Mediation system, mediation method, and program

Abstract

To provide a mediation system capable of generating a mediation plan taking into account changes in the behavior of each system to multiple individual systems in which each individual determines their own behavior independently.SOLUTION: Each of multiple individual systems 2 determines its own behavior independently to improve individual key performance indicators. A mediation system 1 includes: a model generation module 107 that generates a simulation model 102 that estimates the behavior of each individual system 2 and a learning model 104 that estimates behavioral changes of each individual system 2 based on time series changes in the simulation model 102; and a mediation search module 110 that generates a mediation plan to propose an assignment among individual systems 2 of certain factors involved in determining the behavior of individual system 2 using the simulation model 102 and the learning model 104 and proposes the mediation plan to the individual systems 2.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an arbitration system, an arbitration method, and a program, and is suitable for application to an arbitration system, an arbitration method, and a program that generate arbitration proposals for a plurality of independent individual systems. [Background technology]

[0002] Large social systems such as supply chains and mobility systems are composed of multiple independent systems, each of which acts to achieve its own objectives. When each system tries to achieve its own objectives, it can sometimes result in a state that is not optimal overall. This can lead to a decline in the functionality of the system or a decline in the profitability of each individual system. In such situations, an external mediator is needed to guide the system toward optimal overall optimization through the sharing of information and suggestions or instructions on actions.

[0003] On the other hand, many systems nowadays incorporate artificial intelligence and machine learning technologies, and these systems change their subsequent behavior by learning from mediation and behavioral experiences.In systems incorporating such artificial intelligence and machine learning technologies, in order to guide the behavior of each system toward overall optimization, it is necessary to generate mediation proposals that take into account the behavioral changes of individual systems.

[0004] As a technique for mediating to coordinate multiple individual systems, for example, Patent Document 1 discloses a behavior rule generation method including: a behavior rule pattern generation step for generating multiple patterns of behavior rule information that associate a first company that makes up a supply chain with behavior logic that corresponds to the state of a second company that makes up the supply chain of the first company; a simulation step for calculating key performance indicators for the entire supply chain for each of the multiple patterns of behavior rule information; and an optimal behavior rule selection step for selecting the behavior rule information to be applied to the supply chain based on the key performance indicators. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-207017 Summary of the Invention [Problem to be solved by the invention]

[0006] In a system in which multiple independent systems (individual systems) act to maximize their individual key performance indicators (KPIs), one possible method for guiding the behavior of each individual system to maximize an overall KPI that differs from the individual KPIs is to deploy an arbitration system that presents arbitration proposals outside the individual systems. It is assumed that the individual systems will learn as a result of arbitration by the arbitration system, and their subsequent behavioral patterns will change. However, existing technologies, including the behavior rule generation method described in Patent Document 1, do not take into account that each individual system may change as a result of the presentation of arbitration proposals. As a result, the arbitration system may not be able to guide the multiple individual systems into an appropriate cooperative relationship.

[0007] The present invention has been made in consideration of the above points, and aims to propose an arbitration system, an arbitration method, and a program that are capable of generating arbitration proposals that take into account changes in the behavior of each individual system, for multiple individual systems that each independently determine their own behavior. [Means for solving the problem]

[0008] In order to solve such problems, the present invention provides an arbitration system that presents arbitration proposals related to the behavior of a plurality of individual systems, each of which independently determines its own behavior so as to increase its individual key performance indicators, and is characterized by comprising: a model generation module that generates a simulation model that estimates the behavior of each of the individual systems and a learning model that estimates changes in the behavior of each of the individual systems based on time-series changes in the simulation model; and a arbitration proposal search module that uses the simulation model and the learning model to generate arbitration proposals that propose the transfer of specified factors related to the behavioral decisions between the individual systems and presents the arbitration proposals to the individual systems.

[0009] In order to solve the above problem, the present invention provides an arbitration method using an arbitration system that presents arbitration proposals related to the behavior of a plurality of individual systems, each of which independently determines its own behavior so as to increase its individual key performance indicators, the arbitration method comprising: a model generation step in which the arbitration system generates a simulation model that estimates the behavior of each of the individual systems and a learning model that estimates changes in the behavior of each of the individual systems based on time-series changes in the simulation model; and a arbitration proposal search step in which the arbitration system uses the simulation model and the learning model to generate arbitration proposals that propose the transfer of predetermined factors related to the behavior determination between the individual systems and present the arbitration proposals to the individual systems.

[0010] In addition, in order to solve such problems, the present invention provides a program executed in an arbitration system that presents arbitration proposals related to the behavior of a plurality of individual systems, each of which independently determines its own behavior so as to increase its individual key performance indicators, the program comprising: a model generation process that generates a simulation model that estimates the behavior of each of the individual systems and a learning model that estimates changes in the behavior of each of the individual systems based on time-series changes in the simulation model; and a arbitration proposal generation process that uses the simulation model and the learning model to generate an arbitration proposal that proposes the transfer of specified factors related to the behavioral determination between the individual systems and outputs the arbitration proposal to the individual systems. [Effects of the Invention]

[0011] According to the present invention, it is possible to generate an arbitration plan that takes into account changes in the behavior of each individual system for a plurality of individual systems that each determine their own behavior independently. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a block diagram showing an example of the configuration of an entire system including an arbitration system 1 according to an embodiment of the present invention. [Figure 2] 10 is a flowchart illustrating an example of a processing procedure of an arbitration system process. [Figure 3] 10 is a flowchart showing an example of a processing procedure of individual system processing. [Figure 4] FIG. 2 is a diagram showing an example of data stored in a simulation model database 101. [Figure 5] FIG. 2 is a diagram showing an example of data stored in a learning model database 103. [Figure 6] 10 is a flowchart illustrating an example of a processing procedure for mediation proposal generation processing. [Figure 7] 10 is a flowchart illustrating an example of a processing procedure for mediation proposal evaluation processing. [Figure 8] 10 is a graph showing an example of the structure of production costs in an individual system 2. [Figure 9] 10 is a graph illustrating a specific example of data stored in the simulation model database 101. [Figure 10] 10 is a graph for explaining an example of an arbitration proposal for the production status in an individual system 2. [Figure 11] 10 is a graph for explaining another example of an arbitration proposal for the production status in the individual system 2. DETAILED DESCRIPTION OF THE INVENTION

[0013] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS In the following description, a program causes a processor to execute a predetermined process, but for the sake of convenience, the program may be described as the executing entity.

