JP2022041852A - Regeneration condition identification device and regeneration condition identification method - Google Patents

Abstract

To provide technology of identifying regeneration conditions for retaining the amount of a binder eluted from regenerated casting sand at a target value.SOLUTION: A regeneration condition identification device (1) comprises one or plural processors practicing a regeneration condition identification step of, from the amount of a binder eluted from casting sand regenerated by a sand regeneration system, identifying regeneration conditions in the sand regeneration system.SELECTED DRAWING: Figure 1

Description

The present invention relates to an apparatus and a method for managing the sand properties of cast sand regenerated by a sand regeneration system.

A technique for managing the sand properties of casting sand used for molding a mold is known. For example, in Patent Document 1, data is recalled and displayed from a knowledge base that describes a method of managing foundry sand, which has been empirically determined by a sand management engineer in the past, by utilizing the measured values of a foundry sand tester. The system is described.

Further, Patent Document 2 describes a computer utilization system for sand optimization for the purpose of reducing the number of rejected castings in a foundry (casting factory). In the system described in Patent Document 2, the processor module calculates the correlation between the value of the parameter such as the pH value of sand, the parameter, the rejection, and the rejection type, and generates a normative solution.

Japanese Unexamined Patent Publication No. 07-051791 (published on February 28, 1995) Special Table 2016-508882 Gazette (published on March 24, 2016)

By the way, the quality of the mold formed by using the regenerated casting sand changes depending on the sand properties of the casting sand (particularly, the amount of the binder (binder) eluted from the casting sand). Therefore, it is important to stabilize the sand properties of the regenerated cast sand, especially the elution amount of the binder. In order to stabilize the elution amount of the binder, it is necessary to appropriately set the regeneration conditions (plant control parameters, etc.) in the casting sand regeneration process. However, how to set the regeneration conditions depends on the empirical rules of the manager of the regeneration process, and it is difficult to set the regeneration conditions appropriately to stabilize the elution amount of the binder. rice field. Even with the techniques described in Patent Documents 1 and 2 described above, it has been difficult to appropriately maintain the elution amount of the binder from the recycled casting sand.

One aspect of the present invention is to realize a technique for specifying a regeneration condition for maintaining an amount of a binder eluted from a casting sand regenerated by a sand regeneration system at a target value.

In order to solve the above problems, the reproduction condition specifying device according to one aspect of the present invention includes one or a plurality of processors that execute the reproduction condition specifying step. In addition, the regeneration condition specifying method according to one aspect of the present invention includes a regeneration condition specifying step.

Then, in the regeneration condition specifying device and the regeneration condition specifying method, the regeneration condition specifying step is performed in the sand regeneration system based on the amount of the adhesive eluted from the casting sand regenerated by the processor by the sand regeneration system. It is a step to identify.

According to one aspect of the present invention, it is possible to specify the regeneration conditions for maintaining the amount of the binder eluted from the foundry sand regenerated by the sand regeneration system at the target value.

It is a figure which shows the structure of the sand management system which concerns on 1st Embodiment of this invention. It is a block diagram which shows the structure of the sand regeneration system included in the sand management system of FIG. It is a block diagram which shows the structure of the regeneration condition specifying apparatus included in the sand management system of FIG. It is a flowchart which shows the flow of the reproduction condition specifying method carried out by the reproduction condition specifying apparatus of FIG. It is a figure which illustrates the genetic algorithm. It is a flowchart which shows the flow of the relational expression specification step carried out by the reproduction condition specification apparatus. It is a flowchart which shows the flow of the process which performs the reproduction condition specifying apparatus. It is a figure which illustrates the content of a crossover. It is a figure which illustrates the content of a subtree mutation. It is a figure which illustrates the content of a hoist mutation. It is a figure which illustrates the content of a point mutation. It is a figure which shows the structure of the sand management system which concerns on the 2nd Embodiment of this invention. It is a flowchart which shows the flow of the regeneration condition specifying method carried out by the regeneration condition specifying apparatus included in the sand management system of FIG. It is a figure exemplifying a trained model and a genetic algorithm. It is a block diagram which shows the structure of the machine learning apparatus included in the sand management system of FIG. It is a flowchart which shows the flow of the machine learning method carried out by the machine learning apparatus of FIG.

[First Embodiment]
(Sand management system)
The configuration of the sand management system S1 according to the first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing a configuration of a sand management system S1.

The sand management system S1 is a system for managing casting sand used for molding a mold. As shown in FIG. 1, the sand management system S1 includes a casting system S11, a sand regeneration system S12, a regeneration condition specifying device 1, and a data logger 2.

The casting system S11 is a system for casting using casting sand (a mixture of recycled sand and new sand described later). The casting system S11 forms a mold by adding an additive containing a binder to the casting sand, kneading the casting sand, and filling the kneaded casting sand with the kneaded casting frame. The binder is, for example, an inorganic material such as water glass. The mold to be molded is, for example, a main mold or a core. Further, the casting system S11 cools the molten metal injected into the mold after pouring the molten metal into the mold. The cooled molten metal solidifies inside the mold to form a casting. The casting system S11 disassembles the mold into sand lumps by applying vibration to the mold, and takes out the casting obtained by cooling.

The sand regeneration system S12 is a system for regenerating casting sand from the sand mass obtained by the casting system S11. The configuration of the sand regeneration system S12 will be described later instead of the reference drawing.

The reproduction condition specifying device 1 is an apparatus for carrying out the reproduction condition specifying method M1. Regeneration condition specifying method M1 refers to a set of regeneration condition data indicating the regeneration conditions in the sand regeneration system S12 and binder data indicating the amount of the binder eluted from the casting sand regenerated by the sand regeneration system S12. Then, a non-linear relational expression representing the relationship between the reproduction condition data and the binder data is specified by using a non-linear regression algorithm, and the relational expression specifying step is included. Further, the reproduction condition specifying method M1 includes a reproduction condition specifying step for specifying the reproduction condition data by an operation using the binder data and the specified relational expression. The details of the configuration of the reproduction condition specifying device 1 and the flow of the reproduction condition specifying method M1 will be described later instead of the reference drawings.

The binder data is data representing the amount of the binder eluted from the regenerated foundry sand. The binder data includes, for example, data indicating at least one of an elution binder amount and a residual binder amount. The elution binder amount means the amount of the binder that elutes from the casting sand at the temperature at which the casting sand is kneaded. The residual binder amount means the amount of the binder that elutes from the casting sand at a predetermined temperature higher than the temperature at the time of kneading. The amount of eluted binder or the amount of residual binder is represented by, for example, pH, conductivity or strength.

The regeneration condition data is data indicating the regeneration conditions in the sand regeneration system S12. The details of the reproduction conditions will be described later instead of the reference drawings.

The data logger 2 collects the binder data and the reproduction condition data, and provides the collected binder data and the reproduction condition data to the reproduction condition specifying device 1. The data logger 2 can be configured by, for example, a PLC (Programmable Logic Controller), an IPC (Industrial PC), or the like.

(Sand regeneration system)
FIG. 2 is a block diagram showing the configuration of the sand regeneration system S12. As shown in FIG. 2, the sand regeneration system S12 includes a crusher 21, a fine dust collector 22, a dust collector 23, a fluidized combustion device 24, a sand cooling device 25, a sand polishing device 26, a dust collector 27, a sand classification device 28, and a sand classification device 28. It is equipped with a dust collector 29.

The crusher 21 crushes the sand mass obtained by the casting system S11 into sand grains. The aggregate of sand grains obtained by crushing the crusher 21 contains fine powders other than the cast sand such as silica and water glass in addition to the sand grains of the cast sand to be regenerated.

The fine powder removing device 22 separates fine powder (silica, water glass, etc.) other than casting sand from the aggregate of sand grains obtained by the crusher 21. The fine powder removing device 22 includes a blower 221 that blows air to the sand grains. Fine powder is collected in the dust collector 23 by the blow of the blower 221. The fine powder removing device 22 measures the air volume in the fine powder removing device 22 and outputs the measured value to the data logger 2.

