JP6530559B2 - Optimization system and method - Google Patents

Description

  The present invention relates to an optimization system and an optimization method for optimizing distribution warehouse operations.

  In distribution warehouses, distribution centers, and other product supply centers, products manufactured by the supplier on the supply side are received, temporarily stored at a predetermined location in the distribution warehouse, and distributed according to the order of the shipping destination on the demand side. The goods are collected from a predetermined place in the warehouse and delivered to the shipping destination. Also, before delivery, distribution processing such as assortment and decoration of collected products and packaging in a form suitable for delivery such as cardboard may be performed.

  In the distribution warehouse business, it is desirable to be able to receive, store, and ship more products at a lower cost in this arrival → storage → collection → distribution → processing → packaging → delivery cycle. The costs referred to here are, for example, the cost for space management such as storage management and land costs necessary for storing goods, and the cost for work such as labor cost necessary for receiving and shipping goods. In order to increase the cost effectiveness, it is necessary to improve the working efficiency of various tasks.

  As for improvement of work efficiency in the distribution warehouse, there has been an effort to optimize in particular work. For example, in Patent Document 1 below, in order to reduce the space cost at the time of product storage and the operation cost at the time of product shipment, the equation model for calculating the space cost and the shipping operation cost from the product arrangement information and the product demand information Disclosed is the use of a formula model that has been created and created to perform optimization for determining the product supply frequency that minimizes costs.

  Further, Patent Document 2 inputs the number of workers in the distribution warehouse, the arrangement of goods, the warehouse layout, and the equipment used, simulates the movement of the worker at the time of work, and calculates the total work time of all the workers. Create a simulation model and disclose, for example, optimization of the number of workers in each work process.

  Further, Patent Document 3 can acquire and reuse knowledge indicating what kind of work should be performed in a certain situation from the history of work in the past, thereby improving the efficiency of the subsequent work. Disclose a work procedure management system that can

  As described above, when the distribution work is optimized, a model for reproducing the work in the warehouse is created in advance according to the purpose of the optimization. For example, a mathematical expression model that takes various attributes (for example, the number of workers, product placement) related to warehouse work as input and calculates expected values or predicted values of objective variables (for example, working hours) or the like as described above There is a simulation model. After creating the model, the objective variable is calculated from the attributes of the warehouse operation based on these models, and the desired attribute of the warehouse operation that maximizes or minimizes the value of the objective variable is determined, and the optimization result is can get. Thus, the development of optimization is roughly divided into an optimization model creation step of calculating the value of the objective variable from the work attribute, and which operation attribute is changed in order to maximize or minimize the objective variable. And a step of determining.

JP, 2010-61260, A JP 2002-269192 A Japanese Patent Application Laid-Open No. 10-105540

  Heretofore, models used for work optimization are individually developed by data analysts according to the purpose of optimization desired to be realized. At the time of model creation, we will learn the work in the field, carry out interviews on issues in the field, etc., and perform trial and error of function system and parameters to calculate the objective variable from the attribute of the warehouse work. However, in such individual development, each optimization point of view "how to calculate the objective variable from the work attribute, which work attribute to change in realizing the maximization or minimization of the objective variable" In addition, model redevelopment is necessary.

  In addition, it is necessary to redesign the parameters of the model each time the work environment in the warehouse changes, such as the change of workers and goods, and the change of the arrangement of goods. Furthermore, in optimization development based on work knowledge and task hearings in the field, the viewpoint of optimization is limited to the scope of task awareness in the field. Therefore, they can not be taken into consideration when there is a truly effective optimization perspective outside of consciousness.

  The present invention aims to automatically generate an optimization model without trial and error of individual model generation and parameter tuning depending on the purpose of optimization desired to be realized.

  An optimization system according to an aspect of the invention disclosed in the present application includes a processor that executes a program, and a storage device that stores the program, and the storage device stores an operation result in a distribution warehouse operation. The work record storage information includes a work variable indicating work results of the distribution warehouse work as the work results, an explanatory variable indicating work attributes of the distribution warehouse work, and the distribution warehouse work And an order variable indicating a work order of each of the distribution warehouse operations, and the processor assigns a first assignment indicating whether or not each combination of values in different explanatory variables is present for the operation results of a certain distribution warehouse operation. A first allocation feature quantity generation process for generating a feature quantity; a first allocation feature quantity generated by the first allocation feature quantity generation process; And variables, based on, and executes a and optimization model generation process of generating an optimization model to optimize the objective variables.

  An optimization system according to another aspect of the invention disclosed in the present application includes a processor that executes a program, and a storage device that stores the program, and the storage device stores the operation results in the distribution warehouse operation. The work record storage information includes, as the work record, a target variable indicating a work result of the distribution warehouse work, an explanatory variable indicating a work attribute of the distribution warehouse work, and the distribution warehouse An order variable indicating an operation sequence of operations is stored for each distribution warehouse operation, and the processor stores a series of distribution warehouses in a continuous section including the distribution warehouse operation for the operation results of a certain distribution warehouse operation A first order in which an explanatory variable group in work is acquired from the work record storage information, and a variation in order of the explanatory variable group in the continuous section is defined. An order feature amount generation process for generating a feature amount; and a second allocation feature amount for classifying the first order feature amount generated by the order feature amount generation process into any of a plurality of ranges An optimization model for optimizing the objective variable based on the allocation feature quantity generation process of 2, the second allocation feature quantity generated by the second allocation feature quantity generation process, and the objective variable And generating an optimization model generation process.

  According to a representative embodiment of the present invention, an optimization model can be automatically generated without trial and error of individual model generation and parameter tuning according to the purpose of optimization desired to be realized. Problems, configurations, and effects other than those described above will be clarified by the description of the following embodiments.

FIG. 1 is a block diagram illustrating an example of a system configuration of a physical distribution system according to a first embodiment. FIG. 2 is a block diagram showing an example of the hardware configuration of the optimization system. FIG. 3 is an explanatory diagram of an example of contents stored in the work result table. FIG. 4 is an explanatory diagram of an example of contents stored in the work instruction table. FIG. 5 is an explanatory view showing an example of the stored contents of the operation pattern table. FIG. 6 is a block diagram of a functional configuration example of the optimization system according to the first embodiment. FIG. 7 is an explanatory diagram of an example of generation of the order feature quantity. FIG. 8 is an explanatory diagram of an example of generation of the allocation feature amount. FIG. 9 is an explanatory view of the labeling process. FIG. 10 is an explanatory view showing an example of calculation of the degree of importance. FIG. 11 is an explanatory diagram of an example of generation of an optimization model. FIG. 12 is an explanatory diagram of a determination example 1 of the replacement variable and the replacement unit. FIG. 13 is an explanatory diagram of a determination example 2 of the replacement variable and the replacement unit. FIG. 14 is an explanatory diagram of an example of replacement of explanatory variables. FIG. 15 is an explanatory view showing a generation example of the work instruction feature quantity and a calculation example of the evaluation value. FIG. 16 is a flowchart illustrating a detailed processing procedure example of the optimization processing by the optimization processing unit. FIG. 17 is an explanatory diagram of an example of storage contents of the evaluation result table. FIG. 18 is a block diagram of a functional configuration example of the optimization system according to the second embodiment. FIG. 19 is an explanatory view of a presentation example 1 by the presentation unit. FIG. 20 is an explanatory diagram of a presentation example 2 by the presentation unit.

  In the following embodiments, when it is necessary for the sake of convenience, it will be described divided into a plurality of sections or embodiments, but unless otherwise specified, they are not mutually unrelated, one is the other Some or all of the variations, details, supplementary explanations, etc. are in a relation. In addition, in the following embodiments, when referring to the number of elements (including, for example, the number, numerical value, amount, range), unless explicitly stated otherwise and in principle when clearly limited to a specific number, It is not limited to the specific number, and may be more or less than the specific number.