[0014] Note that the following description and drawings are examples for explaining the present invention, and have been omitted or simplified as appropriate for clarity of explanation. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention. The present invention is not limited to the embodiments, and all application examples consistent with the concept of the present invention are included in the technical scope of the present invention. Those skilled in the art can make various additions and modifications to the present invention within the scope of the present invention. The present invention can also be implemented in various other forms. Unless otherwise specified, each component may be plural or singular.

[0015] In the following explanation, various types of information may be described using expressions such as "table," "list," "queue," etc., but the various types of information may also be expressed using data structures other than these. To indicate that it is not dependent on the data structure, "XX table," "XX list," etc. may be referred to as "XX information." When describing the content of each piece of information, expressions such as "identification information," "identifier," "name," "ID," "number," etc. are used, but these are interchangeable.

[0016] Furthermore, although the following description may describe processing performed by executing a program, the program is executed by at least one processor (e.g., a CPU) to appropriately perform a predetermined process using storage resources (e.g., memory) and / or interface devices (e.g., communication ports), and therefore the processor may be the subject of the processing. Similarly, the subject of the processing performed by executing a program may be a controller, device, system, computer, node, storage system, storage device, server, management computer, client, or host having a processor. Furthermore, the subject of the processing performed by executing a program (e.g., a processor) may include a hardware circuit that performs some or all of the processing. For example, the subject of the processing performed by executing a program may include a hardware circuit that performs encryption and decryption, or compression and decompression. The processor operates as a functional unit that realizes a predetermined function by operating in accordance with the program. Apparatuses and systems including a processor are apparatuses and systems that include these functional units.

[0017] A program may be installed on a device such as a computer from a program source. The program source may be, for example, a program distribution server or a non-transitory storage medium readable by a computer. When the program source is a program distribution server, the program distribution server includes a processor (e.g., a CPU) and a non-transitory storage resource, and the storage resource may further store a distribution program and a program to be distributed. Then, the processor of the program distribution server may execute the distribution program, thereby distributing the program to be distributed to other computers. Also, in the following description, two or more programs may be realized as one program, and one program may be realized as two or more programs.

[0018] Furthermore, in the following description, when describing elements of the same type without distinguishing between them, the common portion (the portion excluding the subscript and subscript) of the reference sign including the subscript and subscript may be used, and when describing elements of the same type while distinguishing between them, the reference sign including the subscript and subscript may be used. Specifically, for example, when describing a "simulation model" without any particular distinction, it is written as "simulation model 102," whereas when describing individual simulation models 102 while distinguishing between them, it may be written as "simulation model 102-1," "simulation model 102-2," ... "simulation model 102-N." As another method for distinguishing between elements of the same type in drawings, etc., it may be written as "simulation model #1," "simulation model #2," ... "simulation model #N."

[0019] (1) System configuration Fig. 1 is a block diagram showing an example of the configuration of an overall system including an arbitration system 1 according to one embodiment of the present invention. The overall system shown in Fig. 1 is configured by connecting a plurality of independent individual systems 2 (individual systems 2-1 to 2-N) and an arbitration system 1 having a function of presenting arbitration proposals to these individual systems 2 via a communication system (network) 3.

[0020] In this embodiment, each individual system 2 has the same hardware configuration and functions, and the following describes an example of the configuration of the individual system 2, focusing on individual system #1 (individual system 2-1) shown in Fig. 1. However, this does not restrict multiple individual systems 2 from having different hardware configurations or functions.

[0021] The business operators (which may be considered different businesses for each individual system) that operate each individual system 2 (individually, individual systems 2-1 to 2-N) are businesses that receive orders from customers, manufacture products, and deliver them. Each business operator uses its own individual system 2-1 to 2-N at predetermined intervals (for example, once a week) to aggregate information on production status and order estimates in its own system and determine the production quantity for the following week. Furthermore, each business operator can transfer at least a portion of its own orders to another business operator depending on the order status.

[0022] The arbitration system 1 has the function of collecting information from multiple individual systems 2 at the above-mentioned predetermined period and proposing to each individual system 2 the number of orders to be transferred to another individual system 2 as an arbitration proposal. Each individual system 2 then determines how many orders to actually transfer based on the arbitration proposal proposed by the arbitration system 1. The processing flow by the arbitration system 1 will be described later with reference to Figure 2, and the processing flow by an individual system 2 will be described later with reference to Figure 3.

[0023] The internal configurations of the arbitration system 1 and the individual systems 2 shown in FIG. 1 will be described in detail below.

[0024] (1-1) Mediation System 1 The arbitration system 1 is a system that has the function of generating and presenting arbitration proposals (in this example, the number of orders to be transferred) to multiple independent individual systems 2. The arbitration system 1 has, as hardware, a storage device 10, an input device 11, an output device 12, a processor 13, a memory 14, and a bus 15 that connects them.

[0025] The input device 11 is a device that accepts input of data. Specific examples of the input device 11 include a keyboard, a mouse, a touch panel, a numeric keypad, a scanner, a microphone, a sensor, and a communication interface. The output device 12 is a device that outputs data. Specific examples of the output device 12 include a display, a printer, a speaker, and a communication interface.

[0026] The processor 13 is a processor that controls the arbitration system 1, and specifically is, for example, a CPU (Central Processing Unit). The processor 13 reads a program stored in the storage device 10 into the memory 14 and executes it. The memory 14 serves as a working area for the processor 13, and is, for example, a RAM (Random Access Memory).

[0027] The storage device 10 stores programs executed by the arbitration system 1, data used when executing the programs, etc. The storage device 10 is, for example, a storage device such as an HDD (Hard Disk Drive) or SSD (Solid State Drive) mounted on the device of the arbitration system 1, but is not limited to this, and at least a portion of the data stored in the storage device 10 may be held on a recording medium externally connected to the device of the arbitration system 1, or in a cloud environment connected to the arbitration system 1 via the network 3.

[0028] As shown in FIG. 1, the storage device 10 stores a database 101 for simulation models, a simulation model 102, a database 103 for learning models, a learning model 104, an environmental model 105, an input / output program 106, a model generation module 107, and a mediation proposal search module 110.

[0029] The simulation model database 101 is a data area for storing information required to generate the simulation model 102. Details will be described later with reference to Fig. 4, but the simulation model database 101 stores information previously transmitted to each individual system 2, information previously received from each individual system 2, and the like.

[0030] The simulation model 102 is an inference model that has a function of predicting the number of order transfers output by each individual system 2. The simulation model 102 is, for example, a function approximation program implemented by a neural network.