The dust collector 23 collects the fine powder removed from the sand grains by the fine powder removing device 22. The dust collector 23 measures the air volume in the dust collector 23 and outputs the measured value to the data logger 2.

The fluidized incinerator 24 is an incinerator that heats granular casting sand from which fine powder has been removed. The fluidized and burned apparatus 24 conveys the cast sand from the fine powder removing device 22 by suction or pressure feeding, and heats the conveyed cast sand. By heating, the water of crystallization contained in the crystals of water glass evaporates. As the water of crystallization evaporates, voids are formed in the crystals of water glass. As a result, the binder such as water glass contained in the cast sand becomes brittle, and the sand becomes easy to be crushed.

A preset amount (hereinafter, also referred to as “casting sand input amount”) of casting sand is charged into the fluidized combustion apparatus 24. Here, the casting sand input amount is, for example, the weight or volume of the casting sand charged into the fluidized and burned apparatus 24 per unit time (in the case of continuous processing) or per treatment (in the case of batch processing). Point to that. The flow burning apparatus 24 provides the data logger 2 with the set casting sand input amount.

Further, the fluidized roasting device 24 measures the heating temperature of the fluidized roasting apparatus 24 and outputs the measured value to the data logger 2.

The sand cooling device 25 cools the heated casting sand. The sand cooling device 25 is, for example, an air-cooled cooling device that cools casting sand by taking in outside air. The sand cooling device 25 lowers the temperature of the cast sand of about 400 degrees to a temperature suitable for the next step (for example, 80 degrees or less).

The sand polishing device 26 is a device that peels the residue of a binder such as water glass from the casting sand by colliding the granular casting sands with each other and makes the casting sand spherical. The sand polishing device 26 conveys the cast sand from the sand cooling device 25 by suction or pressure feeding. Further, the sand polishing device 26 rotates a rotating drum (not shown). Due to the centrifugal force generated by the rotation of the rotating drum, the foundry sand is rolled up, and the foundry sand collides with each other to be polished, and the residue is removed. The casting sand obtained by the sand polishing device 26, that is, the casting sand after the residue adhering to the surface is removed, is also referred to as "recycled sand" below. The sand polishing device 26 measures the polishing current value and outputs the measured value to the data logger 2.

Light of the residue such as water glass peeled from the casting sand by the sand polishing device 26 is collected in the dust collector 27. The dust collector 27 measures the air volume in the dust collector 27 and outputs the measured value to the data logger 2.

The sand classification device 28 separates the regenerated sand and the residue such as water glass peeled from the regenerated sand. Further, the sand classification device 28 adds fresh sand, that is, unused cast sand, to the reclaimed sand from which the residue has been separated. The mixture of recycled sand and fresh sand is used as casting sand in the casting system S11.

The sand classification device 28 includes a blower 281 that blows air onto the regenerated sand. Residues are collected in the dust collector 29 by the blow of the blower 281. The sand classification device 28 measures the air volume in the sand classification device 28 and outputs the measured value to the data logger 2.

Further, a preset amount (hereinafter, also referred to as “regenerated sand input amount”) of regenerated sand is charged into the sand classification device 28. Here, the reclaimed sand input amount is, for example, the weight or volume of the regenerated sand charged into the sand classification device 28 per unit time (in the case of continuous processing) or per treatment (in the case of batch processing). Point to. The sand classification device 28 provides the data logger 2 with the set amount of recycled sand input.

Further, a preset amount of new sand (hereinafter, also referred to as “new sand input amount”) is charged into the sand classification device 28. Here, the amount of fresh sand charged refers to, for example, the weight or volume of fresh sand charged into the sand classification device 28 per unit time (in the case of continuous processing) or per treatment (in the case of batch processing). .. The sand classification device 28 provides the set new sand input amount to the data logger 2.

The regeneration condition data collected by the data logger 2 from the sand regeneration system S12 includes some or all of the following data.
・ Air volume in the fine powder removal device 22 ・ Air volume in the dust collector 23 ・ Cast sand input amount into the fluidized cultivating device 24 ・ Heating temperature of the fluidized cultivating device 24 ・ Polishing current value of the sand polishing device 26 ・ Air volume of the dust collector 27 ・ Sand classification The amount of recycled sand input to the device 28 and the amount of new sand input to the sand classification device 28 That is, the regeneration condition data collected by the data logger 2 is the air volume in the fine dust removing device 22 and the dust collector 23, and the casting sand to the fluidized burning device 24. The input amount and heating temperature, the polishing current value of the sand polishing device 26, the air volume for removing the residue from the casting sand, and the regenerated sand when the regenerated sand regenerated by the sand regenerating system S12 and the new sand are mixed. Includes data indicating at least one of the input amount and the input amount of fresh sand.

(Configuration of playback condition specifying device)
The configuration of the reproduction condition specifying device 1 will be described with reference to FIG. FIG. 3 is a block diagram showing the configuration of the reproduction condition specifying device 1.

The reproduction condition specifying device 1 is realized by using a general-purpose computer, and includes a processor 11, a primary memory 12, a secondary memory 13, an input / output interface 14, a communication interface 15, and a bus 16. .. The processor 11, the primary memory 12, the secondary memory 13, the input / output interface 14, and the communication interface 15 are connected to each other via the bus 16.

The reproduction condition specifying program P1, the binder data, and the reproduction condition data are stored in the secondary memory 13. The processor 11 expands the reproduction condition specifying program P1, the binder data, and the reproduction condition data stored in the secondary memory 13 on the primary memory 12. Then, the processor 11 executes each step included in the reproduction condition specifying method M1 according to the instruction included in the reproduction condition specifying program P1 expanded on the primary memory 12. The binder data and the reproduction condition data developed on the primary memory 12 are used when the processor 11 executes the relational expression specifying step M11 (described later) of the reproduction condition specifying method M1. The fact that the playback condition specifying program P1 is stored in the secondary memory 13 means that the source code or the execution format file obtained by compiling the source code is stored in the secondary memory 13. Point to.

Devices that can be used as the processor 11 include, for example, a CPU (Central Processing Unit), a GPU (Graphic Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), an FPU (Floating point number Processing Unit), and a PPU. (Physics Processing Unit), a microcontroller, or a combination thereof can be mentioned. The processor 11 is sometimes called an "arithmetic unit".

Further, as a device that can be used as the primary memory 12, for example, a semiconductor RAM (Random Access Memory) can be mentioned. The primary memory 12 is sometimes referred to as a "main storage device". Devices that can be used as the secondary memory 13 include, for example, a flash memory, an HDD (Hard Disk Drive), an SSD (Solid State Drive), an ODD (Optical Disk Drive), and an FDD (Floppy (registered trademark) Disk Drive). , Or a combination of these. The secondary memory 13 is sometimes called an "auxiliary storage device". The secondary memory 13 may be built in the reproduction condition specifying device 1, or may be another computer (for example, a cloud server) connected to the reproduction condition specifying device 1 via the input / output interface 14 or the communication interface 15. It may be built in the computer that constitutes the. In the present embodiment, the storage in the reproduction condition specifying device 1 is realized by two memories (primary memory 12 and secondary memory 13), but the present invention is not limited to this. That is, the storage in the reproduction condition specifying device 1 may be realized by one memory. In this case, for example, a certain storage area of the memory may be used as the primary memory 12, and the other storage area of the memory may be used as the secondary memory 13.

An input device and / or an output device is connected to the input / output interface 14. Examples of the input / output interface 14 include interfaces such as USB (Universal Serial Bus), ATA (Advanced Technology Attachment), SCSI (Small Computer System Interface), and PCI (Peripheral Component Interconnect). Examples of the input device connected to the input / output interface 14 include a data logger 2. The data acquired in the reproduction condition specifying method M1 is input to the reproduction condition specifying device 1 via the data logger 2 and stored in the primary memory 12. Further, examples of the input device connected to the input / output interface 14 include a keyboard, a mouse, a touch pad, a microphone, or a combination thereof. Examples of the output device connected to the input / output interface 14 include a display, a projector, a printer, a speaker, headphones, or a combination thereof. The information provided to the user in the reproduction condition specifying method M1 is output from the reproduction condition specifying device 1 via these output devices. The reproduction condition specifying device 1 may include a keyboard that functions as an input device and a display that functions as an output device, such as a laptop computer. Alternatively, the reproduction condition specifying device 1 may have a built-in touch panel that functions as both an input device and an output device, such as a tablet computer.

