Cutting fluid quantity adjustment device and system

Abstract

Cutting fluid quantity adjustment device (1, 1', 1'') for adjusting a delivery quantity of the cutting fluid from at least one cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) which is included in a cutting fluid supply device (3) for supplying the cutting fluid to a machining area of ​​a machine tool for machining a workpiece, characterized in that it a data acquisition unit (34) configured to acquire at least data indicating a machining state by the machine tool and data relating to the cutting fluid supplied by the cutting fluid supply device (3); a preprocessing unit (36) configured to generate data for machine learning based on the data obtained by the data acquisition unit (34); and a machine learning device (100, 100') configured to perform machine learning processing related to the quantity of cutting fluid dispensed from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) in an environment where the workpiece is machined by the machine tool, based on the data generated by the preprocessing unit (36), includes.

Description

BACKGROUND OF THE INVENTION 1. Field of the invention

[0001] The present invention relates to a cutting fluid quantity adjusting device and a cutting fluid quantity adjusting system. 2. Description of the state of the art

[0002] A machine tool having at least one cutting fluid nozzle sprays a water-soluble or water-insoluble cutting fluid onto the tool during cutting or the like of a workpiece to improve the lubricity of a tool, cool the tool, flush away chips, and the like. The amount of cutting fluid used during machining of the workpiece by the machine tool varies depending on the cutting conditions and specifications, the type of cutting fluid, the ambient temperature, and the like. Therefore, an operator determines an appropriate discharge amount of cutting fluid per unit time used during workpiece machining based on experience.

[0003] Typically, the cutting fluid discharge amount per unit time specified by the operator remains unchanged during machining of the workpiece. Therefore, a predetermined amount of cutting fluid is discharged from a cutting fluid discharge nozzle, for example, to a position where the cutting fluid is not sprayed onto the workpiece. This causes a deterioration in the efficiency of cutting fluid use. As a countermeasure to this problem, Japanese Patent Laid-Open No. 2017-113849 discloses a technique for adjusting the amount of cutting fluid discharged from nozzles based on the discharge areas of the cutting fluid from the respective nozzles and the shape of a workpiece.

[0004] However, the cutting heat generated during machining of the workpiece varies depending on the type of tool, machining conditions, the purpose of machining (rough machining or finishing), the workpiece material, etc. Therefore, it is difficult to adjust the cutting fluid discharge rate to a discharge rate suitable for cooling the workpiece and tool unless these circumstances are taken into account.

[0005] Regarding the cutting fluid, there are different types of cutting fluids, such as water-soluble cutting fluids, water-insoluble cutting fluids, and cutting fluids with different viscosities. The cooling and lubricating effects vary depending on the type of cutting fluid. The cooling effect also varies depending on the temperature of the cutting fluid. Therefore, the operator or machine tool must adjust the amount of cutting fluid discharged from the nozzles, taking these different conditions into account.

[0006] WO 2012 / 119 723 A2 describes a method for supplying coolant to cutting tools, in which a constant coolant quantity is detected by flow sensors and regulated in a speed-controlled or speed-regulated manner on at least one tool. DE 10 2018 001 893 A1 describes a cleaning process optimization device comprising a machine learning device that learns a cleaning condition when cleaning an object to be cleaned. JP 2013-193 175 A describes a cooling device with oil flow sensors and oil temperature sensors for detecting the temperature of grinding oil before or after its use on a workpiece, wherein pumps control the grinding oil flow rate based on an estimated amount of heat generated by the grinding part and a heat balance of the grinding part such that the workpiece is brought to a target temperature. SUMMARY OF THE INVENTION

[0007] Therefore, it is an object of the present invention to provide a cutting fluid amount adjusting device and a cutting fluid amount adjusting system capable of adjusting a discharge amount of a cutting fluid to an appropriate discharge amount in consideration of cutting heat generated by machining a workpiece, a type of the cutting fluid, the temperature of the cutting fluid, and the like.

[0008] The present invention provides a cutting fluid amount adjusting device for adjusting a discharge amount of the cutting fluid from at least one cutting fluid nozzle included in a cutting fluid supply device for supplying the cutting fluid to a machining section of a machine tool for machining a workpiece.The cutting fluid amount adjusting device includes a data acquisition unit configured to acquire at least data indicating a machining state by the machine tool and data related to the cutting fluid supplied from the cutting fluid supply device; a pre-processing unit configured to generate data used in machine learning based on the data acquired by the data acquisition unit; and a machine learning device configured to perform machine learning processing related to the discharge amount of the cutting fluid from the cutting fluid nozzle in an environment in which machining of the workpiece is performed by the machine tool, based on the data generated by the pre-processing unit.

[0009] Alternatively, the present invention provides a cutting fluid amount adjusting system in which a plurality of devices are connected to each other via a network, wherein the plurality of devices includes a cutting fluid amount adjusting device having at least one learning unit.

[0010] According to one aspect of the present invention, it is possible to set the appropriate amount of the cutting fluid in consideration of the cutting heat generated by machining a workpiece, a type of the cutting fluid, the temperature of the cutting fluid, and the like. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The above and other objects and features of the present invention will become clear from the following description of embodiments taken in conjunction with the accompanying drawings. Fig. 1 is a schematic hardware configuration diagram of a cutting fluid amount adjusting device according to an embodiment of the present invention; Fig. 2 is a schematic functional block diagram of a control device according to a first embodiment of the present invention; Fig. 3 is a diagram showing condition data relating to a workpiece; Fig. 4 is a diagram showing condition data related to a cutting fluid nozzle; Fig. 5 is a diagram illustrating an operation of a learning unit according to the first embodiment; Fig. 6 is a schematic functional block diagram of a cutting fluid amount adjusting device according to a second embodiment of the present invention; Fig. 7 is a diagram illustrating an operation of a discharge quantity determining unit; Fig. 8 is a diagram illustrating an operation of a learning unit according to a third embodiment of the present invention; Fig. 9 is a schematic functional block diagram of a cutting fluid amount adjusting device according to a fourth embodiment of the present invention; Fig. 10 is a schematic functional block diagram of a cutting fluid amount adjusting device according to a fifth embodiment of the present invention; Fig. 11 is a schematic functional block diagram of a cutting fluid amount adjusting device according to a sixth embodiment of the present invention; Fig. 12 is a diagram showing an example of a system with a three-tier structure including a cloud server, a fog computer, and an edge computer; Fig. 13 is a schematic hardware configuration diagram of one form of cutting fluid amount adjusting device incorporated in a computer; Fig. 14 is a schematic structural diagram of a cutting fluid amount adjusting system according to a seventh embodiment of the present invention; Fig. 15 is a schematic structural diagram of a cutting fluid amount adjusting system according to an eighth embodiment of the present invention; and Fig. 16 is a schematic structural diagram of a cutting fluid amount adjusting system according to a ninth embodiment of the present invention. DETAILED EXPLANATION OF THE PREFERRED EMBODIMENTS

[0012] Hereinafter, some embodiments of the present invention will be explained with reference to the drawings.

[0013] Fig. 1 is a schematic hardware configuration diagram showing a main portion of a cutting fluid amount adjusting device having a machine learning device according to an embodiment. A cutting fluid amount adjusting device 1 according to this embodiment can be implemented, for example, in a control device for controlling a machine tool. The adjusting device 1 in this embodiment can be implemented as a computer such as a personal computer provided together with a control device for controlling a machine tool, a management device connected to the control device via a wired or wireless network, an edge computer, a fog computer, and a cloud server. This embodiment exemplifies that the cutting fluid amount adjusting device 1 is implemented in the control device for controlling the machine tool.

[0014] The cutting fluid quantity adjusting device 1 includes a CPU (central processing unit) 11, a ROM (read-only memory) 12, a RAM (random access memory) 13, and a non-volatile memory 14. The cutting fluid quantity adjusting device 1 further includes an interface 15, a PMC (programmable machine control unit) 16, an I / O (input / output) unit 17, and interfaces 18, 19, and 21. The adjusting device 1 is connected to a cutting fluid supply device 3 for supplying the cutting fluid via the I / O unit 17. The adjusting device 1 further includes a servo motor 50, a spindle motor 62, a display / MDI (manual data input) unit 70, and a control panel 71.

[0015] The CPU 11 included in the cutting fluid amount adjusting device 1 according to this embodiment is a processor for controlling the adjusting device 1 as a whole. The CPU 11 reads a system program stored in the ROM 12 via a bus 20 and controls the entire cutting fluid amount adjusting device 1 according to the system program. Temporary calculation data and display data, various data input by an operator via an input unit (not shown), and the like are temporarily stored in the RAM 13.

[0016] The non-volatile memory 14 is backed up by, for example, a battery not shown, to be configured as a memory in which the storage state is maintained even when the cutting fluid amount adjusting device 1 is turned off.

[0017] The non-volatile memory 14 stores a program read from an external device 72 via the interface (INT) 15, a program input via the display / MDI unit 70, and various data acquired from components of the cutting fluid amount adjusting device 1, the machine tool, the cutting fluid supply device 3, and the like. The various data stored in the non-volatile memory 14 include, for example, information related to tools such as tool types; cutting conditions such as spindle speed, feed rate, and cutting depth; machining processes such as rough machining and fine machining; a workpiece such as a workpiece material and workpiece shape; cutting fluid such as a type and temperature of the cutting fluid; discharge areas of the nozzles; and discharge amount of the cutting fluid.

[0018] The programs and various data stored in non-volatile memory 14 can be developed into RAM 13 during execution or use. Various system programs, such as a known analysis program, have been pre-written into ROM 12. The system programs pre-written into ROM 12 include a system program for controlling communication with a machine learning device 100 described below.

[0019] The interface 15 is an interface for connecting the cutting fluid amount adjusting device 1 to the external device 72 such as an adapter. Programs, various parameters, and the like are read from the external device 72 into the adjusting device 1 via the interface 15. The programs, various parameters, and the like edited in the adjusting device 1 can be stored in an external memory via the external device 72. The PMC 16 controls the input and output of signals between the PMC 16 and the machine tool and peripheral devices of the machine tool, such as the cutting fluid supply device 3, via the I / O unit 17 according to a sequence program incorporated into the adjusting device 1.