[0031] The learning model database 103 is a data area for storing information necessary to generate the learning model 104. Details will be described later with reference to Fig. 5, but the learning model database 103 stores, for example, parameters of the simulation model 102 from the past to the present.

[0032] The learning model 104 is an inference model that has the function of predicting how the simulation model 102 will change. The learning model 104 is, for example, a function approximation program implemented using a neural network. More specifically, the learning model #1 (learning model 104-1) corresponding to the business operator #1 (not shown) of the individual system #1 is a program that predicts the parameters of the simulation model at the next point in time based on the parameters of the simulation model #1 and the parameters of the arbitration proposal to be given to the business operator #1 (which may be read as the individual system #1).

[0033] The environmental model 105 is a model that predicts information that an individual system 2 acquires from outside its own system. The environmental model 105 is, for example, a program that predicts the number of orders that each individual system 2 will receive from other individual systems 2 in the next (e.g., next week).

[0034] The input / output program 106 is a program having a function of receiving data using the input device 11 and transmitting data using the output device 12 .

[0035] The model generation module 107 is a module including programs for generating the simulation model 102 and the learning model 104. The model generation module 107 includes a simulation model generation program 108 that generates the simulation model 102 using data stored in the simulation model database 101, and a learning model generation program 109 that generates the learning model 104 using data stored in the learning model database 103.

[0036] The mediation proposal search module 110 is a program that determines a mediation proposal to be output to the individual system 2. The mediation proposal search module 110 has mediation proposal generation parameters 111, a mediation proposal generation program 112, and a mediation proposal evaluation program 113. The mediation proposal generation parameters 111 are parameters of a function used when the mediation proposal generation program 112 generates a mediation proposal. The mediation proposal generation program 112 is a program that has the function of generating a mediation proposal using the simulation model 102, the learning model 104, and the mediation proposal generation parameters 111. Details of the mediation proposal generation process performed by the mediation proposal generation program 112 will be described later with reference to FIG. 6. The mediation proposal evaluation program 113 is a program that evaluates the mediation proposal generated by the mediation proposal generation program 112. Details of the mediation proposal evaluation process performed by the mediation proposal evaluation program 113 will be described later with reference to FIG. 7.

[0037] (1-2) Individual System 2 The individual system 2 has, as its hardware, a storage device 20, an input device 21, an output device 22, a processor 23, a memory 24, and a bus 25. The functions and specific examples of each piece of hardware are the same as those described above in the description of the hardware configuration of the arbitration system 1. In the overall system shown in this embodiment, N individual systems 2 having similar configurations are connected to the arbitration system 1 via a communication system (network) 3.

[0038] The storage device 20 stores an input / output program 201, a behavior determination program 202, behavior determination parameters 203, a parameter update program 204, and a database 205 as programs executed by the individual system 2 and data used when executing the programs.

[0039] The input / output program 201 is a program having the function of receiving data using the input device 21 and transmitting data using the output device 22 .

[0040] The behavioral decision program 202 is a program that acquires the current production status, the current order status, and arbitration proposal information from the arbitration system 1, and determines the number of orders to be transferred to another individual system 2 using the behavioral decision parameters 203 and data stored in the database 205. Specifically, for example, when the number of products in production is greater than a certain number and the number of orders received this week is greater than a certain number, the behavioral decision program 202 is a program that transfers orders by the proportion specified in the behavioral decision parameters 203 relative to the number presented in the arbitration proposal.

[0041] The behavior decision parameter 203 is parameter information consisting of one or more numerical values ​​used when the behavior decision program 202 decides on an action. Specifically, the behavior decision parameter 203 is, for example, a numerical value that determines the ratio of the number presented in the mediation proposal to be actually transferred.

[0042] The parameter update program 204 is a program that has a function of updating the behavioral decision parameters 203 using information stored in the database 205. Specifically, the parameter update program 204 is a program that adjusts the behavioral decision parameters 203 based on past data, for example, so as to maximize the expected value of the profit (individual KPI).

[0043] The database 205 stores information on past production status, order status, presented arbitration proposals, and agreements on the arbitration proposals. The information stored in the database 205 may include, for example, information on estimates of the number of orders received in the past, past behavioral decision parameters 203, etc.

[0044] (2) Processing The following describes the processing executed by the arbitration system 1 and the individual systems 2 in the overall system having the configuration described in (1).

[0045] 2 is a flowchart showing an example of the processing procedure of the arbitration system processing. The arbitration system processing is a process executed by the arbitration system 1, which uses data collected from multiple individual systems 2 and various models and data held by the arbitration system 1 to learn a model, repeatedly generates and evaluates arbitration proposals, and finally outputs a decided arbitration proposal.

[0046] According to Figure 2, first, the arbitration system 1 uses the input / output program 106 to receive information from each individual system 2 regarding the current production status and order status (number of orders), as well as information on whether the order was transferred in accordance with the previous arbitration proposal, via the input device 11, and stores this information in the simulation model database 101 (step S20).

[0047] Next, the arbitration system 1 uses the model generation module 107 to perform a model generation process for generating the simulation model 102 and the learning model 104 (step S21). In the model generation process, the simulation model generation program 108 of the model generation module 107 uses the data stored in the simulation model database 101 in step S20 to learn the existing simulation model 102, thereby generating the simulation model 102 for this arbitration. Also, in the model generation process, the learning model generation program 109 of the model generation module 107 uses the data stored in the learning model database 103 in step S20 to learn the learning process (time change) of the past simulation model 102, thereby generating the learning model 104 for this arbitration. The relationship between the data stored in the simulation model database 101 and the simulation model 102 will be described later with reference to FIG. 4. The relationship between the data stored in the learning model database 103 and the learning model 104 will be described later with reference to FIG. 5.

[0048] After performing model learning in step S21, the arbitration system 1 repeatedly generates and evaluates arbitration proposals using the arbitration proposal search module 110 in steps S22 to S27, and generates a arbitration proposal to be finally output. The processing of each step will be described.

[0049] The mediation proposal search module 110 first sets mediation proposal generation parameters 111 (step S22). The mediation proposal generation parameters 111 may be inherited from the previous process, or may be newly set (updated) according to some rule, for example, randomly.

[0050] Next, the mediation proposal search module 110 executes a mediation proposal generation process for generating a mediation proposal using the mediation proposal generation program 112 (step S23). The mediation proposal generation process will be described in detail later with reference to FIG.

