it is required to transport the goods at the minimum cost, the logistics network optimization unit 130 adds up the total cost when the goods pass through each edge. The formula to minimize the above two items is written as Equation (9).

number

[0144] Here, D' is the set of warehouses that are the target of load concentration, and D'' is the set of warehouses that reduce the load. The optimization range can be further limited, not just by simply limiting the proportion of GW roles, but also by limiting the region, etc.

[0145] The subsequent processing is the same as in the first embodiment and will not be described.

[0146] Although several embodiments of the present invention have been described above, these embodiments are presented as examples and are not intended to limit the scope of the invention. The present invention can be embodied in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments are included within the scope and spirit of the invention, as well as within the scope of the inventions described in the claims and their equivalents. For example, in the above embodiments, the "weight" and "volume" of a package are synonymous, but in a broader sense, the "volume" of a package may be determined based on volume, etc., instead of or in addition to weight.

[0147] The above description can be summarized as follows, for example: The following summary may include supplementary explanations to the above description and explanations of modifications of the above embodiment.

[0148] A logistics network planning system (e.g., logistics network planning system 100) includes an input unit (e.g., input / output unit 140), a logistics network analysis unit (e.g., logistics network analysis unit 120), a logistics network optimization unit (e.g., logistics network optimization unit 130), and an output unit (e.g., input / output unit 140). The input unit inputs existing logistics network data, which is data related to an existing logistics network. An "existing logistics network" is composed of multiple luggage routes for multiple luggage. For each luggage, the luggage route is a route consisting of two or more warehouses from the luggage's departure point (first departure warehouse) to its arrival point (last arrival warehouse), i.e., a series of warehouses from the departure point to the arrival point. The existing logistics network data may be an example of past performance data. The existing logistics network data includes, for example, luggage data 101, vehicle data 102, and warehouse data 103. The existing logistics network data may be input from outside the logistics network planning system or may be read from a storage device. The logistics network analysis unit analyzes the existing logistics network data to identify logistics network characteristics (characteristics of the existing logistics network). The logistics network optimization unit constructs an optimization problem based on logistics network characteristics to minimize a target based on one or more indicators, and derives an optimal cargo route by solving the optimization problem. The output unit outputs logistics network plan data, which is data representing a planned logistics network including the derived cargo route. This enables the creation of an optimal logistics network plan with minimal changes to entities such as warehouses and transporters (transportation means). Any indicator can be used as the indicator, such as transportation cost, the volume of cargo to the warehouse with the highest workload, or the negative value of the difference in workload between the warehouse with the highest workload and the warehouse with the lowest workload. For example, the target may be transportation cost, a weighted sum of transportation cost and the volume of cargo to the warehouse with the highest workload, or a weighted sum of transportation cost and the negative value of the difference in workload between the warehouse with the highest workload and the warehouse with the lowest workload.

[0149] The existing logistics network data may include, for each of a plurality of packages, data representing the volume of the package and the transporter for each warehouse along the package's route, as well as data representing the maximum load capacity for each transporter. The logistics network optimization unit may determine, for each of the warehouses, an upper limit on the capacity between the warehouses based on the sum of the maximum load capacities of the transporters that pass through the warehouses, based on the existing logistics network data. The logistics network optimization unit may construct an optimization problem so that one or more packages whose total volume exceeds the upper limit on the capacity between the warehouses do not pass through the warehouses. This allows for the creation of an optimal logistics network plan with minimal changes to the actual situation. An example will be described with reference to FIG. 16.

[0150] When constructing the optimization problem, the logistics network optimization unit 130 may generate path candidates, and the luggage route may be derived by selecting a luggage route from the path candidates. The following (1) to (6) may be assumed. According to the assumptions below, as shown in FIG. 16, the candidates that can be executed in parallel (cases where luggage 1 and luggage 2 can be transported in parallel) are candidates b and c. In addition, in the following explanation, the element represented by "dxx" ("xx" is a number) is a warehouse. (1) The inputs are (1a) to (1c) below. (1a) Data representing a graph with warehouses as vertices and directed edges between warehouses with transportation records in an existing logistics network (for example, at least a part of the graph data 104). (1b) Data including the weight, departure point, and arrival point for each package (for example, at least a part of package data 101). (1c) Data on multiple path candidates for multiple packages (for each package, data on each of one or more path candidates for the package may be a series of warehouses from the store from which the package is to be delivered to the store where the package is to be delivered). (2) The constraint is the upper limit of the sum of the cargo volumes that can flow on the edge (the sum of the loads of one or more vehicles moving between warehouses), and is, for example, the value represented by the capacity 524 for each edge. In this example, the upper limit of the sum of the capacities of each edge is "70". (3) The objective is to minimize transportation costs. The transportation costs may be determined based on at least one of the following indicators: transportation costs between warehouses (e.g., fuel costs), travel distance, travel time, and environmental impact (e.g., CO2 emissions). (4) The decision variable is whether or not a path candidate is selected (for example, a binary value of "0" or "1"). (5) There are two luggage items, luggage 1 and luggage 2, as shown below. Note that for each luggage item, path candidates may be generated with a possible configuration within an upper limit. In this example, the upper limit of the number of path candidates for each luggage item may be 2. Furthermore, the "capacity" of a luggage item may be determined based on volume or the like instead of or in addition to weight. (5a) The capacity of baggage 1 is "50." The path candidates for baggage 1 are d10 → d11 → d13 → d15 (thick solid line) and d10 → d12 → d14 → d15 (thick dashed dot line). (5b) The capacity of baggage 2 is "50." The path candidates for baggage 2 are d11 → d13 → d15 (thick dashed line) and d11 → d14 → d15 (thick chain double-dashed line). (6) The cost function is “1” on all edges.

