Tool selection device and machine learning device

Abstract

Tool selection device (1) for selecting a type of tool that can be used when machining a workpiece under a specified condition, wherein the tool selection device (1) a machine learning device (100) which learns, for a given condition, a type of tool that can be used in the machining of a workpiece, includes, whereby the machine learning device (100) a condition monitoring unit (106) that monitors machining condition data (S1), cutting condition data (S2), machining result data (S3), and tool data (S4) as state variables (S) indicating a current environmental state; and a learning unit (110) which, using the state variables (S), learns the distribution of the machining condition data (S1), the cutting condition data (S2), and the data related to the machining result (S3) in relation to the tool data (S4), exhibits wherein the distribution is configured in a multidimensional space which has axes that display the machining condition data (S1), the cutting condition data (S2) and the machining result data (S3), and generates different clusters for each tool type, each of which displays a tool type-based trend of the learned distribution based on the learned distribution, wherein the tool selection device (1) is further configured to To determine priorities for the respective tool types based on the cluster densities of the different clusters, to select a tool type for a given condition that can be used when machining the workpiece based on the learned distribution trend based on the tool type, and to control a machine tool and an actuator in order to use the selected tool type when machining the workpiece under the specified condition.

Description

BACKGROUND OF THE INVENTIONField of the invention

[0001] The present invention relates to a tool selecting device that selects an appropriate tool and determines an appropriate cutting condition, and a machine learning device. Description of the state of the art

[0002] To machine a workpiece with a machine tool, a tool must be selected for machining and cutting conditions must be determined for machining using the selected tool. When selecting a tool and determining cutting conditions, a tool to be used and cutting conditions (spindle rotation speed, feed rate) are provisionally determined during the machining planning period using a tool catalog or a tool selection device prepared by a tool manufacturer. The cutting conditions are adjusted by repeating trial machining operations using the provisionally determined tool under the provisionally determined cutting conditions. In this way, a tool and cutting conditions are determined when it is confirmed that no problems have occurred.

[0003] As a conventional technique for determining a cutting condition for use in machining using a machine tool, JP 2017-064837 A discloses a technique for determining a cutting condition based on the conditions of a tool load allowable during machining.

[0004] A cutting condition depends heavily on machine characteristics such as spindle output and rigidity, or on a machining condition such as the material quality of a workpiece or a machining method. However, the information provided by a tool catalog or tool selector is for general use, so trial machining and verification of machining results must be repeated many times to derive a condition corresponding to a machine to be used for machining, or a machining condition. This is time-consuming and labor-intensive. Furthermore, when adjusting the cutting condition, a machining program or tool data must also be corrected accordingly. This process also requires time.In addition, the cutting condition must be further adjusted depending on whether an operation is to be performed in which cycle time is given priority, or an operation in which machining accuracy or tool life is to be given priority.

[0005] On the other hand, the technique disclosed in JP 2017-06487 A allows for the determination of a cutting condition. However, the cutting condition can be determined during processing using a specific tool, so the technique is not useful for selecting another suitable tool according to the purpose of the processing.

[0006] From US 5,801,963 A (D1) a method for predicting optimal machining conditions and for selecting machining parameters and tool inserts for turning operations is known, wherein the method comprises the following steps: (1) development of a process model for linking performance variables and process parameters for turning operations; (2) selection of one of the performance variables for optimization, while maintaining the remaining performance variables and process parameters as constraints for the optimization process; and (3) application of non-linear programming techniques to the process model in order to determine optimal cutting conditions for a specific tool insert or to select an optimal tool insert for specific machining performance requirements.JP H03-294147 A is directed to a tool selection system in which machining data representing characteristics of a machining object and tool data relating to a tool to be used are input by a user. The tool selection system then determines a similar case from the stored data and provides estimation results. JP S64-45545 A is similarly directed to an automatic tool selection device of a machine tool, wherein an optimal tool is automatically selected based on stored tool information and machining information for each machining point, taking into account the minimization of machining preparation time and machining time.From DE 10 2017 105 224 A1, a machine learning device is known which learns laser processing condition data of a laser processing system, wherein the laser processing condition data are optimized based on observed data on the processing quality of a workpiece. SUMMARY OF THE INVENTION

[0007] Therefore, it is an object of the present invention to provide a tool selection device capable of selecting a suitable tool and determining a suitable cutting condition. This object is achieved by a tool selection device according to claim 1 or 2.