[0051] Next, the mediation proposal search module 110 uses the mediation proposal evaluation program 113 to evaluate the mediation proposal generated in step S23 and executes a mediation proposal evaluation process to calculate its evaluation value (step S24). The evaluation value calculated in step S24 is data consisting of at least one or more numerical values, and is an index in which the larger each value is, the more desirable the state is. The mediation proposal evaluation process will be described in detail later with reference to FIG. 7.

[0052] Next, the mediation proposal search module 110 stores the mediation proposal obtained in step S23 and the evaluation value obtained in step S24 in the memory 14 or the like (step S25). At this time, the mediation proposal search module 110 may also store the variables and parameters calculated in the mediation proposal generation process (step S23) and the mediation proposal evaluation process (step S24) in the memory 14 or the like.

[0053] Next, the settlement proposal search module 110 (settlement proposal evaluation program 113) determines whether the evaluation value of the settlement proposal obtained in step S24 satisfies the search termination condition (step S26).

[0054] The search termination condition is a condition determined using one or more of the numerical values ​​included in the evaluation value. Specifically, for example, if the evaluation value is composed of an "overall KPI" for the overall system, "individual KPIs" for each individual system 2, and a "predicted value" of the degree of cooperation in each individual system 2 after a certain period of time, this evaluation value is expressed as one overall KPI, N individual KPIs, and N predicted values, i.e., a total of 2N+1 numerical values. In this case, the search termination condition can be whether or not the weighted sum of the 2N+1 numerical values ​​constituting the evaluation value exceeds a predetermined threshold. Note that the "overall KPI" is an index that is desired to be maximized in the entire system including the N individual systems 2, and the "individual KPI" is an index that is desired to be maximized in each individual system 2, and the overall KPI and the individual KPIs may be different indexes.

[0055] In the above specific example, the weight parameters corresponding to the 2N+1 numerical values ​​of the evaluation value and the threshold value used to determine the termination condition may be given fixed values ​​in advance, or may be set by a user or the like using the input device 11 from the outside each time the arbitration system processing is executed, or may be set by the arbitration proposal search module 110 based on the arbitration results up to the previous time.

[0056] If the search termination condition is not satisfied in the determination of step S26 (NO in step S26), the mediation proposal search module 110 returns to step S22, updates the mediation proposal generation parameter 111 to another value, and repeats the process. At this time, the mediation proposal search module 110 may update the mediation proposal generation parameter 111 according to a predetermined rule (for example, randomly), or may adjust and update the mediation proposal generation parameter 111 based on the mediation proposal and evaluation value obtained in the most recent steps S23 to S24.

[0057] If the search termination condition is met in the determination of step S26 (YES in step S26), the reconciliation proposal search module 110 stores the reconciliation proposal and evaluation value generated in the most recent steps S23 to S24 as final results in the memory 14 or the like (step S27). Note that in step S27, the reconciliation proposal search module 110 may store, together with the reconciliation proposal and evaluation value, numerical values ​​and parameters calculated in the reconciliation proposal generation process (step S23) and the reconciliation proposal evaluation process (step S24), such as information on the prediction of future behavior of the individual system 2.

[0058] Following step S27, the arbitration system 1 performs arbitration proposal output processing to output the information stored in the memory 14 in step S27 (step S28). The arbitration proposal output processing is performed by the arbitration proposal search module 110 calling the input / output program 106, and specifically, for example, notifies each individual system 2 of the arbitration proposal via the network 3 and displays the arbitration proposal on the output device 12.

[0059] Thereafter, the arbitration system 1 ends the current arbitration system process and returns to a state of waiting for the next input reception from the individual system 2 (step S20).

[0060] 3 is a flowchart showing an example of the processing procedure of the individual system processing. The individual system processing is executed by each individual system 2, and includes transmitting information about the production status and order status of the individual system to the arbitration system 1, and determining and implementing actions in the individual system based on the arbitration plan decided by the arbitration system 1.

[0061] According to FIG. 3, first, the individual system 2 uses the input / output program 201 to receive the production status and order status (order quantity) of its own system from the input device 21 and stores it in the database 205 (step S30).

[0062] Next, the individual system 2 executes a behavior-determining parameter update process (step S31) using the parameter update program 204. Specifically, the parameter update program 204 determines behavior-determining parameters that maximize the individual KPI of the behavior of its own system based on the past data stored in the database 205, and updates the behavior-determining parameters 203 with the determined parameters.

[0063] Next, the individual system 2 uses the input / output program 201 to transmit the information on the production status and order status acquired in step S30 from the output device 22 via the network 3 to the arbitration system 1 (step S32). The data transmitted in step S32 is received by the arbitration system 1 in the processing of step S20 in Fig. 2. The arbitration system 1 then executes the processing of steps S21 to S27, and sends the decided arbitration proposal to the individual system 2 in step S28.

[0064] Next, the individual system 2 receives the arbitration proposal output from the arbitration system 1 in step S28 of FIG. 2 via the network 3 at the input device 21, and stores it in the database 205 (step S33).

[0065] Next, the individual system 2 executes a behavior decision process to decide the current behavior of its own system using the behavior decision program 202 (step S34). Specifically, for example, the behavior decision program 202 determines the production quantity and the number of orders to be transferred to another individual system 2 based on the information (production status, order status) stored in the database 205, the behavior decision parameters 203 determined in step S31, and the arbitration proposal received from the arbitration system 1 in step S33.

[0066] Next, the individual system 2 carries out the action determined in step S34 (step S35), and ends the series of processes. In step S35, the individual system 2 transfers or accepts the order in accordance with the action content determined in step S34, and then starts production for the final ordered quantity.

[0067] 4 is a diagram showing an example of data stored in the simulation model database 101. The simulation model database 101 holds information required to generate simulation models 102-1 to 102-N that predict the behavior of the individual systems 2-1 to 2-N.

[0068] Specifically, as shown in Fig. 4, for each past arbitration (executing the arbitration system process in Fig. 2 constitutes one arbitration), data 40-1 from the previous arbitration, data 40-2 from the second previous arbitration, ..., data 40-T from the Tth previous arbitration, are stored in the simulation model database 101. T is an arbitrary integer, and may be a predetermined number set in advance or may be changeable by the user or a program.

[0069] More specifically, data 40 at each point in time includes, for each individual system 2, data 41 relating to the production status and order status in that individual system 2, data 42 relating to the order transfer status in that individual system 2, and data 43 relating to the reconciliation proposal presented to that individual system 2. Data 41 (data 41-1 to 41-N) and data 42 (data 42-1 to 42-N) correspond to the information data received from each individual system 2 in step S20 of Fig. 2. Data 43 (data 43-1 to 43-N) corresponds to the data of the reconciliation proposal notified to each individual system 2 in step S28 of Fig. 2.

