WO2022102160A1 - Product quantity prediction system - Google Patents

Product quantity prediction system Download PDF

Info

Publication number
WO2022102160A1
WO2022102160A1 PCT/JP2021/023518 JP2021023518W WO2022102160A1 WO 2022102160 A1 WO2022102160 A1 WO 2022102160A1 JP 2021023518 W JP2021023518 W JP 2021023518W WO 2022102160 A1 WO2022102160 A1 WO 2022102160A1
Authority
WO
WIPO (PCT)
Prior art keywords
business hours
change
physical quantity
store
predicted
Prior art date
Application number
PCT/JP2021/023518
Other languages
French (fr)
Japanese (ja)
Inventor
孝光 坂井
雄大 石塚
利男 加藤
洋市 山村
悠一 栗脇
Original Assignee
株式会社アイシン
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 株式会社アイシン filed Critical 株式会社アイシン
Priority to JP2022561268A priority Critical patent/JPWO2022102160A1/ja
Publication of WO2022102160A1 publication Critical patent/WO2022102160A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management

Definitions

  • the present invention relates to a physical quantity prediction system.
  • Patent Document 1 it is known to predict the sales volume of a product based on the past sales performance.
  • the present invention has been made in view of the above problems, and an object of the present invention is to provide a technique for predicting a physical quantity after a change in business hours when the business hours are changed.
  • the quantity forecast system is based on the business hours acquisition department that acquires changes in the business hours of the store and the actual quantity in the business hours before the change, and the quantity corresponding to the business hours after the change. It is provided with a physical quantity prediction unit for predicting.
  • the physical quantity prediction system predicts the physical quantity in the business hours after the change based on the actual physical quantity in the business hours before the change. Therefore, when the business hours are changed, the quantity after the change can be predicted.
  • FIG. 3A is a diagram showing an example of actual physical quantity
  • FIG. 3B is a diagram showing an example of input / output of machine learning. It is a flowchart of a physical quantity prediction process.
  • FIG. 1 is a block diagram showing a configuration of a physical quantity prediction system according to the present invention.
  • the physical quantity prediction system 10 is realized by a server operated from the administrator terminal 100.
  • the administrator terminal 100 can be realized by, for example, a general-purpose computer, a mobile terminal, or the like, and includes a display, an operation input unit, and the like (not shown).
  • the manager causes the physical quantity prediction system 10 to predict the physical quantity by operating the operation input unit of the manager terminal 100 in order to create a delivery plan of the product to a plurality of stores.
  • the delivery plan is a plan for one or more delivery vehicles to visit a plurality of stores in a predetermined order and carry out cargo handling to deliver the products to each store.
  • the delivery plan is created by defining the order of arrival at a plurality of stores and the period for cargo handling at each store.
  • FIG. 2A is a diagram showing an example of a delivery plan.
  • the store as the delivery destination is represented by identification information such as SA to SF .
  • the delivery plan for 24 hours is partially omitted.
  • the delivery plan defines at least the time when the cargo handling work begins.
  • the period for carrying out cargo handling work at the same store is within 30 minutes. Of course, the period may change depending on the distance between the store and the parking lot, the amount of luggage, and the like.
  • the products are classified into chilled, cooked rice, normal temperature, and frozen, and the delivery vehicle delivers the products to each store from each distribution base corresponding to the classification.
  • C, R, D, and F indicate the time zones for delivering chilled, cooked rice, normal temperature, and frozen products to each store, respectively.
  • the number of delivery vehicles departing from each base is arbitrary, and if there are multiple delivery vehicles, each delivery vehicle will share one of the multiple stores for delivery. In FIG. 2A, for the sake of simplicity, it is assumed that one delivery vehicle departs from each base.
  • the delivery vehicle that delivers the chilled ( C ) product departs from the chilled product base and then delivers in the order of the stores SB, SC , SD , SE , SF , and SA . ..
  • the delivery vehicle departing from the base of the chilled product starts cargo handling at store SB at 5:00 , finishes cargo handling at store SB , and moves to the next delivery destination, store SC.
  • the completion time is 5:30.
  • the time for starting cargo handling at the store SC is 5:30
  • the time for completing the transfer to the store SD, which is the next delivery destination is 6:00.
  • the delivery plan is defined by specifying the period during which the delivery vehicle departing from each base performs cargo handling work at each delivery destination store.
  • a plan can be created in which products belonging to the same category are delivered N times (N is 1 or more) on the same day.
  • N is 1 or more
  • the chilled product is delivered to each delivery destination store three times on the same day.
  • chilled products are delivered three times from 7:00, 10:30, and 16:30.
  • delivery of goods of the same category at different time zones is referred to as "flight”.
  • the delivery of the chilled (C) product and the delivery of the cooked rice (R) product exist three times a day, and the normal temperature (D) product delivery and the frozen (F) delivery.
  • Products are delivered once a day.
  • the delivery destination is a store
  • the parcel is a product sold at each store.
  • the peak time P is a time zone in which the store as the delivery destination is congested, and in the present embodiment, a delivery plan is created so that cargo handling does not occur at the peak time P.
  • the mode of the delivery plan is not limited to the mode of this example, and may be defined in various modes.
  • the physical quantity prediction system 10 includes a control unit 20 including a CPU, RAM, ROM, etc., a recording medium 30, and a communication unit 40.
  • the communication unit 40 includes a circuit for exchanging information with the administrator terminal 100.
  • the control unit 20 can communicate with the administrator terminal 100 via the communication unit 40.
  • the recording medium 30 records the store information 30a, the physical quantity actual 30b, the calendar information 30c, and the learned model 30d.
  • the store information 30a is defined for each store to which the product is delivered, and includes information such as a store ID, a store name, store location information, peak time, and business hours.
  • the business hours include, for example, information indicating a 24-hour business or, in the case of non-24-hour business, a business start time and a business end time. In addition, for stores that are scheduled to be changed from 24-hour business to non-24-hour business in the future, the change date is also included.
  • the store information 30a is input or changed by, for example, an administrator who operates the administrator terminal 100.
  • the actual quantity of goods 30b includes the actual quantity of goods for each store and each flight for each day in the past fixed period. Specifically, for example, as shown in FIG. 3A, the quantity is included in association with the date, the day of the week, the weekday / holiday classification, the weather, the temperature, the flight, and the store.
  • the day of the week or weekday / holiday classification is recorded by acquiring the day of the week or weekday / holiday classification corresponding to the date of the corresponding day from the calendar information 30c.
  • the physical quantity prediction system 10 can communicate with a weather information server 200 that provides weather information including future weather and temperature in each region via a communication unit 40.
  • the weather and temperature are recorded by acquiring the weather and temperature of the corresponding day from the weather information server 200.
  • C1, C2, and C3 indicate the first, second, and third chilled flights, respectively, and R1, R2, and R3 refer to the first, second, and third flights of cooked rice. Shows.
  • the physical quantity is represented by the number of containers for transporting goods such as a number weight. The number of containers actually used to deliver the products ordered from each store on the relevant flight on the relevant day is recorded.
  • the calendar information 30c is information in which a date is associated with a day of the week or a weekday / holiday classification.
  • the control unit 20 can acquire the day of the week and the weekday / holiday classification of the designated date by referring to the calendar information 30c.
  • the trained model 30d is a machine learning model generated by the control unit 20 by the function of the physical quantity prediction unit 21b described later. Details will be described later.
  • the control unit 20 can execute the program stored in the recording medium 30 or the ROM in the control unit 20.
  • the physical quantity prediction program 21 can be executed as this program.
  • the control unit 20 functions as a business hour acquisition unit 21a, a physical quantity prediction unit 21b, and a delivery plan acquisition unit 21c.
  • the control unit 20 acquires the change in the business hours of the store. That is, the control unit 20 refers to the store information 30a and acquires the business hours of the delivery destination store.
  • the control unit 20 has the business hours of the store on the date designated by the administrator as the day on which the quantity is to be predicted (hereinafter referred to as the prediction target date) and the business hours of the store before the prediction target date (machine learning teacher data described later). (Business hours of stores during the hiring period) is compared for each store, and if there is a store whose business hours are changed, the store whose business hours are changed is specified.
  • the business hours of the store SA are shortened from 6:00 to 23:00 (the non - business hours are from 23:00 to 6:00 the next day), and the business hours of the store SD are shortened.
  • An example is shown in which the business hours are shortened from 6:00 to 0:00 the next day (non-business hours are from 0:00 to 6:00).
  • the control unit 20 predicts the physical quantity corresponding to the business hours after the change based on the actual physical quantity in the business hours before the change.
  • the control unit 20 generates in advance a machine-learned model based on the actual physical quantity in the past fixed period, and stores it in the recording medium 30 (learned model 30d).
  • a machine-learned model based on the actual physical quantity in the past fixed period, and stores it in the recording medium 30 (learned model 30d).
  • a combination of day of the week, weekday / holiday classification, weather, maximum temperature, minimum temperature, flight, and store is input, and the actual quantity corresponding to the combination is input. This is a machine-learned model using the output teacher data.
  • the control unit 20 acquires a physical quantity that does not consider the change in business hours for each store and flight on the forecast target date (date after the change in business hours) using the trained model 30d. Therefore, the control unit 20 acquires the day of the week and the weekday / holiday classification of the prediction target day with reference to the calendar information 30c. Further, the control unit 20 acquires the weather and temperature of the forecast target day from the weather information server 200. Then, the control unit 20 inputs each acquired information, the flight, and the store into the learned model 30d, and acquires the physical quantity of each flight and the store on the prediction target date.
  • the trained model 30d can be additionally learned and updated at regular intervals based on teacher data based on daily physical quantity results, but in the present embodiment, the trained model 30d predicts the physical quantity when the change in business hours is not considered. Since it is supposed to be used for this purpose, it is a model learned from the actual quantity of goods before the change of business hours of stores SA and SD (that is, the period of 24-hour business).
  • the control unit 20 extracts out of the N flights in the store where the business hours are changed, the flights whose quantity is estimated to be affected by the change in the business hours.
  • the opportunity (hours) for selling the product increases or decreases, and as a result, the quantity ordered from the store increases or decreases, so that the quantity increases or decreases. Therefore, the quantity may be affected by the change in business hours.
  • D normal temperature
  • F frozen
  • the one flight is extracted as the flight whose quantity is affected.
  • C chilled
  • cooked rice (R) delivered by multiple flights a day the affected flights are extracted as follows in this embodiment.
  • Focusing on store SA for example, chilled has three delivery opportunities per day.
  • the C1 flight at 7:00 includes a product expected to sell in the morning including the peak time in the morning
  • the C2 flight at 10:30 includes the peak time at noon.
  • Products that are expected to sell by the evening including the above are included, and it is expected that the C3 flight at 16:30 will sell within about 14 hours around the C1 flight the next morning, including the peak time of the night. Products are included.
  • control unit 20 also considers the rice R, which has three flights per day, as the flight whose quantity is affected by the change in business hours, which is the flight closest to the business end time. Then, the control unit 20 predicts that the quantity of R3 flight of the store SA after the change of business hours is QR / 2 (QR is the quantity of R3 flight of the store SA before the change).
  • the control unit 20 considers that each of these one flight is affected by the change in business hours.
  • the control unit 20 sets the value obtained by multiplying the physical quantity Q F before the change by 17/24 as the predicted physical quantity of the frozen F after the change.
  • the quantity of the forecast target day when the change of business hours is not considered is first predicted by using the machine-learned model based on the actual quantity of the quantity in the business hours before the change. Multiply the forecasted quantity by the time ratio associated with the change in business hours to obtain the predicted value of the quantity after the change in business hours. Therefore, according to the present embodiment, when the business hours are changed, the quantity after the change can be predicted.
  • the control unit 20 predicts the physical quantity for each of a plurality of stores including the store whose business hours are changed. That is, for the flights estimated to be affected by the change in business hours of the store whose business hours are changed, the control unit 20 is a value obtained by multiplying the predicted physical quantity acquired using the trained model 30d by the time ratio. Is the predicted value.
  • the control unit 20 uses the output result of the trained model 30d as the predicted quantity for flights other than the flights estimated to be affected by the change in business hours of the store whose business hours are changed. Further, the control unit 20 uses the output result of the learned model 30d as the predicted quantity for each flight of the store whose business hours are not changed.
  • the control unit 20 uses the function of the delivery plan acquisition unit 21c to obtain a plurality of quantities based on the predicted quantity. Get a delivery plan to deliver goods to the store.
  • the control unit 20 acquires a delivery plan for each flight. That is, the control unit 20 outputs the predicted quantity of each store, the position of each store, the position of the delivery base (shipping point) of the target flight, and the like to the VRP (Vehicle Routing Problem) server 300 for each flight.
  • the VRP server 300 is made to create a delivery plan, and the created delivery plan is acquired.
  • the VRP server 300 is a server that creates a delivery plan in which a vehicle departing from a delivery base delivers a product to each store by a known algorithm.
  • C3 flight, R3 flight, D flight, and F flight are flights whose quantity is expected to change with a change in business hours.
  • the control unit 20 can acquire a delivery plan based on the changed quantity for these flights.
  • the cargo handling time can be shortened, so that a delivery plan indicating that the required time of the delivery plan can be shortened can be obtained.
  • FIGS. 2A and 2B show that there is only one delivery vehicle, delivery may be performed by a plurality of delivery vehicles.
  • a delivery plan was adopted in which all stores were open 24 hours a day, and C3 flights were shared and delivered by V vehicles, but some stores started non-24 hours. In that case, a delivery plan for delivering C3 flights with fewer delivery vehicles than V units can be obtained.
  • FIG. 4 is a flowchart showing a physical quantity prediction process executed by the control unit 20.
  • the administrator operates the administrator terminal 100 to input a change in the business hours of the store, and the administrator terminal 100 gives an instruction to start the physical quantity prediction processing. It is started when the control unit 20 acquires the information via the communication unit 40.
  • the control unit 20 acquires a machine-learned model using the past physical quantity actual results 30b by the function of the physical quantity prediction unit 21b (step S100). That is, the control unit 20 acquires the learned model 30d recorded on the recording medium 30. Subsequently, the control unit 20 acquires the weather information by the function of the physical quantity prediction unit 21b (step S105). That is, the control unit 20 acquires the weather information including the weather, the temperature, etc. of the prediction target day from the weather information server 200.
  • the control unit 20 acquires the calendar information 30c by the function of the physical quantity prediction unit 21b (step S110). That is, the control unit 20 refers to the calendar information 30c and acquires the day of the week and the weekday / holiday classification of the prediction target day. Subsequently, the control unit 20 predicts the physical quantity for each flight and each store on the prediction target day by using the function of the physical quantity prediction unit 21b (step S115). That is, the control unit 20 inputs the day of the week of the prediction target day, the weekday / holiday classification of the prediction target day, the weather and temperature of the prediction target day, the flight, and the store in the trained model 30d, and the flight and the store on the prediction target day. Get the predicted value of the physical quantity of.
  • the control unit 20 acquires the business hours of the delivery destination store by the function of the business hours acquisition unit 21a (step S120). That is, the control unit 20 refers to the store information 30a and acquires the business hours of each store of the delivery destination on the prediction target date. Subsequently, the control unit 20 determines whether or not there is a store whose business hours are changed in the delivery destination store by the function of the business time acquisition unit 21a (step S125). That is, the control unit 20 compares the business hours of the store on the prediction target date with the business hours of the stores before the prediction target date (the business hours of the store during the adoption period of the machine learning teacher data) for each store. Determine if there is a store whose business hours have changed.
  • the control unit 20 When it is determined in step S125 that there is a store whose business hours are changed, the control unit 20 reflects the change in the physical quantity due to the change in the business hours of the delivery destination store in the physical quantity prediction by the function of the physical quantity prediction unit 21b. (Step S130). That is, in the present embodiment, the control unit 20 identifies a store and a flight in which the quantity is estimated to be affected by the change in business hours, and multiplies the estimated quantity of the store and the flight by a time ratio. Acquire the quantity forecast value of the store and flight after changing the business hours.
  • control unit 20 After completing step S130, or if it is not determined in step S125 that there is a store whose business hours are changed, the control unit 20 ends the physical quantity prediction process. After that, the control unit 20 causes the VRP server 300 to create a delivery plan based on the predicted quantity for each flight and each store, and acquires the created delivery plan.
  • the physical quantity prediction system 10 and the VRP server 300 may be configured by the same device.
  • the physical quantity prediction system 10 may be composed of a plurality of systems.
  • some functions of the physical quantity prediction system 10 may be realized by the administrator terminal 100 or a cloud server.
  • the user of the administrator terminal 100 may exist at the distribution base or at the ordering party. Further, the administrator terminal 100 may be provided in the vehicle, may be a portable terminal, or the like. Further, at least a part of each unit (business hour acquisition unit 21a, physical quantity prediction unit 21b, delivery plan acquisition unit 21c) constituting the physical quantity prediction system 10 may be divided into a plurality of devices and exist. Further, it is possible to assume a configuration in which a part of the configuration of the above-described embodiment is omitted, or a configuration in which the processing is varied or omitted.
  • the business hours acquisition department should be able to acquire changes in the business hours of the store.
  • the change of business hours is the "shortening" or "extension" of business hours due to the change of at least one of the business start time and business end time, and the business start time and business end time do not change the length of business hours. May be any of the "shifts".
  • the change of the business hours may be configured to be input by the administrator, or may be configured to be automatically acquired periodically from a server that manages the business hours of the store, for example. Further, when a store whose business hours are changed is detected, the physical quantity prediction process may be automatically executed.
  • the physical quantity prediction unit may be able to predict the physical quantity corresponding to the business hours after the change based on the actual physical quantity in the business hours before the change, and may adopt various configurations other than the above-described embodiment. .. For example, when a delivery plan acquisition unit for acquiring a delivery plan for delivering products having different quantities to a plurality of stores is provided, a configuration for predicting the quantity of each flight and each store is adopted, but the delivery plan acquisition unit is adopted. If this is not provided, it is not always necessary to predict the quantity for each store or each flight.
  • the amount of goods on the forecast target day is predicted from the past physical quantity results with similar conditions such as weather and days, and the business hours of multiple stores before the change of business hours are By multiplying the time ratio with the sum as the denominator and the sum of the business hours of multiple stores after the business hours as the numerator, the shipment volume of chilled products for the forecast target day when the business hours are changed is predicted. May be.
  • a machine learning model may be used or a method such as multivariate analysis may be used for predicting the physical quantity before the change of business hours.
  • the quantity of the forecast target day may be predicted by multiplying the basic quantity by a coefficient according to the day of the week, weekday / holiday classification, weather, temperature, store size, store location condition, and the like.
  • a machine learning model is used for physical quantity prediction
  • various configurations can be adopted for the input and output configurations of the machine learning model in addition to the above embodiments.
  • a machine learning model may be generated for each store or each store and flight.
  • the inputs and outputs of the machine learning model are not limited to the examples of the above embodiments.
  • the physical quantity as the output of the machine learning model is the number of containers for each type. May be good.
  • the physical quantity prediction unit may be configured to predict the physical quantity of a store whose business hours are newly changed based on the change in physical quantity in another store whose business hours have been changed in the past. Specifically, for example, when it can be obtained from the past physical quantity actual that the physical quantity changed from Q X to Q Y on average when the business hours were changed from X hours to Y hours in the past, the change ratio of time.
  • the configuration is such that the quantity is predicted without considering the change in business hours, and the predicted quantity is corrected according to the change in business hours, but the configuration is not limited to that.
  • a machine learning model may be used to make predictions based on changes in the quantity of other stores whose business hours have changed in the past.
  • the quantity prediction unit can predict the quantity of deliveries (feces) that are estimated to be affected by changes in business hours out of N deliveries (feces) per day. good. If the business hours are changed and some of the hours that were previously business hours become non-business hours, it can be estimated that the products that were sold during those hours will not sell. Therefore, it can be estimated that the order quantity of the products sold in the time zone decreases, and as a result, the quantity of the delivered flights decreases.
  • the flight to which the product was delivered and the time zone in which the product was sold can be specified from the delivery record, POS data, and the like.
  • the present invention when the business hours are shortened, it is estimated that the flight immediately before the business end time is affected by the change in the business hours, but the present invention is not limited to this.
  • the shortened business hours can have an impact on flights other than those closest to the closing time.
  • the physical quantity prediction unit identifies the number of reductions in the physical quantity (for example, the number of transport containers) due to the M 1 % reduction in the products delivered to the store SA by C2 flight, and the physical quantity before the change in business hours. It is possible to calculate the quantity after changing the business hours by subtracting the decrease number from.
  • M2 % of the products delivered by C3 flight immediately after the change of business end time at store SA have not been sold in the midnight time zone (time zone when it is changed to non-business hours). If it sells at another time), it can be estimated that M2 % of the products on C3 flight to store SA are not affected by the change in business hours. Therefore, it may be considered that the M2 % of the products of the C3 flight to the store SA does not decrease and the (100- M2 )% decreases with respect to the time ratio, and the quantity after the change of business hours may be predicted.
  • the flight to deliver the products that sell during the relevant time period is determined (determined in consideration of the quantity of other flights, the expiration date of the product, etc.). You may predict that the quantity of the stool will increase.
  • the time zone that was previously business hours and the time zone to which flights were assigned may be changed to non-business hours.
  • the goods that should have been delivered by the relevant flight will not be delivered, so that the quantity may increase in the flights before and after the relevant flight.
  • the fourth chilled flight C4 flight
  • store SA since the time zone from 23:00 to 6:00 will be changed to non - business hours at store SA, store SA will not be able to receive chilled products on this C4 flight.
  • the present invention can also be applied as a program or a method.
  • the above systems, programs, and methods may be realized as a single device or may be realized by using parts shared with each part provided in the vehicle, including various aspects. It is a program.
  • some of them are software and some of them are hardware, so they can be changed as appropriate.
  • the invention is also established as a recording medium for a program for controlling an apparatus.
  • the recording medium of the software may be a magnetic recording medium or a semiconductor memory, and any recording medium developed in the future can be considered in exactly the same way.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Quality & Reliability (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

[Problem] To provide a technology for predicting a post-change product quantity when business hours are changed. [Solution] Configured is a product quantity prediction system provided with: a business hours acquisition unit that acquires a change of business hours of a shop; and a product quantity prediction unit that predicts a product quantity corresponding to post-change business hours on the basis of a past product quantity record during pre-change business hours.