[0020] The cutting fluid supply device 3 is a device capable of supplying cutting fluid to a machining area of ​​the machine tool. The supply device 3 includes a tank for storing the cutting fluid, a pump for supplying the cutting fluid from the tank via a cutting fluid supply path, and one or more fluid nozzles connected to the cutting fluid path for discharging the cutting fluid to the machining area of ​​the machine tool. The cutting fluid amount adjusting device 1 adjusts the discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles via a signal output from the PMC 16. The cutting fluid supply device 3 outputs the amounts of the cutting fluid discharged from the respective cutting fluid nozzles to the CPU 11 via the MPC 16.

[0021] The display / MDI unit 70 is a manual data input device that includes a display, a keyboard, and the like. The interface (INT) 18 receives a command and data from the keyboard in the display / MDI unit 70 and forwards the command and data to the CPU 11. The interface (INT) 19 is connected to the control console 71, which includes a manual pulse generator used for manually driving shafts and the like.

[0022] The cutting fluid amount adjusting device 1 further includes a shaft control circuit 30 for controlling a shaft of the machine tool and a servo amplifier 40 provided for driving the servo motor 50. The shaft control circuit 30 receives a shaft movement amount command from the CPU 11 and then outputs a shaft movement command to the servo amplifier 40. The servo amplifier 40 receives the command and then drives the servo motor 50 capable of moving the shaft included in the machine tool. The shaft servo motor 50 includes a position / speed detector. The servo motor 50 feeds a position / speed feedback signal from the position / speed detector back to the shaft control circuit 30 and then performs position and speed feedback control.

[0023] In the hardware structure diagram of Fig. 1, only one shaft control circuit 30, one servo amplifier 40, and one servo motor 50 are shown. However, in fact, as many shaft control circuits 30, servo amplifiers 40, and servo motors 50 are prepared as the number of shafts included in the machine tool as the control target. For example, for a machine tool having three linear shafts, three shaft control circuits 30, three servo amplifiers 40, and three servo motors 50 are prepared in the cutting fluid amount adjusting device 1. For a machine with five shafts, five shaft control circuits 30, five servo amplifiers 40, and five servo motors 50 are prepared in the adjusting device 1.

[0024] The cutting fluid quantity adjusting device 1 further includes a spindle control circuit 60 and a spindle amplifier 61 provided for rotating the spindle motor 62. The spindle control circuit 60 receives a spindle rotation command to a spindle of the machine tool and then outputs a spindle speed signal to the spindle amplifier 61. The spindle amplifier 61 receives the spindle speed signal, rotates the spindle motor 62 for the spindle at the commanded rotation speed, and drives a tool. The spindle motor 62 is combined with a position encoder 63. The position encoder 63 outputs a feedback pulse in synchronization with the spindle revolutions. The feedback pulse is read by the CPU 11.

[0025] The cutting fluid amount adjusting device 1 includes a machine learning device 100. The interface (INT) 21 in the adjusting device 1 is an interface for connecting the adjusting device 1 to the machine learning device 100. The machine learning device 100 includes a processor 101 for controlling the entire machine learning device 100, a ROM 102 in which a system program and the like are stored, a RAM 103 for performing temporary storage during processing related to machine learning, and a non-volatile memory 104 used for storing a learning model and the like.

[0026] The machine learning device 100 can observe information that can be acquired by the cutting fluid amount adjustment device 1 via the interface 21. The information that can be observed by the machine learning device 100 includes, for example, information related to tools such as types of tools; information related to cutting conditions such as spindle speed, feed rate, and cutting depth; information related to machining processes such as rough machining and fine machining; information related to a workpiece such as a workpiece material and a workpiece shape; information related to the cutting fluid such as a type and temperature of the cutting fluid; information related to discharge areas of nozzles; and information related to discharge amounts of the cutting fluid.The setting device 1 receives the information output from the machine learning device 100, and then performs processing operations such as controlling the machine tool and a cutting fluid discharge device, displaying on the display / MDI unit 70, and transmitting the information to other devices via a network not shown.

[0027] Fig. Figure 2 is a schematic functional block diagram of the cutting fluid quantity adjusting device 1 according to the first embodiment and the machine learning device 100 provided therein. The adjusting device 1 in this embodiment has components required when the machine learning device 100 performs supervised learning (learning mode). Fig. The function blocks shown in Figure 2 are implemented by the CPU 11, which is in the Fig. 1, and the processor 101 in the machine learning device 100 execute their respective system programs and control the operations of the components of the setting device 1 and the machine learning device 100.

[0028] The cutting fluid quantity adjustment device 1 in this embodiment includes a control unit 32, a data acquisition unit 34, and a preprocessing unit 36. The machine learning device 100 included in the adjustment device 1 includes a learning unit 110.

[0029] The Fig. The non-volatile memory 14 shown in Figure 1 has an acquired data memory 52 for storing data acquired from a machine tool 2, the cutting fluid supply device 3, and the like. The non-volatile memory 104 in the machine learning device 100 shown in Figure 1 Fig. 1, has a learning model memory 130 for storing a learning model constructed by machine learning by the learning unit 110.

[0030] The control unit 32 executes the system program read from the ROM 12 through the CPU 11, which is in the Fig. 1. The control unit 32 is mainly implemented by performing arithmetic processing using the RAM 13 and the non-volatile memory 14 by the CPU 11 and control processing for the machine tool 2 and the cutting fluid supply device 3 via the shaft control circuit 30, the spindle control circuit 60, and the PMC 16. The control unit 32 is a functional means for controlling the operation of the machine tool 2 and the cutting fluid supply device 3 based on a program 54 for control included in the Fig. 1 shown non-volatile memory 14.

[0031] The control unit 32 has a general control function required to control components of the machine tool 2 such that a movement command is sent to the servo motor 50 ( Fig. 1) and the spindle motor ( Fig. 1) is output to control the shafts included in the machine tool 2 according to the control program 54. The control unit 32 outputs a command to adjust the amounts of cutting fluid discharged from the respective cutting fluid nozzles to the cutting fluid supply device 3.

[0032] Furthermore, the control unit 32 obtains information regarding a machining state by the machine tool 2 and a state of the cutting fluid supplied from the cutting fluid supply device 3 from the machine tool 2 and the cutting fluid supply device 3, and then outputs this information to the data acquisition unit 34.The data that the control unit 32 acquires from the machine tool 2 and the cutting fluid supply device 3 and outputs to the data acquisition unit 34 includes, for example, information related to tools such as types of tools, information related to cutting conditions such as spindle speed, feed rate, and cutting depth, information related to machining processes such as rough machining and fine machining, information related to a workpiece such as a workpiece material and a workpiece shape, information related to the cutting fluid such as a type and temperature of the cutting fluid, information related to discharge areas of nozzles, and information related to a discharge amount of the cutting fluid.

[0033] The data acquisition unit 34 executes the system program read from the ROM 12 through the CPU 11, which is in the Fig. 1. The data acquisition unit 34 is mainly implemented by the CPU 11 performing arithmetic processing using the RAM 13 and the non-volatile memory 14. The data acquisition unit 34 is considered as functional means for storing, for example, data related to a machining state by the machine tool 2 input from the control unit 32, data related to a state of the cutting fluid supplied from the cutting fluid supply device 3, and data related to the evaluation for a cutting fluid amount input by the operator from the display / MDI unit 70, in the acquired data memory 52.

[0034] The data acquisition unit 34 correlates the data relating to the machining state of the machine tool 2, input from the control unit 32, and the data relating to the state of the cutting fluid supplied by the cutting fluid supply device 3, with the data relating to the evaluation of the amount of cutting fluid discharged from the cutting fluid nozzles. The data acquisition unit 34 stores the correlated data mentioned above as acquired data in the acquired data storage 52.

[0035] The preprocessing unit 36 ​​executes the system program read from the ROM 12 through the CPU 11, which is in the Fig. 1. The preprocessing unit 36 ​​is mainly implemented by the CPU 11 performing arithmetic processing using the RAM 13 and the nonvolatile memory 14. The preprocessing unit 36 ​​is considered to be functional means for generating learning data used in machine learning by the machine learning device 100 based on the data acquired by the data acquisition unit 34.

[0036] The preprocessing unit 36 ​​generates learning data obtained by converting (digitizing, sampling, or the like) the data acquired by the data acquisition unit 34 (and stored in the acquired data storage 52) into a unified form processed in the machine learning device 100. For example, when the machine learning device 100 performs supervised learning, the preprocessing unit 36 ​​generates, as learning data, a set of state data S and label data L having a predetermined form during learning.

[0037] The state data generated by the preprocessing unit 36 ​​in this embodiment includes at least the following seven types of data S1 to S7. More specifically, the state data S includes at least tool data S1 containing information related to tools used in machining a workpiece by the machine tool 2, machining condition data S2 containing information related to machining conditions in machining the workpiece by the machine tool 2, workpiece data S3 containing information related to the workpiece machined by the machine tool 2, cutting fluid data S4 containing information related to the cutting fluid supplied to a machining area of ​​the machine tool 2 by the cutting fluid supply device 3, machining process data S5,which include information relating to a machining process for the workpiece by the machine tool 2, cutting fluid discharge position data S6 which includes information relating to discharge positions of the cutting fluid by the cutting fluid nozzles included in the delivery device 3, and cutting fluid discharge amount data S7 which includes information relating to an amount of the cutting fluid discharged from the cutting fluid nozzles included in the delivery device 3.

[0038] The tool data S1 is defined as a line of data indicating the types and materials of the tools used in machining the workpiece by the machine tool 2. Regarding the types of tools, the tools can be classified, for example, according to the shapes of the tools and their use methods during machining, such as a cutting tool, a milling tool, and a drilling tool, and each can be represented by uniquely identifiable numerical values. Regarding the materials of the tools, for example, tool materials such as high-speed steel and cemented carbide can each be represented by uniquely identifiable numerical values. Information regarding the tools specified by the operator for the cutting fluid amount adjustment device 1 and the machine tool 2 can be acquired. The tool data S1 can be generated based on the acquired information regarding the tools.

[0039] The machine condition data S2 is defined as a line of data containing machining conditions such as spindle speed, feed rate, and cutting depth, which were commanded or specified when machining the workpiece by the machine tool 2, as elements. Regarding these machining conditions, numerical values ​​indicating the values ​​of the respective machining conditions in predetermined units can be used. The values ​​of the respective machining conditions can be commanded by the program 54 for the controller or set as standard values ​​for the controller. Therefore, the values ​​of the respective machining conditions can be generated by obtaining the program 54 for the controller or the standard values ​​for the controller.