[0070] 4 also shows an image of the relationship between the simulation model 102 generated by the simulation model generation program 108 in the model generation process (step S21 in FIG. 2) in the arbitration system process and the data stored in the simulation model database 101. Specifically, it is shown that the simulation model #1 (simulation model 102-1) generated in the immediately previous (most recent) arbitration system process is a model that uses data 41-1 related to the production status and order status of individual system #1 at the time of the immediately previous arbitration and data 43-1 related to the arbitration proposal presented to individual system #1 as input variables, and data 42-1 related to the transfer status of individual system #1 at the time of the first arbitration (the number of orders transferred to other systems and the number of orders transferred from other systems) as the objective variable.

[0071] 4 is merely an example and is not limited to this. As another example, the simulation model 102 may be a model whose objective function is the ratio of the number of actual transfers to the number of transfers presented in the mediation proposal, as will be described later in FIG. 9.

[0072] 5 is a diagram showing an example of data stored in the learning model database 103. The learning model database 103 holds information necessary to generate learning models 104-1 to 104-N that predict the learning process of the simulation models 102-1 to 102-N.

[0073] Specifically, as shown in Fig. 5, for each past arbitration (executing the arbitration system process in Fig. 2 constitutes one arbitration), data 50-1 from the previous arbitration, data 50-2 from the second previous arbitration, ..., data 50-T from the Tth previous arbitration, are stored in the learning model database 103. T is an arbitrary integer, and may be a predetermined number set in advance or may be changeable by the user or the program.

[0074] More specifically, the data 50 at each point in time includes data 51 related to the simulation model 102 of each individual system 2 at that point in time, and data 52 related to the reconciliation proposal presented to each individual system 2 at that point in time. The data 51 (data 51-1 to 51-N) is, for example, parameters used in the simulation models 102-1 to 102-N. Furthermore, the data 52 (data 52-1 to 52-N) corresponds to the data of the reconciliation proposal notified to each individual system 2 in step S28 of FIG. 2.

[0075] 5 also shows an image of the relationship between the learning model 104 generated by the learning model generation program 109 in the model generation process (step S21 in FIG. 2) in the arbitration system process and the data stored in the learning model database 103. Specifically, it shows that the learning model #1 (learning model 103-1) generated in the immediately previous (most recent) arbitration system process is a model that uses data 50 from past arbitrations, i.e., data 50-2 to 50-N from arbitrations two times before to T times before, as input variables and data 50-1 from the first arbitration as a response variable.

[0076] 6 is a flowchart showing an example of the processing procedure for the mediation proposal generation process. As described in the explanation of step S23 in FIG.

[0077] 6, first, the mediation proposal generation program 112 initializes the mediation proposal generation parameters 111, reads out the simulation models 102-1 to 102-N from the simulation model database 101, and stores them in the memory 14 (step S60). In step S60, the mediation proposal generation program 112 may initialize the mediation proposal generation parameters 111 according to a predetermined rule, for example, randomly, or may use other methods, for example, parameters used when generating the previous mediation proposal, as initial values.

[0078] Next, the mediation proposal generation program 112 initializes a step variable t to 0 and sets a predicted maximum number of steps T (step S61). The step variable t used in the mediation proposal generation process is a variable indicating how many mediation proposals have been generated from the current point in time (t=0), and the predicted maximum number of steps T is a variable that determines how many mediation proposal candidates the mediation proposal generation program 112 will output at most. The predicted maximum number of steps T is set to, for example, a fixed value according to a preset rule.

[0079] Next, the arbitration proposal generation program 112 generates an arbitration proposal based on the arbitration proposal generation parameters 111 set in step S60 (step S62). For example, when the sum of the number of products being produced (number of products in production) and the number of products ordered (number of orders) in each individual system 2 exceeds a value specified by the arbitration proposal generation parameters 111, the arbitration proposal generation program 112 generates an arbitration proposal that proposes that orders for its own system be transferred to other individual systems 2 by the amount that exceeds the sum.

[0080] Next, the mediation proposal generation program 112 predicts, for each individual system 2, what behavior each individual system 2 (or its operator) will take when the mediation proposal generated in step S62 is presented to the individual systems 2-1 to 2-N (step S63), using the simulation model 102 stored in the memory 14 in step S60. A behavior prediction value for each individual system 2 is calculated based on the prediction in step S63.

[0081] Next, the mediation proposal generation program 112 uses the learning models 104-1 to 104-N to update the simulation model 102 stored in the memory 14 (step S64) in order to predict how the behavior of each individual system 2 (or its operator) will change when the mediation proposal generated in step S62 is presented to the individual systems 2-1 to 2-N. That is, the simulation model 102 after learning is calculated by the processing of step S64.

[0082] Next, the arbitration proposal generation program 112 predicts the production status, order status, etc. of each business operator (each individual system 2) at the time of the next arbitration using the environmental model 105 (step S65). By the processing of step S65, predicted values ​​of the production status and order status at the time of the next arbitration are calculated.

[0083] Next, the arbitration proposal generation program 112 stores in the memory 14 the behavior prediction value calculated in step S63, the learned simulation model 102 calculated in step S64, and the prediction values ​​of the production status and order status at the time of the next arbitration calculated in step S65 (step S66). Hereinafter, this information stored in the memory 14 in step S66 will be referred to as the "arbitration proposal and accompanying information" in step t.

[0084] Next, the mediation proposal generating program 112 compares whether the step variable t is equal to or greater than the predicted maximum number of steps T set in step S61 (step S67).

[0085] If the step variable t is less than the predicted maximum number of steps T in step S67 (NO in step S67), the mediation proposal generation program 112 updates the mediation proposal generation parameter 111, assigns t+1 to the step variable t (step S68), and returns to the processing of step S62. In step S68, the mediation proposal generation parameter 111 is updated according to a predetermined rule, for example, randomly.

[0086] If the step variable t is equal to or greater than the predicted maximum number of steps T in step S67 (YES in step S67), the mediation proposal generation program 112 stores the mediation proposals and accompanying information from step 0 to step T in the storage device 10 (step S69), and terminates the mediation proposal generation process.

[0087] 7 is a flowchart showing an example of the processing procedure for the mediation proposal evaluation process. As described in the explanation of step S24 in FIG.