  Furthermore, in the following embodiments, it is needless to say that the constituent elements (including, for example, element steps) are not necessarily essential unless specifically stated and considered to be obviously essential in principle. Yes. Similarly, in the following embodiments, when referring to the shapes and positional relationships of components, the shapes or the like are substantially similar to the shapes or the like unless specifically stated otherwise and in principle clearly considered otherwise. It shall include similar ones. The same applies to the above numerical values and ranges.

  In optimization, other explanatory variables are changed to maximize or minimize an objective variable. In the following, the minimization of the objective variable will be described as a main example, but even the maximization of the objective variable can be realized by appropriately replacing the relevant determination conditions (for example, determination in replacement adoption determination step S1606 described later) conditions). Further, in the present embodiment, the “system” is one computer or a plurality of computers communicatively linked.

<System configuration example of distribution system>
FIG. 1 is a block diagram illustrating an example of a system configuration of a physical distribution system according to a first embodiment. The physical distribution system 100 includes a management system 101, an optimization system 102, a work result table 111, a work instruction table 112, and an optimization work instruction table 113. In the following description, although the management system 101 and the optimization system 102 are described separately, they do not necessarily have to be separately configured, and the functions of each of the management system 101 and the optimization system 102 on one system are described. It may be configured as a subsystem having

  The management system 101 is a system that manages distribution warehouse work. Specifically, for example, the management system 101 accesses the work result table 111 and the work instruction table 112, and the work results of various work (for example, arrival, shipment, collection, distribution processing, packaging) related to the distribution warehouse Manage work orders. The management system 101 may also manage information on the arrangement of goods and the attributes (for example, weight) of goods. The management system 101 issues a work instruction with reference to the optimization work instruction table 113. The management system 101 is, for example, a warehouse management system (WMS).

  The optimization system 102 is a system that optimizes the physical distribution work instruction. Specifically, for example, with reference to the work result table 111, the optimization system 102 optimizes the work instruction which is an entry of the work instruction table 112, and generates the optimization work instruction table 113. After generating the optimization work instruction table 113, the optimization system 102 optimizes the work instruction which is an entry of the optimization work instruction table 113, and updates the optimization work instruction table 113.

  The work instruction table 112 is work instruction storage information that stores a work instruction as an entry. Specifically, for example, in the case of a work instruction related to a receiving operation, the work instruction table 112 is created by the management system 101 based on the product arrival information from the product supplier to the distribution warehouse. Further, in the case of the work instruction regarding the shipping work, the work instruction table 112 is created by the management system 101 based on the product ordering information from the product demander to the distribution warehouse. Thus, each entry in the work instruction table 112 is, for example, a work instruction for confirming what kind of product, who, how many, and which supplier.

  The work record table 111 is work record storage information that stores a work record as an entry. Specifically, for example, in the case of work results related to arrival work, the work result table 111 is created by the management system 101 based on the results of goods arrival work performed based on the goods arrival information from the product supplier to the distribution warehouse. Be done. Further, in the case of the work instruction regarding the shipping work, the work instruction table 112 is created by the management system 101 based on the results of the product ordering work performed based on the product ordering information from the product demander to the distribution warehouse. Thus, each entry of the work result table 111 is, for example, a work result indicating what kind of product, who, how many, and which supplier.

  In the work record collection 121, the management system 101 records the work record performed by the worker based on the work instruction issued by the management system 101 as an entry in the work record table 111. Specifically, for example, the work results are transmitted to the management system 101 by the operator operating the handy terminal, or transmitted to the management system 101 from various sensors mounted on the facilities in the warehouse. It is recorded in 111.

  In the work instruction issue 122, the work instruction accumulated in the optimization work instruction table 113 is issued by the management system 101 to the worker of the warehouse. Specifically, for example, printing of a work instruction to the worker, display on the handy terminal of the worker, and display on other facilities in the warehouse are performed. As described above, in the physical distribution system 100, the optimized work instruction can be given to the worker, and the working efficiency can be improved.

<Hardware Configuration Example of Optimization System 102>
FIG. 2 is a block diagram showing an example of the hardware configuration of the optimization system 102. As shown in FIG. The optimization system 102 includes a processor 201, a storage device 202, an input device 203, an output device 204, and a communication interface (communication IF 205). The processor 201, the storage device 202, the input device 203, the output device 204, and the communication IF 205 are connected by a bus. The processor 201 controls the optimization system 102. The storage device 202 is a work area of the processor 201. Also, the storage device 202 is a non-temporary or temporary storage medium that stores various programs and data. Examples of the storage device 202 include a read only memory (ROM), a random access memory (RAM), a hard disk drive (HDD), and a flash memory. The input device 203 inputs data. Examples of the input device 203 include a keyboard, a mouse, a touch panel, a ten key, and a scanner. The output device 204 outputs data. Examples of the output device 204 include a display and a printer. The communication IF 205 connects to the network to transmit and receive data.

<Operation result table and work instruction table>
FIG. 3 is an explanatory view showing an example of the stored contents of the work result table 111, and FIG. 4 is an explanatory view showing an example of the stored contents of the work instruction table 112. As shown in FIG. The work result table 111 and the work instruction table 112 show, as an example, a work result of shipping work and a work instruction example. Each row of the work result table 111 is an entry indicating the work result related to the shipping work. Each row of the work instruction table 112 is an entry indicating a work instruction related to the shipping work. Each column of the work result table 111 and the work instruction table 112 shows columns related to shipping work.

  The work result table 111 and the work instruction table 112 have, as common columns, a column related to the work order and a column related to the work attribute. The column relating to the work order is a storage area that defines the work order, and has a work group number column 300 and a work number column 301. The work group number column 300 is a storage area for storing a work group number for each entry. The work number column 301 is a storage area for storing a work number for each entry. The work number is a number that defines the order of work. In this example, work is performed in ascending order of work numbers. The work group field number is a number that defines the order of work groups when a series of work (for example, the work groups identified by work numbers 1 to 4) is one unit. In this example, work groups are executed in ascending order of work group numbers.

  The columns relating to work attributes are storage areas that define work attributes, and worker ID column 302, product ID column 303, product weight column 304, work location column 305, X coordinate column 306, Y coordinate And a column 307. The worker ID column 302 is a storage area for storing a worker ID for each entry. The worker ID is identification information that uniquely identifies the worker. The product ID column 303 is a storage area for storing a product ID for each entry. The item ID is identification information that uniquely identifies an item.

  The item weight column 304 is a storage area for storing the item weight for each entry. The item weight is information indicating the weight of the item identified by the item ID. The work place column 305 is a storage area for storing a work place for each entry. The work place is a place in the distribution warehouse where the worker specified by the worker ID works on the product specified by the product ID. The X coordinate column 306 and the Y coordinate column 307 are attached attribute columns of the work place column 305. The X-coordinate column 306 and the Y-coordinate column 307 are storage areas for storing X-coordinate values and Y-coordinate values specifying the work place in the distribution warehouse for each entry.

  In addition, the work result table 111 has a column related to the work result. The column related to the work result is a storage area that defines the work result, and includes a work time column 308 and a work time column 309. The work time column 308 is a storage area for storing work time for each entry. The work time column 309 is a storage area for storing the work start time (or end time) for each entry.