Another computer is connected to the communication interface 15 by wire or wirelessly via a network. Examples of the communication interface 15 include interfaces such as Ethernet (registered trademark) and Wi-Fi (registered trademark). Available networks include PAN (Personal Area Network), LAN (Local Area Network), CAN (Campus Area Network), MAN (Metropolitan Area Network), WAN (Wide Area Network), GAN (Global Area Network), or , Internetworks including these networks. Internetwork may be an intranet, an extranet, or the Internet. Data acquired by the reproduction condition specifying device 1 from another computer (for example, data logger 2) in the reproduction condition specifying method M1 (for example, binder data, reproduction condition data), and reproduction condition specification in the reproduction condition specifying method M1. The data provided by the device 1 to other computers is transmitted and received via these networks.

In the present embodiment, a configuration is adopted in which the reproduction condition specifying method M1 is executed using a single processor (processor 11), but the present invention is not limited thereto. That is, a configuration may be adopted in which the reproduction condition specifying method M1 is executed using a plurality of processors. In this case, a plurality of processors that cooperate to execute the reproduction condition specifying method M1 may be provided in a single computer and may be configured to be able to communicate with each other via a bus, or may be distributed to a plurality of computers. It may be provided so as to be able to communicate with each other via a network. As an example, a mode in which a processor built in a computer constituting a cloud server and a processor built in a computer owned by a user of the cloud server cooperate to execute the reproduction condition specifying method M1 can be considered. ..

Further, in the present embodiment, a configuration is adopted in which the binder data and the reproduction condition data are stored in the memory (secondary memory 13) built in the same computer as the processor (processor 11) that executes the reproduction condition specifying method M1. However, the present invention is not limited to this. That is, a configuration may be adopted in which the binder data and the reproduction condition data are stored in a memory built in a computer different from the processor that executes the reproduction condition specifying method M1. In this case, the computer having a built-in memory for storing the binder data and the reproduction condition data is configured to be able to communicate with each other via a network with the computer having a built-in processor that executes the reproduction condition specifying method M1. As an example, the binder data and the reproduction condition data are stored in the memory built in the computer constituting the cloud server, and the processor built in the computer owned by the user of the cloud server executes the reproduction condition specifying method M1. It is conceivable that the mode is to be used.

Further, in the present embodiment, a configuration in which the binder data and the reproduction condition data are stored in a single memory (secondary memory 13) is adopted, but the present invention is not limited thereto. That is, a configuration may be adopted in which the binder data and the reproduction condition data are distributed and stored in a plurality of memories. In this case, the plurality of memories for storing the binder data and the reproduction condition data may or may not be a single computer (a computer having a built-in processor that executes the reproduction condition specifying method M1). It may be provided in a plurality of computers (which may or may not include a computer having a processor for executing the reproduction condition specifying method M1). You may be. As an example, a configuration in which the binder data and the reproduction condition data are distributed and stored in the memory built in each of the plurality of computers constituting the cloud server can be considered.

(Measuring method of binder data)
An example of a measuring method for measuring the binder data of cast sand, which is a mixture of recycled sand and fresh sand, produced by the sand regeneration system S12 will be described.

The measurement of the binder data may be performed by a user such as the administrator of the sand regeneration system S12, or the device for measuring the binder data may execute the measurement process of the binder data. When the user performs this, the user collects a part of the regenerated casting sand and puts the collected casting sand into a temperature-controlled pure water-filled measuring cup with an electrode in a predetermined amount. The user stirs the charged casting sand in pure water for a predetermined time to elute the binder and measures the pH. Alternatively, the user applies a voltage to the electrodes to measure the conductivity. The user inputs the binder data representing the measured pH or conductivity into the data logger 2.

When the measuring device measures the binder data, the measuring device (not shown) collects a part of the regenerated casting sand, and the collected casting sand is filled with pure water whose temperature is controlled. Add a predetermined amount to the measuring cup. The measuring device stirs the charged casting sand in pure water for a predetermined time to elute the binder and measures the pH. Alternatively, the measuring device applies a voltage to the electrodes to measure the conductivity. The measuring device outputs the binder data representing the measured pH or conductivity to the data logger 2.

A method for measuring another binder data will be described. A part of the recycled casting sand is collected, and a predetermined amount of the collected casting sand is put into a measuring cup filled with pure water whose temperature is controlled. The measuring device stirs the charged casting sand in pure water for a predetermined time to elute the binder. Then, a test piece is molded from this cast sand and the strength of the test piece is measured. The strength of this test piece is the adhesive data. The pressure-sensitive adhesive data may be input to the data logger 2 by the user in the same manner as described above, or may be output to the data logger 2 by the measuring device.

(Flow of playback condition identification method)
The flow of the reproduction condition specifying method M1 will be described with reference to FIG. FIG. 4 is a flowchart showing the flow of the reproduction condition specifying method M1.

The reproduction condition specifying method M1 includes a relational expression specifying step M11, a relational expression output step M12, a determination step M13, a reproduction condition specifying step M14, and a control step M15.

In the relational expression specifying step M11, the processor 11 determines the regeneration condition data x1, x2, ..., Xn indicating the regeneration conditions in the sand regeneration system S12, and the amount of the binder eluted from the casting sand regenerated by the sand regeneration system S12. This is a step of specifying a non-linear relational expression representing the relationship between the reproduction condition data and the coagulant data by using a non-linear regression algorithm with reference to the set with the coagulant data y representing the above. In this embodiment, the processor 11 uses a genetic algorithm to specify the relational expression y = f (x1, x2, ..., Xn). A specific example of the relational expression specifying step M11 will be described later instead of the reference drawing.

The relational expression output step M12 is a step in which the processor 11 outputs the relational expression y = f (x1, x2, ..., Xn) specified in the relational expression specifying step M11. In the present embodiment, the processor 11 outputs (displays) the relational expression y = f (x1, x2, ..., Xn) to the display. At this time, the processor 11 may display a graph representing the relational expression on the display.

The user who visually confirmed the relational expression y = f (x1, x2, ..., Xn) output on the display has the relational expression y = f (x1, x2, ..., Xn) specified by the reproduction condition specifying device 1. Determine if it is an appropriate relation. Then, the user who has completed this determination performs a user operation for inputting the determination result into the reproduction condition specifying device 1.

The determination step M13 determines whether or not the relational expression y = f (x1, x2, ..., Xn) specified by the processor 11 in the relational expression specifying step M11 is an appropriate relational expression according to the above-mentioned user operation. It is a step to judge. When it is determined in the determination step M13 that "it is an appropriate relational expression", the processor 11 executes the reproduction condition specifying step M14 described later. On the other hand, when it is determined in the determination step M13 that "it is not an appropriate relational expression", the processor 11 re-executes the processing after the above-mentioned relational expression specifying step M11. When the processing after the relational expression specifying step M11 is executed again, the processor 11 changes the set of the binder data and the reproduction condition data used in the relational expression specifying step M11, or changes the parameters used in the genetic algorithm. You may perform the processing such as.

The reproduction condition specifying step M14 uses the relational expression y = f (x1, x2, ..., Xn) specified in the relational expression specifying step M11 to regenerate the binder data y in order to match the target value. This is a step of specifying what value the condition data x1, x2, ..., Xn should be set to. That is, in the present embodiment, the processor 11 specifies the reproduction condition data by the operation using the binder data and the relational expression.

In the present embodiment, the processor 11 substitutes the target value of the binder data y into the relational equation y = f (x1, x2, ..., Xn) to obtain the reproduction condition data x1, x2, ..., Xn. Obtain an equation that is an unknown. Then, the processor 11 obtains the set values of the reproduction condition data x1, x2, ..., Xn by solving this equation. When the number of reproduction condition data x1, x2, ..., Xn, which are unknowns, is two or more, the solution of the above equation cannot be uniquely determined. In this case, the processor 31 sets at least one of the solutions of the above equations as the set values of the reproduction condition data x1, x2, ..., Xn.