[0088] According to Fig. 7, first, the mediation proposal evaluation program 113 initializes a step variable t to 0 and sets the maximum number of evaluation steps T (step S70). The step variable t used in the mediation proposal evaluation process is a variable indicating how many mediation proposals from the current time (t = 0) the mediation proposal evaluation program 113 is evaluating, and the maximum number of evaluation steps T is a variable that determines how many mediation proposal candidates the mediation proposal evaluation program 113 will evaluate at most. The maximum number of evaluation steps T is set to, for example, a fixed value according to a preset rule. Note that the predicted maximum number of steps T used in the mediation proposal generation process and the maximum number of evaluation steps T used in the mediation proposal evaluation process may be the same value.

[0089] Next, the mediation proposal evaluation program 113 acquires the mediation proposal and accompanying information in step t, i.e., the mediation proposal to be presented to the individual system 2, the predicted behavior value of the individual system 2, and the simulation model 102 after learning of the individual system 2, and stores them in the memory 14, etc. (step S71).

[0090] Next, based on the information saved in step S71, the mediation proposal evaluation program 113 calculates the individual KPI of each individual system 2 (step S72). The individual KPI of each individual system 2 (individual KPI) is, for example, the profit that each individual system 2 has gained by implementing the action selected in step t.

[0091] Next, the mediation proposal evaluation program 113 calculates an overall KPI based on the information saved in step S71 (step S73). The overall KPI is an index that is desired to be maximized for the entire system including the multiple individual systems 2-1 to 2-N, and is, for example, the sum of the individual KPIs of each individual system 2 calculated in step S72.

[0092] Next, the mediation proposal evaluation program 113 calculates the degree of cooperation of each individual system 2 based on the information saved in step S71 (step S74). The degree of cooperation is, for example, a value obtained by evaluating how many transfers have actually been made compared to the number of transfers proposed in the mediation proposal presented by the mediation system 1. Specifically, for example, the mediation proposal evaluation program 113 can calculate the degree of cooperation by using the simulation model 102 after learning at step t saved in step S71 to check whether the individual system 2 will accept the transfer proposal under various circumstances.

[0093] Next, the mediation proposal evaluation program 113 stores in the memory 14 the individual KPI calculated in step S72, the overall KPI calculated in step S73, and the evaluation value of the degree of cooperation of each individual system 2 calculated in step S74 (step S75). Hereinafter, these numerical values ​​stored in the memory 14 in step S75 will be referred to as the "total evaluation value" in step t.

[0094] Next, the mediation proposal evaluation program 113 compares whether the step variable t is equal to or greater than the maximum number of evaluation steps T set in step S70 (step S76).

[0095] If the step variable t is less than the maximum number of evaluation steps T in step S76 (NO in step S76), the mediation proposal evaluation program 113 assigns t+1 to the step variable t (step S77) and returns to the processing of step S71.

[0096] If the step variable t is equal to or greater than the maximum number of evaluation steps T in step S76 (YES in step S76), the mediation proposal evaluation program 113 calculates a final evaluation value based on all evaluation values ​​from step 0 to step T, stores the final evaluation value in memory 14 (step S78), and terminates the mediation proposal evaluation process. Note that the "final evaluation value" is an index that represents the overall merit of the mediation proposal defined by multiple numerical values ​​for the period from the present time to a predetermined future time point corresponding to the maximum number of evaluation steps T, and specifically, for example, it can be calculated as the average value of each of the individual KPI, overall KPI, and degree of cooperation from step 0 to step T.

[0097] (3) Specific examples Below, with reference to Figures 8 to 14, we will explain specific examples of the model generation process (step S21 in Figure 2), the mediation proposal generation process (step S23 in Figure 2), and the mediation proposal evaluation process (step S24 in Figure 2), as well as the effects obtained from the mediation system 1.

[0098] In the specific example described below, the arbitration system 1 is a system that arbitrates production allocation problems, and in the simplest example, the number of businesses is N=2 (that is, two individual systems 2 in the overall system).

[0099] Within the overall system, two individual systems #1 and #2 are systems operated independently by two branches #1 (operator #1) and #2 (operator #2) belonging to the same company. Each branch independently receives product orders and carries out the production and delivery of the products. Each branch sells products at a fixed price, and its goal is to maximize gross profit, which is the selling price minus production costs (individual KPI). Meanwhile, the goal of the headquarters (the entire system) that oversees each branch is to maximize the total gross profit of branch #1 and branch #2 (overall KPI).

[0100] As shown in Figure 8 below, the cost of producing a product varies depending on the number of products in production (number of products in production), and tends to increase as the number of products in production increases. This is due to the additional costs incurred from overtime and outsourcing when the factory's production capacity is exceeded. Each branch office starts production of product orders once a week, but can transfer orders to other branches from week to week.

[0101] Figure 8 is a graph showing an example of the structure of production costs in individual system 2. In the graph shown in Figure 8, the horizontal axis represents the number of products being produced at the factory at the time of order receipt (number of products in production), and the vertical axis represents the production cost required to produce one additional product. The production cost is assumed to be the same for branch office #1 and branch office #2.

[0102] Graph 801, shown by a solid line in Figure 8, represents the time series change in production costs when an order received at a branch office is produced at the branch office's factory. As graph 801 shows, when an order is received at a branch office, the product can be produced at a constant production cost of "5" as long as the number of products being produced does not exceed 30 units, which is the factory's normal production capacity. On the other hand, if the number of products being produced exceeds the capacity of 30 units, production costs increase, and when the number of products being produced reaches 60 units, the production cost becomes "15", which is equal to the selling price, and profit becomes "0".

[0103] Graph 802, shown by a dashed line in Figure 8, represents the time series change in production costs when an order transferred from another branch is produced at the branch's factory. In this graph 802, compared to the production cost of an order received at the branch (graph 801), a cost of "5" associated with transferring an order between branches is added uniformly to the production cost. With this cost structure, if the number of products currently being produced at the branch is 45 or more and the number of products currently being produced at the other branch is 15 or more fewer than the number of products currently being produced at the branch, transferring the branch's order to the other branch will increase the total profit of all branches. In other words, the overall KPI can be improved.

[0104] On the other hand, if we focus on the home branch, the profits of the branch that transferred the order will decrease (individual KPI will decrease), while the profits of the branch to which the order is transferred will increase (individual KPI will increase). Therefore, if each branch simply acts to maximize its own profits, no transfers from the home branch to other branches will occur unless the number of products being produced exceeds 60 and the branch is in the red. However, in this case, the increase in the total profits of all branches will be smaller than when the moderate transfers described above are made between branches. In other words, if each branch (each individual system 2) takes actions to maximize its own individual KPI, it is possible that the overall KPI will not be maximized.