Description

物量予測システムPhysical quantity prediction system
　本発明は、物量予測システムに関する。 The present invention relates to a physical quantity prediction system.
　従来、過去の販売実績に基づいて商品の販売量を予測することが知られている（例えば、特許文献１）。 Conventionally, it is known to predict the sales volume of a product based on the past sales performance (for example, Patent Document 1).
特開２００５－１９４０７３号公報Japanese Unexamined Patent Publication No. 2005-194073
　しかし、従来の技術では、１日当たりの営業時間が過去の販売実績における営業時間と異なっている場合に、適切に販売量を予測できない。
  本発明は、上記課題にかんがみてなされたもので、営業時間が変更される場合に、変更後の物量を予測する技術の提供を目的とする。 However, with the conventional technique, when the business hours per day are different from the business hours in the past sales results, the sales volume cannot be predicted appropriately.
The present invention has been made in view of the above problems, and an object of the present invention is to provide a technique for predicting a physical quantity after a change in business hours when the business hours are changed.
　上記の目的を達成するため、物量予測システムは、店舗の営業時間の変更を取得する営業時間取得部と、変更前の営業時間での物量実績に基づいて、変更後の営業時間に対応する物量を予測する物量予測部と、を備える。 In order to achieve the above objectives, the quantity forecast system is based on the business hours acquisition department that acquires changes in the business hours of the store and the actual quantity in the business hours before the change, and the quantity corresponding to the business hours after the change. It is provided with a physical quantity prediction unit for predicting.
　すなわち、物量予測システムでは、変更前の営業時間での物量実績に基づいて、変更後の営業時間での物量を予測する。そのため、営業時間が変更される場合に、変更後の物量を予測することができる。 That is, the physical quantity prediction system predicts the physical quantity in the business hours after the change based on the actual physical quantity in the business hours before the change. Therefore, when the business hours are changed, the quantity after the change can be predicted.
物量予測システムの構成を示すブロック図である。It is a block diagram which shows the structure of the physical quantity prediction system. 図２Ａおよび図２Ｂは、配送計画の例を示す図である。2A and 2B are diagrams showing an example of a delivery plan. 図３Ａは物量実績の例を示す図であり、図３Ｂは機械学習の入出力例を示す図である。FIG. 3A is a diagram showing an example of actual physical quantity, and FIG. 3B is a diagram showing an example of input / output of machine learning. 物量予測処理のフローチャートである。It is a flowchart of a physical quantity prediction process.
　ここでは、下記の順序に従って本発明の実施の形態について説明する。
（１）物量予測システムの構成：
（２）物量予測処理：
（３）他の実施形態： Here, embodiments of the present invention will be described in the following order.
(1) Configuration of physical quantity prediction system:
(2) Physical quantity prediction processing:
(3) Other embodiments:
　（１）物量予測システムの構成：
  図１は、本発明にかかる物量予測システムの構成を示すブロック図である。本実施形態において物量予測システム１０は、管理者端末１００から操作されるサーバによって実現される。管理者端末１００は、例えば、汎用コンピュータや携帯端末等で実現可能であり、図示しないディスプレイや操作入力部等を備えている。管理者は、複数の店舗に対する商品の配送計画を作成するため、管理者端末１００の操作入力部を操作することにより、物量予測システム１０に物量を予測させる。 (1) Configuration of physical quantity prediction system:
FIG. 1 is a block diagram showing a configuration of a physical quantity prediction system according to the present invention. In the present embodiment, the physical quantity prediction system 10 is realized by a server operated from the administrator terminal 100. The administrator terminal 100 can be realized by, for example, a general-purpose computer, a mobile terminal, or the like, and includes a display, an operation input unit, and the like (not shown). The manager causes the physical quantity prediction system 10 to predict the physical quantity by operating the operation input unit of the manager terminal 100 in order to create a delivery plan of the product to a plurality of stores.
　本実施形態においては、例えばコンビニエンスストアの各店舗に配送する商品の物量を予測することを想定している。昨今、コンビニエンスストア各社では、２４時間営業から営業時間を短縮することが議論されている。営業時間を短縮する店舗が現れた場合の物量の変化を予測して変化後の物量での配送計画を予め作成し、作成した配送計画によって営業時間の変更に即応した無駄の少ない配送を実現できることが望ましい。 In this embodiment, it is assumed that, for example, the quantity of products to be delivered to each convenience store is predicted. Recently, convenience store companies are discussing shortening business hours from 24-hour business. It is possible to predict changes in the quantity of goods when a store that shortens business hours appears, create a delivery plan with the changed quantity in advance, and realize less wasteful delivery in response to changes in business hours by the created delivery plan. Is desirable.
　本実施形態において、配送計画は、１台以上の配送車両が複数の店舗を決められた順序で訪問し、各店舗に商品を届ける荷役を行うための計画である。本実施形態において配送計画は、複数の店舗に到着する順序および各店舗で荷役を行う期間が定義されることによって作成される。 In the present embodiment, the delivery plan is a plan for one or more delivery vehicles to visit a plurality of stores in a predetermined order and carry out cargo handling to deliver the products to each store. In the present embodiment, the delivery plan is created by defining the order of arrival at a plurality of stores and the period for cargo handling at each store.
　図２Ａは、配送計画の一例を示す図である。図２Ａにおいて、配送先である店舗は、ＳＡ～ＳＦ等の識別情報で表記されている。また、図２Ａにおいては、２４時間分の配送計画が一部省略して表記されている。配送計画では、荷役作業を開始する時刻が少なくとも定義される。図２Ａに示す例において、同一の店舗で荷役作業を行う期間が３０分以内であることが想定されている。むろん、当該期間は、店舗と駐車場との距離や荷物の量等に応じて変化しても良い。 FIG. 2A is a diagram showing an example of a delivery plan. In FIG. 2A, the store as the delivery destination is represented by identification information such as SA to SF . Further, in FIG. 2A, the delivery plan for 24 hours is partially omitted. The delivery plan defines at least the time when the cargo handling work begins. In the example shown in FIG. 2A, it is assumed that the period for carrying out cargo handling work at the same store is within 30 minutes. Of course, the period may change depending on the distance between the store and the parking lot, the amount of luggage, and the like.
　本実施形態において、商品は、チルド、米飯、常温、フローズンに分類されており、当該分類に対応する各物流拠点から配送車両が商品を各店舗に配送する。図２Ａにおいて、Ｃ，Ｒ，Ｄ，Ｆは、それぞれ、チルド、米飯、常温、フローズンの商品を各店舗に配送する時間帯を示している。各拠点から出発する配送車両の数は任意であり、複数台である場合には各配送車両で複数の店舗のいずれかを分担して配送を行う計画になる。図２Ａにおいては、簡単のため、各拠点から１台の配送車両が出発する状態が想定されている。 In the present embodiment, the products are classified into chilled, cooked rice, normal temperature, and frozen, and the delivery vehicle delivers the products to each store from each distribution base corresponding to the classification. In FIG. 2A, C, R, D, and F indicate the time zones for delivering chilled, cooked rice, normal temperature, and frozen products to each store, respectively. The number of delivery vehicles departing from each base is arbitrary, and if there are multiple delivery vehicles, each delivery vehicle will share one of the multiple stores for delivery. In FIG. 2A, for the sake of simplicity, it is assumed that one delivery vehicle departs from each base.
　例えば、図２Ａにおいて、チルド（Ｃ）の商品を配送する配送車両は、チルド商品の拠点を出発後、店舗ＳＢ，ＳＣ，ＳＤ，ＳＥ，ＳＦ，ＳＡの順に配送を行う。また、チルド商品の拠点を出発した配送車両が店舗ＳＢで荷役を開始する時刻は５：００であり、店舗ＳＢでの荷役を終了して次の配送先である店舗ＳＣに移動を完了する時刻は５：３０である。また、店舗ＳＣにて荷役を開始する時刻は５：３０であり、次の配送先である店舗ＳＤに移動を完了する時刻は６：００である。このように、各拠点から出発する配送車両が各配送先の店舗で荷役作業を行う期間が特定されることで、配送計画が定義される。 For example, in FIG. 2A, the delivery vehicle that delivers the chilled ( C ) product departs from the chilled product base and then delivers in the order of the stores SB, SC , SD , SE , SF , and SA . .. In addition, the delivery vehicle departing from the base of the chilled product starts cargo handling at store SB at 5:00 , finishes cargo handling at store SB , and moves to the next delivery destination, store SC. The completion time is 5:30. Further, the time for starting cargo handling at the store SC is 5:30 , and the time for completing the transfer to the store SD, which is the next delivery destination, is 6:00. In this way, the delivery plan is defined by specifying the period during which the delivery vehicle departing from each base performs cargo handling work at each delivery destination store.
　本実施形態においては、同一の分類に属する商品を同日にＮ回（Ｎは１以上）配送する計画が作成され得る。例えば、図２Ａに示す例であれば、同日にチルドの商品が配送先の各店舗に３回配送される。店舗ＳＡであれば、７：００からと、１０：３０からと、１６：３０からの３回チルドの商品が配送される。これらを区別するため、本明細書では、同一の分類の商品の異なる時間帯における配送を「便」と呼ぶ。例えば、図２Ａに示す例において、チルド（Ｃ）の商品の配送と、米飯（Ｒ）の商品の配送は１日に３便ずつ存在し、常温（Ｄ）の商品の配送と、フローズン（Ｆ）の商品の配送は１日に１便ずつ存在する。 In the present embodiment, a plan can be created in which products belonging to the same category are delivered N times (N is 1 or more) on the same day. For example, in the example shown in FIG. 2A, the chilled product is delivered to each delivery destination store three times on the same day. At store SA , chilled products are delivered three times from 7:00, 10:30, and 16:30. To distinguish between them, in this specification, delivery of goods of the same category at different time zones is referred to as "flight". For example, in the example shown in FIG. 2A, the delivery of the chilled (C) product and the delivery of the cooked rice (R) product exist three times a day, and the normal temperature (D) product delivery and the frozen (F) delivery. ) Products are delivered once a day.
　なお、本実施形態において配送先は店舗であり、荷物は各店舗で販売される商品である。ピークタイムＰは、配送先である店舗が混雑する時間帯であり、本実施形態においては、当該ピークタイムＰにおいて荷役が発生しないように配送計画が作成される。なお、配送計画の態様は、本例の態様に限定されず、種々の態様で定義されて良い。 In this embodiment, the delivery destination is a store, and the parcel is a product sold at each store. The peak time P is a time zone in which the store as the delivery destination is congested, and in the present embodiment, a delivery plan is created so that cargo handling does not occur at the peak time P. The mode of the delivery plan is not limited to the mode of this example, and may be defined in various modes.
　物量予測システム１０は、ＣＰＵ，ＲＡＭ，ＲＯＭ等を備える制御部２０、記録媒体３０、通信部４０を備えている。通信部４０は、管理者端末１００と情報の授受を行う回路を備えている。制御部２０は、通信部４０を介して管理者端末１００と通信を行うことができる。 The physical quantity prediction system 10 includes a control unit 20 including a CPU, RAM, ROM, etc., a recording medium 30, and a communication unit 40. The communication unit 40 includes a circuit for exchanging information with the administrator terminal 100. The control unit 20 can communicate with the administrator terminal 100 via the communication unit 40.