[0008] A tool selection device according to the present invention implements a mechanism that uses machine learning to derive a suitable tool and cutting condition (a rotational speed, a feed rate) for a peak spindle load that satisfies a machining condition (the material quality of a workpiece, a machining type, the cutting depth, the cutting width) and required operating specifications (speed, accuracy, tool life). A machining program and tool data are automatically rewritten to adapt to the tool and cutting condition derived by the mechanism.

[0009] A worker uses the above mechanism at the time of inputting a machining program and workpiece information during a setup process, and can thereby select a tool and set a cutting condition corresponding to the machining without performing any trial operation or performing an operation to change the machining program and tool data.

[0010] A tool selecting device according to one form of the present invention selects a type of tool usable in machining a workpiece under a set condition, and includes a machine learning device that learns a type of tool usable in machining a workpiece for a set condition.The machine learning device includes a state observation unit that observes data related to a machining condition, data related to a cutting condition, data related to a machining result, and data related to a tool as state variables indicating a current environmental state; and a learning unit that learns the distribution of the data related to the machining condition, the data related to the cutting condition, and the data related to the machining condition with respect to the data related to the tool using the state variables.

[0011] A tool selecting device according to another form of the present invention selects a type of tool usable in machining a workpiece under a set condition, and has a machine learning device that has learned a type of tool usable for machining a workpiece for a set condition.The machine learning device includes a state observation unit that observes data related to a machining condition, data related to a cutting condition, and data related to a machining result as state variables indicating a current environmental state; a learning unit that has learned the distribution of the data related to the machining condition, the data related to the cutting condition, and the data related to the machining result with respect to data related to a tool used in machining; and a determination unit that determines a tool type usable under a condition set by the state variables based on the state variables observed by the state observation unit and a learning result by the learning unit, and outputs the determined tool type.

[0012] The determination unit can determine the cutting condition that can be set together with the tool type determined to be usable and output the cutting condition.

[0013] A machine learning apparatus according to one form of the present invention learns, for a set condition, a type of tool usable in machining a workpiece, and includes a state observation unit that observes data related to a machining condition, data related to a cutting condition, data related to a machining result, and data related to a tool as state variables indicating a current environmental state; and a learning unit that learns, using the state variables, the distribution of the data related to the machining condition, the data related to the cutting condition, and the data related to the machining condition with respect to the data related to the tool.

[0014] A machine learning device according to another form of the present invention is a machine learning device that has learned, for a set condition, a tool type usable in machining a workpiece, and includes a state observation unit that observes data related to a machining condition, data related to a cutting condition, and data related to a machining result as state variables indicating a current environmental state; a learning unit that has learned the distribution of the data related to the machining condition, the data related to the cutting condition, and the data related to the machining result with respect to data related to a tool used in machining;and a determination unit that determines, based on the state variables observed by the state observation unit and a learning result by the learning unit, a tool type usable under a condition set by the state variables and outputs the determined tool type;

[0015] According to the present invention, the selection of a tool, the derivation of a cutting condition, and its application can be easily performed to correspond to application purposes at a location such as mechanical properties, a machining condition, and important points (speed, accuracy, tool life) during operation, thereby shortening the time required for a setup process. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a schematic hardware configuration diagram of a tool selecting device according to a first embodiment; Fig. 2 is a schematic functional block diagram of the tool selecting device according to the first embodiment; Fig. 3 is an explanatory diagram showing a cluster analysis conducted by a learning unit; Fig. Figure 4 is a schematic functional block diagram illustrating one form of a system having tool selection devices; and Fig. Figure 5 is a schematic functional block diagram illustrating another form of the system having tool selectors. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] Fig. 1 is a schematic hardware configuration diagram showing the main components of a tool selecting device according to an embodiment of the present invention.