  In each of the columns 300 to 309 in the work result table 111 and the work instruction table 112, an axis attribute 311 and a type 312 are set. The axis attribute 311 defines an attribute of a column. The axis attribute 311 of the column related to the work order is "order" indicating the work order, the axis attribute 311 of the column related to the work result is "purpose" indicating the objective variable, and the axis attribute 311 of the column related to the work attribute is It is "explanation" which shows an explanatory variable. The objective variable is a variable that you want to minimize at optimization time. The type 312 also defines the type of information stored in the columns 300-309. The type 312 includes “numeric value”, “time”, and “character”.

<Operation pattern table>
FIG. 5 is an explanatory view showing an example of the stored contents of the operation pattern table. The operation pattern table 500 is a table that defines the operation 502 and the viewpoint 503 for each of the types 501 of the columns 300 to 309. The operation 502 is information that defines the method of calculating the order feature amount for the information stored in the columns 300 to 309 identified by the type 501. The viewpoint 503 is information indicating the purpose for generating the order feature of the operation 502. The order feature quantity is a feature quantity that characterizes the work order, and is generated by the operation 502. In addition, the same type 501 has one or more operations 502. When there are a plurality of types of operations 502, which type 501 to use may be associated in advance with the columns 300 to 309 of the work result table 111. Further, when generating the feature amount, the specific operation 502 may be selected by the operation of the operator.

<Functional Configuration Example of Optimization System 102>
FIG. 6 is a block diagram of a functional configuration example of the optimization system 102 according to the first embodiment. The optimization system 102 includes an order feature quantity generation unit 601, an allocation feature quantity generation unit 602, a calculation unit 603, a selection unit 604, an optimization model generation unit 605, a determination unit 606, and an optimization processing unit 607. And. Specifically, these are realized, for example, by causing the processor 201 to execute a program stored in the storage device 202. The flow of optimization processing by the optimization system 102 follows the arrow connecting the order feature quantity generation unit 601 to the optimization processing unit 607 shown in FIG.

  Further, the optimization system 102 stores an operation pattern table 500, a work result feature amount table 611, a model feature amount table 612, and an optimization model M. Specifically, these are realized by, for example, information stored in the storage device 202. The operation pattern table 500 is a master table, and the work result feature amount table 611, the model feature amount table 612, and the optimization model M are intermediately generated data.

  The order feature quantity generation unit 601 generates an order feature quantity of the explanatory variable with reference to the work result table 111 and the calculation pattern table 500. Specifically, for example, the order feature quantity generation unit 601 generates an explanatory variable group in a series of distribution warehouse work in a continuous section including a certain distribution warehouse work with respect to the operation results of a certain distribution warehouse work in the operation results table 111 , And generates an order feature that defines the variation of the order of the explanatory variable group in the continuous section. More specifically, for example, the ordinal feature quantity generation unit 601 specifies, from the work result table 111, a time-series entry group including an entry that is a generation source of the ordinal feature quantity. The order feature quantity generating unit 601 refers to the operation pattern table 500 to specify the operation 502 of the explanatory variable group of the specified entry group. The order feature quantity generation unit 601 calculates an order feature quantity by the specified operation 502.

  FIG. 7 is an explanatory diagram of an example of generation of the order feature quantity. FIG. 7 illustrates an example in which the order feature quantity is generated with the value of the work attribute surrounded by the thick frame of the work result table 111 as a generation source and stored in the work result table 111. Here, the operation 502 of the type “character” in the work place column 305 is “UniqNum”. “Uniq Num” is an operation method of counting the number of occurrences of the generator in the target section. The target section of operation 502 is the value of the time series work attribute including the generator. That is, the order feature quantity is a feature quantity indicating how much the value of the work attribute is included in the target section.

  In this example, the value of the work attribute from “C03” three entries before the generation source 701 in the entry E21 to “B02” one entry after the generation source 701 is set as a target section (arrows in FIG. 7). display). The target section is assumed to be set in advance. As the order feature amount is larger, it indicates that there are more tasks at the work location “B02” that is the generation source 701 in the target section. In this case, the work place “B02”, which is the generation source 701, appears three times in the target section. Therefore, the order feature quantity 711 of the generation source 701 is “3”. The order feature quantity generation unit 601 stores the order feature quantity 711 in the UniqueNum work place column 750.

  In addition, the operation 502 of the type “numerical value” of the X coordinate column 306 is “Var” and “Max”. “Var” is an arithmetic method of calculating the variance value of the generation source in the target section. “Max” is an arithmetic method for obtaining the maximum value of the generator in the target section. The Var which is the order feature quantity 721 of the generation source 702 is the variance value “134” of the X coordinate values “60”, “40”, “30”, “50”, and “30” in the target section. Similarly, Max which is the order feature quantity 722 of the generation source 702 is the maximum value “60” of the X coordinate values “60”, “40”, “30”, “50”, and “30” in the target section. The order feature quantity generation unit 601 stores the order feature quantities 721 and 722 in the VarX coordinate column 761 and the MaxX coordinate column 762. In this manner, the sequential feature quantities related to the value of the work attribute are held in the work result feature quantity table 611.

  Returning to FIG. 6, the allocation feature quantity generation unit 602 generates an allocation feature quantity with reference to the work result table 111 and the work result feature quantity table 611. The allocated feature quantity is a feature quantity to be assigned to the work result. The allocation feature quantity includes a feature quantity (first allocation feature quantity) when the type 312 is an explanatory variable that is a character and a feature quantity (second allocation feature quantity) when the type 312 is a numerical value.

  The first allocation feature quantity is a feature quantity allocated to the combination of the explanatory variables in the work result table 111, that is, an allocation feature quantity indicating the presence or absence of a combination of values in different explanatory variables. Further, the second allocated feature value is a feature value to be assigned according to the size of the quantity of the explanatory variable in the work result table 111 or the work result feature value table 611, that is, one of a plurality of order feature values. It is an allocation feature quantity for classifying into.

  FIG. 8 is an explanatory diagram of an example of generation of the allocation feature amount. Here, a set of columns in which the first allocation feature quantity and the second allocation feature quantity are stored is referred to as a work record feature quantity column. Specifically, for example, worker ID [WK01] × work place [A01] column 801 to worker ID [WK03] × work place [A03] column 809, worker ID × work result feature amount column related to work place It is called 800. Further, the VarX coordinate [low] column 811 to the VarX coordinate [high] column 813 are referred to as a work result feature amount column 810 related to the VarX coordinate.

  First, the first allocation feature amount will be described. The combination of the worker ID 32 and the work place 35 in the entry E11 of the work result table 111 is “WK01” and “A01”. Therefore, in the worker ID [WK01] × working place [A01] column 801 in the entry corresponding to the entry E11 in the work result feature quantity table 611, “1” indicating the presence of the combination is the first allocated feature quantity It is set as 81. In this entry, “0”, which indicates the absence of the combination, is the worker performance feature amount column 800 regarding the worker ID × worker location other than worker ID [WK01] × worker location [A01] column 801. It is set as an assigned feature value of 1 (see worker ID [WK03] × work place [A03] column 809).

  Next, the second allocation feature quantity will be described. The VarX coordinate 76 in the entry of the work result feature quantity table 611 corresponding to the entry E24 is “153”. “153” is classified as a high value in the VarX coordinate column 761. In this case, “1” indicating the presence of “153” of the VarX coordinate 76 is set as the second allocated feature value 83 in the VarX coordinate [high] column 813. In the entry, “0” indicating the absence of the VarX coordinate 76 is set as the second allocation feature for the work result feature quantity column 810 regarding the VarX coordinate other than the VarX coordinate [high] column 813 in the entry. (See VarX coordinate column 811.) Here, the generation process of the allocation feature amount will be described more specifically using the labeling process.