The control step M15 is a step in which the processor 11 controls the sand regeneration system S12 so that the value of the reproduction condition data becomes the set value specified in the reproduction condition specifying step M14. That is, in the present embodiment, the processor 31 controls the sand regeneration system S12 using the regeneration condition data specified in the regeneration condition specifying step M14.

In the present embodiment, the processor 11 controls the fine powder removing device 22, the sand polishing device 26, and the sand classifying device 28 so that the air volume becomes the set value specified in the regeneration condition specifying step M14. Further, the processor 31 controls the fluidized and burned apparatus 24 so that the casting sand input amount and the heating temperature become the set values specified in the regeneration condition specifying step M14. Further, the processor 31 controls the sand polishing device 26 so that the polishing current value becomes the set value specified in the reproduction condition specifying step M14. Further, the processor 31 controls the sand classification device 28 so that the reclaimed sand input amount and the new sand input amount become the set values specified in the regeneration condition specifying step M14.

In this embodiment, a genetic algorithm is used as an algorithm for specifying a non-linear relational expression representing the relationship between the binder data y and the reproduction condition data x1, x2, ..., Xn. However, the present invention is not limited to this. That is, as an algorithm for specifying a non-linear relational expression representing the relationship between the binder data y and the reproduction condition data x1, x2, ..., Xn, a non-linear regression algorithm other than a genetic algorithm such as logistic regression is used. May be good.

(Specific example of relational expression specific step)
A specific example of the relational expression specifying step M11 included in the reproduction condition specifying method M1 will be described with reference to FIGS. 5 and 6. FIG. 5 is a diagram illustrating the genetic algorithm GA. FIG. 6 is a flowchart illustrating the flow of the relational expression specifying step M11 executed by the processor 11. In the example of FIG. 5, the genetic algorithm GA comprises first generation G1 to fourth generation G4.

In the relational expression specifying step M11 according to this specific example, the relational expression y = f (x1, x2, ..., Xn) is specified by using a genetic algorithm. Here, the genetic algorithm prepares a plurality of individuals i in which solution candidates are expressed by genes, preferentially selects an individual i having a high fitness Di, and repeats operations such as crossover and mutation to solve the solution. Refers to the algorithm to be searched. In the present embodiment, the individual i represents a non-linear relational expression in a tree structure, and the operators and arguments included in the relational expression are represented by the nodes of the tree.
The fitness Di is given by the fitness function.

The processor 11 executes the relational expression specifying step M11 using a predetermined module (hereinafter, referred to as “A module”). The A module is a module that executes a genetic algorithm. In Module A, the processor 31 begins by creating a simple random population that represents the relationship between known independent variables and their dependent variable targets in order to predict new data. Next, the processor 31 evolves the population to generate the next generation population by selecting the most suitable individual from the population to be genetically engineered. By the above operation, the relational expression that best indicates the above relation is specified.

In this specific example, a module that executes genetic programming is used as the A module. Genetic programming is an extension of a genetic algorithm and uses a tree structure to represent genotypes. The flow of the relational expression specifying step M11 shown in FIG. 6 is an example, and the method for specifying the relational expression using the genetic algorithm GA is not limited to the method shown in FIG. Various other methods can be adopted as a method for specifying the relational expression using the genetic algorithm GA.

In step M121, the processor 31 acquires the binder data and the reproduction condition data. In this operation example, the processor 31 acquires the binder data and the reproduction condition data by reading the binder data and the reproduction condition data stored in the secondary memory 33.

In step M122, the processor 11 acquires a parameter (hereinafter referred to as “individual parameter”) used in the genetic algorithm GA. The individual parameters are, for example, the number of generated individuals N, the tournament size Nt, the crossover probability Pc, the mutation probability Pms, the number of evolutionary generations Ng, the operator Oj used for the syntax tree, the maximum depth d of the syntax tree, and the event occurrence probability Pk1 to. Includes Pk5. For example, the user inputs the value of each individual parameter to the reproduction condition specifying device 1.

The number of generated individuals N represents the number of individuals i included in the set. The tournament size Nt is the number of individuals i randomly selected from the current generation set. Mutation probability Pms is the probability that a gene will mutate. The operators Oi used in the syntax tree are, for example, Max, Min, sqrt (root), log (natural logarithm), +,-, ×, ÷, sin (radians), cos (radians), tan (radians), abs. , Neg, inv. Max is an operator that selects the maximum value. Min is an operator that selects the minimum value. neg is an operator that makes the sign negative. inv is an operator that sets an argument close to zero to 0.

The event occurrence probabilities Pk1 to Pk5 are the probabilities that the operations m1 to m5 are selected as operations for evolving the next-generation set. The processor 11 evolves the next-generation set by any of the operations m1 to m5. The sum of the event occurrence probabilities Pk1 to Pk5 is 1. As an example, the values of the event occurrence probabilities Pk1, Pk2, Pk3, Pk4, and Pk5 are "0.1", "0.2", "0.3", "0.4", and "0.1", respectively. Is. The operations m1 to m5 will be described later instead of the reference drawings.

In step M123, the processor 11 randomly generates N individual i based on the specified individual parameters (operator Oi used for the syntax tree, maximum depth d of the syntax tree, etc.), and first Generates a set of N individuals i that are the current generation.

In step M124, the processor 11 calculates the fitness Di of each individual i included in the set of the current generation. The fitness Di is given by the fitness function.

In step M125, the processor 11 randomly extracts an individual i having a tournament size Nt from the current generation set, selects the individual i having the highest fitness Di among them, and adds it to the next generation set. The individual i selected in step M125, that is, the individual i added to the next-generation set is also referred to as a "winner tree".

The processor 11 repeats the process of step M125 until the number of next-generation individuals reaches N, which is the same as that of the current generation, that is, while the number of next-generation individuals has not reached N (step M126; NO). When the number of next-generation individuals reaches N (step M126; YES), the processor 11 executes the process of step M127.

In step M127, the processor 11 executes a process of evolving the next generation set. The details of step M127 will be described later instead of the drawings to be referred to.

In step M130, the processor 11 overwrites the next generation set with the current generation set. In step M131, the processor 11 determines whether or not the number of evolutionary generations Ng has been reached. If the number of evolutionary generations Ng has not been reached (step M131; NO), the processor 11 returns to the process of step M124. On the other hand, when the number of evolutionary generations has reached Ng (step M131; YES), the processor 11 proceeds to the process of step M132.

In step M132, the processor 11 identifies the individual i having the highest fitness Di from the individuals i included in the current generation set. Through the above processing, the processor 11 specifies a non-linear relational expression representing the relationship between the binder data and the reproduction condition data.

FIG. 7 is a flowchart illustrating the flow of step M127 executed by the processor 11. In step M201, the processor 11 selects one of the operations m1 to m5 based on the event occurrence probabilities Pk1 to Pk5 set by the user. When the operation m1 is selected (step M201; "operation m1"), the processor 11 proceeds to the process of step M202. When the operation m2 is selected (step M201; "operation m2"), the processor 11 proceeds to the process of step M211. When the operation m3 is selected (step M201; “operation m3”), the processor 11 proceeds to the process of step M221. When the operation m4 is selected (step M201; "operation m4"), the processor 11 proceeds to the process of step M231. When the operation m5 is selected (step M201; "operation m5"), the processor 11 ends the process.

The operation m1 is a crossover. Crossover is a method of mixing genetic material between individuals. In the case of crossover, in step M202, the processor 11 randomly selects a subtree included in each winner tree for the winner tree included in the next generation set.

In step M203, processor 11 produces a next-generation set for donors. The content of the process in step M203 is the same as the content of steps M123 to M125 in FIG. That is, the processor 11 first randomly generates N individual i based on the individual parameter specified by the user, and generates a set of N individual i (hereinafter referred to as "donor set"). Next, the processor 11 calculates the fitness Di of each individual i included in the donor set. Next, the processor 11 randomly picks out the individual i having the number of tournament size Nt from the donor set, selects the individual i having the highest fitness Di among them, and adds it to the next-generation donor set. The individual i selected by this process is also referred to as a "donor tree". The processor 11 repeats the donor tree selection process until the number of next-generation donor trees reaches N.