[0017] A tool selecting device 1 may be embodied as a control unit for controlling a machine such as a machine tool, may be embodied as a personal computer provided together with a control unit for controlling a machine, or may be embodied as a computer such as a cell computer, a host computer, or a cloud server connected to a control unit via a network. Fig. 1 illustrates an example in which the tool selection device 1 is designed as a control unit for controlling a machine tool.

[0018] A CPU 11 included in the tool selector 1 according to the present embodiment is a processor that controls the entire tool selector 1, reads a system program stored in a ROM 12 via a bus 20, and controls the entire tool selector 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 a RAM 13.

[0019] A non-volatile memory 14 is configured as a memory whose storage state is retained even after the power to the tool selector 1 is turned off, since backup is provided by, for example, a battery (not shown). In the non-volatile memory 14, for example, a machining program read from an external device 72 via an interface 15, a machining program input via a display / MDI unit 70, and various data acquired from the components of the tool selector 1 or the machine tool (for example, a machining state input by a worker (the material quality of a workpiece, a machining type, the cutting depth, the cutting amount), tool information, a cutting condition (the spindle rotation speed, the feed rate), a spindle load during machining under the cutting condition, etc.) are stored.The machining programs and various data stored in the nonvolatile memory 14 can be developed into the RAM 13 upon execution or use. Furthermore, various system programs (including a system program for controlling communication with a machine learning device 100 (described later)) such as well-known analysis programs, etc., have been written in advance into the ROM 12.

[0020] The interface 15 connects the tool selector 1 to an external device 72 such as an adapter. Programs, various parameters, and the like are read from the external device 72. Furthermore, programs and various parameters, etc., edited in the tool selector 1 can be stored in an external storage medium via the external device 72. Based on a sequence program incorporated in the tool selector 1, a programmable machine controller (PMC) 16 outputs signals to a machine tool and a peripheral device thereof (e.g., an actuator such as a robot hand for tool exchange) via an I / O unit 17, and controls the machine tool and the peripheral device. Furthermore, the PMC 16 receives signals from various switches, etc.on an operation panel provided on the main body of the machine tool, performs necessary signal processing on the signals, and forwards the processed signals to the CPU 11.

[0021] The display / MDI unit 70 is a manual data input device equipped with a display and a keyboard, etc. An interface 18 receives a command and data from the keyboard of the display / MDI unit 70 and forwards the command and data to the CPU 11. An interface 19 is connected to an operation panel 71 equipped with a manual pulse generator used for manually driving axes, etc.

[0022] An axis control circuit 30 for controlling axes included in the machine tool receives a command regarding a movement amount of an axis from the CPU 11 and outputs a command for the axis to a servo amplifier 40. The servo amplifier 40 receives the command and drives a servo motor 50 that moves the axis included in the machine tool. The servo motor 50 for the axis has a built-in position / speed detector, feeds a position / speed feedback signal from the position / speed detector back to the axis control circuit 30, and performs position / speed feedback control. It should be noted that in the hardware configuration diagram in Fig. 1, an axis control circuit 30, a servo amplifier 40, and a servo motor 50 are shown. However, in practice, as many axis control circuits 30, servo amplifiers 40, and servo motors 50 are prepared as there are axes in a machine tool to be controlled.

[0023] A spindle control circuit 60 receives a spindle rotation command for a machine tool and outputs a spindle speed signal to a spindle amplifier 61. The spindle amplifier 61 receives the spindle speed signal, rotates a spindle motor 62 of the machine tool at the commanded rotation speed, and drives a tool. A position encoder 63 is coupled to the spindle motor 62. The position encoder 63 outputs a feedback pulse in synchronization with the rotation of the spindle. The feedback pulse is read by the CPU 11.

[0024] An interface 21 connects the tool selection device 1 to the machine learning device 100. The machine learning device 100 includes a processor 101 that controls 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 processes related to machine learning, and a non-volatile memory 104 used to store a learning model and the like. The machine learning device 100 can process various information that can be acquired by the tool selection device 1 via the interface 21 (for example, a machining condition (the material quality of a workpiece, a machining type, the cutting depth, the cutting amount, etc.) input by the worker), a cutting condition (the spindle rotation speed, the feed rate), and the operating state (the spindle load, etc.).during machining). In addition, the tool selection device 1 displays a tool selection and a cutting condition suggestion output from the machine learning device 100 on the display / MDI unit 70, and selects a tool and sets a cutting condition (corrects, for example, a machining program or sets tool data) based on the selection made by the operator who checked the display.