[0151] The existing logistics network data may include, for each of a plurality of packages, data representing the departure and arrival times for each warehouse that constitutes the package route for the package, the transporters for each warehouse along the package route, and data representing the region of each warehouse. The logistics network analysis unit may include, for example, a warehouse role estimation unit 121. For each warehouse where unloading and loading has been performed, if the transporter that transported the package to the warehouse is different from the transporter that transported the package from the warehouse to its next destination warehouse (i.e., if the package is transferred from one transporter to another), the logistics network analysis unit may assign a warehouse role to the warehouse based on whether the package's transportation to the warehouse was a branch line transport or a trunk line transport, and whether the package's transportation from the warehouse to its next destination warehouse was a branch line transport or a trunk line transport. Branch line transport is transport between warehouses in the same region. Trunk line transport is transport between warehouses in different regions. The logistics network feature may include, for each warehouse, warehouse role features relating to one or more warehouse roles of the warehouse in a certain period. The logistics network optimization unit may formulate an optimization problem such that the smaller the deviation from the warehouse role, the smaller the transportation cost.

[0152] In existing logistics networks that have been built over time, there are operational practices (for example, warehouses that focus on branch line transport cannot receive cargo from main lines), but it is possible to create optimal plans for the logistics network while adhering to these practices (i.e., with minimal changes to the actual warehouse).In addition, over the years, there is a possibility that the original warehouse design may have diverged from the current situation, but even if such a divergence has occurred, the current warehouse role can be estimated and the optimal cargo route can be derived based on the estimated warehouse role, making it possible to create appropriate logistics network plans.

[0153] The "warehouse role characteristics" may be based on the respective proportions or frequencies of one or more warehouse roles for each warehouse over a certain period of time. For example, the warehouse role characteristics may include the proportions or frequencies of estimated warehouse roles for each warehouse, or may include the warehouse role finally estimated from the proportions or frequencies of estimated warehouse roles (e.g., the warehouse role with the highest proportion or frequency). The logistics network optimization unit may construct an optimization problem such that the greater the deviation from the warehouse role, the greater the degree to which transportation costs increase as the proportion of warehouse roles increases. This is expected to result in a more appropriate logistics network plan.

[0154] As an example, for each warehouse role, the larger the value obtained by subtracting the proportion or frequency of that warehouse role (for example, the coefficients 514 to 517 described above) from a predetermined value (for example, "1"), the larger the correction constant becomes, and if the warehouse role of a warehouse on a cargo route matches a warehouse role with a large correction constant, this means that there is a large deviation from the warehouse role, and transportation costs will increase. This is expected to prevent the creation of a logistics network plan that has a large deviation from the warehouse role.

[0155] Specifically, for example, in Figure 17, assume that d21 serves as an external / internal gateway, and d22 does not serve as a gateway. The possible routes from d20 to d22 are d20 → d21 → d22 and d20 → d22. When only costs such as travel distance and fuel costs are taken into consideration, d20 → d22 is more likely to be selected than d20 → d21 → d22. However, if d20 → d22 is selected, d22 will need to be prepared for trunk line transportation. In other words, d22 will need to be changed. Therefore, the optimization problem can be constructed so that costs are higher when a route that requires a change in the role of d22 is selected. When constructing the optimization problem, in addition to a correction constant based on the magnitude of deviation from the warehouse role, it is also better to realize load distribution based on the similarity of the warehouses, as described below.

[0156] The existing logistics network data may include time-series data on the workload for each warehouse in the existing logistics network. For each warehouse, the workload for that warehouse may be based on the unloading volume and loading volume. The logistics network optimization unit may formulate an optimization problem to minimize the amount of cargo sent to the warehouse with the highest workload. This is expected to avoid creating a logistics network plan in which the workload (load) is concentrated in a certain warehouse.