In step M204, the processor 11 randomly selects a subtree included in the donor tree (hereinafter referred to as “donor subtree”).

In step M205, processor 11 swaps subtrees in the winner tree. In the present embodiment, the processor 11 removes the subtree selected in step M202 from the winner tree, and transplants the donor subtree selected in step M203 to the place where the subtree was. That is, the processor 11 replaces the subtree contained in the winner tree with the donor subtree. The winner tree in which this subtree is replaced becomes the descendant (individual) of the next generation.

The operation m2 is an operation for mutating the subtree. By mutating the subtree, extinct functions and operators can be reintroduced into the population and diversity can be maintained. In this case, in step M211 the processor 11 randomly selects a subtree included in the winner tree.

In step M212, the processor 11 randomly generates a subtree. In step M213, processor 11 swaps subtrees in the winner tree. In the present embodiment, the processor 11 removes the subtree selected in step M202 from the winner tree, and transplants the subtree generated in step M212 to the place where the removed subtree was. That is, the processor 11 replaces the subtree included in the winner tree with the subtree generated in step M212. The winner tree in which this subtree is replaced becomes the descendant (individual) of the next generation.

Operation m3 is a hoist mutation. Hoist mutation is a mutation operation that fights the bulge of the tree. In step M221, the processor randomly selects a subtree contained in the winner tree. In step M222, the processor 11 randomly selects a subtree included in the subtree selected in step M221.

In step M223, the processor 11 winds up the subtree selected in step M222 to the position of the original subtree (subtree selected in step M221). The winner tree to which this subtree is rolled up will be the descendants (individuals) of the next generation.

The operation m4 is a point mutation. Point mutation is the operation of reintroducing extinct relations and operators into a population in order to maintain diversity. In step M231, processor 11 randomly selects a node in the winner tree. In step M232, the processor 11 replaces the node selected in step M231 with another node. This replaces the relational expression represented by the winner tree with another relational expression that requires the same number of arguments as the original node. The winner tree obtained by the replacement will be the descendants (individuals) of the next generation.

The operation m5 is reproduction. In this case, the winner tree is duplicated and included in the next generation without modification.

8 to 11 are diagrams illustrating the contents of operations performed on the winner tree. FIG. 8 is a diagram illustrating the contents of the operation m1 (crossover). In the example of FIG. 8, the subtree tr111 of the winner tree tr11 is replaced with the subtree tr121 of the donor tree tr12 to become the winner tree tr13. The winner tree tr13 will be the descendant (individual) of the next generation.

FIG. 9 is a diagram illustrating the contents of the operation m2 (subtree mutation). In the example of FIG. 9, the subtree tr111 of the winner tree tr11 is replaced by the subtree tr22 to become the winner tree tr23. The winner tree tr23 will be the descendants (individuals) of the next generation.

FIG. 10 is a diagram illustrating the content of the operation m3 (hoist mutation). In the example of FIG. 10, the subtree tr1121 of the winner tree tr11 is wound up to the position of the subtree tr112 to become the winner tree tr31. The winner tree tr31 will be the descendant (individual) of the next generation.

FIG. 11 is a diagram illustrating the contents of the operation m4 (point mutation). In the example of FIG. 11, the node n21 and the node n34 included in the winner tree tr11 are replaced with the node n421 and the node n434 to become the winner tree tr41. The winner tree tr41 will be the descendant (individual) of the next generation.

As described above, according to the present embodiment, the regeneration condition specifying device 1 specifies a non-linear relational expression representing the relationship between the binder data and the regeneration condition data by the nonlinear regression algorithm, and the binder data is obtained. The reproduction condition data is specified by substituting the target value into the relational expression and performing the calculation. Thereby, it is possible to specify the regeneration condition data for stabilizing the sand properties of the cast sand regenerated by the sand regeneration system S12.

[Second Embodiment]
(Structure of sand management system)
The configuration of the sand management system S2 according to the second embodiment of the present invention will be described with reference to FIG. FIG. 12 is a block diagram showing the configuration of the sand management system S2. For convenience of explanation, the same reference numerals are given to the members having the same functions as the members described in the above embodiment, and the description thereof will not be repeated.

The sand management system S2 includes a casting system S11, a sand regeneration system S12, a regeneration condition specifying device 1B, a data logger 2, and a machine learning device 5. Of these, the casting system S11, the sand regeneration system S12, and the data logger 2 have the same configurations as those in the first embodiment described above.

The machine learning device 5 is a device for implementing the machine learning method M2. The machine learning method M2 constructs a training data set DS (a set of binder data and reproduction condition data) using the data provided by the data logger 2, and machine learning using the training data data set. This is a method for constructing a trained model LM1 by (supervised learning). As the trained model LM1, for example, an algorithm such as a neural network model such as a convolutional neural network or a recurrent neural network, a regression model such as linear regression, or a tree model such as a regression tree can be used. The configuration of the machine learning device 5 and the details of the flow of the machine learning method M2 will be described later instead of the drawings to be referred to.

The trained model LM1 is a trained model in which the correlation between the input data and the output data is machine-learned. The input of the trained model LM1 includes the above-mentioned reproduction condition data x1, x2, ..., Xn. The output of the trained model LM1 includes the binder data y described above.

[Reproduction condition specifying device]
The reproduction condition specifying device 1B includes a data collecting device 111. The data acquisition device 111 temporarily holds the input data to the trained model LM1 and the output data from the trained model LM1, and holds the input data in the next-stage nonlinear regression algorithm (for example, the genetic algorithm GA). And the output data are synchronized and input. The data collecting device 111 is, for example, a buffer provided with information storage means (for example, secondary memory 13) such as a semiconductor memory and a hard disk, and input / output means. In the example of FIG. 12, the configuration in which the data acquisition device 111 is included in the reproduction condition specifying device 1B is illustrated, but even if the data collecting device 111 is configured as a device separate from the reproduction condition specifying device 1B. good. Further, the data acquisition device 111 may be configured by a single piece of hardware, or may be configured by hardware and software.

When the data acquisition device 111 is configured as a device separate from the reproduction condition specifying device 1B, the data acquisition device 111 includes, for example, a memory (not shown) and a processor (not shown).

The reproduction condition specifying device 1B implements the reproduction condition specifying method MB1. The details of the flow of the reproduction condition specifying method MB1 will be described later instead of the reference drawings. In the secondary memory 13 of the reproduction condition specifying device 1B, the learned model LM1 is stored in addition to the reproduction condition specifying program P2, the binder data, and the reproduction condition data. The processor 11 expands the trained model LM1 stored in the secondary memory 13 on the primary memory 12. The trained model LM1 developed on the primary memory 12 is used when the processor 11 executes the estimation step MB11 (described later) of the reproduction condition specifying method MB1. The fact that the trained model LM1 is stored in the secondary memory 13 means that the parameters defining the trained model LM1 are stored in the secondary memory 13.

[Flow of playback condition identification method]
The flow of the reproduction condition specifying method MB1 will be described with reference to FIG. FIG. 13 is a flowchart showing the flow of the reproduction condition specifying method MB1. The reproduction condition specifying method MB1 includes an estimation step MB11, a relational expression specifying step MB12, a relational expression output step M12, a determination step M13, a reproduction condition specifying step M14, and a control step M15. Of these steps, the processing of the relational expression output step M12, the determination step M13, the reproduction condition specifying step M14, and the control step M15 is the same as the processing of each step described with reference to FIG. 4 in the first embodiment described above. ..

The estimation step MB11 is a step in which the processor 11 estimates the output data from the input data using the trained model LM1. In the estimation step MB11, the processor 11 acquires the reproduction condition data collected by the data logger 2, and inputs the input data including the acquired reproduction condition data x1, x2, ..., Xn to the trained model LM1 to output the data. Get the data. The output data output by the trained model LM1 includes the above-mentioned binder data. The input data input to the trained model LM1 and the output data output from the trained model LM1 are temporarily held in the data acquisition device 111.