[0025] Fig. 2 is a schematic functional block diagram of the tool selecting device 1 and the machine learning device 100 according to the first embodiment.

[0026] The functional blocks that are in Fig. 2 are executed by the CPU 11 provided in the tool selection device 1 and the processor 101 of the Fig. 1, which executes respective system programs for controlling the operation of the components of the tool selection device 1 and the machine learning device 100.

[0027] The tool selecting device 1 of the present embodiment includes a control unit 34 that controls motors such as the servo motors 50 and the spindle motor 62 provided in the machine tool 2 and controls a peripheral machine (not shown) of the machine tool 2 based on the machining program or machining conditions stored in the non-volatile memory 14, the setting of the cutting conditions, or the like, and a tool selecting unit 36 ​​that displays, on the display / MDI unit 70, a tool type determined by the machine learning device 100 as usable in machining or other conditions, and sets the tool type selected by the worker or other conditions as information for use in machining.

[0028] On the other hand, the machine learning device 100 provided in the tool selection device 1 includes software (a learning algorithm, etc.) and hardware (the processor 101, etc.) for performing self-learning of data related to a tool with respect to data related to a machining condition (the material quality of a workpiece, a machining type, the cutting depth, the cutting amount, etc.), data related to a machining result (the spindle load during machining, and the like), and data related to a cutting condition (the spindle rotation speed, the feed rate), and for performing self-learning of the determination of data related to a tool with respect to data related to an input cutting condition (the material quality of a workpiece, a machining type, the cutting depth, the cutting amount, etc.).), data related to the cutting condition (spindle rotation speed, feed rate), and data related to the machining result (spindle load, machining accuracy, etc., during machining) through so-called machine learning. What is learned by the machine learning device 100 provided in the tool selection device 1 corresponds to a model structure that indicates the correlation of data related to a tool with data related to the machining condition (the material quality of a workpiece, a machining type, the cutting depth, the cutting amount, etc.), data related to the cutting condition (spindle rotation speed, feed rate), and data related to the machining result (spindle load, machining accuracy, etc., during machining).

[0029] As in the function blocks in Fig. As shown in Fig. 2, the machine learning device 100 provided in the tool selection device 1 includes a state observation unit 106, a learning unit 110, and a determination unit 122. The state observation unit 106 observes state variables S, which include machining condition data S1 containing data related to the machining condition (the material quality of a workpiece, a machining type, the cutting depth, the cutting amount, etc.), cutting condition data S2 containing data related to the cutting condition (the spindle rotation speed, the feed rate), machining result data S3 containing data related to machining data (the spindle load, the machining accuracy, etc., during machining), and tool data S4 containing data related to the tool.The learning unit 110 learns, by using the state variable S, data related to the tool in conjunction with data related to the machining condition (the material quality of a workpiece, a machining type, the cutting depth, the cutting amount, etc.), data related to the cutting condition (the spindle rotation speed, the feed rate), and data related to the machining result (the spindle load during machining, etc.). And the determination unit 122 determines, by using a learned model learned by the learning unit 110, data related to the tool in conjunction with data related to the machining condition (the material quality of a workpiece, a machining type, the cutting depth, the cutting amount, etc.).), data related to the cutting condition (spindle rotation speed, feed rate) and data related to the machining result (spindle load, machining accuracy, etc., during machining).

[0030] Among the state variables S observed by the state observation unit 106, the machining condition data S1 can be obtained as a machining condition set by a worker during machining using the machine tool 2. The machining condition may include, for example, the material quality of a workpiece to be machined, a machining type such as direct thread cutting or face milling, and the cutting amount or depth of cutting of the workpiece by a tool. The state observation unit 106 observes a machining condition input to the machine tool 2 or the control unit, or a machining condition set in the machining program, as machining condition data S1.