　また、記録媒体３０には、店舗情報３０ａと物量実績３０ｂとカレンダー情報３０ｃと学習済モデル３０ｄが記録されている。店舗情報３０ａは、商品の配送先である店舗毎に定義され、店舗ＩＤ、店舗名、店舗の位置情報、ピークタイム、営業時間等の情報を含んでいる。営業時間には、例えば、２４時間営業や、２４時間営業でない場合には営業開始時刻と営業終了時刻を示す情報が含まれる。また、今後２４時間営業から非２４時間営業に変更される予定の店舗については、変更日も含まれる。店舗情報３０ａは、例えば、管理者端末１００を操作する管理者によって入力や変更が行われる。 Further, the recording medium 30 records the store information 30a, the physical quantity actual 30b, the calendar information 30c, and the learned model 30d. The store information 30a is defined for each store to which the product is delivered, and includes information such as a store ID, a store name, store location information, peak time, and business hours. The business hours include, for example, information indicating a 24-hour business or, in the case of non-24-hour business, a business start time and a business end time. In addition, for stores that are scheduled to be changed from 24-hour business to non-24-hour business in the future, the change date is also included. The store information 30a is input or changed by, for example, an administrator who operates the administrator terminal 100.
　物量実績３０ｂは、過去の一定期間の各日の店舗毎および便毎の物量の実績が含まれる。具体的には例えば、図３Ａに示すように、日付、曜日、平日／休日区分、天候、気温、便、店舗に対応付けて物量が含まれる。曜日や平日／休日区分は、カレンダー情報３０ｃから該当日の日付と対応する曜日や平日／休日区分を取得することにより記録される。物量予測システム１０は、通信部４０を介して各地域の将来の天候や気温等を含む天候情報を提供する天候情報サーバ２００と通信可能である。天候や気温は、天候情報サーバ２００から該当日の天候や気温を取得することにより記録される。 The actual quantity of goods 30b includes the actual quantity of goods for each store and each flight for each day in the past fixed period. Specifically, for example, as shown in FIG. 3A, the quantity is included in association with the date, the day of the week, the weekday / holiday classification, the weather, the temperature, the flight, and the store. The day of the week or weekday / holiday classification is recorded by acquiring the day of the week or weekday / holiday classification corresponding to the date of the corresponding day from the calendar information 30c. The physical quantity prediction system 10 can communicate with a weather information server 200 that provides weather information including future weather and temperature in each region via a communication unit 40. The weather and temperature are recorded by acquiring the weather and temperature of the corresponding day from the weather information server 200.
　図３ＡにおいてＣ１、Ｃ２、Ｃ３は、それぞれチルドの第１便、第２便、第３便を示しており、Ｒ１，Ｒ２，Ｒ３は、米飯の第１便、第２便、第３便を示している。本実施形態において物量は、例えば番重などの商品運搬用の容器の個数で表されることを想定している。該当日の該当便において、各店舗から発注された商品を配送するために実際に用いられた容器の個数が記録される。 In FIG. 3A, C1, C2, and C3 indicate the first, second, and third chilled flights, respectively, and R1, R2, and R3 refer to the first, second, and third flights of cooked rice. Shows. In the present embodiment, it is assumed that the physical quantity is represented by the number of containers for transporting goods such as a number weight. The number of containers actually used to deliver the products ordered from each store on the relevant flight on the relevant day is recorded.
　カレンダー情報３０ｃは、日付と曜日や平日／祝日の区分とを対応付けた情報である。制御部２０は、カレンダー情報３０ｃを参照することにより、指定した日付の曜日や、平日／祝日の区分を取得することができる。学習済モデル３０ｄは、後述する物量予測部２１ｂの機能により、制御部２０によって生成される機械学習モデルである。詳細は後述する。 The calendar information 30c is information in which a date is associated with a day of the week or a weekday / holiday classification. The control unit 20 can acquire the day of the week and the weekday / holiday classification of the designated date by referring to the calendar information 30c. The trained model 30d is a machine learning model generated by the control unit 20 by the function of the physical quantity prediction unit 21b described later. Details will be described later.
　制御部２０は、記録媒体３０やＲＯＭに記憶されたプログラムを制御部２０で実行することができる。本実施形態においては、このプログラムとして物量予測プログラム２１を実行可能である。物量予測プログラム２１が実行されると、制御部２０は、営業時間取得部２１ａ、物量予測部２１ｂ、配送計画取得部２１ｃとして機能する。 The control unit 20 can execute the program stored in the recording medium 30 or the ROM in the control unit 20. In the present embodiment, the physical quantity prediction program 21 can be executed as this program. When the physical quantity prediction program 21 is executed, the control unit 20 functions as a business hour acquisition unit 21a, a physical quantity prediction unit 21b, and a delivery plan acquisition unit 21c.
　営業時間取得部２１ａの機能により、制御部２０は、店舗の営業時間の変更を取得する。すなわち、制御部２０は、店舗情報３０ａを参照し、配送先の店舗の営業時間を取得する。制御部２０は、物量を予測したい日として管理者が指定した日付（以降、予測対象日という）における店舗の営業時間と、予測対象日より前の店舗の営業時間（後述する機械学習の教師データの採用期間における店舗の営業時間）とを店舗毎に比較し、営業時間が変更になっている店舗が存在する場合、営業時間が変更される店舗を特定する。 By the function of the business hour acquisition unit 21a, the control unit 20 acquires the change in the business hours of the store. That is, the control unit 20 refers to the store information 30a and acquires the business hours of the delivery destination store. The control unit 20 has the business hours of the store on the date designated by the administrator as the day on which the quantity is to be predicted (hereinafter referred to as the prediction target date) and the business hours of the store before the prediction target date (machine learning teacher data described later). (Business hours of stores during the hiring period) is compared for each store, and if there is a store whose business hours are changed, the store whose business hours are changed is specified.
　本実施形態では、配送先の複数の２４時間営業の店舗のうちの一部が非２４時間営業に変更になる例を想定している。具体的には例えば、図２Ｂは、店舗ＳＡの営業時間が６：００から２３：００までに短縮され（２３：００から翌日６：００までが非営業時間となる）、店舗ＳＤの営業時間が６：００から翌日０：００までに短縮される（０：００から６：００まで非営業時間となる）となる例を示している。 In this embodiment, it is assumed that some of the plurality of 24-hour stores of the delivery destination are changed to non-24-hour business. Specifically, for example, in FIG. 2B, the business hours of the store SA are shortened from 6:00 to 23:00 (the non - business hours are from 23:00 to 6:00 the next day), and the business hours of the store SD are shortened. An example is shown in which the business hours are shortened from 6:00 to 0:00 the next day (non-business hours are from 0:00 to 6:00).
　物量予測部２１ｂの機能により、制御部２０は、変更前の営業時間での物量実績に基づいて、変更後の営業時間に対応する物量を予測する。本実施形態において、制御部２０は、過去の一定期間における物量実績に基づいて機械学習したモデルを予め生成し、記録媒体３０に保存している（学習済モデル３０ｄ）。本実施形態における機械学習モデルは、例えば図３Ｂに示すように、曜日、平日／休日区分、天候、最高気温、最低気温、便、店舗の組み合わせを入力とし、当該組み合わせに対応する実際の物量を出力する教師データを用いて機械学習したモデルである。 By the function of the physical quantity prediction unit 21b, the control unit 20 predicts the physical quantity corresponding to the business hours after the change based on the actual physical quantity in the business hours before the change. In the present embodiment, the control unit 20 generates in advance a machine-learned model based on the actual physical quantity in the past fixed period, and stores it in the recording medium 30 (learned model 30d). In the machine learning model in this embodiment, for example, as shown in FIG. 3B, a combination of day of the week, weekday / holiday classification, weather, maximum temperature, minimum temperature, flight, and store is input, and the actual quantity corresponding to the combination is input. This is a machine-learned model using the output teacher data.
　制御部２０は、まず、予測対象日（営業時間変更後の日付）における店舗および便毎に、営業時間変更を考慮しない物量を、学習済モデル３０ｄを用いて取得する。そのために、制御部２０は、予測対象日の曜日や平日／休日区分を、カレンダー情報３０ｃを参照して取得する。また、制御部２０は、予測対象日の天候および気温を、天候情報サーバ２００から取得する。そして、制御部２０は、取得した各情報と便および店舗を学習済モデル３０ｄに入力し、予測対象日における便および店舗毎の物量を取得する。 First, the control unit 20 acquires a physical quantity that does not consider the change in business hours for each store and flight on the forecast target date (date after the change in business hours) using the trained model 30d. Therefore, the control unit 20 acquires the day of the week and the weekday / holiday classification of the prediction target day with reference to the calendar information 30c. Further, the control unit 20 acquires the weather and temperature of the forecast target day from the weather information server 200. Then, the control unit 20 inputs each acquired information, the flight, and the store into the learned model 30d, and acquires the physical quantity of each flight and the store on the prediction target date.
　なお学習済モデル３０ｄは、毎日の物量実績に基づく教師データにより一定期間毎に追加学習され更新されうるが、本実施形態において学習済モデル３０ｄは、営業時間の変更を考慮しない場合の物量を予測するために用いることを想定しているため、店舗ＳＡやＳＤの営業時間の変更前（すなわち２４時間営業の期間）における物量実績によって学習されたモデルである。 The trained model 30d can be additionally learned and updated at regular intervals based on teacher data based on daily physical quantity results, but in the present embodiment, the trained model 30d predicts the physical quantity when the change in business hours is not considered. Since it is supposed to be used for this purpose, it is a model learned from the actual quantity of goods before the change of business hours of stores SA and SD (that is, the period of 24-hour business).
　続いて、制御部２０は、営業時間が変更される店舗におけるＮ回の便のうち、営業時間の変更に物量が影響を受けることが推定される便を抽出する。営業時間が変更されることにより商品が売れる機会（時間）が増減し、その結果、店舗からの商品の発注量が増減するため物量が増減する。そのため、営業時間の変更に物量が影響を受けうる。１日に１便のみで配送される常温（Ｄ）やフローズン（Ｆ）については、当該１便が、物量が影響を受ける便として抽出される。１日に複数便で配送されるチルド（Ｃ）や米飯（Ｒ）については、本実施形態では次のように影響を受ける便が抽出される。 Subsequently, the control unit 20 extracts out of the N flights in the store where the business hours are changed, the flights whose quantity is estimated to be affected by the change in the business hours. By changing the business hours, the opportunity (hours) for selling the product increases or decreases, and as a result, the quantity ordered from the store increases or decreases, so that the quantity increases or decreases. Therefore, the quantity may be affected by the change in business hours. For normal temperature (D) and frozen (F), which are delivered only once a day, the one flight is extracted as the flight whose quantity is affected. For chilled (C) and cooked rice (R) delivered by multiple flights a day, the affected flights are extracted as follows in this embodiment.