The relational expression specifying step MB12 specifies a non-linear relational expression representing the relationship between the reproduction condition data input to the trained model LM1 by the processor 11 and the binder data estimated in the estimation step MB11 by the genetic algorithm GA. It is a step to do. The data collection device 111 inputs the input data and the output data of the trained model LM1 temporarily held to the genetic algorithm GA with temporal synchronization. In this way, the data collection device 111 operates as a kind of buffer that collects the data of the trained model LM1 (in a certain amount of information unit) and inputs it to the next-stage genetic algorithm GA.

FIG. 14 is a diagram illustrating the trained model LM1 and the genetic algorithm GA. The trained model LM1 is a deep neural network composed of a plurality of layers. The trained model LM1 includes a plurality of layers including an input layer LX to which input data is input, a hidden layer LY, and an output layer LZ to output output data. Each layer has a structure in which multiple nodes are connected by edges.

Each layer has a function called the activation function, and the edges can have weights. The output value of each node is calculated from the output value of the node of the layer before connecting to that node. That is, the output value of each node is calculated from the output value of the node of the previous layer, the value of the weight of the connection edge, and the activation function of the layer.

In the example of FIG. 14, the input layer LX includes nodes X1, X2, X3, X4. That is, in this example, the input data input to the trained model LM1 includes the output values x1, x2, x3, x4 of the nodes X1, X2, X3, X4.

The hidden layer LY includes nodes Y1, Y2, and Y3. The output values y1, y2, y3 of the nodes Y1, Y2, Y3 are calculated from the output values x1, x2, x3, x4 of the node of the input layer LX, which is the layer before the hidden layer LY.

The output layer LZ includes the node Z1. The output value z1 of the node Z1 is calculated from the output values y1, y2, y3, y4 of the node of the hidden layer LY, which is the layer before the output layer LZ. That is, in this example, the output data of the trained model LM1 is the output value z1 of the node Z1.

The data acquisition device 111 temporarily holds the input data to the trained model LM1 and the output data of the trained model LM1, and inputs the held data to the next-stage genetic algorithm GA.

The method for specifying the relational expression in the relational expression specifying step MB12 is the same as the method for specifying the relational expression specifying step M11 shown in the first embodiment described above. When the relational expression is specified, the processor 11 executes the relational expression output step M12 and the determination step M13. Further, the processor 11 proceeds to the reproduction condition specifying step M14, substitutes the binder data into the specified relational expression, and performs the calculation to specify the reproduction condition data. Further, in the control step M15, the processor 11 controls the sand regeneration system S12 using the specified regeneration condition data.

According to the present embodiment, the reproduction condition specifying device 1 estimates the output of the trained model LM1 using the non-linear relational expression specified by the genetic algorithm GA, and uses the estimated output of the trained model LM1. , Specify the regeneration conditions for maintaining the pressure-sensitive adhesive data at the target value. That is, the reproduction condition specifying device 1 can estimate the reproduction condition from the input data without using the trained model LM1.

Further, in the present embodiment, the data acquisition device 111 buffers the input data and the output data of the trained model LM1 and inputs them to the genetic algorithm GA with temporal synchronization. Since the output from the trained model LM1 and the input to the genetic algorithm GA can be time-synchronized, the time until the next-stage genetic algorithm GA outputs the final solution can be pre-read.

Further, in the present embodiment, the data collection device 111 functions as a kind of buffer for inputting the data of the trained model LM1 together (in a certain amount of information unit) to the next-stage genetic algorithm GA. As a result, wasteful memory space and wasteful arithmetic processing can be reduced in the generational arithmetic processing of the next-stage genetic algorithm GA, and the generational arithmetic processing of the genetic algorithm can be made more efficient.

[Configuration of machine learning device]
The configuration of the machine learning device 5 will be described with reference to FIG. FIG. 15 is a block diagram showing the configuration of the machine learning device 5.

The machine learning device 5 is realized by using a general-purpose computer, and includes a processor 51, a primary memory 52, a secondary memory 53, an input / output interface 54, a communication interface 55, and a bus 56. The processor 51, the primary memory 52, the secondary memory 53, the input / output interface 54, and the communication interface 55 are connected to each other via the bus 56.

The machine learning program P5 and the learning data set DS are stored in the secondary memory 53. The learning data set DS is a set of teacher data DS1, DS2, and so on. The teacher data DS1, DS2 ... Is a set of the binder data and the reproduction condition data. The processor 51 expands the machine learning program P5 stored in the secondary memory 53 on the primary memory 52. Then, the processor 51 executes each step included in the machine learning method M2 according to the instruction included in the machine learning program P5 expanded on the primary memory 52. The learning data set DS stored in the secondary memory 53 is constructed in the learning data set construction step M21 (described later) of the machine learning method M2, and in the trained model construction step M22 (described later) of the machine learning method M2. It will be used. Further, the trained model LM1 constructed in the trained model construction step M22 of the machine learning method M2 is also stored in the secondary memory 53. The fact that the machine learning program P5 is stored in the secondary memory 53 means that the source code or the executable file obtained by compiling the source code is stored in the secondary memory 53. .. Further, the fact that the trained model LM1 is stored in the secondary memory 53 means that the parameters defining the trained model LM1 are stored in the secondary memory 53.

Devices that can be used as the processor 51 include, for example, a CPU (Central Processing Unit), a GPU (Graphic Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), an FPU (Floating point number Processing Unit), and a PPU. (Physics Processing Unit), a microcontroller, or a combination thereof can be mentioned. The processor 51 is sometimes called an "arithmetic unit".

Further, as a device that can be used as the primary memory 52, for example, a semiconductor RAM (Random Access Memory) can be mentioned. The primary memory 52 is sometimes referred to as a "main storage device". Devices that can be used as the secondary memory 53 include, for example, a flash memory, an HDD (Hard Disk Drive), an SSD (Solid State Drive), an ODD (Optical Disk Drive), an FDD (Floppy Disk Drive), or these. The combination of can be mentioned. The secondary memory 53 is sometimes referred to as an "auxiliary storage device". The secondary memory 53 may be built in the machine learning device 5, or constitutes another computer (for example, a cloud server) connected to the machine learning device 5 via the input / output interface 54 or the communication interface 55. It may be built into the computer). In the present embodiment, the storage in the machine learning device 5 is realized by two memories (primary memory 52 and secondary memory 53), but the present invention is not limited to this. That is, the memory in the machine learning device 5 may be realized by one memory. In this case, for example, a certain storage area of the memory may be used as the primary memory 52, and the other storage area of the memory may be used as the secondary memory 53.

An input device and / or an output device is connected to the input / output interface 54. Examples of the input / output interface 54 include interfaces such as USB (Universal Serial Bus), ATA (Advanced Technology Attachment), SCSI (Small Computer System Interface), and PCI (Peripheral Component Interconnect). Examples of the input device connected to the input / output interface 54 include a data logger 2. The data acquired in the machine learning method M2 is input to the machine learning device 5 via the data logger 2 and stored in the primary memory 52. Examples of the input device connected to the input / output interface 54 include a keyboard, a mouse, a touch pad, a microphone, or a combination thereof. The data acquired from the user in the machine learning method M2 is input to the machine learning device 5 via these input devices and stored in the primary memory 52. Examples of the output device connected to the input / output interface 54 include a display, a projector, a printer, a speaker, headphones, or a combination thereof. The information provided to the user in the machine learning method M2 is output from the machine learning device 5 via these output devices. The machine learning device 5 may include a keyboard that functions as an input device and a display that functions as an output device, such as a laptop computer. Alternatively, the machine learning device 5 may have a built-in touch panel that functions as both an input device and an output device, such as a tablet computer.