[0031] Among the state variables S observed by the state observation unit 106, the cutting condition data S2 can be obtained as a cutting condition set by a worker during machining using the machine tool 2. The cutting condition may include, for example, the spindle rotation speed, the feed rate, and the like. The state observation unit 106 observes the cutting condition input to the machine tool 2 or the control unit, or the cutting condition set in the machining program, as cutting condition data S2.

[0032] Among the state variables S observed by the state observation unit 106, the machining result data S3 may be obtained as the maximum value of the feed axis load, the spindle load, or the like detected during machining of a workpiece using a tool attached to the machine tool 2 (observed as tool data S4) under the specified machining condition (observed as machining condition data S1) and the cutting condition (observed as cutting condition data S2) by the machine tool 2 or a sensor attached to the machine tool 2, etc.), or may be obtained as the machining error indicating a dimensional error of the machined workpiece from a design value, or the like.For the machining result data S3, a value obtained by detecting a physical quantity that caused a machining defect, a defect of the machine tool 2, a breakage of the tool, or the like during the machining being performed, data for evaluating the result of the machining, or a value input by the worker, or the like may be used.

[0033] Among the state variables S observed by the state observation unit 106, the tool data S4 can be acquired as information related to a tool for use in machining using the machine tool 2. The tool-related information may include, for example, information that can uniquely identify a tool type (e.g., the model number of the tool), and may include, if necessary, the manufacturer of the tool, the material quality (hardness) of the tool, etc. Mainly, for the tool-related information, data input by the worker or data included in a machining command specified by a higher-level device such as a cell computer can be used.

[0034] The learning unit 110 performs cluster analysis based on the state variables S (the machining condition data S1, the cutting condition data S2, the machining result data S3, the tool data S4) according to an arbitrary learning algorithm, which is generally referred to as machine learning, and stores (learns) a cluster generated as a result of the cluster analysis as a learned model. The learning unit 110 generates a cluster based on a predetermined number of the state variables S (the machining condition data S1, the cutting condition data S2, the machining result data S3, the tool data S4) obtained during normal machining of a workpiece.For the state variables S used to generate a cluster, collected data (big data) obtained, for example, via a wired / wireless network from the machine tool 2 installed in a factory can be used. As a result of such learning, the learning unit 110 analyzes the distribution of the machining condition (machining condition data S1), the cutting condition (cutting condition data S2), and the operating state (machining result data S3) of the machine tool 2 based on the tool type (tool data S4) as a cluster set.

[0035] Fig. Figure 3 is a diagram showing an example of cluster sentences generated by the learning unit 110. It should be noted that in Fig. 3 For convenience, the data distribution space is three-dimensional, and the axes respectively indicate the cutting depth (machining condition data S1), the cutting speed (cutting condition data S2), and the spindle load (machining result data S3). However, in practice, the data distribution is given in a multidimensional space with axes indicating the respective data points obtained as state variables S (except for the tool type (tool data S4).

[0036] As in Fig. 3, the learning unit 110 generates different clusters for at least each tool type. As shown in Fig. As shown in Figure 3, clusters generated by the learning unit 110 each indicate the tool-type-based trend of the operating state distribution with respect to the machining condition and the cutting condition. That is, the clusters each indicate the trend of the operating state (spindle load) obtained when machining is performed using a predetermined tool under the specified machining condition (cutting depth) and the specified cutting condition (cutting speed). The clusters are used when the determination unit 122 makes a determination to select a tool under constraints such as the machining condition, cutting condition, and machining state.

[0037] The determination unit 122 determines which tool is suitable for use based on the learned model (a cluster of the machining condition (the machining condition data S1), the cutting condition (the cutting condition data S2), and the operating state of the machine tool 2 (the machining result data S3) for each of tool types (the tool data S4)) obtained by performing learning based on the state variables S (the machining state data S1, the cutting condition data S2, the machining result data S3, the tool data S4) acquired by the learning unit 110 when machining a workpiece was normally performed, and on the basis of newly observed (inputted) machining state data S1, cutting condition data S2, and machining result data S3.