[0040] The workpiece data S3 is defined as a data line that specifies the material of the workpiece being machined by the machine tool 2 and the shape of the machined workpiece. Regarding the workpiece material, for example, workpiece materials such as aluminum or iron can each be represented by uniquely identifiable numerical values. Regarding the shape of the workpiece, for example, as shown in Fig. 3, a table included in the machine tool 2 may be divided into predetermined areas and the shape of the workpiece may be represented as a line of data indicating heights in the areas (average values ​​of the heights in the areas) of the workpiece placed on the table.

[0041] In the embodiment shown in Fig. As shown in Figure 3, for convenience of explanation, the table is viewed from one side and divided into several laterally arranged regions. However, in actuality, a two-dimensional plane of an upper surface of the table can be divided into several rectangular regions, and the shape of the workpiece can be specified as a line of data indicating the heights of the workpiece in each rectangular region.The workpiece data S3 can be generated, for example, based on information regarding the workpiece set by the operator for the cutting fluid amount adjusting device 1 and the machine tool 2, the shape of the workpiece measured before machining is started, data obtained from a CAD (Computer Aided Design) or a CAM (Computer Aided Manufacturing) device, and data regarding a machining progress obtained by analyzing the program 54 for the controller.

[0042] The cutting fluid data S4 is defined as a data line indicating a type of cutting fluid delivered to the machining area of ​​the machine tool 2 and the temperature of the cutting fluid. Regarding the type of cutting fluid, for example, types of cutting fluid such as a water-soluble type A1 and a water-insoluble type N1 can each be represented by uniquely identifiable numerical values. Regarding the temperature of the cutting fluid, a numerical value indicating a temperature value of the cutting fluid in a predetermined unit can be used.Information regarding the cutting fluid set by the operator for the cutting fluid amount adjusting device 1 and the machine tool 2, or a value detected by a sensor or the like included in the cutting fluid supplying device 3, may be acquired to generate the cutting fluid data S4 based on the acquired information regarding this cutting fluid.

[0043] The machining process data S5 is defined as data indicating a process of machining the workpiece by the machine tool 2. Regarding the machining process, for example, machining processes such as rough machining, intermediate finishing, and finishing can each be represented by uniquely identifiable numerical values. The machining process data S5 can be generated based on data acquired from the CAD / CAM device before the start of machining, data related to the machining process obtained by analyzing the program 54 for the controller, and the like.

[0044] The cutting fluid discharge position data S6 is defined as a data line indicating discharge positions of the cutting fluid through the respective cutting fluid nozzles included in the cutting fluid supply device 3. For example, the discharge positions of the cutting fluid through the cutting fluid nozzles can be as shown in Fig. 4 as a data line indicating to which of predetermined divided areas of the table included in the machine tool 2 the cutting fluid discharged from the cutting fluid nozzles is sprayed.

[0045] In the embodiment shown in Fig. 4, a cutting fluid nozzle 301a discharges cutting fluid 302 to areas a and b. Therefore, the cutting fluid discharge position data with respect to the cutting fluid nozzle 301a can be represented as (1, 1, 0, 0, 0), where the numeral 1 indicates that the cutting fluid 302 is sprayed onto the areas, and the numeral 0 indicates that the cutting fluid 302 is not sprayed onto the areas. In the embodiment shown in Fig. 4, for convenience of explanation, the table is viewed from one side and divided into a plurality of laterally arranged regions. However, in actuality, a two-dimensional plane of an upper surface of the table may be divided into a plurality of rectangular regions, and discharge positions of the cutting fluid 302 may be indicated as a row of data indicating whether or not the cutting fluid 302 is sprayed onto the respective rectangular regions. The operator can set the cutting fluid discharge position data S6 in advance according to the arrangement of the cutting fluid nozzles from the display / MDI unit 70.

[0046] The cutting fluid discharge amount data S7 is defined as data indicating the amounts of cutting fluid discharged from the respective cutting fluid nozzles included in the cutting fluid supply device 3. The amounts of cutting fluid discharged from the cutting fluid nozzles may be discharge amounts of cutting fluid per unit time represented by numerical values ​​in a predetermined unit. Alternatively, the amounts of cutting fluid discharged from the cutting fluid nozzles may be represented, for example, by numerical values ​​indicated in multiple stages (e.g., ten stages). The cutting fluid discharge amount data S7 may be generated based on the cutting fluid discharge amounts of the cutting fluid nozzles obtained from the supply device 3.

[0047] The label data L generated by the pre-processing unit 36 ​​includes at least discharge amount adequacy data L1, which is data regarding an evaluation indicating whether or not the discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles are sufficient in a machining state in which the above-mentioned state data S is acquired.

[0048] The discharge amount adequacy data L1 is defined as data relating to each of the cutting fluid nozzles that assumes a label value regarding a judgment indicating whether the discharge amount of the cutting fluid is sufficient or not. As an example, the operator observing a machining state of the workpiece by the machine tool 2 can generate the discharge amount adequacy data L1 based on an input operation performed using the display / MDI unit 70. Regarding the generation based on the operator's input operation, the operator can directly input whether the discharge amounts of the cutting fluid are sufficient or not for the respective cutting fluid nozzles.Alternatively, the operator may adjust the amount of cutting fluid discharged from any one of the cutting fluid nozzles, and, if the same machining condition is maintained for a certain period of time in that condition, generate discharge amount adequacy data L1 for the cutting fluid nozzle whose discharge amount has been adjusted, which determines that the amount of cutting fluid discharged after adjustment is sufficient. For example, the temperatures of parts of the workpiece may be measured using a temperature sensor such as a thermometer mounted on the machine tool 2. The discharge amount adequacy data L1 may be generated based on the measurement result.

[0049] In the embodiment shown in Fig. As shown in FIG. 4, when the temperature of the workpiece is measured in regions a to e and the temperature at the position of region c does not decrease sufficiently, a label value indicating that the discharge amounts for the cutting fluid nozzles 301b, 301c, and 301d are insufficient may be generated. In this case, all of these cutting fluid nozzles 301a, 301c, and 301d are cutting fluid nozzles that spray the cutting fluid onto region c. Further, the discharge amount adequacy data L1 may be generated using an optical sensor, a distance sensor, or the like mounted on the machine tool 2, based on whether or not chips have been sufficiently removed from parts of the workpiece.

[0050] The learning unit 110 executes the system program read from the ROM 102 through the processor, which is in the Fig. 1. The learning unit 110 is mainly implemented by the processor 101 performing arithmetic processing using the RAM 103 and the non-volatile memory 104. The learning unit 110 in this embodiment performs machine learning using the learning data generated by the preprocessing unit 36. The learning unit 110 generates, by a well-known supervised learning method, a learning model that has learned the evaluation of the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles with respect to a machining state by the machine tool 2 and a supply state of the cutting fluid by the cutting fluid supply device 3. Furthermore, the learning unit 110 stores the generated learning model in the learning model memory 130.

[0051] The supervised learning method performed by the learning unit 110 includes, for example, a multilayer perceptron method, a recurrent neural network method, a long short term memory method, and a convolutional neural network method.

[0052] The learning unit 110 in this embodiment generates learning models for the respective cutting fluid nozzles, which models correlate the evaluation of the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles with the machining state by the machine tool 2 and the state of the cutting fluid supplied by the cutting fluid supply device 3, and then learn the relationship. For example, as shown in Fig. 4, five cutting fluid nozzles are present, the learning unit 110 generates five learning models that have learned the discharge quantities of the respective cutting fluid nozzles.

[0053] Fig. 5 is a diagram showing an overview of the generation of a learning model for each of the cutting fluid nozzles using the state data S and the label data L by the learning unit 110. As shown in Fig. As shown in Figure 5, the tool data S1, the machining condition data S2, the workpiece data S3, the cutting fluid data S4, and the machining process data S5 can be used jointly among several cutting fluid nozzles. Fig. 5, condition data common to the above-mentioned cutting fluid nozzles are designated by reference numeral 3010.

[0054] The learning unit 110 adds the cutting fluid discharge position data S6, the cutting fluid discharge amount data S7, and the discharge amount data L1 generated for each of the cutting fluid nozzles to these state data and generates learning models for the respective cutting fluid nozzles using these data. Fig. 5, the state data (data S6 and S7) specific to the cutting fluid nozzles 301a and 301b are designated by reference numerals 3011a and 3011b, respectively. Label data (discharge amount appropriate data L1) of the cutting fluid nozzles 301a and 301b are designated by reference numerals 3012a and 3012b, respectively. Furthermore, learning models for the cutting fluid nozzles 301a and 301b are designated by reference numerals 3013a and 3013b, respectively.

[0055] The learning model thus generated by the learning unit 110 in this embodiment estimates, for the respective cutting fluid nozzles, when the machining state by the machine tool 2 and the state of the cutting fluid supplied from the cutting fluid supply device 3 are given, whether the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles under these states are sufficient or not.

[0056] The learning unit 110 is an essential component during a learning stage. However, after the learning unit 110 completes the evaluation of the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles, the learning unit 110 is not always an essential component. For example, when a machine learning device 100 for which learning has been completed is delivered to a customer, it is possible that the machine learning device 100 is delivered after the learning unit 110 has been removed.

[0057] The cutting fluid amount adjustment device 1 according to this embodiment, which has the above-described structure, generates learning models for the respective cutting fluid nozzles, which have correlated the evaluation of the discharge amount of the cutting fluid discharged from the cutting fluid nozzles with the machining state of the machine tool 2 and the state of the cutting fluid supplied by the cutting fluid supply device 3, and learned the relationship. Using the learning models thus generated, an estimator 120 described below can perform estimation processing required to determine more appropriate amounts of the cutting fluid discharged from the cutting fluid nozzles in the acquired state based on the state data acquired from the machine tool 2 and the supply device 3.