  FIG. 9 is an explanatory view of the labeling process. The allocated feature quantity generation unit 602 generates an intermediate table 900 in which each value or range of values of the explanatory variable is a column. When the type of the explanatory variable is “character” in the entry of the intermediate table 900 corresponding to the work result table 111, the allocated feature amount generation unit 602 of the intermediate table 900 corresponding to the value of the explanatory variable in the work result table 111. In the column, the value “1” indicating that the explanatory variable exists is set.

  For example, for the worker ID “WK01” surrounded by a thick frame, the assigned feature amount generation unit 602 sets “1” in the worker ID [WK01] column 911 in the corresponding entry of the intermediate table 900, and performs the operation In the person ID [WK02] column 912 and the worker ID [WK03] column 913, “0” is set.

  Further, when the type of the explanatory variable is “numerical value” in the entry of the intermediate table 900 corresponding to the work result table 111, the allocated feature quantity generation unit 602 corresponds to the range of the value of the explanatory variable in the work result table 111. In the column of the intermediate table 900, a value "1" indicating that the explanatory variable exists is set. The allocated feature quantity generating unit 602, for example, for the item weights “300”, “125”, “125”, and “120” surrounded by a bold frame, the item weight in the corresponding entry in the intermediate table 900 is 600 kg. Item weight [g] [low] (-600) column 921 indicating low is set to “1”, item weight [g] [mid] (600-1300) column 922, item weight [g] [0] is set in the [high] (1300-) column 923.

  As a result, the value set in the intermediate table 900 becomes a logical value of “1” or “0”. The allocated feature quantity generating unit 602 calculates the logical product of the combination of explanatory variables using the intermediate table 900, and the calculated logical product value is stored in the column defining the combination of the explanatory variables in the work record characteristic quantity table 611. Set

  Returning to FIG. 6, the calculation unit 603 calculates the degree of importance in the work result feature amount table 611. Specifically, for example, when the first allocation feature quantity is generated for a plurality of types of combinations, for example, the first allocation feature quantity and the target variable are calculated for each of the plurality of types of combinations. Based on the importance of the combination is calculated. In addition, the calculating unit 603 determines, for each of the plurality of types of order feature quantities, the importance regarding the order feature quantity that is the generation source of the second allocation feature quantity based on the second allocation feature quantity and the target variable. Calculate

  The importance is a quantity that represents the importance and contribution of the work result feature in each work result feature column in describing the objective variable of the work result feature table 611. As the importance, for example, for each work record feature quantity column, the work record feature quantity is used as an explanatory variable, and a determination coefficient when multiple regression analysis is performed on a target variable (work time) is used.

  The coefficient of determination is an index indicating whether the obtained multiple regression equation is true, and the larger the value, the better. The square of the multiple correlation coefficient obtained by the multiple regression analysis is the determination coefficient. As another method, analysis of variance may be performed using the value of the objective variable when the work performance feature quantity in each work performance feature quantity column is “1”, and the fluctuation ratio of the variance analysis may be used. In addition to the above, it does not depend on the operation result feature amount in each operation result feature amount column, as long as the amount by which the objective variable can be explained can be determined quantitatively.

  FIG. 10 is an explanatory view showing an example of calculation of the degree of importance. Here, the case of using multiple regression analysis will be described. The objective variable is set as the work record feature quantity in the work time column 308, and the explanatory variable is set as the worker record × the work record feature quantity (first allocated feature quantity) in the work record feature quantity column 800 related to the work place. The calculating unit 603 obtains the determination coefficient “0.3” by solving the equation for the number of entries using multiple regression analysis. Further, the objective variable is set as the work record feature quantity in the work time column 308, and the explanatory variable is set as the work record feature quantity (second allocated feature quantity) in the work record feature quantity column 810 regarding VarX coordinates. The calculating unit 603 obtains the determination coefficient “0.2” by solving the equation for the number of entries using multiple regression analysis.

  Returning to FIG. 6, the selection unit 604 selects the work result feature quantity used to generate the optimization model M based on the degree of importance calculated by the calculation unit 603. Specifically, for example, the selection unit 604 holds a threshold of importance. For example, the threshold value of the determination coefficient which is the degree of importance is 0.3. In the case of the example of FIG. 10, the selecting unit 604 uses the work record feature quantity of the work record feature quantity column 800 for the worker ID × the work place having the determination coefficient equal to or more than the threshold value to generate the optimization model M Select as the actual feature value.

  In addition, the selection unit 604 may select the work result feature quantity of the top n work result feature quantity columns with high importance. Thus, the selecting unit 604 generates the model feature amount table 612 from the work result feature amount table 611. The model feature amount table 612 is a table obtained by removing the work result feature amounts not selected by the selection unit 604 from the work result feature amount table 611.

  The optimization model generation unit 605 generates the optimization model M using the model feature amount table 612. Specifically, for example, similarly to the calculation unit 603, the relationship between the objective variable and the explanatory variable group is executed by performing multiple regression analysis with the work record feature amount as an explanatory variable and using the objective variable (working time) as the objective variable. The multiple regression equation is generated as the optimization model M. In addition to multiple regression analysis, an optimization model M may be generated by using a neural network. Other than this, the method is not limited to these as long as it is a method of obtaining a relational expression between the objective variable and the explanatory variable.

  FIG. 11 is an explanatory diagram of an example of generation of the optimization model M. FIG. 11 shows an example of generating the optimization model M by multiple regression analysis. Thereby, the values of ai and b shown in FIG. 11 are determined. N is the number of entries of the model feature amount table 612, and i is an arbitrary integer of 1 or more and N or less.

  In the case where the optimization model generation unit 605 generates the optimization model M by the multiple regression equation, if the calculation unit 603 has already obtained the multiple regression equation when obtaining the determination coefficient as the importance, the multiple regression The formula may be diverted. Thereby, the speed of generation of the optimization model M can be increased.

  Returning to FIG. 6, the determination unit 606 accesses the work instruction table 112 and determines replacement variables from the explanatory variables of the work instruction table 112 based on the work result feature amount column used to generate the optimization model M. And determine the replacement unit from the ordinal variable. The replacement variable is a variable (value in the column) to be replaced at the time of optimization processing by the optimization processing unit 607. A replacement unit is a variable that represents a group when replacing a replacement variable, and sections in which variables of the replacement unit are continuously the same value are always replaced as one group.

  Specifically, for example, the determination unit 606 is a column group (for example, the work result feature amount column 800) that defines a combination of explanatory variables that are different in the work result feature amount column used to generate the optimization model M. In this case, one of the explanatory variables constituting the combination is determined as a replacement variable. Further, in this case, the determination unit 606 determines the work group number column 300 as a replacement unit. In addition, if the work performance feature quantity column used to generate the optimization model M is a column group (for example, work performance feature quantity column 810) that defines the same explanatory variables, the determination unit 606 determines the work number column 301 Decide on exchange variables and exchange units.

  Thus, by determining the replacement variable and the replacement unit, the optimization processing unit 607 can obtain the optimized work instruction by using the optimization model M in a state where the explanatory variables are exchanged.

  FIG. 12 is an explanatory diagram of a determination example 1 of the replacement variable and the replacement unit. The example in FIG. 12 is a determination example in the case where the work result feature quantity column used to generate the optimization model M is a column group that defines a combination of different explanatory variables. Here, the work result feature quantity column used to generate the optimization model M is assumed to be the work result feature quantity column 800 for the worker ID × work place used to generate the optimization model M in FIG.

  In columns 300 to 307 of the work instruction table 112, the columns of the allocated feature quantities constituting the combination of the work result feature quantity column 800 related to the worker ID × the work place are the worker ID column 302 and the work place column 305. The determination unit 606 determines one of the explanatory variables in the worker ID column 302 and the work place column 305 as a replacement variable. For example, the determination unit 606 preferentially determines an explanatory variable including a variable corresponding to an order variable that is a replacement unit as a replacement variable.