Another computer is connected to the communication interface 55 by wire or wirelessly via a network. Examples of the communication interface 55 include interfaces such as Ethernet (registered trademark) and Wi-Fi (registered trademark). Available networks include PAN (Personal Area Network), LAN (Local Area Network), CAN (Campus Area Network), MAN (Metropolitan Area Network), WAN (Wide Area Network), GAN (Global Area Network), or , Internetworks including these networks. Internetwork may be an intranet, an extranet, or the Internet. The data (for example, the trained model LM1) provided by the machine learning device 5 to another computer (for example, the reproduction condition specifying device 1) is transmitted / received via these networks.

In the present embodiment, a configuration in which the machine learning method M2 is executed using a single processor (processor 51) is adopted, but the present invention is not limited to this. That is, a configuration may be adopted in which the machine learning method M2 is executed using a plurality of processors. In this case, a plurality of processors that execute the machine learning method M2 in cooperation with each other may be provided in a single computer and may be configured to be able to communicate with each other via a bus, or may be distributed to a plurality of computers. It may be provided and configured to be able to communicate with each other via a network. As an example, a mode in which a processor built in a computer constituting a cloud server and a processor built in a computer owned by a user of the cloud server cooperate to execute the machine learning method M2 can be considered.

Further, in the present embodiment, a configuration is adopted in which the learning data set DS is stored in the memory (secondary memory 53) built in the same computer as the processor (processor 51) that executes the machine learning method M2. , The present invention is not limited to this. That is, a configuration may be adopted in which the learning data set DS is stored in a memory built in a computer different from the processor that executes the machine learning method M2. In this case, the computer with the built-in memory for storing the learning data set DS is configured to be able to communicate with each other via the network with the computer with the built-in processor that executes the machine learning method M2. As an example, the learning data set DS is stored in the memory built in the computer constituting the cloud server, and the processor built in the computer owned by the user of the cloud server executes the machine learning method M2. Conceivable.

Further, in the present embodiment, a configuration in which the learning data set DS is stored in a single memory (secondary memory 53) is adopted, but the present invention is not limited to this. That is, a configuration may be adopted in which the learning data set DS is distributed and stored in a plurality of memories. In this case, a plurality of memories for storing the learning data set DS are provided in a single computer (which may or may not be a computer having a built-in processor that executes the machine learning method M2). It may be distributed to a plurality of computers (which may or may not include a computer having a processor for executing the machine learning method M2). As an example, a configuration in which the learning data set DS is distributed and stored in the memory built in each of a plurality of computers constituting the cloud server can be considered.

Further, in the present embodiment, a configuration is adopted in which the reproduction condition specifying method M1 and the machine learning method M2 are executed by using different processors (processors 11 and 51), but the present invention is not limited thereto. That is, the reproduction condition specifying method M1 and the machine learning method M2 may be executed using the same processor. In this case, by executing the machine learning method M2, the trained model LM1 is stored in the memory built in the same computer as this processor. Then, when the reproduction condition specifying method M1 is executed, this processor uses the learned model LM1 stored in the memory to specify the relational expression and the reproduction condition.

[Flow of machine learning method]
The flow of the machine learning method M2 will be described with reference to FIG. FIG. 16 is a flowchart showing the flow of the machine learning method M2.

The machine learning method M2 includes a training data set construction step M21 and a trained model construction step M22.

The learning data set construction step M21 is a step in which the processor 51 constructs a learning data set DS, which is a set of teacher data DS1, DS2, .... Each teacher data DSi (i = 1, 2, ...) Contains binder data and regeneration condition data. The binder data included in the teacher data DSi is the same data as the input data input to the trained model LM1. In the learning data set construction step M21, the processor 51 acquires this data by the same method as that of the reproduction condition specifying device 1. Further, the teacher data DSi includes reproduction condition data. In the learning data set construction step M21, the processor 11 associates the binder data with the reproduction condition data and stores them in the secondary memory 13. By repeating the above process by the processor 11, the learning data set DS is constructed.

The trained model construction step M22 is a step in which the processor 51 constructs the trained model LM1. In the trained model building step M22, the processor 51 builds the trained model LM1 by supervised learning using the training data set DS. Then, the processor 11 stores the constructed trained model LM1 in the secondary memory 53.

〔summary〕
The regeneration condition specifying device according to the first aspect is one or a plurality of processors that execute a regeneration condition specifying step for specifying the regeneration conditions in the sand regeneration system from the amount of the pressure-sensitive adhesive eluted from the cast sand regenerated by the sand regeneration system. It is equipped with.

According to the above configuration, the regeneration condition specifying device specifies the regeneration condition data for maintaining the sand properties of the cast sand regenerated by the sand regeneration system at the target value.

The reproduction condition specifying device according to the second aspect has the following characteristics in addition to the characteristics of the reproduction condition specifying device according to the first aspect. That is, in the regeneration condition specifying device according to the second aspect, the processor sets the regeneration condition data and the binder data representing the amount of the binder eluted from the casting sand regenerated by the sand regeneration system. With reference to this, a non-linear regression algorithm for specifying a non-linear relational expression representing the relationship between the reproduction condition data and the binder data is further executed, and in the reproduction condition specifying step, the relational expression specifying step is further executed. The reproduction condition data is specified by the calculation using the binder data and the relational expression.

According to the above configuration, the regeneration condition specifying device specifies the regeneration condition data for maintaining the sand properties of the cast sand regenerated by the sand regeneration system at the target value.

The reproduction condition specifying device according to the third aspect has the following characteristics in addition to the characteristics of the reproduction condition specifying device according to the first or second aspect. That is, in the regeneration condition specifying device according to the third aspect, the processor executes a control step of controlling the sand regeneration system using the regeneration condition data specified in the regeneration condition specifying step.

According to the above configuration, the regeneration condition specifying device can stabilize the sand properties of the cast sand regenerated by the sand regeneration system by controlling the sand regeneration system using the specified regeneration condition data.

The reproduction condition specifying device according to the fourth aspect has the following characteristics in addition to the characteristics of the reproduction condition specifying device according to the second aspect. That is, in the regeneration condition specifying device according to the fourth aspect, the binder data is the elution binder amount representing the amount of the binder eluted from the casting sand at the temperature at which the casting sand is kneaded, or the temperature. Includes data indicating at least one of the residual binder amounts, which represent the amount of binder that elutes from the foundry sand at higher temperatures.

According to the above configuration, the regeneration condition specifying device can stabilize the elution binder amount or the residual binder amount of the cast sand regenerated by the sand regeneration system.

The reproduction condition specifying device according to the fifth aspect has the following characteristics in addition to the characteristics of the reproduction condition specifying device according to the second aspect. That is, in the regeneration condition specifying device according to the fifth aspect, the regeneration condition data is the air volume for separating the casting sand and the fine powder in the sand regeneration system, and the casting sand to the incineration device for heating the casting sand. The input amount, the heating temperature, the polishing current value of the sand polishing device for polishing the casting sand, the air volume for removing the residue from the casting sand, and the regenerated sand and the new sand regenerated by the sand regeneration system. It contains data indicating at least one of the input amount of the regenerated sand and the input amount of the new sand at the time of mixing.

According to the above configuration, the regeneration condition specifying device heats the casting sand and the air volume for classifying the casting sand and the fine powder in the sand regeneration system as the regeneration conditions for maintaining the sand properties of the casting sand at the target value. The amount of casting sand input to the incineration device, the heating temperature, the polishing current value of the sand polishing device for polishing the casting sand, the air volume for removing the residue from the casting sand, and the regeneration regenerated by the sand regeneration system. At least one of the amount of sand and the amount of new sand can be specified.

The reproduction condition specifying device according to the sixth aspect has the following characteristics in addition to the characteristics of the reproduction condition specifying device according to the second or fifth aspect. That is, in the reproduction condition specifying device according to the sixth aspect, the processor specifies the non-linear relational expression by the genetic algorithm in the relational expression specifying step.

According to the above configuration, the regeneration condition specifying device can specify the regeneration condition data for maintaining the sand properties of the cast sand regenerated by the sand regeneration system at the target value.