[0058] As an alternative embodiment of the cutting fluid amount adjusting device 1 according to this embodiment, the preprocessing unit 36 ​​may further generate, as state data, in addition to the tool data S1, the machining condition data S2, the workpiece data S3, the cutting fluid data S4, the machining process data S5, the cutting fluid discharge position data S6, and the cutting fluid discharge amount data S7, spindle position data S8 indicating the position of the spindle of the machine tool 2. The spindle position data S8 may be data indicating on which of the Fig. 3 on the table of the machine tool 2, the spindle is present, and so on. When machining the workpiece in the machine tool 2, a large amount of heat is generated at a contact position of a tool attached to the spindle and the workpiece. Therefore, by adding the spindle position data S8 to the data used in learning, it is possible to learn information related to a discharge amount of the cutting fluid to the position where the tool and the workpiece are in contact. Furthermore, by adding the spindle position data S8 to the data used in learning, it is possible to use the data for estimating the discharge amount of the cutting fluid to the position where the tool and the workpiece are in contact.

[0059] Fig. 6 is a schematic functional block diagram of the cutting fluid quantity adjusting device 1 according to a second embodiment and the machine learning device 100 provided therein. The adjusting device 1 in this embodiment has components required when the machine learning device 100 estimates the evaluation of the discharge quantities of the cutting fluid discharged from the respective cutting fluid nozzles (estimation mode). Fig. 6 are implemented by the CPU 11, which is shown in Fig. 1, and the processor 101 in the machine learning device 100 execute their respective system programs and control the operations of the components of the setting device 1 and the machine learning device 100.

[0060] The cutting fluid quantity adjustment device 1 according to this embodiment includes, in addition to the control unit 32, the data acquisition unit 34, and the preprocessing unit 36, a discharge quantity determination unit 38. The machine learning device 100 included in the adjustment device 1 specifically includes the estimation unit 120. The acquired data storage 52 for storing the data acquired from the machine tool 2, the cutting fluid supply device 3, and the like is provided in the non-volatile memory 14 included in Fig. 1. The learning model memory 130 for storing a learning model constructed by machine learning by the learning unit 110 described in the first embodiment is provided in the non-volatile memory 104 in the machine learning device 100 shown in Fig. 1 is provided.

[0061] The control unit 32 and the data acquisition unit 34 in this embodiment have functions similar to those of the control unit 32 and the data acquisition unit 34 in the first embodiment.

[0062] The preprocessing unit 36 ​​in this embodiment, in the stage of estimating the discharge amounts of cutting fluid discharged from the respective cutting fluid nozzles, which is performed using a learning model by the machine learning device 100, converts data acquired by the data acquisition unit 34 into a unified form, which is processed in the machine learning device. A conversion form may be, for example, digitization or sampling. Through this conversion, the preprocessing unit 36 ​​in this embodiment generates the state data S having a predetermined form, which is used in the estimation by the machine learning device 100.

[0063] In this embodiment, when generating the state data S, the preprocessing unit 36 ​​generates the cutting fluid discharge amount data S7 indicating a commanded discharge amount for the cutting fluid nozzle commanded by the discharge amount determination unit 38. For example, the preprocessing unit 36 ​​generates the tool data S1, the machining condition data S2, the workpiece data S3, the cutting fluid data S4, the machining process data S5, and the cutting fluid discharge position data S6 based on the data acquired by the data acquisition unit 34. Furthermore, based on a command from the discharge amount determination unit 38, the preprocessing unit 36 ​​generates the cutting fluid discharge amount data S7 indicating the commanded discharge amount for the commanded cutting fluid nozzle.In an initial period of the estimation, the preprocessing unit 36 ​​may generate cutting fluid discharge amount data S7 indicating predetermined discharge amounts of the cutting fluid for the respective cutting fluid nozzles.

[0064] The estimation unit 120 carries out the processing by the processor 101, which is in the Fig. 1, executes a system program read from the ROM 102. The estimation unit 120 is mainly implemented by the processor 120 performing arithmetic processing using the RAM 103 and the non-volatile memory 104. The estimation unit 120 estimates, based on the state data S generated by the pre-processing unit 36, the evaluation of the discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles, which is performed using the learning model stored in the learning model memory 130.

[0065] The estimation unit 120 in this embodiment inputs the state data S supplied by the preprocessing unit 36 ​​into the learning model (whose parameters are determined, so to speak) generated by the learning unit 110 to estimate and output the evaluation of the discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles. The result of the estimation by the estimation unit 120 is output to the discharge amount determination unit 38.

[0066] The discharge quantity determination unit 38 executes the system program read from the ROM 12 through the CPU 11, which is in the Fig. 1. The discharge amount determination unit 38 is mainly implemented by the CPU 11 performing arithmetic processing using the RAM 13 and the non-volatile memory 14. The discharge amount determination unit 38 is considered to be functional means for determining the amounts of cutting fluid discharged from the respective cutting fluid nozzles based on an estimation result of the evaluation of the discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles estimated by the estimation unit 120.

[0067] The discharge amount determination unit 38 searches the discharge amount estimation result output by the estimation unit 120 for the respective cutting fluid nozzles. If the estimation result indicates that "the discharge amounts are sufficient," the discharge amount determination unit 38 instructs the pre-processing unit 36 ​​to generate state data in which the discharge amounts of the cutting fluid nozzles are reduced by one level. Conversely, if the estimation result indicates that "the discharge amounts are insufficient," the discharge amount determination unit 38 instructs the pre-processing unit 36 ​​to generate state data in which the cutting fluid discharge amounts are increased by one level. The discharge amount determination unit 38 repeats this processing to search for the minimum cutting fluid discharge amounts for the respective cutting fluid nozzles at which the estimation result that "the discharge amounts are sufficient" is obtained.Moreover, the discharge amount determination unit 38 determines the discharge amounts of the cutting fluid obtained by the search as discharge amounts of the cutting fluid from the cutting fluid nozzles.

[0068] With reference to Fig. 7, a process of determining a discharge amount of the cutting fluid at a cutting fluid nozzle by the discharge amount determining unit 38 will be described. In the embodiment shown in Fig. 7, it is assumed that a discharge amount 3020 of the cutting fluid from the cutting fluid nozzle can be adjusted in ten stages. In determining the cutting fluid discharge amount 3020 of the cutting fluid nozzle (in the case of Fig. 7, the cutting fluid discharge amount of the cutting fluid nozzle 301a), first, the preprocessing unit 36 ​​generates the state data S including cutting fluid discharge amount data S7 indicating a predetermined discharge amount (for example, a maximum value × 1.0).

[0069] The estimation unit 120 estimates the evaluation of the discharge amount 3020 of the cutting fluid based on state data S, that is, the data 3010 and 3011a generated by the preprocessing unit 36. When the estimation unit 120 outputs an estimation result 302 indicating that "the discharge amount is sufficient," the discharge amount determination unit 38 instructs the preprocessing unit 36 ​​to generate state data S in which the discharge amount 3020 is reduced by one level. Then, the discharge amount determination unit 38 estimates the evaluation of the discharge amount 3020 by the estimation unit 120 based on the generated state data S.When this is repeated and the evaluation or estimation result 3022 from the estimation unit 120 changes from "sufficient discharge amount" to "insufficient discharge amount," the discharge amount determination unit 38 determines that the cutting fluid discharge amount 3020 at the immediately preceding stage of the command to the pre-processing unit 36 ​​is an appropriate cutting fluid discharge amount (the smallest cutting fluid discharge amount whose evaluation is "sufficient discharge amount"). The discharge amount determination unit 38 determines the discharge amount based on this determination as the cutting fluid discharge amount of the cutting fluid nozzle. The discharge amount determination unit 38 performs this processing for all cutting fluid nozzles.

[0070] The search for a discharge amount of the cutting fluid by the discharge amount determination unit 38 can begin with the maximum discharge amount, as described above. Alternatively, the discharge amount determination unit 38 can search for the discharge amount from a minimum discharge amount value to larger values, or from an intermediate discharge amount value to larger and smaller values. The minimum discharge amount value can be searched using a known algorithm such as a binary search.

[0071] After determining the discharge amounts 3020 of the cutting fluid for all cutting fluid nozzles, the discharge amount determining unit 38 commands the control unit to set the discharge amounts of the cutting fluid to the determined discharge amounts.

[0072] The cutting fluid amount adjusting device 1 according to this embodiment, which has the above-described structure, can adjust amounts of the cutting fluid discharged from the respective cutting fluid nozzles to smaller amounts with which the workpiece or the like can be sufficiently cooled.

[0073] As an alternative embodiment of the cutting fluid amount adjusting device 1 according to this embodiment, the preprocessing unit 36 ​​may further generate, as state data, in addition to the tool data S1, the machining condition data S2, the workpiece data S3, the cutting fluid data S4, the machining process data S5, the cutting fluid discharge position data S6, and the cutting fluid discharge amount data S7, the spindle position data S8 indicating the position of the spindle of the machine tool 2. The spindle position data S8 may be data indicating on which of the Fig. 3 on the table of the machine tool 2, the spindle is present, and so on. During machining of the workpiece in the machine tool 2, a large amount of heat is generated at a contact position of a tool mounted on the spindle and the workpiece. Therefore, when estimating the discharge amounts of the cutting fluid by using the spindle position data S8, it is possible to estimate a sufficient amount of cutting fluid discharged to the position where the tool and the workpiece are in contact.

[0074] In the following, the cutting fluid quantity adjusting device 1 according to a third embodiment of the present invention will be described. The adjusting device 1 according to this embodiment has functional blocks similar to those in the Fig. 2. The adjustment device 1 in this embodiment includes components required when the machine learning device 100 performs supervised learning (learning mode).

[0075] The control unit 32 and the data acquisition unit 34 in this embodiment have functions similar to those of the control unit 32 and the data acquisition unit 34 in the first embodiment.

[0076] The pre-processing unit 36 ​​in this embodiment generates, as state data S, the tool data S1, the machining condition data S2, the workpiece data S3, the cutting fluid data S4, the machining process data S5, and the cutting fluid discharge position data S6. The pre-processing unit 36 ​​in this embodiment generates, as label data L in a state of machining in which the above-described state data S is acquired, appropriate discharge amount data L2, which is data regarding appropriate discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles.

[0077] The appropriate discharge amount data L2 is defined as data that uses, as a label value, an appropriate discharge amount of cutting fluid for each of the cutting fluid nozzles, in relation to current states of the machine tool 2 and the cutting fluid supply device 3. For example, an experienced operator observing a state of machining of a workpiece by the machine tool 2 can generate the appropriate discharge amount data L2 based on an input operation performed using the display / MDI unit 70. For example, the operator can set an amount of cutting fluid discharged from any one of the cutting fluid nozzles, and, if the same machining state continues for a predetermined time in that state, generate the appropriate discharge amount data L2 based on the discharge amounts of cutting fluid from the respective cutting fluid nozzles at that time.