　店舗ＳＡに注目すると、例えばチルドは、１日当たり３便、配送の機会がある。本実施形態においては、例えば、７：００のＣ１便には、朝のピークタイムを含む午前中に売れることが期待される商品が含まれ、１０：３０のＣ２便には、正午のピークタイムを含む夕方頃までに売れることが期待される商品が含まれ、１６：３０のＣ３便には夜のピークタイムを含み翌日の朝のＣ１便頃の約１４時間の間に売れることが期待される商品が含まれる。 Focusing on store SA, for example, chilled has three delivery opportunities per day. In the present embodiment, for example, the C1 flight at 7:00 includes a product expected to sell in the morning including the peak time in the morning, and the C2 flight at 10:30 includes the peak time at noon. Products that are expected to sell by the evening including the above are included, and it is expected that the C3 flight at 16:30 will sell within about 14 hours around the C1 flight the next morning, including the peak time of the night. Products are included.
　本実施形態において、制御部２０は、これまで営業時間であった２３：００から６：００までの７時間が非営業時間となることによって、Ｃ３便で店舗ＳＡに配送される商品の販売の機会が１４時間から７時間（＝１４－７）に減少すると見積もる。すなわち本実施形態において、制御部２０は、営業終了時刻の直近の便を、営業時間の変更に物量が影響を受ける便として抽出する。 In the present embodiment, the control unit 20 sells a product delivered to the store SA by C3 flight by changing the 7 hours from 23:00 to 6:00, which was the business hours so far, into non-business hours. It is estimated that the opportunity will decrease from 14 hours to 7 hours (= 14-7). That is, in the present embodiment, the control unit 20 extracts the flight closest to the business end time as a flight whose quantity is affected by the change in business hours.
　さらに、制御部２０は、変更前の営業時間の長さＴ１を分母とし変更後の営業時間の長さＴ２を分子とする時間比（Ｔ２／Ｔ１）を、変更前の営業時間における物量Ｑに乗じた値に基づいて、営業時間変更後の物量を予測する。店舗ＳＡのＣ３便の場合、変更前の物量をＱＣとすると、制御部２０は、ＱＣ×７／１４＝ＱＣ／２を営業時間変更後のＣ３便の物量と予測する。本実施形態において制御部２０は、１日当たり３便存在する米飯Ｒについても同様に、営業終了時刻の直近の便であるＲ３を、営業時間の変更に物量が影響を受ける便と見なす。そして、制御部２０は、営業時間変更後の店舗ＳＡのＲ３便の物量を、ＱＲ／２と予測する（ＱＲは、変更前の店舗ＳＡのＲ３便の物量）。 Further, the control unit 20 sets the time ratio (T 2 / T 1 ) with the length T 1 of the business hours before the change as the denominator and the length T 2 of the business hours after the change as the numerator, and the business hours before the change. Based on the value multiplied by the physical quantity Q in, the physical quantity after the change of business hours is predicted. In the case of the C3 flight of the store SA , assuming that the quantity before the change is QC, the control unit 20 predicts that QC × 7/14 = QC / 2 is the quantity of the C3 flight after the change in business hours. In the present embodiment, the control unit 20 also considers the rice R, which has three flights per day, as the flight whose quantity is affected by the change in business hours, which is the flight closest to the business end time. Then, the control unit 20 predicts that the quantity of R3 flight of the store SA after the change of business hours is QR / 2 (QR is the quantity of R3 flight of the store SA before the change).
　本実施形態において、１日１便の常温ＤおよびフローズンＦについては、制御部２０は、これら１便がそれぞれ営業時間変更の影響を受けると見なす。店舗ＳＡの場合、営業時間が２４時間から１７時間（＝２４－７）に減少するため、変更前の物量ＱＤに１７／２４を乗じた値を、変更後の常温Ｄの便の予測物量とする。フローズンＦについても同様に、制御部２０は、変更前の物量ＱＦに１７／２４を乗じた値を変更後のフローズンＦの予測物量とする。 In the present embodiment, for the normal temperature D and the frozen F of one flight per day, the control unit 20 considers that each of these one flight is affected by the change in business hours. In the case of store SA , the business hours will decrease from 24 hours to 17 hours (= 24-7), so the value obtained by multiplying the quantity QD before the change by 17/24 is predicted for the flight at room temperature D after the change. It is a physical quantity. Similarly for the frozen F, the control unit 20 sets the value obtained by multiplying the physical quantity Q F before the change by 17/24 as the predicted physical quantity of the frozen F after the change.
　以上のように、本実施形態によれば、変更前の営業時間での物量実績に基づいて機械学習したモデルを用いて、まず営業時間変更を考慮しない場合の予測対象日の物量を予測し、予測物量に営業時間変更に伴う時間比を乗じて営業時間変更後の物量の予測値とする。従って本実施形態によれば、営業時間が変更される場合に、変更後の物量を予測することができる。 As described above, according to the present embodiment, the quantity of the forecast target day when the change of business hours is not considered is first predicted by using the machine-learned model based on the actual quantity of the quantity in the business hours before the change. Multiply the forecasted quantity by the time ratio associated with the change in business hours to obtain the predicted value of the quantity after the change in business hours. Therefore, according to the present embodiment, when the business hours are changed, the quantity after the change can be predicted.
　物量予測部２１ｂの機能により、制御部２０は、営業時間が変更される店舗を含む複数の店舗毎の物量を予測する。すなわち、制御部２０は、営業時間が変更される店舗の、営業時間の変更に影響されると推定される便については、学習済モデル３０ｄを用いた取得した予測物量に時間比を乗じた値を予測値とする。制御部２０は、営業時間が変更される店舗の、営業時間の変更に影響されると推定される便以外の便については、学習済モデル３０ｄの出力結果を予測物量とする。また、制御部２０は、営業時間が変更されない店舗の各便について、学習済モデル３０ｄの出力結果を予測物量とする。 By the function of the physical quantity prediction unit 21b, the control unit 20 predicts the physical quantity for each of a plurality of stores including the store whose business hours are changed. That is, for the flights estimated to be affected by the change in business hours of the store whose business hours are changed, the control unit 20 is a value obtained by multiplying the predicted physical quantity acquired using the trained model 30d by the time ratio. Is the predicted value. The control unit 20 uses the output result of the trained model 30d as the predicted quantity for flights other than the flights estimated to be affected by the change in business hours of the store whose business hours are changed. Further, the control unit 20 uses the output result of the learned model 30d as the predicted quantity for each flight of the store whose business hours are not changed.
　以上のようにして、予測対象日における全ての店舗の全ての便の物量の予測値を取得した後、制御部２０は、配送計画取得部２１ｃの機能により、予測された物量に基づいて複数の店舗に商品を配送するための配送計画を取得する。本実施形態において、制御部２０は、便毎の配送計画を取得する。すなわち、制御部２０は、便毎に、各店舗の予測物量と、各店舗の位置と、対象の便の配送拠点（出荷地点）の位置等を、ＶＲＰ（Vehicle Routing Problem）サーバ３００に出力し、ＶＲＰサーバ３００に配送計画を作成させ、作成された配送計画を取得する。ＶＲＰサーバ３００は、配送拠点を出発した車両が各店舗に商品を配送する配送計画を公知のアルゴリズムによって作成するサーバである。 As described above, after acquiring the predicted value of the quantity of all flights of all the stores on the forecast target date, the control unit 20 uses the function of the delivery plan acquisition unit 21c to obtain a plurality of quantities based on the predicted quantity. Get a delivery plan to deliver goods to the store. In the present embodiment, the control unit 20 acquires a delivery plan for each flight. That is, the control unit 20 outputs the predicted quantity of each store, the position of each store, the position of the delivery base (shipping point) of the target flight, and the like to the VRP (Vehicle Routing Problem) server 300 for each flight. , The VRP server 300 is made to create a delivery plan, and the created delivery plan is acquired. The VRP server 300 is a server that creates a delivery plan in which a vehicle departing from a delivery base delivers a product to each store by a known algorithm.
　本実施形態において、例えばＣ３便、Ｒ３便、Ｄ便、Ｆ便（図２Ｂを参照）は、営業時間の変更に伴って物量が変化することが予測される便である。制御部２０は、これらの便について変化後の物量に基づく配送計画を取得することができる。営業時間が短縮する店舗が出現し物量が減少することにより、例えば荷役時間が短縮できるため、配送計画の所要時間が短縮できることを示す配送計画が取得されうる。また、図２Ａおよび図２Ｂの例では配送車両が１台であることを示しているが、複数の配送車両で配送を行っていても良い。その場合例えば、全ての店舗で２４時間営業を行っていた期間のＣ３便はＶ台の車両で分担して配送する配送計画が採用されていたが、一部の店舗で非２４時間営業を開始する場合にはＶ台よりも少ない配送車両でＣ３便の配送を実施する配送計画が取得されうる。 In the present embodiment, for example, C3 flight, R3 flight, D flight, and F flight (see FIG. 2B) are flights whose quantity is expected to change with a change in business hours. The control unit 20 can acquire a delivery plan based on the changed quantity for these flights. By the appearance of stores with shortened business hours and the reduction of the quantity of goods, for example, the cargo handling time can be shortened, so that a delivery plan indicating that the required time of the delivery plan can be shortened can be obtained. Further, although the examples of FIGS. 2A and 2B show that there is only one delivery vehicle, delivery may be performed by a plurality of delivery vehicles. In that case, for example, a delivery plan was adopted in which all stores were open 24 hours a day, and C3 flights were shared and delivered by V vehicles, but some stores started non-24 hours. In that case, a delivery plan for delivering C3 flights with fewer delivery vehicles than V units can be obtained.
　（２）物量予測処理：
  次に、制御部２０が実行する物量予測処理を説明する。図４は、制御部２０が実行する物量予測処理を示すフローチャートである。本実施形態において、当該物量予測処理は、管理者が管理者端末１００を操作して店舗の営業時間の変更を入力し、管理者端末１００から物量予測処理の開始指示が行われたことを、制御部２０が通信部４０を介して取得した場合に開始される。 (2) Physical quantity prediction processing:
Next, the physical quantity prediction process executed by the control unit 20 will be described. FIG. 4 is a flowchart showing a physical quantity prediction process executed by the control unit 20. In the present embodiment, in the physical quantity prediction processing, the administrator operates the administrator terminal 100 to input a change in the business hours of the store, and the administrator terminal 100 gives an instruction to start the physical quantity prediction processing. It is started when the control unit 20 acquires the information via the communication unit 40.
　物量予測処理が開始されると、制御部２０は、物量予測部２１ｂの機能により、過去の物量実績３０ｂを用いて機械学習されたモデルを取得する（ステップＳ１００）。すなわち、制御部２０は、記録媒体３０に記録されている学習済モデル３０ｄを取得する。続いて、制御部２０は、物量予測部２１ｂの機能により、天候情報を取得する（ステップＳ１０５）。すなわち、制御部２０は、天候情報サーバ２００から予測対象日の天候、気温等を含む天候情報を取得する。 When the physical quantity prediction process is started, the control unit 20 acquires a machine-learned model using the past physical quantity actual results 30b by the function of the physical quantity prediction unit 21b (step S100). That is, the control unit 20 acquires the learned model 30d recorded on the recording medium 30. Subsequently, the control unit 20 acquires the weather information by the function of the physical quantity prediction unit 21b (step S105). That is, the control unit 20 acquires the weather information including the weather, the temperature, etc. of the prediction target day from the weather information server 200.