[0038] Using Fig. 3 as an example, the operation of the determination unit 122 will be described. When new machining condition data S1, new cutting condition data S2, and new machining result data S3 are observed (inputted) in a state where a cluster obtained by normal machining of a workpiece has been generated, the determination unit 122 analyzes the relationship between these data and a cluster for each tool type, and determines a tool type corresponding to the cluster to which the newly inputted state variables S belong as a tool type usable under the set conditions. Note that the state variables S input in this case may be accompanied by a specific range.For example, in a case where a spindle load condition that the spindle load is equal to or lower than L1 is input as machining result data S3, tool types corresponding to all clusters overlapping this range can be determined as tool types usable under the set condition.

[0039] It is possible that the determination unit 122 not only simply determines a tool type, but also may determine the priorities of tool types for use under the set condition based on the distances of the positions of newly observed (input) machining condition data S1, cutting condition data S2, and machining result data S3 in the cluster space from the centers of respective clusters, or on the basis of the cluster densities of respective clusters at the positions of newly observed (input) machining condition data S1, cutting condition data S2, and machining result data S3.

[0040] Furthermore, after determining a new usable tool type, the determination unit 122 can determine, based on the observed (input) machining condition data S1, cutting condition data S2, and machining result data S3, the range of a machining condition or cutting condition that can be set when machining is performed using a tool of the determined tool type. For example, after determining a usable tool type, the determination unit 122 can determine the maximum value of the cutting speed in a range in which the machining result (spindle load, etc.) set in the cluster for the tool type can be maintained.

[0041] As described above, if a tool type capable of normal machining under the specified machining condition and cutting condition to achieve the desired machining result can be automatically determined without involving any calculation or estimation, a worker can quickly make a determination to select an appropriate tool merely by inputting a required machining condition, a required cutting condition, and a desired machining result (or reading them from a CAD / CAM, etc.).

[0042] Then, tool types (and the priorities given to the tool types, the range of the machining condition or the cutting condition) determined by the determination unit 122 are output to the tool selection unit 36. The tool selection unit 36 ​​displays one or more tool types usable under the determined conditions on the display / MDI unit 70. The worker selects a tool type as a tool to be used in machining by operating the display / MDI unit 70, and sets the selected tool type in the machining program, etc., stored in the nonvolatile memory 14.If the determination unit 122 has determined the machining condition or cutting condition ranges that can be set for machining using respective tool types, the tool selection unit 36 ​​further displays the machining condition or cutting condition ranges to prompt the worker to change or not change the set machining condition or cutting condition. Through the tool selection unit 36, the machining condition or cutting condition set by the worker is set in the machining program stored in the nonvolatile memory 14 or a setting area provided in the nonvolatile memory 14, or the like.

[0043] Fig. 4 illustrates a system 80 including the machine tools 2 according to one embodiment.

[0044] The system 80 comprises a plurality of machine tools 2 with the same machine structure and a network 82 through which the machine tools 2 are connected to one another. At least one of the machine tools 2 is configured as the machine tool 2 with the aforementioned tool selection device 1. The machine tools 2 each have a general machine tool structure necessary for machining workpieces.

[0045] In the system 80 having the above configuration, the machine tool 2 among the plurality of machine tools 2 that has the tool selecting device 1 can automatically and accurately obtain, for each of the machining-related conditions set for the machine tools 2 (including the machine tools 2 that do not have the tool selecting device 1), a tool type usable under the condition set for the machine tool 2 by using the learning result by the learning unit 100. Furthermore, the tool selecting device 1 of the at least one machine tool 2 can be configured to perform the same learning for all the machine tools 2 based on the state variables S obtained for each of the other machine tools 2, and allow the learning result to be commonly used by all the machine tools 2.Therefore, the system 80 can improve the learning speeds or reliability of various data detected by the machine tools 2 upon receiving input of a wider variety of data sets (including the state variables S).

[0046] Fig. 5 illustrates the system 80 provided with the machine tools 2 according to another form.