[0078] The learning unit 110 in this embodiment performs machine learning using learning data generated by the preprocessing unit 36. The learning unit 110 generates a learning model, which has learned discharge amounts of the cutting fluid discharged from the cutting fluid nozzles with respect to a machining state by the machine tool 2 and a supply state of the cutting fluid by the cutting fluid supply device 3, by a known supervised learning method. The learning unit 110 stores the generated learning model in the learning model storage 130. Examples of the supervised learning method performed by the learning unit 110 include a multilayer perceptron method, a recurrent neural network method, a long-short-term memory method, and a convolutional neural network method.

[0079] The learning unit 110 in this embodiment generates learning models for the respective cutting fluid nozzles, which have correlated the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles with the machining state by the machine tool 2 and the state of the cutting fluid supplied by the cutting fluid supply device 3, and have learned the relationship. For example, as shown in Fig. 4, five cutting fluid nozzles are present, the learning unit 110 generates five learning models that have learned the discharge quantities of the respective cutting fluid nozzles.

[0080] Fig. Fig. 8 is a diagram showing an overview of the generation of a learning model 3013 for each of the cutting fluid nozzles 301 using the state data S and the label data L by the learning unit 110. As shown in Fig. As shown in Figure 8, among the state data S, the tool data S1, the machining condition data S2, the workpiece data S3, the cutting fluid data S4, and the machining process data S5 can be shared among a plurality of cutting fluid nozzles. The learning unit 110 adds the cutting fluid discharge position data S6 (cutting fluid nozzle state data 3011) and the appropriate discharge amount data L2 (cutting fluid nozzle label data 3012) generated for each of the cutting fluid nozzles 301 to these shared state data 3010 to generate a learning model 3013 for the respective cutting fluid nozzles using this data.

[0081] In this way, the learning model generated by the learning unit 110 in this embodiment estimates, for the respective cutting fluid nozzles, when the machining state by the machine tool 2 and the state of the cutting fluid supplied by the cutting fluid supply device 3 are given, more appropriate discharge amounts of the cutting fluid discharged from the cutting fluid nozzles under these states.

[0082] The learning unit 110 is an essential component during a learning stage. However, after the learning unit 110 completes the evaluation of the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles, the learning unit 110 is not always an essential component. For example, when a machine learning device 100 for which learning has been completed is delivered to a customer, it is possible that the machine learning device 100 is delivered after the learning unit 110 has been removed.

[0083] The cutting fluid amount adjusting device 1 according to this embodiment, which has the structure described above, generates a learning model for the respective cutting fluid nozzles, which has related the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles to the machining state 2 and the state of the cutting fluid supplied from the cutting fluid supply device 3 and has learned this relationship.

[0084] The cutting fluid amount adjustment device 1 according to this embodiment must perform learning based on appropriate discharge amounts of the cutting fluid from the respective cutting fluid nozzles set by the operator. Therefore, it is difficult to collect learning data required for sufficient learning. However, once sufficient machine learning is completed, the estimation unit 120 described below can directly estimate more appropriate amounts of the cutting fluid discharged from the cutting fluid nozzles in the acquired state using the thus-generated learning model based on the state data S acquired from the machine tool 2 and the cutting fluid supply device 3.

[0085] Fig. Fig. 9 is a schematic functional block diagram of the cutting fluid amount adjusting device 1 according to a fourth embodiment and the machine learning device 100 provided therein. The adjusting device 1 in this embodiment has components required when the machine learning device 100 estimates the evaluation of the discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles (estimation mode). Fig. The function blocks shown in Figure 9 are implemented by the CPU 11, which is shown in Fig. 1, and the processor 101 in the machine learning device 100 execute their respective system programs and control the operations of the components of the setting device 1 and the machine learning device 100.

[0086] The cutting fluid quantity adjusting device 1 in this embodiment includes the control unit 32, the data acquisition unit 34, the preprocessing unit 36, and the discharge quantity determination unit 38. The machine learning device 100 included in the adjusting device 1 includes the estimation unit 120. The acquired data storage 52 for storing data acquired from the machine tool 2, the cutting fluid supply device 3, and the like is provided in the Fig. 1. The learning model memory 130 for storing a learning model constructed by machine learning by the learning unit 110 described in the first embodiment is provided in the non-volatile memory 104 in the manner shown in Fig. 1 shown machine learning device 100.

[0087] The control unit 32 and the data acquisition unit 34 in this embodiment have functions similar to those of the control unit 32 and the data acquisition unit 34 in the first embodiment.

[0088] The preprocessing unit 36 ​​in this embodiment, in the stage of estimating the discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles, which is performed using a learning model by the machine learning device 100, converts data acquired by the data acquisition unit 34 into a unified form, which is processed in the machine learning device. The form of uniform conversion may be, for example, digitization or sampling. Through this conversion, the preprocessing unit 36 ​​according to this embodiment generates the state data S having a predetermined form, which is used in the estimation by the machine learning device 100.The preprocessing unit 36 ​​in this embodiment generates the tool data S1, the machining condition data S2, the workpiece data S3, the cutting fluid data S4, the machining process data S5, and the cutting fluid discharge position data S6 based on the data acquired by the data acquisition unit 34.

[0089] The estimation unit 120 in this embodiment estimates, based on the state data S generated by the preprocessing unit 36, the evaluation of the discharge amount of the cutting fluid discharged from the respective cutting fluid nozzles, which is made using the learning model stored in the learning model memory 130. The estimation unit 120 in this embodiment inputs the state data S input by the preprocessing unit 36 ​​into the learning model (the parameters of which are determined) generated by the learning unit 110. Further, based on the input state data S, the estimation unit 120 estimates more appropriate discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles and outputs the more appropriate discharge amounts. The result of the estimation by the estimation unit 120 is output to the discharge amount determination unit 38.

[0090] The discharge amount determination unit 38 in this embodiment determines an estimation result of the discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles, estimated by the estimation unit 120, as the amounts of cutting fluid discharged from the respective cutting fluid nozzles. The discharge amount determination unit 38 further commands the control unit 32 to adjust the discharge amounts of the cutting fluid to the determined discharge amounts.

[0091] The cutting fluid amount adjusting device 1 according to this embodiment, which has the structure described above, can adjust the amounts of the cutting fluid discharged from the respective cutting fluid nozzles to smaller amounts with which the workpiece or the like can be sufficiently cooled.

[0092] Fig. Figure 10 is a schematic functional block diagram of the cutting fluid quantity adjustment device 1 and the machine learning device 100 in a fifth embodiment. The cutting fluid quantity adjustment device 1 in this embodiment includes components required when the machine learning device 100 performs supervised learning (learning mode). Fig. 10 shown function blocks are implemented by the CPU 11, which is in the Fig. 1, and the processor 101 in the machine learning device 100 execute their respective system programs and control the operations of the components of the adjusting device 1 and the machine learning device 100.

[0093] The cutting fluid quantity adjusting device 1 in this embodiment includes the control unit 32, the data acquisition unit 34, and the preprocessing unit 36. The machine learning device 100 in the adjusting device 1 includes the learning unit 110 having a first learning unit 112 and a second learning unit 114. The acquired data storage 52 for storing data acquired from the machine tool 2, the cutting fluid supply device 3, and the like is provided in the non-volatile memory 14. Fig. 1. The learning model memory 130 for storing a learning model constructed by the machine learning by the first learning unit 112 and the second learning unit 114 is provided in the non-volatile memory 104 in the machine learning device 100 shown in Fig. 1 is provided.

[0094] The control unit 32, the data acquisition unit 34, and the preprocessing unit 36 ​​in this embodiment have functions similar to those in the first embodiment.

[0095] The learning unit 110 in this embodiment includes the first learning unit 112 and the second learning unit 114. Like the learning unit described in the first embodiment, the first learning unit 112 performs machine learning using learning data generated by the preprocessing unit 36, and generates, for each cutting fluid nozzle, a first learning model that correlates the evaluation of the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles with a machining state by the machine tool 2 and a state of the cutting fluid supplied by the cutting fluid supply device 3, and learns the relationship.The first learning model generated by the first learning unit 112 estimates, for the respective cutting fluid nozzles, when the machining state by the machine tool 2 and the state of the cutting fluid supplied by the cutting fluid supply device 3 are given, whether the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles in these states are sufficient or not.

[0096] Like the learning unit 110 described in the third embodiment, the second learning unit 114 performs machine learning using learning data and generates a second learning model for each cutting fluid nozzle. The second learning model correlates the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles with a machining state by the machine tool 2 and a state of the cutting fluid supplied by the cutting fluid supply device 3, and learns the relationship. The second learning unit 114 analyzes a learning result of the first learning model generated by the first learning unit 112. The second learning unit 114 performs machine learning based on the analysis result to generate the second learning model.The second learning unit 114 performs, for example, processing on the first learning model similar to that performed by the discharge amount determination unit 38 described in the second embodiment, and then searches for the respective cutting fluid nozzles a minimum discharge amount of cutting fluid at which an estimation result that "the discharge amounts are sufficient" is obtained for the states of the machine tool 2 and the cutting fluid supply device 3. The second learning unit 114 performs machine learning using the state of the machine tool 2 and the cutting fluid supply device 3 and the discharge amount of cutting fluid obtained by the search to generate the second learning model for the respective cutting fluid nozzles.

[0097] The cutting fluid amount adjusting device 1 according to this embodiment generates learning data using the first learning model generated by the same machine learning method as that of the learning unit 110 in the adjusting device 1 described in the first embodiment. Furthermore, the adjusting device 1 generates the second learning model using the generated learning data. While the adjusting device 1 described in the first embodiment can relatively easily generate learning data and perform machine learning, it must also perform search processing to determine cutting fluid discharge amounts of the cutting fluid nozzles using the generated learning model.Although the adjusting device 1 described in the third embodiment can directly estimate more appropriate cutting fluid discharge amounts of the cutting fluid nozzles using the learning model, on the other hand, it is difficult to collect learning data.