　続いて、制御部２０は、物量予測部２１ｂの機能により、カレンダー情報３０ｃを取得する（ステップＳ１１０）。すなわち制御部２０は、カレンダー情報３０ｃを参照し、予測対象日の曜日や平日／休日区分を取得する。続いて、制御部２０は、物量予測部２１ｂの機能により、機械学習モデルを用いて予測対象日の便毎および店舗毎の物量を予測する（ステップＳ１１５）。すなわち制御部２０は、学習済モデル３０ｄに、予測対象日の曜日、予測対象日の平日／休日区分、予測対象日の天候および気温、便、店舗を入力し、予測対象日における当該便および店舗の物量の予測値を取得する。 Subsequently, the control unit 20 acquires the calendar information 30c by the function of the physical quantity prediction unit 21b (step S110). That is, the control unit 20 refers to the calendar information 30c and acquires the day of the week and the weekday / holiday classification of the prediction target day. Subsequently, the control unit 20 predicts the physical quantity for each flight and each store on the prediction target day by using the function of the physical quantity prediction unit 21b (step S115). That is, the control unit 20 inputs the day of the week of the prediction target day, the weekday / holiday classification of the prediction target day, the weather and temperature of the prediction target day, the flight, and the store in the trained model 30d, and the flight and the store on the prediction target day. Get the predicted value of the physical quantity of.
　続いて、制御部２０は、営業時間取得部２１ａの機能により、配送先の店舗の営業時間を取得する（ステップＳ１２０）。すなわち制御部２０は、店舗情報３０ａを参照し、予測対象日における配送先の各店舗の営業時間を取得する。続いて、制御部２０は、営業時間取得部２１ａの機能により、配送先の店舗に営業時間が変更される店舗が有るか否かを判定する（ステップＳ１２５）。すなわち、制御部２０は、予測対象日における店舗の営業時間と、予測対象日より前の店舗の営業時間（機械学習の教師データの採用期間における店舗の営業時間）とを店舗毎に比較し、営業時間が変更になっている店舗が存在するか否かを判定する。 Subsequently, the control unit 20 acquires the business hours of the delivery destination store by the function of the business hours acquisition unit 21a (step S120). That is, the control unit 20 refers to the store information 30a and acquires the business hours of each store of the delivery destination on the prediction target date. Subsequently, the control unit 20 determines whether or not there is a store whose business hours are changed in the delivery destination store by the function of the business time acquisition unit 21a (step S125). That is, the control unit 20 compares the business hours of the store on the prediction target date with the business hours of the stores before the prediction target date (the business hours of the store during the adoption period of the machine learning teacher data) for each store. Determine if there is a store whose business hours have changed.
　ステップＳ１２５において、営業時間が変更される店舗が有ると判定された場合、制御部２０は、物量予測部２１ｂの機能により、配送先の店舗の営業時間の変更に伴う物量変化を物量予測に反映させる（ステップＳ１３０）。すなわち、本実施形態において、制御部２０は、営業時間の変更に物量が影響されることが推定される店舗および便を特定し、当該店舗および便の物量予測値に時間比を乗じることによって、営業時間変更後の当該店舗および便の物量予測値を取得する。 When it is determined in step S125 that there is a store whose business hours are changed, the control unit 20 reflects the change in the physical quantity due to the change in the business hours of the delivery destination store in the physical quantity prediction by the function of the physical quantity prediction unit 21b. (Step S130). That is, in the present embodiment, the control unit 20 identifies a store and a flight in which the quantity is estimated to be affected by the change in business hours, and multiplies the estimated quantity of the store and the flight by a time ratio. Acquire the quantity forecast value of the store and flight after changing the business hours.
　ステップＳ１３０を終了後、または、ステップＳ１２５で営業時間が変更される店舗が有ると判定されなかった場合、制御部２０は、物量予測処理を終了する。その後、制御部２０は、便毎および店舗毎に予測された物量に基づいて、ＶＲＰサーバ３００に配送計画を作成させ、作成された配送計画を取得する。 After completing step S130, or if it is not determined in step S125 that there is a store whose business hours are changed, the control unit 20 ends the physical quantity prediction process. After that, the control unit 20 causes the VRP server 300 to create a delivery plan based on the predicted quantity for each flight and each store, and acquires the created delivery plan.
　（３）他の実施形態：
  以上の実施形態は本発明を実施するための一例であり、他にも種々の実施形態を採用可能である。例えば、物量予測システム１０とＶＲＰサーバ３００は同じ装置で構成されてもよい。また、物量予測システム１０が、複数のシステムで構成されてもよい。例えば、物量予測システム１０の一部の機能が管理者端末１００で実現されても良いし、クラウドサーバで実現されても良い。 (3) Other embodiments:
The above embodiment is an example for carrying out the present invention, and various other embodiments can be adopted. For example, the physical quantity prediction system 10 and the VRP server 300 may be configured by the same device. Further, the physical quantity prediction system 10 may be composed of a plurality of systems. For example, some functions of the physical quantity prediction system 10 may be realized by the administrator terminal 100 or a cloud server.
　さらに、管理者端末１００の利用者は、物流拠点に存在しても良いし、発注元に存在しても良い。さらに、管理者端末１００は、車両に備えられていても良いし、可搬型の端末等であっても良い。また、物量予測システム１０を構成する各部（営業時間取得部２１ａ、物量予測部２１ｂ、配送計画取得部２１ｃ）の少なくとも一部が複数の装置に分かれて存在していても良い。また、上述の実施形態の一部の構成が省略される構成や、処理が変動または省略される構成も想定し得る。 Further, the user of the administrator terminal 100 may exist at the distribution base or at the ordering party. Further, the administrator terminal 100 may be provided in the vehicle, may be a portable terminal, or the like. Further, at least a part of each unit (business hour acquisition unit 21a, physical quantity prediction unit 21b, delivery plan acquisition unit 21c) constituting the physical quantity prediction system 10 may be divided into a plurality of devices and exist. Further, it is possible to assume a configuration in which a part of the configuration of the above-described embodiment is omitted, or a configuration in which the processing is varied or omitted.
　営業時間取得部は、店舗の営業時間の変更を取得することができればよい。営業時間の変更は、営業開始時刻および営業終了時刻の少なくともいずれか一方が変化することによる営業時間の「短縮」や「延長」、営業時間の長さは変化しないが営業開始時刻および営業終了時刻が「シフト」することのいずれであってもよい。営業時間の変更は、管理者が入力する構成であってもよいし、例えば店舗の営業時間を管理するサーバから定期的に自動的に取得する構成であってもよい。また、営業時間が変更される店舗を検出した場合に、物量予測処理が自動的に実行されてもよい。 The business hours acquisition department should be able to acquire changes in the business hours of the store. The change of business hours is the "shortening" or "extension" of business hours due to the change of at least one of the business start time and business end time, and the business start time and business end time do not change the length of business hours. May be any of the "shifts". The change of the business hours may be configured to be input by the administrator, or may be configured to be automatically acquired periodically from a server that manages the business hours of the store, for example. Further, when a store whose business hours are changed is detected, the physical quantity prediction process may be automatically executed.
　物量予測部は、変更前の営業時間での物量実績に基づいて、変更後の営業時間に対応する物量を予測することができればよく、上記実施形態の他にも様々な構成を採用してよい。例えば、複数の店舗にそれぞれ量が異なりうる商品を配送するための配送計画を取得する配送計画取得部を備える場合、便毎および店舗毎の物量を予測する構成を採用するが、配送計画取得部を備えない場合は、必ずしも店舗毎や便毎に物量を予測しなくてもよい。例えばチルド商品の拠点の１日の出荷物量を予測する場合、天候や曜日などの条件が類似する過去の物量実績から、予測対象日の物量を予測し、営業時間変更前の複数店舗の営業時間の和を分母とし営業時間変更後の複数店舗の営業時間の和を分子とする時間比を乗じることで、営業時間が変更される場合の予測対象日１日のチルド商品の出荷量を予測する構成であってもよい。 The physical quantity prediction unit may be able to predict the physical quantity corresponding to the business hours after the change based on the actual physical quantity in the business hours before the change, and may adopt various configurations other than the above-described embodiment. .. For example, when a delivery plan acquisition unit for acquiring a delivery plan for delivering products having different quantities to a plurality of stores is provided, a configuration for predicting the quantity of each flight and each store is adopted, but the delivery plan acquisition unit is adopted. If this is not provided, it is not always necessary to predict the quantity for each store or each flight. For example, when predicting the amount of luggage for one sunrise at a base of chilled products, the amount of goods on the forecast target day is predicted from the past physical quantity results with similar conditions such as weather and days, and the business hours of multiple stores before the change of business hours are By multiplying the time ratio with the sum as the denominator and the sum of the business hours of multiple stores after the business hours as the numerator, the shipment volume of chilled products for the forecast target day when the business hours are changed is predicted. May be.
　また、営業時間変更前の物量の予測には、機械学習モデルを用いられてもよいし、多変量解析等の手法が用いられても良い。基本の物量に、曜日、平日／休日の区分、天候、気温、店舗規模、店舗立地条件、等に応じた係数を乗じて補正することで、予測対象日の物量を予測してもよい。 Further, a machine learning model may be used or a method such as multivariate analysis may be used for predicting the physical quantity before the change of business hours. The quantity of the forecast target day may be predicted by multiplying the basic quantity by a coefficient according to the day of the week, weekday / holiday classification, weather, temperature, store size, store location condition, and the like.
　物量の予測に機械学習モデルが用いられる場合も、機械学習モデルの入力と出力の構成は上記実施形態の他にも種々の構成を採用可能である。店舗毎および便毎に物量を予測する場合、例えば、店舗毎や、店舗および便毎に機械学習モデルが生成されてもよい。機械学習モデルの入力や出力も、上記実施形態の例に限定されない。また、商品の配送に用いられる番重等の運搬用容器として、大きさ等が異なる複数種類の容器が用いられる場合、機械学習モデルの出力としての物量は、種類毎の容器の個数であってもよい。 Even when a machine learning model is used for physical quantity prediction, various configurations can be adopted for the input and output configurations of the machine learning model in addition to the above embodiments. When predicting the quantity for each store and each flight, for example, a machine learning model may be generated for each store or each store and flight. The inputs and outputs of the machine learning model are not limited to the examples of the above embodiments. Further, when a plurality of types of containers having different sizes are used as transport containers for weights and the like used for delivering products, the physical quantity as the output of the machine learning model is the number of containers for each type. May be good.