[0105] Fig. 9 is a graph illustrating a specific example of data stored in the simulation model database 101. Fig. 9 is a graph of various data for 100 weeks going back from the current time "O" regarding the data stored in the simulation model database 101 by the arbitration system 1 regarding individual system 2 of branch office #1.

[0106] Graph 900 in FIG. 9(A) is a graph showing data on the number of products in production at branch #1 in chronological order. The horizontal axis indicates the week in which the data was acquired. For example, "O-25" means that the data is acquired 25 weeks prior to the current "O." Similar horizontal axes are used in other figures referred to below, and explanations will be omitted. The vertical axis indicates the number of products in production at branch #1 at the time the data was acquired (number of products in production).

[0107] Graph 910 in Figure 9(B) is a graph showing, in chronological order, the number of orders (transferred) transferred from branch office #1 to another branch office (branch office #2) and the number of orders (transferred) transferred from the other branch office (branch office #2) to branch office #1. As with graph 910, the horizontal axis indicates the week in which the data was obtained. The vertical axis indicates the number of orders transferred or transferred in the week in which the data was obtained.

[0108] Graph 920 in Figure 9(C) is a graph showing the ratio of the number of transfers actually made to the number of transfers proposed in the mediation proposal over time. As with graphs 910 and 920, the horizontal axis indicates the week in which the data was acquired. The vertical axis indicates the ratio of the number of orders actually transferred to other branches in the week in which the data was acquired to the number of transfers indicated in the mediation proposal presented in the week prior to the week in which the data was acquired (hereinafter referred to as the "mediation proposal acceptance rate").

[0109] In the model generation process (step S21 in Figure 2) within the arbitration system processing, the arbitration system 1 uses the data represented by graphs 900, 910, and 920 to generate a simulation model 102 that takes the number of products in production and the number of transfers presented in the arbitration proposal as inputs and the arbitration proposal acceptance rate as output, by implementing it using a function implemented, for example, by a neural network and performing supervised learning.

[0110] Furthermore, in the model generation process (step S21 in FIG. 2 ), the arbitration system 1 generates a learning model 104 that estimates changes in the behavior of the individual system 2 by analyzing the changes between a simulation model 102 generated using only data from a certain period and a simulation model 102 generated using data from another period. In the specific example shown in FIG. 9 , the learning model 104-1 calculates that the arbitration proposal acceptance rate changes significantly every 25 weeks from the data in graph 920, for example, by performing a Fourier transform on graph 920 to analyze the frequency. Furthermore, the learning model 104-1 calculates that if the number of orders transferred within a certain period (e.g., 25 weeks) is greater than a predetermined number, the arbitration proposal acceptance rate will increase in the next 25-week period, for example, by analyzing the correlation between the arbitration proposal acceptance rate at a different point in time and the number of orders transferred in the previous 25 weeks.

[0111] Fig. 10 is a graph for explaining an example of an arbitration proposal for the production status in individual system 2. Fig. 10 is a graph showing the number of products in production going back 20 weeks from the present (i.e., "O" to "O-20") and the number of orders transferred between branch office #1 and branch office #2 (transferred amount, transferred amount) using data stored in simulation model database 101 by arbitration system 1 regarding individual systems 2 of branch office #1 and branch office #2.

[0112] 10, the horizontal axis represents the week in which the data was acquired, and the vertical axis represents the number of products in production at each branch at the time the data was acquired. Graph 1001, shown by a solid line in Fig. 10, represents the change in the number of products in production at branch #1, and graph 1002, shown by a dashed dotted line, represents the change over time in the number of products in production at branch #2.

[0113] According to Figure 10, from week O-19 to around week O-11, branch #1 transferred orders to branch #2 in accordance with the arbitration proposal (transferred amount 1003, transferred amount 1004). After that, the number of products in production at branch #1 decreased and the number of products in production at branch #2 increased, but because the difference between the two was not large, no transfers occurred between the branches.

[0114] In the situation of Figure 10 described above, if the arbitration system 1 generates an arbitration proposal that maximizes the overall KPI without taking the learning process into consideration, the difference in the number of products being produced between branch #1 and branch #2 at the current time O is 15 or less, and according to the cost structure shown in Figure 8, even if the transfer were to occur, the overall KPI would not increase, so the arbitration system 1 would not propose a transfer (it would generate an arbitration proposal that does not propose a transfer).

[0115] FIG. 11 is a graph for explaining another example of an arbitration proposal for the production status in individual system 2. Like FIG. 10, FIG. 11 is a graph showing the number of products being produced and the transfer of order quantities at each branch using data related to individual systems 2 of branch #1 and branch #2, and explanation of the graph configuration etc. will be omitted. The difference from FIG. 10 is that in FIG. 11, arbitration system 1 proposes an arbitration proposal taking into account the learning process.

[0116] The mediation proposal that does not propose a transfer, as described in Figure 10, predicts that the rate at which branch office #1 will accept the mediation proposal will decrease when the mediation proposal evaluation process (step S24 in Figure 2, Figure 7) is performed by the mediation proposal evaluation program 113, and the evaluation value of the degree of cooperation of branch office #1 calculated in step S74 in Figure 7 will be low.

[0117] On the other hand, in FIG. 11, a mediation proposal is proposed at the current time "O" to transfer orders (transfer amount 1101) to branch office #2. This mediation proposal reduces overall profits in the short term, resulting in a low evaluation value for the overall KPI calculated in step S73 of FIG. 7, but a high evaluation value for the degree of cooperation of branch office #2 calculated in step S74. A high evaluation value for the degree of cooperation in the long term makes the transfer proposed by the mediation proposal more likely to be accepted, which is expected to have the effect of improving the overall KPI in the future. To improve the evaluation value for the degree of cooperation, the search termination condition of the mediation system processing (step S26 of FIG. 2) can be programmed to require, for example, that the evaluation value for the degree of cooperation included in the "final evaluation value" be above a certain level (above a predetermined threshold). As a result, the mediation system 1 can generate and output a mediation proposal such as that shown in FIG. 11, thereby improving the long-term degree of cooperation and the future overall KPI.