[0098] By combining these two learning methods, the cutting fluid amount adjustment device 1 according to this embodiment can generate a learning model that can relatively easily collect learning data and directly estimate more appropriate cutting fluid discharge amounts of the cutting fluid nozzles using the collected learning data. The learning model (the second learning model) generated in this way can be used, for example, in estimating the cutting fluid discharge amounts of the cutting fluid nozzles in the adjustment device 1 according to the fourth embodiment.

[0099] Fig. 11 is a schematic functional block diagram of the cutting fluid quantity adjusting device 1 according to a sixth embodiment and the machine learning device 100 provided therein. Fig. 11 are implemented by the CPU 11 in the setting device 1, which is shown in Fig. 1, and the processor 101 in the machine learning device 100 execute their respective system programs and control the operations of the components of the setting device 1 and the machine learning device 100.

[0100] The cutting fluid quantity adjusting device 1 in this embodiment includes the control unit 32, the data acquisition unit 34, and the preprocessing unit 36. The machine learning device 100 in the adjusting device 1 includes, in addition to the learning unit 110, a decision-making unit 122. The acquired data storage 52 for storing data acquired from the machine tool 2, the cutting fluid supply device 3, and the like is provided in the non-volatile memory 14. Fig. 1. The learning model memory 130 for storing a learning model constructed on the basis of machine learning by the learning unit 110 is provided in the non-volatile memory 104 in the machine learning device 100 shown in Fig. 1 is provided.

[0101] The control unit 32 and the data acquisition unit 34 in this embodiment have functions similar to those in the first embodiment.

[0102] In this embodiment, when the machine learning device 100 performs reinforcement learning, the preprocessing unit 36 ​​generates, as learning data, a set of the state data S and the determination data D having a predetermined form during learning. The preprocessing unit 36 ​​generates the tool data S1, the machining condition data S2, the workpiece data S3, the cutting fluid data S4, the machining process data S5, the cutting fluid discharge position data S6, and the cutting fluid discharge amount data S7 as state data S.The pre-processing unit 36 ​​generates, as determination data D in a state of machining in which the above-mentioned state data S is acquired, discharge amount determination data D1 which is data for determining the discharge amounts of the cutting fluid discharged from the respective cutting fluid nozzles, and workpiece determination data D2 which is data for determining a cooling state of the workpiece and a chip removal state.

[0103] The discharge amount determination data D1 is defined for each of the cutting fluid nozzles as data serving as a reference for determining whether the discharge amount of the cutting fluid is sufficiently small. The discharge amount determination data D1 may be data indicating whether or not the amount of cutting fluid discharged from the cutting fluid nozzle has increased or decreased, for example, as a result of adjusting the amount of cutting fluid discharged from the cutting fluid nozzle. The discharge amount determination data D1 may be data including a degree of increase or decrease in the amount of cutting fluid discharged from the cutting fluid nozzle.

[0104] The workpiece condition determination data D2 is defined as data used as a reference for determining, for example, whether a workpiece is sufficiently cooled or whether chips adhere to the surface of the workpiece. An operator observing a machining condition of the workpiece by the machine tool 2 or a similar condition can generate the workpiece determination data D2 based on an input operation performed using the display / MDI unit 70. Alternatively, when chips adhere to the surface of the workpiece, the operator can operate the display / MDI unit 70 and then input the presence of a problem at the location of the workpiece where the chips adhere.Thereby, the pre-processing unit 36 ​​can determine that the amount of cutting fluid discharged from the cutting fluid nozzle at the position corresponding to the chip attachment point is small and then generate workpiece state determination data D2 indicating this fact.

[0105] The temperature of portions of the workpiece can be measured using a temperature sensor such as a thermography device mounted on the machine tool 2. The workpiece condition determination data D2 can be generated based on the measurement result. In the example shown in Fig. 4, the temperature of the workpiece in the areas a to e may be measured, and workpiece determination data D2 may be generated which indicates that the discharge amounts with respect to the cutting fluid nozzles 301b, 301c, and 301d (all of which are cutting fluid nozzles that spray the cutting fluid onto the area c) are small when the temperature decrease is insufficient, and otherwise indicates that the discharge amounts are correct.

[0106] Furthermore, the workpiece condition determination data D2 may be generated using an optical sensor, a distance sensor, or the like mounted on the machine tool 2, based on whether or not the chips are sufficiently removed from the portions of the workpiece. The workpiece condition determination data D2 may include a degree of cooling and a degree corresponding to a remaining state of the chips.

[0107] The learning unit 110 in this embodiment performs machine learning using learning data generated by the preprocessing unit 36. The learning unit 110 generates, through a well-known reinforcement learning method, a learning model for each cutting fluid nozzle. The learning model has learned a setting behavior for the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles with respect to a machining state by the machine tool 2 and a supply state of the cutting fluid by the cutting fluid supply device 3. Furthermore, the learning unit 110 stores the generated learning model in the learning model memory 130.Reinforcement learning is a method in which a cycle of observing a current state (i.e., an input) of an environment in which a learning target exists, performing a predetermined behavior (i.e., an output) in the current state, and awarding a certain reward for the behavior is repeated in a trial-and-error manner, and a measure to maximize the total reward is learned as an optimal solution. The measure to maximize the total reward, in the machine learning device 100 of the present invention, means a behavior of adjusting the discharge amounts of the cutting fluid from the cutting fluid nozzles. Examples of the reinforcement learning method by the learning unit 110 include Q-learning.

[0108] In Q-learning by the learning unit 110, for example, a reward R may be set to a positive reward R when the discharge amount determination data D1 indicates that "the cutting fluid amount has decreased," and a negative reward R when the discharge amount determination data D1 indicates that "the cutting fluid amount has increased." The reward R may be set to the positive reward R when the workpiece condition determination data D2 indicates that "the cutting fluid amount is appropriate," and a negative reward R when the workpiece condition determination data D2 indicates that "the cutting fluid amount is small." Further, the reward R may be a larger positive or negative reward R depending on the magnitude of the degree indicated by the operation determination data D1 or the workpiece condition determination data D2.

[0109] The learning unit 110 may be configured to use a neural network as a value function Q (a learning model), receive state data S and a behavior a as inputs to the neural network, and output a value (a result y) of the behavior a in the state. When the learning unit 110 is configured in this way, a neural network comprising an input layer, an intermediate layer, and an output layer may be used as the learning model. However, it is also possible to configure the learning unit 110 to perform more effective learning and derivation by using a so-called deep learning method in which a neural network forming three or more layers is used. The learning model generated by the learning unit 110 is stored in the learning model memory 130 provided in the non-volatile memory 104.The learning model stored in the learning model memory 130 is used in determining a setting behavior for discharge amounts of the cutting fluid discharged from the cutting fluid nozzles by the decision-making unit 122.

[0110] The learning unit 110 is an essential component during a learning stage. However, after the learning of the adjustment behavior for the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles has been completed by the learning unit 110, the learning unit 110 is not always an essential component. For example, when a machine learning device 100 for which learning has been completed is delivered to a customer, it is possible that the machine learning device 100 is delivered after the learning unit 110 has been removed.

[0111] The decision-making unit 122 executes the system program read from the ROM 102 through the processor 101 in the cutting fluid quantity adjusting device 1, which is in Fig. 1. The decision-making unit 122 is mainly implemented by the processor 101 performing arithmetic processing using the RAM 103 and the non-volatile memory 104. The decision-making unit 122 calculates, based on the state data S input from the pre-processing unit 36, an optimal solution of the adjustment behavior for the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles, which adjustment behavior was made using the learning model stored in the learning model memory 130. The decision-making unit 122 further outputs the adjustment behavior for the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles calculated as the optimal solution.

[0112] The decision-making unit 122 in this embodiment can calculate a reward given when the behavior is retained in a current state by inputting the state data S (the tool data S1, the machining condition data S2, the workpiece data S3, the cutting fluid data S4, the machining process data S5, the cutting fluid discharge position data S6, and the cutting fluid discharge amount data S7) input from the preprocessing unit 36 ​​and the setting behavior for the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles (the determination of a discharge amount of the cutting fluid for each of the cutting fluid nozzles) as input data to the learning model created by the reinforcement learning by the learning unit 110 (the parameters of which are determined).Therefore, the decision-making unit 122 calculates the reward for the adjustment behavior for the cutting fluid discharge amount that can currently be used, compares the calculated multiple rewards, and determines the adjustment behavior for a discharge amount of the cutting fluid discharged from the cutting fluid nozzle for which the largest reward is calculated as the optimal solution. The optimal solution of the adjustment behavior for the discharge amount of the cutting fluid discharged from the cutting fluid nozzle determined by the decision-making unit 122 can be input to the discharge amount determination unit 38 and used in determining a discharge amount of the cutting fluid discharged from the cutting fluid nozzle.In addition, the optimal solution of the setting behavior can also be used, for example, to be output to and displayed on the display / MDI unit 70 or to be sent and output to a fog or cloud computer via a wired / wireless network not shown.

[0113] In the cutting fluid amount adjusting device 1 having the above-described structure, it is possible to appropriately adjust the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles in an environment in which the machine tool 2 and the cutting fluid supply device 3 operate.

[0114] In the seventh to ninth embodiments below, embodiments configured as part of a system in which a cutting fluid amount adjusting device 1 according to the first to sixth embodiments is connected to a plurality of devices including a cloud server, a host computer, a fog computer, and an edge computer (such as a robot control unit and a control device) via a wired or wireless network will be explained.

[0115] As in Fig. As shown in FIG. 12, in the seventh to ninth embodiments described below, a system is adopted which is configured to be logically divided into three layers of a layer including a cloud server 6 and the like, a layer including fog computers 7 and the like, and a layer including edge computers 8 (such as a robot control unit and a control device included in cells 9) in a state where each of a plurality of devices is connected to a network. In such a system, the cutting fluid amount adjusting device 1 according to one form of the present invention may also be installed in any one of the cloud server 6, the fog computers 7, and the edge computers 8.With such a structure, the tuning device 1 can share data used in machine learning-related processing among the respective multiple devices via the network, perform distributed learning, and collect generated learning models in the fog computers 7 and the cloud server 6 for large-scale analysis. The tuning device 1 can further reuse the generated learning models, for example, mutually.

[0116] In the system that Fig. As shown in Figure 12, a plurality of cells 9 are deployed in factories at different locations, respectively. The fog computers 7 in a higher layer manage the respective cells in a predetermined unit (for example, a factory unit or a unit of multiple factories of the same manufacturer). The cloud server 6 in an even higher layer collects the data collected and analyzed by the fog computers 7 and performs analysis and / or other processing of the data. Information obtained as a result of the analysis and / or data processing can be used in control and the like in the respective edge computers 8.