　物量予測部は、過去に営業時間が変更された他の店舗における物量の変化に基づいて、新たに営業時間が変更される店舗の物量を予測する構成であってもよい。具体的に例えば、過去に営業時間をＸ時間からＹ時間に変更した場合の物量が平均してＱＸからＱＹに変化したことが過去の物量実績から取得できた場合に、時間の変化比と物量の変化比の関係を用いて、新たに営業時間がＸ時間からＺ時間に変更になる場合に物量がｑＸからどのように変化するかを予測する構成を採用してもよい。より具体的には、Ｙ／Ｘ：ＱＹ／ＱＸ＝Ｚ／Ｘ：ｑＺ／ｑＸの関係式からｑＺを算出してもよい。 The physical quantity prediction unit may be configured to predict the physical quantity of a store whose business hours are newly changed based on the change in physical quantity in another store whose business hours have been changed in the past. Specifically, for example, when it can be obtained from the past physical quantity actual that the physical quantity changed from Q X to Q Y on average when the business hours were changed from X hours to Y hours in the past, the change ratio of time. A configuration may be adopted in which the relationship between the physical quantity and the change ratio of the physical quantity is used to predict how the physical quantity will change from q X when the business hours are newly changed from X hours to Z hours. More specifically, q Z may be calculated from the relational expression of Y / X: Q Y / Q X = Z / X: q Z / q X.
　また上記では、まず営業時間変更を考慮しない物量の予測を行い、その予測物量を、営業時間の変更に応じて補正するという２段階の処理を行う構成であるが、その構成に限定されない。例えば、機械学習モデルを用いて過去に営業時間が変更された他の店舗の物量変化に基づく予測を行っても良い。その場合、機械学習モデルの入力として、曜日、平日／休日、天候、最高気温、最低気温、便、店舗規模、店舗の立地環境（オフィス街、住宅街、幹線道路沿い、駅前、等）、店舗周辺人口、営業開始時刻、営業終了時刻を採用し、機械学習モデルの出力として物量を採用してもよい。 Further, in the above, the configuration is such that the quantity is predicted without considering the change in business hours, and the predicted quantity is corrected according to the change in business hours, but the configuration is not limited to that. For example, a machine learning model may be used to make predictions based on changes in the quantity of other stores whose business hours have changed in the past. In that case, as input of the machine learning model, day of the week, weekday / holiday, weather, maximum temperature, minimum temperature, flight, store scale, store location environment (office district, residential area, along the main road, in front of the station, etc.), store Peripheral population, business start time, business end time may be adopted, and physical quantity may be adopted as the output of the machine learning model.
　物量予測部は、１日当たりＮ回の配送（便）のうちの営業時間の変更に物量が影響を受けることが推定される配送（便）について、物量を予測することができる構成であってもよい。営業時間が変更されてこれまで営業時間であった一部の時間帯が非営業時間となる場合、当該時間帯で売れていた商品が売れなくなると推定できる。そのため当該時間帯で売れていた商品の発注量が減少することが推定でき、その結果、配送する便の物量が減少すると推定できる。なお商品が配送された便や売れた時間帯は、配送実績やＰＯＳデータ等から特定可能である。上記実施形態においては、営業時間が短縮される場合に、営業終了時刻の直近の便が、営業時間の変更に物量が影響を受けると推定する構成であったがこれに限定されない。営業時間が短縮されることにより、営業終了時刻の直近の便以外にも、影響を及ぼしうる。 The quantity prediction unit can predict the quantity of deliveries (feces) that are estimated to be affected by changes in business hours out of N deliveries (feces) per day. good. If the business hours are changed and some of the hours that were previously business hours become non-business hours, it can be estimated that the products that were sold during those hours will not sell. Therefore, it can be estimated that the order quantity of the products sold in the time zone decreases, and as a result, the quantity of the delivered flights decreases. The flight to which the product was delivered and the time zone in which the product was sold can be specified from the delivery record, POS data, and the like. In the above embodiment, when the business hours are shortened, it is estimated that the flight immediately before the business end time is affected by the change in the business hours, but the present invention is not limited to this. The shortened business hours can have an impact on flights other than those closest to the closing time.
　例えば、店舗ＳＡにＣ２便で配送される商品のＭ１％は店舗ＳＡにおいてこれまで深夜の時間帯（非営業時間に変更になる時間帯）に売れていた場合に、営業時間変更後は、Ｃ２便のＭ１％は売れないと推定できる。そのため、店舗ＳＡのＣ２便が、営業時間減少の影響を受けると推定することができる。この場合に、物量予測部は、Ｃ２便で店舗ＳＡに配送される商品がＭ１％減少することによる物量（例えば運搬用容器の個数）の減少数を特定し、営業時間変更前の物量から減少数を減算して営業時間変更後の物量を算出することができる。 For example, if M1% of products delivered to store SA by C2 flight have been sold at store SA in the middle of the night (time when it is changed to non-business hours), after the business hours are changed. Can be estimated that M1 % of C2 flights cannot be sold. Therefore, it can be estimated that the C2 flight of the store SA is affected by the decrease in business hours. In this case, the physical quantity prediction unit identifies the number of reductions in the physical quantity (for example, the number of transport containers) due to the M 1 % reduction in the products delivered to the store SA by C2 flight, and the physical quantity before the change in business hours. It is possible to calculate the quantity after changing the business hours by subtracting the decrease number from.
　また、店舗ＳＡにおいて変更後の営業終了時刻の直近のＣ３便で配送される商品のＭ２％はこれまで深夜の時間帯（非営業時間に変更になる時間帯）に売れていなかった（別の時間帯に売れる）場合、店舗ＳＡへのＣ３便の商品のＭ２％は営業時間変更に影響を受けないと推定できる。従って店舗ＳＡへのＣ３便の商品のＭ２％は減少せず、（１００－Ｍ２）％は時間比で減少すると見なし、営業時間変更後の物量を予測してもよい。 In addition, M2 % of the products delivered by C3 flight immediately after the change of business end time at store SA have not been sold in the midnight time zone (time zone when it is changed to non-business hours). If it sells at another time), it can be estimated that M2 % of the products on C3 flight to store SA are not affected by the change in business hours. Therefore, it may be considered that the M2 % of the products of the C3 flight to the store SA does not decrease and the (100- M2 )% decreases with respect to the time ratio, and the quantity after the change of business hours may be predicted.
　また例えば、営業時間が変更されてこれまで非営業時間であった一部の時間帯が営業時間となる場合、当該時間帯で売れることが予測される商品を配送する便の物量が増えると推定できる。そのため、店舗規模や立地条件が類似の他店舗のＰＯＳデータを参照し、当該時間帯に売れる商品を配送する便を決定し（他の便の物量や商品の賞味期限等を考慮して決定されてよい）、当該便の物量が増加すると予測してもよい。 Also, for example, if business hours are changed and some hours that were previously non-business hours become business hours, it is estimated that the quantity of flights that deliver products that are expected to sell in those hours will increase. can. Therefore, by referring to the POS data of other stores with similar store size and location conditions, the flight to deliver the products that sell during the relevant time period is determined (determined in consideration of the quantity of other flights, the expiration date of the product, etc.). You may predict that the quantity of the stool will increase.
　なお、これまで営業時間であった時間帯であり便が割り当てられていた時間帯が、非営業時間に変更になることもあり得る。その場合、当該便で配送されるはずだった商品が配送されないこととなるので、当該便の前後の便において物量が増加することもあり得る。具体的には例えば図２Ａの例にさらに、これまで深夜２：００にチルドの第４便（Ｃ４便）が存在していたとする。しかし、店舗ＳＡにおいて２３：００から６：００までの時間帯が非営業時間に変更になるためこのＣ４便でのチルドの商品を店舗ＳＡは受け取れないこととなる。営業時間変更前の１６：３０のＣ３便の物量をＱｃ３、営業時間変更前の２：００のＣ４便の物量をＱｃ４、営業時間変更後の１６：３０のＣ３便の物量をｘとすると、式（１）の関係にあると見なすことができる。
  （Ｑｃ３＋Ｑｃ４）：１４時間（１７時から７時）
＝ｘ：７時間（１７時から２３時までの６時間＋６時から７時までの１時間）　…（１）
  式（１）より、ｘ＝（Ｑｃ３＋Ｑｃ４）／２となり、Ｑｃ３およびＱｃ４の値によっては営業時間変更前の１６：３０のＣ３便の物量Ｑｃ３よりも変更後の１６：３０のＣ３便の物量ｘの方が増加しうる（例えば２≦Ｑｃ３＜Ｑｃ４の場合、ｘ＞Ｑｃ３となる）。 It is possible that the time zone that was previously business hours and the time zone to which flights were assigned may be changed to non-business hours. In that case, the goods that should have been delivered by the relevant flight will not be delivered, so that the quantity may increase in the flights before and after the relevant flight. Specifically, for example, in the example of FIG. 2A, it is assumed that the fourth chilled flight (C4 flight) has been present at 2:00 midnight. However, since the time zone from 23:00 to 6:00 will be changed to non - business hours at store SA, store SA will not be able to receive chilled products on this C4 flight. The quantity of C3 flight at 16:30 before the change of business hours is Q c3 , the quantity of C4 flight at 2:00 before the change of business hours is Q c4 , and the quantity of C3 flight at 16:30 after the change of business hours is x. Then, it can be considered that there is a relationship of the equation (1).
(Q c3 + Q c4 ): 14 hours (17:00 to 7:00)
= X: 7 hours (6 hours from 17:00 to 23:00 + 1 hour from 6:00 to 7:00) ... (1)
From equation (1), x = (Q c3 + Q c4 ) / 2, and depending on the values of Q c3 and Q c4 , the quantity of C3 flights at 16:30 before the change in business hours is 16:30 after the change from Q c3 . The quantity x of C3 stool can be increased (for example, in the case of 2 ≦ Q c3 <Q c4 , x> Q c3 ).
　さらに、本発明は、プログラムや方法としても適用可能である。また、以上のようなシステム、プログラム、方法は、単独の装置として実現される場合もあれば、車両に備えられる各部と共有の部品を利用して実現される場合もあり、各種の態様を含むものである。例えば、以上のようなシステムで実現される方法、プログラムを提供することが可能である。また、一部がソフトウェアであり一部がハードウェアであったりするなど、適宜、変更可能である。さらに、装置を制御するプログラムの記録媒体としても発明は成立する。むろん、そのソフトウェアの記録媒体は、磁気記録媒体であってもよいし半導体メモリであってもよいし、今後開発されるいかなる記録媒体においても全く同様に考えることができる。 Furthermore, the present invention can also be applied as a program or a method. In addition, the above systems, programs, and methods may be realized as a single device or may be realized by using parts shared with each part provided in the vehicle, including various aspects. It is a program. For example, it is possible to provide a method and a program realized by the above system. In addition, some of them are software and some of them are hardware, so they can be changed as appropriate. Further, the invention is also established as a recording medium for a program for controlling an apparatus. Of course, the recording medium of the software may be a magnetic recording medium or a semiconductor memory, and any recording medium developed in the future can be considered in exactly the same way.
  １０…物量予測システム、２０…制御部、２１…物量予測プログラム、２１ａ…営業時間取得部、２１ｂ…物量予測部、２１ｃ…配送計画取得部、３０…記録媒体、３０ａ…店舗情報、３０ｂ…物量実績、３０ｃ…カレンダー情報、３０ｄ…学習済モデル、４０…通信部、１００…管理者端末、２００…天候情報サーバ、３００…ＶＲＰサーバ 10 ... physical quantity prediction system, 20 ... control unit, 21 ... physical quantity prediction program, 21a ... business hours acquisition unit, 21b ... physical quantity prediction unit, 21c ... delivery plan acquisition unit, 30 ... recording medium, 30a ... store information, 30b ... physical quantity Achievements, 30c ... Calendar information, 30d ... Learned model, 40 ... Communication unit, 100 ... Administrator terminal, 200 ... Weather information server, 300 ... VRP server