The reproduction condition specifying device according to the seventh aspect has the following characteristics in addition to the characteristics of the reproduction condition specifying device according to the second, fifth or sixth aspect. That is, in the regeneration condition specifying device according to the seventh aspect, the processor uses a trained model in which the correlation between the binder data and the regeneration condition data is machine-learned, and the regeneration condition is obtained from the binder data. Further executing the estimation step of estimating the data, the processor represents the relationship between the binder data input to the trained model in the relational expression specifying step and the reproduction condition data estimated in the estimation step. The non-linear relational expression is specified.

According to the above configuration, the regeneration condition specifying device can specify the regeneration condition data for maintaining the sand properties of the cast sand regenerated by the sand regeneration system at the target value without using the trained model.

The regeneration condition specifying method according to the eighth aspect is a regeneration condition specifying step in which one or a plurality of processors specify the regeneration conditions in the sand regeneration system from the amount of the pressure-sensitive adhesive eluted from the cast sand regenerated by the sand regeneration system. Includes.

According to the above configuration, it is possible to specify the regeneration condition data for stabilizing the sand properties of the cast sand regenerated by the sand regeneration system.

[Appendix 1]
In each of the above-described embodiments, in the reproduction condition specifying step M14, the processor 11 substitutes the target value of the binder data y into the relational equation y = f (x1, x2, ..., Xn) to reproduce the reproduction condition data. We obtained equations with x1, x2, ..., Xn as unknowns. Then, the processor 31 obtained the set values of the reproduction condition data x1, x2, ..., Xn by solving this equation. The data to be substituted into the relational expression is not limited to that shown in the above-described embodiment. For example, in addition to the target value of the binder data y, some set values of the reproduction condition data x1, x2, ..., Xn may be substituted into the above relational expression. In this case, the processor 31 obtains the remaining set values of the reproduction condition data x1, x2, ..., Xn by solving the equation obtained by substituting each value into the relational expression.

[Appendix 2]
In the second embodiment described above, as the trained model LM1, a trained model machine-learned by supervised learning is used, but a trained model machine-learned by unsupervised learning may be used. Also for a trained model machine-learned by unsupervised learning, the processor 11 specifies a non-linear relational expression representing the relationship between the input and the output of the trained model by a genetic algorithm.

[Appendix 3]
In the second embodiment described above, the machine learning device 5 has built the trained model LM1, but the trained model LM1 may be built in advance by a device other than the machine learning device 5. In this case, the reproduction condition specifying device 1 implements the above-mentioned reproduction condition specifying method MB1 using a trained model constructed in advance by another device.

[Appendix 4]
In the above-mentioned second embodiment, the trained model LM1 in which the reproduction condition data x1, x2, ..., Xn is input and the binder data y is output is used, but the trained model according to the present invention is described above. It is not limited to the one shown in the above-described embodiment. The trained model may be, for example, a trained model in which a part of the reproduction condition data and the binder data are input and one of the reproduction condition data is output. Further, the trained model may be, for example, a trained model in which the binder data is input and the reproduction condition data is output.

[Appendix 5]
In the second embodiment described above, the processor 11 identifies the relational expression by a genetic algorithm using a non-linear relational expression representing the relation between the input data and the output data of the trained model as a solution candidate. The method for specifying the relational expression representing the relationship between the input data and the output data is not limited to the method shown in the second embodiment described above. For example, the processor 11 has a non-linear first relational expression representing a relationship between a numerical value representing input data and each output value of some or all nodes belonging to the hidden layer, and each output value of the node. The first relational expression and the second relational expression may be specified by a genetic algorithm using a second non-linear relational expression representing a relation with a numerical value representing output data as a solution candidate. In this case, the processor 11 solves, for example, a simultaneous equation of the specified first relational expression and the second relational expression, and thereby represents a non-linear relational expression representing the relationship between the numerical value representing the input data and the numerical value representing the output data. To identify.

Further, the algorithm for specifying the non-linear relational expression representing the relationship between the input data and the output data of the trained model is not limited to the genetic algorithm, and may be another non-linear algorithm. As an example, the processor 11 may specify a non-linear relational expression expressing the relation between the input data and the output data by using the Monte Carlo method. In this case, as an example, the processor 11 creates a plurality of relational expressions by randomly selecting the length of the relational expression and the elements (variables, operators, etc.) of the relational expression, and the created plurality of relational expressions. By selecting the relational expression with the smallest error, the relational expression that derives the output data from the input data is constructed.

Further, in the second embodiment described above, the processor 11 specifies a non-linear relational expression representing the relationship between the input and the output of the trained model LM1 by non-linear regression, and substitutes the amount of the adhesive into the specified relational expression. The reproduction conditions were specified by the calculation. The method for specifying the reproduction conditions is not limited to that shown in the above-described embodiment. As an example, the processor 11 may specify the reproduction condition represented by the reproduction condition data obtained by inputting the pressure-sensitive adhesive data into the trained model LM1.

[Appendix 6]
The present invention is not limited to the above-described embodiment, and various modifications can be made within the scope of the claims, and other technical means obtained by appropriately combining the technical means disclosed in the above-mentioned embodiments. Embodiments are also included in the technical scope of the present invention.

1, 1B Playback condition specification device 2 Data logger 5 Machine learning device 11, 51 Processor 12, 52 Primary memory 13, 53 Secondary memory 14, 54 Input / output interface 15, 55 Communication interface 16, 56 Bus M1, MB1 Playback condition specification Method M11, MB12 Relational expression specific step M12 Relational expression output step M13 Judgment step M14 Playback condition specification step M15 Control step M2 Machine learning method M21 Learning data set construction step M22 Trained model construction step MB11 Estimating step S1, S2 Sand management system S11 Casting system S12 Sand regeneration system

Claims (8)

It comprises one or more processors that perform a regeneration condition specifying step that identifies the regeneration conditions in the sand regeneration system from the amount of adhesive eluted from the foundry sand regenerated by the sand regeneration system.
A playback condition specifying device characterized by this. The processor
The regeneration condition data and the binding are referred to with reference to a set of the regeneration condition data indicating the regeneration conditions and the binder data representing the amount of the binder eluted from the casting sand regenerated by the sand regeneration system. Further execution of the relational expression identification step, in which the non-linear relational expression representing the relationship with the agent data is specified by using the non-linear regression algorithm, is further executed.
In the regeneration condition specifying step, the regeneration condition data is specified by an operation using the binder data and the relational expression.
Equipped with one or more processors to run
The reproduction condition specifying apparatus according to claim 1. The reproduction condition specifying device according to claim 1 or 2, wherein the processor executes a control step of controlling the sand regeneration system using the reproduction condition data specified in the reproduction condition specifying step. The binder data includes the amount of the binder that elutes from the casting sand at the temperature at which the casting sand is kneaded, and the amount of the binder that elutes from the casting sand at a temperature higher than the temperature. Contains data indicating at least one of
The reproduction condition specifying apparatus according to claim 2. The regeneration condition data includes the air volume for separating the casting sand and the fine powder in the sand regeneration system, the amount of the casting sand input to the incinerator that heats the casting sand, the heating temperature, and the polishing of the casting sand. The polishing current value of the sand polishing apparatus to be used, the air volume for removing the residue from the casting sand, the input amount of the regenerated sand when mixing the regenerated sand regenerated by the sand regeneration system and the new sand, and the above. Includes data indicating at least one of the amount of fresh sand input,
The reproduction condition specifying apparatus according to claim 2. The processor identifies the non-linear relational expression by a genetic algorithm in the relational expression specifying step.
The reproduction condition specifying apparatus according to claim 2 or 5. The processor further executes an estimation step of estimating the binder data from the regeneration condition data using a trained model in which the correlation between the binder data and the regeneration condition data is machine-learned.
The processor identifies the non-linear relational expression representing the relationship between the reproduction condition data input to the trained model and the binder data estimated in the estimation step in the relational expression specifying step.
The reproduction condition specifying apparatus according to claim 2, 5 or 6. A regeneration condition specifying step, wherein one or more processors specify the regeneration conditions in the sand regeneration system from the amount of the pressure-sensitive adhesive eluted from the foundry sand regenerated by the sand regeneration system.
A method for specifying a reproduction condition, which comprises.

Images