[0117] Fig. 13 is a schematic hardware configuration diagram in which a cutting fluid amount adjusting device is installed in a computer such as a cloud server or a fog computer.

[0118] In this embodiment, a computer-implemented cutting fluid quantity adjustment device 1' includes a CPU 311, a ROM 312, a RAM 313, and a non-volatile memory 314. The cutting fluid quantity adjustment device 1' further includes interfaces 317, 318, 319, and 321. The above-mentioned components in the adjustment device 1' are interconnected via a bus 320.

[0119] The CPU 311 is a processor capable of controlling the cutting fluid quantity adjusting device 1' as a whole. The CPU 311 reads a system program stored in the ROM 312 via the bus 320 to control the entire adjusting device 1' according to the system program. The RAM 313 temporarily stores various data and the like. The various data temporarily stored in the RAM 313 include, for example, temporary calculation and display data, and further various types of data input by an operator via an input section (not shown).

[0120] The non-volatile memory 314 is backed up by, for example, a battery (not shown) to be configured as a memory in which the stored state is maintained even when the cutting fluid amount adjusting device 1' is turned off. The non-volatile memory 314 stores a program input via an input device 371 and various data acquired from a robot via the components of the adjusting device 1' and a network 5. The program and various data stored in the non-volatile memory 314 can be developed into the ROM 313 during execution or use. Various system programs, such as a known analysis program, have been pre-written in the ROM 312. The system programs pre-written in the ROM 312 include a system program for controlling exchange with the machine learning device 100 described below.

[0121] The cutting fluid quantity adjustment device 1' is connected to the wired or wireless network 5 via the interface 319. At least one machine tool 2' (a machine tool having a control device), other cutting fluid quantity adjustment devices 1, the edge computers 8, the fog computers 7, the cloud server 6, and the like are connected to the network 5. Data is exchanged reciprocally between the adjustment device 1' and the elements connected to the network 5 via the network 5.

[0122] The cutting fluid quantity adjusting device 1' according to this embodiment is further connected to a display device 370 via the network 5 and further to an input device 371 via the interface 318.

[0123] Data read into the memory, data obtained as a result of executing a program, or the like, and other data are output to the display device 370 for display via the interface 317. The input device 371, which is composed of a keyboard, a pointing device, and the like, outputs a command based on an operator's operation, data, or the like to the CPU 311 via the interface 318.

[0124] The interface 321 is intended to connect the cutting fluid quantity adjustment device 1' to the machine learning device 100. The machine learning device 100 has components that implement the Fig. 1 are the same.

[0125] When the cutting fluid amount adjusting device 1' is configured in this way in a computer such as the cloud server or the fog computer, the functions of the adjusting device 1' are the same as those described in the first to sixth embodiments, except that the acquisition of information from the machine tool 2' and the cutting fluid supply device 3 and the command for adjusting the discharge amounts of the cutting fluid discharged from the cutting fluid nozzles are exchanged to the supply device 3 via the network 5.

[0126] Since the machine tool 2' includes the control device, the control unit 32 in the cutting fluid quantity adjustment device 1' does not control the machine tool 2', but rather acquires information related to a machining state by the machine tool 2' from the control device in the machine tool 2'. The control unit 32 in the adjustment device 1' indirectly controls the cutting fluid supply device 3 via the control device in the machine tool 2' and further acquires a supply state of the cutting fluid via the control device included in the machine tool 2'.

[0127] Fig. 14 is a schematic configuration diagram of a cutting fluid amount adjustment system according to a seventh embodiment, including the cutting fluid amount adjustment device 1'. In this system, a cutting fluid amount adjustment system 500 includes a plurality of cutting fluid amount adjustment devices 1 and 1', a plurality of machine tools 2', and the network for connecting the adjustment devices 1 and 1' and the machine tools 2'.

[0128] In the cutting fluid amount adjustment system 500, the cutting fluid amount adjustment device 1' including the machine learning device 100 adjusts amounts of cutting fluid discharged from the respective cutting fluid nozzles in the supply device 3 set as the adjustment target, using a learning result of the learning unit 110, based on a machining state by the machine tool 2' and a delivery state by the cutting fluid supply device 3. At least one adjustment device 1' learns, based on the state variable S and the label data L or determination data D obtained by each of the other plurality of adjustment devices 1 and 1', adjustment of the amounts of cutting fluid discharged from the respective cutting fluid nozzles with respect to machining states by the machine tools 2 and 2' and a delivery state of the cutting fluid by the supply device 3, which are common to all the adjustment devices 1 and 1'.The system 500 may be configured so that all adjustment devices 1 and 1' share the learning result. Therefore, the system 500 may receive input from a wider variety of data sets (including state variables S and label data L or determination data D) to improve the speed and reliability of learning.

[0129] Fig. 15 is a schematic configuration diagram of a system according to an eighth embodiment, in which a machine learning device and a cutting fluid amount adjustment device are configured in different devices. According to the eighth embodiment, a cutting fluid amount adjustment system 500' includes at least one machine learning device 100 configured as part of a computer such as a cloud server, a host computer, or a fog computer, a plurality of cutting fluid amount adjustment devices 1'', and the network 5 for interconnecting the cutting fluid amount adjustment devices 1'' and the computer. Fig. 15 shows an embodiment in which the machine learning device 100 is configured as part of the fog computer 7. As in the schematic hardware structure of the cutting fluid quantity adjusting device 1' shown in Fig. 13, the hardware structure of the computer is implemented by connecting hardware such as the CPU 311, the RAM 313, and the non-volatile memory 314, which hardware is included in a general computer, via the bus 320.

[0130] In the cutting fluid amount adjustment system 500' having the above-described configuration, the machine learning device 100 learns, based on the state variables S and the label data L or the determination data D obtained for each of the plurality of cutting fluid amount adjustment devices 1'', the adjustment of amounts of the cutting fluid discharged from the respective cutting fluid nozzles included in the cutting fluid supply device 3 with respect to a machining state by the machine tool 2 and a supply state of the cutting fluid by the supply device 3, which are common to all the adjustment devices 1''. Furthermore, the machine learning device 100 can adjust amounts of the cutting fluid discharged from the respective cutting fluid nozzles in the supply device 3 using the learning result.By constructing the system 500', regardless of the time and location where each of the plurality of adjustment devices 100'' is located, it is possible to connect a required number of adjustment devices 1'' to the machine learning device 100 when necessary.

[0131] Fig. 16 is a schematic configuration diagram of a cutting fluid amount adjustment system 500'' according to a ninth embodiment, wherein the system 500'' includes a machine learning device 100' and the cutting fluid amount adjustment device 1. The system 500'' includes at least one machine learning device 100' installed in a computer such as an edge computer, a fog computer, a host computer, or a cloud server, a plurality of adjustment devices 1, and the wired or wireless network 5 for connecting the cutting fluid amount adjustment devices 1 and the computer. Fig.16 shows an embodiment in which the machine learning device 100' is configured as part of the fog computer 7.

[0132] In the cutting fluid amount adjustment system 500'' having the above-described configuration, the fog computer 7 including the machine learning device 100' acquires, from the respective cutting fluid amount adjustment devices 1, learning models obtained as a result of machine learning by the machine learning devices 100 in the adjustment devices 1. The machine learning device 100' in the fog computer 7 performs knowledge optimization and streamlining processing based on these multiple learning models to newly generate an optimized and streamlined learning model. Furthermore, the machine learning device 100' forwards the generated learning model to the respective adjustment devices 1.

[0133] The optimization and refinement of the learning model performed by the machine learning device 100' involves generating a distillation model based on the plurality of learning models obtained from the cutting fluid amount adjusting devices 1. In this case, the machine learning device 100' in this embodiment generates input data that is supplied to the learning model. Then, the machine learning device 100' newly generates a learning model (a distillation model) by performing learning from the beginning using the outputs obtained as a result of inputting the input data to the respective learning models. As described above, the distillation model thus generated is forwarded to the adjusting devices 1 via an external storage medium or the network 5 and then used.

[0134] As another example of the optimization and refinement of the learning model performed by the machine learning device 100', it is also conceivable to perform the following process. That is, in a process of performing distillation on the multiple learning models acquired from the cutting fluid amount adjusting devices 1, the machine learning device 100' analyzes a distribution of the output data of the learning models with respect to the input data by a general statistical method, extracts outliers from a set of the input data and the output data, and then performs distillation using the set of input data and the output data from which the outliers have been excluded.By performing such a process, it is possible to exclude an exceptional estimation result from the set of input data and output data obtained by the respective learning models, and to generate a distillation model using the set of input data and output data from which the exceptional estimation result was excluded. The distillation model generated in this way can be used as a more generic learning model compared to the learning model generated by the plurality of cutting fluid amount adjusting devices 1.

[0135] It is also possible to introduce another method for optimizing and streamlining general learning models, as appropriate. For example, this method involves analyzing the learning models and then optimizing the hyperparameters of the learning models based on the analysis results.

[0136] In the cutting fluid amount adjustment system 500'' according to this embodiment, the machine learning device 100' may be arranged, for example, in the fog computer 7 configured with respect to the plurality of cutting fluid amount adjustment devices 1 acting as edge computers. Learning models generated by the respective adjustment devices 1 may be collected and stored in the fog computer 7. The machine learning device 100' in the fog computer 7 may perform the operation of performing optimization and streamlining based on the stored plurality of learning models and then transmit the optimized and streamlined learning models to the adjustment devices 1 as needed.

[0137] The cutting fluid quantity adjustment system 500'' according to this embodiment can collect the learning models collected and stored in the fog computer 7 and the learning models optimized and streamlined in the fog computer in a higher-level host computer or cloud server to apply them to the intellectual work in a factory or at a manufacturer of the cutting fluid quantity adjustment devices 1 that use these learning models. This application to the intellectual work includes, for example, the construction and distribution of a more generic learning model in a higher-level server, support for maintenance work based on an analysis result of the learning models, analysis of the performance and the like of the respective adjustment devices 1, and application to the development of a new machine.

[0138] Embodiments of the present invention have been described above. However, the present invention is not limited to the above-described embodiments, but can be embodied in various forms by adding appropriate modifications.