Claims (7)

  1. 　店舗の営業時間の変更を取得する営業時間取得部と、
    　変更前の営業時間での物量実績に基づいて、変更後の営業時間に対応する物量を予測する物量予測部と、
    を備える物量予測システム。     The business hours acquisition department that acquires changes in store business hours,
    Based on the actual quantity of goods in the business hours before the change, the quantity forecasting department that predicts the quantity corresponding to the business hours after the change,
    A physical quantity prediction system equipped with.
  2. 　前記物量予測部においては、変更前の営業時間と変更後の営業時間の変化量に基づいて、変更後の営業時間に対応する物量が予測される、
    請求項１に記載の物量予測システム。     In the physical quantity prediction unit, the physical quantity corresponding to the changed business hours is predicted based on the amount of change between the business hours before the change and the business hours after the change.
    The physical quantity prediction system according to claim 1.
  3. 　前記物量予測部においては、店舗毎に時間帯毎の商品の過去の売れた量が取得され、変更後に非営業時間となる時間帯において変更前に売れた商品の量を減算して変更後の営業時間に対応する物量が予測される、
    請求項１または請求項２に記載の物量予測システム。     In the physical quantity prediction unit, the past sold amount of the product for each time zone is acquired for each store, and the quantity of the product sold before the change is subtracted in the non-business hours after the change to be changed. The quantity corresponding to business hours is predicted,
    The physical quantity prediction system according to claim 1 or 2.
  4. 　変更前の営業時間の長さを分母とし変更後の営業時間の長さを分子とする時間比を、変更前の営業時間における物量に乗じた値に基づいて、営業時間変更後の物量が予測される、
    請求項１から請求項３のいずれか一項に記載の物量予測システム。     The quantity after the change of business hours is predicted based on the time ratio with the length of the business hours before the change as the denominator and the length of the business hours after the change as the numerator, multiplied by the quantity in the business hours before the change. Be done,
    The physical quantity prediction system according to any one of claims 1 to 3.
  5. 　前記物量予測部においては、営業時間が変更される店舗を含む複数の店舗毎の物量が予測され、
    　予測された物量に基づいて複数の店舗に商品を配送するための配送計画を取得する配送計画取得部を備える、
    請求項１から請求項４のいずれか一項に記載の物量予測システム。     In the physical quantity prediction unit, the physical quantity for each of a plurality of stores including the store whose business hours are changed is predicted, and the physical quantity is predicted.
    It has a delivery plan acquisition unit that acquires a delivery plan for delivering goods to multiple stores based on the predicted quantity.
    The physical quantity prediction system according to any one of claims 1 to 4.
  6. 　過去に営業時間が変更された他の店舗における物量の変化に基づいて、新たに営業時間が変更される店舗の物量が予測される、
    請求項１から請求項５のいずれか一項に記載の物量予測システム。     Based on changes in the quantity of other stores whose business hours have changed in the past, the quantity of stores whose business hours have changed is predicted.
    The physical quantity prediction system according to any one of claims 1 to 5.
  7. 　営業時間が変更される店舗に商品の配送が１日当たりＮ回（Ｎは１以上）行われる場合に、Ｎ回の配送のうちの営業時間の変更に物量が影響を受けることが推定される配送について、物量が予測される、
    請求項１から請求項６のいずれか一項に記載の物量予測システム。     When products are delivered N times a day (N is 1 or more) to a store whose business hours are changed, it is estimated that the quantity will be affected by the change in business hours out of N times. About, the quantity is predicted,
    The physical quantity prediction system according to any one of claims 1 to 6.
PCT/JP2021/023518 2020-11-12 2021-06-22 Product quantity prediction system WO2022102160A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2022561268A JPWO2022102160A1 (en) 2020-11-12 2021-06-22