[0157] Specifically, for example, if the path candidates for cargo 1 to 3 are as shown in FIG. 18, and only cost is considered, the load (total cargo volume entering) will be concentrated on d30, which has the lowest cost. To avoid such a concentration of load, modeling is performed to reduce the load on the warehouse with the highest load. For example, a decision variable L indicating the maximum incoming cargo volume may be provided in the objective function. A constraint may be added to limit the total volume entering each warehouse to L (minimization of the maximum value). The constraint is that the total volume of cargo transported from warehouse i to warehouse j must be less than or equal to the upper capacity limit corresponding to edge (i, j), and an optimization problem is constructed to minimize the maximum total cargo volume entering warehouse j.

[0158] The existing logistics network data may include time-series data on the workload for each warehouse in the existing logistics network. For each warehouse, the workload may be based on the unloading volume and loading volume. The logistics network features may include similarity features related to the similarity of the workload time series between warehouses. The logistics network optimization unit may formulate an optimization problem to avoid two or more similar warehouses being included in the same cargo route. This is expected to enable the creation of a logistics network plan in which the loads on warehouses are distributed (for example, as described with reference to Figures 9A and 9B).

[0159] Specifically, for example, when generating path candidates, it is assumed that the fewer the number of transit warehouses, the lower the cost. Because the number of path candidates is exponential, it is difficult to generate all path candidates. Therefore, the logistics network optimization unit 130 may first enumerate path candidates for each package that have a maximum number of transit warehouses or less. This "maximum" may be input by the user or may be predetermined. The enumeration of path candidates may be performed using a depth-first search or the like. The logistics network optimization unit 130 may calculate the cost of each path candidate taking into account the warehouse role. For example, the cost may be the sum of the costs corresponding to the edges (i, j) that make up the path candidate. The logistics network optimization unit 130 may set existing paths from the existing logistics network. The logistics network optimization unit 130 may identify warehouses (e.g., warehouses with similar workload time series) similar to a warehouse with a high load (e.g., a warehouse with a load above a threshold or a warehouse with the highest load) from the existing logistics network, and set path candidates that do not pass through the similar warehouses and existing paths as path candidates. In other words, the path candidate other than the existing path may be a path candidate that does not pass through a warehouse similar to the warehouse with the high load.

[0160] The existing logistics network data may include time-series data on the workload for each warehouse in the existing logistics network. The logistics network optimization unit may formulate the optimization problem to maximize the workload difference between the warehouse with the highest workload and the warehouse with the lowest workload. This increases the number of packages passing through a specific warehouse, which is expected to improve the loading rate of transporters passing through that warehouse and improve efficiency.

[0161] The output unit may provide a UI (User Interface) screen according to the logistics network plan data. An example of the UI screen is the logistics network screen 1500 shown in FIG. 15. The UI screen may include a graph of a logistics network consisting of the derived optimal parcel routes, with warehouses as vertices and directed edges between warehouses on the parcel routes, and information about multiple warehouses corresponding to the vertices constituting the graph. This allows a user (e.g., a planner) to easily understand the proposed logistics network plan, thereby facilitating the development of a transportation plan. For example, the information about the multiple warehouses in the graph may include information about warehouses on a parcel route selected by the user from the graph, information representing the warehouse role of the warehouse (e.g., the warehouse ratio with the highest rate or frequency), or information representing the role ratio of the warehouse. The display mode of the vertices in the graph may be a display mode (e.g., shape, color, or size) corresponding to the warehouse role characteristic (e.g., the warehouse ratio with the highest rate or frequency) corresponding to the vertex.

[0162] The output unit may output the logistics network plan data as the logistics network data to a transportation planning unit (e.g., transportation planning system 40) that generates a transportation plan for multiple packages based on logistics network data in which multiple package routes are defined and target package data for multiple packages to be transported. The transportation planning unit may generate a transportation plan for multiple packages to be transported based on the target package data and the logistics network plan data. This is expected to contribute to the generation (drafting) of an appropriate transportation plan. [Explanation of symbols]

[0163] 100: Logistics network planning system, 101: Baggage data, 102: Vehicle data, 103: Warehouse data, 120: Logistics network analysis unit, 121: Warehouse role estimation unit, 122: Similar warehouse classification unit, 130: Logistics network optimization unit, 131: Baggage route optimization unit, 140: Input / output unit

Claims

1. an input unit for inputting existing logistics network data, which is data relating to an existing logistics network consisting of a plurality of cargo routes, each of which has two or more warehouses as components, from a first departure warehouse to a last arrival warehouse of the cargo; a logistics network analysis unit that analyzes the existing logistics network data and identifies logistics network characteristics that are characteristics of the existing logistics network; a logistics network optimization unit that formulates an optimization problem for minimizing a target based on one or more indicators based on the logistics network features and derives an optimal cargo route by solving the optimization problem; an output unit that outputs logistics network plan data that is data representing a planned logistics network including the derived cargo route; A logistics network planning system equipped with:

2. The existing logistics network data includes, for each of the plurality of packages, data representing a departure time and arrival time for each warehouse constituting the package route of the package, and a transporter for each warehouse on the package route, and also includes data representing a region for each warehouse; The logistics network analysis unit assigns a warehouse role to each warehouse where unloading and loading has been performed, for each piece of luggage, if the transporter that transported the luggage to the warehouse is different from the transporter that transported the luggage from the warehouse to its next destination warehouse, depending on whether the transportation of the luggage to the warehouse is branch line transportation or trunk line transportation, and whether the transportation of the luggage from the warehouse to its next destination warehouse is branch line transportation or trunk line transportation; The branch transportation is transportation between warehouses in the same region, The trunk transportation is transportation between warehouses in different regions, The logistics network characteristics include, for each warehouse, a warehouse role characteristic relating to one or more warehouse roles of the warehouse during a certain period; the logistics network optimization unit constructs the optimization problem so that the smaller the deviation from the warehouse role, the smaller the transportation cost. The logistics network planning system according to claim 1 .

3. The warehouse role characteristics may be based on a percentage or frequency of each of one or more warehouse roles for each warehouse over a period of time for that warehouse; the logistics network optimization unit constructs the optimization problem so that the degree to which the transportation cost increases as the deviation from the warehouse role increases increases as the proportion or frequency of the warehouse role decreases; The logistics network planning system according to claim 2.

4. the existing logistics network data includes time-series data of the amount of work for each warehouse in the existing logistics network, For each warehouse, the workload for that warehouse is based on the unloading and loading volumes; The logistics network features include a similarity feature related to a time series similarity of workloads between warehouses. the logistics network optimization unit formulates the optimization problem so as to avoid two or more warehouses that are similar to each other being included in the same cargo route; The logistics network planning system according to claim 1 .

5. the existing logistics network data includes time-series data of the amount of work for each warehouse in the existing logistics network, For each warehouse, the workload for that warehouse is based on the unloading and loading volumes; The logistics network optimization unit constructs the optimization problem so as to minimize the amount of cargo sent to the warehouse with the largest workload. The logistics network planning system according to claim 1 .

6. the existing logistics network data includes time-series data of the amount of work for each warehouse in the existing logistics network, For each warehouse, the workload for that warehouse is based on the unloading and loading volumes; the logistics network optimization unit constructs the optimization problem so as to maximize the difference in workload between the warehouse with the highest workload and the warehouse with the lowest workload; The logistics network planning system according to claim 1 .

7. the output unit provides a UI (User Interface) screen according to the logistics network plan data; The UI screen is a graph of a logistics network consisting of the derived multiple optimal cargo routes, with warehouses as vertices and directed edges between warehouses on the cargo routes; Information about a plurality of warehouses corresponding to a plurality of vertices constituting the graph; having The logistics network planning system according to claim 1 .

8. the output unit outputs the logistics network plan data as the logistics network data to a transportation planning unit that generates a transportation plan for the plurality of packages based on logistics network data in which a plurality of package routes are defined and target package data related to the plurality of packages to be transported; the transportation planning unit generates a transportation plan for the plurality of packages to be transported based on the target package data and the logistics network plan data; The logistics network planning system according to claim 1 .

9. the existing logistics network data includes, for each of the plurality of packages, data representing the volume of the package and a transporter for each warehouse on the package route of the package, and data representing a maximum load capacity for each transporter; the logistics network optimization unit determines, for each of the warehouses, an upper capacity limit between the warehouses based on the sum of the maximum load capacities of transporters that pass through the warehouses, based on the existing logistics network data; the logistics network optimization unit constructs the optimization problem so that one or more packages whose total capacity exceeds an upper capacity limit between warehouses do not pass through the warehouses; The logistics network planning system according to claim 1 .

10. The target is transportation cost. The logistics network planning system according to claim 1 .

11. The target is a weighted sum of transportation cost and the volume of goods going to the busiest warehouse. The logistics network planning system according to claim 1 .

12. The target is a weighted sum of the transportation cost and the negative value of the difference in workload between the warehouse with the highest workload and the warehouse with the lowest workload. The logistics network planning system according to claim 1 .

13. Inputting existing logistics network data, which is data on an existing logistics network consisting of a plurality of cargo routes, each of which has two or more warehouses as components, from a first departure warehouse of the cargo to a last arrival warehouse; Analyzing the existing logistics network data to identify logistics network characteristics that are characteristics of the existing logistics network; Formulating an optimization problem based on the logistics network characteristics to minimize a target based on one or more indicators, and deriving an optimal cargo route by solving the optimization problem; Outputting logistics network plan data, which is data representing a planned logistics network including the derived cargo route. A logistics network planning method that uses a computer.

Images