[0118] The above describes the arbitration of a production allocation problem involving multiple individual systems 2 that produce products as a specific example of arbitration by the arbitration system 1. However, the arbitration system 1 according to this embodiment can be widely applied to an overall system that encompasses multiple individual systems 2, each of which independently determines its own behavior so as to increase its individual key performance indicator (KPI). In generating an arbitration proposal, the arbitration system 1 calculates the transfer quantity so as to increase an overall key performance indicator (overall KPI) that is different from the individual key performance indicators in such an overall system. Furthermore, in the specific example of the production allocation problem described above, the arbitration proposal generated by the arbitration system 1 proposed the number of orders to be transferred to other branches (other individual systems 2). However, the content of the arbitration proposal generated by the arbitration system 1 according to this embodiment is not limited to the number of "orders" to be transferred, and may be any proposal that proposes the transfer of "predetermined factors" related to the determination of the behavior of the individual systems 2.

[0119] From the above, it can be said that the arbitration system 1 according to this embodiment is an arbitration system that presents arbitration proposals related to the behavior of multiple individual systems 2, each of which independently determines its own behavior (e.g., product production) so as to increase its individual key performance indicators, and is a system comprising a model generation module 107 and an arbitration proposal search module 110. Here, the model generation module 107 is a module (program) that generates a simulation model 102 that estimates the behavior of each individual system 2 and a learning model 104 that estimates changes in the behavior of each individual system 2 based on time-series changes in the simulation model 102. The model generation module 107 and the arbitration proposal search module 110 are modules (programs) that use the simulation model 102 and the learning model 104 to generate arbitration proposals that propose the transfer of predetermined factors (e.g., order quantity) related to the determination of the behavior of each individual system 2 between the individual systems, and present the proposals to the individual systems 2. Thus, the arbitration system 1 can generate arbitration proposals that take into account changes in the behavior of each individual system 2. [Explanation of symbols]

[0120] 1. Mediation System 2 Individual Systems 3. Communication Systems (Networks) 10 Storage device 11,21 Input devices 12,22 Output Devices 13,23 processor 14,24 memory 15,25 bus 101 Simulation model database 102 Simulation Model 103 Database for learning models 104 Learning Model 105 Environmental Model 106 Input / Output Program 107 Model Generation Module 108 Simulation model generation program 109 Learning Model Generation Program 110 Mediation Proposal Search Module 111 Mediation Proposal Generation Parameters 112 Mediation proposal generation program 113 Mediation Proposal Evaluation Program 201 Input / Output Program 202 Behavioral Decision Program 203 Behavioral Decision Parameters 204 Parameter Updater 205 databases

Claims

1. An arbitration system that proposes arbitration proposals related to actions to a plurality of individual systems, each of which independently determines its own actions so as to increase an individual key performance indicator, a model generation module that generates a simulation model that estimates the behavior of each of the individual systems and a learning model that estimates changes in the behavior of each of the individual systems based on time-series changes in the simulation model; a reconciliation proposal search module that uses the simulation model and the learning model to generate a reconciliation proposal that proposes a transfer of predetermined factors related to the determination of the behavior between the individual systems and presents the reconciliation proposal to the individual systems; An arbitration system comprising:

2. The reconciliation proposal search module calculates the reconciliation proposal so that an overall key performance indicator, which is different from the individual key performance indicators, is increased in an overall system including the plurality of individual systems in generating the reconciliation proposal. The arbitration system according to claim 1 .

3. The mediation proposal search module: Mediation proposal generation parameters; a reconciliation proposal generation unit that generates the reconciliation proposal using the simulation model, the learning model, and the reconciliation proposal generation parameters; a mediation proposal evaluation unit that evaluates the mediation proposal generated by the mediation proposal generation unit; and The mediation proposal evaluation unit calculates, as evaluation values ​​of the mediation proposal, the individual key performance indicators for each of the individual systems, the overall key performance indicators for the overall system, and an evaluation value of a degree of cooperation that evaluates the degree of cooperation with the mediation proposal in each of the individual systems. The arbitration system according to claim 2 .

4. for a predetermined period from the present time to a predetermined future time, the generation of the simulation model and the learning model by the model generation module and the generation of the reconciliation proposal by the reconciliation proposal search module are repeatedly executed, with each of the individual systems being a predetermined cycle in which the individual systems decide on their actions as a unit; The mediation proposal evaluation unit calculates a final evaluation value of the mediation proposal over the predetermined period based on the individual key performance indicators calculated during the repetition, the overall key performance indicator, and the evaluation value of the degree of cooperation. The arbitration system according to claim 3 .

5. The mediation proposal search module: repeatedly executing generation of the reconciliation proposal by the reconciliation proposal generation unit and evaluation of the reconciliation proposal by the reconciliation proposal evaluation unit while updating the reconciliation proposal generation parameters; When the evaluation value of the degree of cooperation included in the final evaluation value of the mediation proposal calculated by the mediation proposal evaluation unit is equal to or greater than a predetermined threshold, the mediation proposal is determined as the mediation proposal to be presented to the individual system.

5. The arbitration system according to claim 4.

6. The arbitration proposal search module proposes, in the arbitration proposal presented to each of the individual systems, a transfer amount for transferring orders received by the individual system itself to another individual system. The arbitration system according to claim 1 .

7. The arbitration proposal search module generates the arbitration proposal under a condition that the production cost of an order transferred from another individual system is set higher than the production cost of an order received by the individual system itself. The arbitration system according to claim 6 .

8. 1. A mediation method by a mediation system that proposes mediation proposals related to actions to a plurality of individual systems, each of which independently determines its own actions so as to increase its individual key performance indicators, the method comprising: a model generation step in which the arbitration system generates a simulation model that estimates the behavior of each of the individual systems and a learning model that estimates changes in the behavior of each of the individual systems based on time-series changes in the simulation model; a reconciliation proposal search step in which the reconciliation system uses the simulation model and the learning model to generate a reconciliation proposal that proposes the transfer of predetermined factors related to the behavior determination between the individual systems and presents the reconciliation proposal to the individual systems; An arbitration method comprising:

9. A program executed in an arbitration system that proposes arbitration proposals related to actions to a plurality of individual systems, each of which independently determines its own actions to increase its individual key performance indicators, a model generation process for generating a simulation model that estimates the behavior of each of the individual systems and a learning model that estimates changes in the behavior of each of the individual systems based on time-series changes in the simulation model; a reconciliation proposal generation process that uses the simulation model and the learning model to generate a reconciliation proposal that proposes the transfer of predetermined factors related to the determination of the behavior between the individual systems and outputs the reconciliation proposal to the individual systems; A program characterized by executing

Images