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2020188542 2020-11-12
JP2020-188542 2020-11-12

Publications (1)

Publication Number Publication Date
WO2022102160A1 true WO2022102160A1 (en) 2022-05-19

Family

ID=81600912

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2021/023518 WO2022102160A1 (en) 2020-11-12 2021-06-22 Product quantity prediction system

Country Status (2)

Country Link
JP (1) JPWO2022102160A1 (en)
WO (1) WO2022102160A1 (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002024350A (en) * 2000-07-03 2002-01-25 Kasumi Co Ltd Retail store managing system
JP2002297958A (en) * 2001-03-29 2002-10-11 Fujitsu Ltd Daily delivery goods ordering optimization support system, its method and program
JP2003256019A (en) * 2002-03-01 2003-09-10 Ishida Co Ltd Production instructing system, production instructing method, and production instructing program
JP2004152205A (en) * 2002-11-01 2004-05-27 Hitachi Ltd Forecast value evaluation method, system and program

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002024350A (en) * 2000-07-03 2002-01-25 Kasumi Co Ltd Retail store managing system
JP2002297958A (en) * 2001-03-29 2002-10-11 Fujitsu Ltd Daily delivery goods ordering optimization support system, its method and program
JP2003256019A (en) * 2002-03-01 2003-09-10 Ishida Co Ltd Production instructing system, production instructing method, and production instructing program
JP2004152205A (en) * 2002-11-01 2004-05-27 Hitachi Ltd Forecast value evaluation method, system and program

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
"BI (Business Intelligence) Revolution; 1st edition", 19 November 2009, NTT PUBLISHING CO., LTD., JP, ISBN: 978-4-7571-2246-8, article BUSINESS INTELLIGENCE PROMOTION CENTER OF DATA TECHNOLOGY DEVELOPMENT HEADQUARTERS OF NTT DATA CORPORATION: "2.1 CPM and BPM in the Mai chain", pages: 87 - 96, XP009537327 *

Also Published As

Publication number Publication date
JPWO2022102160A1 (en) 2022-05-19

Similar Documents

Publication Publication Date Title
US10776750B2 (en) System and method for fulfilling e-commerce orders from a hierarchy of fulfilment centres
Silbermayr et al. A multiple sourcing inventory model under disruption risk
US20170185933A1 (en) Distribution-Independent Inventory Approach under Multiple Service Level Targets
Rana et al. Dynamic pricing policies for interdependent perishable products or services using reinforcement learning
CN111295681A (en) Demand prediction using a weighted hybrid machine learning model
US8700443B1 (en) Supply risk detection
US20050288993A1 (en) Demand planning with event-based forecasting
US20110225023A1 (en) Prioritized Product Distribution
US20190258979A1 (en) Method and system for transfer order management
US9466043B1 (en) System and method for generating shipment forecasts for materials handling facilities
US20180308039A1 (en) System and Method for Dynamically Establishing A Regional Distribution Center Truck Flow Graph to Distribute Merchandise
JPH096862A (en) Promise management system
US20210224732A1 (en) Distribution-Independent Inventory Approach under Multiple Service Level Targets
US20200320467A1 (en) Computerized inventory redistribution control system
EP1521197A1 (en) Projected sales
CN115699057A (en) Short life cycle sales curve estimation
WO2022102160A1 (en) Product quantity prediction system
JP7308073B2 (en) Logistics management system
JPH0798741A (en) Integrated management method for production sales article
Ulmer et al. Dynamic service area sizing in urban delivery
Kocer et al. A production queueing-inventory system with two-customer and a server subject to breakdown
EP1530143A2 (en) Determination of best transportation guidelines
WO2023056115A1 (en) Optimized tree ensemble based demand model
REMBULAN et al. An Integrated Model of Continuous Review Inventory and Vehicle Routing Problem with Time Windows
Bašljan et al. Prediction of perishable goods deliveries by GRU neural networks for reduction of logistics costs

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21891408

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022561268

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21891408

Country of ref document: EP

Kind code of ref document: A1