  For example, when the replacement unit is the work group number column 300, the work group number "1" corresponds to the worker ID "WK01", the work group number "2" corresponds to the worker ID "WK03", and the work group The number "3" corresponds to the worker ID "WK02". On the other hand, in the case of the work group number “1”, the work place is “A01”, “C03”, and “B02”, which is not supported. Therefore, the determination unit 606 determines the worker ID column 302 as a replacement variable. When none of the explanatory variables correspond to the work group number column 300, the determination unit 606 randomly determines any explanatory variable as a replacement variable.

  FIG. 13 is an explanatory diagram of a determination example 2 of the replacement variable and the replacement unit. The example of FIG. 13 is a determination example in the case where the work result feature quantity column used to generate the optimization model M is a column group that defines the same explanatory variables. Here, the work result feature quantity column used to generate the optimization model M is assumed to be the work result feature quantity column 810 regarding the VarX coordinate. The work result feature amount column 810 related to the VarX coordinate is a set of the VarX coordinate [low] column 811, the VarX coordinate [mid] column 812, and the VarX coordinate [high] column 813 which are the same type. Therefore, the explanatory variable is only the VarX coordinate column 306. Therefore, the determination unit 606 determines the work number column 301 as a replacement variable and a replacement unit.

  Returning to FIG. 6, the optimization processing unit 607 replaces the replacement variables in the work instruction table 112 according to the replacement unit, and executes the optimization process using the work instruction table 112 after replacement. Specifically, for example, the optimization processing unit 607 generates a work instruction feature amount table using the work instruction table 112 after replacement. The work instruction feature amount table is a table obtained by converting an explanatory variable which is a work instruction of the work instruction table 112 into an allocation feature amount.

  The specific conversion method is the same as the generation example of the allocated feature quantity described in FIGS. More specifically, when replacement is determined as shown in FIG. 12, it is the same as the example of generation of the first allocation feature shown in FIGS. 8 and 9, and replacement is determined as shown in FIG. If it is, it is the same as the generation example of the second allocation feature quantity shown in FIGS. That is, in FIGS. 7 to 9, although the work result feature amount table 611 including the allocation feature amount is generated from the work result table 111, in the optimization processing unit 607, the generation source is not the work result table 111 but the work instruction table 112. Then, in the same manner as in FIGS. 7 to 9, the work instruction feature quantity table including the allocation feature quantity is generated from the work instruction table 112.

  Then, the optimization processing unit 607 calculates the prediction variable for each entry by giving the allocation feature amount in the work result feature amount column to the optimization model M. The predictor variable is a predictor value of the objective variable. Finally, the optimization processing unit 607 calculates a statistical value of each calculated prediction variable, and uses it as an evaluation value. The statistical value is a statistical representative value such as the mean value, maximum value, minimum value, median value, and total value of the prediction variable. That is, the evaluation value is a predictive variable, that is, a value indicating the feature of the predictive value of the objective variable.

  FIG. 14 is an explanatory diagram of an example of replacement of explanatory variables. FIG. 14 shows an example of replacement using the replacement variable and the replacement unit determined in the determination example 1 of FIG. For example, worker IDs “WK01” of three consecutive entries whose work group number is “1” and worker IDs “WK03” of two consecutive entries whose work group number is “2”; Is replaced. In the exchange variable, the value of the exchange source and the value of the exchange destination are randomly determined. Since the work place column 305 is not a replacement variable, replacement is not performed. As a result of this replacement, the worker IDs of three consecutive entries whose work group number is "1" are all replaced with "WK03", and the work of two consecutive entries whose work group number is "2" All the person IDs are replaced with "WK01".

  FIG. 15 is an explanatory view showing a generation example of the work instruction feature quantity and a calculation example of the evaluation value. (A) shows the work instruction table 112 after replacement shown in FIG. (B) shows a combination of explanatory variables and generation of assigned feature quantities in the case of using the worker ID and the work place, that is, a generation example of the work instruction feature quantity table 1500. As described above, the work instruction feature quantity table 1500 is generated by the process as shown in FIGS. 8 and 9. Here, a work having a work instruction feature amount column 1501 regarding worker ID x work place including worker ID [WK01] x work place [A01] column 801 to worker ID [WK03] x work place [A03] column 809 An indicated feature amount table 1500 is generated.

  (C) shows the example of calculation of an evaluation value. The evaluation value is a statistical value of a predictor variable. Here, the working time column 308 designated as the objective variable in the work result table 111 is used as a prediction variable. The optimization processing unit 607 gives the assignment feature amount of the work instruction feature amount column 1501 related to the worker location x work place in the work instruction feature amount table 1500 to the optimization model M, thereby making an entry of the work instruction feature amount table 1500 Calculate the working time that is the forecast variable of each Then, the optimization processing unit 607 calculates a statistical value of the calculated working time, and uses it as an evaluation value.

  FIG. 16 is a flowchart illustrating a detailed processing procedure example of the optimization processing by the optimization processing unit 607. The optimization processing unit 607 first sets the index i of the number of times of replacement to i = 0 (step S1601). Next, the optimization processing unit 607 increments the index i (step S1602). Then, the optimization processing unit 607 executes the work instruction replacement process (step S1603).

  The work instruction replacement process (step S1603) is, for example, the process shown in FIG. In the work instruction replacement process (step S1603), the combination of the replacement source value and the replacement destination value in the replacement variable is held, and the work instruction replacement process (step S1603) is performed so as not to be the same combination from next time. Run. This allows more replacement patterns to be covered.

  Next, the optimization processing unit 607 executes feature quantity restoration processing (step S1604). The feature quantity restoration process (step S1604) is a process for restoring the allocated feature quantity in the work instruction table 112 after replacement, and more specifically, for example, the work instruction feature quantity table 1500 shown in FIG. Is a process of generating

  Next, the optimization processing unit 607 executes an evaluation value calculation process (step S1605). In the evaluation value calculation process (step S1605), a prediction variable is calculated by giving the allocation feature amounts of the work instruction feature amount table 1500 to the optimization model M, and the statistical value is calculated as an evaluation value. Specifically, for example, the process shown in FIG.

  Thereafter, the optimization processing unit 607 determines whether the replacement in the index i is appropriate based on the evaluation value (step S1606). Specifically, for example, the optimization processing unit 607 compares the evaluation value before replacement with the evaluation value after replacement, and determines whether the evaluation value after replacement is smaller than the evaluation value before replacement. Do. The optimization processing unit 607 acquires the evaluation value before replacement from the evaluation result table 1700 described later with reference to FIG. When the index i is i = 1, an evaluation value before replacement does not exist, so an evaluation value sufficiently large as an evaluation value is set as an initial value in the entry where the replacement count of the evaluation result table 1700 is zero.

  Further, in this example, the prediction variable is the work time column 308, and the evaluation value becomes higher as the work time is shorter. Therefore, in step S1606, it is determined whether the evaluation value after replacement is smaller than the evaluation value before replacement, and if smaller, replacement is determined as appropriate. On the other hand, for example, in the case of an evaluation value where the evaluation is higher as the prediction variable is larger, it is determined whether the evaluation value before replacement is larger than the evaluation value after replacement. As described above, the determination processing in step S1606 may be the determination content according to the nature of the prediction variable.

  If it is determined that replacement is appropriate (step S1606: YES), the optimization processing unit 607 determines adoption of replacement in the work instruction replacement process (step S1603) in the index i (step S1607), and step S1610 Migrate to

  On the other hand, when it is determined that replacement is not appropriate (step S1606: No), the optimization processing unit 607 determines not to adopt replacement in the work instruction replacement process (step S1603) in the index i (step S1608), The work instruction table 112 is returned to the state before the replacement (step S1609), and the process proceeds to step S1610.