[0139] For example, in the above-mentioned embodiments, the cutting fluid amount adjusting device 1 and the machine learning device 100 are described as devices having different CPUs (processors). However, the machine learning device 100 can be implemented by the CPU 11 in the cutting fluid amount adjusting device 1 and the system program stored in the ROM 12.

Claims

[1] Cutting fluid quantity adjusting device (1, 1', 1'') for adjusting a discharge quantity of the cutting fluid from at least one cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) included in a cutting fluid supply device (3) for supplying the cutting fluid to a machining area of ​​a machine tool for machining a workpiece, characterized by that they a data acquisition unit (34) configured to acquire at least data indicating a machining state by the machine tool and data relating to the cutting fluid supplied from the cutting fluid supply device (3); a preprocessing unit (36) configured to generate data used in machine learning based on the data acquired by the data acquisition unit (34); and a machine learning device (100, 100') configured to perform, on the basis of the data generated by the pre-processing unit (36), machine learning processing related to the discharge amount of the cutting fluid from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) in an environment in which the machining of the workpiece is performed by the machine tool, includes. [2] Cutting fluid quantity adjusting device (1, 1', 1'') according to claim 1, characterized by , that the preprocessing unit (36) as data used in supervised learning by the machine learning device (100, 100'), state data including at least tool data containing information relating to a tool used in machining the workpiece by the machine tool, machining condition data containing information relating to machining conditions in machining the workpiece by the machine tool, workpiece data containing information relating to the workpiece machined by the machine tool, cutting fluid data containing information relating to the cutting fluid supplied to the machining area of ​​the machine tool by the cutting fluid supply device (3), machining process data containing information relating to a machining process of the workpiece by the machine tool, cutting fluid discharge position data,which includes information relating to a discharge position of the cutting fluid through the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), and cutting fluid discharge amount data including an amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) included in the cutting fluid supply device (3), and label data including at least discharge amount adequacy data indicating the adequacy of the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), and, the machine learning device (100, 100') comprises a learning unit (110) configured to generate, on the basis of the state data and the label data, a learning model that relates the machining state by the machine tool and a state of the cutting fluid supplied by the cutting fluid supply device (3) to the appropriateness of the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), has. [3] Cutting fluid quantity adjusting device (1, 1', 1'') according to claim 1, characterized by , that the preprocessing unit (36) uses, as data used in an estimation by the machine learning device (100, 100'), state data including at least tool data containing information relating to a tool used in machining the workpiece by the machine tool, machining condition data containing information relating to machining conditions in machining the workpiece by the machine tool, workpiece data containing information relating to the workpiece machined by the machine tool, cutting fluid data containing information relating to the cutting fluid supplied to the machining area of ​​the machine tool by the cutting fluid supply device (3), machining process data containing information relating to a machining process of the workpiece by the machine tool, cutting fluid discharge position data,which includes information relating to a discharge position of the cutting fluid through the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), and cutting fluid discharge amount data including an amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) included in the cutting fluid supply device (3), and, the machine learning device (100, 100') a learning model memory (130) configured to store a learning model relating the machining state by the machine tool and a state of the cutting fluid supplied by the cutting fluid supply device (3) to the appropriateness of the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e); and an estimation unit (120) configured to estimate the appropriateness of the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) on the basis of the state data using the learning model stored in the learning model memory (130), and the cutting fluid quantity adjusting device (1, 1', 1'') further comprises a discharge quantity determining unit (38) configured to search for a minimum discharge quantity of the cutting fluid estimated as good by the estimation unit (120) and to determine the discharge quantity of the cutting fluid obtained by the search as a quantity of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), includes. [4] Cutting fluid quantity adjusting device (1, 1', 1'') according to claim 1, characterized by , that the preprocessing unit (36) uses, as data used in supervised learning by the machine learning device (100, 100'), state data including at least tool data containing information relating to a tool used in machining the workpiece by the machine tool, machining condition data containing information relating to machining conditions in machining the workpiece by the machine tool, workpiece data containing information relating to the workpiece machined by the machine tool, cutting fluid data containing information relating to the cutting fluid supplied to the machining area of ​​the machine tool by the cutting fluid supply device (3), machining process data containing information relating to a machining process of the workpiece by the machine tool, and cutting fluid discharge position data,which includes information relating to a discharge position of the cutting fluid through the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), and label data including at least appropriate discharge amount data including an amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), and, the machine learning device (100, 100') comprises a learning unit (110) configured to generate a learning model based on the state data and the label data, which model relates the machining state by the machine tool and a state of the cutting fluid supplied by the cutting fluid supply device (3) to the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), has. [5] Cutting fluid quantity adjusting device (1, 1', 1'') according to claim 1, characterized by , that the preprocessing unit (36) uses, as data used in an estimation by the machine learning device (100, 100'), state data including at least tool data containing information relating to a tool used in machining the workpiece by the machine tool, machining condition data containing information relating to machining conditions in machining the workpiece by the machine tool, workpiece data containing information relating to the workpiece machined by the machine tool, cutting fluid data containing information relating to the cutting fluid supplied to the machining area of ​​the machine tool by the cutting fluid supply device (3), machining process data containing information relating to a machining process of the workpiece by the machine tool, and cutting fluid discharge position data,which includes information relating to a discharge position of the cutting fluid through the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), and, the machine learning device (100, 100') a learning model memory (130) configured to store a learning model relating the machining state by the machine tool and a state of the cutting fluid supplied by the cutting fluid supply device (3) to the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e); and an estimation unit (120) configured to estimate an amount of cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) on the basis of the state data using the learning model stored in the learning model memory (130), and the cutting fluid quantity adjusting device (1, 1', 1'') further comprises a discharge quantity determining unit (38) which is designed to determine the quantity of cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) on the basis of an estimation result of the estimation unit (120), includes. [6] Cutting fluid quantity adjusting device (1, 1', 1'') according to claim 1, characterized by , that the preprocessing unit (36) as data used in supervised learning by the machine learning device (100, 100'), state data including at least tool data containing information relating to a tool used in machining the workpiece by the machine tool, machining condition data containing information relating to machining conditions in machining the workpiece by the machine tool, workpiece data containing information relating to the workpiece machined by the machine tool, cutting fluid data containing information relating to the cutting fluid supplied to the machining area of ​​the machine tool by the cutting fluid supply device (3), machining process data containing information relating to a machining process of the workpiece by the machine tool, cutting fluid discharge position data,which includes information relating to a discharge position of the cutting fluid through the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), and cutting fluid discharge amount data including an amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) included in the cutting fluid supply device (3), and label data including at least discharge amount adequacy data indicating the adequacy of the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), and, the machine learning device (100, 100') has a learning unit (110) that generates a first learning unit (112) that is configured to generate, on the basis of the state data and the label data, a first learning model that relates the machining state by the machine tool and a state of the cutting fluid supplied by the cutting fluid supply device (3) to the appropriateness of the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e); and a second learning unit (114) configured to generate, based on a result of estimation processing generated using the first learning model (112) generated by the first learning unit (110), a second learning model relating the machining state by the machine tool and the state of the cutting fluid supplied by the cutting fluid supply device (3) to the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), has. [7] Cutting fluid quantity adjusting device (1, 1', 1'') according to claim 1, characterized by , that the preprocessing unit (36) as data used in reinforcement learning by the machine learning device (100, 100'), state data including at least tool data containing information relating to a tool used in machining the workpiece by the machine tool, machining condition data containing information relating to machining conditions in machining the workpiece by the machine tool, workpiece data containing information relating to the workpiece machined by the machine tool, cutting fluid data containing information relating to the cutting fluid supplied to the machining area of ​​the machine tool by the cutting fluid supply device (3), machining process data containing information relating to a machining process of the workpiece by the machine tool, cutting fluid discharge position data,which includes information relating to a discharge position of the cutting fluid through the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), and cutting fluid discharge amount data including an amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) included in the cutting fluid supply device (3), and determination data including at least discharge amount determination data for determining the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), and workpiece condition determination data for determining a condition of the workpiece, and, the machine learning device (100, 100') a learning unit (110) configured to generate, on the basis of the state data and the determination data, a learning model that relates the machining state by the machine tool and a state of the cutting fluid supplied by the cutting fluid supply device (3) to a setting behavior for the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e); and a decision-making unit (122) configured to determine, on the basis of the state data, an adjustment of the discharge amount of the cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e), which is made using the learning model generated by the learning unit (110), and the cutting fluid quantity adjusting device (1, 1', 1'') further comprises a discharge quantity determining unit (38) configured to determine the quantity of cutting fluid discharged from the cutting fluid nozzle (301a, 301b, 301c, 301d, 301e) based on the determination of the decision-making unit (122), includes. [8] Cutting fluid quantity adjustment system (500, 500', 500''), wherein several devices are connected to each other via a network (5), characterized by that the plurality of devices comprise at least a first cutting fluid quantity adjusting device (1, 1', 1'') which is designed to be the cutting fluid quantity adjusting device (1, 1', 1'') according to one of claims 2, 4, 6 and 7. [9] Cutting fluid quantity adjustment system (500, 500', 500'') according to claim 8, characterized by , that the plurality of devices comprise a computer containing the machine learning device (100, 100'), the computer acquires at least one learning model generated by the learning in the learning unit (110) of the first cutting fluid quantity adjusting device (1, 1', 1''); and the machine learning device (100, 100') contained in the computer performs optimization and tightening based on the acquired learning model. [10] Cutting fluid quantity adjustment system (500, 500', 500'') according to claim 8, characterized by , that the plurality of devices comprise a second cutting fluid quantity adjusting device (1, 1', 1'') which is different from the first cutting fluid quantity adjusting device (1, 1', 1''); and a learning model by the learning unit (110) contained in the first cutting fluid quantity adjusting device (1, 1', 1'') is also used by the second cutting fluid quantity adjusting device (1, 1', 1''). [11] Cutting fluid quantity adjustment system (500, 500', 500'') according to claim 8, characterized by , that the plurality of devices comprise a second cutting fluid quantity adjusting device (1, 1', 1'') which is different from the first cutting fluid quantity adjusting device (1, 1', 1''); and Data observed in the second cutting fluid quantity adjusting device (1, 1', 1'') can be used via the network (5) in the learning by the learning unit (110) included in the first cutting fluid quantity adjusting device (1, 1', 1'').

Images