[0047] The system 80 is formed of a plurality of machine tools 2 having the same machine structure and the machine learning device 100, which is connected to the machine tools 2 via a network 82 and is installed in a computer such as a cell computer, a host computer, or a cloud server. Each of the machine tools 2 has a tool selection device 1', which is designed as a control unit for the machine tool 2. It should be noted that the tool selection device 1' in the Fig. 5, the non-volatile memory 14, the control unit 34 and the tool selection unit 36, which with reference to Fig. 2 described.

[0048] In the system having the above-described configuration 80, the machine learning device 100 performs the same learning for all the machine tools 2 based on the state variables S (and the label data L) obtained for each of the machine tools 2. Using the learning result by the machine learning device 100, the selection of a tool usable under a condition set by each of the machine tools 2 and a suggested cutting condition therefor are sent to the tool selecting device 1' provided in the machine tool 2. Then, in the tool selecting device 1', based on the selection of a tool and the suggested cutting condition obtained by the machine learning device 100, a suitable tool for use by the machine tool 2 can be selected and a cutting condition therefor can be determined.

[0049] With this structure, if necessary, the required number of machine tools 2 can be connected to the machine learning device 100 regardless of the respective positions of the machine tools or regardless of timing.

[0050] Embodiments of the present invention have been described above. However, the present invention is not limited to the above-mentioned embodiments; various embodiments can be implemented by making appropriate changes thereto.

[0051] For example, the learning algorithm executed by the machine learning device 100, the calculation algorithm executed by the machine learning device 100, and the control algorithm executed by the tool selecting device 1 are not limited to those described above, but various algorithms may be used therefor.

Claims

[1] Tool selection device (1) for selecting a type of tool which can be used in machining a workpiece under a specified condition, the tool selection device (1) a machine learning device (100) which learns a type of tool that can be used in machining a workpiece for a set condition, includes, where the machine learning device (100) a state observation unit (106) that observes machining condition data (S1), cutting condition data (S2), machining result data (S3), and tool data (S4) as state variables (S) indicating a current environmental state; and a learning unit (110) which learns the distribution of the machining condition data (S1), the cutting condition data (S2), and the data related to the machining result (S3) with respect to the tool data (S4) using the state variables (S), has, wherein the distribution is configured in a multi-dimensional space having axes indicating the machining condition data (S1), the cutting condition data (S2) and the machining result data (S3), respectively, and generates different clusters for respective tool types, each indicating a tool type-based trend of the learned distribution based on the learned distribution, wherein the tool selection device (1) is further configured to To determine priorities for the respective tool types based on the cluster densities of the different clusters, to select a tool type that can be used in machining the workpiece for a given condition based on the trend of the learned distribution based on the tool type, and to control a machine tool and an actuator to use the selected tool type in machining the workpiece under the designated condition. [2] Tool selection device (1) for selecting a type of tool which can be used in machining a workpiece under a specified condition, the tool selection device (1) a machine learning device (100) which has learned a type of tool that can be used for machining a workpiece for a set condition, includes, where the machine learning device (100) a state observation unit (106) that observes machining condition data (S1), cutting condition data (S2), and machining result data (S3) as state variables (S) indicating a current environmental state; a learning unit (110) that has learned the distribution of the machining condition data (S1), the cutting condition data (S2), and the machining result data (S3) with respect to tool data (S4) of a tool used in machining, wherein the distribution is configured in a multi-dimensional space having axes indicating the machining condition data (S1), the cutting condition data (S2), and the machining result data (S3), respectively, and the learning unit (110) has generated different clusters for respective tool types, each indicating a tool type-based trend of the learned distribution based on the learned distribution; and a determination unit (122) that determines priorities for the respective tool types based on the cluster densities of the various clusters, and determines a tool type that can be used under a condition set by the state variables (S) and outputs the determined tool type based on the state variables (S) observed by the state observation unit (106) and the tool type-based trend of the distribution learned by the learning unit (110), has, wherein the tool selection device (1) is further configured to: to control a machine tool and an actuator to use the specific tool type under the condition specified by the state variables (S). [3] The tool selecting device (1) according to claim 2, wherein the determining unit (122) determines the cutting condition that can be set together with the tool type determined to be usable, and outputs the cutting condition.

Images