  In step S1610, the optimization processing unit 607 sets the evaluation result as the new entry in the evaluation result table 1700 as the new entry or not (step S1607 or S1608) whether or not to adopt the replacement with the index i replacement count. It registers (step S1610).

  Then, the optimization processing unit 607 performs the termination determination of the optimization processing (step S1611). If it is determined that the process does not end (step S1611: NO), the process returns to step S1602, and the optimization processing unit 607 increments the index i (step S1602). On the other hand, when it is determined that the process is ended (Step S1611: Yes), the optimization processing unit 607 saves the work instruction table 112 in the current exchange state as the optimization work instruction table 113 (Step S1612). Finish the conversion process.

  Here, the end determination (step S1611) will be specifically described. The end determination (step S1611) is a process of determining whether the replacement number has been executed a sufficient number of times. For example, the optimization processing unit 607 compares the evaluation value at the initial stage, for example, index i = 1 with the latest evaluation value to calculate the reduction rate. The reduction rate is calculated by the latest evaluation value / the initial evaluation value. If the reduction rate is equal to or less than the threshold th1, the optimization processing unit 607 determines that the optimization processing is to be ended.

  In addition, when the index i reaches the threshold th2, the optimization processing unit 607 may determine to end the optimization processing. In addition, when the latest evaluation value reaches the target value, the optimization processing unit 607 may determine to end the optimization processing. Further, when there is no more new replacement pattern, the optimization processing unit 607 may determine to end the optimization processing. In addition, when a predetermined time has elapsed since the start of the optimization process, the optimization processing unit 607 may determine that the optimization process is to end.

  FIG. 17 is an explanatory diagram of an example of storage contents of the evaluation result table. The evaluation result table 1700 includes a replacement number column 1701, a replacement determination result column 1702, and an evaluation value column 1703. The replacement count column 1701 is a storage area for storing an index i for each entry. The initial value is "0". The replacement determination result column 1702 is a storage area for storing whether or not to adopt the number of replacements of the index i (step S1607 or S1608). There is no initial value. The evaluation value column 1703 is a storage area for storing the evaluation value obtained in the evaluation calculation process (step S1605) with the number of times of input of the index i. The initial value is set to a sufficiently large value.

  In this example, as the number of times of replacement increases, the evaluation value decreases, so that it is possible to detect appropriate replacement by repeatedly executing the optimization process. That is, the optimization processing unit 607 stores, as the optimization work instruction table 113, the work instruction table 112 indicating the replacement state when the evaluation value becomes the minimum. Therefore, it is possible to provide an operator with an optimization operation instruction that reduces the operation time more than the operation instruction before optimization.

  As described above, according to the first embodiment, the optimization model M can be automatically generated without trial and error of individual optimization model M generation and parameter tuning according to the purpose of optimization desired to be realized. This can reduce the cost of generating the optimization model M manually.

  In addition, it is possible to support efficient work improvement by setting model attributes with a truly effective optimization viewpoint and setting work attributes for optimization, without staying within the scope of task awareness in the field. It becomes. In addition, maintenance costs can be reduced by automatically updating the optimization model M and the parameters in the optimization model M in accordance with changes in the work environment. That is, when the work environment changes, the optimization model M and parameters in the optimization model M can be automatically updated only by changing the contents of the work instruction table 112, and maintenance costs can be reduced. .

  The second embodiment is an example in which the optimization viewpoint in optimizing the work instruction is presented to the worker prior to the generation of the optimization model M. This presentation process is executed before the operation result feature amount to be used for generating the optimization model M is selected by the selection unit 604 after the calculation unit 603 calculates the importance. By presenting the optimization viewpoint before generating the optimization model M, the operator can select which combination of the combination of explanatory variables in the work instruction feature quantity column is valid and which combination is invalid. It can be confirmed. In the second embodiment, the presentation process is mainly described, and the other processes are the same as the first embodiment, and thus the description thereof is omitted.

<Functional Configuration Example of Optimization System 102>
FIG. 18 is a block diagram of a functional configuration example of the optimization system 102 according to the second embodiment. The optimization system 102 has a presentation unit 1800 in addition to the configuration described in the first embodiment. Specifically, the presentation unit 1800 is realized, for example, by causing the processor 201 to execute a program stored in the storage device 202.

  The presentation unit 1800 generates the optimization viewpoint and presents it to the worker by outputting the optimization viewpoint to the output device 204 (for example, a display). Specifically, for example, with respect to each feature amount group in the work result feature amount table 611, the presentation unit 1800 presents to the operator what optimization viewpoint they can have. The presentation method of the optimization viewpoint differs depending on whether the feature amount group to be presented is a combination of a plurality of different assignment feature amounts or a plurality of same type assignment feature amounts.

  FIG. 19 is an explanatory diagram of a presentation example 1 by the presentation unit 1800. The presentation example 1 shown in FIG. 19 is an example of presenting an optimization viewpoint in the case where the work result feature quantity to be presented is a combination of a plurality of different allocation feature quantities. Specifically, for example, the presentation unit 1800 determines that the assigned feature quantity is a combination of “worker ID” and “working place”. The first optimization viewpoint presentation information 1900 is generated as a statistical feature based on the amount) and the target variable.

  In this case, the presentation unit 1800 prepares a table in which the two explanatory variables constituting the allocation feature amount are the vertical axis and the horizontal axis. The presentation unit 1800 obtains the correlation coefficient between the value of the objective variable and the value of each feature amount in the entry indicating each work result of the work result feature amount table 611, and stores the correlation coefficient in the cell of the corresponding table. Thereby, the first optimization viewpoint presentation information 1900 is generated. In each cell constituting the matrix, the numerical value on the left is the calculated correlation coefficient, and the numerical value in parentheses on the right is the number of entries used for calculation of the correlation coefficient.

  For example, in the upper left cell, the presentation unit 1800 has 42 explanatory variables in the work result feature quantity column 800 for the worker ID [WK01] × work place [A01], and 42 purposes of the work time column 308 Using variables, it is shown that the correlation coefficient "0.4" is obtained by multiple regression analysis. The correlation coefficient indicates that the value of the objective variable increases when each work record in the work record feature quantity table 611 matches the feature quantity, that is, “worker of worker ID: WK01 works When working at the place [A01], the working time [s] tends to be large. In the same way, it also indicates that “When the worker with worker ID: WK01 works at the work place [C03], the work time [s] tends to be smaller” (correlation coefficient is − 0.3).

  Further, the correlation coefficient is not limited as long as the method can determine whether the value of the objective variable increases or decreases when each work record matches a certain feature amount. For example, it may be the average value of the objective variable in the case where each feature value is matched.

  In addition, the presentation unit 1800 extracts the maximum value and the minimum value of the correlation coefficient from one sequence in the vertical direction or the horizontal direction of the first optimization viewpoint presentation information 1900. Then, the presentation unit 1800 generates tendency information indicating a tendency of change of the target variable due to mutual assignment between two explanatory variables constituting the allocation feature quantity.

  Specifically, for example, by referring to the vertical column of the worker [WK01], the presentation unit 1800 refers to “the worker [WK01], the work place [C03] is ○ (mean of OK, and so on) And the work location [A01] is x (meaning NG, the same applies hereinafter) to generate first trend information 1901. In addition, the presentation unit 1800 refers to the vertical column of the worker [WK03], and for the “worker [WK03], the working place [C03] is ○ and the working place [A01] is x”. To generate trend information 1902 of

  Thus, by presenting the trend information 1901 and 1902 to the worker, the worker can see the presented trend information and grasp the relationship between the explanatory variables at the current time. Specifically, for example, in the case of the above-described two pieces of trend information 1901 and 1902, “worker [WK01] and worker [WK03] are respectively not good at work locations. The worker [WK01] is good at the work place [A01] and is not good at the work place [C03]. The worker [WK03] is the opposite. And so on.

  Furthermore, the presentation unit 1800 extracts a combination in which cells having relatively large correlation coefficients and small cells are paired with each other in a plurality of columns in the first optimization viewpoint presentation information 1900. An implementation plan for optimization can be presented. Specifically, for example, in the case of the two trend information 1901 and 1902, the trend regarding the work place of the first trend information 1901 and the trend regarding the work place of the second trend information 1902 are just opposite.

  Therefore, for the worker [WK01], the presentation unit 1800 reassigns the work location [A01] assigned to the worker [WK01] to [C03], and conversely, to the worker [WK03] The work location [C03] may be reassigned to [A01]. Can be generated and presented.

  FIG. 20 is an explanatory diagram of a presentation example 2 by the presentation unit 1800. The presentation example 2 shown in FIG. 20 is an example of presenting an optimization viewpoint in the case where the work result feature quantity to be presented is a plurality of same kind of allocation feature quantities. Specifically, for example, the presentation unit 1800 performs each work record regarding “VarX coordinate [low]”, “VarX coordinate [mid]”, and “VarX coordinate [high]” whose allocated feature value is “VarX coordinate”. The second optimization viewpoint presentation information 2000 is generated as a statistical feature based on the feature (the second assigned feature) and the objective variable.

  In this case, the presentation unit 1800 prepares a table in which an explanatory variable, which is a plurality of same kind of allocated feature quantities, is taken as a horizontal axis. The presentation unit 1800 obtains a correlation coefficient between the value of the target variable and the value of each allocation feature amount in the work result feature amount table 611, and stores the correlation coefficient in the cell of the corresponding table. Thereby, the second optimization viewpoint presentation information 2000 is generated. In each cell, the value on the left is the calculated correlation coefficient, and the value in parentheses on the right is the number of entries used for calculation of the correlation coefficient.

  For example, the second optimization viewpoint presentation information 2000 indicates that “the variance (variation) of the X coordinates is [low] (small) tends to reduce the objective variable”. Also, the presentation unit 1800 extracts the maximum value and the minimum value of the correlation coefficient from the second optimization viewpoint presentation information 2000. Then, the presentation unit 1800 generates tendency information 2001 indicating a tendency of change of the target variable due to mutual assignment among the three explanatory variables constituting the plurality of same type of allocation feature quantities.

  For example, the presentation unit 1800 can present, to the operator, tendency information 2001 indicating a change in the target variable due to the order relationship of the work order direction, such as “If x variation is large, x if small variation”. The worker sees the presented trend information 2001 and grasps the work prediction that “variations in X coordinates, that is, the work time will be shortened if the work places are arranged closer to each other compared to the work before and after”. be able to. In addition, it is possible to grasp the work prediction that “the work order should be interchanged so that the variation before and after the X coordinate is reduced”.

  In addition, after the selecting unit 604 presents the optimization viewpoint to the worker by the presenting unit 1800, the selecting unit 604 may receive the optimization viewpoint selected by the selection input from the worker. Selection of the optimization viewpoint is made by the input device.

  According to the second embodiment, the optimization system can present the operator with an easy-to-understand optimization viewpoint prior to the optimization process. In this way, the operator's intention can be reflected and optimized.

  As described above, according to this embodiment, the optimization model M can be automatically generated without performing trial and error of individual model generation and parameter tuning according to the purpose of optimization desired to be realized. This can reduce the cost of generating the optimization model M manually.

  In addition, when the first allocation feature quantity is generated for a plurality of types of combinations, a statistical contribution to an objective variable can be obtained by calculating the level of importance for each of the plurality of types of combinations. Therefore, by comparing the degree of importance, the combination to be optimized can be easily identified.

  Also, by presenting the viewpoint of optimization to the worker in an easy-to-understand manner, the worker can use the management system 101 with a sense of satisfaction with the optimization viewpoint devised by the optimization system 102.

  In addition, by optimizing the work instruction table 112 by replacing the values in the explanatory variables, the optimization target in the optimization model M having the truly effective optimization viewpoint without staying within the scope of the task awareness of the site. It is possible to set the work attribute of Therefore, it is possible to support efficient work improvement by using the work instruction after optimization.

  Further, the optimization process may be executed again using the optimization work instruction table 113 according to the change of the work environment. Thereby, parameters in the optimization model M and the optimization model M can be automatically updated, and maintenance costs can be reduced.

  Also, in the above-described embodiment, an example of generating the optimization model M using the first allocation feature and the second allocation feature has been described, but the first allocation feature and the second allocation feature The optimization model M may be generated using a quantity and / or a quantity.

  The present invention is not limited to the embodiments described above, and includes various modifications and equivalent configurations within the scope of the appended claims. For example, the embodiments described above are described in detail to explain the present invention in an easy-to-understand manner, and the present invention is not necessarily limited to those having all the configurations described. Also, part of the configuration of one embodiment may be replaced with the configuration of another embodiment. In addition, the configuration of another embodiment may be added to the configuration of one embodiment. In addition, another configuration may be added to, deleted from, or replaced with part of the configuration of each embodiment.

  In addition, each configuration, function, processing unit, processing means, etc. described above may be realized by hardware, for example, by designing part or all of them with an integrated circuit, etc., and the processor realizes the respective functions. It may be realized by software by interpreting and executing the program to

  Information such as programs, tables, and files for realizing each function can be stored in a memory, a hard disk, a storage device such as a solid state drive (SSD), or a recording medium such as an IC card, an SD card, or a DVD.

  Further, control lines and information lines indicate what is considered to be necessary for explanation, and not all control lines and information lines necessary for mounting are shown. In practice, it can be considered that almost all configurations are mutually connected.

Claims (15)

A processor that executes a program, and a storage device that stores the program;
The storage device has operation result storage information for storing operation results in a distribution warehouse operation, and the operation result storage information is an objective variable indicating the operation result of the distribution warehouse operation as the operation results, and the distribution warehouse An explanatory variable indicating a work attribute of a work and an order variable indicating a work order of the distribution warehouse work are stored for each distribution warehouse work,
The processor is
A first allocation feature quantity generation process for generating a first allocation feature quantity indicating presence or absence of a combination of values in different explanatory variables for an operation result of a certain distribution warehouse operation;
An optimization model generation process of generating an optimization model for optimizing the objective variable based on the first allocation feature quantity generated by the first allocation feature quantity generation process and the objective variable;
An optimization system characterized by performing. The optimization system according to claim 1, wherein
The processor is
When the first allocation feature quantity is generated for a plurality of types of combinations, the target variable is generated based on the first allocation feature quantity and the objective variable for each of the plurality of types of combinations. Calculation processing for calculating the importance of the combination to explain
Performing a selection process of selecting the first allocation feature quantity based on the respective degrees of importance calculated by the calculation process;
In the optimization model generation process, the processor generates an optimization model that optimizes the objective variable based on the first allocation feature value selected by the selection process and the objective variable. An optimization system characterized by The optimization system according to claim 1, wherein
The processor is
For each of the combinations of values in the different explanatory variables, a statistical feature based on the first assignment feature indicating the presence or absence of the combination of the values and the target variable is calculated and output to the output device An optimization system characterized by performing presentation processing. The optimization system according to claim 1, wherein
The processor is
With regard to the operation results of the certain distribution warehouse operation, an explanatory variable group in a series of distribution warehouse operations in the continuous section including the certain distribution warehouse operation is acquired from the operation result storage information, and An order feature quantity generation process for generating a first order feature quantity that defines variation in order of an explanatory variable group;
A second allocation feature quantity generation process for generating a second allocation feature quantity for classifying the first order feature quantity generated by the order feature quantity generation process into any of a plurality of ranges;
In the optimization model generation process, the processor is based on the first allocation feature quantity or a second allocation feature quantity generated by the second allocation feature quantity generation process, and the target variable. An optimization system characterized by generating an optimization model for optimizing the objective variable. The optimization system according to claim 4, wherein
The processor is
A first calculation process of calculating a first importance degree regarding the combination based on the first allocation characteristic amount and the objective variable;
A second order related to the first order feature, which is a generation source of the second assigned feature, for describing the target variable based on the second assigned feature and the objective variable. A second calculation process for calculating the degree of importance;
The first allocated feature value or the second allocated feature value is determined based on the first importance calculated by the first calculation process and the second importance calculated by the second calculation process. Execute a selection process of selecting one of the allocation feature quantities.
In the optimization model generation process, the processor generates an optimization model for optimizing the objective variable based on the allocation feature value selected by the selection process and the objective variable. Optimization system. The optimization system according to claim 1, wherein
The storage device has work instruction storage information for storing a work instruction in the physical distribution warehouse work, and the work instruction storage information is an explanatory variable indicating a work attribute of the physical distribution warehouse work as the work instruction, and the physical distribution It stores an order variable indicating the work order of the warehouse work for each distribution warehouse work,
The processor is
The work instruction storage information is obtained by replacing the values in any one of the combinations of the different explanatory variables that are the generation sources of the first allocation feature amount among the explanatory variable groups in the work instruction storage information. Run optimization process to optimize
In the optimization processing, the processor generates, for the work instruction of the work instruction storage information after replacement, a third allocation feature that indicates the presence or absence of the combination of the different explanatory variables, and the third allocation feature. The evaluation value indicating the feature of the objective variable in the work instruction storage information after replacement is calculated based on the objective variable in the work instruction storage information and the optimization model, and the evaluation value is calculated. Optimizing the work instruction storage information on the basis of. A processor that executes a program, and a storage device that stores the program;
The storage device has operation result storage information for storing operation results in a distribution warehouse operation, and the operation result storage information is an objective variable indicating the operation result of the distribution warehouse operation as the operation results, and the distribution warehouse An explanatory variable indicating a work attribute of a work and an order variable indicating a work order of the distribution warehouse work are stored for each distribution warehouse work,
The processor is
Regarding the operation results of a certain distribution warehouse operation, an explanatory variable group in a series of distribution warehouse operations in a continuous section including the certain distribution warehouse operation is acquired from the operation result storage information, and the explanation in the continuous intervals An order feature quantity generation process for generating a first order feature quantity that defines variation in order of variable groups;
A second allocation feature quantity generation process for generating a second allocation feature quantity for classifying the first order feature quantity generated by the order feature quantity generation process into any of a plurality of ranges;
Optimization model generation processing for generating an optimization model for optimizing the objective variable based on the second allocation feature amount generated by the second allocation feature amount generation processing and the objective variable.
An optimization system characterized by performing. The optimization system according to claim 7, wherein
When the second allocation feature quantity is generated for a plurality of types of first order feature quantities, the second allocation feature quantity and the purpose for each of the plurality of first order feature quantities A calculation process for calculating the importance of the first order feature that is a generation source of the second allocation feature based on a variable;
Performing a selection process of selecting the second allocation feature amount based on the respective degrees of importance calculated by the calculation process;
In the optimization model generation process, the processor generates an optimization model that optimizes the objective variable based on the second allocation feature value selected by the selection process and the objective variable. An optimization system characterized by The optimization system according to claim 7, wherein
The processor is
For each of the second allocation feature quantities in the plurality of ranges corresponding to the first order feature quantity, a statistical feature quantity is calculated based on the second allocation feature quantity and the objective variable. An optimization system characterized by performing presentation processing to output to an output device. The optimization system according to claim 7, wherein
The storage device has work instruction storage information for storing a work instruction in the physical distribution warehouse work, and the work instruction storage information is an explanatory variable indicating a work attribute of the physical distribution warehouse work as the work instruction, and the physical distribution It stores an order variable indicating the work order of the warehouse work for each distribution warehouse work,
The processor is
Performing an optimization process of optimizing the work instruction storage information by replacing values in an explanatory variable of a generation source of the first order feature amount in the work instruction storage information;
In the optimization processing, the processor generates a second order feature amount for a work instruction of the work instruction storage information after replacement, and classifies the second order feature amount into any one of the plurality of ranges. Relates to the work instruction storage information after replacement based on the fourth allocation feature amount, the target variable in the work instruction storage information, and the optimization model. An optimization system comprising: calculating an evaluation value; and optimizing the work instruction storage information based on the evaluation value. The optimization system according to claim 1 or 7, wherein
The optimization system characterized in that the distribution warehouse operation is a collection operation in a distribution warehouse, an arrival operation, a distribution processing operation, or a packaging operation. An optimization method executed by an optimization system having a processor that executes a program and a storage device that stores the program.
The storage device has operation result storage information for storing operation results in a distribution warehouse operation, and the operation result storage information is an objective variable indicating the operation result of the distribution warehouse operation as the operation results, and the distribution warehouse An explanatory variable indicating a work attribute of a work and an order variable indicating a work order of the distribution warehouse work are stored for each distribution warehouse work,
The optimization method is
The processor is
A first allocation feature quantity generation process for generating a first allocation feature quantity indicating presence or absence of a combination of values in different explanatory variables for an operation result of a certain distribution warehouse operation;
An optimization model generation process of generating an optimization model for optimizing the objective variable based on the allocation feature quantity generated by the first allocation feature quantity generation process and the objective variable;
An optimization method characterized by performing. The optimization method according to claim 12, wherein
The processor is
When the first allocation feature quantity is generated for a plurality of types of combinations, the combination is related to the combination based on the first allocation feature quantity and the objective variable for each of the plurality of types of combinations Calculation processing to calculate the degree of importance,
Performing a selection process of selecting the first allocation feature quantity based on the respective degrees of importance calculated by the calculation process;
In the optimization model generation process, the processor generates an optimization model that optimizes the objective variable based on the first allocation feature value selected by the selection process and the objective variable. An optimization method characterized by The optimization method according to claim 12, wherein
The processor is
For each of the combinations of values in the different explanatory variables, a statistical feature based on the first assignment feature indicating the presence or absence of the combination of the values and the target variable is calculated and output to the output device An optimization method characterized by executing a presentation process. The optimization method according to claim 12, wherein
The storage device has work instruction storage information for storing a work instruction in the physical distribution warehouse work, and the work instruction storage information is an explanatory variable indicating a work attribute of the physical distribution warehouse work as the work instruction, and the physical distribution It stores an order variable indicating the work order of the warehouse work for each distribution warehouse work,
The optimization method is
The processor is
The work instruction storage information is obtained by replacing the values in any one of the combinations of the different explanatory variables that are the generation sources of the first allocation feature amount among the explanatory variable groups in the work instruction storage information. Run optimization process to optimize
In the optimization processing, the processor generates, for the work instruction of the work instruction storage information after replacement, a third allocation feature that indicates the presence or absence of the combination of the different explanatory variables, and the third allocation feature. An evaluation value regarding the work instruction storage information after replacement is calculated based on the objective variable in the work instruction storage information and the optimization model, and the work instruction storage is calculated based on the evaluation value. An optimization method characterized by optimizing information.

Images